Voluntary Disclosures That Disavow Mandatory Disclosures The Case of Stock Options

For example, Intel includes the following disavowal statement in the stock option footnote from their fiscal 2001 annual report: The Black-Scholes option valuation model was developed fo

Preliminary Comments welcome Please do not quote without permission.

The authors thank David Aboody, Frank Acito, Dave Farber, Tom Linsmeier, Luann Lynch, SriniRangan, Steve Rock, Jerry Salamon, Paul Simko, and participants at accounting workshops at UCLA, University of Connecticut, University of Colorado, Michigan State University, and University of Virginia for comments We also thank Hyung Soon (Ross) Park and Christine Reinoehl for research assistance Walt Blacconiere gratefully acknowledges the financial support

of Ernst & Young, L.L.P and the Kelley School of Business at Indiana University

we predict that some firms disavow due to concerns about the reliability of the stock option compensation amount Second, we predict that some disavowals are motivated by executive compensation costs Based on a sample of over 1,300 firms during 2001, we find that over 18% disavow pro forma income adjusted for stock option compensation Our evidence provides only weak support for the prediction that disavowals are related to reliability concerns However, there is stronger support for our predictions regarding compensation concerns We find that sample firms are more likely to assert that pro forma income is unreliable in cases where the CEO’s stock option compensation is high relative to total to the CEO’s total compensation and incases where the CEO’s total compensation is “excessive” These latter results suggest that executive compensation costs influence disavowal decisions.

1 Introduction

The Financial Accounting Standards Board (FASB) issued SFAS 123, “Accounting for Stock Options,” in 1995 This accounting standard requires that firms calculate the expense for stock options granted to employees based on the fair value of the options on the grant date However, SFAS 123 gives firms a choice between income statement recognition versus footnote disclosure Specifically, a firm either can recognize stock option compensation expense as a determinant of reported net income in the income statement (i.e., income statement recognition)

or the firm can forego income statement recognition and instead disclose pro forma income

figures in the footnotes that assume stock option compensation was expensed (i.e., footnote disclosure) Although the exposure draft of SFAS 123 required income statement recognition and the FASB encourages income statement recognition, the vast majority of U.S firms

historically elected footnote disclosure.1

Among firms that use footnote disclosure, some firms include language that disavows thecalculation of the stock option compensation expense amount For example, Intel includes the following disavowal statement in the stock option footnote from their fiscal 2001 annual report:

The Black-Scholes option valuation model was developed for use in estimating the fair value of traded options that have no vesting restrictions and are fully transferable In addition, option valuation models require the input of highly subjective assumptions, including the expected stock price volatility Because the company's employee stock options have characteristics significantly different from those of traded options, and because changes in the subjective input assumptions can materially affect the fair value estimate, in the opinion of management, the existing models

do not necessarily provide a reliable single measure of the fair value of employee stock options.

1 According to The Wall Street Journal Online (April 22, 2003), only five publicly-held firms (Boeing

Co., Hawaiian Electric Industries, Winn-Dixie Stores Inc., Level 3 Communications Inc., and MacDermid Inc) treated stock option compensation as an expense in prior periods However, many additional firms started reporting stock option compensation expense in net income more recently For example, Weil and

Cummin [2004] state that “nearly 500 U.S.-listed companies have begun expensing stock option pay or

intend to start” (p C1) See Aboody, Barth, and Kasznik (2004) for an examination of the factors

associated with firms’ decisions to voluntarily recognize stock-based compensation expense under SFAS 123.

The research question addressed in the paper is: why do firms disavow the mandated accounting disclosure concerning stock option compensation expense? More specifically, this study investigates the decision to disavow stock option compensation expense disclosed in the footnotes We posit that this decision is made because the firm (or the firm’s manager) perceives that the benefits of disavowing outweigh the costs.2 We investigate two hypotheses regarding thedecision to disavow First, we predict that some disavowals are motivated by concerns about the reliability of the stock option compensation amount In effect, these firms are conveying

concerns regarding the validity of the option pricing model and/or assumptions used to compute the compensation amount Second, we expect that some disavowals are motivated by

compensation-related incentives If executive compensation appears to be excessive and/or the CEO receives a substantial portion of total compensation from stock option grants, we expect that the firm is more likely to disavow stock option compensation reported in the footnotes

Three characteristics of this particular setting motivate our investigation First, the effect

of stock option compensation expense on earnings appears to be significant Botosan and

Plumlee [2001] find that pro forma fully diluted earnings per share adjusted for stock option compensation is at least 22% lower than the amount based on reported net income for a sample

100 of “America’s fastest-growing” firms (as identified by Fortune magazine).3 More recently, Taub [2004] notes that earnings per share for the S&P 500 would be reduced by an estimated 7.4% if stock option compensation expense is included in reported net income

Second, stock option accounting has been and continues to be an important and

2 For example, benefits could include the perception that the firm is enhancing the quality of the financial statement disclosures by providing additional useful information to stakeholders (e.g., alerting

shareholders about legitimate reliability concerns in estimating stock option compensation expense) However, firms might be hesitant to include statements that explicitly contradict information in their audited financial statements that is required under GAAP Such disavowals could be interpreted as the firm having lower quality financial reporting.

3 As an extreme example, Yahoo! Inc reported net income of over $70 million for fiscal 2000 while disclosing stock option compensation adjusted pro forma loss in the footnotes of over $1.2 billion.

controversial issue The FASB proposed income statement recognition for stock option

compensation in 1993, but relaxed its position and ultimately allowed either income statement recognition or footnote disclosure in the face of significant lobbying and the threat of

Congressional action.4 On March 31, 2004, the FASB issued an Exposure Draft (titled Based Payment”) that calls for income statement recognition of estimated stock option

“Share-compensation expense Again, the FASB’s current proposals have generated significant

controversy.5

Third, the voluntary disavowal of a mandated accounting disclosure included in audited financial statements is relatively rare in practice Although almost all firms provide general caveats regarding estimates that are included in financial statements,6 a specific statement

questioning the usefulness of accounting information is much less common The disavowal statements examined in our study relate to a specific measure of income that must be reported in

an audited footnote under GAAP Thus, we believe that an investigation of the decision to disavow stock option compensation expense disclosed in the footnotes is warranted

We perform analysis on a sample of over 1,300 firms from the Execucomp database Of these firms, 248 (18.5%), representing a wide cross-section of publicly-traded firms, include a

4 For example, Botosan and Plumlee [2001] note that there were over 1,700 comment letters related to the exposure draft of SFAS 123 Revsine et al [2002] note that a Senate bill was introduced that would have eliminated the FASB’s independence by requiring the Securities and Exchange Commission to approve all new FASB standards.

5 For example, Senators Enzi and Reid have proposed a bill that would require expensing stock option compensation on a limited basis but would allow an exemption for small businesses and start-ups firms in

a response to the FASB proposal In a statement following a Senate hearing last week on the FASB proposal, Enzi said that the FASB “is ill equipped to conduct economic impact studies of the accounting standards it adopts … It was evident that FASB is not listening to small businesses and not taking their concerns seriously.” (Wells [2003]).

6 For example, Wal-Mart’s fiscal 2001 annual report includes the following wording in their footnote:

“The preparation of consolidated financial statements in conformity with generally accepted accounting principles requires management to make estimates and assumptions These estimates and assumptions affect the reported amounts of assets and liabilities They also affect the disclosure of contingent assets and liabilities at the date of the consolidated financial statements and the reported amounts of revenues and expenses during the reporting period Actual results may differ from these estimates.”

disavowal of stock option compensation (or the resulting pro forma income) in the footnotes of their fiscal 2001 financial statements Our evidence provides weak support for the prediction that disavowals are related to reliability concerns in measuring stock option compensation Although the standard deviation of stock price volatility is related positively to the decision to disavow, this relation is only significant in univariate tests However, our results suggest that concerns regarding executive compensation do influence the decision to disavow We find a significant relation between disavowals and both the ratio of the CEO’s stock option

compensation to total compensation and a measure of “excessive” CEO total compensation As expected, we find that firms disavow in cases where the stock option compensation expense materially affects return on assets Finally, our results suggest that smaller firms are more likely

to disavow

This study contributes to accounting research in at least two areas First, this study contributes to prior and ongoing research related to stock option accounting (examples include Matsunaga [1995], Aboody [1996], among many others) by investigating disavowals of stock option accounting disclosures This study most directly extends research that examine firms’ lobbying behavior prior to the issuance of SFAS 123 (Dechow, Hutton, and Sloan [1996] and Hill, Shelton, and Stevens [2002]), as well as studies that investigate the question of whether firms manage stock option compensation expense under SFAS 123 (Balsam, Mozes, and

Newman [2003] and Aboody, Barth, and Kasznik [2003])

Second, because disavowals are voluntary, we also potentially contribute to the extensive literature concerning voluntary disclosure (see Healy and Palepu [2001] and Core [2001] for recent reviews of this research) In particular, this study is related to recent research that has

focused on voluntary pro forma earnings disclosures (examples include Lougee and Marquardt

[2004], Frederickson and Miller [2004], among others) In these cases, the firm voluntarily provides two sets of financial information, one based on GAAP and one based on the firm's or industry's own definition The presence of the latter is an implicit disavowal of the mandated GAAP information In contrast, our study considers an explicit disavowal of a mandatory footnote disclosure

To the best of our knowledge, this is the first study to examine the determinants of a voluntary disclosure that explicitly disavows a mandated disclosure.7 Since disavowals

seemingly circumvent the intent of mandated disclosures, it is important to understand whether disavowals are motivated by concerns about the disclosed information or for other reasons (e.g., opportunism)

The remainder of the paper is organized as follows Section two discusses accounting forstock option compensation and develops the hypotheses investigated in the paper Section three presents the research design including discussion of the sample, model, and variables Section four presents results of our empirical analysis, and section five concludes

2 Background and Hypothesis Development

2.1 Accounting for Stock Option Compensation

Currently, the accounting for stock options is governed by Accounting Principles Board Opinion (APB) No 25 and Statement of Financial Accounting Standards (SFAS) No 123 APB

25 (issued in October 1972) defines stock option expense as the measurement date difference between the stock price and option exercise price, multiplied by the number of options Since the measurement date is defined as the date at which the exercise price and number of options are known, the measurement date of fixed option grants is the grant date Most employee stock

7 Frederickson, Hodge, and Pratt [2004] have a concurrent study that uses an experiment to investigate the effect of disavowals on investor judgment.

option grants have a fixed exercise price that equals the stock price at the grant date, implying that the associated expense is zero under APB 25

The FASB reconsidered the accounting for stock-based compensation and issued the exposure draft of SFAS 123 in 1993 The exposure draft proposed that stock option expense be based on option fair values at the grant date, estimated using an option-pricing model The FASB proposal was consistent with Black and Scholes [1973] who demonstrated that the value

of an option is not solely defined by the difference between the stock price and the exercise price.8 However, in response to political pressure, the FASB subsequently modified its proposal

to allow firms to substitute the measurement provisions in SFAS 123 with those in APB 25

SFAS 123 (issued in October 1995) defines stock option expense as the option’s fair value at the measurement date Fair value is to be calculated using an option pricing model whose inputs include the option’s exercise price, the current stock price, the expected life of the option, expected dividend yield, expected risk-free interest rate, and, for publicly traded firms, the expected stock price volatility.9 Once a fair value per option has been calculated, it is

multiplied by the number of options expected to vest and the resulting amount is amortized over the vesting period.10

If a firm elects, instead, to use APB 25 to measure stock option expense, it must provide footnote disclosure of a variety of items, including pro forma net income Virtually all firms

8 SFAS 123 does not dictate a specific option pricing model However, based on a search of the EDGAR database, Hodder, Maydew, McAnally, and Weaver [2004] find that only eight firms did not use the Black-Scholes model to estimate stock compensation expense for 2002

9 Use of the expected life of the option, as opposed to its stated term to maturity, reflects the fact that employees systematically exercise options early because employee stock options are nontransferable (Huddart and Lang [1996]).

10 Use of the number of options expected to vest, rather than the number granted, reflects the fact that not all granted options vest because some employees terminate employment before the end of the vesting period.

initially applied the measurement provisions of APB 25, hence disclose pro forma net income Other disclosures required under SFAS 123 include inputs to the option fair value calculation (i.e., option exercise price, expected option life, dividend yield, risk-free interest rate, and stock price volatility), estimated fair value of options granted, vesting period, and number of options granted, forfeited, and exercised.

2.2 Saliency and the Form of Executive Stock Option Disclosures

SFAS 123 generated significant political controversy The FASB position suggests that stock options are an expense that can be reliably measured and, therefore, should be recognized

in determining net income Its position is based on the argument that issuing stock options transfers claims on the firm’s equity from existing shareholders to employees Because the employees provide services to the firm, the value of the transferred ownership claims represents

a cost of generating revenue

In contrast, opponents argue that no expense should be recognized on the income

statement because the amount of the expense cannot be reliably measured Option pricing models are based on assumptions – such as constant volatility – that are often not descriptively valid Opponents also question the applicability of option price models to the valuation of nontransferable, forfeitable employee stock options held by risk averse, undiversified executives

More recently, significant stock price declines in high technology industries and a rise in the number of accounting restatements and related scandals have reopened the debate about the appropriate accounting treatment of stock options Partly in response to pressure from

shareholder activists, several bills have been introduced into Congress For example, Senators Levin and McCain had sponsored a bill that would disallow the tax deductibility of employee

stock options unless the options were also treated as an expense for financial reporting purposes Related, the International Accounting Standards Board issued International Financial Reporting Standard (IFRS) 2, a global standard that would require the income statement recognition of stock option compensation expense under the fair value method Likewise, the FASB has issued

an Exposure Draft that would require expensing of stock option compensation starting in 2005

Much of the debate about stock option accounting has implicitly focused on the argumentthat shareholders will not pay sufficient attention to unrecognized stock option compensation that

is disclosed in the footnotes to the financial statements (e.g., see Greenspan [1999], Ohl [2000] and Orr [2001], each of whom espouses this view.) Related, companies devote resources to lobbying about disclosure versus recognition even though the terms of stock option grants can beinferred from information in footnotes and proxy statements This behavior reflects a belief that investors fail to fully use information that is presented less saliently

Research in psychology supports this belief Psychologists have repeatedly examined how people form predictions in settings where the variable of interest is a stochastic function of multiple cues (e.g., Kruschke and Johansen [1999]) A consistent finding is that cue saliency affects judgments and decisions For example, individuals place greater weight on particular cues when those cues are made salient than when they are not Psychology research also has documented that due to limited information processing abilities, the way in which information is presented affects judgments and decisions (Payne, Bettman and Johnson [1993]).11 Consistent with the psychology literature, experimental studies in accounting have documented that the form in which information is presented influences investors’ trading and/or valuation judgments

11 Libby, Bloomfield, and Nelson [2001] and Maines [1995] provide detailed surveys of the experimental literature on financial information processing.

in a variety of contexts.12 There also is a growing body of archival accounting evidence

consistent with saliency effects in securities prices.13 Thus, the academic literature supports the argument that stakeholders may be influenced by the saliency of stock option disclosures

2.3 Hypothesis 1: Stock Option Disavowal and Reliability

The recognition of an item in the financial statements (or the disclosure of the item in the footnotes) introduces the issue of how the item should be measured (e.g., what is the appropriate way to measure stock option compensation) Concepts Statement 5, paragraph 3, states

“Measurement involves choice of an attribute by which to quantify a recognized item and choice

of a scale of measurement (often called “unit of measure”).” Most accountants have held views about which is the most relevant and reliable attribute to be measured and about the circumstances needed for recognizing changes in attributes and changes in the amounts of an attribute Some advocate the extant model – a mixture of historical costs and fair values, while others support fair value accounting

strongly-Consistent with the lack of consensus in the accounting profession, Concepts Statement 5does not advocate a particular measurement attribute Rather, it concludes (paragraph 90) that

“information based on current prices should be recognized if it is sufficiently relevant and

reliable to justify the costs involved and more relevant than alternative information.” Further,

12 Past research has examined: recognition versus disclosure of pension liabilities (Harper, Mister, and Strawser [1987]); classification of hybrid financial instruments (Hopkins [1996]); the framing of “bad news” earnings preannouncements (Libby and Tan [1999]); purchase versus pooling-of-interest

accounting treatment for business combinations (Hopkins, Houston, and Peters [2000]); placement of other comprehensive income items (Hirst and Hopkins [1998]); as well as market settings (Dietrich et al [2001]).

13 The effects of information saliency on price have been found in the following contexts: the

reannouncement of gains from debt-equity swaps (Hand [1990]); recognized versus disclosed write-down information in the oil and gas industry (Aboody [1996]); and time series variation in the valuation of post- retirement benefits disclosed in footnotes (Amir [1993]) Additionally, many accounting anomalies may

be viewed as reflecting the tendency of investors to neglect salient information (Sloan [1996], Teoh et al [1998a, 1998b], Bernard and Thomas [1989], Abarbanell and Bushee [1997], and Teoh and Wong [2002]).

Concepts Statement 2, paragraph 59 states, “The reliability of a measure rests on the faithfulness with which it represents what it purports to represent, coupled with an assurance for the user, which comes through verification, that it has that representational quality.” Thus, information is reliable when investors and other stakeholders can use it with confidence

The FASB’s position that firms should recognize employee stock options as an expense indetermining net income is based on two factors: (1) such stock options represent compensation expense and (2) firms can reliably measure the amount of this expense The FASB’s position for the former is based on the argument that issuing stock options to employees transfers claims on the firm’s equity from existing shareholders to employees while its position for the latter is the existence of option pricing models

Opponents of recognizing stock option compensation generally argue that option pricing models do not reliably measure the value of employee stock options for at least two reasons First, some assumptions underlying existing option pricing models are likely to be violated in practice (Black [1988]) For example, not only does the Black-Scholes model assume that the stock’s volatility is known and does not change over the option’s life, it also assumes that stock prices change smoothly (i.e., prices do not “jump up or down” significantly over short time periods) Further, Hodder, Maydew, McAnally, and Weaver’s [2004] results suggest that the subjectivity of the input assumptions for the Black-Scholes model impairs the reliability of stock option compensation expense calculated using this model.14

However, it is likely the Black-Scholes model would provide reasonable estimates of stock option compensation expense for some firms For example, the volatility of stock returns

14 Opponents also argue that extant option valuation models yield unreliable estimates for employee options because they overestimate the value of non-tradable options held by undiversified, risk-averse executives (e.g., Hall and Murphy [2002] show that undiversified executives assign a lower value to executive stock options compared to the Black-Scholes model).

could be relatively constant for a given firm Likewise, the results of Alford and Boatsman [1995] suggest that historical volatility is useful for predicting long-term volatility, so estimates

of expected volatility are likely to be accurate for many firms

This discussion suggests that whether an option pricing model provides reasonable estimates of stock option compensation depends on how well the firm matches the assumptions underlying the option pricing model The less descriptively valid the assumptions are for a particular firm, the less likely the option pricing model will reliably value that firm’s options Because it is likely that some firms match the assumptions underlying option pricing models reasonably well, concern about the reliability of stock option compensation is likely to be a significant issue for only a subset of all firms If managers’ disavowals of SFAS 123 disclosures are motivated by concerns about the reliability of financial statement disclosures, then:

H1: The decision to disavow is related positively to concerns about the reliability of pro

forma income adjusted for stock option compensation expense

2.4 Hypothesis 2: Stock Option Disavowal and Executive Compensation

Regulations that mandate the public disclosure of information about executive pay practices can impose significant costs on executives by increasing the transparency of executive compensation contracts The resulting costs to executives may take several forms First, lower information costs lead to a reduction in shareholders’ monitoring costs and, therefore, increased monitoring of executive pay practices Models of reputation formation such as Diamond [1989] demonstrate that if attempts by a manager to obtain excessive pay are sufficiently penalized, even “bad” managers will be induced to support pay policies that maximize shareholder wealth

as a result of their desire to mimic “good” managers However, Dyck and Zingales [2002] note that such models assume that a manager’s type is revealed with a probability of one In the

absence of public disclosure, such an assumption may not be descriptively valid When

information about pay is publicly disclosed, it is more apt to be widely disseminated and a higherproportion of shareholders are apt to be informed Thus, it is more likely that agency costs arising from sub-optimal pay practices will be constrained when the transparency of

compensation contracts increase

Second, informed stakeholder groups can significantly influence compensation policies

by exerting political pressure For example, Joskow, Rose and Wolfram [1996] provide evidencethat political pressure constrains CEO pay in the electric utility industry Dial and Murphy [1995] argue that General Dynamics responded to political concerns about pay practices by modifying the terms of its performance-based compensation plans Zenner and Perry [1997] andRose and Wolfram [1997] find that firms responded to the $1 million cap on the tax deductibility

of non-performance based pay by substituting performance-based compensation for fixed pay

Third, mandatory disclosure increases the saliency of pay practices to the general public and the coverage of executive pay by the press.15 Negative publicity concerning compensation policies can lead to a substantial loss in the executive’s reputation in the eyes of society at large.16Although nonpecuniary, these reputational costs are likely to matter to executives

Consistent with arguments that public disclosure of compensation information is costly tomanagers, prior research suggests that executives take actions to reduce the transparency of

15 The media plays a critical role in the dissemination of information about executive compensation (Murphy [1999], pp 49-52) The media prefers mandatory disclosure because information that is

mandatory is less apt to be provided by the company on a quid pro quo basis and is less apt to be

selectively disclosed.

16 An example of the reputational effects to executives that can arise from public criticism of various

aspects of a company’s corporate governance structure is a full-page ad that ran in the April 1992 Wall Street Journal Paid for by shareholder activist, Robert Monks, the ad featured a photo of the Sears board

of directors and a caption, “The non-performing assets of Sears.” Embarrassed by the advertisement, directors quickly adopted Monks’ governance reforms (including modifications to the structure of

executive compensation), despite the fact that Monks’ bid for election to the board had failed (Dyck and Zingales [2002]).

compensation by “managing” information related to stock options Murphy [1996] reports evidence consistent with executives managing proxy statement disclosures to lower the implied value of stock options Related, Yermack [1998], Baker [1999] and Aboody, Barth, and Kasznik [2003] find evidence that firms manage stock option compensation expense in an attempt to avoid political costs related to excessive compensation

Similarly, concerns regarding executive compensation appear to influence lobbying behavior related to the accounting for stock options Dechow et al [1996] presents evidence thatlobbying against the SFAS 123 Exposure Draft is related to the relative amount of stock option compensation paid to top executives Related, Hill et al [2002] find that lobbying against disclosure of stock option compensation information proposed by the FASB is related positively

to stock compensation paid to the top five executives

In summary, increased executive compensation disclosures are costly to executives Suchdisclosures increase the likelihood that information about executive compensation policies will

be disseminated to stakeholder groups who can negatively affect compensation contracts and/or impose significant reputational penalties.17 Thus, top executives who receive substantial

compensation from stock options have incentives to reduce the transparency of stock option grants Benefits to reduced transparency include – but are not limited to – increased voting support for stock option plans, reduced shareholder activism, a lower probability of regulatory intervention, and a reduction in various non-pecuniary costs Dechow et al [1996] note that while it is difficult to assess the exact magnitude of costs of this nature, it is safe to assume that

17 Hirshleifer, Lim, and Teoh [2003] develop a model in which misinterpretation of the valuation

implications of stock option disclosures arises as a result of limited shareholder attention Salient

information is given more weight than less salient information or than information absent from the public information set Expensing (not expensing) options at the time they are granted results in undervaluation (overvaluation) In contrast to the polar extremes of expensing (not expensing) options, SFAS 123 requires the amortization of the grant-date value over the vesting period Aboody, Barth, and Kasznik [2001] find no evidence that amortization over the vesting period results in a mismatching of expenses with expected future benefits.

even if such costs are small to the firm, they are large relative to the size of executive

compensation Thus, executives will attempt to reduce these costs If disavowal of stock option compensation expense disclosed in the footnotes is motivated by the cost to the CEO of public scrutiny of compensation policies, then we predict that:

H2: The decision to disavow is related positively to concerns regarding executive

3.2 Model and variables

We use logistic regression to investigate hypotheses concerning the decision to disavow The general model is as follows:

DISAVOWj=β0 + β1 Reliability proxyj + β2 Compensation proxyj + β3 ROADIFj

reliability of estimated stock option compensation (or the resulting pro forma income), (b) a statement related to the appropriateness of Black-Scholes option valuation model for non-traded options, and (c) a statement related to the subjectivity of the input assumptions used in the optionpricing model Table 2, panel A describes the distribution of disavowals Out of 1,342 firms in our sample, 248 (18.5%) disavowed stock option compensation in the footnotes of their fiscal

2001 financial statements by raising at least one of the three issues A total of 191 (14.2%) firms had disavowals that questioned the reliability of pro forma income adjusted for stock option compensation expense, 232 (17.3%) firms had disavowals related to the appropriateness of Black-Scholes option valuation model, and 236 (17.6%) firms had disavowals that raised the subjectivity issue There are 183 (13.6%) firms that raised all three issues in their disavowals

We used the three issues to construct three measures of the dependent variable,

DISAVOW The first is DISRELI, which equals 1 for firms that raised the reliability issue and 0 otherwise The second measure is DISBS, which equals 1 for firms that raised the issue of limitations of the Black-Scholes model and 0 otherwise The final measure is DISASS, which equals 1 for firms that raised the assumption subjectivity issue and 0 otherwise

3.2.2 Independent variables: Reliability concerns

Our first hypothesis predicts that the decision to disavow is related negatively to the descriptive validity of the assumptions underlying the Black-Scholes option pricing model One

of these assumptions is that a stock’s volatility is known and that volatility does not change over the life of the option (Black [1988]) We argue that there is more uncertainty about stock price volatility when historical volatility is itself volatile and when expected future volatility deviates from historical volatility.19 We use two measures to proxy for reliability concerns: SDSDRET

19 The Black-Scholes model also assumes that: stock price changes are smooth, short-term interest rates are constant, there is unlimited borrowing at a single rate, there are no restrictions on short-selling, that transactions costs are zero, investors have constant tax rates, and early exercise of the options (which

and VOLRATIO Under H1, we predict that β1 will be positive for these measures

We define SDSDRET as the standard deviation of the estimated volatility based on CRSPmonthly stock returns for the period 1997-2001.20 Specifically, we first estimate the standard deviation of monthly returns for each individual year (i.e., volatility for the year) Next, we compute the standard deviation of the volatility over the five years We expect that firms with a higher standard deviation in estimated volatility (i.e., higher “volatility in volatility”) will find it more difficult to determine accurate estimates of stock price volatility Thus, we predict that these firms will have greater concerns regarding the reliability of stock option compensation expense Panel B of table 2 reports that the mean of SDSDRET is 0.042 with a range of 0.002 to0.389

We define VOLRATIO as the ratio of the estimated volatility (i.e., standard deviation of CRSP monthly returns) for the 36 month period 2000-2002 to estimated volatility of monthly returns for the 36 month period 1997-1999 Greater volatility in the more recent period raises reliability concerns for two reasons First, non-constant historical volatility suggests non-

constant future volatility, which is inconsistent with the Black Scholes assumption of constant volatility over the option life Second, an increase in volatility makes it more difficult to

estimate future volatility According to the descriptive statistics in Table 2, panel B, more than half the sample faced increasing volatility in monthly stock price returns vis-à-vis prior years (e.g., median VOLRATIO is 1.13)

3.2.3 Independent variables: Compensation concerns

could potentially occur if the firm is, for example, acquired) is prohibited For the purposes of our analysis, we assume that there is no significant variation across our sample firms in the impact of these assumptions on the reliability of Black-Scholes values.

20 This is consistent with Alford and Boatsman [1995] who use monthly returns to predict long-term volatility since monthly returns are approximately normally distributed (consistent with the Black-Scholes model assumptions) while daily returns are not.

Our second hypothesis predicts that the decision to disavow is more likely when the cost

to the CEO of public scrutiny of compensation policies is high Firms that are perceived as paying excessive stock option compensation are more likely to attract the attention of

stakeholders (including the press) who can impose costs on these executives We use three measures to proxy for compensation concerns: SOC%, CEOSO% and XSCOMP We predict that β2 is positive under H2 for each measure

Murphy [1996] argues that managers incur significant nonpecuniary costs from high reported levels of compensation and, therefore, have an incentive to make financial reporting decisions that will reduce reported executive compensation expense Consistent with Murphy’s argument, we expect that the relative size of a CEO’s option compensation will be related

positively to the decision to disavow stock option compensation We use two proxies for the relative size of the option grant First, SOC% is the ratio of stock option compensation granted

to the CEO (Execucomp item BLK_VALU) to the total compensation paid to the CEO

(Execucomp item TDC1) When stock options comprise a larger proportion of the CEO’s total compensation package, we expect that disavowals are more likely to reduce nonpecuniary costs related to stock option compensation Stock options granted to CEOs represented an average of 43.5% total CEO compensation during fiscal 2001 (see table 2, panel B); however, 294 (out of 1,438 with some data) CEOs were granted no stock options during the year while four CEOs received 100% of their compensation in stock options

Second, CEOSO% is the ratio of stock option compensation granted to the CEO

(Execucomp item BLK_VALU) to the total stock option compensation expenses applicable to allemployees of the firm (Compustat item # 399 adjusted for taxes).21 If this ratio is high, it

21 Item # 399 is the difference between reported net income and pro forma income (net of income tax

suggests that the CEO is obtaining a greater proportion of the benefits of stock option grants à-vis employees at lower levels Table 2, panel B reports that the median CEOSO% was just over 10% for 2001.

vis-Our third proxy for compensation concerns is XSCOMP, a measure of excessive CEO total compensation Although we expect that the stock option compensation of the CEO would

be a primary consideration in the decision to disavow, all CEOs who are viewed to be “overpaid”are likely to be concerned about public scrutiny regarding compensation Similar to Murphy [1996] and Aboody, Barth, and Kasznik [2003], we determine XSCOMP by first estimating the following model of “expected” compensation using prior year (i.e., 2000) data for all firms on Execucomp:

ln(TOTAL COMP)i = b1ln(SALES)i + b2BM i + b3GROWTH i + b4ROA i + b5RETi

+ b61-YEAR VOL i + Σn=1to23γnIndustryin + ei (2)TOTAL COMP is total compensation paid to the CEO in thousands of dollars

(Execucomp item TDC1).22 SALES is reported sales in millions of dollars BM is the ratio of the book value of equity to the market value of equity GROWTH is sales growth compared to the prior year RET is the stock return (including reinvestment of dividends) for the year

(Execucomp item TRS1YR) 1-YEAR VOL is the standard deviation of monthly stock prices from CRSP Industry is an indicator variable equal to one (or zero) for each of the 23 industry groups defined in the Global Industry Classification Standard (GICS) classification system (Execucomp item SPINDEX).23

expense) We estimate stock option compensation expense by dividing this amount by 1 – tax rate (i.e., 35%).

22 Specifically, TDC1 is comprised of the following: Salary, Bonus, Other Annual, Total Value of

Restricted Stock Granted, Total Value of Stock Options Granted (using Black-Scholes), Long-Term Incentive Payouts, and All Other Total.

23 GICS was developed by Standard & Poor's in collaboration with Morgan Stanley Capital International

A company is assigned to a single GICS sub-industry based on its principal business Although revenue