Do the Merits Matter More - Michigan Law - University of Michigan

3 downloads 0 Views 564KB Size Report
Daniel Spies provided invaluable research assistance. We thank ... the merits matter more in securities fraud class actions after the passage of the PSLRA? The.
UNIVERSITY OF MICHIGAN JOHN M. OLIN CENTER FOR LAW & ECONOMICS

DO THE MERITS MATTER MORE? THE IMPACT OF THE PRIVATE SECURITIES LITIGATION REFORM ACT MARILYN F. JOHNSON KAREN K. NELSON & A.C. PRITCHARD PAPER #02-011 UPDATED VERSION 2/2006

THIS PAPER CAN BE DOWNLOADED WITHOUT CHARGE AT: MICHIGAN JOHN M. OLIN WEBSITE HTTP://WWW.LAW.UMICH.EDU/CENTERSANDPROGRAMS/OLIN/PAPERS.HTM

1

Do the Merits Matter More? The Impact of the Private Securities Litigation Reform Act (Forthcoming, Journal of Law, Economics, and Organization)

Marilyn F. Johnson Eli Broad College of Business, Michigan State University

Karen K. Nelson Jones Graduate School of Management, Rice University

A.C. Pritchard * Law School, University of Michigan [email protected]

February 2006/Final Version We appreciate the helpful comments of Jennifer Arlen, Bill Beaver, Ronald Mann, and Roberta Romano, as well as workshop participants at the American Enterprise Institute, American Law and Economics Association annual meeting, University of California at Berkeley, Fordham University, George Mason University, University of Michigan, University of Texas, and Vanderbilt University. Daniel Spies provided invaluable research assistance. We thank First Call Corporation for insider trading data and PricewaterhouseCoopers for litigation data. Pritchard received financial support from the Cook Fund at the University of Michigan Law School.

2

Do the Merits Matter More? The Impact of the Private Securities Litigation Reform Act

Abstract: This paper examines the effect of the Private Securities Litigation Reform Act of 1995 (PSLRA) on stockholder lawsuits. We explore the role of restatements, earnings forecasts, and insider trading in the filing and resolution of lawsuits for a sample of high technology firms. Consistent with our predictions, there is a post-PSLRA shift away from litigation based on forward-looking earnings disclosures. Conversely, there is a significantly greater correlation between litigation and both earnings restatements and abnormal insider selling after the PSLRA. Finally, we find a post-PSLRA increase in the likelihood of settlement for cases involving earnings restatements.

Keywords: Securities litigation, litigation risk, accounting fraud, insider trading.

3

1.

INTRODUCTION Do the “merits” matter in securities fraud class actions? This question captured the

attention of both scholars (Alexander, 1991) and legislators (Hearings, 1995) in the early 1990s. Congress eventually concluded that the potentially enormous damages in securities fraud class actions were encouraging frivolous “strike” suits. In Congress’s view, plaintiffs’ lawyers were filing suits “citing a laundry list of cookie-cutter complaints,” typically based on a company’s prior optimistic statements that had not come to fruition. These strike suits were filed “within hours or days” of a substantial drop in the company’s stock price (H.R. Rep., 1995:16). Moreover, plaintiffs’ lawyers had incentives to “file frivolous lawsuits in order to conduct discovery in the hopes of finding a sustainable claim not alleged in the complaint” (S. Rep., 1995:14). To discourage such suits, Congress adopted the Private Securities Litigation Reform Act (“PSLRA”) (1995). The PSLRA erects a series of procedural barriers that have resulted in a higher percentage of securities fraud class actions being dismissed (Foster et al., 2000). The number of suits being filed, however, has not declined. After an initial dip, the number of securities fraud class actions has returned to, and even exceeded, its pre-PSLRA level. (Buckberg et al., 2005). The larger number of filings suggests that the PSLRA may have done little to discourage the filing of frivolous suits, although it may have increased their likelihood of dismissal. Plaintiffs’ lawyers respond that the suits have always been merit driven and that the only thing that has changed post-PSLRA is that meritorious suits are now being dismissed. They argue that the upsurge in filings simply reflects a massive expansion in the amount of fraud being committed (Lerach, 2001), perhaps due to reduced exposure to liability (Bernardo et al., 2000).

4

Consistent with this position, the Securities and Exchange Commission (SEC) has expressed concern about the quality of financial reporting (Levitt, 1998). A more cynical explanation for the surge in filings posits that plaintiffs’ lawyers are incapable – based on the information available to them – of sorting fraud from bad luck. Consequently, they sue on bad news that may reflect either; if they can withstand the issuer’s inevitable motion to dismiss, they can gain access to discovery of the corporation’s internal documents that will allow them to determine whether fraud has been committed. A higher dismissal rate means that plaintiffs’ lawyers need to file more suits in hopes that a reasonable number will make it through to discovery. The above discussion brings us back to our initial question, in slightly revised form: Do the merits matter more in securities fraud class actions after the passage of the PSLRA? The impact of the PSLRA has taken on new significance in the political climate engendered by accounting and insider trading scandals at Enron, Worldcom, and other companies. This study offers evidence on the role that the merits play in securities litigation before and after the passage of PSLRA. Has the PSLRA’s forward-looking safe harbor afforded more protection to companies’ forecasts of earnings information? In enacting the PSLRA, Congress expressed concern that companies failing to meet earnings expectations were vulnerable to securities fraud class action. Fear of liability exposure, it was argued, discouraged firms from making earnings forecasts, despite the importance of such disclosures to the pricing of securities. Congress addressed this concern with a forward-looking safe harbor, which makes it more difficult to bring fraud claims based on projections. Consequently, courts subject these statements to more exacting scrutiny than other general allegations of misleading statements or omissions. In effect, Congress has

5

raised the standard for what counts as “meritorious” for a class of claims that legislators suspected had a very low correlation with fraud. This more stringent definition of merit may discourage claims based on forward-looking statements. Has the law’s higher bar for adequately pleading a securities fraud complaint increased the importance of merit-based factors, including violations of accounting principles and insider trading? The SEC has made accounting fraud an enforcement priority over the last few years, and the increase in options-based compensation has created additional opportunities for insider trading by company managers. Courts recognize both accounting violations and insider trading as supporting an inference of fraudulent intent. Do allegations by plaintiffs’ lawyers correlate with aggressive accounting choices and insider trading that is unusual in amount or timing, or are the allegations simply a formulaic attempt to satisfy pleading standards? Moreover, is the correlation stronger or weaker after the enactment of the PSLRA’s more stringent pleading standard? To address these questions about the impact of the PSLRA, we examine three aspects of class actions: (i) the filing of suits, (ii) the allegations in those suits, and (iii) the resolution of the suits. Our primary research question is whether the PSLRA made forward-looking statements less important, and aggressive accounting and insider trading more important, in explaining the incidence, type of allegations, and resolution of securities fraud class actions. Following prior research, we focus on the high technology sector, consistently a favorite target of plaintiffs’ lawyers. We first examine factors that distinguish firms that are sued from firms that are not sued despite experiencing a contemporaneous stock price drop of similar magnitude. We argue that the PSLRA’s stringent pleading requirements encourage plaintiffs’ lawyers to focus more on objective evidence that the firm and its managers acted with fraudulent

6

intent to mislead investors. Thus, we predict a greater correlation between lawsuit filings and both earnings restatements and unusually high levels of insider stock sales following passage of the PSLRA. Because of the safe harbor protection afforded by the PSLRA, we also predict that lawsuits are less likely to be associated with forward-looking earnings disclosures following the law’s passage. We then examine, for firms that are sued, whether there is a similar shift in the factors explaining the allegations contained in lawsuits and the likelihood of settlement versus dismissal. Our sample consists of 114 computer hardware and software firms (and their respective matches) sued during 1991-2000, with 50 of these lawsuits occurring prior to the PSLRA and the remainder after. We draw two primary conclusions from our empirical analysis. First, forwardlooking earnings disclosures are less important in securities litigation after the PSLRA. The likelihood that an earnings warning will result in a lawsuit filing or an allegation of false or misleading projections is significantly lower in the post-PSLRA period. This evidence is important in light of prior research on pre-PSLRA samples showing that voluntarily preempting bad earnings news frequently triggers securities litigation (e.g., Francis et al., 1994; Skinner, 1997). Our more recent evidence indicates that although earnings warnings continue to be an important factor in securities litigation, the PSLRA has reduced the litigation risk associated with these disclosures, thus achieving part of Congress’s goal in adopting the forward-looking safe harbor. Consistent with Skinner (1997), however, there is no evidence that forward-looking earnings disclosures are associated with lawsuit outcomes (settlement versus dismissal) either before or after the PSLRA. Second, we find a significantly greater correlation between litigation and both earnings restatements and abnormal insider selling after the PSLRA. Earnings restatements are

7

significantly more likely to result in the filing of a lawsuit, an allegation of accounting fraud, and a settlement in favor of the plaintiff after the PSLRA. Moreover, we show that this result is not merely an artifact of the greater number of restatements occurring in the post-PSLRA period (e.g., Wu, 2002). In addition, lawsuit filings and insider trading allegations correlate more significantly with abnormal insider selling than before the PSLRA. This paper is the first to offer a comprehensive study of restatements, earnings forecasts, and insider trading as determinants of litigation, and the first to compare their role in litigation before and after the PSLRA. The evidence in this paper thus contributes to a substantial crossdisciplinary body of research examining the causes and consequences of shareholder litigation, and how changes in legal rules affect the filing and resolution of shareholder claims. Moreover, our evidence informs the debate on the “merits” of securities fraud class actions. Although we cannot directly observe the merits of class actions, we can examine the impact of legislation like the PSLRA to discover how legal rules affect the filing and resolution of shareholder claims. Our evidence suggests a closer relation between factors related to fraud and securities class actions after the passage of the PSLRA, suggesting that Congress may have achieved at least some of its goals. This conclusion, however, comes with a caveat. We do not examine the possibility that the PSLRA may also discourage certain lawsuits that would have been deemed meritorious prior to the statute’s enactment, as suggested by the evidence presented in Choi (2005). The remainder of the paper proceeds as follows. Section 2 provides institutional background on the PSLRA and discusses related research. Section 3 develops our hypotheses. Section 4 describes the sample selection procedure and data collected. Section 5 presents our results. Section 6 concludes the paper with a summary and discussion of our major findings.

8

2.

BACKGROUND

The Private Securities Litigation Reform Act Congress enacted the PSLRA into law on December 22, 1995, by overriding President Clinton’s veto. Congress intended the PSLRA to curtail what many considered to be a rising trend of abusive class action litigation, and thus it contains a number of new hurdles for plaintiffs filing securities fraud complaints. Perhaps most daunting of these hurdles is the PSLRA’s safe harbor for the voluntary disclosure of financial projections and other forward-looking information. To qualify for safe harbor protection, the statements must be identified as forwardlooking and be accompanied by “meaningful” cautionary language discussing important factors that could cause actual results to differ from those projected. In the absence of an appropriate disclaimer, the plaintiff must still prove that the statements were made with actual knowledge that they were false or misleading. A more pervasive barrier to weak claims is the PSLRA’s stringent pleading standard, which makes it more difficult for plaintiffs to allege securities fraud without specific evidence of misconduct. Plaintiffs must specify in their complaint each statement alleged to have been misleading and the reasons why the statement is misleading. In addition, plaintiffs must state with particularity facts giving rise to a “strong inference” that the defendant acted with “the required state of mind,” i.e., with fraudulent intent. Both the forward-looking safe harbor and the pleading standard are reinforced by the PSLRA’s discovery stay. This provision prevents plaintiffs from conducting discovery while a motion to dismiss is pending, and is intended to prevent plaintiffs from engaging in a “fishing

9

expedition” for evidence to build their case, which, it is frequently argued, causes defendants to settle dubious claims simply to avoid the high cost of discovery. Related Literature Several prior studies examine the relation between voluntary disclosure and shareholder litigation. Skinner (1994) and Kasznik and Lev (1995) find evidence consistent with firms’ voluntarily preempting large negative earnings surprises in an attempt to minimize the risk of lawsuit. However, other research suggests that the voluntary disclosure of bad news frequently triggers securities litigation (e.g., Francis et al., 1994; Skinner, 1997), even though it may reduce the expected cost of litigation associated with disappointing earnings news (Skinner, 1997). Although these studies provide evidence on earnings disclosures and securities litigation in the pre-PSLRA period, the expressed intent of the PSLRA’s safe harbor is to make it harder to sue based on voluntary disclosure of forward-looking information. Johnson et al. (2001) examine whether firms changed their disclosure policies in response to the safe harbor. They find a significant increase in the frequency of firms issuing forecasts and the number of forecasts issued in the first year following enactment of the PSLRA, and that the change in disclosure is increasing in firms’ ex ante risk of litigation. 1 Although this evidence suggests that managers believe the safe harbor reduced legal exposure, Johnson et al. (2001) do not examine the relation between disclosure and litigation in the post-PSLRA period. We contribute to the existing literature by investigating disclosure-related litigation under the new regime to provide evidence on the efficacy of the PSLRA’s safe harbor in sheltering forward-looking statements. Other studies examine the relation between accounting problems and shareholder litigation. Kinney and McDaniel (1989) and Palmrose and Scholz (2004) examine the incidence of litigation for firms restating earnings. Kinney and McDaniel (1989) report that approximately

10

14 percent of firms with quarterly restatements from 1976 to 1985 were sued by stockholders or the SEC. Palmrose and Scholz (2004) report a shareholder litigation rate of 38 percent for quarterly and annual restatements announced between 1995 and 1999. 2 In a study of restatements from 1977 to 2000, Wu (2002) reports that 22 percent of restatement firms were the subject of a securities litigation complaint. Moreover, she finds that the number of restatements was fairly stable from the mid-1980’s to mid-1990’s, but increased beginning in 1998. 3 Jones and Weingram (1996b) examine the related issue of whether sued firms are more likely than non-sued firms with a similar single-day stock price decline to have restated earnings within the year preceding or following the decline. For their sample of firms from 1989-1992, they find that litigation risk is significantly higher for firms restating earnings. In separate analyses, they find no evidence that insider trading increases litigation risk. Johnson, Kasznik and Nelson (2000), however, find that stock sales by insiders increase the likelihood of being sued in the pre-PSLRA period. 4 Our paper offers two primary contributions to the above literature. First, we present a more comprehensive analysis of factors affecting securities litigation. Rather than investigate a single attribute of the litigation environment as is generally done in prior research, our analysis includes voluntary earnings disclosures, accounting restatements, and insider trading, as well as governance and market-related variables. We also examine three related aspects of the litigation process – filings, allegations, and outcomes – to obtain a more complete picture of the role played by each of these factors. Second, and more importantly, we compare the characteristics of securities litigation in the pre- and post-PSLRA periods. This analysis leads to a greater understanding of changes in the practice of private securities litigation since the PSLRA.

11

Prior research examining the PSLRA’s impact either focuses on the market reaction to its enactment or on tabulations of lawsuit data. Spiess and Tkac (1997) and Johnson, Kasznik, and Nelson (2000) document that, on average, the PSLRA was wealth-increasing. Specifically, there was a significant negative market reaction to news releases indicating that President Clinton would veto the legislation, followed by a significant positive reaction to the veto override. Moreover, the market response is increasing in firms’ overall risk of litigation, but decreasing in the incremental probability of being sued for fraud (Johnson, Kasznik, and Nelson 2000). Also consistent with this evidence, Johnson, Nelson, and Pritchard (2000) find an overall positive market reaction to a related event, a Ninth Circuit court decision adopting a particularly stringent interpretation of the PSLRA’s pleading standard. The reaction is particularly strong for firms headquartered in the Ninth Circuit and those at greatest risk of being sued in a securities class action. Taken together, this evidence suggests that shareholders considered the PSLRA’s restrictions on securities litigation to be beneficial. Ali and Kallapur (2001), however, argue that the positive abnormal return at the time of the PSLRA’s enactment is more likely a response to the presidential veto rather than the override, and thus shareholders considered the PSLRA to be harmful. The event study evidence is therefore inconclusive. Moreover, stock price reactions do not provide direct evidence of the PSLRA’s effect on securities litigation. Other studies focus more specifically on the effect of the PSLRA on litigation. Studies describing the frequency of lawsuit filings and the type of allegations show that lawsuit filings initially declined under the PSLRA, but subsequently recovered to pre-PSLRA levels (Grundfest and Perino 1997; PricewaterhouseCoopers 2000; Perino 2003). 5 In addition, allegations of accounting irregularities and insider trading increased, while cases based on allegations of false forecasts decreased. These descriptive summaries of allegations in lawsuits, however, do not

12

shed light on the relation between the allegations in the lawsuits and the conduct of firms and their managers. Pritchard and Sale (2005) study how judges applying the PSLRA resolve motions to dismiss securities fraud complaints in the Second and Ninth Circuit. They find that complaints are significantly more likely to be dismissed in the Ninth Circuit. They also find that allegations of insider trading in complaints correlate with dismissal in both circuits, but that the circuits vary in their approaches to allegations of accounting violations and false-forward looking statements. Their study is limited, however, to the allegations made in the complaints and they do not compare the pre-and post-PSLRA periods. Beatty et al. (2001) find that the risk of litigation in connection with an IPO declined significantly following the enactment of the PSLRA, but they do not provide evidence on the determinants of those filings. Finally, Bajaj et al. (2003) find that although mean settlements increased after the passage of the PSLRA, investors recovered a smaller percentage of potential losses. They do not, however, compare systematically whether the determinants of those settlements have changed with the adoption of the PSLRA.

3.

RESEARCH HYPOTHESES A key element of any securities fraud claim is a material misstatement or omission.

Some of the strongest evidence to satisfy this requirement available to plaintiffs’ lawyers is a violation of generally accepted accounting principles (GAAP) that results in an earnings restatement, which is required only when earnings have been materially misstated. Moreover, given the importance that investors place on earnings in valuing a company’s stock, a restatement of earnings is likely to correlate with large expected damages. Thus, we expect the likelihood of a lawsuit filing, accounting allegation, and settlement to be positively associated

13

with the occurrence of an earnings restatement. Our test variable, Restatement, is an indicator equal to one if the firm restated class period earnings. Two features of the PSLRA discussed above likely increase the importance of nonGAAP reporting in securities litigation. First, the heightened pleading standard requires plaintiffs to allege specific evidence of fraud, such as that provided by an earnings restatement, in their complaint. Second, the safe harbor increases protection for forward-looking disclosures, shifting the focus to misstatements of historical information. Consequently, we predict that the association between earnings restatements and filings, accounting allegations, and settlements is greater in the post-PSLRA period. 6 As suggested above, plaintiffs’ lawyers can argue that the firm made false or misleading forward-looking statements, although the PSLRA’s statutory safe harbor makes it particularly difficult to plead and prove such claims. Thus, we predict that voluntary forward-looking earnings disclosures are an important factor in securities litigation, but we expect this association to decline in the post-PSLRA period. We utilize three indicator variables to capture firms’ forward-looking disclosures. First, voluntarily warning of bad earnings news can cause a sudden stock price drop that attracts the attention of the plaintiffs’ bar (e.g., Francis et al., 1994; Skinner, 1997). Thus, the variable Earnings Warning is equal to one if the firm issued a bad news earnings forecast on either the last day of the class period or the day with the most negative stock return during the class period. 7 Second, firms may attempt to avoid litigation by managing expectations downward. Thus, the variable Negative Forecast is equal to one if the firm issues an unfavorable earnings projection at any point during the class period other than on the last day or the minimum return date. Finally, forecasting good earnings news leaves the firm vulnerable to litigation if the

14

projections are not realized. Thus, the variable, Positive Forecast, is equal to one if the firm issues an optimistic earnings forecast during the class period. We expect the likelihood of a lawsuit filing and forecast allegation to be positively associated with Earnings Warning and Positive Forecast, and negatively associated with Negative Forecast. The expected relation between the forecast variables and the likelihood of a settlement is less clear. An earnings warning may trigger plaintiffs’ lawyers to file a lawsuit, but it offers little evidence of fraud. A positive forecast may suggest unwarranted optimism, but it is unlikely to distinguish the firm from many others that are also making forecasts, thus offering only weak evidence of fraud. Finally, a negative forecast may be construed as an effort to discourage litigation, but once the decision to file a lawsuit has been made, it sheds little light on whether the company has made a misrepresentation. Accordingly, we have no prediction for the relation between these variables and the likelihood of a settlement. Another key element of private securities class actions is that the firm and its managers acted with scienter, i.e., fraudulent intent to mislead investors. To establish intent, plaintiffs’ lawyers frequently point to stock sales by executives and directors, arguing that these insiders knowingly perpetrated fraud to inflate stock prices and profit from selling their company stock holdings. Thus, we predict that the likelihood of a lawsuit filing, insider trading allegation, and settlement is associated with stock sales by insiders during the class period. Moreover, as with non-GAAP reporting, the PSLRA’s heightened pleading standard likely increases the importance of insider trading in securities litigation. Thus, we predict that the association between insider trading and filings, insider trading allegations, and settlements is greater in the post-PSLRA period.

15

Niehaus and Roth (1999) find that managers are net sellers during lawsuit class periods, although the level of these sales is not significantly different from managers’ prior trading activity. The difference between the two measures of insider trading suggested by Niehaus and Roth (1999) is important because courts focus on abnormal insider selling as evidence of fraudulent intent (Sale 2002). 8 Accordingly, we measure our insider trading variable, Abnormal Insider Trading, as the difference between the net amount of shares traded (in millions) by directors and executive officers during the class period and the net amount of shares traded by directors and officers during an equal number of days preceding the start of the class period. For example, if the class period is 120 days, we subtract the net amount of insider shares traded during the 120 days prior to the start of the class period from the net amount of insider shares traded during the class period. 9 A negative value indicates abnormal net sales. If litigation decisions accord with judicial doctrine, we expect to observe a negative coefficient estimate on Abnormal Insider Trading. We also consider the role of corporate governance in securities litigation. Although weak governance does not provide direct evidence of fraud, several studies document that weak governance is associated with enforcement actions by the SEC (e.g., Beasley, 1996; Dechow et al., 1996). In addition, Helland and Sykuta (2005) find a relation between board independence and the incidence of securities fraud class actions. Thus, our analysis includes the following governance variables, where the sign of the predicted relation with lawsuit filings, allegations, and settlements is given in parentheses: (i) Avg. Tenure, the average number of years outside directors have served on the Board (–), (ii) Busy, the average number of other directorships held by outside directors (+), (iii) Independent, the percentage of outside directors (–), (iv) Audit

16

Meetings, the number of meetings held by the firm’s audit committee (–), and (v) Independent Audit, the percentage of independent directors on the audit committee (–).

4.

DATA The sample consists of firms in the computer hardware (SIC codes 3570-3577) and

computer software (SIC codes 7370-7379) industries during 1991-2000. The high technology sector has been the most common target for class actions both before and after the PSLRA, unlike other sectors where the incidence of litigation has fluctuated over time for reasons unrelated to passage of the PSLRA. 10 We use the Securities Class Action Alert to identify firms sued in securities fraud class actions in the pre-PSLRA period (1991-1995), and the Stanford Securities Class Action Clearinghouse to identify firms sued in the post-PSLRA period (19962000). Data from these sources, along with disclosures in firms’ periodic SEC filings, discussions of cases in judicial opinions, websites of various claims administrators, and data provided by PricewaterhouseCoopers, were used to identify the date the lawsuit was filed, the class period, the types of allegations contained in the complaint, and the lawsuit outcome. Market and accounting data were obtained from the CRSP and Compustat tapes. Data on accounting restatements and voluntary management earnings forecasts were obtained from a Lexis search of news stories as well as the company’s periodic filings with the SEC. Finally, information regarding corporate governance structure was obtained from firms’ last proxy statement prior to the beginning of the class period, if available; if not, the first available proxy after the beginning of the class period was used. As shown in Table 1, Panel A, our lawsuit sample consists of 114 firms with the requisite data, 50 of which were sued pre-PSLRA and 64 post-PSLRA. The proportion of lawsuits

17

alleging accounting fraud doubled under the PSLRA, from 32 percent in the pre-PSLRA period to 61 percent in the post-PSLRA period. One possible explanation for this increase is that there has been a decline in the quality of financial reporting in recent years (Levitt, 1998). Alternatively, the finding is consistent with plaintiffs’ lawyers believing a lawsuit containing an allegation of accounting fraud is more likely to withstand the higher pleading standard of the PSLRA and altering their filing strategy to maximize the potential for success. Consequently, it is unclear from these descriptive statistics whether accounting problems play a more important role in securities litigation subsequent to the PSLRA. The passage of the PSLRA is also associated with a twofold increase in the proportion of insider trading allegations, from 38 percent in the pre-PSLRA period to 80 percent in the postPSLRA period. 11 This trend is consistent with an increase in insider trading associated with greater stock-based compensation in the post-PSLRA period, but, as with the increase in allegations of accounting fraud, it is also consistent with plaintiffs’ lawyers adapting the form of their complaints to the PSLRA’s standards. In contrast to the accounting and insider trading allegations, Table 1, Panel A reveals that allegations of a false or misleading forecast are less prevalent in the post-PSLRA period. On average, 86 percent of lawsuits in the pre-PSLRA period contained a forecast allegation, compared to 64 percent post-PSLRA. Although the rate of forecast allegations remains relatively high in the post-PSLRA period, only 4 of the post-PSLRA lawsuits are based solely on a forecast allegation, compared to 22 of the pre-PSLRA lawsuits. These findings are consistent with the PSLRA’s safe harbor provision increasing protection for voluntary disclosures of forward-looking information.

18

Table 1, Panel B reports descriptive statistics on lawsuit outcomes. The frequency of dismissals increased after the PSLRA, as did the frequency of lawsuits settled for less than $5 million. 12 The median settlement is also lower in the post-PSLRA period ($5.63 million) than in the pre-PSLRA period ($7.50 million), although the mean settlement post-PSLRA ($17.99 million) is approximately twice the pre-PSLRA mean ($9.26 million). However, differences in the mean and median settlement (with or without dismissals) are not statistically significant across the two periods. To examine factors associated with lawsuit filings, we construct a control sample by matching each lawsuit firm with a firm from the same industry (computer hardware or computer software) that was not sued but which experienced a one-day stock price decline during the class period similar in magnitude to the largest single day stock price drop experienced by the sued firm. Losses suffered by investors are a critical element in determining damages, and thus a stock price decline is a necessary, but not sufficient, condition for litigation. Plaintiffs’ attorneys use price declines as an initial screen in selecting which firms to sue. Not all of these firms are sued, however, as plaintiffs’ attorneys search for other factors useful in establishing liability, such as those discussed in the preceding section, that suggest a lawsuit has sufficient probability of recovery. Table 2 presents descriptive statistics comparing the combined lawsuit and control samples in the pre- and post-PSLRA periods. The findings indicate a significantly greater number of restatements in the post-PSLRA period, consistent with the pattern documented by Wu (2002). In contrast, there is no evidence of a change in abnormal insider trading or forwardlooking earnings disclosures. 13 Finally, only one of the governance variables, Busy, is

19

significantly different between the pre- and post-PSLRA periods. Outside directors serve on significantly fewer boards in the post-PSLRA period. In addition to our primary variables of interest, prior research indicates that market capitalization (Market Cap.) and share turnover (Turnover), which factor into the determination of potential damages, are positively associated with the incidence of lawsuits (e.g., Francis et al., 1994; Jones and Weingram, 1996a). The results in Table 2 indicate that there is no statistical difference in firm size between the pre- and post-PSLRA periods, although share turnover is significantly higher post-PSLRA. Finally, Table 2 reports descriptive statistics for the minimum one-day return (Min. Return) during the class period. The results reveal that Min. Return is significantly more negative in the post-PSLRA period.

5. RESULTS Lawsuit Filings Table 3 presents univariate tests of differences between the lawsuit and control firms in the pre-PSLRA (Panel A) and post-PSLRA (Panel B) periods. Although the incidence of earnings restatements is higher for lawsuit firms in both periods, the difference is only significant following passage of the PSLRA. Similarly, Abnormal Insider Trading is significantly more negative for lawsuit firms only in the post-PSLRA period. In contrast, lawsuit firms issued significantly more positive forecasts and earnings warnings than control firms in both the preand post-PSLRA periods. With respect to governance structure, the comparisons in Table 3 provide mixed evidence. Consistent with expectations, outside directors of lawsuit firms sit on significantly more boards than their counterparts in control firms in both the pre- and post-PSLRA periods.

20

However, contrary to expectations, lawsuit firms have significantly more independent members on the board and audit committee and held significantly more audit committee meetings than control firms in the post-PSLRA period. Finally, Table 3 reveals that, as expected, lawsuit firms are significantly larger and more actively traded than control firms in both periods. The class period single-day stock price drop is also significantly lower for lawsuit firms despite the effort to select a control sample matched on this variable. We examine the implications of this difference in the analysis that follows. Table 4 reports summary statistics from a multinomial logistic regression examining determinants of lawsuit filings. Multinomial logit is an extension of the binary logit model to multiple choices. The procedure estimates the probability of a particular alternative relative to the probabilities of all other alternatives. In the analysis of lawsuit filings, there are three alternatives: (i) no lawsuit, (ii) lawsuit in the pre-PSLRA period, and (iii) lawsuit in the postPSLRA period. 14 The table presents results from a single multinomial logistic regression that permits different coefficient estimates for the pre- and post-PSLRA periods, with non-sued firms as the reference category. The table also reports significance levels from chi-square tests of differences between the two periods. The findings indicate that firms restating class period earnings are significantly more likely to be sued both before and after the PSLRA. However, as predicted, the risk of being sued for an earnings restatement is significantly greater in the post-PSLRA period. Lawsuit filings are also more highly correlated with abnormal insider stock sales in the post-PSLRA period. Specifically, the coefficient estimate on Abnormal Insider Trading is negative in both periods, but is significantly more negative in the post-PSLRA period.

21

The evidence in Table 4 also suggests a change in the relation between voluntary disclosure and securities litigation after enactment of the PSLRA. In the pre-PSLRA period, issuing a forecast of positive earnings news or an earnings warning significantly increased the risk of a lawsuit filing, while issuing a forecast of negative earnings news did not significantly reduce litigation risk. Although earnings warnings continue to be an important trigger of lawsuits in the post-PSLRA period, firms are significantly less likely to be sued for such disclosures. In addition, issuing a forecast of positive earnings news is no longer a significant determinant of lawsuit filings, although the decrease in litigation risk associated with these disclosures is insignificant. Taken together, these results suggest that the PSLRA’s safe harbor provides firms some protection from disclosure-related litigation, consistent with the implications of Johnson et al. (2001). Conditional on the above variables, there is little evidence of an association between governance structure and lawsuit filings in either the pre- or post-PSLRA periods. The signs of the coefficient estimates on our five governance variables are generally in the predicted direction, but only one variable, Busy, is significant, and then only in the pre-PSLRA period. Moreover, there is no evidence of a shift in the relation between corporate governance and securities litigation between the pre- and post-PSLRA periods. 15 Regarding the controls for estimated damages, we find that the likelihood of being sued is increasing in firm size in both periods, as expected, but there is no change in this relation between the pre- and post-PSLRA periods. Share turnover is insignificant in both periods. Finally, the results indicate that although Min. Return is insignificant in the pre-PSLRA period, it is significantly negative in the post-PSLRA period. This finding provides additional evidence that sued firms experience unusually large single-day stock price drops, particularly in the post-PSLRA period.

22

To examine whether the results in Table 4 are due to an inability to find a reasonably comparable control firm for all lawsuit firms in the sample, we first calculate the absolute value of the difference in Min. Return for each matched pair. We then exclude the five percent of matched pairs with the largest difference and re-estimate the model. In this analysis (untabulated), the coefficient estimate on Min. Return is insignificant in both periods, as is the difference between periods. However, there is no change in inferences with respect to our primary variables of interest. Specifically, the risk of being sued for an earnings restatement or abnormal insider selling is significantly greater in the post-PSLRA period, while the risk of being sued upon issuance of an earnings warning is significantly lower. Thus, although the stock price drop of some lawsuit firms is difficult to match, our findings are not driven by these observations. 16 As a further sensitivity check, we examine whether differences in lawsuit determinants between the pre- and post-PSLRA periods are robust to an estimation procedure that explicitly controls for the matched-pair nature of the data. This procedure, often called conditional logistic regression, corrects for possible correlated omitted variables bias due to the matching procedure, particularly if the matching is not perfect. 17 Consistent with Table 4, the untabulated results of this analysis indicate that the coefficient estimate on Restatement is significantly more positive in the post-PSLRA period and the coefficient estimate on Abnormal Insider Trading is significantly more negative. In contrast to Table 4, there is no statistical difference in the coefficient estimate on Earnings Warning between periods. Further analysis, however, reveals that this finding is sensitive to the inclusion of three matched pairs identified as statistical outliers. When these three observations are removed, the coefficient estimate on Earnings Warning is significantly less positive in the post-PSLRA period. Taken together, the results presented in this section

23

provide evidence of a post-PSLRA shift away from litigation based on forward-looking disclosure. Instead, post-PSLRA lawsuits are more likely to be filed when there is evidence of non-GAAP financial reporting or abnormal stock sales by insiders. Lawsuit Allegations We supplement the analysis of lawsuit filings by examining the type of allegations contained in those filings. The evidence reported above indicates that the primary determinants of litigation risk have changed between the pre- and post-PSLRA periods. As a result, it is reasonable to expect a corresponding shift in the type of complaints alleged in pre- and postPSLRA filings. The descriptive evidence reported in Table 1, Panel A indicating an increase in accounting and insider trading allegations and a decrease in forecasting allegations is consistent with such a conclusion. However, it has also been suggested that plaintiffs’ lawyers may simply base their complaints on allegations believed to be sufficient to withstand a motion to dismiss under the legal standards applicable during each period, regardless of whether those allegations are supported by evidence consistent with the allegation (e.g., Grundfest and Perino, 1997). Thus, in this section we explore the determinants of lawsuit allegations before and after the PSLRA. Table 5 presents summary statistics from multinomial logistic regressions for accounting allegations (Panel A), insider trading allegations (Panel B), and forecast allegations (Panel C). 18 The regressions include the variable(s) expected to explain each type of allegation as well as the corporate governance variables. We do not control for estimated damages because lawsuit allegations concern issues of liability rather than investor losses.

24

The findings in Panel A indicate that plaintiffs’ attorneys are more likely to include an allegation of accounting fraud in their complaint if the firm restated class period earnings. Moreover, this association is significantly stronger in the post-PSLRA period. Similarly, Panel B indicates that insider trading allegations are more highly correlated with abnormal stock sales by insiders in the post-PSLRA period. The coefficient estimate on Abnormal Insider Trading is negative and significant only in the post-PSLRA period, and is significantly less than in the prePSLRA period. Finally, Panel C reveals that, of the three forecast variables, only Earnings Warning is significantly associated with the incidence of forecast allegations. Plaintiffs’ attorneys are more likely to allege that the firm issued a false or misleading forecast when negative earnings news is associated with a large single-day stock price decline. However, such earnings warnings are significantly less likely to trigger a false forecast allegation in the postPSLRA period, consistent with the safe harbor reducing exposure to litigation based on forwardlooking disclosure. The results for the governance variables in all three panels of Table 5 provide little evidence of a systematic relation between corporate governance and the types of allegations contained in securities class action complaints. Only one of the five governance variables, Busy, is significantly different between the pre- and post-PSLRA periods in the analysis of accounting and insider trading allegations. The PSLRA’s pleading standard requires specific identification of alleged frauds, with potential sanctions for allegations that are not factually supported. Moreover, the PSLRA’s safe harbor provides protection for forward-looking disclosures. Consistent with these changes, the evidence in this section indicates that allegations of accounting fraud and insider trading are more closely related to objective measures of such behavior in the post-PSLRA period.

25

Moreover, allegations of false forecasts are less likely to follow from warnings of disappointing earnings news. Overall, this evidence supports the notion that the PSLRA has induced plaintiffs’ attorneys to more carefully tailor the allegations in their complaints to the underlying facts. Lawsuit Outcomes Our final analysis compares litigation outcomes in the pre- and post-PSLRA periods. The PSLRA imposes a number of procedural hurdles that affect the likelihood of a lawsuit being dismissed, but the law only tangentially affects the standard for liability and the calculation of damages. We therefore focus our analysis on factors associated with settlement versus dismissal rather than on cross-sectional variation in settlement amounts. For this analysis, we classify a lawsuit as a dismissal if it was resolved in the company’s favor. Thus, dismissals include lawsuits that were dismissed or settled for a small fraction of the firm’s market value. We consider these “nuisance” settlements, defined as a settlement of less than 0.5 percent of firm market value ten days before the end of the class period, as tantamount to a dismissal. 19 The results reported in table 6 indicate that lawsuit outcomes are not significantly associated with earnings restatements in the pre-PSLRA period. In contrast, lawsuits in the postPSLRA period are significantly more likely to result in a settlement if the firm restated class period earnings. The difference between the two periods is significant at the 0.01 level. Abnormal Insider Trading is insignificant in both periods. Results for the three earnings forecast variables are also generally insignificant, consistent with Skinner (1997), who finds no evidence of a relation between lawsuit settlement or dismissal and firms’ voluntary disclosures. The coefficient estimate on Negative Forecast is significant in the pre-PSLRA period, but not in the predicted direction. Thus, these results reveal a sharp contrast between filing/allegation

26

decisions and lawsuit outcomes. Specifically, abnormal stock sales by insiders and earnings warnings are important determinants of lawsuit filings and allegations, but are not a major factor in the outcome of the litigation. Finally, consistent with our previous regressions, there is no evidence of a relation between corporate governance structure and lawsuit outcomes. Sensitivity Analysis We conduct a number of additional sensitivity tests to assess the robustness of our results. First, although we focus on restatements of class period earnings as the strongest and most objective evidence of non-GAAP reporting, we also consider whether estimated discretionary accruals explain lawsuit characteristics in the pre- and post-PSLRA periods. Thus, we reestimate the filing, accounting allegation, and outcome models including abnormal accruals obtained from the modified Jones model (Dechow et al., 1995). 20 Untabulated results indicate that our inferences are not sensitive to the inclusion of this additional accounting variable. The coefficient estimate on abnormal accruals is incrementally positive and significant in the postPSLRA period in the filing and outcome models, and is significantly more positive than in the pre-PSLRA period in all three models. As noted above, the frequency of restatements increased beginning in 1998. Thus, we reestimate the filing, accounting allegation, and outcome models excluding all observations from 1998-2000. The results are robust for this reduced sample for the filing and outcome models. The association between an accounting allegation and a restatement of class period earnings is only marginally stronger in the post-PSLRA period for this reduced sample (p-value = 0.12). However, when we include an expanded set of governance variables, discussed below, in the accounting allegation model, Restatement is significantly more positive in the post-PSLRA

27

period (p-value = 0.06). Thus, our findings are not driven by a trend of increasing restatements in the post-PSLRA period. We also examine whether our results are robust to an alternative specification of the restatements variable that does not include restatements of “non-core” earnings components, such as in-process research and development write-offs (IPR&D). There are a total of five such “non-core” restatements, all in the post-PSLRA period and four of which relate to IPR&D. The results of this untabulated analysis reveal that none of our inferences are changed when we focus on restatements of “core” earnings. Because our governance results are generally insignificant, we re-estimate all regression models including an expanded set of governance variables. Specifically, we include three additional indicator variables identifying whether the firm’s chief executive officer is also the chairman of the board, whether the chief executive office is a company founder, and whether there is an external shareholder who owns at least five percent of the outstanding shares. Consistent with our primary findings, these governance variables are generally insignificant and there is no evidence of a systematic shift in the relation between corporate governance and securities litigation between the pre-and post-PSLRA periods. We also conduct a factor analysis of the governance variables to obtain summary measures of firms’ governance structures. There are three underlying factors, one positively correlated with the independence of the board and the audit committee, another positively correlated with the presence of a chief executive officer that is the board chair or a company founder, and the last positively correlated with the presence of an external blockholder. However, substituting these factors for the individual governance variables does not change any of our inferences. Finally, our results are robust to the inclusion of indicator variables for the computer hardware and software industries.

28

6. SUMMARY AND CONCLUSIONS This paper examines the impact of the Private Securities Litigation Reform Act on securities class action litigation. Our principal finding is that restatements and abnormal insider stock sales, two factors that relate to the likelihood of fraud, play a more important role in explaining the incidence of, and allegations in, litigation post-PSLRA. We also find that lawsuit filings and allegations of false forecasts are significantly less likely to be triggered by voluntary earnings warnings after the PSLRA. This paper thus extends pre-PSLRA findings on earnings disclosures and stockholder litigation (e.g., Skinner, 1994, 1997; Francis et al., 1994; Kasznik and Lev, 1995), and suggests that future research should consider the effects of this shift in the litigation environment on the incentives of management to disclose. Factors relating to damages continue to have explanatory power after the passage of the PSLRA. Damages factors, while unlikely to correlate with fraud, will always play a role in determining the incidence of suit because greater potential damages correlate with correspondingly greater attorneys’ fees. Earnings warnings are also a consistent predictor of litigation and allegations based on forward-looking statements. The failure to meet expectations evidently puts a firm on the “radar screen” of the plaintiffs’ bar, although the effect is not as significant after the PSLRA’s adoption. We believe that our results show a closer relation between factors related to fraud and the filing of securities class actions after the passage of the PSLRA. The efficacy of deterrence necessarily depends upon the accuracy with which sanctions are assessed because deterrence requires both sanctioning wrongdoers and protecting the innocent from sanctions. Accordingly, the evidence that we find of more precise targeting of securities class actions against firms likely

29

to have committed fraud suggests that Congress has achieved at least some of its objectives in adopting the PSLRA. Our study provides some evidence that the “merits” do matter more, at least in the filing of complaints and the allegations included in those complaints. To be sure, this conclusion comes with some caveats. Our model cannot explain all of the variation in the incidence of litigation. The unexplained variation undoubtedly has both merit and non-merit aspects. If the non-merit aspects predominate (an unlikely scenario in our view), our conclusion that there has been a shift in the determinants of litigation may not hold. The inconclusive results from our settlement regression should also be a cautionary note. We find little evidence that the PSLRA has enhanced the sorting process of litigation. One plausible explanation is that the PSLRA has a greater impact in discouraging meritless litigation than it does in enhancing the accuracy of litigation once initiated. The discovery stay, after all, limits the evidence available to the court to assess whether fraud has been committed. 21 Another important question not addressed by our research is whether the PSLRA has deterred lawsuit filings, a principal concern of the law’s critics. Plaintiffs’ lawyers may be unable to prove some meritorious claims under the rigorous constraints imposed by the PSLRA (see Choi (2005) for analysis of this issue). In particular, the forward-looking safe harbor has changed the definition of merit for some claims, which may well have the effect of excluding some claims that would have been deemed meritorious under the pre-PSLRA regime. Finally, the litigation environment may have shifted subsequent to our sample period. The Sarbanes-Oxley Act, passed in 2002, leaves intact the provisions of the PSLRA, but imposes more stringent governance and accounting control requirements on public companies in an effort to deter fraud. These mechanisms may reduce companies’ litigation risk by reducing the overall incidence of fraud. On the other hand, the Sarbanes-Oxley Act also requires the chief executive

30

officer and the chief financial officer to certify their company’s financial results, which could expose these executives and their companies to additional litigation risk because it provides additional evidence of scienter if those financial results are misstated. Litigation risk may also have been increased by the Sarbanes-Oxley Act’s lengthening of the statute of limitations for securities fraud claims. Additional research is needed to assess the impact of that law on securities fraud class actions.

31

APPENDIX Variable Definitions Variable

Definition

Restatement

Indicator variable equal to one if the firm restated class period earnings; zero otherwise

Abnormal Insider Trading

Shares purchased less shares sold (in millions) during the class period by directors, CEOs, COOs, CFOs, Presidents and Vice-Presidents less the same measure for an equal length period preceding the start of the class period, winsorized at the 5% and 95% level

Positive Forecast

Indicator variable equal to one if the firm made a positive forecast during the class period; zero otherwise

Negative Forecast

Indicator variable equal to one if the firm made a negative forecast during the class period other than on the minimum return date or the end of the class period; zero otherwise

Earnings Warning

Indicator variable equal to one if the firm made a negative forecast on either the class end or minimum return dates; zero otherwise

Avg. Tenure

Mean number of years that outside directors have been on the firm’s board

Busy

Mean number of external directorships of public companies held by outside directors

Independent

The percentage of outside directors on the firm’s board

Audit Meetings

Number of meetings held by the audit committee

Audit Independence The percentage of outside directors on the audit committee Market Cap.

Log of market value of common equity (in $ millions) at the end of the fiscal year preceding the beginning of the class period

Min. Return

Minimum one-day return during the class period plus one day after the end of the class period

Turnover

1 – (1 –Turn)X, where Turn is average daily trading volume divided by the number of shares outstanding, and X is the number of trading days during the class period

The governance variables are obtained from the last available proxy statement preceding the beginning of the class period. *

Corresponding author. We appreciate the helpful comments of Jennifer Arlen, Bill Beaver,

32

Ronald Mann, Roberta Romano, and two anonymous referees, as well as workshop participants at the American Enterprise Institute, American Law and Economics Association annual meeting, University of Alabama, University of California at Berkeley, Fordham University, George Mason University, University of Michigan, University of Texas, and Vanderbilt University. Daniel Spies provided invaluable research assistance. We thank PricewaterhouseCoopers for litigation data. Pritchard received financial support from the Cook Fund at the University of Michigan Law School. Johnson received financial support from PricewaterhouseCoopers. 1

In a related context, Baginski et al. (2002) find that, compared to their U.S. counterparts,

managers are more likely to issue forecasts in the less litigious Canadian environment. 2

Thus, while the Kinney and McDaniel (1989) sample predates the PSLRA, the restatements

examined by Palmrose and Scholz (2004) encompass both the pre- and post-PSLRA periods. 3

Untabulated statistics reveal a similar pattern in our data.

4

Other studies examine the association between insider selling and accounting enforcement

actions by the SEC, with mixed results. Dechow et al. (1996) find that insider selling at firms subject to enforcement actions does not differ significantly from a sample of control firms. Beneish (1999), however, finds that managers of firms subject to an enforcement action are significantly more likely to sell their holdings. 5

Most of the initial decline appears to be due to plaintiffs shifting venue to state courts to avoid

the PSLRA. However, the Securities Litigation Uniform Standards Act was passed on November 3, 1998, effectively closing this loophole. 6

An alternative explanation is that plaintiffs’ lawyers are no better at targeting accounting errors

than before the PSLRA; rather, there are simply more errors occurring. Thus, in sensitivity analysis we examine whether the increase in accounting restatements discussed above affects the

33

reported results. 7

Our definition of the class period includes one trading day following the end of the class period

specified in the lawsuit. This is because the revelation of potential fraud that triggers the suit may occur after the market closed, resulting in a negative market reaction the following trading day. The most negative stock return occurs on the last day of the specified class period for 14 percent (16 firms) of our lawsuit sample, while the most negative stock return occurs on the first trading day following the end of the specified class period for an additional 37 percent (42 firms) of our lawsuit sample. 8

For example, the court held in In re Apple Computer (9th Cir.1989) that insider trading is

suspicious only when it is “dramatically out of line with prior trading practices at times calculated to maximize the personal benefit from undisclosed inside information.” 9

To mitigate the effect of outliers, we winsorize Abnormal Insider Trading at the 5% and 95%

levels. 10

For example, litigation against finance companies declined significantly in the post-PSLRA

period due to the end of the savings and loan crisis and an associated reduction in loan loss reserve litigation (Grundfest and Perino 1997). 11

Because a lawsuit may contain multiple allegations, the total number of allegations in any

given year may exceed the number of lawsuits filed. 12

Following Skinner (1997), we include only the cash component of settlements because most

settlements are for cash and it is difficult to accurately value non-cash settlements. We code a case as dismissed in Table 1, Panel B if it was dismissed without a settlement. Some of the cases in our sample were dismissed by the court, but nonetheless subsequently settled by the parties either because the dismissal was granted with leave to refile or the plaintiffs had appealed the

34

dismissal. We code as settlements those cases that although initially dismissed subsequently were settled by the parties, generally for relatively small amounts. 13

At first glance, this result appears inconsistent with Johnson et al. (2001). However, they

consider all earnings forecasts made by high-technology firms in the two years surrounding passage of the PSLRA, whereas we only capture forecasts made by lawsuit and control firms during the class period. Thus, due to differences in research design, it is difficult to draw direct comparisons between the two studies. 14

Similarly, there are three alternatives in the allegation and settlements models discussed

below. 15

The coefficient on Independent is insignificant in our regressions, inconsistent with Helland

and Sykuta (2005) who find a significant negative relation between shareholder litigation and board independence. A number of factors could explain the difference in results, such as our focus on technology firms and our more extensive set of independent variables. Specifically, their analysis includes only two firm-specific controls, Tobin’s Q and an unfavorable audit opinion in a lengthy period (at least eight years) preceding the lawsuit filing. In addition, it is unclear from Helland and Sykuta’s description whether their sample includes only securities fraud class actions or whether it also includes derivative claims under state corporate law. If their sample includes the latter, their finding is not surprising because board independence is a defense to such claims. 16

Because the control sample is not required in our other empirical tests, our primary analysis

retains all lawsuit firms for comparability. 17

See Cram et al. (2003) for further discussion. The procedure requires estimating a logistic

regression, without an intercept term, using variable differences for each matched pair.

35

18

The three types of allegations are generally uncorrelated. In the pre-PSLRA period, the only

significant correlation is between accounting and insider trading allegations (0.26, p-value = 0.07). In the post-PSLRA period, the only significant correlation is between accounting and forecast allegations (–0.33, p-value = 0.01). 19

Inferences are unchanged using settlement cut-off points between 0.1 percent and 1.0 percent

of firm market value. Skinner (1997) considers lawsuits settled for less than $1 million as resolved in the company’s favor, i.e., dismissed. Our primary analysis does not use a fixed dollar amount to determine “nuisance” settlements because there is substantial cross-sectional and intertemporal variation in the size of our lawsuit firms. However, inferences are unchanged when we include all settlements, no matter how small, in the settlement category, or use cut-off points of $1 million or $2 million. Results are also robust when we exclude small settlements that are difficult to classify, defined either using a cut-off point of 0.5 percent of market value or $2 million, from the analysis. 20

Abnormal accruals is an estimate of earnings management obtained by separating out the

portion of total accruals due to management’s exercise of discretion from the estimated “normal” accruals attributable to the firm’s operations and economic environment. 21

The stay is lifted if the complaint survives the motion to dismiss, but courts may dismiss some

complaints that would have been deemed meritorious if the plaintiffs’ lawyer had greater access to information that would allow him or her to adequately plead a complaint.

Johnson 36

REFERENCES Alexander, Janet Cooper. 1991. “Do the Merits Matter? A Study of Settlement in Securities Class Actions,” 43 Stanford Law Review 497-598.

Ali, Ashiq, and Sanjay Kallapur. 2001. “Securities Price Consequences of the Private Securities Litigation Reform Act of 1995 and Related Events,” 76 The Accounting Review 431-460.

Baginski, Stephen P., John M. Hassell, and Michael D. Kimbrough. 2002. “The Effect of Legal Environment on Voluntary Disclosure: Evidence from Management Earnings Forecasts Issued in U.S. and Canadian Markets,” 77 The Accounting Review 25-50.

Bajaj, Mukesh, Sumon C. Mazumdar, and Atulya Sarin. 2003. “Securities Class Action Settlements: An Empirical Analysis,” 43 Santa Clara Law Review 1001-1033.

Beasley, Mark S. 1996. “An Empirical Analysis of the Relation Between the Board of Director Composition and Financial Statement Fraud,” 71 The Accounting Review 443-465.

Beatty, Randolph P., Philip P. Drake, and Chris E. Hogan. 2001. “The Impact of the 1995 Private Securities Litigation Reform Act on Litigation Risk and Auditor

Johnson 37

Compensation in the IPO Market,” Working paper, University of Southern California.

Beneish, Messod D. 1999. “Incentives and Penalties Related to Earnings Overstatements that Violate GAAP,” 74 The Accounting Review 425-457.

Beneish, Messod D., and Mark. E. Vargus. 2002. “Insider Trading, Earnings Quality, and Accrual Mispricing,” 77 The Accounting Review 755-791.

Bernardo, Antonio. E., Eric Talley, and Ivo Welch. 2000. “A Theory of Legal Presumptions,” 16 Journal of Law, Economics, & Organization 1-49. Buckberg, Elaine., Todd Foster, Ronald I. Miller, and Stephanie Plancich. 2005. “Recent Trends in Shareholder Class Action Litigation: Bear Market Cases Bring Big Settlements,” Unpublished Manuscript, National Economic Research Associates.

Choi, Stephen J. 2005. “Do the Merits Matter Less After the Private Securities Litigation Reform Act?” Working paper, New York University.

Cram, Donald P., Iris Stuart, and Vijay Karan. 2003. “Analysis of Matched Samples in Accounting Research,” Working paper, California State University – Fullerton.

Dechow, Patricia M., Richard G. Sloan, and Amy P. Sweeney. 1995. “Detecting Earnings Management,” 70 The Accounting Review 193-225.

Johnson 38

———. 1996. “Causes and Consequences of Earnings Manipulation: An Analysis of Firms Subject to Enforcement Actions by the SEC,” 13 Contemporary Accounting Research 1-36.

Foster, Todd S., Denise N. Martin, Vinita M. Juneja, and Frederick C. Dunbar. 2000. “Trends in Securities Litigation and the Impact of the PSLRA,” Unpublished Manuscript, National Economic Research Associates. Francis, Jennifer, Donna Philbrick, and Katherine Schipper. 1994. “Shareholder Litigation and Corporate Disclosures,” 32 Journal of Accounting Research 137-164.

Grundfest, Joseph A., and Michael A. Perino. 1997. “Securities Litigation Reform: The First Year’s Experience,” Working paper, Stanford University.

Hearings Before the Subcomm. on Telecomm. and Fin., House Comm. on Commerce on Legislation on Securities Fraud Litigation, 1995, 104th Congress, 1st Session.

Helland, Eric, and Michael E. Sykuta. 2005. “Who’s Monitoring the Monitor? Do Outside Directors Protect Shareholders’ Interests?” 40 Financial Review 155-172.

H.R. Rep. No. 104-50, 104th Cong., 1st Sess. (1995).

Johnson 39

Johnson, Marilyn F., Ron Kasznik, and Karen K. Nelson. 2000. “Shareholder Wealth Effects of the Private Securities Litigation Reform Act of 1995,” 5 Review of Accounting Studies 217-233.

———. 2001. “The Impact of Securities Litigation Reform on the Disclosure of Forward-Looking Information by High Technology Firms,” 39 Journal of Accounting Research 297-327.

Johnson, Marilyn F., Karen K. Nelson, and A.C. Pritchard. 2000. “In re Silicon Graphics, Inc.: Shareholder Wealth Effects Resulting from the Interpretation of the Private Securities Litigation Reform Act’s Pleading Standard,” 73 Southern California Law Review 773-810.

Jones, Christopher L., and Seth E. Weingram. 1996a. “The Determinants of 10b-5 Litigation Risk,” Working paper, Stanford Law School.

———. 1996b. “The Effects of Insider Trading, Seasoned Equity Offerings, Corporate Announcements, Accounting Restatements, and SEC Enforcement Actions on 10b-5 Litigation Risk,” Working paper, Stanford Law School.

Kasznik, Ron, and Baruch Lev. 1995. “To Warn or Not To Warn: Management Disclosures in the Face of an Earnings Surprise,” 70 The Accounting Review 113-134.

Johnson 40

Kinney, William R., Jr., and Linda S. McDaniel. 1989. “Characteristics of Firms Correcting Previously Reported Quarterly Earnings,” 11 Journal of Accounting and Economics 71-93.

Lerach, William S. 2001. “An Alarming Decline in the Quality of Financial Reporting,” www.milberg.com. Levitt, Arthur, Jr. 1998. “The Numbers Game,” Remarks delivered at the NYU Conference for Law and Business.

Niehaus, Greg, and Greg Roth. 1999. “Insider Trading, Equity Issues, and CEO Turnover in Firms Subject to Securities Class Actions,” 28 Financial Management 52-72.

Palmrose, Zoe-Vonna, and Susan Scholz. 2004. “The Circumstances and Legal Consequences of Non-GAAP Reporting: Evidence from Restatements,” 21 Contemporary Accounting Research 139-180.

Perino, Michael A. 2003. “Did the Private Securities Litigation Reform Act Work?” 2003 University of Illinois Law Review 913-97.

PricewaterhouseCoopers. 2000. “2000 Securities Litigation Study,” Manuscript

Johnson 41

available at www.10b5.com.

Pritchard, A.C., and Hilary A. Sale. 2005. “What Counts as Fraud? An Empirical Study of Motions to Dismiss Under the Private Securities Litigation Reform Act,” 2 Journal of Empirical Legal Studies 125-149.

Private Securities Litigation Reform Act, 109 Stat. 737 (1995) (codified as amended in scattered sections of 15 U.S.C.)

Sale, Hilary A. 2002. “Judging Heuristics,” 35 University of California at Davis Law Review 903-963.

S. Rep. No. 104-98, 104th Cong., 1st Sess. (1995).

Skinner, Douglas. 1994. “Why Firms Voluntarily Disclose Bad News,” 32 Journal of Accounting Research 38-60.

———. 1997. Earnings Disclosures and Stockholder Lawsuits,” 23 Journal of Accounting and Economics 249-282.

Spiess, Donna K., and Paula Tkac. 1997. “The Private Securities Litigation Reform Act of 1995: The Stock Market Casts its Vote,” 18 Managerial and Decision Economics

Johnson 42

545-561.

Wu, Min. 2002. “Earnings Restatements: A Capital Market Perspective,” Working paper, New York University.

Johnson 43

Table 1 Descriptive Statistics on Lawsuits Panel A: Number of filings and allegations Accounting Allegations Lawsuit Stand Year Lawsuits All Alone Pre-PSLRA: 1991 10 3 (30.0%) 1 1992 8 2 (50.0%) 0 1993 11 3 (27.3%) 0 1994 10 5 (50.0%) 0 1995 11 3 (27.3%) 1 50 16 (32.0%) 2 Post-PSLRA: 1996 11 6 (54.5%) 1 1997 15 8 (53.3%) 0 1998 15 9 (60.0%) 2 1999 15 10 (66.7%) 0 2000 8 6 (75.0%) 0 64 39 (60.9%) 3 Total

114

55 (48.2%)

Panel B: Lawsuit outcomes Outcome Frequency: Dismissed Settled (in millions): < $2 $2 – < $5 $5 – < $10 > $10 Undisclosed or bankrupt Amount (in millions): Mean (including dismissals) Median (including dismissals) Maximum

5

Insider Trading Allegations Stand All Alone

Forecast Allegations Stand All Alone

3 (30.0%) 4 (50.0%) 1 ( 9.0%) 6 (60.0%) 5 (45.5%) 19 (38.0%)

0 0 0 0 1 1

9 ( 90.0%) 8 (100.0%) 10 ( 90.9%) 9 ( 90.0%) 7 ( 63.6%) 43 ( 86.0%)

5 4 7 2 4 22

9 (81.8%) 11 (73.3%) 11 (73.3%) 13 (86.7%) 7 (87.5%) 51 (79.7%)

0 2 1 1 0 4

7 ( 63.6%) 9 ( 60.0%) 10 ( 66.7%) 10 ( 66.7%) 5 ( 62.5%) 41 ( 64.1%)

1 2 1 0 0 4

70 (61.4%)

5

84 ( 73.7%)

26

Pre-PSLRA

Post-PSLRA

18 ( 36.0%)

25 ( 39.1%)

5( 5( 9( 12 ( 31 (

6( 12 ( 7( 13 ( 38 (

16.1%) 16.1%) 29.0%) 38.7%) 62.0%)

15.8%) 31.6%) 18.4%) 34.2%) 59.4%)

1 ( 2.0%) 50 (100.0%)

1 ( 1.6%) 64 (100.0%)

9.26 (5.86) 7.50 (2.30) 55.00

17.99 (10.85) 5.63 (2.00) 259.00

Johnson 44

Table 2 Descriptive Statistics Comparing the pre-PSLRA and post-PSLRA Periods Pre-PSLRA (N = 100) Variable Restatement Abnormal Insider Trading Positive Forecast Negative Forecast Earnings Warning Avg. Tenure Busy Independent Audit Meetings Independent Audit Market Value of Equity Turnover Min. Return

Mean 0.05 –0.01 0.29 0.20 0.48 6.07 1.78 0.60 2.64 0.82 1742.38 0.57 –0.21

Median

Std. Dev.

0.00 0.00 0.00 0.00 0.00 5.50 1.63 0.60 3.00 1.00 247.83 0.63 –0.19

0.22 0.19 0.46 0.40 0.50 4.22 1.12 0.16 1.66 0.25 6699.31 0.32 0.11

Post-PSLRA (N = 128) Mean 0.20 –0.06 0.30 0.18 0.41 5.36 1.47 0.60 2.55 0.83 2294.18 0.67 –0.26

Tests of Differences

Median

Std. Dev.

Mean

Median

0.00 0.00 0.00 0.00 0.00 5.00 1.46 0.60 2.00 1.00 298.67 0.75 –0.25

0.40 0.24 0.46 0.39 0.49 3.49 1.11 0.19 1.92 0.26 7663.67 0.29 0.13

< 0.01 0.12 0.91 0.70 0.32 0.17 0.04 0.81 0.72 0.69 0.16 0.01 < 0.01

< 0.01 0.44 0.91 0.70 0.32 0.18 0.06 0.74 0.41 0.47 0.60 0.01 < 0.01

Variable definitions are in the Appendix. The pre-PSLRA period is 1991-1995 and the post-PSLRA period is 1996-1999.

Johnson 45

Table 3 Comparison of Lawsuit and Control Firms Panel A: Pre-PSLRA period Lawsuit Firms (N = 50) Variable Restatement Abnormal Insider Trading Positive Forecast Negative Forecast Earnings Warning Avg. Tenure Busy Independent Audit Meetings Independent Audit Market Value of Equity Turnover Min. Return

Mean 0.08 –0.01 0.42 0.26 0.82 5.30 2.02 0.61 2.96 0.83 2876.27 0.73 –0.25

Median

Std. Dev.

0.00 0.00 0.00 0.00 1.00 4.83 2.00 0.63 3.00 1.00 749.89 0.83 –0.25

0.27 0.22 0.50 0.44 0.39 3.09 1.05 0.15 1.84 0.25 9237.13 0.26 0.10

Control Firms (N = 50) Mean

Median

0.02 –0.02 0.16 0.14 0.14 6.85 1.53 0.58 2.32 0.80 608.49 0.41 –0.17

0.00 0.00 0.00 0.00 0.00 6.00 1.37 0.60 2.00 1.00 75.48 0.36 –0.15

Tests of Differences

Std. Dev.

Mean

Median

0.14 0.16 0.37 0.35 0.35 5.03 1.16 0.16 1.41 0.26 1652.00 0.31 0.11

0.17 0.85 < 0.01 0.14 < 0.01 0.07 0.03 0.29 0.05 0.53 < 0.01 < 0.01 < 0.01

0.18 0.48 < 0.01 0.14 < 0.01 0.16 0.01 0.38 0.12 0.47 < 0.01 < 0.01 < 0.01

Johnson 46

Table 3 - continued Comparison of Lawsuit and Control Firms Panel B: Post-PSLRA period Lawsuit Firms (N = 64) Variable Restatement Abnormal Insider Trading Positive Forecast Negative Forecast Earnings Warning Avg. Tenure Busy Independent Audit Meetings Independent Audit Market Value of Equity Turnover Min. Return

Mean 0.38 –0.12 0.38 0.20 0.67 5.24 1.72 0.65 3.08 0.88 3888.17 0.79 –0.30

Median 0.00 –0.02 0.00 0.00 1.00 5.25 1.76 0.67 3.00 1.00 807.87 0.88 –0.31

Std. Dev. 0.49 0.28 0.49 0.41 0.47 2.64 1.07 0.18 2.16 0.21 10367.32 0.24 0.15

Control Firms (N =64) Mean

Median

0.02 0.00 0.22 0.15 0.15 5.47 1.22 0.55 2.03 0.78 700.19 0.55 –0.23

0.00 0.00 0.00 0.00 0.00 4.25 1.00 0.59 2.00 1.00 85.75 0.52 –0.20

Tests of Differences

Std. Dev.

Mean

Median

0.13 0.17 0.42 0.37 0.37 4.20 1.10 0.19 1.49 0.30 2398.41 0.29 0.11

< 0.01 < 0.01 0.05 0.49 < 0.01 0.71 0.01 < 0.01 < 0.01 0.03 < 0.01 < 0.01 < 0.01

< 0.01 0.02 0.06 0.50 < 0.01 0.40 < 0.01 < 0.01 < 0.01 0.05 < 0.01 < 0.01 < 0.01

Variable definitions are in the Appendix. The pre-PSLRA period is 1991-1995 and the post-PSLRA period is 1996-1999.

Johnson 47

Table 4 Factors Associated with Lawsuit Filings

Variable

Prediction

Constant Restatement Abnormal Insider Trading Positive Forecast Negative Forecast Earnings Warning Avg. Tenure Busy Independent Audit Meetings Independent Audit Market Cap. Turnover Min. Return

? + – + – + – + – – – + + –

Pseudo R2 N

Pre-PSLRA Coeff. p-value

Post-PSLRA Pre vs. Coeff. p-value Post PSLRA

–5.30 2.42 –2.08 1.03 –0.46 3.32 –0.09 0.60 –1.63 –0.01 0.15 0.49 0.39 0.55

–8.17 4.77 –4.03 0.77 –0.85 2.09 –0.04 0.23 –2.22 0.17 0.60 0.65 1.32 –5.75

< 0.01 0.03 0.10 0.07 0.49 < 0.01 0.27 0.02 0.36 0.98 0.90 < 0.01 0.71 0.82

< 0.01 < 0.01 < 0.01 0.21 0.26 < 0.01 0.65 0.41 0.26 0.27 0.66 < 0.01 0.23 0.01

< 0.01 0.04 0.62 0.53 0.05 0.59 0.10 0.72 0.21 0.72 0.33 0.38 < 0.01

0.70 228

Variable definitions are in the Appendix. The results are for a single multinomial logistic regression that permits different coefficient estimates for firms sued in the pre- and postPSLRA periods, with non-sued firms serving as the reference category. The final column reports p-values for a comparison of the coefficient estimates for the pre- and postPSLRA periods. All p-values are two-tailed. The pre-PSLRA period is 1991-1995 and the post-PSLRA period is 1996-2000.

Johnson 48

Table 5 Factors Associated with Lawsuit Allegations Panel A: Accounting allegations

Variable Constant Restatement Avg. Tenure Busy Independent Audit Meetings Independent Audit Pseudo R2 N

Prediction ? + – + – – –

Pre-PSLRA Coeff. p-value

Post-PSLRA Pre vs. Coeff. p-value Post PSLRA

–2.76 1.85 0.05 0.36 –0.73 0.05 0.96

–3.30 3.12 0.04 –0.23 –0.05 –0.01 2.71

0.06 0.03 0.64 0.19 0.73 0.77 0.55

0.03 < 0.01 0.65 0.38 0.98 0.98 0.07

0.08 0.98 0.05 0.77 0.81 0.35

0.35 114

Panel B: Insider trading allegations

Variable

Prediction

Constant Abnormal Insider Trading Avg. Tenure Busy Independent Audit Meetings Independent Audit

? – – + – – –

Pseudo R2 N

0.22 114

Pre-PSLRA Coeff. p-value

Post-PSLRA Pre vs. Coeff. p-value Post PSLRA

–2.41 0.76 –0.12 0.61 2.36 –0.14 –0.08

–2.08 –2.11 –0.08 0.03 2.98 –0.01 0.67

0.08 0.55 0.25 0.03 0.28 0.37 0.96

0.07 0.02 0.34 0.91 0.07 0.95 0.56

0.02 0.69 0.04 0.78 0.40 0.62

Johnson 49

Table 5 Factors Associated with Lawsuit Allegations Panel C: Forecast allegations

Variable Constant Positive Forecast Negative Forecast Earnings Warning Avg. Tenure Busy Independent Audit Meetings Independent Audit Pseudo R2 N

Prediction ? + – + – + – – –

Pre-PSLRA Coeff. p-value

Post-PSLRA Pre vs. Coeff. p-value Post PSLRA

–0.39 0.40 –0.02 2.29 –0.05 0.34 –2.40 0.02 0.05

0.34 0.19 –0.25 1.16 –0.12 0.05 –2.12 0.10 0.96

0.78 0.49 0.97 < 0.01 0.58 0.22 0.24 0.91 0.97

0.80 0.73 0.71 0.03 0.20 0.85 0.29 0.43 0.51

0.67 0.69 0.09 0.40 0.21 0.86 0.49 0.45

0.21 114

Variable definitions are in the Appendix. The results in each panel are for a single multinomial logistic regression that permits different coefficient estimates for lawsuits containing the specified type of allegation in the pre- and post-PSLRA periods, with lawsuits not containing that allegation serving as the reference category. The final column reports p-values for a comparison of the coefficient estimates for the pre- and post-PSLRA periods. All p-values are two-tailed. The pre-PSLRA period is 1991-1995 and the post-PSLRA period is 1996-2000.

Johnson 50

Table 6 Factors Associated with Settled versus Dismissed Lawsuits

Variable

Prediction

Constant Restatement Abnormal Insider Trading Positive Forecast Negative Forecast Earnings Warning Avg. Tenure Busy Independent Audit Meetings Independent Audit

? + – ? ? ? – + – – –

Pseudo R2 N

Pre-PSLRA Coeff. p-value

Post-PSLRA Pre vs. Coeff. p-value Post PSLRA

–2.06 0.01 –0.65 0.12 1.91 0.34 –0.07 0.16 –2.19 –0.12 2.36

0.17 1.94 –0.38 0.42 1.02 –0.24 –0.07 0.10 –3.92 –0.17 1.51

0.16 0.99 0.52 0.84 < 0.01 0.62 0.47 0.55 0.27 0.43 0.13

0.89 < 0.01 0.70 0.46 0.15 0.68 0.48 0.69 0.05 0.30 0.28

0.01 0.82 0.66 0.23 0.44 0.99 0.84 0.44 0.79 0.62

0.32 112

Variable definitions are in the Appendix. The results are for a single multinomial logistic regression that permits different coefficient estimates for settlements in the pre- and postPSLRA periods, with dismissals serving as the reference category. The final column reports p-values for a comparison of the coefficient estimates for the pre- and postPSLRA periods. All p-values are two-tailed. A lawsuit is classified as a dismissal if it was resolved in the firm’s favor, and includes lawsuits that were dismissed and settled for less than 0.5 percent of the firm’s market value ten days prior to the end of the class period. The pre-PSLRA period is 1991-1995 and the post-PSLRA period is 1996-2000.