Securities Price Consequences of the Private Securities Litigation Reform Act of 1995 and...
I. INTRODUCTION
In response to a perceived litigation explosion, Congress enacted the Private Securities Litigation Reform Act of 1995 (PSLRA), which increases restrictions on private litigants' ability to sue for investment losses from securities fraud. This legislation stimulated
This study examines stock price reactions to events that changed the likelihood of the passage of PSLRA and other related events, to assess whether shareholders considered PSLRA beneficial or harmful. Our analysis focuses on four industries that are most susceptible to securities lawsuits and that consequently are most likely to be affected by the passage of PSLRA. These high-litigation-risk industries are computers, electronics, pharmaceuticals/biotechnology, and retailing (Francis et al. 1994a).
Spiess and Tkac (1997) (hereafter ST) and Johnson et al. (2000) (hereafter JKN) also examine stock price reaction to PSLRA for firms in these industries. They focus on the price reaction to the presidential veto late on the night of 12/19/95 and the congressional votes to override the veto on 12/20/95 (House) and 12/22/95 (Senate). They observe significantly negative abnormal returns for firms in high-litigation-risk industries on 12/18/95, which they attribute to veto rumors, and significantly positive abnormal returns on 12/20/95, which they attribute to the House vote to override the veto. As a result, they conclude that shareholders considered PSLRA beneficial.
We present evidence that is inconsistent with their conclusion. First, we show that conventional statistical procedures overstate the significance of the excess returns on 12/18/95. Second, based on a detailed analysis of press coverage, and results of intertemporal correlation analysis following the market price reversal methodology of Smith (1981) and Collins et al. (1982), we argue that the positive abnormal return observed on 12/20/95 is more likely a response to the presidential veto than to the House override. Third, we undertake a more comprehensive examination of the events during the almost year-long debate on PSLRA and find that significant abnormal returns for events other than the veto suggest that shareholders considered PSLRA harmful. Fourth, we observe significantly negative cumulative abnormal returns during the last three weeks, when the final version of the bill received the congressional approval, presidential veto, and congressional override of the veto, taking the probability of the bill's passage from less than one to one. Fifth, we find a significantly negative stock price reaction to the defeat of an initiative, Proposition 211, intended to reverse many of PSLRA's key provisions (Bencivenga 1996); these abnormal returns are positively correlated with abnormal returns related to the passage of PSLRA. Finally, we find significantly negative abnormal returns to the last securities-litigation-related major judicial or legislative event before the enactment of PSLRA in 1995--the 1994 Supreme Court decision that made it harder for shareholders to sue professional advisors, such as auditors and investment bankers, suspected of willfully aiding and abetting securities violations (Seligman 1994b). Although the timing of the presidential veto and the House vote to override the veto make it difficult to unambiguously interpret the returns on 12/20/95, the entire body of our results is more consistent with the inference that, on average, shareholders in the high-litigation-risk industries considered PSLRA harmful.
There is little evidence in the literature on the effect on shareholders of changes in the U.S. securities-litigation environment. Our study (along with ST and JKN) addresses this void by documenting stock price consequences of PSLRA and other related events. Other studies have attempted to document the effect of PSLRA on shareholders by comparing patterns of securities litigation before and after the Act (e.g., Grundfest and Perino 1997; SEC 1997a). However, such an approach requires several years' data on actual cases and their outcomes under PSLRA to draw reliable conclusions about the Act's effect. Moreover, such an approach requires a subjective determination of the merit of lawsuits because PSLRA benefits shareholders if nonmeritorious lawsuits are curbed to a greater extent than are meritorious ones. The event-study methodology we use avoids these difficulties by providing a more timely and objective assessment of the Act's effect on shareholders.
We have organized the rest of the paper as follows. Section II describes the PSLRA, its effect on shareholders, and events leading to the passage of the Act. Section III explains our research design, and Section IV discusses the results. Section V discusses other events related to PSLRA--Proposition 211 (which sought to overturn key provisions of PSLRA) and Supreme Court decisions related to securities litigation. Section VI shows that our results are robust to using a much broader sample with a continuous litigation-risk measure based on the Jones and Weingram (1996) model. Section VII provides concluding remarks.
II. PRIVATE SECURITIES LITIGATION REFORM ACT
After paying big settlements in cases related to the savings and loan crisis, accountants lobbied for tort reform, and by the early 1990s there was a widespread belief that the litigation explosion had gotten out of hand. Bills introduced in Congress led to the Private Securities Litigation Reform Act of 1995, which contains several provisions (Hamilton 1996). First, the Act makes it more difficult to file frivolous lawsuits by requiring plaintiffs to plead specific facts alleging that management intentionally committed fraud, and it imposes penalties (normally the payment of the other parties' attorneys' fees) on plaintiffs if the lawsuit is found to be frivolous. Second, the Act provides companies a safe harbor for forward-looking disclosures that are accompanied by "meaningful cautionary statements." Third, it moves toward proportional liability instead of joint and several liability for defendants not found to have knowingly violated securities laws (King and Schwartz 1997). This limits the liability of deep-pocket defendants, such as auditors and underwriters, who previously had been liable for damages up to the full amount of the settlement or judgment even if they were responsible for only a small portion of the blame. Finally, the Act allows large shareholders to petition the court to be appointed as lead plaintiff, which is intended to reduce plaintiff lawyers' ability to file frivolous lawsuits for their own gain.
Effect of the Act on Shareholders
Most large securities-fraud cases are class actions filed by lawyers on behalf of investors who bought shares from other investors when the price was, allegedly, artificially high due to misstatements by the company. The sellers reap an undeserved windfall, but cannot be sued because they are innocent. So the injured investors sue the company and its officers and professional advisors who are responsible for the misleading statements. There is a considerable debate whether PSLRA, which restricts private litigants' ability to sue for securities fraud, benefits or hurts shareholders (Avery 1996).
Dead-Weight Losses of Litigation
Proponents of the Act argue that much of the pre-PSLRA litigation was frivolous. According to this view, legal defense costs were high, and it was in the defendants' interest to settle rather than litigate. Therefore, plaintiff lawyers filed nuisance lawsuits simply to receive a settlement, regardless of the actual merits of the case. The company's shareholders bore the cost of securing insurance coverage against losses from such lawsuits, and the costs of defense and settlement not reimbursed by insurers. Shareholders also suffered other adverse effects of the litigation, as managers diverted attention from other issues, companies were less willing to enter industries where the potential for lawsuits was high, and directors, auditors, and professional advisors were hesitant to associate themselves with such industries. The threat of such litigation may have also hurt shareholders by making managers reluctant to disclose forward-looking information (Avery 1996). The dead-weight losses of frivolous litigation suggest that high-litigation-risk firms' shareholders would react positively to PSLRA.
Deterrence Effect of Litigation
Opponents of PSLRA argued that the courts could dismiss and sanction nonmeritorious securities litigation even under the old law (Seligman 1994a). PSLRA's new restrictions may discourage plaintiffs from filing even meritorious suits. President Clinton also expressed a similar concern. His veto message stated that the bill would "have the effect of closing the courthouse doors on investors who have legitimate claims" (Hamilton 1996, 701).(1) Opponents of PSLRA are concerned that the law may provide too much immunity to corporate officers and professional advisors, reducing their incentives to disclose timely and accurate information to shareholders.(2) The resulting information uncertainty would harm shareholders (Lev 1995). This view suggests that high-litigation-risk firms' shareholders would react negatively to PSLRA.
Unexpected Loss of Insurance Value
PSLRA would have reduced shareholders' expected recovery of post-PSLRA investment losses from outside professionals, and from directors' and officers' liability-insurance providers. Firms may have renegotiated fees/premiums paid to these parties to reflect the lower likelihood of lawsuits due to PSLRA. However, to the extent that these fees/premiums are not completely adjusted, these parties would benefit at the expense of the shareholders, suggesting that high-litigation-risk firms' shareholders would react negatively to PSLRA.(3)
Events Leading to the Reform Act
The main difficulty with measuring the effects of legislative events on security prices is identifying when the market first anticipates the effects of the events (Binder 1985). The road to the Private Securities Litigation Reform Act of 1995 was "long and winding" (Avery 1996). The capital market, during the almost year-long legislative debate, would have anticipated the effects of most of the legislative events before they occurred. To mitigate this measurement problem, we follow the Schipper and Thompson (1983) strategy of considering legislative events that are likely to have significantly altered expectations about the bill's passage. For PSLRA, these events include the three sets of full House and Senate votes, and the presidential veto. Table 1 lists these events, and indicates whether they increased or decreased the likelihood of the passage of PSLRA. Unfortunately, news of the veto and the House override of the veto reached the market on the same day (12/ 20/95), and the two events have opposite effects on the likelihood of PSLRA's passage. We interpret subsequent analysis of the financial press reports as indicating that the veto was more surprising than was the House vote. Thus, for ease of exposition, the prediction listed in Table 1 is based on the assumption that the market reaction on 12/20/95 is due primarily to the news of the veto, although in light of the ambiguity induced by the confounding events, we place a question mark next to it.
TABLE 1
Results from Ordinary Least Squares Regression of Daily Portfolio
Returns for High-Litigation-Risk Firms on Market Index and Dummy
Variables for Event Dates Leading to the Enactment of PSLRA
Estimation Period is the 252 Trading Days in the Year 1995(a)
Model: [R.sub.t] = [a.sub.0] + [a.sub.1][R.sub.mt] + [Sigma][a.sub.2i]
[D.sub.i] + [u.sub.t](b)
[a.sub.2i]
Event Predic- (t-stat., Randomization
date, i tion(c) Coeff. p-value) p-value(d)
Intercept 0.07 (2.43, 0.01)(***)
[R.sub.mt] 1.01 (16.69, 0.00)(***)
1/4/95 0.02 (0.06, 0.95) 0.74
1/18/95 0.28 (0.67, 0.50) 0.45
1/25/95 -0.70 (-1.65, 0.10) 0.05(**)
2/9/95 0.93 (2.17, 0.03) 0.04(**)
2/14/95 0.07 (0.17, 0.87) 0.83
2/15/95 0.02 (0.05, 0.96) 0.81
2/16/95 0.09 (0.20, 0.84) 0.94
3/8/95 I 0.00 (0.01, 0.99) 0.99
5/19/95 -0.06 (-0.14, 0.89) 0.79
5/24/95 -0.12 (-0.29, 0.77) 0.69
5/25/95 0.14 (0.34, 0.73) 0.76
6/28/95 I -0.58 (-1.37, 0.17) 0.10(*)
10/24/95 -0.20 (-0.46, 0.65) 0.55
11/16/95 -0.23 (-0.54, 0.59) 0.41
11/28/95 0.06 (0.14, 0.89) 0.94
12/5/95 I -0.71 (-1.66, 0.10) 0.08(*)
12/6/95 I -0.61 (-1.44, 0.15) 0.09(*)
12/11/95 -0.71 (-1.66, 0.10) 0.05(**)
12/18/95 -1.38 (-3.14, 0.00) 0.16
12/20/95 D (?) 1.12 (2.62, 0.01) 0.02(**)
12/21/95 -0.06 (-0.13, 0.90) 0.78
12/22/95 I 0.39 (0.91, 0.36) 0.20
Adj. 0.57
[R.sup.2]
Event
date, i Event Description (Source)(e)
Intercept
[R.sub.mt]
1/4/95 Common Sense Legal Reform Act introduced in the House.
(CPR)
1/18/95 Private Securities Litigation Reform Act of 1995
introduced in the Senate. (CPR)
1/25/95 SEC Chief warns legislators that changes envisaged by
Republicans would impair the nation's securities
regulation system. (NYT)
2/9/95 Legislators modify bill to meet some of SEC's objections.
(WP)
2/14/95 Finance Subcommittee passes the bill, 16-10. (CPR)
2/15/95 SEC rejects revised bill. (WP)
2/16/95 House Commerce Committee passes the bill, 32-10. (CPR)
3/8/95 House passes its bill, 325-99. (WSJ)
5/19/95 SEC asks Senator D'Amato, Chairman, Senate Banking
Committee, to amend the safe harbor provision in the
House bill, which would have given sweeping protection
to companies. (WSJ)
5/24/95 Senator D'Amato complies. (WSJ)
5/25/95 Senate Banking Committee approves the bill, 11-4. (WSJ)
6/28/95 Senate passes its bill, 70-29. (WSJ)
10/24/95 House and Senate staffers hammer out a compromise. (WSJ)
11/16/95 SEC approves safe harbor provisions, making a
presidential veto less likely. (WSJ)
11/28/95 Conference Committee finalizes the bill.
12/5/95 Senate approves the bill, 65-30. Clinton is undecided.
(WSJ)
12/6/95 House approves the bill, 320-102. (WSJ)
12/11/95 President indicates willingness to sign the bill. (WSJ)
12/18/95 Clinton is officially undecided. Rumors are that he will
veto the bill. (National Journal's Congress Daily)
12/20/95 President vetoes bill (on 12/19/95 after market close),
and House overrides veto, 319-100. (WSJ)
12/21/95 Senate poised to override the veto. (WSJ)
12/22/95 Senate overrides veto, 68-30. (WSJ)
Adj.
[R.sup.2]
(*), (**), (***) indicate significance at 10, 5, and 1 percent level or
better, respectively (two-tailed test).
(a) The portfolio of high-litigation-risk firms on a given day consists
of all firms in the following four industries for which daily returns
data are available: computers (SIC codes 3570-3577 and 7370-7374),
electronics (SIC codes 3600-3674), pharmaceuticals/biotechnology (SIC
codes 2833-2836 and 8731-8734), and retailing (SIC codes 5200-5961).
The number of firms in the portfolio ranges from 1,443 to 1,589.
(b) Variable definitions:
[R.sub.t] = high-litigation-risk portfolio daily stock return in
percentage;
[R.sub.mt] = CRSP daily value-weighted market return in percentage; and
[D.sub.i] = 1 if the day corresponds to event date i, and 0 otherwise.
(c) Predictions: Increase (I) or decrease (D) in the likelihood of the
passage of PSLRA.
(d) Randomization p-values are based on the distribution of abnormal
returns for 100 random days with market returns within plus or minus
0.05 percent of that of the test day; the exception is 12/18/95 (see
footnote 10).
(e) Sources: CPR: Congressional Press Releases; WSJ: Wall Street
Journal; NYT: New York Times; WP: Washington Post.
To provide a more comprehensive analysis, we also examine abnormal returns for several other events during the legislative debate on PSLRA: the introduction of the bills in the Congress; congressional committee and subcommittee votes; conference committee deliberations; the Securities and Exchange Commission's (SEC's) position on the legislation; President Clinton's indicated willingness to sign the bill; and the rumor of presidential veto. Since for many of these events it is difficult to assess whether the event increased or decreased the likelihood of the passage of PSLRA, we do not make predictions for this set of events. Table 1 also lists these events.(4)
Republicans won control of Congress in 1994, and soon thereafter introduced the Common Sense Legal Reform Act in the House, and the Private Securities Litigation Reform Act of 1995 in the Senate. SEC Chairman Arthur Levitt warned that changes envisaged by Republicans would impair the nation's securities regulation system by undercutting investors' right to sue companies that inflate the value of their stock (Avery 1996). Both the House and the Senate amended the bills to address some of Levitt's concerns. On 3/8/95 the House approved its bill with a vote of 325-99, and on 6/28/95 the Senate approved its bill with a vote of 70-29. The House and Senate subsequently reconciled the differences in the two bills adopted by their respective chambers, mostly to conform to the more moderate Senate version. On 12/5/95, the Senate approved the bill, 65-30, and the House approved the bill the next day with a vote of 320-102.
The President had until December 19 to sign or veto the bill. Although Congress did not completely address the administration's objections in the bill, it passed; a few days after the House and Senate votes, the president "privately" conveyed his intention to sign the bill (Frisby and Taylor 1995a). However, on the night of December 19, President Clinton vetoed the bill, indicating that he disapproved of certain provisions. The stock market reacted to the veto on December 20. On the same day--December 20--as could be expected given the earlier lopsided vote on the bill, the House voted 319-100 to override the veto. The Appendix provides quotes from several newspaper articles suggesting that President Clinton's veto on the night of December 19 was more surprising than was the House vote to override the veto on December 20. Thus, in our opinion, the stock price reaction of high-litigation-risk firms on December 20 is more likely attributable to the veto than to the House vote.(5) It was not certain, however, that the Senate would be able to override the veto (Frisby and Taylor 1995b). The following day, President Clinton lobbied intensely to defeat the bill, but with little success (Frisby and Taylor 1995c). On December 22, the Senate overrode the veto 68-30, a margin of only two votes.
III. RESEARCH DESIGN
To examine shareholders' response to PSLRA, we estimate, for PSLRA event dates, abnormal returns for a portfolio of firms in four industries that Francis et al. (1994a) assert experienced relatively high incidence of litigation during 1988-1992: computers (SIC codes 3570-3577 and 7370-7374), electronics (SIC codes 3600-3674), pharmaceuticals/biotechnology (SIC codes 2833-2836 and 8731-8734), and retailing (SIC codes 5200-5961).(6) Following Ryngaert and Netter (1988) and Karpoff and Malatesta (1989), we employ the following model to compute abnormal returns:
(1) [R.sub.t] = [a.sub.0] + [a.sub.1][R.sub.mt] + [Sigma][a.sub.2i][D.sub.i] + [u.sub.t],
where [R.sub.t] is the daily return of a portfolio of high-litigation-risk firms, [R.sub.mt] is the value-weighted market index, and [D.sub.i] is a dummy variable that takes a value of 1 when the day corresponds to event date i, and 0 otherwise. The coefficient [a.sub.2i] represents the average abnormal return of high-litigation-risk firms on event date i. Since all our PSLRA events occur in 1995, we estimate equation (1) using daily-return data for 1995. The portfolio of high-litigation-risk firms on a given day consists of all firms in the four industries for which daily returns data are available on the CRSP database. Because of missing returns data, the number of firms in the daily portfolios is not the same; the minimum (maximum) number of firms is 1,443 (1,589). We also estimate equation (1) separately for each of the four industries. The minimum (maximum) numbers of firms in the computers, electronics, pharmaceuticals/biotechnology, and retailing portfolios are 492 (579), 430 (484), 74 (79), and 441 (450), respectively.
Jain (1986) shows that event studies in which firms have the same event date are susceptible to overstated (understated) significance levels when the absolute value of market returns on the event date is high (low). To control for this bias, we conduct a randomization test (Noreen 1989). For each of the event dates, we randomly select 100 dates during the 1976-1995 period with market returns within 0.05 percent of the returns on the test dates. For each of the random dates, we obtain abnormal returns by estimating equation (1). We use daily returns for 250 days preceding the selected random date for estimation purposes. We estimate the significance level of the abnormal return on the test date based on the distribution of the estimated abnormal returns for the 100 random days. Specifically, the two-tailed p-value is the proportion of random days on which the absolute value of the abnormal return exceeds the abnormal return we observe on the test date.
Noreen (1989, Appendix 2A) also demonstrates that the randomization test is just as powerful as the conventional parametric test. Consequently, we base our conclusions about the significance of daily abnormal returns on the randomization p-values, although we also report significance levels from conventional t-tests for comparability with other studies.(7)
IV. RESULTS
Regression Results
Table 1 reports OLS estimates of equation (1) for the portfolio of all high-litigation-risk firms. Of the six dates for which we predict whether the event will increase or decrease the likelihood of passage of PSLRA, abnormal returns are significantly different from zero at better than the 10 percent level (two-tailed test based on the randomization p-values) on four dates. Abnormal return on 6/28/95 (Senate passes its bill) is -0.58 percent (p = 0.10); on 12/5/95 (Senate approves the conference committee bill) is -0.71 percent (p = 0.08); on 12/6/95 (House approves the conference committee bill) is -0.61 percent (p = 0.09); and on 12/20/95 (President vetoes bill and House overrides veto) is 1.12 percent (p = 0.02).(8) The first three dates unambiguously increase probability of passage, and market reacted negatively, inconsistent with ST's and JKN's interpretations. Although we regard 12/20/95 as decreasing the probability of passage (for reasons spelled out later), interpretation of the 12/20/95 return itself is inconclusive because of the two confounding events on this day. Although we regard 12/20/95 as decreasing the probability of passage (for reasons spelled out later), interpretation of the 12/20/95 return itself is inconclusive because of the two confounding events on this day.
Abnormal returns for the portfolio of high-litigation-risk firms are significant for three other dates. On 1/25/95, the SEC chairman indicated that some of the proposed changes to the law would make it too difficult for private litigants to sue, and would "impair the nation's securities regulation system" (Henriques 1995b). The negative abnormal return, -0.70 percent (p = 0.05), suggests that shareholders reacted adversely to the SEC's criticism. The abnormal return on 2/9/95, when the House legislators watered down the bill to meet some of the SEC's objections, is 0.93 percent (p = 0.04). This positive market reaction is consistent with shareholders' preference for a bill that would make it less difficult to sue for fraud. The abnormal return on 12/11/95, when President Clinton indicated his willingness to sign the bill, is -0.71 percent (p = 0.05), consistent with shareholders' unfavorable view of PSLRA.
For each of the dates for which the full sample of high-litigation-risk firms exhibits a coefficient that is significantly different from zero in Table 1, Table 2 reports abnormal returns for each industry sample separately. Of 28 coefficients (seven event dates times four industry portfolios), signs of 26 coefficients are consistent with the signs of the coefficients observed for the full sample. The consistency across the four high-litigation-risk industries provides confidence that the results stem from the litigation-related events.
TABLE 2
For Each Industry in High-Litigation-Risk Category, Results from
Ordinary Least Squares Regression of Daily Portfolio Returns on
Market Index and Dummy Variables for PSLRA Event Dates with
Significant Abnormal Returns in Table 1 Estimation Period is the 252
Trading Days in the Year 1995(a)
Model: [R.sub.t] = [a.sub.0] + [a.sub.1][R.sub.mt] + [Sigma][a.sub.2i]
[D.sub.i] + [u.sub.t](b)
Computers
Independent Coefficient
Event Variable or (t-stat, Randomization
Description Event Date p-value) p-value(c)
Intercept 0.09
(2.41, 0.02)(**)
[R.sub.mt] 1.24
(15.54, 0.00)(***)
SEC cautions 1/25/95 -1.11
against Bill (-1.96, 0.05) 0.01(***)
Bill revised 2/9/95 0.95
to suit SEC (1.69, 0.09) 0.10(*)
Senate passes 6/28/95 -0.66
its Bill (-1.17, 0.24) 0.18
Senate passes 12/5/95 -0.80
revised Bill (-1.43, 0.15) 0.12
House passes 12/6/95 -0.69
revised Bill (-1.22, 0.22) 0.12
President 12/11/95 -0.67
willing to sign (-1.19, 0.23) 0.17
Veto/House 12/20/95 1.41
Override (2.50, 0.01) 0.01(***)
Adj. 0.54
[R.sup.2]
Electronics
Independent Coefficient
Event Variable or (t-stat, Randomization
Description Event Date p-value) p-value(c)
Intercept 0.08
(2.05, 0.04)(**)
[R.sub.mt] 1.21
(15.15, 0.00)(***)
SEC cautions 1/25/95 -0.63
against Bill (-1.12, 0.26) 0.14
Bill revised 2/9/95 1.23
to suit SEC (2.19, 0.03) 0.03(**)
Senate passes 6/28/95 -1.11
its Bill (-1.99, 0.05) 0.02(**)
Senate passes 12/5/95 -1.13
revised Bill (-2.01, 0.04) 0.03(**)
House passes 12/6/95 -0.70
revised Bill (-1.24, 0.21) 0.15
President 12/11/95 -0.54
willing to sign (-0.96, 0.34) 0.22
Veto/House 12/20/95 0.99
Override (1.76, 0.08) 0.05(**)
Adj. 0.50
[R.sup.2]
Pharmaceuticals/Biotech
Independent Coefficient
Event Variable or (t-stat, Randomization
Description Event Date p-value) p-value(c)
Intercept 0.09
(2.33, 0.02)(**)
[R.sub.mt] 0.61
(7.21, 0.00)(***)
SEC cautions 1/25/95 -0.05
against Bill (-0.08, 0.94) 0.85
Bill revised 2/9/95 -0.01
to suit SEC (-0.01, 0.99) 0.99
Senate passes 6/28/95 -0.26
its Bill (-0.44, 0.66) 0.59
Senate passes 12/5/95 0.27
revised Bill (0.46, 0.65) 0.61
House passes 12/6/95 -0.40
revised Bill (-0.68, 0.50) 0.34
President 12/11/95 -0.26
willing to sign (-0.43, 0.67) 0.65
Veto/House 12/20/95 1.25
Override (2.10, 0.04) 0.02(**)
Adj. 0.24
[R.sup.2]
Retailing
Independent Coefficient
Event Variable or (t-stat, Randomization
Description Event Date p-value) p-value(c)
Intercept 0.03
(1.24, 0.21)
[R.sub.mt] 0.60
(10.72, 0.00)(***)
SEC cautions 1/25/95 -0.44
against Bill (-1.11, 0.27) 0.14
Bill revised 2/9/95 0.77
to suit SEC (1.96, 0.05) 0.07(*)
Senate passes 6/28/95 -0.01
its Bill (-0.03, 0.98) 0.98
Senate passes 12/5/95 -0.31
revised Bill (-0.80, 0.42) 0.30
House passes 12/6/95 -0.47
revised Bill (-1.20, 0.23) 0.17
President 12/11/95 -1.02
willing to sign (-2.60, 0.01) 0.01(***)
Veto/House 12/20/95 0.87
Override (2.20, 0.03) 0.06(**)
Adj. 0.35
[R.sup.2]
(*), (**), (***) indicate significance at 10, 5, and 1 percent level
or better, respectively (two-tailed test).
(a) An industry portfolio on a given day consists of all firms for
which daily returns data are available. The industry composition is as
follows: computers (SIC codes 3570-3577 and 7370-7374), electronics
(SIC codes 3600-3674), pharmaceuticals/biotechnology (SIC codes
2833-2836 and 8731-8734), and retailing (SIC codes 5200-5961). The
number of firms in the computer, electronics, pharmaceutical/
biotechnology, and retailing industry portfolios ranges from 492 to
579, 430 to 484, 74 to 79, and 441 to 450, respectively.
(b) Variable definitions:
[R.sub.t] = industry portfolio daily stock return in percentage;
[R.sub.mt] = CRSP daily value-weighted market return in percentage;
and
[D.sub.i] = 1 if the day corresponds to event date i, and 0 otherwise.
(c) Randomization p-values are based on the distribution of abnormal
returns for 100 random days with market returns within plus or minus
0.05 percent of that of the test day.
Alternative Interpretation of the Results
ST and JKN argue that on December 18 there was a negative stock price reaction due to a rumor that President Clinton would veto the bill. They further argue that the positive stock price reaction on December 20 is due primarily to the House vote to override the veto. In contrast to ST and our study, JKN report a significantly positive abnormal return on December 22, attributing it to the Senate override of the veto. ST and JKN conclude that shareholders of high-litigation-risk firms reacted favorably to the passage of PSLRA. We reexamine their evidence and conclusion.
December 18 Abnormal Returns
ST and JKN conclude that abnormal returns on December 18 are significantly negative based on p-values from a conventional t-test, and our Table 1 also reports that a conventional p-value for December 18 suggests statistical significance. However, Jain (1986) shows that the significance level of market-model-based abnormal returns is likely to be overstated when market return on the event date is large (in absolute value)--and the absolute market return on December 18 is relatively large, at -1.7 percent.(9) Table 1 shows that the randomization p-value for the abnormal return of high-litigation-risk firms is 0.16.(10) This result suggests that we cannot reliably conclude that a PSLRA-related event affected stock prices on 12/18/95, which is inconsistent with ST's and JKN's inference that high-litigation-risk firms experienced significantly negative abnormal returns on December 18 due to the veto rumor.
December 20 Abnormal Returns
ST and JKN refer to two newspaper articles on the veto rumor on 12/18/95, and conclude that the House vote on 12/20/95 was more surprising than was the veto on 12/ 19/95 (see part IIA of the Appendix). However, other newspaper articles suggest that the market may not have considered the veto rumor on 12/18/95 to be credible (see part IIB of the Appendix). On the whole, we interpret the body of evidence from the press coverage detailed in the Appendix as suggesting that the veto on the night of December 19 was more surprising than was the House vote on December 20. Thus, we argue that the stock price reaction of high-litigation-risk firms on December 20 is more likely a response to the presidential veto than to the House override vote.
To provide empirical evidence that can help distinguish between our interpretation and that of ST/JKN, we conduct price reversal tests.(11) We examine correlations between abnormal returns on 12/18/95 (the alleged veto rumor date) and 12/20/95 (the veto and House override), as well as the correlation of abnormal returns on each of these dates with the abnormal return on 12/11/95, when President Clinton conveyed his intention to sign the bill (Frisby and Taylor 1995a), and the abnormal return on 12/22/95, when the Senate voted to override the veto. Table 3 reports the expected signs of the correlations under ST's and JKN's vs. our inference for the abnormal returns on 12/18/95 and 12/20/95. The results are consistent with our inference. The abnormal returns on 12/18/95 (the alleged veto rumor date) are not significantly correlated with the abnormal returns on any of the other three dates, suggesting that a PSLRA-related event is unlikely to have significantly affected stock prices on 12/18/95. The abnormal return on 12/20/95 (the veto and House override) is significantly negatively correlated with the abnormal returns on 12/11/95 (President conveys intent to sign the bill) (-0.08, p = 0.00)(12) and 12/22/95 (Senate override) (-0.04, p = 0.08), suggesting that stock price reaction on 12/20/95 is more likely a response to the veto than to the House override vote.(13)
TABLE 3
Spearman Correlations of Abnormal Returns of Firms in
High-Litigation-Risk Industries(a) on the Possible Veto
News Dates (12/18/95 and 12/20/95) with Abnormal Returns
on Other Veto-Related Event Dates
Veto Rumor 12/18/95
Predicted Sign
ST's This
and JKN's Study's Correla-
Interpre- Interpre- tion(b)
Event Date Description tation tation (p-value)
12/11/95 President willing -- no +0.03
to sign Bill prediction (0.18)
12/18/95 Veto rumor
12/20/95 Veto/House -- no +0.00
override prediction (0.94)
12/22/95 Senate override -- no -0.01
prediction (0.59)
Veto/House Override 12/20/95
Predicted Sign
ST's This
and JKN's Study's Correla-
Interpre- Interpre- tion(b)
Event Date Description tation tation (p-value)
12/11/95 President willing + -- -0.08
to sign Bill (0.00)(***)
12/18/95 Veto rumor -- no +0.00
prediction (0.94)
12/20/95 Veto/House
override
12/22/95 Senate override + -- -0.04
(0.08)(*)
(*), (**), (***) indicate significance at 10, 5, and 1 percent level
or better, respectively (two-tailed test).
(a) The number of sample firms on a given day include all firms in the
following four industries for which daily returns data are available:
computers (SIC codes 3570-3577 and 7370-7374), electronics (SIC codes
3600-3674), pharmaceuticals/biotechnology (SIC codes 2833-2836 and
8731-8734), and retailing (SIC codes 5200-5961). The sample size used
for computing the correlations ranges from 1,587 to 1,589.
(b) Two-tailed p-values in parentheses.
December 22 Abnormal Returns
JKN (Table 1) report a significantly positive abnormal return on 12/22/95 for their full sample of high-litigation-risk firms (1.33 percent, p = 0.01), and conclude that the market reacted positively to the Senate vote to override the veto. ST (Table 2) do not obtain a significant result for this date, 0.48 percent (p-value not reported, but insignificant at the 10 percent level), nor do we, 0.39 percent (p = 0.20). JKN's result is driven by an abnormal return for their pharmaceutical/biotechnology industry sample (2.82 percent, p = 0.00). For this industry, however, ST (Table 2) report insignificant abnormal returns, 0.23 percent (p-value not reported), as do we, 0.68 percent (p = 0.29).
JKN's pharmaceuticals/biotechnology sample has 191 firms, ST's has 65 firms, and ours has 74 to 79 firms. The main reason for the difference in the sample sizes is that unlike ST's and our samples, JKN's sample includes firms on CRSP with SIC code 2830.(14) CRSP has not assigned these firms to one of the specific four-digit SIC codes starting with 283: 2833, 2834, 2835, and 2836 (1987 SIC manual). Accordingly, the firms in JKN's pharmaceuticals/biotechnology sample should have litigation-risk characteristics similar to those with SIC codes 2833-2836 in ST's and our pharmaceuticals/biotechnology samples.
However, JKN observe a significant abnormal return on 12/22/95 only in their pharmaceuticals/biotechnology sample (and not across all of their high-litigation-risk industries), and even this appears to be attributable to a subset of the pharmaceuticals/biotechnology sample (because neither ST nor we observe a significant abnormal return for our pharmaceuticals/biotechnology samples on 12/22/95). Thus, evidence that the market reacted positively to the Senate vote is mixed at best.(15)
Even though the financial press expected that the House would easily override the veto, it suggested that there was uncertainty about the Senate override (see the Appendix). If the positive abnormal return on 12/20/95 is due to the veto, then one might question why the abnormal return on 12/22/95, when the Senate voted to override the veto, is not significantly negative.(16) One explanation is that the Senate's veto-override process watered down a key provision. In his veto message, President Clinton indicated that the bill sent to him submitted plaintiffs to a pleading requirement that was more severe than the Second Circuit standard for pleading. The Second Circuit standard considers reckless conduct sufficient for pleading, and the plaintiff is not required to show that the defendant acted willfully. While urging an override of the President's veto, Senator Dodd (one of the bill's sponsors) said on the floor of the Senate that, contrary to the President's interpretation, the bill does not depart from the Second Circuit standard for pleading (141 Congressional Record S19, 068 [1995]). Senator Domenici reiterated this point (141 Congressional Record S19, 150 [1995]). Thus, when the Senate overrode the veto, the bill's legislative history supported a more lenient standard for pleading. An SEC brief, two court rulings, and a statement by President Clinton are consistent with this interpretation. Specifically, the SEC filed a brief in a recent lawsuit to indicate that the legislative history weakening the bill during the veto override should be considered in the judicial decision (SEC 1997b). Also, at least two court rulings have relied on Senator Dodd's statement to interpret the pleading standard in PSLRA (Zeid v. Kimberly and Epstein v. Itron). Finally, President Clinton also says: "(A)fter my veto, the bill's supporters made it clear that they did in fact intend to codify the Second Circuit standard. After this important assurance, the bill passed over my veto" (U.S. Newswire 1998).
Cumulative Returns during December 1995
We conduct an alternative analysis that does not require (a necessarily subjective) identification of specific days on which value-relevant information regarding the increase or decrease in the probability of the passage of PSLRA is disclosed. Specifically, we examine the high-litigation-risk firms' cumulative abnormal returns (CAR) from the day before the congressional vote on the conference committee bill (12/4/95) to the next trading day after its ultimate passage (12/26/95). We include 12/26/95 in this analysis because due to light trading on 12/22/95, stock prices may not have fully impounded the news of the Senate vote until 12/26/95 (see footnote 16). During the 12/4/95 to 12/26/95 period, the probability that PSLRA would pass increased from less than 1.0 to precisely 1.0. Positive (negative) CAR would indicate that shareholders viewed PSLRA as beneficial (not beneficial). Table 4 reports CAR for the full sample of high-litigation-risk firms as well as separately for each of the high-litigation-risk industry portfolios. For the full sample, CAR is negative and significant, -3.48 percent (t = -2.00). CAR is also negative for three of the industry portfolios, although it is significant only for electronics, -4.02 percent (t = -1.80) and retailing, -4.04 percent (t = -2.54). Even if we exclude 12/26/95, CAR for the full sample remains significantly negative, -2.93 percent (t = -1.74). CAR also remains significantly negative for electronics and retailing firms. These results suggest that shareholders viewed PSLRA as not beneficial.(17)
TABLE 4
Cumulative Abnormal Returns for Portfolios of High-Litigation-Risk
Firms from the Trading Day before the Congressional Vote on the
Conference Committee Bill (12/4/95) to the Trading Day after the
Veto Override (12/26/95)(a)
All High-
Risk Firms Computers Electronics
Date CAR t-stat CAR t-stat CAR t-stat
12/4/95 -0.43 -1.00 -0.22 -0.39 -0.55 -0.98
12/5/95 -1.14 -1.85 -1.03 -1.27 -1.67 -2.12
12/6/95 -1.75 -2.32 -1.71 -1.72 -2.36 -2.44
12/7/95 -1.59 -1.82 -1.52 -1.32 -2.06 -1.85
12/8/95 -1.44 -1.48 -1.39 -1.08 -1.74 -1.39
12/11/95 -2.14 -2.01 -2.05 -1.46 -2.27 -1.66
12/12/95 -2.54 -2.20 -2.32 -1.52 -2.96 -2.01
12/13/95 -2.65 -2.15 -2.52 -1.55 -3.05 -1.93
12/14/95 -2.50 -1.91 -2.24 -1.29 -3.03 -1.81
12/15/95 -3.07 -2.23 -2.99 -1.64 -4.05 -2.30
12/18/95 -4.41 -3.06 -4.96 -2.60 -5.00 -2.71
12/19/95 -4.40 -2.92 -4.65 -2.34 -4.92 -2.56
12/20/95 -3.26 -2.08 -3.20 -1.55 -3.92 -1.96
12/21/95 -3.32 -2.04 -3.27 -1.52 -3.97 -1.91
12/22/95 -2.93 -1.74 -2.76 -1.24 -3.77 -1.75
12/26/95 -3.48 -2.00(**) -3.44 -1.50 -4.02 -1.80(*)
Pharma-
ceuticals/
Biotech Retailing
Date CAR t-stat CAR t-stat
12/4/95 0.04 0.06 -0.68 -1.71
12/5/95 0.28 0.33 -0.99 -1.75
12/6/95 -0.14 -0.14 -1.45 -2.11
12/7/95 0.15 0.13 -1.48 -1.86
12/8/95 0.17 0.13 -1.49 -1.67
12/11/95 -0.11 -0.07 -2.51 -2.57
12/12/95 0.37 0.23 -2.88 -2.73
12/13/95 0.06 0.04 -2.88 -2.55
12/14/95 -0.45 -0.25 -2.67 -2.23
12/15/95 -0.19 -0.10 -2.65 -2.10
12/18/95 -1.72 -0.87 -3.53 -2.68
12/19/95 -1.47 -0.71 -4.04 -2.93
12/20/95 -0.20 -0.09 -3.16 -2.20
12/21/95 0.86 0.38 -3.45 -2.32
12/22/95 1.51 0.65 -3.06 -1.98
12/26/95 2.51 1.05 -4.04 -2.54(**)
(*), (**) indicate significance at 10 and 5 percent level or better
(two-tailed test).
(a) The portfolio of high-litigation-risk firms on a given day
consists of all firms in the following four industries for which
daily returns data are available: computers (SIC codes 3570-3577 and
7370-7374), electronics (SIC codes 3600-3674), pharmaceuticals/
biotechnology (SIC codes 2833-2836 and 8731-8734), and retailing
(SIC codes 5200-5961). The number of firms in the portfolio ranges
from 1,443 to 1,589. The number of firms in the computer, electronics,
pharmaceutical/biotechnology, and retailing industry portfolios ranges
from 492 to 579, 430 to 484, 74 to 79, and 441 to 450, respectively.
(b) CAR = cumulative abnormal returns starting from 12/4/95.
Abnormal returns are obtained from OLS regression of daily
portfolio returns on market index and dummy variables for dates
from 12/4/95 to 12/26/95. Estimation period is the 252 trading
days in the year 1995. Model: [R.sub.t] = [a.sub.0] + [a.sub.1]
[R.sub.mt] + [Sigma][a.sub.2i][D.sub.i] + [u.sub.t], where
[R.sub.t] = Portfolio daily stock return in percentage;
[R.sub.mt] = CRSP daily value-weighted market return in percentage;
and [D.sub.i] = 1 if the day corresponds to event date
i, and 0 otherwise.
To summarize, we interpret our evidence of significant abnormal returns for certain events leading to the passage of PSLRA, our analysis of intertemporal correlations of abnormal returns between days with expected reversal of the probability of the passage of PSLRA, and the cumulative abnormal returns from the first congressional vote on the conference committee bill to the final Senate override, as more consistent with the conclusion that shareholders viewed PSLRA as harmful. Nonetheless, the confounding veto and House override on December 20 preclude conclusive evidence. We therefore analyze other events related to the securities litigation environment and PSLRA to shed more light on whether shareholders considered restrictions on their ability to bring class-action lawsuits for securities fraud harmful.
V. OTHER EVENTS RELATED TO PSLRA
Proposition 211
California's Proposition 211 proposed to reverse many of PSLRA's key provisions, less than a year after PSLRA was enacted. A negative stock price reaction by high-litigation-risk firms to the defeat of Proposition 211 would suggest that shareholders considered PSLRA harmful. Also, firms whose stock prices react most negatively to the passage of PSLRA should also have the most negative stock price reaction to the defeat of the proposition. Thus, a positive correlation between abnormal returns associated with the passage of PSLRA and the defeat of Proposition 211 would provide additional evidence that the abnormal returns are due to these litigation-related events, and that the shareholders considered PSLRA harmful. Finally, the direction of the correlation between the abnormal returns on 12/20/95 and the defeat of Proposition 211 can help discern whether the abnormal return on 12/20/95 is more likely due to the veto or to the House vote.
Proposition 211 adopted a "fraud-on-the-market" theory, provided no safe harbor for forward-looking statements, restored joint and several liability for each defendant regardless of the proportionate involvement, and authorized the aiding and abetting liability that the Supreme Court eliminated in 1994. It also allowed punitive damages, prohibited regulation of attorneys' fees, and barred a company from paying the costs of defending its officers who "knowingly or recklessly" engage in deceptive conduct. Though a state ballot measure, it was significant nationally because it permitted lawsuits against any company with shareholders in California (Corcoran 1996; Bencivenga 1996). SEC Chairman Levitt indicated Proposition 211 would have a "nationwide impact" (Donovan 1996), and it was the costliest campaign ever over a California ballot measure (Meyerson 1996).
We examine the stock returns on the day of the Proposition 211 vote (November 5, 1996) and on the day after. On November 5, State News Brief (3:05 PM, Eastern Standard Time) predicted that Proposition 211 would fail. Proposition 211 was eventually defeated by a 3 to 1 margin (Abate 1996). The financial press indicated that the greatest uncertainty for the stock market about the election seemed to be the outcome of Proposition 211. The Associated Press 11/3/96 (Meyerson 1996) reported that "with most market players betting the elections will produce few changes in Washington, a prickly battle over a California ballot proposal looms as the biggest wild card for Wall Street on Tuesday." The Washington Post (Glassman 1996) also made similar observations.
We also consider two other events within a week of the vote. On October 30, 1996, a Field poll indicated that 54 percent of the voters opposed Proposition 211, and that 28 percent were in favor. This was a significant shift since the Field poll in early October, which showed only 31 percent of the voters opposing and 39 percent favoring the proposition (Gunnison 1996). On November 4, the day before the vote, another Field poll indicated that 56 percent of the voters opposed and 25 percent favored the proposition (Yoachum 1996). This news is likely to be informative beyond the October 30 Field poll result, because it indicates that the opposition to Proposition 211 is stable, and the poll result one day before the actual vote is likely to be a more reliable measure of the actual outcome. The events on each of these days indicate a decrease in the likelihood of the passage of Proposition 211, so we classify these events as not being plaintiff-friendly (see panel A of Table 5).(18)
TABLE 5
Proposition 211 Event Dates and Results from Regression of Daily
Portfolio Returns for High-Litigation-Risk Firms on Market Index
and Dummy Variables for the Event Dates Estimation Period is the
261 Trading Days from 10/31/95 to 11/9/96(a)
Model: [R.sub.t] = [a.sub.0] + [a.sub.1][R.sub.mt] + [Sigma]
[a.sub.2i][D.sub.i] + [u.sub.t](b)
Panel A: Proposition 211 Event Dates
Plaintiff
Date Friendly? Event
10/30/96 No Poll shows Proposition 211 trailing: 28%
yes; 54% no; 18% undecided, which is a
major shift since poll conducted in
early October: 39% yes, 31% no; 30%
undecided. Source: San Francisco
Chronicle (Gunnison 1996)
11/4/96 No Poll shows 25% support Proposition 211,
56% oppose it, with 19% undecided.
Source: San Francisco Chronicle (Yoachum
1996)
11/5/96 No Proposition 211 would fail 26% to 74%.
Source: State News Briefs (1996, November
5, 3:05 PM EST)
11/6/96 No Voters rejected Proposition 211 by a
stunning 3-to-1 ratio. Source: The San
Francisco Examiner (Abate 1996)
Panel B: Abnormal Returns for Portfolios of High-Litigation-Risk Firms
All High-Risk Firms
Plaintiff Coefficient Random.
Date, i Friendly? (t-stat., p) p-value(c)
Intercept 0.00
(0.02, 0.98)
[R.sub.mt] 1.30
(26.28, 0.00)(***)
10/30/96 No -0.08
(-0.16, 0.87) 0.45
11/4/96 No -1.21
(-2.28, 0.02) 0.03(**)
11/5/96 No -1.08
(-2.03, 0.04) 0.03(**)
11/06/96 No -0.67
(-1.26, 0.21) 0.14
Adj. [R.sup.2] 0.57
P211XR(d) -3.04
(F-statistic) (8.06)(***)
(p-value) (0.00)
Computers
Coefficient Random.
Date, i (t-stat., p) p-value
Intercept -0.01
(0.15, 0.88)
[R.sub.mt] 1.46
(25.29, 0.00)(***)
10/30/96 0.06
(0.09, 0.93) 0.99
11/4/96 -1.19
(-1.93, 0.05) 0.02(**)
11/5/96 -1.48
(-2.40, 0.02) 0.01(***)
11/06/96 -1.00
(-1.60, 0.11) 0.14
Adj. [R.sup.2] 0.55
P211XR(d) -3.61
(F-statistic) (8.39)(***)
(p-value) (0.00)
Electronics
Coefficient Random.
Date, i (t-stat., p) p-value
Intercept -0.02
(0.45, 0.65)
[R.sub.mt] 1.42
(23.15, 0.00)(***)
10/30/96 0.05
(0.07, 0.94) 0.56
11/4/96 -1.49
(-2.28, 0.02) 0.02(**)
11/5/96 -0.82
(-1.25, 0.21) 0.07(*)
11/06/96 -0.78
(-1.19, 0.23) 0.09(*)
Adj. [R.sup.2] 0.51
P211XR(d) -3.04
(F-statistic) (5.29)(**)
(p-value) (0.02)
Pharmaceuticals/Biotech
Coefficient Random.
Date, i (t-stat., p) p-value
Intercept 0.00
(0.02, 0.98)
[R.sub.mt] 1.01
(11.98, 0.00)(***)
10/30/96 -0.17
(-0.19, 0.85) 0.85
11/4/96 -0.76
(-0.85, 0.40) 0.45
11/5/96 -0.71
(-0.78, 0.44) 0.33
11/06/96 -0.49
(-0.54, 0.59) 0.36
Adj. [R.sup.2] 0.21
P211XR(d) -2.13
(F-statistic) (1.37)
(p-value) (0.24)
Retailing
Coefficient Random.
Date, i (t-stat., p) p-value
Intercept 0.05
(1.59, 0.11)
[R.sub.mt] 0.80
(17.17, 0.00)(***)
10/30/96 -0.75
(-1.51, 0.13) 0.05(**)
11/4/96 -1.03
(-2.06, 0.04) 0.10(*)
11/5/96 -0.68
(-1.36, 0.17) 0.19
11/06/96 0.29
(0.57, 0.57) 0.78
Adj. [R.sup.2] 0.37
P211XR(d) -2.17
(F-statistic) (4.66)(**)
(p-value) (0.03)
(*), (**), (***) indicate significance at 10, 5, and 1 percent level
or better, respectively (two-tailed test).
(a) The portfolio of high-litigation-risk firms on a given day
consists of all firms in the following four industries for which daily
returns data are available: computers (SIC codes 3570-3577 and
7370-7374), electronics (SIC codes 3600-3674), pharmaceuticals/
biotechnology (SIC codes 2833-2836 and 8731-8734), and retailing (SIC
codes 5200-5961). The number of firms in the portfolio ranges from
1,486 to 1,778. The number of firms in the computer, electronics,
pharmaceutical/biotechnology, and retailing industry portfolios ranges
from 516 to 697, 457 to 521, 77 to 87, and 436 to 475, respectively.
(b) Variable definitions:
[R.sub.t] = portfolio daily stock return in percentage;
[R.sub.mt] = CRSP daily value-weighted market return in percentage;
and
[D.sub.i] = 1 if the day corresponds to event date i, and 0 otherwise.
(c) Randomization p-values are based on the distribution of abnormal
returns for 100 random days with market returns within plus or minus
0.05 percent of that of the test day.
(d) P211XR = sum of abnormal returns on 10/30/96, 11/4/96, 11/5/96,
and 11/6/96.
Panel B of Table 5 reports that abnormal returns are negative and significant on November 4, 1996, the day the Field poll results were released, and on November 5, 1996, the election day. The cumulative abnormal return for high-litigation-risk firms for the four event dates related to Proposition 211, P211XR, is -3.04 percent (p = 0.00). P211XR is negative for all four of the high-litigation-risk industries, and is significant for three: -3.61 percent (p = 0.00) for computers, -3.04 percent (p = 0.02) for electronics, and -2.17 percent (p = 0.03) for retailing. The consistency of results across the high-litigation-risk industries provides additional confidence that the abnormal returns are related primarily to Proposition 211. These results suggest that shareholders reacted negatively when Proposition 211's attempt to reverse the provisions of PSLRA did not succeed. This evidence supports the conclusion that shareholders considered PSLRA harmful.
The Spearman correlation for the sample of high-litigation-risk firms between P211XR and PSLRA-related cumulative abnormal return, PSLRAXR (which corresponds to the sum of the abnormal returns on the three sets of full House and Senate votes and on the veto, occurring on 3/8/95, 6/28/95, 12/5/95, 12/6/95, 12/20/95, and 12/22/95, is significantly positive, 0.05 (p = 0.05). This supports the inference that these abnormal returns reflect the litigation-related events, and that the shareholders considered PSLRA harmful. Finally, the Spearman correlation between P211XR and abnormal returns on 12/20/95 is significantly negative, -0.12 (p = 0.00), further suggesting that the positive abnormal return for high-litigation-risk firms on 12/20/95 is more likely due to the veto than to the House vote.
Supreme Court Decisions
We examine stock price reactions to Supreme Court decisions related to securities-fraud cases decided since 1988.(19) A negative (positive) stock price reaction to decisions that made it harder (easier) for shareholders to sue for damages from securities fraud would suggest that shareholders consider restrictions on their rights to sue to be harmful. We are cautious in drawing inferences about PSLRA from these analyses, however, because the specific securities litigation issues a Supreme Court decision addresses may be quite different from the issues the PSLRA addresses.
Panel A of Table 6 lists the Supreme Court decisions and indicates whether each decision made the litigation environment more or less plaintiff-friendly--that is, easier or harder for shareholders to sue for damages from securities fraud. In Basic v. Levinson (485 US 224, 1988), the Supreme Court made the plaintiff's task much easier by upholding the fraud-on-the-market theory. Plaintiffs need not show that they actually decided to purchase or sell the security based on the defendant's misleading statement or omission; they need only plead that they bought or sold relying on the integrity of the market price, which was affected by the statement or omission. During the 1990s, however, the Supreme Court decided several securities cases against plaintiffs. In 1991, the Supreme Court shortened the statutory period of limitations for bringing lawsuits alleging fraud (Lampf Pleva, Lipkind, Prupis & Petigrow v. Gilbertson, 501 US 350). In Reves v. Ernst and Young (113 S. Ct. 1163, 1993), the Court ruled that professional advisors (such as auditors) were not liable under the Racketeer Influenced and Corrupt Organizations Act (RICO) (Ratner 1998). In Central Bank of Denver, N.A. v. First Interstate Bank of Denver (114 S. Ct. 1439, 1994), the Court set aside 25 years of litigation practice by ruling that role 10(b)-5 did not prohibit aiding and abetting securities fraud because it contains no language to that effect. This ruling was a response to perceived securities litigation abuses (Avery 1996)--the same concern that prompted Congress to enact PSLRA (Hamilton 1996).
TABLE 6
Supreme Court Decisions on Securities Lawsuits Since 1988 and Results
from Regression of Daily Portfolio Returns for High-Litigation-Risk
Firms on Market Index and Dummy Variable for the Central Bank of Denver
Decision Date Estimation Period is the 262 Trading Days from 4/19/93 to
4/29/94(a)
Model: [R.sub.t] = [a.sub.0] + [a.sub.1][R.sub.mt] + [Sigma]
[a.sub.2i][D.sub.i] + [u.sub.t](b)
Panel A: Supreme Court Decisions
Plaintiff- Significant
Date Friendly? Market Reaction?
3/7/88 Yes No Basic v. Levinson, 485 US
224. Vote: 4-2. Supreme Court
endorsed fraud-on-the-market
doctrine that enables
plaintiffs to claim damages
by arguing that they relied
on the market price, which
incorporated the misleading
information the defendant
disseminated. Thus,
plaintiffs do not have to
prove that they based their
decisions on actual knowledge
of the misleading information.
6/20/91 No No Lampf, Pleva, Lipkind, Prupis
& Petigrow v. Gilbertson, 501
US 350. Vote: 5-4. Shortened
the statutory period of
limitations for bringing
lawsuits (three years of the
alleged fraud and one year
from the discovery of the
alleged fraud).
3/3/93 No No Reves v. Ernst and Young, 113
S. Ct. 1163. Vote: 7-2. Held
that outside advisors such as
auditors are not liable for
triple damages under RICO.
4/19/94 No Yes Central Bank of Denver, N.A.
v. First Interstate Bank of
Denver, 114 S. Ct. 1439.
Vote: 5-4.
The Court ruled that rule
10(b)-5 did not prohibit
aiding and abetting
securities fraud. This
restricted the ability of
plaintiffs to sue outside
advisors such as auditors.
Panel B: Abnormal Returns for Portfolios of High-Litigation-Risk
Firms--Central Bank of Denver Decision
All High-Risk Firms
Plaintiff Coefficient Random.
Date, i Friendly? (t-stat, p-value) p-value(c)
Intercept 0.14
(5.49, 0.00)(***)
[R.sub.mt] 0.97
(20.63, 0.00)(***)
4/19/94 No -1.16
(-2.91, 0.00) 0.01(***)
Adj. [R.sup.2] 0.63
Computer
Coefficient Random.
Date, i (t-stat, p-value) p-value
Intercept 0.17
(5.06, 0.00)(***)
[R.sub.mt] 1.12
(17.88, 0.00)(***)
4/19/94 -1.47
(-2.77, 0.01) 0.02(**)
Adj. [R.sup.2] 0.56
Electronics
Coefficient Random.
Date, i (t-stat, p-value) p-value
Intercept 0.17
(5.52, 0.00)(***)
[R.sub.mt] 0.97
(16.27, 0.00)(***)
4/19/94 -1.32
(-2.61, 0.01) 0.03(**)
Adj. [R.sup.2] 0.51
Pharmaceuticals/Biotech
Coefficient Random.
Date, i (t-stat, p-value) p-value
Intercept 0.04
(1.19, 0.23)
[R.sub.mt] 1.01
(14.72, 0.00)(***)
4/19/94 -0.29
(-0.50, 0.61) 0.67
Adj. [R.sup.2] 0.45
Retailing
Coefficient Random.
Date, i (t-stat, p-value) p-value
Intercept 0.09
(3.44, 0.00)(***)
[R.sub.mt] 0.80
(16.97, 0.00)(***)
4/19/94 -0.84
(-2.11, 0.03) 0.04(**)
Adj. [R.sup.2] 0.53
(*), (**), (***) indicate significance at 10, 5, and 1 percent level
or better, respectively (two-tailed test).
(a) The portfolio of high-litigation-risk firms on a given day
consists of all firms in the following four industries for which daily
returns data are available: computers (SIC codes 3570-3577 and
7370-7374), electronics (SIC codes 3600-3674), pharmaceuticals/
biotechnology (SIC codes 2833-2836 and 8731-8734), and retailing (SIC
codes 5200-5961). The number of firms in the portfolio ranges from
1,221 to 1,416. The number of firms in the computer, electronics,
pharmaceutical/biotechnology, and retailing industry portfolios ranges
from 421 to 488, 345 to 403, 79 to 87, and 375 to 438, respectively.
(b) Variable definitions:
[R.sub.t] = portfolio daily stock return in percentage;
[R.sub.mt] = CRSP daily value-weighted market return in percentage;
and
[D.sub.I] = 1 if the day corresponds to event date i, and 0 otherwise.
(c) Randomization p-values are based on the distribution of abnormal
returns for 100 random days with market returns within plus or minus
0.05 percent of that of the test day.
We find a significant stock price reaction only for the narrowly decided (5-4) Supreme Court decision on Central Bank of Denver. The lack of significant reaction to the other three Supreme Court decisions is not particularly surprising. For example, Weiss and White (1987) examine seven significant corporate-law-related Delaware court decisions, and find insignificant market reactions to each. The significant stock price reaction to the Central Bank of Denver decision underscores both the importance and the surprise element of this decision. Seligman (1994b, 1430) notes it "is the most important federal securities law decision in several years," and that the decision "was an unexpected result." Our analysis, however, cannot rule out the possibility that the other three Supreme Court decisions affected shareholders' wealth, but the market had already anticipated the decisions.
For brevity, Panel B of Table 6 reports the regression results of only the Supreme Court decision on Central Bank of Denver. Abnormal return for the portfolio of high-litigation-risk firms is significantly negative, -1.16 percent (p = 0.01). Abnormal return is also negative and significant for three high-litigation-risk industries: -1.47 percent (p = 0.02) for computers, -1.32 percent (p = 0.03) for electronics, and -0.84 percent (p = 0.04) for retailing. These results suggest that shareholders reacted negatively to another change in litigation regime that made it harder for shareholders to sue deep-pocket advisors--namely, the Supreme Court decision that 10(b)-5 did not prohibit aiding and abetting securities fraud. This was the last major securities-related judicial or legislative event before the enactment of PSLRA, so the Central Bank of Denver decision results further support the conclusion that shareholders considered restrictions on their rights to sue, such as in PSLRA, to be harmful.
The abnormal return on the day of the Central Bank decision is not significantly correlated with the PSLRA-related cumulative abnormal return (PSLRAXR) or with the abnormal return on 12/20/95. A plausible explanation is that PSLRA does not consider the issue of private securities litigation for aiding and abetting, and that firms that were most sensitive to the aiding and abetting issue were not the ones most sensitive to the provisions considered under PSLRA. We also do not observe a significant correlation between abnormal returns for the Central Bank decision and the defeat of Proposition 211. Even though Proposition 211 proposed to reverse the effect of the Central Bank decision by authorizing the aiding and abetting liability, the proposition had many other provisions that were unrelated to the Central Bank decision (and which were primarily related to PSLRA).
VI. ALTERNATIVE LITIGATION-RISK PROXIES
We have used membership in high-litigation-risk industries to identify firms subject to high litigation risk. Jones and Weingram (1996) estimate a litigation prediction model that includes most of the litigation-risk proxies identified in the literature. They find that the following variables have significant explanatory power: (1) standard deviation of daily returns measured from t - 20 to t - 1 (STNDTW), where t is the date for which litigation risk is assessed; (2) skewness of returns measured from t - 250 to t - 1 (SKEW); (3) market capitalization measured on day t - 1 (MKTCAP); (4) cumulative turnover, estimated as the proportion of outstanding shares that traded from t - 250 to t - 1 (ECTURN); (5) cumulative return from t - 250 to t - 1 (ANRET); and (6) systematic risk, which is the Scholes-Williams adjusted beta, measured using the CRSP value-weighted index and returns from t - 250 to t - 1 (BETA). We refer to the litigation-risk measure based on this model as LITGRISK.(20) As expected, the rank correlation between LITGRISK and the high-litigation-risk industry indicator variable, which takes a value of 1 if a firm belongs to a high-litigation-risk industry and 0 otherwise, is positive and significant (0.30; p = 0.00). However, because the two measures are far from perfectly correlated, we examine the sensitivity of our conclusions to using LITGRISK. This measure also allows us to use a much broader sample, including firms outside the four high-litigation-risk industries.(21)
We examine the association between LITGRISK and market-model-based abnormal returns for PSLRA (PSLRAXR), Proposition 211 (P211XR), and the Supreme Court decision on the Central Bank of Denver case (CBDXR). The estimated OLS standard errors of a regression of the event-date abnormal returns on LITGRISK would be biased due to contemporaneous correlation in returns (Bernard 1987). The portfolio OLS regression procedure, proposed by Sefcik and Thompson (1986), controls for the bias. Also, to reduce the effect of extreme values, we delete the top and bottom 0.5 percent of the values of LITGRISK. Table 7 reports that, as predicted, abnormal returns for PSLRA, Proposition 211, and the Supreme Court decision are significantly negatively associated with LITGRISK, with t-statistics of -2.67, -1.95, and -2.24, respectively.(22) Thus, our inferences are robust to the choice of litigation-risk proxy.(23)
TABLE 7
For PSLRA and Other Related Events, Cross-Sectional Relations Obtained
from the Sefcik and Thompson (1986) Portfolio Regression of Market
Model Abnormal Returns on a Litigation-Risk Measure Based on Jones and
Weingram (1996)(a)
Dependent Variable(b) Intercept LITGRISK(c)
PSLRAXR 0.03 -0.63
(0.07) (-2.67)(***)
P211XR -0.26 -0.50
(-0.57) (-1.95)(*)
CBDXR -0.24 -0.20
(-0.97) (-2.24)(**)
(*), (**), (***) indicate significance at 10, 5, and 1 percent
level or better (two-tailed test).
(a) Entries are estimates and t-statistics using the
Sefcik-Thompson portfolio OLS regression procedure; for estimation
for PSLRA events, Proposition 211 events, and Central Bank of Denver
decision, the sample sizes are 7,681, 8,587, and 8,037 firms,
respectively, and the time period is one year that ends on 12/31/95,
11/6/96, and 4/19/94, respectively.
(b) PSLRAXR = cumulative abnormal return (in percentage) of major PSLRA
events; computed as the sum of the abnormal returns on the event dates
when the probability of the passage of PSLRA is expected to increase,
3/8/95, 6/28/95, 12/5/95, 12/6/95, and 12/22/95, less abnormal returns
on the event date when the probability of the passage of PSLRA is
expected to decrease, 12/20/95. See Table 1 for event description;
P211XR = cumulative abnormal return (%) on the Proposition 211 event
dates (sum of abnormal returns on 10/30/96, 11/4/96, 11/5/96, and
11/6/96). See Table 5 for event description;
CBDXR = cumulative abnormal return (%) on the Central Bank of Denver
decision date, 4/19/94; and
(c) LITGRISK = litigation risk calculated using significant
coefficients from Jones and Weingram (1996, Tables 3 and 12). To reduce
the effect of extreme values, top and bottom 0.5 percent values of
LITGRISK are deleted. Predicted signs on LITGRISK are negative because
each of the three events is plaintiff-unfriendly.
VII. CONCLUSIONS
On December 22, 1995, Congress enacted the Private Securities Litigation Reform Act, which increased restrictions on private litigation for securities fraud. We examine stock price reactions of firms in four high-litigation-risk industries: computers, electronics, pharmaceuticals/biotechnology, and retailing (Francis et al. 1994a). Two other studies on this issue, ST and JKN, conclude that shareholders considered PSLRA beneficial. While we find largely similar daily abnormal returns for event-related days that they examine, we present evidence that the timing of multiple confounding events makes the interpretation of these daily returns ambiguous. Results from additional analyses, beyond that conducted by ST and JKN, where significant, are largely inconsistent with their interpretation. First, results of intertemporal correlation analysis suggest that the positive abnormal return observed on 12/20/95 is more likely a response to the presidential veto and not to the House override. Second, abnormal returns for events other than the veto during the almost year-long debate on PSLRA suggest that shareholders considered PSLRA harmful. Third, we observe significantly negative cumulative abnormal returns during the last three weeks, when the final version of the bill received the congressional approval, presidential veto, and congressional override of the veto. Fourth, we find a significantly negative stock price reaction to the defeat of a California state ballot initiative, Proposition 211, intended to reverse many of PSLRA's key provisions; these abnormal returns are positively correlated with abnormal returns related to the passage of PSLRA. Finally, we find significantly negative abnormal returns to the 1994 Supreme Court decision that made it harder for shareholders to sue professional advisors suspected of willfully aiding and abetting securities violations. While the confounding timing of events makes it difficult to provide conclusive evidence, we interpret the entire body of evidence as suggesting that, on average, investors in these four high-litigation-risk industries react negatively to the PSLRA's restrictions on their ability to bring securities-related lawsuits.
Investors' disapproval of the PSLRA could be due to a decrease in the unexpected loss of insurance value or due to decrease in deterrence of false or misleading disclosures, which, according to Seligman (1994a), is one of the most desirable attributes of securities law. We cannot directly test the relative contributions of these two effects. Nevertheless, the litigation settlement data in Carleton et al. (1996) for the computer, electronics, and pharmaceutical /biotechnology firms provide an estimate of the maximum stock price effect of the unexpected loss of insurance value due to PSLRA.(24) The abnormal returns that we observe for these industries for some of the PSLRA events are much larger, suggesting that a decrease in deterrence due to increased restrictions on private securities litigation is likely a major factor in shareholders' disapproval of PSLRA.
APPENDIX
Newspaper Articles on Whether the President's Veto on the Night of December 19 Was More Surprising than the House Vote to Override the Veto on December 20
(Sources: Wall Street Journal, Dow Jones News Service, and newspapers covered by the Lexis-Nexis database.)
I. Articles Suggesting That the Veto Was More Surprising
1. Wall Street Journal 12/20/95 (Frisby and Taylor 1995b) lists the sequence of events leading to the veto, indicating that President Clinton had not made up his mind before December 19:
12/15 (Friday)--"Mr. Clinton spent time with top class-action trial lawyers who oppose the bill ... during a thank-you dinner hosted at the White House for the big donors to his campaign." (p. A3)
12/16 and 12/17 (Saturday and Sunday)--"Mr. Clinton, a former law professor, delved into minute details of the bill over the weekend, calling law professors to get their predictions on the legislation's impact." (p. A3)
12/18 (Monday night)--"He [Mr. Clinton] talked with Senator Dodd into the early morning hours about the pleading standards; the senator followed up yesterday by sending the president examples of cases that have been successfully litigated in states where the pleading standards are similar to the ones in the bill." (p. A3)
12/19 (Tuesday night)--"Clinton set up a meeting late night in the Oval Office with his lawyers, aides said, to give the bill's supporters one last chance to resolve his concerns.... [T]he president rejected it just hours before midnight when it automatically would have become law.... "(p. A3)
2. Wall Street Journal 12/20/95 (Frisby and Taylor 1995b): "In the House, support is considered strong enough to override the veto with ease.... The veto sets the stage for a likely override fight in the Senate." (p. A3)
3. Dow Jones News Service (1995a) before the House vote: "House is expected to override the veto ... but Senate Republican leaders are unsure whether they will be able to collect the required two-thirds support."
4. Dow Jones News Service (1995b), after the House vote: "House was expected to override the veto ... but the Senate vote is likely to be more contentious."
5. Washington Post 12/21/95 (Walsh 1995): "Clinton decided to veto the bill just hours before a midnight deadline Tuesday, surprising supporters and opponents alike.... "(p. A20)
6. Wall Street Journal 12/21/95 (Taylor 1995): "Late Tuesday night, Mr. Clinton stunned a coalition of publicly owned companies, accountants and securities firms advocating the bill by vetoing the legislation--after indicating earlier that he planned to sign it." (p. A3)
7. Wall Street Journal 12/22/95 (Frisby and Taylor 1995c): "The veto was unexpected." (p. A3)
IIA. Articles Referenced by Spiess and Tkac (1997) and Johnson et al. (2001) for the News of Veto Rumor on 12/18/95 (Basis for Their Conclusion That the House Vote on 12/20/95 Was More Surprising than Was the Veto on 12/19/1995)
1. National Journal's Congress Daily (1995): "Speculation was rampant on Capitol Hill as congressional sources insisted the president told staff Monday he had `tentatively decided' to veto the legislation."
2. Washington Times (1995): "Word has it that Mr. Clinton is reaching for the veto pen." (p. A20)
IIB. Articles Suggesting That the Market May Not Have Considered the Veto Rumor on 12/18/95 Credible
1. National Journal's Congress Daily (1995): "Clinton late Monday was deciding whether to sign the securities litigation reform bill, White House Press Secretary Michael McCurry said."
2. Associated Press 12/20/95 (Fournier 1995): "As late as Monday aides were saying they were nearly certain that Clinton would sign the bill."
We gratefully acknowledge the comments of Bill Felix, Joseph Grundfest, Laurie Krigman, Lil Mills, Adam Pritchard, Jerry Salamon, Joel Seligman, Lynn Stout, Elliott Weiss, two anonymous reviewers, and the workshop participants at The University of Arizona, the Hong Kong University of Science and Technology, University of Maryland, Purdue University, and Tulane University. Earlier versions of the paper were titled: "Shareholders' Perception of the Securities Litigation Regime in the U.S.: Stock Price Impact of Private Securities Litigation Reform Act of 1995 and Related Events."
(1) Grundfest and Perino (1997) show that after PSLRA plaintiffs must, on average, demonstrate more dramatic wrongdoing to satisfy the new federal pleading standard. Specifically, they report that the average daily-stock-price decline on the disclosure of wrongdoing increased from 19 percent to 31 percent after the enactment of PSLRA.
(2) Seligman (1994a) points out that the SEC has limited resources for enforcing the mandatory disclosure system. Usually, the SEC obtains only consent decrees, whereby corporate wrongdoers agree not to do certain things in the future. Thus, the most effective deterrent preventing corporate officers and others from defrauding shareholders is private litigation.
(3) If shareholders fail to file a lawsuit for pre-PSLRA investment losses during the less restrictive pre-PSLRA regime, the firm's value may increase from the passage of PSLRA. This argument predicts that stocks of high-litigation-risk firms would react positively to PSLRA. However, this prediction is countered by the following arguments. First, the directors' and officers' liability-insurance providers generally cover the potential damage claims against firms. Hence, it is the insurance provider, rather than the firm, who may benefit from the passage of PSLRA. Second, shareholders with pre-PSLRA investment losses had the incentive to sue the firms before the enactment of PSLRA, and in fact, lawsuits surged during the weeks before the passage of PSLRA (Schmitt 1996). To the extent that shareholders filed a lawsuit for pre-PSLRA investment losses before PSLRA's effective date, its passage may not have affected firm value.
(4) We do not examine stock price reaction to congressional hearings and floor debates on PSLRA. We think that the market largely anticipates the substance of these events. For example, on 1/19/95, the House subcommittee on Telecommunication and Finance held the first hearing on the Securities Litigation Reform Act. Avery (1996) reports that three witnesses testified in favor of the bill: a representative of the American Business Conference, the chairman of a high-technology company, and Professor R. Fischel of the University of Chicago Law School; and that William S. Lerach, on behalf of the National Associations of Securities and Commercial Law Attorneys, testified against. The market is unlikely to have been surprised by their testimonies, given the position of the organizations that the witnesses represented, and in the case of Professor Fischel, given his career as a highly paid consultant to corporate defendants in securities-fraud cases (Henriques 1995a). Similar concerns apply to other congressional hearings and floor debates.
(5) ST and JKN attribute the December 20 stock price reaction to news of the House vote. Sections IV and V of our paper provide additional evidence to distinguish between the two interpretations.
(6) O'Brien and Hodges (1993) and Jones and Weingram (1996) also report that high-tech companies were involved in a disproportionately large number of securities-fraud class-action cases. Grundfest and Perino (1997) report that this continues to be the case after PSLRA. We repeat our analysis in a broader sample of firms, using an alternative measure of litigation risk. Results reported in Section VI are consistent with inferences from our primary analysis.
(7) We do not conduct randomization procedures to assess significance levels for cumulative abnormal returns. It would require the random dates to have consecutive days with market returns similar to the market returns on each of the consecutive test days. It would not be possible to obtain sufficient observations. Moreover, with cumulative returns, the effect of the bias due to an unusually small or large market return on a particular date is likely to be relatively small.
(8) ST's and our sample firms are from the same industries. For the dates that are common between ST (Table 2) and our Table 1, abnormal returns are very similar in magnitude. ST do not report t-statistics or p-values (other than to indicate that they are below the 5 or 10 percent levels), making it difficult to compare significance levels across the two studies. There are two dates in Table 1 where our conventional p-values suggest that the abnormal returns are significant, whereas ST classifies abnormal returns on these two dates as insignificant. On 12/5/95 and 12/11/95, our conventional p-values are 0.10, whereas ST indicates p-values of greater than 10 percent (they do not report the actual values). However, the magnitudes of abnormal returns on these two dates are very similar across the two studies (-0.80 percent and -0.86 percent in ST, and -0.71 percent and -0.71 percent in our study). JKN's sample is different from both ST's and our sample, however, except for one event date, their results (JKN, Table 1) are similar to ST's and ours for events that are common across our studies. We discuss the exception in the next section.
(9) The main reasons cited for the large decline in stock prices were growing concerns about the Washington budget imbroglio, which had already partially shut down the government, and a weakening economy with more signs of earnings difficulty (Kansas 1995).
(10) The randomization procedure is similar to the one described in Section III, with the following exception. There were so few dates with such large absolute market returns that we had to use market returns within plus or minus 0.10 percent, instead of 0.05 percent, from the test-date market return. Even so, for the entire 1976-1995 period, we found only 37 days with market returns between -1.8 and -1.6 percent. For sensitivity analysis, we select random days between 1976 and 1995 on which the market return is within plus or minus 0.15 percent (0.20 percent, 0.25 percent) of -1.7 percent, and obtain 54 (71, 96) such days. The randomization p-value turns out to be 0.15 (0.17, 0.14), which is very close to the 0.16 value reported in Table 1.
(11) Prior research has used price reversal tests primarily to investigate the stock market effects of mandatory accounting policy changes (e.g., Foster 1980; Smith 1981; Noreen and Sepe 1981; Collins et al. 1982). For example, Smith (1981) and Collins et al. (1982) document a negative correlation between abnormal returns during the period when the FASB proposed elimination of full cost accounting and when the SEC proposed 13 months later to allow full cost accounting, and conclude that this result confirms that the abnormal returns are due to the accounting policy changes.
(12) Alternatively, the negative correlation may not be due to the veto, but to a decrease in the market's confidence on 12/20/95 in the Senate's ability to override the veto. This alternative interpretation also casts doubt on ST's and JKN's conclusion that the market reacted adversely to a decrease in the probability of passage of PSLRA. In any case, a search of the articles in the Wall Street Journal and all newspapers covered by Lexis-Nexis suggests that neither the content of the veto statement nor the details of the House vote (e.g., positions taken by specific House members), is likely to have affected expectations of whether the Senate would override the veto. We compare who voted for or against in the two House votes on 12/5/95 and 12/20/95. Of the members who voted on both occasions, four members changed their votes from yea to nay, and four changed from nay to yea. None of these eight members were either actively involved or influential with respect to PSLRA; e.g., none of them had participated in any of the floor debates on this legislation.
(13) The House vote took place at 12:20 PM on 12/20/95. We tried an intra-day analysis to measure separate abnormal returns for the veto and the House vote, but results were insignificant, in part because intra-day data were available for less than 20 percent of our sample firms.
(14) We thank JKN for providing their sample.
(15) Except for 12/22/95, none of our results are sensitive to whether we exclude firms with SIC code 2830 in the pharmaceuticals/biotechnology sample. However, on 12/22/95 the pharmaceuticals/biotechnology sample that includes firms with SIC code 2830 yields an abnormal return of 2.86 percent. This is about six times the next-highest abnormal return of 0.50 percent for the computer firms (JKN, Table 1, also report a similar result). For other event dates with significant abnormal returns for the full sample, the highest industry abnormal return is never more than twice the next-highest industry abnormal return. Thus, except for 12/22/95, the magnitude of abnormal returns for the full sample of high-litigation-risk firms is not dominated by one industry.
We decided to report in this paper the results for the sample excluding firms with SIC code 2830 for the following reason. Our sample, like ST's, includes firms on CRSP in industries with four-digit SIC codes that Francis et al. (1994a) suggest experience high litigation rates. Thus, our sample is directly comparable to ST's. We elected to use a sample comparable to ST's rather than sample comparable to JKN's because JKN's sample excludes two industries that Francis et al. (1994a) consider high litigation risk--the electronics and retailing industries. Thus, JKN's full sample contains only 489 firms compared to 1,485 firms in ST and 1,443 to 1,589 firms in our study.
(16) Since 12/22/95 was the Friday before Christmas, trading was lighter than usual--volume on New York Stock Exchange was 289.6 million shares, compared with 415.8 million shares the day before. Thus, news of the Senate vote may not have been completely impounded in stock prices on 12/22/95. However, the abnormal returns on 12/22/95 and when the market reopened on 12/26/95 together are also insignificant, -0.17 percent (p = 0.66).
(17) It is unlikely that dilution of the bill is driving the CAR to be negative over the 12/4/95 to 12/26/95 period. The conference committee unveiled on 11/28/95 the version of the bill that was ultimately enacted. Dilution of the bill seems to have occurred only during the Senate's veto-override process, which would be reflected in abnormal returns from 12/21/95 to 12/26/95. Table 4 shows no evidence of a significant change in CAR during that period.
(18) Another recent securities litigation-related legislative event is the passage of Securities Litigation Uniform Standard Act. The law intends to prevent plaintiffs from using state courts to circumvent parts of PSLRA. This law is not expected to have a significant impact on the litigation environment, because the surge in lawsuits at the state level shortly after the passage of PSLRA, which lawmakers used to justify this legislation, was short-lived. The Act thus responds to a problem that does not exist (Painter 1998). We find insignificant abnormal returns for high-litigation-risk firms on the days when news reached the market of the two sets of full House and Senate votes (Dow Jones News Service 1998a, 1998b, 1998c, 1998d) and of the President's signing the bill (U.S. Newswire 1998); the cumulative abnormal return for these event dates is 0.66 percent, t = 0.65.
(19) We consider Supreme Court decisions beginning in 1988 for the following reasons. First, since 1988 there has been a significant increase in litigation activity, attributed primarily to the Supreme Court decision that upheld the fraud-on-the-market theory, which made it much easier to bring lawsuits (Ferguson 1994). Thus, the effect on equity value of any change in litigation regime is likely to be higher in the post-1988 period, and therefore easier to detect. Second, the measures of litigation risk proposed in the literature are based on recent sample periods. For example, the sample period of O'Brien and Hodges (1993) is 1988-1993, of Francis et al. (1994a, 1994b) is 1988-1992, and of Jones and Weingram (1996) is 1991-1992. Thus, the proposed litigation-risk proxies may not be valid before 1988.
(20) Based on Jones and Weingram (1996, Tables 3 and 12) the estimated litigation-risk model with statistically significant variables is as follows: LITGRISK = 6.46 STNDTW - 0.13 SKEW + 0.05 MKTCAP + 3.86 ECTURN - 1.28 ANRET + 0.22 BETA.
(21) Both ST and JKN confine their analyses to firms in high-litigation-risk industries.
(22) We perform the following sensitivity checks. First, the observed association between our abnormal return measure and LITGRISK, which is a function of market capitalization, could be attributable to a lack of control for firm-size in estimating abnormal returns (Banz 1981). To rule out this alternative explanation we estimate the Sefcik and Thompson (1986) portfolio OLS regression models in Table 7 with market capitalization as an additional explanatory variable. The signs and significance levels of the coefficients on the litigation-risk variable, LITGRISK, remain similar to those reported in Table 7--for dependent variables PSLRAXR, P211XR, and CBDXR, t-statistics for the coefficient on LITGRISK are -2.62, -1.87, and, -2.60, respectively. Second, the rank correlation between LITGRISK and each of the return measures, PSLRAXR, P211XR, and CBDXR, is significantly negative at better than the 1 percent level. Finally, we consider cumulative abnormal returns for all PSLRA event dates with significant abnormal returns (see Table 1), instead of PSLRAXR. The portfolio OLS regression procedure estimates the coefficient on LITGRISK to be -0.49 with a t-statistic of -3.90.
(23) The Spearman correlation between abnormal returns on 12/18/95 (the veto rumor date) and LITGRISK is insignificant (p = 0.48), further supporting our argument that high-litigation-risk firms did not exhibit significant abnormal returns on 12/18/95.
(24) Details of the estimation may be obtained from the authors upon request.
REFERENCES
Abate, T. 1996. Prop. 211 opponents to push on: Leaders say win makes Silicon Valley an emerging force in state politics. San Francisco Examiner (November 6): A25.
Avery, J. 1996. Securities litigation reform: The long and winding road to the Private Securities Litigation Reform Act of 1995. The Business Lawyer 51 (February): 335-378.
Banz, R. 1981. The relationship between return and market value of common stocks. Journal of Financial Economics 9 (1): 3-18.
Bencivenga, D. 1996. War over class actions: California ballot seeks to undo reforms. New York Law Journal (October 10): 5.
Bernard, V. L. 1987. Cross-sectional dependence and problems in inferences in market based accounting research. Journal of Accounting Research 25 (2): 1-48.
Binder, J. 1985. Measuring the effects of regulation with stock price data. Rand Journal of Economics 16 (2): 167-183.
Carleton, W., M. Weisbach, and E. Weiss. 1996. Securities class action lawsuits: A descriptive study. Arizona Law Review 38 (2): 491-511.
Collins, D. W., M. S. Rozeff, and W. K. Salatka. 1982. The SEC's rejection of SFAS No. 19: Test of market price reversal. The Accounting Review 57 (1): 1-17.
Corcoran, E. 1996. California voters reject Proposition 211: Silicon Valley fought measure making shareholder fraud suits easier. Washington Post (November 7): D3.
Donovan, K. 1996. War over securities bill heats up in California. National Law Journal (October 21): A6.
Dow Jones News Service. 1995a House override attempt: Two-thirds majority seen. 10:59 AM. (December 20).
--. 1995b. House overrides veto: Senate fight likely. 1:14 PM. (December 20).
--. 1998a. Senate passes bill to stop state securities class actions. 7:40 PM. (May 13).
--. 1998b. House votes to restrict investor suits in state courts. 5:30 PM. (July 22).
--. 1998c. Senate clears national standard for shareholder suits. 6:42 PM. (October 13).
--. 1998d. Congress adopts securities litigation reform bill. 7:08 PM. (October 13).
Dunbar, F., V. Juneja, and D. Martin. 1995. Shareholder litigation: Deterrent value, merit, and litigants' options. Working paper, Washington University Business, Law, and Economics Center.
Ferguson, T. W. 1994. A contract to restore contract to securities law. Wall Street Journal (October 25): A17.
Foster, G. 1980. Accounting policy decisions and capital market research. Journal of Accounting and Economics 2 (1): 29-62.
Fournier, R. 1995. Clinton vetoes bill to restrict securities fraud suits. Associated Press (December 20).
Francis, J., D. Philbrick, and K. Schipper. 1994a. Shareholder litigation and corporate disclosures. Journal of Accounting Research 32 (2): 137-164.
--, --, and --. 1994b. Determinants and outcomes in class action securities litigation. Working paper, University of Chicago.
Frisby, M. K. and J. Taylor. 1995a. Clinton likely to sign suit limits. Wall Street Journal (December 12): A3.
--, and --. 1995b. Clinton vetoes bill limiting securities suits. Wall Street Journal (December 20): A3.
--, and --. 1995c. Senate poised to override securities veto. Wall Street Journal (December 22): A3.
Glassman, J. K. 1996. Stock market scenarios in search of an election. Washington Post (November 3): H1.
Grundfest, J., and M. Perino. 1997. Securities litigation reform: The first year's experience. Working paper, Stanford Law School.
Gunnison, R. B. 1996. Prop. 209 lead dwindles, poll also shows Prop. 211 trailing for the first time. San Francisco Chronicle (October 30