The impact of stock options compensation on earnings and probability of bankruptcy.

Akindayomi, Akinloye ; Warsame, Hussein A.

INTRODUCTION

As the debate around excessive corporate executive compensation heats up in the United States in the era of Troubled Asset Relief Program (TARP) (1), the debate on the efficacy of stock options compensation is not yet settled. The restrictions on executive salaries and bonuses by firms that benefitted from TARP is likely to spread to comparable companies in the US. To avoid high political costs while at the same time keeping the option of providing incentives for managers to optimize firm value, Board of Directors may opt to increase equity related compensations such as stock options.

The objective of this study is to investigate the impact of stock options compensation on earnings and probability of bankruptcy of the firm. Hanlon, Rajgopal and Shevlin 2003 (HRS) document the incentive alignment hypothesis of executive stock options, but the authors use reported operating performance as the dependent measure. We argue that the positive contributions of executive stock options to reported earnings documented in that study could have been exaggerated if one considers the real potentials of earnings management, and so corporate boards and compensation committees should exercise caution in the interpretations of HRS finding. Therefore, in part, we examine executive stock options contributions to other measure of earnings after controlling for earnings management using nondiscretionary earnings as a dependent measure. While we find a positive contribution consistent with incentive alignment, the magnitude of such contribution is substantially lower. This suggests that nondiscretionary earnings will be a better measure of corporate performance as a guide for executive compensation decisions.

Prior studies have examined empirically and analytically a variety of issues ranging from the role of taxes in the decision to grant options (e.g., Klassen and Mawani, 2000), the choice between incentive stock options and nonqualified options (e.g., Austin et al, 1998), the tax deductibility of stock options (e.g., Balsam et al., 1996 & 1997; Mawani, 2003a), to the firm's disclosure behavior around the granting of the options (e.g., Aboody and Kasznik, 2000) as well as the tax and accounting income consideration for the cancellation of executive stock options (e.g., Mawani 2003b). However, very few (e.g., HRS; Kato et al, 2005; Sanders and Hambrick, 2007) have attempted to provide direct evidence of the impact of executive stock options on the firm's earnings. HRS conclude that every dollar of stock options (using Black-Scholes values) granted to the top five executives contributes $3.71 to future operating earnings of the company over the next five years. Kato et al. (2005), using Japanese data and an event study methodology, also conclude that operating performance improves with stock options. However Sanders and Hambrick (2007) have shown that while stock options do affect CEO behaviors, their heavy use produces more losses than gains. Other agency theorists wondered whether the traditional ESO plans for executives are not leading to creative ways of managing earnings while ignoring the cost of equity (Jensen, Murphy, and Wruck, 2004).

These mixed results are manifestations that the question of whether stock options induce mangers to take appropriate actions is still not settled. Researchers using the incentive alignment hypothesis argue that stock options compensation could be utilized to reduce the incentives asymmetry between managers and shareholders (e.g., Rajgopal and Shevlin, 2002; HRS; Mawani, 2003a). However, other researchers using the rent extraction hypothesis argue that this compensation package can be a conduit of transferring wealth from shareholders to management/top executives (e.g., Johnson 2003; Aboody and Kasznik, 2000; Baker, Collins, and Reitenga, 2003).

Our study is motivated by the need to fill this important gap in the literature with the intent to examining the impact of granting options to top corporate executives on the firms' earnings and the probability of bankruptcy, and by extension the value of the firm. We build on the future operating earnings-based model used by HRS which we believe has advantages over models using ex-post stock price performance like that used by Kato et. (2005) Future operating earnings do not suffer from stockholder expectation problem embedded in ex-post price performance of shares. We adjust HRS's model for challenges suggested by HRS and Larker (2003). We use the nondiscretionary component of earnings to avoid problems caused by earnings management. As HRS recognize, if some firms overstate or understate earnings the results "might reflect earnings management as a function of ESO grant values rather than economic payoffs" (HRS, pp 37). We also took into account the alternative "forward-looking" research design suggested by Larcker (2003) to address similar research questions raised by HRS.

Furthermore, we use Altman's Z-score to test suggestions in the literature that ESOs induce managers to take too many risks and may cause financial distress. We use the probability of bankruptcy represented by the Altman's Z-score as a proxy for a change in the cost of equity. In effect, Altman's Z-score is inversely related to the cost of equity. The higher the Altman's Z-Score, the lower is the cost of equity. Results from our models are consistent with the incentive alignment hypothesis and are inconsistent with the overall conclusion of Sanders and Hambrick (2007) that stock options cause more losses than gains. However, they are consistent with Sanders and Hambrick (2007)'s less emphasized result that moderate levels of stock options (20% to 50%) do actually induce executives to become more risk neutral (less risk averse) with performance symmetrically divided between losses and gains. The overall implication of our results is that, at least in our sample of firms, partly compensating top executives with stock options not only induces them to improve earnings, it also motivates them to take moderate risks.

The rest of the paper proceeds as follows. Section 2 provides the theoretical background for the study and the hypotheses tested. Research methodology and design are the subjects of section 3. Section 4 provides the results and findings of the study. The final section provides a summary and the potential limitations/constraints that this study may face.

THEORETICAL BACKGROUND

Executive compensation constitutes a typical problem domain for agency theory. The relationship between the shareholders and the executives of a firm is one in which the two groups have partly differing goals and risk preferences. Executives are thought to be more risk averse than shareholders. This is due to the likelihood that executives, whose incomes and reputation are tied to their firms, may not have as many opportunities as shareholders to effect appropriate levels of diversification for themselves (Eisenhardt, 1989). Shareholders are more likely to be risk neutral, while executives are more likely to be risk averse. The result would be that executives avoid profitable projects with a probability of a downside, which may lead to lower returns. Consistent with seminal works in agency theory (such as Jensen and Meckling, 1976), the solution to the problem is to move the executives' risk-averse preferences to risk-neutrality. Stock options, not only add a feature of outcome-orientation to any salary contract, which is primarily behavior-oriented, but they also increase the firm ownership by executives which decreases opportunism. Basically, any action taken by executives to reward themselves will simultaneously reward the shareholders. This is the incentive alignment perspective that makes some researchers (e.g. HRS; Kato et al) to argue that the motivational potentials of stock options should motivate top executives to act in a way that maximizes firm value.

However, the question that agency theorists were grabbling with lately is whether the resultant executive behavior includes sensible risk taking (Jensen et al, 2004; HRS; Sanders and Hambrick, 2007). Researchers have shown that, while stock options have induced executives to take more risks, there are doubts that these risks are value enhancing. Sanders and Hambrick (2007) show that moderate levels of stock options (20% to 50%) do induce executives to become more risk neutral (less risk averse) with performance symmetrically divided between losses and gains. On the other hand, more option-loaded executives produced more big losses than big gains (Sanders and Hambrick, 2007, p.1070). The extreme results of high option levels are plausible given the fact that stock options bestow on holders the opportunity to participate in the improved or enhanced share price without directly partaking in the downside loss, if it eventually occurs.

RESEARCH METHODOLOGY AND DESIGN

Consistent with the dictates of agency theory, the unit of analysis for this problem domain is the contract between the shareholders (principal) and the executives (agents). Specifically, we will look at the impact of compensating top executives with stock options on earnings and probability of bankruptcy of the firm. Earnings and probability of bankruptcy have direct impact on firm value. However, instead of looking at the value (2) of the firm directly, we will look at the accounting return and a proxy for the risk incurred in earning that return (3). A change in the expected earnings or a change in the rate used to discount the future earnings or the combination of changes could cause a change in the value of the firm. In other words, an increase (decrease) in earnings or decrease (increase) in discount rate will lead to an increase (decrease) in the value of the firm, all else equal. Significant increase in the probability of bankruptcy will normally increase the required rate of return used to discount future earnings thus reducing the value of the firm. However, researchers are yet to agree on whether or not the use of employee/executive stock option is good for the shareholders and how it affects those components of firm value [see for example, Johnson, 2003; Mawani, 2003a; HRS; Kato et al, 2005; Sanders and Hambrick, 2007). Testing for performance both in terms of return (earnings) and in terms of risk may yield more compelling evidence of stock option compensation efficacy.

Earnings

Earnings, in the accounting sense, are generally the difference between revenues and expenses of operating activities. Due to the tendencies of executives/managers to take leverage of their discretionary powers in smoothing earnings, research in the earnings management literature has indicated that reported earnings might not be persistent and thus might not reflect the 'true' earnings components. To estimate "true earnings", accounting scholars proposed several methods to remove the effect of discretionary components of earnings from the reported earnings (see for example, Dechow et al, 1995; Jones, 1991; Gaver, 1995; Reitenga et al, 2002; Baker et al, 2003; Kang and Sivaramakrishnan, 1995; Cohen and Zarowin, 2010). In this study, we follow the approaches of Dechow et al (1995) to calculate 'nondiscretionary earnings' as a proxy for 'true' earnings. Therefore, in order to capture the effect of our measure of earnings on the firm's value vis-a-vis executive stock options compensation, we put forward the following two hypotheses that relate to reported earnings ([H.sub.1]) and to nondiscretionary (true) earnings ([H.sub.2]).

[H.sub.1]: Ceteris paribus, the higher the use of Executive stock options, the higher the reported operating earnings of the firm.

In testing this hypothesis, we try to replicate the results of HRS after adjusting for some missing variables suggested by Larker (2003). Hypothesis 2 adjusts HRS for earnings management.

[H.sub.2]: Ceteris paribus, the higher the use of Executive stock options, the higher the nondiscretionary earnings of the firm.

In hypothesis 1, the dependent measure is the reported operating income and the estimated empirical model, using least squares regression, is presented as:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (1)

Where:

OPINC = Operating Income before depreciation scaled by Sales of firm i at time t.

TA = Total Assets of firm i at time t

BSO = Black-Scholes value of executive stock options granted to top 5 Executives. BSO is also squared to adjust for an observed non-linearity in the relationship between BSO and OPINC.

R&D = Research and development expenses of firm i during the year t - k (k = 0 - 5)

[sigma][(OPINC).sub.i,t-1] = Standard deviation of earnings measures estimated over the prior 5 year, for firm i.

S = is the annual sales in time t.

Equation (1) above is the baseline model of HRS for examining the incentives potential effects of executive stock options. However, this baseline model does not control for previous firm's performance and as argued by Larcker (2003), failure to control for previous firm's performance ([OPINC/S.sub.i,t-1]) might be an essential omission in the HRS baseline model. Therefore, in the spirit of Larcker (2003) argument, we control for firm's previous performance and thus modify equation 1 as:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (2)

In order to examine the effect of earnings management vis-a-vis the use of executive stock options, we replace OPINC/S in equation 1 and 2 with NDE/S (nondiscretionary earnings) as in

(3) and (4) below:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (3)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)

All variables are as described in (1). The industry dummies are based on a two-digit SIC code classification, unless otherwise stated, while the year dummies represents the fiscal year when operating/nondiscretionary is measured. All variables are scaled by sales to control for possible size effects and the possibility of heteroscedascticity. The standard deviation estimated over the prior 5 years is expected to control for the possible relation between firm risk and future earnings. This is consistent with Core et al (1999) specification (see also HRS). Other compensation related variables, such as cash compensation and the number of exercisable options in the money, that could simultaneously impact a firm's performance are also controlled for in the sensitivity analysis section.

Nondiscretionary earnings are measured as nondiscretionary accrual plus cash flow from operations. Nondiscretionary accrual is measured using modified Jones model as specified by Dechow et al (1995) and Gaver et al (1995). This is calculated as:

[NDA.sub.it] = [a.sub.i] + [b.sub.1i]([DELTA][REV.sub.it] - [DELTA][REC.sub.it]) + [b.sub.2i][PPE.sub.it] (5a)

The estimates of [a.sub.i], [b.sub.1], [b.sub.2] are generated from the following model:

[TAC.sub.it] = [a.sub.i] + [b.sub.1i]([DELTA][REV.sub.it] - [DELTA][REC.sub.it]) + [b.sub.2i][PPE.sub.it] + [[epsilon].sub.it] (5b)

[NDE.sub.it] = [NDA.sub.it] + [COP.sub.it] (5c)

Where:

[NDA.sub.it] = Nondiscretionary accruals;

[TAC.sub.it] = total accruals in year t for firm i, and it is calculated as:

[TAC.sub.it] = [DELTA][CA.sub.t] - [DELTA][Cash.sub.t] - [DELTA][CL.sub.t] + [DELTA][CM.sub.t] + [DELTA]Income Taxes [Payable.sub.t] - Depreciation and Amortization [Expense.sub.t]

[NDE.sub.it] = Nondiscretionary earnings;

[COP.sub.it] = cash flow from operations;

[DELTA][REV.sub.it] = revenues in year t less revenues in year t - 1 for firm i;

[DELTA][REC.sub.it] = receivables in year t less receivables in year t - 1 for firm i;

[DELTA] is the change and computed as the difference between time t and t - 1.

[PPE.sub.it] = gross property, plant, and equipment at the end of year t for firm i;

CA = Current Assets

CL = Current Liabilities

CM = current maturities of long term debt.

[[epsilon].sub.it] = error term for firm i;

The above baseline model (and by extension other models, excluding 5) are termed as "backward-looking" design by Larcker (2003) and so he suggests that future research could explore the potentials of "forward-looking" models. Taking up the challenge, and using almost all the variables, we use an alternative model choice to the HRS baseline model. The advantages of such "forward-looking" model include the opportunity to efficiently maximize the sample size.

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

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

Similarly for the nondiscretionary earnings, we have:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (7a)

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (7b)

(Definitions of variables are the same as described above.)

Measure of Risk

The probability of bankruptcy will be used to capture the responsiveness of the firm's cost of discounting the earnings to the use of stock options to compensating top executives. Johnson (2003) argues that the use of stock options may encourage managers to pursue suboptimal goals that maximize firms' earnings in the short term at the expense of long term viability of the firm. The crest of the argument is that, since stock options provide the executives an upside potential without exposing them to a commensurate risk of the downside, managers may take huge risks (Sanders and Hambrick, 2007). On the other hand, if the claim of agency theorists that the use of stock options ameliorates agency problems by aligning the incentives of managers to those of the shareholders holds, then firms whose executives are compensated more with stock options should have lower probability of bankruptcy. As a result, the true relationship between the use of stock options and the probability of bankruptcy becomes an empirical question. We put forward the following hypothesis in the affirmative while acknowledging the possibility of no or negative effect for the aforementioned reasons.

[H.sub.3]: Ceteris paribus, the higher the use of Executive stock options, the lower the probability of bankruptcy

We use the following equation to empirically test the effect of stock options hypothesized in H3.

[PROBNKP.sub.it] = [[micro].sub.0] + [[micro].sub.1][BSO.sub.it] + [[micro].sub.2][ERNVOL.sub.it] + [[micro].sub.3][SIZE.sub.it] + [[micro].sub.4][GROWTH.sub.it] + [[micro].sub.5][LEV.sub.it] + [[OMEGA].sub.it] (8)

Where:

[PROBNKP.sub.it] = probability of bankruptcy of firm i at time t. This is measured using the Altman (1968) Z score.

[ERNVOL.sub.it] = earnings volatility of firm i at time t. This is measured as the standard deviation of the firm's earnings per share over the sample period.

[SIZE.sub.it] = size of firm i at time t. This is measured as total assets at [sub.t-1]

[LEV.sub.it] = leverage of firm i at time t. This is measured as the prior year long term debt to total equity capital of the firm.

[GROWTH.sub.it] = captures the market to book value over the prior 5 years.

[[OMEGA].sub.it] = error term.

Industries dummies will also be used to capture and control for the cross sectional industry effects.

The inclusion of [BSO.sub.it] in equation (8) is only an attempt to establish empirical relationship, not causation, between the use of executive stock options and the failure of the firm. There are too many reasons and potential causes of corporate failures/bankruptcy that will prevent us from claiming causality in this regard.

There is consistent evidence in the literature that the degree of firms' earnings volatility is an increasing function of the firms cost of capital (see for example, Patell, 1976; Goel & Thakor, 2003; Lacina, 2004; DeFond & Hung, 2003). ERNVOL is added to capture the effect of earnings volatility. Earnings volatility is a decreasing function of the quality of earnings in that the more volatile a firm's earnings are, the noisier the investors' assessments of such earnings with the potential consequence of diminishing the earnings' perceived quality. As a result, before informed investment decisions could be made, additional search costs are implicitly imposed on investors as they will require additional sources of information to allow for desirable interpretations and then make informed judgments of such firm's volatile earnings. Goel and Thakor (2003) suggest that "an increase in the volatility of reported earnings will magnify these shareholders' trading losses." No doubt, such additional costs will be impounded in the required rate of returns for investment in such firms with the attendant increase in the firm's cost of capital. Alternative explanation for the possible increase in the cost of capital as a result of a firm's earnings volatility could be that since firms with high volatile earnings will need to provide other types of disclosures and information to market participants so as to mitigate the possible negative market reactions, such contingent additional information are not costless. (4)

LEV is expected to capture the operational uncertainty caused by cost of debt. Ahmed et al (2002) empirically document that operational uncertainty is one of the sources of "bondholder-shareholder conflicts over dividend costs" and that mitigating such conflicts could translate into the reduction in the firm's debt costs, and thus consequently increasing the value of the firm, all else equal. Titman and Wessels (1988) as cited by Dittmar (2004) provide evidence that the firm's cost of debt increases the probability of a firm's susceptibility to bankruptcy or financial distress (See Ngo, 2002; Mao, 2003).

GROWTH captures the relationship between probability of bankruptcy and book-to-market values of firms. The extant literature shows that firms with high probability of bankruptcy Z-score on average have low book-to-market values (see Hahn et al, 2010; and Zaretzky & Zumwalt, 2007 for a review of this literature).

The proxy for the probability of bankruptcy (PROBNKP it), the Altman (1968) Z score, will be calculated for individual sample firms over the sample period as follows:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (9)

Generally, higher Z-score corresponds to lower probability of bankruptcy. If a company has a Z-Score above 3, it is considered to be healthy and, therefore, unlikely to enter bankruptcy. If the score is lower than 1.8, the firm is in danger of bankruptcy. But if the Score is between 1.8 and 3, it is in a grey area (Altman, 1968)

Sample Selection

This study covers all US firms with available data in the Execucomp database as well as the Compustat tapes. The Execucomp database contains the compensation data for the top five executives of individual firms in the S&P 1500 (comprising those in the S&P 500 index, S&P 400 mid cap index and the S&P 600 small cap index). This data coverage begins in 1992. We extract the necessary data regarding the Black-Scholes value of an option from this database. For the entire model, we start with an initial sample of 2,507 firms with 17,970 firm years.

After interpolating and intersecting data from the two databases, deleting missing observations and conducting other data screening exercises, we have for the 'backward-looking' research design 858 firms with 2,579 firm-years. The forward-looking design comprises three different model categories viz: n + 1, Sum n + 1 + 2 and Sum n + 1 + 2 + 3 (where n is the grant year). Therefore, the first has 1,666 firms spanning 8,384 firm-years; the second has 1,476 firms with 6,666 firm-years and the third has 1,283 firms covering 5,357 firm-years (5). We believe that the larger sample size and the longer sample period relative to HRS better maximize the generalizability of findings in this critically important area of compensation research in empirical accounting.

To avoid complications caused by differences in reporting rules, the sample firms are required to be incorporated in the US. This is consistent with Matsunaga (1995). Also, regulated firms such as utilities companies (SIC codes 4900-4999) and financial institutions (SIC codes 6000-6099) are excluded so as to control for the differential incentives and motivational situations faced by executives operating in those regulatory environments relative to their counterparts in the non-regulated industries.

ANALYSES AND RESULTS

In this section, we present the empirical results obtained in the study and discus the implications of the findings for extant and future research in the area. Commencing with the descriptive statistics for the sampled firms in the 'backward-looking model, panel A of table 1 shows that the average firm in the sample generates annual sales worth of 5.4 billion (median $1.7 billion) with an operating margin of approximately 15%. The average firm in the sample has assets worth $5 billion (median $1.6 billion) with asset turnover rate of approximately 0.90. This suggests that firms in this category are fairly large and profitable. The average value (BSO) of the executive stock options granted to the top five executives of the sample firms is $7.758 million (median $2.7 million). This is approximately 0.4% of operating revenues, which is very similar to that reported in HRS.

Results

The coefficients from the regression and implied sensitivity analyses undertaken for the respective models to estimate payoffs using Black-Scholes values of executive stock option

Tables 2, 3 and 4, show information for firms in the 'forward-looking' models. Similar conclusion about size and profitability of firms in the respective sample category could be reached with the above descriptive information. Panel B of these tables shows the correlation matrix of the individual variables of interest in the respective models and virtually all the correlations are significant at the conventional significance thresholds.

Regression Results

The coefficients from the regression and implied sensitivity analyses undertaken for the respective models to estimate payoffs using Black-Scholes values of executive stock option grants are presented in tables 5 to 14 for both backward-looking and forward-looking models. Discussion of the results vis-a-vis their implications are concurrently presented as well.

Recall that due to the nonlinearity of the executive stock options and the respective performance measures, a second order term was introduced. BSO/S is the first order term while its square is the second order-term. Consistent with the findings of HRS, the regression coefficients of this second-order term was significantly negative in all the model specifications. This significantly negative coefficient suggests concavity, meaning that executive stock option grants increase performance at a reducing rate. Arguably, the inclusion of the second-order term does appear to correct omitted variable bias and does not seem to have induced our results. This is because there was no single situation of sign-switching of any of the regression coefficients of the primary variable of interest (which is BSO/S), the first-order term, as a result of the inclusion of the second-order term, but instead, the measure of goodness of fit statistic (adjusted R-Square) is consistently improved across all models. Similarly, we include lag of dependent measures in the respective models so as to control for prior year performance. This is important because of the mean reverting nature of the performance measures. Recall that HRS do not control for this in their baseline regression model which is primarily 'backward-looking'. Therefore, as a result of the compelling econometric justification for the inclusion of the second order term, as well as lagged performance measures, which is consistent with theoretical reasoning, considerable amount of our discussions will centre on the nonlinear coefficients of both prior and current performance measures with occasional references to the linear results for comparison purposes, where necessary.

Backward-looking design

Tables 5 and 6 contain the regression coefficients of the lagged design in panel A. Linear specifications of the respective models are presented in columns 1 and 2, while their nonlinear counterparts are contained in columns 3 and 4. There are 858 firms with 2,579 usable firm year observations.

The coefficients of the primary variable of interest in the model which is the additive sum of BSO/S, and perhaps the [(BSO/S).sup.2], show positive and negative directions respectively in all the model specifications. Looking at the nonlinear without previous performance measures of column 3, panel A of tables 5 and 6, for reported performance, the additive sum of these variables are respectively 0.348 and -0.171; 0.317 and -0.187 for nondiscretionary earnings. Column 4 shows the nonlinear with previous performance measures results. It shows the additive sum coefficients for BSO/S and [(BSO/S).sup.2] as 0.408 and -0.115, respectively in the reported performance model and 0.213 and -0.042 in the nondiscretionary earnings model. All these coefficients are highly significant.

The positive direction of these coefficients with respect to the first order term (BSO/S) implies positive contributions of executive stock options grants to both of our performance measures (reported earnings and nondiscretionary earnings). In other words, regardless of which earnings performance measures (reported, or 'true' earnings), corporate use of executive stock options positively impacts corporate performance. These findings provide extended, stronger and corroborative support for the findings of HRS. If the coefficients on BSO/S were to have been negative, consistent with the agency theory literature, then there is evidence of rent extraction.

Notwithstanding the above assertion, it is important to note the impact of introducing previous performance measures on the results. Column 4 shows that introducing lagged dependent variable actually increases BSO contributions for reported earnings (from 0.348 to 0.408), but reduces same contribution with respect to nondiscretionary earnings (from 0.317 to 0.213). We interpret these findings to mean that the improvement in earnings attributable to the granting of stock options to executives is not as high as implied by reported earnings when one controls for earnings management and prior year's earnings performance.

Panel B and C provide corroborative evidence of the results presented in panel A of tables 5 and 6. These panels show economic sensitivity (following HRS) of various BSO distributions to the performance measures. This is computed as the change in each of the dependent measures scaled by change in BSO/S, showing the economic impact, i.e. the dollar value, on performance measures of changing the median BSO up or down to next quartile cutoff, which in this instance is first and third quartile respectively. Specifically, focusing on the reported operating income without prior performance measure, if one moves from the quartile 1 BSO/S cutoff value of 0.0005 to the median of 0.0012, the dependent measure, OPINC/S, would increase from 0.0002 to 0.0004 indicating an implied sensitivity of 0.35. Similarly, the equivalent sensitivity for moving from the median to the 3rd quartile cutoff is 0.35, note that without approximating to two decimal places, in absolute decimal terms, this value is less than 0.35. According to HRS, the small slide in the implied sensitivities due to a shift from the median to quartile 3 of BSO/S indicates that the second-order effect of BSO/S is "economically" inconsequential, but that failure to consider this second-order term "appears to create a significant omitted variable in the linear specification".

From the implied sensitivity calculations, our results show that there is positive economic contribution of executive stock option grants to firm performance measures. For example, without prior performance measures, a dollar grant of executive stock options to top 5 corporate executives increase future reported operating performance by $1.35 and future nondiscretionary earnings by $1.32. With lagged performance measures, future reported operating performance increases by S1.41 and nondiscretionary earnings by $1.21 Overall, while the BSO-performance relation is positive, there is still some evidence of earnings management. For example, while the reported income shows $1.41 increment in BSO contribution to future operating performance, if the concept of 'true' earnings is considered as in nondiscretionary earnings, the contribution is only $1.21 or a reduction of 14%. This reduction is economically significant given that the average value of stock options granted by our sample firms is $7.8m in the backward model and around $4.5m in the forward model.

The other variable of interest in the empirical analysis is the research and development expenditure. R&D is an investment expenditure that should impact the future performance of the firm. Without controlling for this type of investment capital expenditure, one might run the risk of excessively attributing BSO performance payoffs (which may involve overestimating or underestimating error), hence the importance of this variable in the empirical design. Controlling for prior performance makes a difference in the sign of the coefficients of this variable in the operating income model. This thus implies that while it might appear that there is a positive contribution of the R&D expenditure to future operating performance, once prior performance is controlled for, this might not be the case. The same variable in HRS is positive (but HRS do not control for prior performance) and our result in column 4 of the panel A of table 5 challenges this result. Column 4 of table 6 also portrays a similar result. However, with respect to the nondiscretionary earnings measure, there is a consistently positive contribution of R&D expenditure to this future performance measure. If nondiscretionary earnings measure is truly a measure of 'true' earnings, then we will submit that managers do make positive net present value investment commitments in research and development expenditure.

Forward-looking design

As Larcker (2003) appropriately noted, the 'backward-looking' design approach employed by HRS is susceptible to quite a few limitations and criticisms and so can be improved upon. Some of the criticisms according to Larcker include its restrictive sample size, restrictive sample period, and the real potential reduction in the model explanatory power (6). He therefore suggested a 'forward-looking' research design choices.

Responding to this challenge, we will re-investigate the research question by re-specifying the empirical models using the 'forward-looking' empirical design in the following sequence: n + 1 (i.e. Year + 1), Sum n + 1 + 2 (i.e. SumYear + 1 + 2) and Sum n + 1 + 2 + 3 (i.e. SumYear + 1 + 2 + 3); where n is the grant year.

Year +1 Empirical Model

With this model, we estimate the option-performance payoffs of granting executive stock options to top 5 corporate executives in year n and the contribution of such new grants to future performance in year n+1, after controlling for necessary variables like corporate capital expenditures in tangible assets and research and development expenditure, prior performance measures as well as total cash compensation to these target executives.

There are 1,666 firms with 8,384 usable number of firm year observations for this empirical model. The regression coefficients and the implied sensitivity analysis for this model, is contained in tables 7 and 8.

The primary variable of interests are BSO/S and [(BSO/S).sup.2]. These variables show highly significant positive and negative coefficients signs respectively. For the operating income dependent measure, the coefficients are 0.373 and -0.249 without prior performance; 0.229 and 0.172 with prior performance. Nondiscretionary earnings measure has 0.247 and -0.124, and -0.147 and -0.078 for model without prior performance and that with prior performance respectively. One of the important implications of these coefficients is that the second-order term returning negative coefficients consistently in each of the models attests to the concavity nature of the BSO-performance relation, meaning that while executive stock options grants to top 5 corporate executives increase future performance, such relation is at a decreasing rate. This also attests to the nonlinear nature of the BSO-performance relation.

Another note worthy of mention is the fact that the coefficients of BSO/S in each of the models are consistently reduced when prior performances are controlled for. For example, for reported earnings dependent measure, it reduces from 0.373 to 0.229 and from 0.247 to 0.147 for nondiscretionary earnings dependent measure. This speaks to the fact that without controlling for this important variable, apart from the serious omitted variable bias that such exclusion might introduce into the models, the payoff estimates attributable to the BSO/S variable will be wrongly overestimated (7).

Similarly, it is important to mention that the coefficients of BSO/S are highest in reported operating performance measure model (0.373 and 0.229) compared to those of nondiscretionary performance measure model (0.247 and 0.147). This consistent trend in significant coefficients reduction empirically supports our conjecture that performance contributions of executive stock options grants to top 5 corporate executives as indicated in the reported operating performance might be overestimated relative to concepts of 'true' earnings as reflected in nondiscretionary earnings measure. However, it is important to note that, notwithstanding the probable performance contributions overestimations, corporate grants of executive stock options positively impact future performance, whether it is accrual-earnings (susceptible to earnings management) or future performance measures that are substantially 'accrual-free'. The results of the implied sensitivity analysis contained in panels C and D of the respective tables corroborates the position above. This analysis shows that a dollar grant of executive stock options to top 5 corporate executive contributes $1.37 to future operating income without controlling for prior performance and $1.23 when prior performance is controlled for. Similarly, $1.25 and $1.15 are contributed to nondiscretionary earnings without and with prior performance respectively. These dollar contribution amounts support the discussions above concerning the need to control for prior performance on one hand, and earnings management potentials of managers to expansively maximize their option payoffs on the other hand. In all, consistent with HRS evidence, our findings make it difficult to reject the incentive alignment hypothesis of corporate executive stock option grants, as evidence supporting rent extraction hypothesis is largely absent in our findings.

Other variables in the various models display expected trend and significant coefficients characteristics. The TA/S variable produces -0.136 and -0.151 with respect to the reported operating income dependent measure without and with prior performance measures. Also, for the nondiscretionary earnings, the coefficients are -0.187 and -0.072 respectively for with or without controlling for prior performance. We believe that the negative significant coefficients of this variable is actually reflecting assets turnover characteristics and so, it might not be inappropriate to interpret the coefficients in absolute terms as these significant coefficients indicate that managers productively utilize their corporate tangible assets in generating future earnings.

The coefficients of the capital expenditure on research and development expenditure (R&D/S) also show patterns that appear similar to productive corporate performance. The highly significant coefficients are 0.252 and 0.072 for reported operating income dependent measure, and 0.314 and 0.061 for nondiscretionary earnings dependent measure without and with prior performance respectively.

In addition, the variable controlling for the total cash compensation components of top 5 corporate executive, (TCC/S) shows surprising coefficients signs, in the respective models. These coefficients respectively without and with prior performance are -0.179 and -0.066, and -0.504 and -0.257 for the reported operating earnings and nondiscretionary earnings dependent measures respectively. We believe that it is important to control for this variable so as to determine whether, after remunerating top 5 corporate executives with regular salaries and cash bonuses as well as other forms of cash compensation, executive stock options grants are still capable of impacting positively future performance. HRS do not control for this variable in their baseline model (8), but we consider this a potential source of omitted variable bias and so decide to control for it in our study, especially if one considers the analytical argument of Tian (2004) on the substitution effect of cash compensation for options. He argues that the value or the incentive effects of an option to executives reduces quickly as more cash pay is substituted for options.

Interestingly but surprisingly and somewhat puzzling, this variable (TCC/S) shows highly significant negative coefficients consistently across all the respective models. This suggests that remunerating top 5 executives with salary and other cash bonuses effectively de-motivates them and thus reduces future performance measures. While we might agree to a reasonable extent with the fact that top corporate executives cannot be effectively motivated by only cash compensation in the glowing era of executive stock options, we would have expected this variable to be insignificant or at best less significant. But the intriguing thing is that even recent studies in the compensation literature find (what we will call) same anomaly significant negative coefficients (see HRS). Matolcsy (2000) documents what he refers to as "counterintuitive findings", a significant negative relationship between CEO's cash compensation and corporate performance. A completely different interpretation that we can give in this instance is that if a firm uses increasing amount of cash to compensate its top executives, investable cash for worthy positive net present value investment opportunities declines and this could reduce future corporate performance. Future studies that aim at resolving this somewhat counterintuitive finding can be a wonderful contribution to the compensation literature.

The coefficients of the previous performance measures in the respective models exhibit expected pattern or directions, that is, positively related to future performance measures. Findings for the Sumyear +1 +2 and SumYear +1 +2 +3 empirical models are substantially similar with the Year + 1 model (See tables 9 through 12), thus allowing generalization regarding the three forward-looking models.

Overall, both the lagged model (i.e. 'backward-looking') design and the 'forward-looking' model design findings collectively and consistently provide strong evidence of incentive alignment hypothesis, meaning that it is in the interests of shareholders to remunerate top corporate executives with executive stock options as this corporate granting behavior strongly motivates executives towards improving future corporate performance, an action that will be in the interest of shareholders. The evidence becomes more compelling as the findings consistently hold if one considers not only reported operating performance measures, but the other measure of earnings believed to reflect the concept of 'true' performance. The latter performance measure is devoid of managers earnings management actions, motivations for which are stronger when there are opportunities to maximize compensation payoffs such as one can find in executive stock options.

Probability of bankruptcy as a proxy for cost of discounting earnings

As explained earlier, the value of the firm can be explained by corporate earnings and the cost of discounting the earnings. While the above analyses, results and discussions center substantially on the earnings components (numerator) of the concept of the value of the firm, we will be examining the twin of this (denominator) in this section, and this is the cost of discounting the earnings using a measure of the probability of bankruptcy as developed by Altman Z-Score as a proxy (9). We do not use bond rating as a measure of firms' financial soundness for three essential reasons. First, extant research reveals that usually, bonds attract serious analysts' attentions during their first issuance or at infrequent extraordinary or special events, and such attentions diminish substantially thereafter (Holthausen and Leftwich, 1986). Second, corroborating this position, Wilson and Fabozzi (1990) provide evidence of the discontinuous nature of bond ratings. The final reason is the fact that Howe (1997) notes that there is usually a delay between when the corporate conditions change and when the ratings of the underlying bonds is actually done. Hence, bond ratings may provide a distorting lag that can generate otherwise inappropriate empirical findings to our research question in this instance.

Another potential alternative to the use of accounting-based measures as ingredients in probability of bankruptcy prediction model is stock market information. However, the challenge would be how to extract relevant probability of bankruptcy information from stock prices (see Beaver, 1968; Ohlson, 1980 and Cheung, 1991). This challenge becomes compelling if one considers the fact that the stock market may be inefficiently positioned (as it is often the case) to incorporate in a timely fashion, all relevant and publicly available information into the security prices (see for example, Sloan, 1996).

The results for the empirical investigation relating to this measure is contained in table 13, where we have the descriptive statistics and correlation matrix coefficients, and table 14 where the regression coefficients are presented. These results are discussed in sequence below.

Descriptive statistics

Here, we present the descriptive statistics of the sample relating the use of executive stock options to remunerate top 5 corporate executives and the probability of corporate failure, as measured by the Altman's Z-Score. In this sample, we have 1,507 firms with firm year observations totaling 8,217. The firms that on the average granted approximately $4.3 million (median $1.7 million) in executive stock options to its top 5 executives, measured by the Black-Scholes option value as reported by the Execucomp data base,, are considered large, profitable and employ sizeable amount of long term debt components in their capital structures, as measured by the size of their assets, earnings per share composition and the leverage status respectively. Large number of firms in the sample also shows promising growth status as measured by the market-to-book value ratio.

On the average, the firms in the sample made approximately $4 billion (median $1.2 billion) in revenue with $3.668 billion (median $1.0 billion) in tangible assets, and carrying long term debt of a little above 30% (median 29%) of their invested capital. On the average, the firms in the sample have approximately 4.79 (median 3.47) Z-Score suggesting a relatively low probability of bankruptcy. According to the bankruptcy prediction model of Atman (1968), if the model returns a value less than 1.81, there is a high probability of bankruptcy and if a value greater than 3.0 is produced, then there is low probability of bankruptcy. The values between 1.81 and 3 are in grey areas. The firms in the sample on the average have lower probability of corporate failure.

Regression results of the probability of bankruptcy model

The results for the regression coefficients are presented in table 14. The variables contained in the model are BSO/S, SIZE, TCC/S, ERNVOL, GROWTH and LEV. As was done in the testing of the effect on earnings, we scaled these variables mainly to minimize heteroscedascticity effects on the models as well as allowing for cross-sectional pooling of sampled firms with varying scale levels. The adjusted R-Square of the empirical model is 0.213. The primary variable of interest in the model is the BSO/S and as shown in Table 14, its coefficient is highly significant. This coefficient and its positive sign suggest that a point increase in the use of executive stock options to remunerate top 5 corporate executives leads to 0.046 point increase in the Altman Z-Score statistic, thus implying lower probability of corporate failure. This result corroborates the earnings components results discussed above.

The variable that captures earnings volatility (ERNVOL) appears to support the above comments. This variable has a positive coefficient of 0.032. This coefficient is significant (t-value of 3.04) suggesting that companies with higher earnings volatility have lower probability of corporate failure as a point increase in the volatility measure increases the Z-score by 0.032. However, the relationship between the use of stock options and corporate earnings volatility is worth mentioning. Empirically, there is a positive relationship between the use of this form of compensation package and the measure of earnings volatility. This means that the more the options used to remunerate top 5 corporate executives, the more volatile are corporate earnings.

In other words, granting stock options encourages managers to increase corporate volatility as the value of the options increase, among others, in the volatility of underlying stock returns, implying that stock options presage future volatility. Similarly, larger firms (captured by SIZE) have lower volatile returns and that companies with high volatile earnings are less levered, as such companies may not be attractive debtor-customers to lenders. Also, note that the relationship between the volatility variable and the corporate growth status is positive, suggesting that high growth firms are more likely to experience high earnings volatility. Cui and Mak (2002) document that this category of firms faces substantial operating uncertainty and business risk and that these usually lead in the direction of "significant variation in their profit rate, making accounting figures less informative about managerial performance", all of which will likely translate into corporate volatility.

Notwithstanding Cui and Mark (2002) position, the data here produce empirical results consistent with the original rationale for granting options which is to encourage managers into aggressive but profitable risk-taking behavior. The quality of such risk taking activities of executives (as empirically shown in this paper) is reflected in the fact that the volatility of corporate earnings does not result into increased chances of corporate failure. In fact, it actually reduces it.

Overall, the message here is that granting stock options presages future volatility and thus can increase the potential of corporate failure, consequently leading to high probability of bankruptcy especially in high growth firms with considerable high earnings volatility. We must admit that this conclusion is based on the fact that financial indicators determine corporate chances of bankruptcy. However, research in strategic management and related literature suggests that financially sound and economically worthy corporations can file for bankruptcy for strategic reasons (see for example, Moulton and Thomas, 1993; Shrader and Hickman, 1993; Bell, 1994; Tavakolian, 1995; Daily, 1996; Foust, 2000; Bhattacharya et al, 2007). Rose-Green and Dawkins (2002) distinguish between "financial bankruptcies" and "strategic bankruptcies", claiming that firms in the former categories are more likely to exhibit unimpressive financial indices than firms in the latter group. They conjecture and find that the market reaction to corporate bankruptcy situation discriminates between these two bankruptcy motivations and appropriately penalizes those firms that are compelled into bankruptcy by financial reasons more than those who choose to be strategically 'bankrupt'. Therefore, on the strength of these findings, the rationale for bankruptcy is not a first-order concern for our study as the market appropriately sees through this and reacts accordingly.

Sensitivity Analysis

Robustness checks are conducted to subject the sensitivities of the empirical findings presented and discussed above to alternative scalar choice, intensity of the research and developments expenditure as well as varied sample period. Unreported results indicate that findings are substantially comparable with those of the main analysis.

In order to control for possible firm specific effects, i.e. firm-specific shocks that are constant over time, we run fixed effect regression using the STATA statistical software. The magnitudes of the coefficients closely approximate those presented earlier. For example, for the Year + 1 empirical model, the coefficients of the primary variables of interest i.e. BSO/S and [(BSO/S).sup.2] in the new regression are 0.231 and - 0.176 respectively for the reported earnings after controlling for prior performance. These were respectively 0.229 and -0.172 in the main regressions. In both instances, these coefficients are significant at 1% significance level, although the adjusted R-Squared is slightly higher in the fixed effects regression (0.552 as against 0.536).

Further, since almost half of the companies in the Compustat database have missing values for R&D, we assign zero to many firms in our sample for the R&D variable. As indicated earlier, this is consistent with the approach maintained in the prior literature. Notwithstanding, we subject our empirical findings to a sensitivity test with regard to R&D variable by considering the research and development-only-firms in order to rule out the possibility that this variable could have driven the empirical results. For the Year + 1 forward-looking model, firm year observations reduces from 8,384 to 4,256 and the number of firms in the sample drops to 874 from 1,666. The coefficients of BSO/S and [(BSO/S).sup.2] in the new regression are 0.329 (0.244) and - 0.254 (-0.143) respectively for the reported earnings (nondiscretionary earnings) after controlling for prior performance. The dollar contributions of the reported earnings (nondiscretionary earnings) are $1.33 ($1.24) albeit an increase over the full sample of $1.23 ($1.15) respectively. These findings suggest a consistent positive contribution pattern in the performance benefits of executive stock option grants.

In addition, in order to address the concerns of potential confounding effects of the relatively scanty 1992 executive compensation data in our sample since 1992 was the first year Execucomp Database emerges, we remove observations for that year resulting into a shortened sample size. For Year + 1 empirical model, this exercise results into a loss of 250 firm year observations of only seven firms, producing 8,134 instead of 8,384 firm year observations and 1,659 instead of 1,666 firms contained in the full sample. BSO/S and [(BSO/S).sup.2] have coefficients of 0.228 (0.147) and -0.172 (-0.078) respectively for reported earnings (nondiscretionary earnings). The dollar contribution is exactly the same amount with the main analysis, i.e. $1.23 ($1.15).

In order to investigate whether the empirical findings are sensitive to alternative scalar choices, we restate the model using current year value of total assets. We consider this analysis worthwhile more importantly because the coefficient of the variable TA/S is consistently negative in virtually all empirical models in the main analyses. Recall that we interpreted this to mean that the variable is actually exhibiting the asset turnover relations in the models, considering the fact that it is scaled with sales. Therefore, in order to further examine this, we scale this variable and other variables in the model by total asset and the coefficient sign of the variable TA/S becomes positive in all the models in addition to the variables of interests displaying consistent coefficients in signs and magnitude. For example, for a Year + 1 model, reported performance (nondiscretionary earnings) after controlling for prior performances produces BSO/S and [(BSO/S).sup.2] equal to 0.201 (0.150) and -0.094 (0.090) respectively.

It must be noted that the pattern of consistent results of the sensitivity analyses with the main analyses holds across the all the empirical models be it 'backward-looking-design or 'forward-looking-design'.

Overall, the theme or tenor of the findings remains substantially unaffected as a result of these sensitivity and additional analyses. Notwithstanding, it is important to mention that like any other research endeavor especially of empirical nature, certain caveats could weaken or impact the conclusions or inferences from the findings of this study. For example, the sample selection criteria may induce survivorship bias, even though such criteria appear reasonable and acceptable in the domain of empirical accounting research. Also, one cannot completely rule out the potential bias of correlated omitted variables as it will be extremely difficult, if not impossible to envisage and account for all relevant variables in a model. Bearing in mind that it is always tricky to appropriately foretell the direction, level and magnitude of any bias if it exists; noting these caveats is considered appropriate. In addition, we must mention that there is the real concern about the potential problem(s) of endogeneity, and that the tenor of our empirical results may change if appropriate instrumental variables are found in this setting. This is another promising area for future research efforts in this area of compensation research.

Similarly, the total generalizability of this study's findings cannot be guaranteed. This is because, we only consider a somewhat short time-series of new executive stock option grants, spanning only 10 years (i.e. 1992-2001), and performance measures of only 12 years (i.e. 1993-2004) (10). This thus speaks to the generalizability of the empirical findings reported in this study beyond this time frame. Also it should be recalled that this study uses executive stock options value measured by the Black-Scholes option pricing model. This model is not immune from criticisms among academics, compensation consultants and practitioners alike, as they have consistently pointed to its shortcomings. Therefore, the findings of this study can only be as good as this option pricing model. Finally, there could be measurement error in the variables of choice and this could limit the interpretations of the findings of this study.

Notwithstanding the potential limitations highlighted above, the theme of this study and its findings contribute to the compensation literature and empirical accounting studies in significant dimensions. For example, the findings of this study provide some of the first evidence and probing insights into the option-performance relation within the dynamics of corporate earnings and the cost of discounting such earnings. In this study, we exclude financial firms and other firms in regulated industries. It could be a fruitful future research effort to examine the option-payoffs relations in these industries. The starting point for such studies would be to take care of or control for the peculiarities of these industries vis-a-vis the unique agency relationship and earnings management incentives that subsist in them. In addition, given the relatively short sample period of this study, subsequent studies could evaluate the robustness and thus, the generalizability of this study's findings to longer time periods and by extension, larger cross-section of sample firms.

CONCLUSION

Granting stock options is a strategic corporate activity aimed at achieving certain corporate objectives, theoretically in the overall shareholders' ultimate interests. Executive stock options compensation has continued to remain an increasingly substantial component of management compensation packages.

Not many studies have provided direct evidence of the impact of executive stock options on the primary components of firm value which include earnings and cost of discounting the earnings. A notable exception is the study of Hanlon, Rajgopal and Shevlin, 2003 (HRS) which examined the executive stock options vis-a-vis future earnings of the firm. However, our findings extend HRS findings by showing in part that nondiscretionary measure could be a more appropriate guide to compensation committees and corporate boards when making executive compensation decisions. In fact, our findings could have potential public policy implications and ramifications giving the contemporariness of executive compensations in the debates surrounding current global economic turmoil. Generally, studies on employees/executive stock options appear to assume that the value of the firm is impacted by the use of this compensation package and thus build the focus of their investigations on this premise (see for example Akindayomi, 2010 for a review of relevant literature). While such an assumption could be well placed, it is yet sufficiently unclear which component of the firm value is individually or jointly impacted by the use of stock options to compensate executives. Therefore, this study is motivated by the need to fill this important gap (and generally the taken-for-granted view) in the literature, with the intent to examining the impact of granting options to top corporate executives on the firms' earnings, cost of capital and by extension the value of the firm.

The concept of accounting earnings and the cost of discounting such earnings are central to the value of the firm. Theoretically, therefore, the effect of using stock options to compensate executives should be reflected in those two major components of the firm value i.e. the earnings component and the cost of capital or discount rate associated with the earnings. Thus, central to this study is the firms' cost of discounting earnings, as well as the various measures of earnings. The volatility of the firm's earnings and the probability of bankruptcy are used to capture the responsiveness of the firm's cost of capital to the use of stock options to compensating top executives, while the measures of earnings employed are the reported operating earnings and 'nondiscretionary' earnings. Overall, both the lagged model (i.e. 'backward-looking') design and the 'forward-looking' model design findings collectively and consistently provide strong evidence of incentive alignment hypothesis, meaning that it is in the interests of shareholders to remunerate top corporate executives with executive stock options as this corporate granting behavior strongly motivates executives towards improving future corporate performance, an action that will be in the interest of shareholders. The evidence becomes more compelling as the findings consistently hold if one considers not only reported operating performance measures, but the other measure of the earnings believed to reflect the concept of 'true' performance as such a performance measure is devoid of managers earnings management actions, motivations for which are stronger when there are opportunities to maximize compensation payoffs like one can find in executive stock options. In other words, we could not find support for the competing rent extraction hypothesis, as executive stock option grants improve future corporate performance as measured by the earnings measures.

Corroboratively, the empirical findings in relation to the proxy of cost of discounting earnings as measured by the Altman Z-Score statistic of bankruptcy probability also reinforce the earnings components findings, even as volatility increases in executive stock option grants.

ACKNOWLEDGEMENTS

The authors acknowledge helpful comments by Amin Mawani and Steven Balsam on the earlier version of this paper, as well as feedback by colleagues in University of Calgary and University of Texas - Pan American, the participants of the AAA Southeastern 2008 meeting in Birmingham, Alabama, and the participants of Canadian Academic Accounting Association (CAAA) 2009 meeting in Montreal. This paper was nominated for best paper award in the AAA Southeastern 2008 meeting in Birmingham, Alabama. It then received the "Distinguished Research Award" from the Academy of Accounting and Financial Studies. Akindayomi A, also likes to thank members of his dissertation committee as this piece is a part-product of his doctoral dissertation at University of Calgary. The authors will also like to acknowledge the research grant by the CGA Alberta Faculty Fellowship.

REFERENCES

Aboody, D., & R. Kasznik (2000). CEO stock option awards and the timing of corporate voluntary disclosures. Journal of Accounting and Economics, 29(1), 73-100.

Ahmed, A. S., B. K. Billings, R. M. Morton & M. S. Harris (2002). The role of accounting conservatism in mitigating bondholder-shareholder conflicts over dividend policy and in reducing debt costs. The Accounting Review, 77(4), 867-890.

Akindayomi, A. (2010). Executive stock options and firm valuation: A literature review and research framework. International Journal of Business Excellence, 3(4), 470-493.

Altman, E. I. (1968). Financial ratios, discriminant analysis, and the prediction of corporate bankruptcy. Journal of Finance, 23(4), 589-609.

Antle, R., & A. Smith (1986). An empirical investigation of the relative performance evaluation of corporate executives. Journal of Accounting Research, 24(1), 1-39.

Austin, J. R., J. J. Gaver, & K. M. Gaver (1998). The choice of incentive stock options vs. nonqualified options: A marginal tax rate perspective. Journal of the American Taxation Association, 20(2), 1-21.

Baker, T., D. Collins & A. Reitenga (2003). Stock option compensation and earnings management incentives. Journal of Accounting, Auditing and Finance, 18(4), 557.

Balsam, S., & D. H. Ryan (1996). Response to tax law changes involving the deductibility of executive compensation: A model explaining corporate behavior. Journal of the American Taxation Association, 18(Supplement), 1-11.

Balsam, S., R. Halperin & H. A. Mozes (1997). Tax costs and nontax benefits: The case of incentive stock options. Journal of the American Taxation Association, 19(2), 19-37.

Beaver, W. (1968). Market prices, financial ratios, and the prediction of failure. Journal of Accounting Research, 6(2), 179 -192.

Bell, T. (1994). Discussion of auditor litigation and modified reporting on bankrupt clients. Journal of Accounting Research, 32(Supplement), 1-31.

Bernard, V. L. (1995). The Feltham-Ohlson framework: implications for empiricists. Journal of Accounting Research, 11(2), 733-747.

Bhattacharya, N., M. Dawkins & L. Bamber (2007). Systematic share price fluctuations after bankruptcy fillings and the investors who drive them. Journal of Financial and Quantitative Analysis, 42(2), 399-420.

Clement, M., R. Frankel, & J. Miller (2003). Confirming management earnings forecasts, earnings uncertainty, and stock returns. Journal of Accounting Research, 41(4), 653-679.

Chen, K. C., & K. C. Wei (1993). Creditors' decisions to waive violations of accounting-based debt covenants. The Accounting Review, 68(2), 218-232.

Cheung, J. (1991). A review of option-pricing theory in accounting research. Journal of Accounting Literature, 10, 51 -84

Cohen, D.A., & P. Zarowin (2010). Accrual-based and real earnings management activities around seasoned equity offerings. Journal of Accounting and Economics, 50(1), 2-19.

Core, J., R. Holthausen & D. Larcker (1999). Corporate governance, chief executive officer compensation, and firm performance. Journal of Financial Economics, 51(3), 371-406.

Cui, H. & Y. T. Mak (2002). The relationship between managerial ownership and firm performance in high R&D firms. Journal of Corporate Finance, 8(4), 313-336.

Daily, C. (1996). Governance patterns in bankruptcy reorganizations. Strategic Management Journal, 17(5): 355-375.

Dechow, P.M., R. Sloan & A.P. Sweeney (1995). Detecting earnings management. The Accounting Review, 70(2), 193-225

DeFond, M. L. & M. Hung (2003). An empirical analysis of analysts' cash flow forecasts. Journal of Accounting and Economics, 35(1), 73-100.

Dittmar, A. (2004). Capital structure and corporate spin-offs. Journal of Business, 77(1), 9-54.

Eisenhardt, K. (1989). Agency theory: An assessment and review. Academy of Management Review, 14(1), 57-74.

Feltham, G.A. & J.A. Ohlson (1995). Valuation and clean surplus accounting for operating and financial activities. Contemporary Accounting Research, 11(2), 689-731.

D. Foust (2000, March). Chapter 11 never looked so good, Business Week, 44.

Gaver, J K., J. Gaver & J.R. Austin (1995). Additional evidence on bonus plans and income management. Journal of Accounting and Economics, 19(1), 3-28.

Goel, A.M. & A. V. Thakor (2003). Why do firms smooth earnings? The Journal of Business, 76(1), 151-192.

Hahn, T., M. O'Neil & J. Swisher (2010). What does book-to-market proxy: Risk or investor sentiment. Academy of Accounting and Financial Studies Journal, 14(3), 85-100.

Hanlon, M., S. Rajgopal & T. Shevlin (2003). Are executive stock options associated with future earnings? Journal of Accounting and Economics, 36(1-3), 3-43.

Holthausen, R. & R. Leftwich (1986). The effect of bond rating changes on common stock prices. Journal of Financial Economics, 7(1), 57-89.

Howe, J. (1997). Credit analysis for corporate bonds: The handbook of fixed income securities (pp. 371-404). Chicago, IL: Irwin Professional Publishing.

Jensen, M. C. & W. H. Meckling (1976). Theory of the firm: Managerial behavior, agency costs and ownership structure. Journal of Financial Economics, 3(4), 305-360.

Jensen, M. C., K. J. Murphy & E. G. Wruck (2004). Remuneration: Where we've been, how we got to here, what are the problems, and how to fix them. Harvard NOM Working Paper No. 04-28; ECGI - Finance Working Paper No. 44/2004. Available at SSRN: http://ssrn.com/abstract=561305 or DOI: 10.2139/ssrn.561305.

Johnson, C. H. (2003). Stock and stock option compensation: A bad idea. Canadian Tax Journal, 51(3), 1259-90.

Jones, J. (1991). Earnings management during import relief investigations. Journal of Accounting Research, 29(2), 193-228.

Kang, S. & K. Sivaramakrishnan (1995). Issues in testing earnings management and an instrumental variable approach. Journal of Accounting Research, 33(2), 353-367.

Kato, H. .K., M., Lemmon, M., Luo & J. Schallheim (2005). An empirical examination of the cost and benefits of executive stock options: Evidence from Japan. Journal of Financial Economics, 78(2), 435-461

Klassen, K. J. & A. Mawani (2000). The impact of financial and tax reporting incentives on option grants to Canadian CEOs. Contemporary Accounting Research, 17(2), 227-62.

Lacina, M. J. (2004). Characteristics of firms that voluntarily issue a revenue forecasts with no accompanying earnings forecast. American Business Review, 22(1), 124-131.

Larcker, D. F. (2003). Discussion of: Are executive stock options associated with future earnings? Journal of Accounting and Economics, 36(1-3), 91-103.

Mao, C.X. (2003) Interaction of debt agency problems and optimal capital structure: Theory and evidence. Journal of Financial and Quantitative Analysis, 38(2), 399-437.

Matolcsy, Z. P. (2000). Executive cash compensation and corporate performance during different economic cycles. Contemporary Accounting Research, 17(4), 671-692.

Matsunaga, S. R. (1995). The effects of financial reporting costs on the use of employee stock options. The Accounting Review, 70(1), 1-26.

Mawani, A. (2003a). Tax deductibility of employee stock options. Canadian Tax Journal, 51(3), 1230-58.

Mawani, A. (2003b). Cancellation of executive stock options: Tax and accounting income considerations. Contemporary Accounting Research, 20(3), 495-517.

Moulton, W. N. & H. Thomas (1993). Bankruptcy as a deliberate strategy: Theoretical consideration and empirical evidence. Strategic Management Journal 14(2), 125-135.

Ngo, M. V. (2002). Agency costs and the demand and supply of secured debt and asset securitization. Yale Journal on Regulation, 19(2), 413-471.

Ohlson, J. (1980). Financial ratios and the probabilistic prediction of bankruptcy. Journal of Accounting Research, 19(1), 109-131

Ohlson, J. A. (1995). Earnings, book values, and dividends in equity valuation. Journal of Accounting Research, 11(2), 661-687.

Patell, J. (1976). Corporate forecasts of earnings per share and stock price behavior: Empirical tests. Journal of Accounting Research, 246-76.

Rajgopal, S. & T. Shevlin (2002). Empirical evidence on the relation between stock option compensation and risk taking. Journal of Accounting and Economics, 33(2), 145-171.

Reitenga, A., S. Buchheit & T. Baker (2002). CEO bonus pay, tax policy, and earnings management. Journal of the American Taxation Association, 24(2), 1-23.

Rose-Green, E. & M. Dawkins (2002). Strategic bankruptcies and price reactions to bankruptcy filings. Journal of Business Finance and Accounting, 29 (9&10), 1319-1335.

Sanders, W. G. & D. C. Hambrick (2007). Swinging for the fences: The effects of CEO stock options on company risk taking and performance. Academy of Management Journal, 50(5), 1055-1078.

Sloan, R. G. (1996). "Do stock prices fully reflect information in accruals and cash flows about future earnings? The Accounting Review, 71(3): 289-315.

Shrader, M. & K. Hickman (1993). Economic issues in bankruptcy and reorganization. Journal of Applied Business Research, 9(3), 110-118.

Tavakolian, H. (1995). Bankruptcy: An emerging corporate strategy. S.A.M. Advanced Management Journal, 60(2): 18-22.

Tian, Y. S. (2004). Too much of a good incentive? The case of executive stock options. Journal of Banking and Finance, 28(6), 1225-1245.

Titman, S. & R. Wessels. (1988). The determinants of capital structure choice. Journal of Finance, 43 (1), 1-19.

Wilson, R. & F. Fabozzi (1990). The New corporate bond market. Chicago, IL:

Zaretzky, K. & J. K. Zumwalt (2007). Relation between distress risk, book-to-market ratio and return premium. Managerial Finance, 3(10), 788-97.

Akinloye Akindayomi, University of Texas-Pan American

Hussein A. Warsame, University of Calgary

ENDNOTES

(1.) Our assertion is informed by the fact that many firms that participate in TARP have been under intense scrutiny of the regulators and the congress such that the congress insists that the firms must pay back their TARP obligations before paying out the usual big cash compensation to executives. For more on TARP, see the Emergency Economic Stabilization Act of 2008, and Public Law 110-343.

(2.) Value of the firm can be demonstrated using the framework of firm valuation model as developed by the Feltham-Ohlson, 1995 (hereinafter referred to as FO).

[P.sub.t] = [bv.sub.t] + [[infinity].summation over ([tau]=1)][R.sup.-[tau].sub.f][E.sub.t][[x.sup.a.sub.t+[tau]]]

Where:

[P.sub.t] = market value of the firm's equity, at time t

[bv.sub.t] = book value of the firm's equity at time t

[R.sub.f] = the firm's cost of capital or the discounting rate of the earnings. FO suggests that [R.sub.f] be calculated as one plus the risk-free interest rate.

[x.sup.a] = the abnormal earnings; Et = the expectation operator

(3.) Clement et al (2003) used a variation of the firm valuation model viz: [P.sub.t] = k x [[infinity].summation over (t=1)][E.sub.t]/[(1+r).sup.[tau]]. However, one of the implicit inferences in FO framework is that in order to determine the value of the firm, one does not necessarily have to forecast future dividends, a view Bernard (1995) applauds and describes as taking accounting researcher's away from the "traditional mainstream view"; notwithstanding, some researchers still use it as a starting point in evaluating the effect of the primary components of the firm values viz earnings and cost of capital which are still relevant even in the FO framework. But in order to reflect the distinctive relevance of accounting numbers to the value of the firm, this study will align with the conceptual inferences of the FO valuation model.

(4.) For example, DeFond and Hung (2003) identify cash flow forecasts as one of the information sources that have to be released to the market by firms with high volatile earnings so that "market participants could identify the persistent components in earnings."

(5.) The discrepancies in the number of firms and firm-years between and within the backward and forward looking models are mainly due to the stronger data requirement constraints imposed by their underlying characteristics, as the final sample in each of these categories contains only firms and firm-year observations with required compensation and financial data. Also note that we use firm-years and not firm-quarters or other potentially usable periods because the Execucomp which is the source of our stock options data is available on annual basis.

(6.) We must admit that the lagged design results presented above are effectively challenged by Larcker's observations on the research design choice. We therefore, re-examine the research question using the 'forward-looking' design below.

(7.) For more on this, see our discussions surrounding this relation in the section on the lagged results above.

(8.) Instead, they do so as part of their sensitivity analysis, while they mention that their results remain qualitatively similar, we strongly believe, that the performance contribution attributed to executive stock options in their baseline model might be somehow overstated.

(9.) For example, Chen and Wei (1993) documented that firms with less likelihood of becoming bankrupt (i.e. with lower probability of bankruptcy) are more likely to enjoy waiver opportunity from creditors. This suggests that the cost of debt and by extension the cost of operations of such firms is likely to be lower relative to firms with high probability of bankruptcy.

(10.) HRS considered an eight-year and a three-year of time-series of option grants and payoffs relations respectively.

TABLE 1: {BACKWARD LOOKING DESIGN}
DESCRIPTIVE STATISTICS AND CORRELATION MATRIX

Panel A: Descriptive Statistics

 N = 2,579: F = 858

Variables Mean Std. Median Q1 Q3
 deviation

OPINC ($billion) 0.845 2.028 0.239 0.091 0.717
NDE ($billion) 0.322 1.009 0.082 0.024 0.259
SALES ($billion) 5.395 11.151 1.737 0.730 4.977
BSO grants (Smillion) 7.758 18.819 2.684 0.865 7.512
ASSETS ($billion) 5.050 12.382 1.564 0.654 4.611
OPINC/S 0.149 0.206 0.140 0.087 0.206
NDE/S 0.070 0.190 0.060 0.020 0.110
TA/S 1.083 0.794 0.887 0.621 1.281
BSO/S 0.004 0.009 0.001 0.0005 0.003
R&D/S 0.043 0.181 0.004 0.000 0.037

Panel B: Correlation Matrix

Variables OPINC/S NDE/S TA/S BSO/S R&D/S

OPINC/S 1
NDE/S 0.435 1
TA/S 0.343 0.290 1
BSO/S 0.303 0.514 0.382 1
R&D/S 0.202 0.536 0.522 0.491 1

Note on Panel A:

The 'backward-looking' design model is estimated using 2,579 firm-year
observations for a total of858 firms with no missing data. The firm y
span through 1998 to 2001. OPINC is annual operating income; NDE is
nondiscretionary earnings; Sales is annual sales, BSO is Black-Sch
value of options grants to top 5 corporate executives as per Execucomp,
ASSETS is year-end balance sheet value of total assets (TA) and R&D is
research and development expenditure. Missing values of R&D are set to
zero.

Note on Panel B:

Variables are as described above scaled by sales. All correlations are
significant at conventional thresholds except otherwise indicated
superscript NS

TABLE 2: {FORWARD LOOKING DESIGN} {YEAR + 1}
DESCRIPTIVE STATISTICS AND CORRELATION MATRIX

Panel A: Descriptive Statistics

 N = 8,384: F = 1,666

Variables Mean Std. Median Q1 Q3
 deviation

OPINC (Sbillion) 0.611 1.650 0.160 0.058 0.467
NDE (Sbillion) 0.236 0.872 0.034 -0.007 0.167
SALES (Sbillion) 4.089 10.057 1.216 0.494 3.497
BSO grants (Smillion) 4.428 11.171 1.673 0.645 4.263
ASSETS (Sbillion) 3.805 10.983 0.991 0.384 2.952
OPINC/S 0.150 0.148 0.140 0.080 0.020
NDE/S 0.020 0.145 0.030 -0.010 0.070
TA/S 1.010 0.921 0.820 0.590 1.180
BSO/S 0.003 0.004 0.001 0.0004 0.004
R&D/S 0.030 0.071 0.001 0.000 0.033

Panel B: Correlation Matrix

Variables OPINC/S NDE/S TA/S BSO/S TCC/S

OPINC/S 1
NDE/S 0.670 1
TA/S 0.117 -0.120 1
BSO/S 0.201 0.065 0.190 1
TCC/S 0.020NS -0.320 0.301 0.434 1
R&D/S 0.256 0.196 0.279 0.360 0.375

Panel A: Descriptive Statistics

N = 8,384: F = 1,666

Variables

OPINC (Sbillion)
NDE (Sbillion)
SALES (Sbillion)
BSO grants (Smillion)
ASSETS (Sbillion)
OPINC/S
NDE/S
TA/S
BSO/S
R&D/S

Panel B: Correlation Matrix

Variables R&D/S

OPINC/S
NDE/S
TA/S
BSO/S
TCC/S
R&D/S 1

Note on Panel A:

The 'forward-looking' design model {Year + 1} is estimated using 8,384
firm-year observations for a total of 1,666 firms with no missing data
Firm years span through 1992 to 2001. OPINC is annual operating income
following the year of grant; NDE is nondiscretionary earnings following
the year of grant; following the year ofgrant, Sales is annual sales,
BSO is Black-Scholes value of options grants to top 5 corporate
executives as per Execucomp, ASSETS is year-end balance sheet value of
total assets (TA), TCC is cash compensation for top 5 corporate
executives as per Execucomp, and R&D is research and development
expenditure. Missing values of R&D are set to zero.

Note on Panel B:

Variables are as described above scaled by sales. All correlations are
significant at conventional thresholds except otherwise indicated as a
superscript NS.

TABLE 3: {FORWARD LOOKING DESIGN}
{SUMYEAR + 1 + 2}
DESCRIPTIVE STATISTICS AND CORRELATION MATRIX

Panel A: Descriptive Statistics

 N = 6,666: F = 1,476

Variables Mean Std. Median Q1 Q3
 deviation

OPINC1 ($billion) 1.335 3.554 0.357 0.135 1.050
NDE1 ($billion) 0.516 1.754 0.078 -0.013 0.367
SALES ($billion) 9.034 22.517 2.707 1.089 7.720
BSO grants ($million) 4.687 10.677 1.811 0.703 4.564
ASSETS ($billion) 3.984 11.302 1.020 0.401 3.165
OPINC1/S 0.150 0.121 0.140 0.090 0.020
NDE1/S 0.020 0.112 0.030 -0.010 0.070
TA/S 0.480 0.393 0.390 0.280 0.560
BSO/S 0.002 0.004 0.001 0.0002 0.002
R&D/S 0.009 0.014 0.002 0.000 0.014

Panel B: Correlation Matrix

Variables OPINC1/S NDE1/S TA/S BSO/S TCC/S

OPINC1/S 1
NDE1/S 0.662 1
TA/S 0.225 -0.065 1
BSO/S 0.121 -0.059 0.154 1
TCC/S 0.060 -0.351 0.213 0.442 1
R&D/S 0.356 0.312 0.036 0.204 0.175

Panel A: Descriptive Statistics

N = 6,666: F = 1,476

Variables

OPINC1 ($billion)
NDE1 ($billion)
SALES ($billion)
BSO grants ($million)
ASSETS ($billion)
OPINC1/S
NDE1/S
TA/S
BSO/S
R&D/S

Panel B: Correlation Matrix

Variables R&D/S

OPINC1/S
NDE1/S
TA/S
BSO/S
TCC/S
R&D/S 1

Note on Panel A:

The 'forward-looking' design model {SumYear + 1 + 2} is estimated
using 6,666 firm-year observations for a total of 1,476 firms with no
missing data. Firm years span through 1992 to 2001. OPINC1 is sum of
operating income for two years following the grant year; NDE1 is sum
of nondiscretionary earnings for two years following the grant year;
Sales is annual sales, BSO is Black-Scholes value of options grants
to top 5 corporate executives as per Execucomp, ASSETS is year-end
balance sheet value of total assets (TA), TCC is cash compensation for
top 5 corporate executives as per Execucomp, and R&D is research and
development expenditure. Missing values of R&D are set to zero.

Note on Panel B:

Variables are as described above scaled by sales. All correlations are
significant at conventional thresholds except otherwise indicated as a
superscript NS.

TABLE 4: {FORWARD LOOKING DESIGN}
{SUMYEAR + 1 + 2 + 3}
DESCRIPTIVE STATISTICS AND CORRELATION MATRIX

Panel A: Descriptive Statistics

 N = 5,357: F = 1,283

Variables Mean Std. Median Q1
 deviation

OPINC2S ($billion) 2.061 5.402 0.546 0.197
NDE2 ($billion) 0.840 2.711 0.128 -0.021
SALES ($billion) 12.866 29.887 3.943 1.587
BSO grants ($million) 5.065 12.627 1.587 0.748
ASSETS ($billion) 3.660 8.358 1.015 0.396
OPINC22/S 0.150 0.099 0.140 0.100
NDE2/S 0.020 0.116 0.030 -0.010
TA/S 0.285 0.107 0.267 0.199
BSO/S 0.001 0.004 0.000 0.0002
R&D/S 0.010 0.013 0.004 0.000

Panel B: Correlation Matrix

Variables OPINC2/S NDE2/S TA/S BSO/S

OPINC2/S 1
NDE2/S 0.709 1
TA/S 0.275 0.074 1
BSO/S 0.152 -0.004 NS 0.140 1
TCC/S 0.078 -0.408 0.137 0.364
R&D/S 0.489 0.367 0.293 0.237

Panel A: Descriptive Statistics

 N = 5,357: F = 1,283

Variables Q3

OPINC2S ($billion) 1.587
NDE2 ($billion) 0.607
SALES ($billion) 11.265
BSO grants ($million) 4.727
ASSETS ($billion) 2.993
OPINC22/S 0.200
NDE2/S 0.070
TA/S 0.353
BSO/S 0.001
R&D/S 0.014

Panel B: Correlation Matrix

Variables TCC/S R&D/S

OPINC2/S
NDE2/S
TA/S
BSO/S
TCC/S 1
R&D/S 0.257 1

Note on Panel A:

The 'forward-looking' design model {SumYear + 1 + 2 + 3} is estimated
using 5,357firm-year observations for a total of 1,283 firms with no
missing data. Firm years span through 1992 to 2001. OPINC2 is sum of
operating income for three years following the grant year; NDE2 is sum
of nondiscretionary earnings for three years following the grant year;
Sales is annual sales, BSO is Black-Scholes value of options grants to
top 5 corporate executives as per Execucomp, ASSETS is year-end balance
sheet value of total assets (TA) and R&D is research and development
expenditure. Missing values of R&D are set to zero.

Note on Panel B:

Variables are as described above scaled by sales. All correlations are
significant at conventional thresholds except otherwise indicated as a
superscript NS.

TABLE 5: {BACKWARD LOOKING DESIGN}
ESTIMATION OF PAYOFFS USING BLACK-SCHOLES VALUES OF BSO GRANTS
{N= 2,579; F= 858}

Panel A: {Regression Coefficients}

 LINEAR

 1 2

Variable {Dependent: OPINC/S} Coefficient Coefficient

TA/S 5 0.07 *** -0.127 ***
[5.summation over (k=0)] 0.191 *** 0.218 ***
 [[alpha].sub.2,k]
 [(BSO/S).sub.i,t-k]
[5.summation over (k=0)]
 [[alpha].sub.3,k]
 [(BSO/S).sup.2.sub.i,t-k]
[5.summation over (k=0)] -0.091 *** -0.137 ***
 [[alpha].sub.4,k]
 [(R&D/S).sub.i,t-k]
[sigma][(OPINC/S).sub.it-1] 0.034 0.088 ***
[(OPINC).sub.t-1]/S 0.634 ***
Adj. [R.sup.2] without dummies 0.274 0.49
Adj. [R.sup.2] overall 0.448 0.574

{N= 2,579; F= 858}

Panel A: {Regression Coefficients}

 NONLINEAR

 3 4

Variable {Dependent: OPINC/S} Coefficient Coefficient

TA/S 5 0.094 *** -0.113 ***
[5.summation over (k=0)] 0.348 *** 0.408 ***
 [[alpha].sub.2,k]
 [(BSO/S).sub.i,t-k]
[5.summation over (k=0)] -0.171 *** -0.115 ***
 [[alpha].sub.3,k]
 [(BSO/S).sup.2.sub.i,t-k]
[5.summation over (k=0)] 0.067 *** -0.07 ***
 [[alpha].sub.4,k]
 [(R&D/S).sub.i,t-k]
[sigma][(OPINC/S).sub.it-1] -0.032 0.06 ***
[(OPINC).sub.t-1]/S 0.627 ***
Adj. [R.sup.2] without dummies 0.326 0.513
Adj. [R.sup.2] overall 0.475 0.59

Panel B: Economic effects sensitivity of various BSO distribution
{without previous performance}

 LINEAR

 BSO/S Effect on Implied
Distribution Cutoff OPINC/S Sensitivity

FIRST 0.0005 0.0001 0.19
MEDIAN 0.0012 0.0002 0.19
THIRD 0.0033 0.0006

 NONLINEAR

 BSO/S Effect on Implied
Distribution Cutoff OPINC/S Sensitivity

FIRST 0.0005 0.0002 0.35
MEDIAN 0.0012 0.0004 0.35
THIRD 0.0033 0.0012

Panel C: Economic effects sensitivity of various BSO distribution
{with previous performance}

 LINEAR

Distribution Cutoff BSO/S Effect on Implied
 OPINC/S Sensitivity

FIRST 0.0005 0.0001 0.22
MEDIAN 0.0012 0.0003 0.22
THIRD 0.0033 0.0007

 NONLINEAR

Distribution Cutoff BSO/S Effect on Implied
 OPINC/S Sensitivity

FIRST 0.0005 0.0002 0.41
MEDIAN 0.0012 0.0005 0.41
THIRD 0.0033 0.0014

Note on Panel A:

***, ** and * represent significance levels at 0.01, 0.05 and 0.10
respectively.

The 'backward-looking' design model is estimated using 2,579 firm-year
observations for a total of 858 firms with no missing data. The firm
years span through 1998 to 2001. OPINC is annual operating income;
Sales is annual sales, BSO is Black-Scholes value of options grants
to top 5 corporate executives as per Execucomp, ASSETS is year-end
balance sheet value of total assets (TA) and R&D is research and
development expenditure. Missing values of R&D are set to zero. All
variables are scaled by sales. Years are indexed by t and firms by i,
time and industry dummies are suppressed for expositional convenience.
Panel A contains regression coefficient estimates. Columns 1 and 3
contain coefficients without previous performance while columns 2 and
4 cover estimates with previous performance. Columns 1 to 2 and
columns 3 to 4 are for linear and nonlinear models respectively.

Note on Panel B and C:

Implied sensitivity analyses in panel B and C refer to the change
in OPINC/S scaled by change in BSO/S.

TABLE 6: {BACKWARD LOOKING DESIGN}
ESTIMATION OF PAYOFFS USING BLACK-SCHOLES VALUES OF BSO GRANTS
{N= 2,579; F = 858}

Panel A: {Regression Coefficients}

 LINEAR NONLINEAR

 1 2 3

Variable {Dependent: NDE/S} Coefficient Coefficient Coefficient

TA/S 5 -0.181 *** -0.084 *** -0.136 ***
[5.summation over (k=0)] 0.288 *** 0.128 *** 0.317 ***
 [[alpha].sub.2,k]
 [(BSO/S).sub.i,t-k]

[5.summation over (k=0)] -0.187 ***
 [[alpha].sub.3,k]
 [(BSO/S).sup.2.sub.i,t-k]

[5.summation over (k=0)] 0.171 *** -0.017 *** 0.363 ***
 [[alpha].sub.4,k]
 [(BSO/S).sub.i,t-k]

[sigma][(NDE/S).sub.it-1] 0.523 *** -0.281 *** 0.266 ***

[(NDE).sub.t-1]/S 1.091 ***

Adj. [R.sup.2] without 0.697 0.781 0.733
 dummies

Adj. [R.sup.2] overall 0.730 0.794 0.756

Panel A: {Regression Coefficients}

 NONLINEAR

 4

Variable {Dependent: NDE/S} Coefficient

TA/S 5 -0.072 ***
[5.summation over (k=0)] 0.213 ***
 [[alpha].sub.2,k]
 [(BSO/S).sub.i,t-k]

[5.summation over (k=0)] -0.042 ***
 [[alpha].sub.3,k]
 [(BSO/S).sup.2.sub.i,t-k]

[5.summation over (k=0)] 0.104 ***
 [[alpha].sub.4,k]
 [(BSO/S).sub.i,t-k]

[sigma][(NDE/S).sub.it-1] 0.325 ***

[(NDE).sub.t-1]/S 0.993 ***

Adj. [R.sup.2] without 0.794
 dummies

Adj. [R.sup.2] overall 0.804

Panel B: Economic effects sensitivity of various BSO distribution
{without previous performance}

 LINEAR

Distribution Cutoff BSO/S Effect on Implied
 NDE/S Sensitivity

FIRST 0.0005 0.0001 0.29
MEDIAN 0.0012 0.0004 0.29
THIRD 0.0033 0.0033

 NONLINEAR

Distribution Cutoff BSO/S Effect on Implied
 NDE/S Sensitivity

FIRST 0.0005 0.0002 0.32
MEDIAN 0.0012 0.0004 0.32
THIRD 0.0033 0.0010

Panel C: Economic effects sensitivity of various BSO distribution
{with previous performance}

 LINEAR

Distribution Cutoff BSO/S Effect on Implied
 NDE/S Sensitivity

FIRST 0.0005 0.0001 0.13
MEDIAN 0.0012 0.0002 0.13
THIRD 0.0033 0.0004

 NONLINEAR

Distribution Cutoff BSO/S Effect on Implied
 NDE/S Sensitivity

FIRST 0.0005 0.0001
MEDIAN 0.0012 0.0003
THIRD 0.0033 0.0007

Note on Panel A:

***, ** and * represent significance levels at 0.01, 0.05 and 0.10
respectively.

The 'backward-looking' design model is estimated using 2,579 firm-year
observations for a total of 858 firms with no missing data. The firm
years span through 1998 to 2001. NDE is nondiscretionary earnings;
Sales is annual sales, BSO is Black-Scholes value of options grants to
top 5 corporate executives as per Execucomp, ASSETS is year-end
balance sheet value of total assets (TA) and R&D is research and
development expenditure. Missing values of R&D are set to zero. All
variables are scaled by sales. Years are indexed by t and firms by i,
time and industry dummies are suppressed for expositional convenience.
Panel A contains regression coefficient estimates. Columns 1 and 3
contain coefficients without previous performance while columns 2 and
4 cover estimates with previous performance. Columns 1 to 2 and
columns 3 to 4 are for linear and nonlinear models
respectively.

Note on Panel B and C:

Implied sensitivity analyses in panel B and C refer to the change in
NDE/S scaled by change in BSO/S.

TABLE 7: {FORWARD LOOKING DESIGN}
{YEAR + 1}
ESTIMATION OF PAYOFFS USING BLACK-SCHOLES VALUES OF BSO GRANTS
{N= 8,384; F= 1,666}

Panel A: {Regression Coefficients without Previous Performance}

 1 2 3

Variable {Dependent: OPINC/S} Coefficients t-statistic p-value

TA/S -0.138 -12.28 .000
BSO/S 0.131 12.18 .000
[(BSO/S).sup.2]
RD/S 0.243 21.28 .000
TCC/S -0.170 -15.63 .000
Adj. [R.sup.2] without dummies 0.100
Adj. [R.sup.2] overall 0.316

Panel A: {Regression Coefficients without Previous Performance}

 4 5 6

Variable {Dependent: OPINC/S} Coefficients t-statistic value

TA/S -0.136 -12.21 .000
BSO/S 0.373 13.39 .000
[(BSO/S).sup.2] -0.249 -9.40 .000
RD/S 0.252 22.07 .000
TCC/S -0.179 -16.47 .000
Adj. [R.sup.2] without dummies 0.115
Adj. [R.sup.2] overall 0.323

Panel B: {with previous performance}

TA/S -0.152 -16.35 .000
BSO/S 0.062 6.90 .000
[(BSO/S).sup.2]
RD/S 0.065 6.57 .000
TCC/S -0.059 -6.40 .000
[(OPINC).sub.t-1]/S 0.567 62.08 .000
Adj. [R.sup.2] without dummies 0.478
Adj. [R.sup.2] overall 0.533

TA/S -0.151 -16.28 .000
BSO/S 0.229 9.90 .000
[(BSO/S).sup.2] -0.172 -7.83 .000
RD/S 0.072 7.29 .000
TCC/S -0.066 -7.15 .000
[(OPINC).sub.t-1]/S 0.563 61.76 .000
Adj. [R.sup.2] without dummies 0.483
Adj. [R.sup.2] overall 0.536

Panel C: Economic effects sensitivity of various BSO distribution
{without previous performance}

 LINEAR NONLINEAR

Distribution BSO/S Effect on Implied Effect on
Cutoff OPINC/S Sensitivity BSO/S OPINC/S

FIRST 0.0004 0.0001 0.13 0.0004 0.0002
MEDIAN 0.0012 0.0002 0.13 0.0012 0.0004
THIRD 0.0035 0.0005 0.0035 0.0013

 NONLINEAR

Distribution Implied
Cutoff Sensitivity

FIRST 0.37
MEDIAN 0.37
THIRD

Panel D: Economic effects sensitivity of various BSO distribution
{with previous performance}

 LINEAR NONLINEAR

Distribution BSO/S Effect on Implied BSO/S Effect on
Cutoff OPIN/S Sensitivity OPINC/S

FIRST 0.0004 0.0000 0.06 0.0004 0.0001
MEDIAN 0.0012 0.0001 0.06 0.0012 0.0003
THIRD 0.0035 0.0002 0.0035 0.0008

 NONLINEAR

Distribution Implied
Cutoff Sensitivity

FIRST 0.23
MEDIAN 0.23
THIRD

Notes on Panels A & B:

The 'forward-looking' design model {Year + 1} is estimated using 8,384
firm-year observations for a total of 1,666 firms with no missing data.
Firm years span through 1992 to 2001. OPINC is annual operating income
following the year of grant {the dependent measure}; Sales is annual
sales, BSO is Black-Scholes value of options grants to top 5 corporate
executives as per Execucomp, ASSETS is year-end balance sheet value of
total assets (TA), TCC is cash compensation for top 5 corporate
executives as per Execucomp and R&D is research and development
expenditure. Missing values of R&D are set to zero. All variables are
scaled by sales. Years are indexed by t and firms by i, time and
industry dummies are suppressed for expositional convenience. Panel A
is with respect to estimates without previous performance while Panel
B covers estimates with previous performance. Columns 1 to 3 and
columns 4 to 6 are for linear and nonlinear models respectively in
both panels.

Note on Panel C and D:

Implied sensitivity analyses in panel C and D refer to the change
in OPINC/S scaled by change in BSO/S.

TABLE 8: {FORWARD LOOKING DESIGN} {YEAR + 1}
ESTIMATION OF PAYOFFS USING BLACK-SCHOLES VALUES OF BSO GRANTS
{N= 8,384; F = 1,666}

Panel A: {Regression Coefficients without Previous Performance}

 1 2 3

Variable Coefficients t-statistic p-value
{Dependent: NDE/S}

TA/S -0.188 -16.81 .000
BSO/S 0.128 11.91 .000
[(BSO/S).sup.2]
RD/S 0.310 27.26 .000
TCC/S -0.500 -46.14 .000
Adj. [R.sup.2] without 0.250
 dummies
Adj. [R.sup.2] overall 0.324

 4 5 6

Variable Coefficients t-statistic p-value
{Dependent: NDE/S}

TA/S -0.187 -16.76 .000
BSO/S 0.247 8.90 .000
[(BSO/S).sup.2] -0.124 -4.67 .000
RD/S 0.314 27.57 .000
TCC/S -0.504 -46.43 .000
Adj. [R.sup.2] without 0.256
 dummies
Adj. [R.sup.2] overall 0.325

Panel B: {with previous performance}

 1 2 3

Variable Coefficients t-statistic p-value
{Dependent: NDE/S}

TA/S -0.072 -7.49 .000
BSO/S 0.072 7.86 .000
[(BSO/S).sup.2]
RD/S 0.058 5.44 .000
TCC/S -0.254 -25.07 .000
(NDE)t-1/S 0.543 57.27 .000
Adj. [R.sup.2] without 0.498
 dummies
Adj. [R.sup.2] overall 0.515

 4 5 6

Variable Coefficients t-statistic p-value
{Dependent: NDE/S}

TA/S -0.072 -7.47 .000
BSO/S 0.147 6.22 .000
[(BSO/S).sup.2] -0.078 -3.45 .001
RD/S 0.061 5.72 .000
TCC/S -0.257 -25.30 .000
(NDE)t-1/S 0.541 57.15 .000
Adj. [R.sup.2] without 0.500
 dummies
Adj. [R.sup.2] overall 0.516

Panel C: Economic effects sensitivity of various BSO distribution
{without previous performance}

 LINEAR

Distribution Cutoff BSO/S Effect on Implied
 NDE/S Sensitivity

FIRST 0.0004 0.0001 0.13
MEDIAN 0.0012 0.0002 0.13
THIRD 0.0035 0.0004

 NONLINEAR

Distribution Cutoff BSO/S Effect on Implied
 NDE/S Sensitivity

FIRST 0.0004 0.0001 0.25
MEDIAN 0.0012 0.0003 0.25
THIRD 0.0035 0.0009

Panel D: Economic effects sensitivity of various BSO distribution {with
previous performance}

 LINEAR

Distribution Cutoff BSO/S Effect on Implied
 NDE/S Sensitivity

FIRST 0.0004 0.0000 0.07
MEDIAN 0.0012 0.0001 0.07
THIRD 0.0035 0.0002

 NONLINEAR

Distribution Cutoff BSO/S Effect on Implied
 NDE/S Sensitivity

FIRST 0.0004 0.0001 0.15
MEDIAN 0.0012 0.0002 0.15
THIRD 0.0035 0.0005

Notes on Panels A & B:

The 'forward-looking' design model {Year + 1} is estimated using 8,384
firm-year observations for a total of 1,666 firms with no missing data.
Firm years span through 1992 to 2001. NDE is nondiscretionary earnings
following the year of grant {the dependent measure}; Sales is annual
sales, BSO is Black-Scholes value of options grants to top 5 corporate
executives as per Execucomp, ASSETS is year-end balance sheet value of
total assets (TA), TCC is cash compensation for top 5 corporate
executives as per Execucomp and R&D is research and development
expenditure. Missing values of R&D are set to zero. All variables are
scaled by sales. Years are indexed by t and firms by i, time and
industry dummies are suppressed for expositional convenience. Panel A
is with respect to estimates without previous performance while Panel
B covers estimates with previous performance. Columns 1 to 3 and
columns 4 to 6 are for linear and nonlinear models respectively in
both panels.

Note on Panel C and D:

Implied sensitivity analyses in panel C and D refer to the change in
NDE scaled by change in BSO/S.

TABLE 9
{FORWARD LOOKING DESIGN}
{SUMYEAR + 1 + 2}

ESTIMATION OF PAYOFFS USING BLACK-SCHOLES VALUES OF BSO GRANTS
{N= 6,666; F = 1,476}

Panel A: {Regression Coefficients without Previous Performance}

 1 2 3

Variable {Dependent: Coefficients t-statistic p-value
OPINC1/S}

TA/S 0.016 1.32 .187
BSO/S 0.045 4.00 .000
[(BSO/S).sup.2]
RD/S 0.347 26.42 .000
TCC/S -0.084 -7.32 .000
Adj. [R.sup.2] without 0.175
 dummies
Adj. [R.sup.2] overall 0.380

 4 5

Variable {Dependent: Coefficients t-statistic
OPINC1/S}

TA/S 0.013 1.12
BSO/S 0.210 9.51
[(BSO/S).sup.2] -0.174 -8.67
RD/S 0.339 25.86
TCC/S -0.104 -8.91
Adj. [R.sup.2] without 0.187
 dummies
Adj. [R.sup.2] overall 0.386

Panel B: {with previous performance}

 1 2 3

Variable {Dependent: Coefficients t-statistic p-value
OPINC1/S}

TA/S -0.193 -17.06 .000
BSO/S 0.049 4.97 .000
[(BSO/S).sup.2]
RD/S 0.204 17.18 .000
TCC/S -0.067 -6.72 .000
(OPINC)t-1/S 0.502 45.48 .000
Adj. [R.sup.2] without 0.427
 dummies
Adj. [R.sup.2] overall 0.528

 4 5

Variable {Dependent: Coefficients t-statistic
OPINC1/S}

TA/S -0.193 -17.17
BSO/S 0.180 9.32
[(BSO/S).sup.2] -0.138 -7.88
RD/S 0.199 16.76
TCC/S -0.083 -8.17
(OPINC)t-1/S 0.498 45.29
Adj. [R.sup.2] without 0.434
 dummies
Adj. [R.sup.2] overall 0.532

Panel C: Economic effects sensitivity of various BSO
distribution {without previous performance}

 LINEAR

Distribution BSO/S Effect on Implied
Cutoff OPINC1/S Sensitivity

FIRST 0.0002 0.0000 0.05
MEDIAN 0.0016 0.0000 0.05
THIRD 0.0015 0.0001

 NONLINEAR

Distribution BSO/S Effect on Implied
Cutoff OPINC1/S Sensitivity

FIRST 0.0002 0.0000 0.21
MEDIAN 0.0016 0.0001 0.21
THIRD 0.0015 0.0003

Panel D: Economic effects sensitivity of various BSO
distribution {with previous performance}

 LINEAR

Distribution BSO/S Effect on Implied
Cutoff OPINC1/S Sensitivity

FIRST 0.0002 0.0000 0.05
MEDIAN 0.0016 0.0000 0.05
THIRD 0.0015 0.0001

 NONLINEAR

Distribution BSO/S Effect on Implied
Cutoff OPINC1/S Sensitivity

FIRST 0.0002 0.0000 0.18
MEDIAN 0.0016 0.0001 0.18
THIRD 0.0015 0.0003

Notes on Panels A & B:

The 'forward-looking' design model {SumYear + 1 + 2} is
estimated using 6,666 firm-year observations for a total
of 1,476 firms with no missing data. Firm years span
through 1992 to 2001. OPINC1 is sum of operating income
for two years following the grant year {the dependent
measure}; OPINC is annual operating income, Sales is annual
sales, BSO is Black-Scholes value of options grants to top
5 corporate executives as per Execucomp, ASSETS is year-end
balance sheet value of total assets (TA), TCC is cash
compensation for top 5 corporate executives as per Execucomp
and R&D is research and development expenditure. Missing
values of R&D are set to zero. All variables are scaled by
sales. Years are indexed by t and firms by i, time and
industry dummies are suppressed for expositional convenience.
Panel A is with respect to estimates without previous
performance while Panel B covers estimates with previous
performance. Columns 1 to 3 and columns 4 to 6 are for linear
and nonlinear models respectively in both panels.

Note on Panel C and D:

Implied sensitivity analyses in panel C and D refer to the
change in OPINC1/S scaled by change in BSO/S.

TABLE 10: {FORWARD LOOKING DESIGN}
{SUMYEAR + 1 + 2}
ESTIMATION OF PAYOFFS USING BLACK-SCHOLES VALUES OF BSO GRANTS
{N= 6,666; F = 1,476}

Panel A: {Regression Coefficients without Previous Performance}

 1 2 3

Variable Coefficients t-statistic p-value
{Dependent: NDE1/S}

TA/S -0.065 -5.26 .000
BSO/S 0.039 3.32 .001
[(BSO/S).sup.2]
RD/S 0.400 29.29 .000
TCC/S -0.475 -39.93 .000
Adj. [R.sup.2] without 0.270
 dummies
Adj. [R.sup.2] overall 0.330

 4 5 6

Variable Coefficients t-statistic p-value
{Dependent: NDE1/S}

TA/S -0.067 -5.43 .000
BSO/S 0.167 7.26 .000
[(BSO/S).sup.2] -0.135 -6.46 .000
RD/S 0.394 28.82 .000
TCC/S -0.491 -40.55 .000
Adj. [R.sup.2] without 0.277
 dummies
Adj. [R.sup.2] overall 0.334

Panel B: {with previous performance} TA/S

 1 2 3

Variable Coefficients t-statistic p-value
{Dependent: NDE1/S}

TA/S -0.076 -7.38 .000
BSO/S 0.042 4.30 .000
[(BSO/S).sup.2]
RD/S 0.187 15.37 .000
TCC/S -0.307 -29.21 .000
(NDE)t-1/S 0.513 52.51 .000
Adj. [R.sup.2] without 0.512
 dummies
Adj. [R.sup.2] overall 0.528

 4 5 6

Variable Coefficients t-statistic p-value
{Dependent: NDE1/S}

TA/S -0.078 -7.57 .000
BSO/S 0.158 8.17 .000
[(BSO/S).sup.2] -0.122 -6.94 .000
RD/S 0.182 14.94 .000
TCC/S -0.321 -30.10 .000
(NDE)t-1/S 0.512 52.59 .000
Adj. [R.sup.2] without 0.516
 dummies
Adj. [R.sup.2] overall 0.531

Panel C: Economic effects sensitivity of various BSO distribution
{without previous performance}

 LINEAR

Distribution BSO/S Effect on Implied
Cutoff NDE1/S Sensitivity

FIRST 0.0002 0.0000 0.04
MEDIAN 0.0016 0.0000 0.04
THIRD 0.0015 0.0001

 NONLINEAR

Distribution BSO/S Effect on Implied
Cutoff NDE1/S Sensitivity

FIRST 0.0002 0.0000 0.17
MEDIAN 0.0016 0.0001 0.17
THIRD 0.0015 0.0003

Panel D: Economic effects sensitivity of various BSO distribution
{with previous performance}

 LINEAR

Distribution BSO/S Effect on Implied
Cutoff NDE1/S Sensitivity

FIRST 0.0002 0.0000 0.04
MEDIAN 0.0016 0.0000 0.04
THIRD 0.0015 0.0001

 NONLINEAR

Distribution BSO/S Effect on Implied
Cutoff NDE1/S Sensitivity

FIRST 0.0002 0.0000 0.16
MEDIAN 0.0016 0.0001 0.16
THIRD 0.0015 0.0002

Notes on Panels A & B:

The 'forward-looking' design model {SumYear + 1 + 2} is
estimated using 6,666 firm-year observations for a total
of 1,476 firms with no missing data. Firm years span
through 1992 to 2001. NDE1 is sum of nondiscretionary
earnings for two years following the grant year {the
dependent measure}; NDE is nondiscretionary earnings,
Sales is annual sales, BSO is Black-Scholes value of
options grants to top 5 corporate executives as per
Execucomp, ASSETS is year-end balance sheet value of
total assets (TA) and R&D is research and development
expenditure. Missing values of R&D are set to zero. All
variables are scaled by sales. Years are indexed by t and
firms by i, time and industry dummies are suppressed for
expositional convenience. Panel A is with respect to
estimates without previous performance while Panel B
covers estimates with previous performance. Columns 1 to
3 and columns 4 to 6 are for linear and nonlinear models
respectively in both panels.

Note on Panel C and D:

Implied sensitivity analyses in panel C and D refer to the
change in NDE1/S scaled by change in BSO/S.

TABLE 11
{FORWARD LOOKING DESIGN}
{SUMYEAR + 1 + 2 + 3}
ESTIMATION OF PAYOFFS USING BLACK-SCHOLES VALUES OF BSO GRANTS
{N= 5,357; F = 1,283}

Panel A: {Regression Coefficients without Previous Performance}

 1 2 3

Variable Coefficients t-statistic p-value
{Dependent: OPINC2/S}

TA/S 0.055 4.33 .000
BSO/S 0.055 4.68 .000
[(BSO/S).sup.2]
RD/S 0.405 28.15 .000
TCC/S -0.096 -7.93 .000
Adj. [R.sup.2] without 0.264
 dummies
Adj. [R.sup.2] overall 0.395

 4 5

Variable Coefficients t-statistic
{Dependent: OPINC2/S}

TA/S 0.055 4.33
BSO/S 0.148 7.35
[(BSO/S).sup.2] -0.105 -5.68
RD/S 0.395 27.28
TCC/S -0.109 -8.90
Adj. [R.sup.2] without 0.267
 dummies
Adj. [R.sup.2] overall 0.398

Panel B: {with previous performance}

 1 2 3

Variable Coefficients t-statistic p-value
{Dependent: OPINC2/S}

TA/S -0.118 -10.41 .000
BSO/S 0.073 7.32 .000
[(BSO/S).sup.2]
RD/S 0.193 14.77 .000
TCC/S -0.050 -4.87 .000
(OPINC)t-1/S 0.548 45.76 .000
Adj. [R.sup.2] without 0.502
 dummies
Adj. [R.sup.2] overall 0.567

 4 5

Variable Coefficients t-statistic
{Dependent: OPINC2/S}

TA/S -0.118 -10.36
BSO/S 0.130 7.61
[(BSO/S).sup.2] -0.064 -4.10
RD/S 0.187 14.31
TCC/S -0.059 -5.59
(OPINC)t-1/S 0.546 45.53
Adj. [R.sup.2] without 0.503
 dummies
Adj. [R.sup.2] overall 0.568

Panel C: Economic effects sensitivity of various BSO distribution
{without previous performance}

Distribution LINEAR
Cutoff

 BSO/S Effect on Implied
 OPINC2/S Sensitivity

FIRST 0.0002 0.0000 0.06
MEDIAN 0.0004 0.0000 0.06
THIRD 0.0011 0.0001

Distribution NONLINEAR
Cutoff

 Effect on Implied
 BSO/S OPINC2/S Sensitivity

FIRST 0.0002 0.0000 0.15
MEDIAN 0.0004 0.0001 0.15
THIRD 0.0011 0.0002

Panel D: Economic effects sensitivity of various BSO distribution
{with previous performance}

Distribution LINEAR
Cutoff

 BSO/S Effect on Implied
 OPINC2/S Sensitivity

FIRST 0.0002 0.0000 0.07
MEDIAN 0.0004 0.0000 0.07
THIRD 0.0011 0.0001

Distribution NONLINEAR
Cutoff

 Effect on Implied
 BSO/S OPINC2/S Sensitivity

FIRST 0.0002 0.0000 0.13
MEDIAN 0.0004 0.0001 0.13
THIRD 0.0011 0.0001

Notes on Panels A & B:

The 'forward-looking' design model {SumYear + 1 + 2 + 3} is estimated
using 5,357 firm-year observations for a total of 1,283 firms with no
missing data. Firm years span through 1992 to 2001. OPINC2 is sum of
operating income for three years following the grant year {the
dependent measure}; OPINC is annual operating income, Sales is annual
sales, BSO is Black-Scholes value of options grants to top 5 corporate
executives as per Execucomp, ASSETS is year-end balance sheet value of
total assets (TA), TCC is cash compensation for top 5 corporate
executives as per Execucomp and R&D is research and development
expenditure. Missing values of R&D are set to zero. All variables are
scaled by sales. Years are indexed by t and firms by i, time and
industry dummies are suppressed for expositional convenience. Panel
A is with respect to estimates without previous performance while
Panel B covers estimates with previous performance. Columns 1 to 3
and columns 4 to 6 are for linear and nonlinear models respectively
in both panels.

Note on Panel C and D:

Implied sensitivity analyses in panel C and D refer to the change in
OPINC2/S scaled by change in BSO/S.

TABLE 12
{FORWARD LOOKING DESIGN}
{SUMYEAR + 1 + 2 + 3}
ESTIMATION OF PAYOFFS USING BLACK-SCHOLES VALUES OF BSO GRANTS
{N= 5,357; F = 1,283}

Panel A: {Regression Coefficients without Previous Performance}

Variable 1 2 3
{Dependent: NDE2/S}

 Coefficients t-statistic p-value

TA/S -0.035 -2.93 .003
BSO/S 0.068 6.08 .000
[(BSO/S).sup.2]
RD/S 0.450 33.15 .000
TCC/S -0.589 -51.74 .000
Adj. [R.sup.2] without 0.411
 dummies
Adj. [R.sup.2] overall 0.462

Variable 4 5
{Dependent: NDE2/S}

 Coefficients t-statistic

TA/S -0.035 -2.96
BSO/S 0.167 8.81
[(BSO/S).sup.2] -0.112 -6.46
RD/S 0.438 32.18
TCC/S -0.603 -52.21
Adj. [R.sup.2] without 0.417
 dummies
Adj. [R.sup.2] overall 0.466

Panel B: {with previous performance} TA/S

Variable 1 2 3
{Dependent: NDE2/S}

 Coefficients t-statistic p-value

TA/S -0.066 -6.82 .000
BSO/S 0.083 9.13 .000
[(BSO/S).sup.2]
RD/S 0.166 13.54 .000
TCC/S -0.373 -36.98 .000
(NDE)t-1/S 0.547 52.52 .000
Adj. [R.sup.2] without 0.63
 dummies
Adj. [R.sup.2] overall 0.647

Variable 4 5
{Dependent: NDE2/S}

 Coefficients t-statistic

TA/S -0.066 -6.84
BSO/S 0.152 9.87
[(BSO/S).sup.2] -0.078 -5.56
RD/S 0.160 12.99
TCC/S -0.384 -37.45
(NDE)t-1/S 0.544 52.36
Adj. [R.sup.2] without 0.633
 dummies
Adj. [R.sup.2] overall 0.649

Panel C: Economic effects sensitivity of various BSO distribution
{without previous performance}

Distribution LINEAR
Cutoff
 Effect on Implied
 BSO/S NDE2/S Sensitivity

FIRST 0.0002 0.0000 0.07
MEDIAN 0.0004 0.0000 0.07
THIRD 0.0011 0.0001

Distribution NONLINEAR
Cutoff

 BSO/S Effect on Implied
 NDE2/S Sensitivity

 0.0002 0.0000 0.17
 0.0004 0.0001 0.17
FIRST 0.0011 0.0002
MEDIAN
THIRD

Panel D: Economic effects sensitivity of various BSO distribution
{with previous performance}

Distribution LINEAR
Cutoff
 Effect on Implied
 BSO/S NDE2/S Sensitivity

FIRST 0.0002 0.0000 0.08
MEDIAN 0.0004 0.0000 0.08
THIRD 0.0011 0.0001

Distribution NONLINEAR
Cutoff

 BSO/S Effect on Implied
 NDE2/S Sensitivity

 0.0002 0.0000 0.15
 0.0004 0.0001 0.15
FIRST 0.0011 0.0002
MEDIAN
THIRD

Notes on Panels A & B:

The 'forward-looking' design model {SumYear + 1 + 2 + 3} is estimated
using 5,357 firm-year observations for a total of 1,283 firms with
no missing data. Firm years span through 1992 to 2001. NDE2 is sum of
nondiscretionary earnings for three years following the grant year {the
dependent measure}; NDE is annual operating income, Sales is annual
sales, BSO is Black-Scholes value of options grants to top 5 corporate
executives as per Execucomp, ASSETS is year-end balance sheet value of
total assets (TA), TCC is cash compensation for top 5 corporate
executives as per Execucomp and R&D is research and development
expenditure. Missing values of R&D are set to zero. All variables are
scaled by sales. Years are indexed by t and firms by i, time and
industry dummies are suppressed for expositional convenience. Panel
A is with respect to estimates without previous performance while
Panel B covers estimates with previous performance. Columns 1 to 3
and columns 4 to 6 are for linear and nonlinear models respectively
in both panels.

Note on Panel C and D:

Implied sensitivity analyses in panel C and D refer to the change in
NDE2/S scaled by change in BSO/S.

TABLE 13
{PROBABILITY OF BANKRUPTCY DESIGN}
DESCRIPTIVE STATISTICS AND CORRELATION MATRIX
{N= 8,217; F= 1,507}

Panel A: Descriptive Statistics

Variables Mean Std. Median Q1 Q3
 deviation

SALES ($billion) 4.052 9.888 1.234 0.505 3.515
BSO grants 4.333 10.707 1.670 0.646 4.230
 ($million)
ASSETS ($billion) 3.668 9.151 1.001 0.393 2.960
PROBNKP 4.790 6.356 3.470 2.310 5.340
EPS (ERNVOL) 0.640 5.976 0.870 0.290 1.550
LEV 30.670 94.262 29.140 9.770 44.580
GROWTH 4.440 11.577 2.760 1.850 4.460
SIZE /S 1.010 0.912 0.820 0.590 1.117
BSO/S 0.003 0.004 0.001 0.0004 0.003

Panel B: Correlation Matrix

Variables ZSCORE/S TA/S BSO/S TCC/S EPS/S

PROBNKP/S 1
SIZE/S 0.002 (NS) 1
BSO/S 0.204 0.158 1
TCC/S 0.338 0.271 0.439 1
EPS/S 0.036 -0.048 0.014 (NS) -0.002 (NS) 1
GROWTH/S 0.305 0.066 0.173 0.363 -0.005
LEV/S -0.096 0.130 0.108 0.332 0.001

Variables MV/S DTC/S

PROBNKP/S
SIZE/S
BSO/S
TCC/S
EPS/S
GROWTH/S 1
LEV/S 0.070 1

Note on Panel A:

The probability of bankruptcy design is estimated using 8,217 firm-year
observations for a total of 1,507 firms with no missing data. Sales(S)
is annual sales, BSO is Black-Scholes value of options grants to top 5
corporate executives as per Execucomp, ASSETS (a measure of SIZE) is
year-end balance sheet value of total assets (TA), PROBNKP is the
Altman Z-score, EPS is earnings per share before extraordinary items
and discontinued operations, the standard deviation of which is used to
measure firm's volatility (ERNVOL), TCC is cash compensation for top 5
corporate executives as per Execucomp, LEV is long term debt to total
capital and GROWTH is Market to Book value ratio.

Note on Panel B:

Variables are as described above scaled by sales. All correlations are
significant at conventional thresholds except otherwise indicated as a
superscript NS.

TABLE 14
{PROBABILITY OF BANKRUPTCY DESIGN}
REGRESSION COEFFICIENTS ESTIMATES
{N= 8,217; F= 1,507}

Panel A: Prior 5 Year Growth Status

Variable Coefficients t-statistic p-value
{Dependent: PROBNKP}

SIZE/S -0.082 -7.02 .000
BSO/S 0.046 4.00 .000
TCC/S 0.327 25.55 .000
ERNVOL/S 0.032 3.04 .002
GROWTH/S 0.196 18.08 .000
LEV/S -0.216 -20.53 .000
Adj. [R.sup.2] without 0.206
 dummies
Adj. [R.sup.2] overall 0.213

Panel B: Current Year Growth Status SIZE/S

SIZE/S -0.081 -6.92 .000
BSO/S 0.041 3.60 .000
TCC/S 0.317 24.91 .000
ERNVOL/S 0.031 3.02 .003
GROWTH/S 0.225 20.81 .000
LEV/S -0.211 -20.19 .000
Adj. [R.sup.2] without 0.215
 dummies
Adj. [R.sup.2] overall 0.233

The probability of bankruptcy design is estimated using 8,217 firm-year
observations for a total of 1,507 firms with no missing data. Firm
years span through 1992 to 2001. Sales is annual sales, BSO is
Black-Scholes value of options grants to top 5 corporate executives as
per Execucomp, ASSETS (a measure of SIZE) is year-end balance sheet
value of total assets (TA), TCC is cash compensation for top 5
corporate executives as per Execucomp, PROBNKP is the Altman Z-score,
ERNVOL/S, measuring volatility is the standard deviation of earnings
per share before extraordinary items and discontinued operations,
LEV/S) is long term debt to total capital and GROWTH is Market to Book
value ratio. While the growth measure in Panel A is the average prior
5 year period, the corresponding measure in Panel B is the year t
measure. All variables are scaled by sales. Years are indexed by t
and firms by i, time and industry dummies are suppressed for
expositional convenience.