Do the merits matter? Empirical evidence on shareholder suits from options backdating litigation.
INTRODUCTION I. DERIVATIVE LITIGATION AND OPTIONS BACKDATIN A. Do the Merits Matter in Stockholder Litigation? B. The Backdating Scandal II. DATA AND METHODOLOGY A. Methodology for Identifying Backdating Activity B. Description of Backdating Variables C. Data on Litigation over Backdating III. RESULTS A. Targeting B. Dismissal C. Settlement D. Special Litigation Committees IV. IMPLICATIONS FOR CORPORATE LAW A. Filing of Derivative Claims B. Dismissal and Settlement of Claims C. Special Litigation Committees CONCLUSION
Corporate managers are deterred from wrongdoing by both public and private enforcement. While some types of corporate malfeasance may result in criminal or civil sanctions at the hands of the government, the staff and budget of regulators are limited. For this reason, corporate law relies heavily on private enforcement through state law derivate suits and federal securities class actions. (1) The efficiency and effectiveness of private enforcement are therefore of central concern for corporate law.
A threshold question is whether the merits of legal claims matter in stockholder litigation. Private enforcement relies heavily on the plaintiffs' bar to identify and prosecute promising cases. Since suits are initiated by plaintiffs' attorneys and settled by corporate managers using firm or insurance company dollars, the risk of strike suits and collusive settlements is high. (2) The problem confronting efforts to answer this question is that the merits of claims of corporate malfeasance are generally unobservable. To get around this problem, prior research on stockholder litigation has relied on variables that can be observed and are assumed to correlate with legal merit: the presence of an accounting restatement, for example, or a parallel SEC investigation. Such measures are noisy proxies for the merits of cases. Moreover, such a strategy restricts the researcher to legal claims actually litigated and omits cases of malfeasance that never resulted in a claim being filed. This is a significant omission, as the deterrence function of litigation depends critically on the probability that bad acts reliably lead to litigation.
This Article takes a novel approach to studying stockholder litigation by identifying a context in which it is possible to quantify breaches of duty across a large universe of firms, both sued and unsued. Specifically, we study cases arising out of the stock options backdating scandal. Backdating involved the falsification of the grant dates for stock options used to compensate key employees in order to covertly increase the employees' compensation. (3) Executive stock options typically set an exercise price (4) equal to the stock price on the day the option was issued, and as a result, options issued on days when the stock price happened to be low were more valuable to executives. (5) By falsifying grant dates, backdaters were able to create an appearance that grants were issued on dates in the past when the stock price happened to be low, while accounting for them as though they had been issued on the falsified date. Backdating involved the manipulation of stock option grants with statistically measurable consequences: grants were more likely to be issued on favorable dates. Because of this practice, and because option grants are publicly reported, we are able to calculate the likelihood that individual firms engaged in manipulative practices over a very large sample. This methodology provides a measure of the merits (6) of potential backdating claims that is both more precise and available for a larger universe of firms than in other types of litigation.
The data for this study includes hand-collected data on private shareholder litigation--both derivative suits under state law and class actions under the federal securities laws--alleging options backdating. For derivative suits, we have collected extensive information on each case, including the full set of claims pre-consolidation, the decision on the motion to dismiss, the use and recommendations of special litigation committees, and information about attorneys' fees and settlement. We supplement this data with information about the presence of securities class actions from public reports, public lists of SEC investigations, and the disclosure of backdating activities in media and analyst reports.
We combine this legal data with data on option grants and statistical simulation to estimate both the probability that a firm engaged in backdating and the extent to which the executive recipients of option grants benefited from that backdating. These measurements provide a more precise picture of the merits than in other shareholder litigation contexts. It is possible, therefore, to produce a firm-level, ex ante estimation of the merits of backdating claims for a large sample of firms. Armed with these estimates of merit, we investigate whether shareholder litigation, in the aggregate, targets the "right" firms, how the merits affect the progress and disposition of the litigation, and whether the recoveries correlate to the merits of the claim.
We find that a majority of firms that likely engaged in backdating were never publicly linked to the practice. Among those firms publicly alleged to have engaged in backdating--in analyst reports or news coverage--a majority, but not all, were named in a derivative suit. Even fewer firms were targets of securities class actions. The sued firms were more likely to have backdated and have higher total reversal around likely-backdated option grants than the publicly implicated but unsued firms. Similarly, firms targeted in class actions, which were a subset of the firms sued derivatively, show more egregious patterns of backdating than firms subject only to derivative claims and also larger stock price drops when backdating activity was revealed. These results suggest that the incidence of lawsuits--even controlling for public revelations of backdating and SEC investigations--was linked to merits-related measures of backdating activity.
We also find that sued firms differ from unsued firms along other dimensions. Derivatively sued firms were larger than implicated but unsued firms. Additionally, firms that were targets of SEC investigations were more likely to face both class action and derivative suits. We observe a race-to-the-courthouse effect in derivative litigation, with multiple lawsuits targeting a single firm and making similar allegations. This effect was strongest among large firms and firms investigated by the SEC, though the number of complaints was not otherwise related to our measures of merit.
We find no strong predictors of the disposition of the motion to dismiss in either derivative or class action suits. None of the covariates is significant in our regressions for either derivative or class action claims, and even a sensitive non-parametric test does not distinguish the merits of dismissed claims from non-dismissed claims. This may reflect that the motion to dismiss often turns on legal rules that are not directly related to the alleged wrongful activity, such as demand on the board in the derivative context. (7)
We find some evidence that the size of settlements is related to the merits of cases. For shareholder derivative suits, we use the size of attorneys' fees as a proxy for the settlement amount, and we find that fees are related to the level of backdating activity. For class action suits, we find no relationship between settlement amounts and either the backdating probability or the total value extracted through backdating. However, we do find that when the SEC conducted an investigation, cases had larger settlements.
We also examine the use of special litigation committees (SLCs) as a tool to regain corporate control over derivative litigation. We find that the use of SLCs was both strongly related to the number of complaints filed and an indicator variable reflecting a very high probability of actual backdating activity. Interestingly, only a minority of SLCs recommended that the company seek dismissal of the claims, contrary to the common claim that SLCs recommend dismissal as a matter of course.
The welfare implications of stockholder litigation in general and the derivative suit in particular are hotly disputed in corporate law scholarship. Some scholars have concluded that the derivative suit is in need of radical reform, (8) if not complete abolition. (9) In important ways, state corporate law has responded to these critiques by forcing derivative plaintiffs to run a demanding gauntlet of procedural requirements. (10) Others have a more positive view of derivative litigation, seeing it as an important bulwark against managerial opportunism. (11) The debate over derivative litigation has gained particular importance in the aftermath of the recent ATP Tour, Inc. v. Deutscher Tennis Bund case. In ATP Tour, the Delaware Supreme Court held that corporate bylaws that shift the cost of derivative litigation to unsuccessful plaintiffs are lawful. (12) Such bylaws would increase the expected costs of pursuing a stockholder claim and thus may reduce--perhaps dramatically--the volume of stockholder litigation. This, of course, is the ambition of those who promote such bylaws, but there is a concomitant risk that such a wholesale approach would eliminate meritorious suits as surely as it would eliminate nuisance suits. If derivative litigation is unrelated to the underlying merit of claims, then there would be little worth preserving. The fee shifting debate in Delaware was at least temporarily resolved when, over the objection of the U.S. Chamber of Commerce, Delaware passed an amendment to the Delaware General Corporation Law (DGCL) in 2015 that effectively prohibited corporations from adopting bylaws that would shift the cost of litigation to losing plaintiffs. (13)
The data we present here suggests that derivative litigation--at least in the backdating context--is more responsive to the underlying merits than many observers suspect. Sued companies have higher measures of both the magnitude and probability of backdating than unsued firms, and we find a significant relationship between attorneys' fees and merit. The relationship between measures of legal merit and litigation outcomes suggests that efforts to reform the pathologies of stockholder litigation should be approached with caution.
At the same time, one of the unique features of our data here--the availability of measures of legal merit and potential damages--counsels against generalizing our findings. The measures of merit and damages we compute are based on publicly available information, so plaintiffs' attorneys could have relied on them just as we have here. Accordingly, the responsiveness of derivative litigation in this context may not necessarily mean that derivative litigation works well in contexts where the merits are obscured from the view of researchers and plaintiffs' attorneys. Nevertheless, in light of the fact that the continued existence of shareholder derivative liability is a matter of open debate, identifying a context in which derivative suits are merits-related is important.
The information we present about backdating litigation is also important in its own right. While existing studies have analyzed the prevalence of backdating, the price impact of being implicated in the backdating scandal, and SEC investigations of backdating, ours is the first study to offer a comprehensive picture of private shareholder litigation involving backdating allegations.
I. DERIVATIVE LITIGATION AND OPTIONS BACKDATING
Do the merits of corporate legal claims affect the incidence and outcomes of stockholder suits? The question cuts to the heart of corporate law's design, which relies on private enforcement to animate substantive rights. This Part outlines the importance of the do-the-merits-matter question and the elusive quest to answer it. It also introduces the stock option backdating scandal of 2006 and 2007 and explains why that episode offers a unique opportunity to investigate this basic corporate law question.
A. Do the Merits Matter in Stockholder Litigation?
The utility of shareholder suits has been the subject of longstanding and contentious debate. (14) Shareholder litigation could be a tool that harnesses the self-interest of plaintiffs' attorneys to deter misconduct at public companies. Alternatively, it could be a mechanism that operates chiefly to benefit plaintiffs' attorneys and defendants at the expense of shareholders. (15) The answer carries important policy implications: if stockholder litigation fails to focus on actual wrongdoing, there can be little hope that it in fact deters wrongdoing or that it delivers compensation to those who suffer from it. If there is no connection between the merits and the operation of litigation, stockholder litigation--an expensive system of private enforcement--would be in need of deep reform if not outright abolition. (16)
The principal procedural hurdles in stockholder litigation, for both derivative and securities suits, have been shaped by the desire to inhibit meritless lawsuits. The requirement that a stockholder first make a demand on the board of directors in derivative suits, for example, has traditionally been justified as a bulwark against nuisance suits. (17) Similarly, courts and commentators have defended the SLC, which allows a board committee to wrest control of the derivative claims from the complaining stockholder, as a mechanism for weeding out meritless claims. (18) In the securities context, the demanding pleading requirements and specialized mechanism for selecting a lead plaintiff introduced by the Private Securities Law Reform Act of 1995 were directly aimed at reducing the volume of suits. (19)
A prominent and persistent focus of reformers' attention has been attorneys' fees. Under the prevailing rule in the United States (the so-called American Rule), each party to litigation pays its own attorneys regardless of the outcome. The alternative rule--known as the English Rule--forces the losing party to pay the winner's legal fees and expenses. (20) A longstanding suspicion is that the American Rule induces meritless strike suits because plaintiffs (and their attorneys) can sue and offer to settle for less than the cost of defending the suit. Assuming that stockholders would be less likely to bring derivative suits if forced to bear the corporation's cost of defense, many states in the middle of the twentieth century adopted security-for-expenses statutes. (21) These statutes entitle the corporation to demand that any small stockholder bringing a suit post security for the corporation's legal expenses, from which the corporation could recover at the termination of the suit. The target was explicitly stockholder suits with no merit. The Governor of New Jersey, upon signing that state's statute, noted that the legislation was intended to "deter the filing of irresponsible suits by persons who either have no legitimate cause of action or who institute such action more for the personal gain of a settlement out of court than in the interest of the corporation or its stockholders." (22) At the time, many academic observers feared these statutes would upset a careful equilibrium in corporate law and fundamentally damage private enforcement of fiduciary duties. (23) Through a combination of creative pleading by plaintiffs, forgiving amendments by legislatures, and lenient interpretations by courts, (24) such dire predictions did not come to passes and indeed it was not long before commentators proclaimed the revival of the derivative suit. (26)
The American Rule remains a target of reform, especially in the wake of the Delaware Supreme Court's May 2014 decision in ATP Tour, Inc. v. Deutscher Tennis Bund and the subsequent legislative response. ATP Tour, Inc., the governing body of men's professional tennis and a Delaware nonstock company, had adopted a bylaw that purported to require any league owner who initiated litigation against the Tour to reimburse the Tour's legal fees if the plaintiff did not "obtain a judgment on the merits that substantially achieves, in substance and amount, the full remedy sought." (27) Answering certified questions from a federal court about the validity and enforceability of the bylaw, the Delaware Supreme Court held that such a bylaw is valid in a nonstock corporation and also noted that it would be enforceable if adopted for a proper purpose. (28) In analyzing the proper purpose inquiry, the court observed that "[t]he intent to deter litigation ... is not invariably an improper purpose." (29)
The decision commanded immediate and widespread attention for its implication that a public company might use a bylaw to shift from the American to the English Rule. (30) Critics of stockholder suits saw in the decision the promise of curtailing meritless litigation. Stephen Bainbridge, for example, noted that "we are faced with a world in which runaway frivolous litigation is having a major deleterious effect on U.S. capital markets," and fee-shifting bylaws offer "an appropriate means of addressing the problem through private ordering." (31) On the other side, plaintiffs' attorneys feared the worst, describing the decision as a "disaster" that "caused Delaware to secede from the union" by forsaking the American Rule. (32) In the wake of the ATP Tour decision, the Delaware bar proposed a legislative amendment that would prohibit fee-shifting bylaws, but that proposal was initially derailed by lobbying effort by the Business Roundtable. While some intrepid public companies adopted fee-shifting bylaws after ATP, (33) Delaware ultimately amended the DGCL to make fee-shifting bylaws explicitly impermissible over resistance from some business groups. (34)
Both the existing structure of class-based stockholder litigation and the debate over reforms sparked by ATP turns on whether stockholder suits bear some relationship to merit. That question has always been in the background of judicial opinions on stockholder suits, (35) and academics have worked to supply an answer for generations. (36) The challenge in testing the connection between the merits and outcomes in stockholder litigation is that the merits are usually hidden. (37) Plaintiffs' attorneys may search through mountains of documents and depose scores of potential witnesses before they can determine whether a claim has merit. The researcher--with neither access to the fruits of discovery nor the time to review them--has little hope of estimating a lawsuit's merit. (38)
Early academic work on the merits of stockholder litigation drew inferences from observed variation in settlement amounts and attorneys' fees. (39) Janet Cooper Alexander's influential study, for example, examined nine securities class actions and found little variation in the settlement amounts, (40) with most settling for twenty-five percent of the alleged stockholder loss. (41) The critical assumption of Alexander's study was that the merits varied across the cases, (42) and the invariance of the settlement amount suggested that the merits did not affect settlements. (43) Alexander's study, alongside the contemporaneous work of Roberta Romano, (44) suggested that the outcomes of stockholder litigation had little relationship to the underlying strength of the claims, and thus the fundamental mechanism of deterrence through litigation was broken. (45)
Most work in this area has focused on federal securities litigation, and the common empirical approach is to rely on variables that ought to correlate with merit. James Bohn and Stephen Choi, for example, used an ambitious variety of proxies for merit in their 1996 study of securities class actions against IPOs prior to the Private Securities Litigation Reform Act of 1995 (PSLRA). (46) They examined underwriter quality, the fraction of total holdings that insiders sell in the IPO, the fraction of outsiders on the pre-offering board of directors, the potential damages (the difference between the IPO price and the price at the end of the class period, multiplied by the number of shares offered), and the capital market reaction to the filing of the securities suit. (47)
In post-PSLRA work, Marilyn Johnson, Karen Nelson, and A.C. Pritchard examined what they termed "factors related to fraud": accounting restatements, earnings forecasts, and insider trading. (48) Additionally, drawing on an insight of Joseph Grundfest, (49) Choi used as proxies for merit two measures based on the ultimate settlement amount in his study of post-PSLRA securities IPO litigation. (50) This study treated a suit as meritorious if it settled for more than $2 million, (51) reasoning that "the maximum amount defendants will settle a nuisance claim typically will not exceed $2 million." (52) As an alternative measure, the study treated suits as meritorious if the settlement amounts exceeded five percent of the IPO offering amount. (53)
James Cox and Randall Thomas have employed a measure they term "provable loss." (54) This is a market-adjusted measure of the stock's abnormal returns in response to the disclosure of the information that corrected the alleged misstatement or omission underlying the securities claim. (55) The larger the share of provable loss that a case recovers, the stronger the inference that it was meritorious. (56) Robert Thompson and Randall Thomas considered an extensive catalog of factors in trying to assess the merit of cases in their study of stockholder litigation in the Delaware Court of Chancery during 1999 and 2000. (57) They considered the involvement of controlling stockholders, the unadjusted merger premia offered by acquirers, and the size of the monetary settlement to be signals of potential merit. (58) They also relied on various case characteristics that were "perhaps the most commonly discussed perceived 'abusive' features of representative litigation." (59)
Some recent papers have attempted to quantify the merits of certain discrete legal claims more precisely in other contexts. In their study of mutual fund litigation, one of the authors of this Article and John Morley examined lawsuits between 2000 and 2009 alleging excessive fee liability under section 36(b) of the Investment Company Act. (60) During the period they studied, the legal standard for 36(b) liability was that the fee in question must be "so disproportionately large that it bears no reasonable relationship to the services rendered and could not have been the product of arm's-length bargaining." (61) For each fund in the universe of mutual funds subject to 36(b), Curtis and Morley determined the fund's unadjusted expense ratio and produced variables to measure the excessiveness of the fees as alternative estimates of merit. (62) They found a "somewhat modest" relationship between fund fees and the incidence of a lawsuit. (63) By contrast, they found a strong relationship between fee litigation and fund family size. (64)
Charles Korsmo and one of the authors of this article examined litigation challenging mergers and acquisitions involving Delaware-incorporated target firms. (65) They collected the universe of merger transactions between 2004 and 2012 where both of the major associated legal remedies--fiduciary class actions and stockholder appraisal petitions--were available. (66) To estimate the merits of either legal claim, they computed a merger premium residual by subtracting the actual merger premium from a predicted merger premium based on size, industry, and year. (67) They also used a going-private dummy as a proxy for merit. (68) They found very little relationship between the merit estimates and the incidence or intensity of fiduciary class actions. (69) By contrast, the incidence of appraisal--where plaintiffs can proceed only on their own behalf and not on behalf of the entire class of absent shareholders--was strongly associated with low premium residuals. (70) The Korsmo and Myers paper shares a basic similarity with the Curtis and Morley paper in that the legal claims in both contexts are relatively one-dimensional. For this reason, the legal claims in those contexts are susceptible to an unusually transparent quantitative measure of merit. (71)
Stock options backdating is of interest to us because it presents an even more tantalizing opportunity to measure the merits of legal claims. Derivative suits commonly allege some fiduciary duty violation, and evaluating the merits of such a claim requires particularized knowledge that is unavailable without the benefit of discovery. (72) Even with discovery, it would be quixotic to quantify such an evaluation of merit. But the gravamen of each backdating lawsuit is that the board retroactively granted option awards. As explained below, we can generate firm-level estimates of the likelihood of backdating and the magnitude of the damages. These measures of backdating activity provide unique insight into the merits of backdating shareholder and derivative claims. To be clear, we cannot capture every element of the legal claims that might be relevant. For example, in a derivative suit, the number of directors who participated in the backdating scheme might be relevant to whether demand is excused. In a securities suit, the state of mind of the directors or officers participating in the scheme would be relevant. Nevertheless, the ability to capture with reasonable precision the underlying wrongful activity provides a unique opportunity to evaluate stockholder litigation.
B. The Backdating Scandal
Attorneys described the backdating scandal of 2006 as "one of the broadest ever to sweep across corporate America," (73) rivaled only by the bribery and illegal payments scandal of the Watergate era. (74) Backdating was first identified by financial economists. In the late 1990s, David Yermack studied option grants to CEOs of Fortune 500 companies and discovered that companies making option awards outperformed the market by more than two percent over the succeeding fifty days. (75) He speculated that option grants were made in advance of favorable corporate news. (76) In 2004, Erik Lie circulated a paper showing negative abnormal returns before unscheduled option grants in addition to positive abnormal returns afterward. (77) He offered a new hypothesis to explain that pattern: "[T]he awards might be timed ex post facto, whereby the grant date is set to be a date in the past on which the stock price was particularly low." (78) The SEC saw the paper and opened a handful of investigations, (79) which became public and attracted the attention of news media and plaintiffs' attorneys. (80))
The story took off after the Wall Street Journal published an extensive front-page report on backdating on March 18, 2006. (81) This report--unlike the academic papers--identified individual firms that may have backdated, highlighting well-timed grants at eight firms. (82) In the succeeding weeks, attention on backdating increased as the Wall Street Journal and other media outlets reported on additional companies that may have engaged in backdating. (83) Wall Street analysts further fanned the flames: in May 2006, Merrill Lynch produced two reports that attracted considerable attention, identifying backdating companies in the high technology industry, and the Center for Financial Research & Analysis identified additional companies with potential backdating problems from among the one hundred companies with the largest option grants before Sarbanes-Oxley (SOX). (84) As the scrutiny directed at companies named as potential backdaters in the media grew, many boards of directors initiated voluntary reviews of their own past grant practices. (85) By the end of the summer, backdating had become one of the biggest corporate scandals in a generation.
For implicated firms, backdating had some predictable consequences. Correcting for backdating could lead to the need to restate past financial results. (86) Companies might also violate debt covenants in the course of restating their financials, forcing them to deal with a default. (87) Backdating allegations also shook investors' confidence in a company's management and in its financial reports. (88) The losses in market capitalization following backdating allegations--on the order of hundreds of millions of dollars per firm--were far out of proportion to any estimate of the out-of-pocket costs associated with handling backdating allegations. (89)
Backdating also gave rise to a substantial amount of public and private litigation. The SEC and DOJ investigated over one hundred firms and filed enforcement actions against some of them. Some worried at the time that the SEC did not pursue cases with adequate vigor. (90) But as Choi, Wiechman, and Pritchard have shown, the SEC poured enormous resources into backdating investigations, diverting attention from other areas. (91) Of the criminal prosecutions, (92) a few resulted in convictions--of former executives from Comverse Technology, Monster World, KB Home, and Brocade Communications--but not many. One commentator declared that the results had "not been as good as the earlier corporate fraud prosecutions." (93)
Somewhat surprisingly, federal securities suits never gained much traction in the backdating context. There were only thirty-six such suits filed and, while there were several huge settlements, most of the cases were notable for settling for less than many observers expected. (94) The common explanation for the absence of much securities litigation is that the stock prices of implicated companies did not drop sufficiently on backdating news to generate large damages. (95) The longer term market penalty uncovered by academic studies, (96) of course, suggests that the market reactions were in fact sometimes large, but plaintiffs' attorneys nevertheless opted against bringing many securities claims.
Derivative litigation was overwhelmingly the enforcement mechanism of choice for plaintiffs' attorneys. As we show below, shareholders filed over 600 derivative lawsuits based on backdating allegations. (97) In a derivative claim, the damages do not depend on the stock price reaction to backdating news. Instead, damages depend on the magnitude of the ill-gotten gains associated with the practice. (98)
There has been very little empirical work on private litigation over backdating. Two studies that documented the extent of the multiforum trend in stockholder litigation relied on surveys of where backdating derivative suits were filed. (99) In addition, Bernile and Jarrell investigated whether the stock price movements they documented could have been due to anticipation of stockholder lawsuits, which would impose costs on firms. (100) Not distinguishing between securities lawsuits and shareholder derivative lawsuits, they found that plaintiffs targeted firms where backdating likely took place and firms with more assets. (101)
II. DATA AND METHODOLOGY
We combine data on stock option grants with hand-collected data on the incidence of stockholder derivative litigation alleging backdating. We augment this data with additional information on securities class actions and SEC investigations. This Part describes our data and important variables used in the analysis below.
A. Methodology for Identifying Backdating Activity
We obtained data on stock option awards at public companies from the Thomson Financials Insider Filing database. The data set includes all insider transactions reported on Form 4. We construct a database of option grants by filtering on insider transactions to identify grants between January 1, 1996, and August 29, 2002, when changes to reporting rules made backdating much more difficult. We eliminate grants to low-level executives, as in Bizjak, Lemmon, and Whitby. (102) This leaves a dataset of 181,852 option grants across 8520 firms.
To determine whether a particular option grant is backdated, we compute the return on the underlying stock in the twenty trading days before the grant and the 20 days after the grant. (103) The post-grant change in price minus the pre-grant change in price is the reversal. Intuitively, the reversal measures the depth of the "V" around the option grant and captures both increases in price after the grants and avoidance of losses before the grant. We compute the reversal for all grants in the dataset. We then compute the reversal for 1,000,000 hypothetical grant dates by making random draws from our sample of firms and dates over the same time period as our sample of option grants.
To estimate the probability that a grant is backdated, we match the grant to the randomly generated sample of hypothetical grant dates with the same volatility level and compute the proportion of hypothetical grants with lower reversal. Volatility is measured as the standard deviation of daily stock returns and captures the tendency of a firm's share prices to change quickly. More volatile stocks are more likely to have a high or low reversal on a given date than less volatile stocks, so controlling for volatility is important. The result is a volatility-controlled estimate of the likelihood that a grant on a randomly chosen date would have a reversal at least as large as the observed reversal of the grant. This can be understood as a probabilistic measure of the likelihood that a grant was backdated.
To estimate backdating at the company level, we use a similar technique. For a company with k option grant dates in the sample, we draw 1000 samples of k options at random from the set of hypothetical grants with the same volatility decile. We compute the cumulative reversal across all k option grants for the firm in question as well as all 1000 draws of k hypothetical grants. We estimate the probability of the cumulative reversal occurring randomly by observing the proportion of random draws that have cumulative reversal larger than the observed reversal. The estimated probability that a company engaged in backdating is the proportion of random draws of k grants with lower total reversal.
Using these company-level estimates of the likelihood that the observed reversals are random, it is possible to estimate the proportion of likely backdating firms in the sample. The company-level probabilities are p-values, and we can take p < 0.05 percent as the confidence cutoff for backdating firms. Thirteen percent of the firms meet this cutoff. It does not follow, of course, that all of these firms backdated. Five percent of the non-backdating firms would have p < 0.05 by chance. While the true proportion of non-backdating firms is not directly observable, it is nevertheless possible to estimate this quantity using statistical methods developed for measuring false discoveries in mixed samples. (104)
To estimate the proportion of firms with aggregate reversal with p < 0.05 that did not in fact engage in backdating, we first choose a cutoff p-value [lambda], below which we assume that no firms engaged in backdating. We use [lambda] = 0.6. That is, we assume that firms whose cumulative reversal was worse than sixty percent of randomly generated reversals did not backdate. We then estimate the number of firms with p < 0.05 that did not backdate as follows:
[W (p < 0.05)/N] * [W (p > 0.6) N [1/1 - [lambda]]]
Where W(p < y) is the number of firms with a probability of random reversal less than y, and N is the number of firms in the sample. Intuitively, this uses the proportion of firms with high p-values that likely did not engage in backdating to estimate the proportion of firms with p < 0.05 that did not backdate. Figure 1 illustrates why this estimation intuitively makes sense.
Based on the equation above, we estimate that there are 855 firms in our sample--or about 10%% of total firms with option grant information--that can be said with 95% confidence to have positive cumulative reversal around option grants that is not the result of random variation. This is lower than the estimate of aggregate options backdating activity of Edelson and Whisenant, who estimate that about 16% of their sample of 4008 firms engaged in backdating with 95% probability using a different methodology for measuring grant probabilities, but still much higher than the number of firms publicly implicated in the practice of backdating. (105)
Our methodology identifies non-random patterns in firm grants. It is of course possible that these non-random patterns are not the result of backdating, but of some other practice like spring-loading, (100) where the disclosure of positive news is timed to follow a grant of options. The effect for the grant recipient is essentially the same--options values are increased--but the legal analysis is different because backdating involves a deliberate violation of a stockholder- approved options plan and the filing of a demonstrably false document with the Securities and Exchange Commission. (107)
To determine the proportion of backdating (as opposed to, say, spring loading) in our sample, we compare our primary sample of grants to grants where backdating was not possible but other methods of increasing option value, like spring loading, could have still worked. In particular, we compare our main sample of grants in the pre-Sarbanes-Oxley era, when companies had a long period in which to report option grants, to grants in the post-Sarbanes-Oxley era that were reported within one day of being issued. (108) Applying the methodology described above to the samples of grants, we find that the incidence of unusually lucky grants is 80% lower among grants for which backdating was impossible compared with our main sample of pre-SOX grants. (109) This suggests, consistent with the finance literature, that the majority of abnormal performance around grants was due to backdating. (110)
B. Description of Backdating Variables
Based on the methodology described above, we derive three measures of backdating activity. First, we use one minus the firm-level p-value on cumulative reversals around grant dates as a measure of the overall degree to which a firm is likely to have engaged in backdating. For convenience of reference, we refer to this as the Firm-Level Backdating Probability. While this is a reasonable name for the variable, it is subject to some measurement error. For example, a firm that backdated only once but issued many options grants may be overlooked by this measure if the cumulative effect of the truly random grants conceals the effect of the backdated grant. Such a firm may be measured to have a low backdating probability even if the single grant could be shown to be backdated with near-certainty. Second, if a firm engaged in other practices that created abnormal reversal, such as timing bad news before option grants and positive news after options grants, then this could be incorporated into the backdating measure, even if the firm did not actually backdate options. Nevertheless, the firm-level measure is helpful in identifying firms that backdated frequently and with measurable economic consequences.
We also compute the p-value of individual option grants. We term grants that have reversal with an estimated likelihood of occurring randomly of less than 0.05 "lucky" grants. (111) Of course, some significant proportion of lucky grants were, in fact, the result of luck, since there is a one in twenty likelihood of such a grant resulting from chance. Moreover, firms that issued many grants are more likely to have lucky grants. To get a more detailed view of firm-level backdating activity, we use the percentage of each firm's grants that were lucky to construct the variable Percent Lucky Grants, which implicitly controls for the frequency with which firms granted options. This measures the incidence of backdating but, again, does not fully capture the scope of backdating activity as it does not distinguish large grants from small ones.
To measure the total effect of backdating on firms' financial reports, we compute the Total Abnormal Reversal. For each lucky grant, we multiply the number of options times the reversal, from twenty days prior to the grant to twenty days after, times the exercise price of the option grant. If we assume that the average reversal of non-backdated grants was zero, then this measure approximates the degree to which the option was "in the money" on the day it was issued and thus the compensation expense that should have been recorded as a result. Note that this value is different from the actual economic impact of the backdating, since the accounting standards at the time did not attempt an actual valuation of the options. (112)
For our regression results, we use both the Firm-Level Backdating Probability and the Total Abnormal Reversal as the variables of interest. Firm-Level Backdating Probability captures the likelihood that a company backdated while Total Abnormal Reversal correlates with the value taken out of the firm through backdating. These variables constitute our main estimates of the merits of backdating derivative cases.
We also collect data on other measures of backdating activity that help validate our merits measures. First, our data includes a list of firms investigated by the SEC for backdating. (113) One might assume, and indeed we confirm below, that the SEC focused on the worst offenders, so this variable is a strong indicator of significant backdating activity. Second, our data includes the size of restatements issued by firms that acknowledged backdating issues. While not every firm that backdated restated its financials, the size of the resulting restatement for those firms that did is a measurement of how widespread and aggressive the backdating activity was. (114) Third, we code whether each firm appeared on one of several lists of likely backdaters produced by media and investment analysts. In unreported regressions, we confirm that all these measures of backdating are strongly correlated with Firm-Level Backdating Probability and Total Abnormal Reversal with p < 0.01.
We use Firm-Level Backdating Probability and, separately, Total Abnormal Reversal as estimates of the strength of a stockholder suit making backdating allegations. The variable Firm-Level Backdating Probability supplies an estimate of legal liability in a fiduciary-duty-based claim because backdating was a straightforward violation of law that gave rise to damages. (115) The variable Total Abnormal Reversal captures the extent to which the options were in the money when granted, and since the likely remedy in a fiduciary based-suit would be rescission in one form or another, (116) this variable correlates with the magnitude of the potential damages. Together, these variables are estimates of liability and damages in a fiduciary suit. They measure the merits of a claim that the company could bring against those who engaged in backdating.
In derivative litigation, the stockholder seeks to stand in the shoes of the company to press its claims. Thus, our measures of merit only apply to private litigation in which the private plaintiffs' attorney successfully gained control of the case. To do that, the private suit must survive a motion to dismiss for failure to make demand on the board. (117) Only when the board is conflicted--meaning that the board members themselves are the targets of the backdating suit--can the private attorney avoid the demand requirement. (118) To investigate this possibility, we examined the effect of board turnover on our results. In unreported tests, we found that director turnover at the backdating company between 2001 and 2006 had no effect on our targeting or litigation outcomes below.
C. Data on Litigation over Backdating
We compiled data on stockholder derivative cases by hand. As the backdating story grew in the news, media entities and other analysts compiled lists of firms implicated in backdating. We collected all four of these lists and combined them into one master list of firms implicated in backdating. (119) We find 264 firms where backdating was publicly alleged to have occurred, of which the option grant and other stock price data necessary to be included in the sample was available for 255. (120)
For each of these 255 firms, we examined its periodic disclosure filings with the SEC to determine whether the firm had attracted stockholder derivative litigation. (121) For each firm with disclosed litigation, we collected data on all of the complaints against the firm by examining documents in court dockets, SEC disclosures, and contemporaneous news reports. For each company, we compiled the filing date, the venue, the filing attorneys, and the lead counsel (if the case was consolidated). We note the ultimate outcomes of cases: whether settled, consolidated with other cases, or dismissed. If the case was settled, we collected the details of the settlement including any amounts paid to the company, option repricings or cancellations, corporate governance changes, the plaintiffs' attorneys' fee, (122) any D&O insurance payment (if disclosed), and whether anyone objected to the settlement. If a court dismissed the case, we noted the grounds. We also noted if the case ended in some other way, such as bankruptcy or merger.
In order to identify securities class action claims, we used a comprehensive list from Kurtzman Carson Consultants, LLC. (123) The list includes the companies sued, whether the case was dismissed, and the size of any settlement paid to the class. (124)
One variable we could not collect was the plaintiffs' stock holdings. The plaintiffs' holdings were rarely noted in court pleadings. When disclosed, however, the holdings appeared trivial. For example, in a fight over which of three plaintiffs would be lead in a consolidated case involving Power Integrations, the judge requested information on their holdings. One plaintiff did not respond, another claimed to hold two shares, and the third plaintiff was unreachable, according to his attorney. (125)
The filing dates of the derivative cases were clustered between early 2006 and early 2007, mirroring public attention on the phenomenon. Figure 2 presents a histogram of derivative filings over time. There were only a handful of backdating cases filed before the Wall Street Journal published its front page story on March 16, 2006. (126) Also, 90% of all cases were filed between April 2006 and March 2007.
As has been noted in other work, the venue selections by plaintiffs' attorneys were skewed towards federal court and, even when filed in state court, out of the state of incorporation. (127) Table 1 shows the venues selected by firms, sorted by the selecting firms' state of incorporation.
Please note: Illustration(s) are not available due to copyright restrictions.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Abstract into II. Data and Methodology C. Data on Litigation over Backdating, p. 291-319|
|Author:||Curtis, Quinn; Myers, Minor|
|Publication:||University of Pennsylvania Law Review|
|Date:||Jan 1, 2016|
|Previous Article:||Time to drop the infield fly rule and end a common law anomaly.|
|Next Article:||Do the merits matter? Empirical evidence on shareholder suits from options backdating litigation.|