1 u. .L.

x: :t

D: A. .
2.... «u h!’
5P5...) 8.)... .
.1... "Pinyin" 411‘s}! ; I
v 2111!... t.

.1 3. \.
‘ ‘QﬁJIC‘l‘lI

al .I .iaanJOazsrinl
* 4.. $5.? .V _ .. ,. - nang‘

‘ l .. ,. 1 1:3... 3 , , . aha?

.nﬁmwi. awggﬁmimxnmuwwﬁkwm 34“: 4:35 {Lip m. , ,

. J“ .,

.

.3
1.5.6.

.3 Hz!»-

Jhuhﬁ. .

as:

h-
. : ..
uﬁvﬂWnpdz
.,..

3575!. van 51..

5513.0»,

,1.
.

 

.100 6

 

LIBRARY
Michigan State
University

 

 

 

This is to certify that the
dissertation entitled

THE INTERACTION OF LAW AND CORPORATE
DECISIONS

presented by
YUN ZHU

has been accepted towards fulﬁllment
of the requirements for the

Doctoral degree in Finance

(($759: , MW KW

Major Professor’s Signature
.Jun 1 Z 7 ) Z00 S

Date

MSU is an Afﬁrmative Action/Equal Opportunity Institution

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

2/05 c/C_-c_©/Iat———IR m.m.15‘

THE INTERACTION OF LAW AND CORPORATE DECISIONS

Bv

d

Yun Zhu

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment ofthe requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Finance

2005

ABSTRACT
THE INTERACTION OF LAW AND CORPORATE DECISIONS
By
Yun Zhu

This dissertation studies the interaction of law and corporate decisions. The ﬁrst
essay investigates the litigation risk hypothesis (LRH) ofthe Initial Public Offering (IPO)
underpricing puzzle based on evidence from the Private Securities Litigation Reform Act
of I995 (PSLRA). Lowry and Shu (2002) construct a simultaneous equations system and
ﬁnd evidence supporting the LRH. On one hand, issuers expecting higher litigation risk
will underprice their [PCs more as a form of insurance against future lawsuits (insurance
effect). On the other hand. issuers with higher IPO underpricing have a better chance of
deterring lawsuits (deterrence effect). Using the passage of the Private Securities
Litigation Reform Act in I995 as a stronger instrumental variable than the one used in
their studies, we ﬁnd empirical results countering the insurance effect of the LRH and
ﬁnd that the reported systematic link between litigation risk and lPO underpricing is
mainly driven by the choice of their instrumental variable. Furthermore, using a slightly
different sample period than theirs, we ﬁnd no signiﬁcant deterrence effect of the LRH
either prior to or following the enactment of the PSLRA. Their results are not persistent
in our sample period. In summary, our results are against the LRH for both insurance
effect and deterrence effect. We additionally show that the PSLRA did reduce IPO ﬁrms‘
litigation risk and that there is less “deep pocket“ digging and less “racing to the

courthouse” in the post-PSLRA period.

In the second essay. we study the role that the legal environment might play in
recent corporate scandals. Some legislators believe the PSLRA, instead of deterring
frivolous lawsuits as it was intended to, deterred meritorious lawsuits and resulted in
many of the recent corporate frauds. They further argue that the Sarbanes-Oxley Act of
2002 (SOA) is an important counter to the PSLRA. Using several event study
approaches, we investigate shareholder reaction to a series of legislative events related to
the passage of the PSLRA and the SOA. From the shareholders’ point of view, we ﬁnd
both statistically and economically signiﬁcant negative abnormal returns during the
passage of the PSLRA in the group of high litigation risk ﬁrms that are more likely to
face meritorious lawsuits; conversely, there is no signiﬁcant stockholder reaction to the
passage of PLSRA in the group of high litigation risk ﬁrms which are likely to face
frivolous lawsuits. Furthermore, shareholders of both groups react negatively and
signiﬁcantly to the passage of the SOA. Our results suggest that shareholders believed the
PSLRA would deter meritorious lawsuits instead of frivolous lawsuits as it was intended
to. In addition, the PSLRA reduces the important legal monitoring achieved by
meritorious lawsuits and increases the incidence of fraud. Even more disappointing,
shareholders did not believe the recently passed SOA would counter the effect of the

PSLRA by bringing better protections for investors against corporate fraud.

Copyright by
Yun Zhu
2005

To my parents and husband

ACKNOWLEDGMENTS

I would like to express my gratitude to the members of my dissertation committee
— Dr. C. Edward Fee. Dr. Naveen Khanna, and Dr. Jennifer Marietta-Westberg - for their
insightful comments. continuous support and tremendous commitment of time. I am
especially indebted to my dissertation chairs, Dr. C. Edward Fee and Dr. Naveen Khanna.
Their knowledge, guidance, encouragement, patience, and trust helped me through the
toughest time of my PhD life and led me towards the ﬁnish of this dissertation. I am
grateful to Dr. Jennifer Marietta-Westberg for her help and dedication of time. I cannot
imagine having a better dissertation committee. I also thank Dr. Jeffrey Wooldridge of
the Economics department. Dr. Elliot Spoon of the College of Law, and Dr. James
Marshall for their guidance during the early stage of this dissertation.

I want to thank the Department Chair, Dr. Geoffrey Booth, for accepting me into
the doctoral program and helping me in every way possible throughout the ﬁnance
program. I also appreciate the knowledge and wisdom passed on to me by all the faculty
members in the Finance Department, in particular, Dr. Richard Baillie, Dr. Kirt Butler,
Dr. Long Chen, Dr. Zsuzsanna F luck, Dr. Charles Hadlock, Dr. Jun-Koo Kang, Dr. Wei-
Lin Liu, Dr. Mark Schroder, and Dr. Richard Simonds. I also want to thank my fellow
. doctoral students for their friendship and support during my stay in the program.

I would like to thank my parents for their continued love and support. Last, my _
deepest gratitude goes to my husband, Erman Wang, for his love. understanding,
optimism, encouragement and sacriﬁces. which have been necessary for my completion

of the doctoral program and which always help me to continue when I feel discouraged.

vi

TABLE OF CONTENTS

LIST OF TABLES .......................................................................................................... IX

CHAPTER I

IS IPO UNDERPRICING RELATED TO LITIGATION RISK? EVIDENCE
BASED ON THE PRIVATE SECURITIES LITIGATION REFORM ACT OF 1995

............................................................................................................................................. l
i . I INTRODUCTION .................................................................................................... l
I .2 RELATED LITERATURE ......................................................................................... 5
1.2.1 Litigation risk hrtpothesis ............................................................................ 5
1.2.2 The Private Securities Litigation Reform .4 et of] 9.95 ................................ 7
L3 DATA ................................................................................................................. l0
1 . 3 .1 [PO class action lawsuit decisions and the impact of the PSLRA ............ I 3
1.3.2 Descriptive statistics of sued and non-sued [1’0 ﬁrms in the pre-PSLRA
period vs. post-PSLRA period ................................................................................... l 4
I .4 METHODS .......................................................................................................... 20
1.4.] T wo-stage estimation method introduced by Maddala (1983) 22
1.4.2 Generated instrument approach introduced by Wooldridge 2()01,)......... 23
1.4.3 Instrumental variables and control variables ........................................... 24
1.5 EMPIRICAL RESULTS .......................................................................................... 32
1.5.1 Regression results using the simultaneous equations system ................... 32
l.5.l 1 Insurance effect results ......................................................................... 32
l.5.l .2 Deterrence effect results ....................................................................... 34
1.5.2 Sensitivity analysis .................................................................................... 36
1.6 CONCLUSION ...................................................................................................... 41
CHAPTER 2
A MICRO EXAMPLE OF LAW AND FINANCE: A STUDY OF THE PRIVATE
SECURITIES LITIGATION REFORM ACT OF 1995 AND THE SARBANES-
OXLEY ACT OF 2002 ................................................................................................... 42
2.l INTRODUCTION .................................................................................................. 42
2.2 RELATED LITERATURE ....................................................................................... 45
2. 2.1 The Private Securities Litigation Reform Act (if/995 .............................. 45
2.2.2 The Sarbanes-Oxley Act of 2002 ............................................................... 48
2.3 RIasI-zARCH DESIGN ............................................................................................ SI
2. 3.] Measure of Litigation Risk Level .............................................................. 5]
2. 3.2 Measure of Earnings Management ............................................................ 5 2
2. 3. 3 llvleasure of Shareholder Rights ................................................................ 5 4

vii

2. 3. 4 Legislative Event Dates ............................................................................. 56

 

2.4 EVENT STUDY METHODS AND EMPIRICAL RESUI. Is .......................................... 58
2. 4. 1 Cumulative Abnormal Return Approach .................................................. 58
2. 4.2 Equal- Weighted Portfolio OLS Approach ................................................ 62
2. 4.3 Generalized Least Square Approach ........................................................ 66

2.5 CONCLUsION ...................................................................................................... 72

APPENDIX 1

TABLES OF CHAPTER 1 ............................................................................................. 73

APPENDIX 2

TABLES OF CHAPTER 2 ....................................................... -- 90

REFERENCES ...... - - - - - 115

 

viii

LIST OF TABLES

TABLE 1.1 DISTRIBUTION OF SAMPLE FIRMS ACROSS YEARS ............. 74

TABLE 1.2 LAWSUIT STATUS AND RESOLUTION CHARACTERISTICS 75

TABLE 1.3 DESCRIPTIVE STATISTICS OF SUED VS. NON-SUED IPOS
BEFORE AND AFTER THE ENACTMENT OF THE PRIVATE SECURITY
LITIGATION REFORM ACT OF 1995 -- 76

 

TABLE 1.4 DESCRIPTIVE STATISTICS OF INSTRUMENTAL VARIABLES

 

 

AND MATCHED SAMPLE PROXIES ................ - -- -- - 79
TABLE 1.5 INSURANCE EFFECT REGRESSION RESULTS - 82
TABLE 1.6 DETERRENCE EFFECT REGRESSION RESULTS ...................... 84

TABLE 1.7 REGRESSION RESULTS OF THE DETERRENCE EFFECT IN
THE PRE- AND POST-PSLRA PERIODS ....... -- - - - - - - 86

 

TABLE 1.8 PERFORMANCE COMPARISON OF PSLRA AND MATCHED

 

TURNOVER 88
TABLE 2.1 DISTRIBUTION OF EARNING MANAGEMENT LEVEL ........... 91
TABLE 2.2 . DISTRIBUTION OF SHAREHOLDER RIGHTS LEVEL .............. 94
TABLE 2.3 LEGISLATIVE EVENT DATES 96

 

TABLE 2.4 COMPARISONS OF PORTFOLIO CARS AROUND PSLRA AND
SOA EVENT DATES - - 99

 

TABLE 2.5 RESULTS OF EQUAL-WEIGHTED PORTFOLIO OLS
APPROACH FOR DIFFERENT EM AND HIGH LITIGATION RISK GROUPS
101

 

TABLE 2.6 RESULTS OF EQUAL-WEIGHTED PORTFOLIO OLS
APPROACH FOR DIFFERENT SHAREHOLDER RIGHTS AND HIGH
LITIGATION RISK GROUPS ..... ........ - - - 104

 

dlit

lav

Chapter 1

IS IPO UNDERPRICING RELATED TO LITIGATION RISK? EVIDENCE

BASED ON THE PRIVATE SECURITIES LITIGATION REFORM ACT OF 1995

1.1 Introduction

Initial public offerings (lPOs) typically experience a signiﬁcant increase of the
ﬁrst day closing price from the offer price. Loughran and Ritter (2004) document a l5%
ﬁrst-day return during I990-l998. As summarized by Ritter (2003), there are at least
eight potential explanations for [PC underpricing including dynamic information
acquisition. prospect theory, corruption, the winner's curse, informational cascades,
signaling, marketing event, and litigation risk. This paper concentrates on the discussion
of the litigation risk hypothesis (LRH).

The debate on the LRH has persisted for decades. Supporters argue that issuers
and investment banks may use IPO underpricing as insurance against possible future
lawsuits, because lowering the offer prices reduces both the probability of being sued and
the potential damages that plaintiffs can recover (lbbotson (1975). Tinic (1988), Hughes
and Thakor (I992), Hensler (1995), and Lowry and Shu (2002) etc.). Opponents are
unconvinced whether lPO underpricing actually deters lawsuits given that these lawsuits
are typically triggered by unfavorable (and probably unexpected) news in the aftermarket,
regardless of offer price level. They also argue that underpricing seems too costly to be
used as insurance against lawsuits, since the frequency of lPO lawsuits, as well as the

size ofthe settlement, tends to be low (lbbotson. Sindelar. and Ritter (I988), Alexander

(le

pink

(I993), Drake and Vetsuypens (I993). Keloharju (I993), Prabhala and Puri (I999). and
Ritter (2003) etc.). This paper belongs in the group of opponents.

The LRH predicts that ﬁrms with higher litigation risk should have higher IPO
underpricing in general. On December 22, I995. Congress enacted the Private Securities
Litigation Reform Act of 1995 (PSLRA), which adopted a series of procedural hurdles to
securities fraud class actions (described in detail later). Evidence has been shown that this
law reduced ﬁrms’ litigation risk (Grundfest and Perino (I997), Beatty, Drake, and
Hogan (2002), Perino (2003), Bajaj, Mazumdar, and Sarin (2003) etc.). Interestingly,
instead of a decrease in IPO underpricing as predicted by the LRH, Loughran and Ritter
(2004) ﬁnd an increase in IPO underpricing in the years I995 and after. According to
their numbers, the average underpricing in every year from I995 to I998 was higher than
that in any year between WM and I994. Though this observation alone is not deﬁnitive
evidence against the LRH. it does serve as a motivation for the empirical tests of this
papen

Our tests of LRH are based on the simultaneous equations system used in Lowry
and Shu (2002). They argue that IPO ﬁrms’ litigation risk and IPO underpricing are
actually two interdependent endogenous variables. The relationship between these two
can be expressed through two effects: the insurance effect and the deterrence effect. On
one hand, issuers expecting higher litigation risk will underprice their IPOs more as a
form of insurance against future lawsuits. On the other hand, issuers with higher IPO
underpricing have a better chance of deterring lawsuits. They ﬁnd strong evidence
supporting both effects using the two-stage estimation method suggested in Maddala

(I983). Our tests differ from theirs in four major respects. First, we use a different

instrumental variable in testing the insurance effect. The choice of the instrumental
variable is very critical in the two-stage estimation. We show that the one they use
(matched ﬁrms' stock turnover) may lead to biased regression results because it may be
strongly related to the investor interest in IPO aftermarket trading, which is likely to be
positively related to [PO underpricing. Fortunately. the passage of the PSLRA of 1995
offers us a more appropriate one. which is signiﬁcantly correlated with ﬁrms’ litigation
risk and completely exogenous in estimating IPO underpricing. Second, in addition to the
two-stage estimation method, we use an alternative generated instrument approach
developed in Wooldridge (200]). We argue that this method is better in terms of
coefﬁcient estimation and standard error and test statistics adjustments. Third. our sample
of sued IPOs includes all lPO-related class action lawsuits, which include those ﬁled
under Section II or 12 of the Securities Act of 1933 (I933 Act) as well as those ﬁled
under Rule l0b-5 ofthe Securities Exchange Act of 1934 (1934 Act). Though Lowry and
Shu (2002) include only cases ﬁled under Section I l of the Securities Act of I933. we
argue that all three laws should be included in identifying an IPO-related class action
lawsuit from both legal and technical perspectives. Fourth, our tests cover the post-
PSLRA period when the legal environment is signiﬁcantly changed.

Due to the usage of the two-stage estimation method and the new instrumental
variable, we actually test both the signiﬁcance of the LRH and the impact of the PSLRA
of I995 simultaneously. Therefore, this paper contributes to the literature in two ways.
First, our results offer evidence against the LRH of [PO underpricing. We ﬁnd results
opposite to earlier research by using a new exogenous instrumental variable, a new

methodology. a new sample selection standard. and a new sample period. Second, we test

CL

in"

the effectiveness ofthe PSLRA on IPO ﬁrms' litigation risk using a multivariate test after
controlling for IPO underpricing. We also test the sued and non-sued ﬁrms' characteristic
differences in both the pre- and post-PSLRA periods. Both of these two approaches
disclose the impact of the PSLRA better than prior research does.

There are four major results of this paper. First, our empirical results counter the
LRH of IPO underpricing for both insurance and deterrence effects. Issuers expecting
higher litigation risk do not underprice their IPOs more. Moreover, issuers with higher
IPO underpricing do not have an advantage in deterring lawsuits. Second. we show the
signiﬁcant insurance effect found in the earlier research is driven by their choice of
instrumental variable. The signiﬁcance disappears after we use a new and stronger
instrument. Third. we further ﬁnd no signiﬁcant deterrence effect either prior to or
following the enactment of the PSLRA. The deterrence effect found in the earlier
research is not there in our sample period. Fourth, the PSLRA of 1995 does reduce the
IPO ﬁrms’ litigation risk in our sample period (I990 to I998). We ﬁnd less “deep
pocket" digging and less “racing to the courthouse” in the post-PSLRA period.

The rest of this chapter is organized as follows. Section 1.2 further discusses the
related literature about the LRH and the PSLRA. Section 1.3 describes the data selection
process and provides descriptive statistics of sued and non-sued ﬁrms‘ characteristic
differences in both pre- and post-PSLRA periods. Section 1.4 discusses the simultaneous
equations system and alternative testing approaches. Section 1.5 presents evidence on the
impact of the PSLRA and the relationship between IPO underpricing and litigation risk.

Section L6 is the conclusion.

1.2 Related literature
1.2.1 Litigation risk hypothesis

lbbotson (I975) ﬁrst suggests the hypothesis that issuers and investment banks
may use IPO underpricing as insurance against possible future lawsuits. Tinic (1988)
models expected legal liabilities as the increasing function of IPO offer price. Hughes and
Thakor (I992) analyze the trade-off between expected litigation cost and current revenue
loss from underpricing in a game theory model, while Hensler (1995) studies it in a
single-period setting. Both models predict a positive relationship between expected
litigation risk and IPO underpricing.

Tinic (I988) further supports the LRH by showing that average IPO underpricing
has significantly increased following the enactment of the Securities Act of I933, which
is assumed to have increased ﬁrms‘ litigation risk. On the other hand, Prabhala and Puri
(1999) ﬁnd that the ﬁndings in Tinic (1988) can be explained by differences in IPO risk
without recourse to possible differences in litigation risk. Drake and Vetsuypens (I993)
ﬁnd that IPO underpricing of sued ﬁrms is the same as that of non-sued ﬁrms. They
conclude that underpricing does not deter lawsuits, which. are typically triggered by
unfavorable news in the aftermarket. In addition, Alexander (1993) opposes the lawsuit
avoidance theory by arguing that the low frequency of IPO lawsuits and low settlement
costs compared to the big loss from underpricing barely make litigation avoidance an
important justiﬁcation for IPO underpricing. In Finland, where IPO issuers have almost
no legal liabilities, Keloharju (I993) nevertheless ﬁnds signiﬁcant IPO underpricing. He

thus concludes that IPO underpricing cannot be explained by the lawsuit avoidance

U1

theory. Additionally. lbbotson. Sindelar. and Ritter (I988) and Ritter (2003) point out
that underpricing seems too costly to be used as insurance against lawsuits.

In a recent paper, Lowry and Shu (2002) argue that none of the previous research
has taken care of the endogeneity of IPO underpricing and litigation risk. They examine
the LRH from two angles. First, issuers expecting higher litigation risk might use greater
IPO underpricing as insurance against future lawsuits, “the insurance effect”. Second,
more highly underpriced IPOs may deter more lawsuits. “the deterrence effect”. Using a
simultaneous equations system, they ﬁnd both signiﬁcant insurance and deterrence
effects. Ljungqvist (2004). however. has concerns about the validity and strength oftheir
choice of instrumental variable. With a stronger instrumental variable. this paper ﬁnds
evidence against the LRH.

From a legal perspective, supporters argue the LRH is sound and logical. IPO-
related class action lawsuits are mainly brought under Section I l and I2 of the Securities
Act of 1933 and Rule lOb-S of the Securities Exchange Act of I934. For all the Section
II cases, plaintiffs’ damages are calculated based on the difference between IPO offer
price and the price on lawsuit ﬁling day or sale price. Therefore. lowering the offer price
is likely to decrease both the probability of being sued and potential damages. Alexander
(I993), however, points out that this argument ignores the fact that “investors can also —
and almost invariably do — bring claims under the I934 Act along with claims under the
I933 Act”. The availability of the I934 Act makes the LRH practically ineffective
because the damages under Rule 10b-5 are based on shareholders’ purchase price, which
can be either the offer price or the aftermarket price. As a result. regardless of how much

issuers underprice their lPOs. they can be sued as long as the aftermarket price is high

enough to cover the plaintiffs‘ and lawyers’ legal costs. It may be argued that lowering
the offer price can at least reduce some damages in lawsuits; nevertheless, it is debatable

whether these gains can compensate for the big proceeds typically lost in underpricing.

1.2.2 The Private Securities Litigation Reform Act of 1995
On December 22, I995, Congress enacted the Private Securities Litigation
Reform Act of I995, which amended the Securities Act of I933 and the Securities
Exchange Act of I934. This new law adopted a series of procedural obstacles to
securities fraud class actions to discourage abusive lawsuits, thereby reducing ﬁrms’
litigation risk. These obstacles include the following:
o The replacement of “joint and several liabilities” by “proportionate liability”
reduces the plaintiffs’ incentive of looking for “deep pockets”;
- Plaintiffs’ stay of discovery in a pending motion to dismiss reduces ﬁrms’
defending costs;
0 A new lead plaintiff appointment process and the prohibition of bonus payment to
lead plaintiffs stop lawyers from hiring “professional plaintiffs”;
o A heightened pleading standard increases the cost of ﬁling lawsuits;
o A new 90-day bounce back rule stops cases from rushing into courts;
0 The mandatory sanction of Rule l l of the Federal Rules of Civil Procedure deters

any party from making frivolous lawsuits;

o The comprehensive disclosure of the provisions of settlement and attorney fees

and the limitation on attorney fees to a "reasonable percentage” of any damage
. . I
award reduce attorneys proﬁts from lawsuits.

Evidence of decreased litigation risk following the PSLRA exists in both court

practices2 and academic research. The debate on the effectiveness of the PSLRA mainly
concentrates on two aspects: whether the PSLRA decreases ﬁrms’ litigation risk and

whether the PSLRA deters meritorious along with frivolous cases. This paper is more

related to the ﬁrst issue.3 Opponents such as Grundfest and Perino (I997) think the
PSLRA has no effect on ﬁrms’ litigation risk since plaintiffs can simply shift lawsuits to
state courts. They ﬁnd that about 26% of class action claims moved to state courts, but
they still ﬁnd a 7% drop in total litigation volume. Bajaj, Mazumdar, and Sarin (2003)
ﬁnd there is an immediate drop in the number of federal cases, down from 19] in I995 to
I 19 in I996. Although the number of state cases increases from 65 to 80 during the same
years, there is still an overall drop in total cases. They also ﬁnd this shift tendency
diminishes very quickly. The number of federal cases promptly rises in both 1997 and
I998 while the number of state cases continues to drop. Perino (2003) uses data from

I996 to 200i and draws a similar conclusion.

 

| Another important revision by the PSLRA is a “safe harbor" for forward-looking statements, which
protects ﬁrms from lawsuits aimed at forward-looking statements. This hurdle, however. does not apply to
IPO—related lawsuits.

2 Courts appear to apply the PSLRA in their decisions. In Steckman. et al. v. Ilart Brewing. Inc. et al , the
court dismissal order states, “The Court also notes the impact of the...Reform Act of I995 on this
decision...Through this Act, Congress has encouraged the use of motions to dismiss in certain securities
cases. One of the salient features of the I995 Reform Act is...to protect against the abusive practice of
ﬁling lawsuits like this one ‘with only faint hope that the discovery process might lead eventually to some

lausible cause ofaction.’...” This case was dismissed with prejudice on December 24. I996.

The merits issue is discussed in Avery (I996), Spiess and Tkac (I997), Prichard (I999), Johnson.
Kasznik, and Nelson (2000), Ali and Kallapur (200i ), and Johnson, Nelson. and Pritchard (2000. 2002) etc.

So far. all studies listed above are univariate tests. Since the PSLRA is designed
to reduce ﬁrms’ litigation risk, it may change the characteristics of ﬁrms going public.
Particularly. some small and risky companies who prefer staying private may go to the
IPO market with reduced litigation risk. Without controlling for ﬁrms’ characteristics.
these univariate comparisons cannot tell much about litigation risk change. Although
some of them have studied sued ﬁrms’ characteristic changes after the PSLRA, none
have used non-sued ﬁrms as a benchmark to control for ﬁrms’ characteristic movement.
To our knowledge. Beatty, Drake. and Hogan (2002) are the only ones who conduct both
univariate and multivariate tests. They ﬁnd a decreased litigation risk of IPO ﬁrms
following the PSLRA. They, however, do not control for IPO underpricing in their
regressions. In this paper, we test sued and non-sued ﬁrms’ characteristic differences in
both pre- and post-PSLRA periods. Furthermore, we offer a multivariate test on the
effectiveness of the PSLRA on IPO ﬁrms’ litigation risk after controlling for IPO

underpricing.

L;

\L

Li

1.3 Data
Data on US. common stock initial public offerings issued between 1990 and

I9984 are collected from the Thomson Financial Securities Data (SDC) database. Closed-
end funds, unit offerings, REITs, ﬁnancial ﬁrms, reverse LBOs, ADRs, spin-offs, lPOs
with an offer price below $5.00 per share, ﬁrms without exchange information in SDC
and ﬁrms listed in the small capital market or the OTC market are excluded.

The list of federal class action lawsuits is obtained from the Security Class Action
Alert newsletter (SCAA) between January I990 and August 2002 as having been sued
under the 1933 Act and/or the 1934 Act. We choose the SCAA because it is the only
resource covering our entire sample period. Other case information has been collected by
searching the Public Access to Court Electronic Records service (PACER)
[http://pacer.psc.uscourts.gov], the Stanford Securities Class Action Clearinghouse
[http://securities.stanford.edu], the Gilardi and Co. class action administration website
[http://www.gilardi.com/allcases.htmI], company SEC ﬁlings. Lexis-Nexis newswire and
Lexis-Nexis legal research.

Because of data availability limitations, our sample includes only cases ﬁled in
federal courts. Although this may slightly bias our results, we believe the bias is minor

for several reasons. As introduced in the last section, the overall lawsuit rate still

 

4 Our sample does not include lPOs issued in the years 1999 and 2000 for the following reasons. Loughran
and Ritter (2004) deﬁne these two years as the Internet bubble period, which has a 65% average IPO
underpricing, more than 4 times the 15% in the I990-I998 period. In addition, the stock market downturn
in 200l brought more than 300 “IPO allocation” lawsuits, which is more than double the total number of
IPO lawsuits in our entire sample period. These extreme numbers and events may obscure the relationship
between litigation risk and IPO underpricing as well as the effectiveness of the PSLRA. Beatty, Drake, and
Hogan (2002) do not include any of the “IPO allocation” lawsuits because they argue that “IPO allocation
lawsuits are claims alleging that underwriters engaged in undisclosed practices in connection with the
distribution of IPO shares, rather than claims of fraud ﬁled against the IPO ﬁrms.“ Hao (2004) studies the
LRH with these “IPO allocation” lawsuits only. We offer more discussion on this issue in Section I.5.

l0

decreases signiﬁcantly after the enactment of the PSLRA, although some cases shift to
state courts. Furthermore, state security laws are characterized by a great diversity of
language and interpretation and they apply only to transactions within states (Ratner
(1998), p. 9). Damages that can be recovered from state courts are much smaller than
those from federal courts. which naturally deters case shifts. In addition, the majority of
IPO related cases are ﬁled in federal courts because lPOs in our sample are typically
offered and traded nationally or worldwide.5

An IPO ﬁrm is identiﬁed as sued ifthe class action lawsuits are related to its IPO,
as shown in case documents. These cases are generally ﬁled under Section II and 12 of
the 1933 Act, Rule 10b-5 of the 1934 Act. or both. Lowry and Shu (2002) argue that the
LRH is more applicable to Section 11 cases because damages under Section II are
directly related to the offer price; however, we disagree on this point. Damages under
Section 12 and Rule 10b-5 are related to investors‘ purchase prices, which in fact do
cover the offer price.

First, in our sample, there is not a single case ﬁled under Section 12 that is not
also ﬁled under Section 11. Therefore, excluding Section 12 cases is irrelevant in this
paper, though we argue Section 12 does cover the offer price.6 Second, Rule 10b-5

applies to any purchase or sale by any person of any security. According to Ratner

 

5 . . .

Beatty, Drake, and Hogan (2002) Identify only 13 IPO-related state cases from 1991 to 1999. Given that
their IPO sample size is much larger than ours is, we believe there will be no signiﬁcant amount of state
cases leﬁ after our stricter screening process.

6 Section 12 applies to any person who “offers or sells” a security by means of a prospectus or oral
communication which contains a material misstatement or omission. In an IPO-related case. underwriters
are usually the defendants under Section 12 because “liability under Section 12 extends to any person who
solicits the purchase” (Ratner (1998), p. 88-90). Consequently, Section 12 applies to those investors
purchasing from underwriters in offering. and their purchase prices do include the offer price. Technically,
Section I I and 12 are almost inseparable for IPO-related cases since underwriters are also defendants under
Section I 1 as long as they sign the registration statement.

11

Il‘.

I”>‘
. \\

III 0}
\\ L

Iii

slal

(1998), the "sale” includes the initial public offering and Rule 10b-5 does not exclude
. 7 . - .
Section 11. As a result. an IPO-related suIt could be brought under Rule 10b-5 against
the same defendants as Section I l and 12 claims. including both issuers and underwriters

(Alexander (1993)), and Rule 10b-5 does cover shares bought at the offer price.8 In
summary, we believe all three laws relate to the IPO offer price; therefore. our sample
includes all IPO related class action lawsuits under any of these three laws.

Table 1.1 lists the number of lPOs, the number and percentage of IPO-related
class action lawsuits. and the average months between offering month and lawsuit ﬁling
month by issue year. In our sample, there were 2.681 lPOs issued from 1990 to 1998,
where 1,605 were issued before the PSLRA enactment and 1,076 after. Due to the 3-year
statute of limitation, the lawsuits related to 1993-1995 lPOs could be ﬁled before or after
the PSLRA enactment. Consequently, they may or may not be affected by the PSLRA
since this law only applies to the cases ﬁled after December 22, 1995 (Lerach (1998)). To
avoid this issue, we exclude 1994 and 1995 lPOs from our ﬁnal sample. 1993 lPOs are
still included because the latest 1993 IPO case was ﬁled before the enactment.9 The ﬁnal
sample includes 946 lPOs issued between 1990 and 1993, and 1,076 lPOs issued between
1996 and 1998. Among them, 6.8% of the 1990-1993 IPOs and 4.7% of the 1996-1998

lPOs were sued. Overall, the lawsuit rate dropped by 2.1% following the enactment.

 

7 “The court held that the issuance by a corporation of its own shares was a "sale’ under Rule 10b-5.
[...]The court has also held that suit can be brought under Rule 10b-5 to recover damages resulting from
misstatements in a 1933 Act registration statement, even though such misstatements give rise to a speciﬁc
right ofaction under Section II ofthe 1933 Act." (Ratner(l998), p. 140-143)

An example is Zishka et al. v. American Pad & Paper Company et al. Its court complaint states, “This is
a suit on behalf of purchasers of American Pad & Paper Company...stock between 7/2/96 and 12/ 17/97
(the ‘Class Period’), alleging violations of ...SEC Rule lOb-5.” Here, 7/2/96 was the IPO offering date.
One defendant of this case bought shares at offer price. while the other one bought at aﬁermarket price.
This case was ﬁled under Rule 10b-5, but not Section 1 l.

9 Results are statistically indifferent from those reported in this paper when we exclude 1993 lPOs in our
sample.

12

 

 

fur

l‘i“

GUI

 

 

d\;

6.1

l’)

 

Furthermore, there was a signiﬁcant decrease of 4.8% in the percentage of cases from
1993 to 1996. These results suggest that the PSLRA did deter many lawsuits. at least in
our sample period, which is consistent with earlier research. In total, we ﬁnd I 14 cases

(5.6%) in 1990-1998 (with 1994 and 1995 excluded).

1.3.1 IPO class action lawsuit decisions and the impact of the PSLRA
Table 1.2 lists case decisions and the average decision time length between

lawsuit ﬁling month and case decision month.'0 For settled cases, we further present the
average settlement amount and settlement/proceed ratio.

Out of 64 cases ﬁled prior to the PSLRA, 67.2% of these cases were settled, while
64% of the 50 cases ﬁled following the PSLRA were settled. Compared to this small
3.2% decline in settlement rate from the pre- to the post-PSLRA period. there was a big
11% decrease in settlement rate from 1993 to 1996. Again, the PSLRA shows the most
impact in the year 1996, which is similar to the ﬁndings of Bajaj, Mazumdar, and Sarin
(2003). In contrast to the decreased settlement rate, the overall dismissal/termination rate
increased 6.31% following the PSLRA. This result is opposite to Bajaj, Mazumdar, and
Sarin (2003) but consistent with Foster et al. (2000). ” Interestingly, the
dismissal/termination rate of 1996 IPO cases was 14.5% more than that of 1993 IPO
cases. For post-1996 IPO cases, the dismissal/termination rate dropped gradually again.

Consistent with other research, our results show the PSLRA did reduce ﬁrms’ litigation

 

'0 The case decision month is based on the date of District court ﬁnal judgment with prejudice, the date of
preliminary judgment, the date on the settlement document from the Gilardi and C 0. class action
administration web site, the date listed in ﬁrms’ SEC ﬁlings. or the case close date from court dockets.

H The sample of Bajaj, Mazumdar, and Sarin (2003) includes all securities class action lawsuits in all court
levels, which is different from our sample coverage. In addition. their sample ends in 1999 with many cases
still pending during the research while there are only two cases pending in our sample.

13

1,.
Um

III»

)4

5L

risk for a while. but lawyers and plaintiffs eventually found a new way to make their
cases successful.

The decision speed of cases was also changed by the PSLRA. The average
decision time length of settled cases dropped by 1.3 months following the enactment. In
contrast, it took longer for cases to be dismissed in the post-PSLRA period. The average
decision time length of dismissed/terminated cases after the enactment was 1.6 months

more than before. Foster et al. (2000) interpret this increase as reﬂecting the effect of

staying discovery while a motion to dismiss is pending.'2 The average settlement amount
increased $2.7 million following the enactment. The average settlement/proceed ratio
increased 8.5%. These increases, however, were mostly from the settlement results of
1998 IPO cases. In fact, 1996 IPO cases had low settlement amounts and low
settlement/proceed ratios. In summary, our results support earlier studies. The PSLRA
did reduce ﬁrms" litigation risk. but it only did so for a few years, which also supports

our sample period selection.

1.3.2 Descriptive statistics of sued and non-sued IPO ﬁrms in the pre—PSLRA
period vs. post-PSLRA period

Descriptive statistics are shown in Table 1.3. Panel A compares IPO ﬁrms’
characteristics in the pre-PSLRA period to those in the post-PSLRA period. Panel B
compares the characteristics of sued and non-sued ﬁrms in both the pre—PSLRA period
and the post-PSLRA period. IPO ﬁrms’ characteristic data are from SDC and CRSP.

Most ofthe results in Panel A and the pre-PSLRA period of Panel B are consistent with

 

12 . . . . . . .
Under the PSLRA, a plaintiff cannot commence discovery during the pendency of any motion to dIsmIss

unless the court ﬁnds that particular discovery is necessary to preserve evidence or prevent undue
prejudice. 15 U.S.C. § 77Z-l(b)(3l

l4

6.17

and

 

\IL‘rI

 

n1) \

PSI

\\(1

sli

earlier ﬁndings; however, the results in the post—PSLRA period of Panel B show a
somewhat different story.

Initial return is deﬁned as the percentage change between the offer price (USPR)

and the ﬁrst closing price. '3 As Panel A shows. the average IPO initial return
signiﬁcantly increases by 5.2% following the enactment of the PSLRA. Panel B shows
no signiﬁcant difference between the initial returns of sued and non-sued ﬁrms in the pre-
PSLRA period. The average initial return of sued ﬁrms. however, is 21.7% larger than
that of non-sued ﬁrms in the post-PSLRA period. More underpricing seems not to deter
but to trigger lawsuits in the post-PSLRA period. Still, because of the simultaneity issues,
multivariate tests in section 1.5 give interpretations that are more appropriate.

Price update is deﬁned as the percentage change between the midpoint of the
ﬁling price range (MFILE) and the offer price from SDC. As shown in Panel A, the price
update level does not change much following the enactment. Panel B shows sued ﬁrms
have a higher price update than non-sued ﬁrms have both prior to and following the
enactment, but the test’s signiﬁcance nearly vanishes in the post-PSLRA period. In other
words, ﬁrms pricing more aggressively are more likely to be sued, but this likelihood
slightly decreases under the impact of the PSLRA.

Market capitalization is equal to the ﬁrst closing price multiplied by the number
of shares outstanding after offering in all markets (OUTPF) from SDC. Proceeds raised

equal the offer price multiplied by the shares offered in all markets (TOTSHSMIL) from

 

'3 Following Lowry and Schwert (2004). to determine the ﬁrst closing price of a particular issue, we use
the ﬁrst available closing price data from CRSP within 14 calendar days aﬁer the offering date. If CRSP
data are not available, we try to obtain the closing price on the ﬁrst trading day (PRSDAY) from SDC. If
that is still not available, initial return is equal to the percentage change between offer price and closing
price one day after offer (PCT 1 DAY) from SDC.

15

“.11

III

(II’

III;

€11

SDC.M Panel A shows that both size measures increase signiﬁcantly following the
enactment, which may be caused by stock market valuation uncertainty (Loughran and
Ritter (2004)). Panel B shows sued ﬁrms have $19.6 million more median market
capitalization and $3.9 million more median proceeds raised than non-sued ﬁrms have in
the pre-PSLRA period (both are signiﬁcant at the 1% level); however, the size
differences become much smaller in the post-PSLRA period. Following the enactment,
sued ﬁrms have only $4 million more median market capitalization and $1.1 million
more median proceeds raised than non-sued ﬁrms have (neither is signiﬁcant). As in
Grundfest and Perino (1997)”, our results suggest that plaintiffs’ and lawyers’ incentive
looking for a “deep pocket” gets much weaker under the impact of proportionate
liability” ofthe PSLRA.

The lead underwriter rank measure is the same as that used in Loughran and Ritter
(2004).17 Panel A shows that lPOs are underwritten by more prestigious investment
banks following the enactment. Panel B shows that lPOs underwritten by more
prestigious investment banks were sued more in the pre-PSLRA period. Following the

enactment, however, the underwriter rank difference between sued and non-sued ﬁrms is

 

4 . . . . . . . .
' Both market capitalization and proceeds raised are measured In millions of 1983 dollars usmg the
Consumer Price Index.

‘5 Grundfest and Perino (1997) do not compare the difference between sued and non-sued ﬁrms, and their
sample period only covers one year aﬂer the passage of the PSLRA.

‘6 Before the PSLRA, liability of defendant was "joint and several", which means each defendant was
liable for the entire amount of damages regardless of how much fault he or she was actually at. Therefore,
plaintiffs could recover all of their damages simply against one single "deep pocket" defendant. The
PSLRA adopted the proportionate liability, under which each defendant’s liability con'esponds to the
percentage of his or her responsibility, measured as a percentage of the total fault of all persons who caused
or contributed to the loss. As a result. plaintiffs’ incentive for going after “deep pocket” is much weaker
under the PSLRA.

'7 It is constructed in a way similar to Carter, Dark, and Singh ( 1998). The data is available on Dr. Ritter’s

web page [http://bear.cba.uﬂ.edu/ritteﬂipodatahtm]. We take the average when an IPO has more than one
lead underwriter.

LII

only 0.47, nearly half of the 09] prior to the enactment. The test signiﬁcance is also
gone. Once again, the “deep pocket” effect fades under the impact of the PSLRA.

Stock volatility is measured as the standard deviation of daily returns from CRSP
during the ﬁrst 25 days starting with the offering date (day 0 to day 24). Panel A shows
that IPO stock volatility signiﬁcantly increases in the post-PSLRA period. This could
reflect the higher uncertainty of the IPO market in the late 19905. In addition, there could
be more risky ﬁrms going public following the enactment due to the lower litigation risk.

Panel B shows no signiﬁcantly different stock volatility between sued and non-sued ﬁrms

prior to the enactment.'8 Following it, however, the average standard deviation of sued
ﬁrms is 1.6% higher than that of non-sued ﬁrms at the 1% signiﬁcance level. Given that
it is harder to bring a lawsuit under the new law, plaintiffs have to target those ﬁrms with
a much steeper price drop, who thus could have done something more obviously wrong.
Age is deﬁned as the difference between IPO offering year and the ﬁrm’s
founding year, following Loughran and Ritter (2004).'9 On average, IPO ﬁrms are 3.4
years younger following the enactment. This might suggest that underwriters are willing
to take younger ﬁrms to the IPO market with lower litigation risk under the PSLRA.

Panel B shows sued lPOs are younger than non-sued lPOs. The average age difference

 

'8 We choose a 25-day interval because the shortest time length between offering date and lawsuit ﬁling
date is 25 days in our sample. This time length is different from Lowry and Shu (2002). They calculate
stock volatility using daily returns over a one-year interval beginning one month after the offering date. As
Table 1.1 shows, a large fraction of lawsuits are ﬁled within one year after offering, so their volatility
measure may exaggerate the stock’s true volatility because lawsuits themselves will cause an increase in
stock volatility. Hence, we believe our measure is better. The results in Panel B are contrary to their
ﬁndings, which conﬁrms that their measure is affected by lawsuits.

'9 The founding year data is available on Dr. Jay Ritter’s web page
[http://bear.cba.uf1.edu/ritter/ipodatahtm]. This age measure is different from Lowry and Shu (2002), who
classiﬁes an IPO being ﬁve or more years old if it has ﬁve or more years of earnings information on SDC.
We think their classiﬁcation will bring in noises, because nearly 60% of our sample misses earning
information on SDC.

l7

il’lil

Ll~l1\

lr\x

between sued and non-sued lPOs is signiﬁcant at the 5% level prior to the enactment.
With a lower uncertainty level, older ﬁrms seem subject to a lower litigation risk.

IPO ﬁrms are labeled as high-tech (HIGHTECHCODE), venture capitalist backed
(VE), traded on the NYSE/AMEX (EXCHC), and with secondary offering in all markets
(TOTSECSHROMIL) according to SDC classiﬁcations.

More high-tech ﬁrms go public in the post-PSLRA period. High-tech ﬁrms are
subject to more uncertainties, which cause higher litigation risk. Consistent with this

prediction, we ﬁnd the high-tech proportion of sued ﬁrms is higher than that of non-sued

ﬁrms in both pre- and post-PSLRA periods.20 Slightly more IPO ﬁrms choose to be listed
on the NYSE or AMEX following the enactment. Furthermore, there is a very small
difference between sued and non-sued IPOs in terms of NYSE/AMEX listed proportion
in both pre- and post-PSLRA periods. 14.1% less of lPOs are backed by venture
capitalists following the enactment than those prior to it. In addition. 9.6% more of sued
lPOs are backed by venture capitalists than non-sued IPOs in the pre-PSLRA period. In
the post-PSLRA period, however, actually 2.5% fewer of sued lPOs are backed by
venture capitalists than non-sued lPOs. This reversal again reﬂects the impact of the
proportionate liabilities of the PSLRA.

We use secondary offering as a proxy for insider selling. Panel A shows that 8%
less of IPOs have a secondary offering following the enactment. This is consistent with
the prospect theory of Loughran and Ritter (2004) that issuers concentrate more on
aftermarket proﬁt than IPO proceeds in the late 19905. The signaling hypothesis states

that insiders retaining a great portion of ﬁrm shares can signal ﬁrms’ good quality

 

20 This is different from the results of Lowry and Shu (2002); they ﬁnd no signiﬁcant difference in the
high-tech proportion between sued and non-sued ﬁrms.

18

(Leland and Pyle (1977) and Grinblatt and Hwang (1989), etc.). Consistent with this
hypothesis, the proportion of sued lPOs having a secondary offering is higher than that of

non-sued lPOs. Consistent with Foster et al. (2000), we ﬁnd that there are more insider

trading cases following the enactment.2| 10.7% more of sued lPOs have a secondary
offering than non-sued IPOs in the post-PSLRA period. In the pre-PSLRA period,
however, only 8.4% more of sued lPOs have a secondary offering than non—sued lPOs.

In summary, the PSLRA of 1995 does reduce IPO ﬁrms’ litigation risk in our
sample period (1990 to 1998). We ﬁnd less “deep pocket” digging, less “racing to the
courthouse”, and more insider trading related cases in the post-PSLRA period. Since the
PSLRA has a strong impact on IPO ﬁrms, in fact, we can predict that its impact on most
non-IPO ﬁrms would be stronger because the “safe harbor” does not apply to cases

related to lPOs, tender offers, and penny stocks, etc.

 

2' The PSLRA raises the scienter requirement, which requires defendants to “know or have reason to know
that the information was non-public and had been obtained improperly by selective revelation or
otherwise.” Under the new law, plaintiffs more oﬁen search for a pattern of ofﬁcers and directors
liquidating their stocks during the class period to plead motive and opportunity. Therefore, we expect to see
more insider trading related cases in the post-PSLRA period.

19

1.4 Methods

We apply the simultaneous equations system to test the LRH of IPO underpricing.
Under this model set-up. the relationship between IPO underpricing and ﬁrms’ litigation
risk can be expressed as two simultaneously decided effects: the insurance effect and the
deterrence effect. On one hand, IPO ﬁrms expecting higher litigation risk purchase more
insurance, i.e., they underprice their lPOs by a greater amount. On the other hand, lPOs
with higher underpricing should have lower probabilities of being sued after offering, i.e..
they have a better chance to deter lawsuits. While initial return is observed directly,
litigation risk is observed only as a dichotomous variable — lawsuit dummy. Hence, these

two effects are captured by the following equations:

0 o o o o o *
Insurance effect: initial return = y. litigation risk + (3. X. + p. (1.1)
*
Deterrence effect: litigation risk = y; initial return + [33 X3 + pg (1.2)

lawsuit: 1 If litigation risk >0 (1.3)
0 otherwise

where
initial Return = the amount of underpricing for IPO ﬁrm i (primary variable of
interest);
litigation Rigki = the unobservable probability oflitigation for IPO ﬁrm i (primary

variable of interest);

X. = a vector of exogenous control variables for equations (1.1);
X2 = a vector of exogenous control variables for equations (1.2):
lawsuit = dummy variable equals to 1 if ﬁrm i is sued. and 0 otherwise;

20

111 and p3 = unobservable i.i.d. bivariate normal error terms.

Equation (1.1) expresses the insurance effect of the LRH. We accept the
hypothesis ify| is signiﬁcantly positive. Equation (1.2) expresses the deterrence effect of
the LRH. We accept the hypothesis if 73 is signiﬁcantly negative. The LRH of IPO
underpricing is accepted only when both y; and y2 are signiﬁcant and have the right signs.

According to (Maddala (1983). p. 244). the reduced forms of this simultaneous

equations system are:

initial return = I]. X + v. (1.4)
litigation risk' = 112 x + v2 (1.5)

where
X = a vector of all exogenous variables in the system including
instrumental variables (the union ole and X2);
v] and v2 = unobservable i.i.d. bivariate normal error terms.

Since litigation risk is observed only as a dichotomous variable — lawsuit dummy,

. 2 .
we can only estimate I12 /02, where 02 =var(vz). Therefore. equation (1.5) should be

written as:

*
. . . . in: litigation risk no V? a. .
litigation risk : V = “ X +——=— = TIZX +v7 (1.6)
0.7 0") 02 b

b .-

 

 

The structural equations (1.1) and (1.2) are now written as:

21

TL.

C.

(II.

Ii;

Ctr

 

1 o o o o o o I **
Insurance effect: initial return = yloz litigation risk + [3. X. + p. (1.7)

. . . . H 7 . . . :8 .
Deterrence effect: litigation risk 2 —2—Inlllal return + —2 .X 7 + 4 (1.8)
02 a, " 0.,

1.4.1 Two-stage estimation method introduced by Maddala (1983)

We ﬁrst estimate both effects (equation (1.7) and (1.8)) using the two-stage
estimation procedure recommended by (Maddala (1983), p. 245). For the insurance

effect, the ﬁrst stage is estimating equation (1.6) by probit maximum likelihood (ML) to

A
obtain index function TI; X. The second stage is estimating equation (1.7) by ordinary

A
least squares (OLS) after substituting II: X for litigation risk . Parameters yloz and [3.

are estimated here. For the deterrence effect. the ﬁrst stage is estimating equation (1.4) by

A
OLS to obtain linear projection function IIl X . The second stage is estimating equation

A

(1.8) by probit ML after substituting I‘II X for initial return. Parameters Yz/O’z and [32/02
are estimated here. The asymptotic covariance matrixes are given in (Maddala (1983), p.
245). We use the non-parametric bootstrap to obtain standard errors for coefﬁcients.
There are two drawbacks to this method. First, coefﬁcients y], yz and [32 are not
identiﬁable. We can only estimate scaled coefﬁcients 7.02, yz/oz and [33/03. Second, the

computing procedure of asymptotic covariance matrixes can be very difﬁcult. Regardless

of these defects, the regression coefﬁcients are still consistent. We are able to test

1“)
Tx)

ul

{-1
7’7

CU

 

 

 

whether these coefﬁcients are signiﬁcantly different from zero and understand the

statistical signiﬁcance of both insurance and deterrence effects.

1.4.2 Generated instrument approach introduced by Wooldridge (2001)
We further use a more applicable method to estimate the insurance effect. Given
that the endogenous variable (litigation risk.) in the insurance effect equation is observed

as a dichotomous variable (lawsuit), equation (1.1) is called a dummy endogenous
variable model (Heckman (1978)). We can estimate it using a generated instrument
approach introduced by (Wooldridge (2001), p. 621).

The detailed procedure is as follows: (1) Estimate equation (1.6) by probit ML to

A
obtain the ﬁtted probabilities F(I1:i2 X ). which is the predicted probability of lawsuit=l.

A
.. . * . . . . . . “ .
(Ii) Usmg F(fI2 X ) as a generated instrumental variable for litigation risk , estimate

equation (1.1) by two-stage least squares (ZSLS).22 The ZSLS in step (ii) is equivalent to

the following two steps: (a) Estimate the following equation (1.9) by OLS to obtain linear
A 9* A .
projection function 6170—12 X) + 61Xl . (b) Estimate equation (1.1) by OLS after
A A* /\ t .
substituting 6F(II2 X)+6Xl for litigation risk . Parameters y. and B. are estimated
here.

A

litigation riski = s F( n; X) + ex. +t (1.9)

 

l) o D o
2“ This generated instrumental variable is uncorrelated With the error term. In, and highly correlated With
* I u u
the endogenous variable, litigation risk , so it meets both assumptions for instrumental variables.

23

 

' 1",
‘0) k; . s

it?" x

 

1.4.3

 

 

where

T = unobservable i.i.d. normal error term.

In short, this approach is estimating the dummy endogenous variable model

A

(equation (1.1)) by ZSLS using a generated instrumental variable F( fl; X). As discussed

in (Wooldridge (2001), p. 622 and 623), there is no special consideration in estimating a
dummy endogenous variable model by ZSLS, but this generated instrumental variable
approach offers estimators that are more efﬁcient.

There are two nice features of this approach compared to the two-stage estimation
method of Maddala (1983). First, we can actually estimate yi directly and avoid yloz.
Second, we can ignore whether the speciﬁcation of equation (1.6) and the estimation of
II; are correct or not. Therefore, the 2SLS standard errors and test statistics are

asymptotically valid (Wooldridge (2001), p. 1 l7 and 623).

1.4.3 Instrumental variables and control variables

In this subsection, we discuss instrumental variables and control variables. The
descriptive statistics for most of these variables are listed in Table 1.3. Table 1.4 shows
the descriptive statistics of instrumental variables and the comparison of alternative
matched sample proxies. Panel A compares the characteristics of these variables in the
pre-PSLRA period to that in the post-PSLRA period. Panel B compares sued and non-

sued ﬁrms in the whole sample period.

for

 

ITIJ

Sl’i

In the insurance effect estimation, an exogenous instrumental variable is required

for litigation risk. We ﬁrst discuss the matched sample stock turnover used in Lowry and

Shu (2002). IPO ﬁrms’ stock turnover equals [l-I1[=..22387(1- volume tradedt/total sharesl)]

with t = 0 as IPO offering date.23 Panel A shows the average IPO stock turnover
signiﬁcantly increases by 1.9% following the enactment. Since shareholder damages are
increasing in the number of shares traded at the allegedly misleading prices, stock
turnover should be positively related to ﬁrms’ litigation risk. Consistent with this
conjecture, Panel B shows that the average stock turnover of sued lPOs is 10.5% higher
than that of non-sued lPOs. Both tests are at the 1% signiﬁcance level. Because the post-

IPO stock turnover is not in managers’ information set at the time of offering. the

matched sample stock turnover (matched turnover) is further constructed.24 Table 1.4

shows that matched turnover has characteristics similar to IPOs’ own stock turnover; 23

therefore, we believe matched turnover is a good proxy for IPOs’ own stock turnover.
They argue that the stock turnover of similar ﬁrms prior to IPO should not

directly affect IPO initial return; therefore, matched turnover can be used as the

instrumental variable for litigation risk. On the contrary, we believe matched turnover is

 

23 As Lowry and Shu (2002) point out, this turnover measure is a direct input in estimating damaged shares
under the proportional trading model, which is widely accepted by courts as the appropriate way to
calculate damages in class action lawsuits. For ﬁrms listing on NASDAQ, we divide the trading volume by
a factor of two.

24 Speciﬁcally, for each IPO ﬁrm, we select all ﬁrms with the same three-digit SIC code and market
capitalization within 80% to 120% of the IPO ﬁrm as the matched sample. If there is no matched ﬁrm
found, we select all ﬁrms with the same two-digit SIC code and market capitalization within 80% to 120%
ofthe IPO ﬁrm. If still no matched ﬁrm is found, we select all ﬁrms with the same one-digit SIC code and
market capitalization within 80% to 120% of the IPO ﬁrm. If still no matched ﬁrm is found, we select all
ﬁrms that have market capitalization within 80% to 120% of the IPO ﬁrm. For each matched ﬁrm in a
particular IPO ﬁrm’s matched sample, stock turnover is calculated based on its daily returns over a one-
year interval prior to this IPO’s offering date. This IPO ﬁrm’s matched turnover is calculated by averaging
the stock turnovers of all matched ﬁrms in its matched sample.

25 Matched turnover increases signiﬁcantly after the enactment of the PSLRA. Sued ﬁrms have
signiﬁcantly higher matched turnover. The correlation coefﬁcient between IPOs’ own stock tumover and
matched turnover is 0.15.

25

III'l

Mi

Jill

(I:

AIL

1111

§
hi ‘r-
\‘1‘

not exogenous enough to be a good instrumental variable. Reese (1998) ﬁnds that lPOs
with greater “investor interest level” prior to offerings tend to have higher initial returns
and higher initial trading volumes (scaled by shares issued), which cause more
information production and brokerage market making services, then further cause more
“investor interest” and higher long term trading volume. In short, the “investor interest
level” prior to offering leads to a positive relationship between an IPO’s initial return and
its stock turnover. Busaba and Chang (2002) show that informed investors will bid low in
the premarket if they are interested in the potential proﬁt from aftermarket trading;
therefore, IPO underpricing is positively related with investors’ interest in aftemiarket
trading. Other than these two papers, there has been quite an amount of empirical

evidence showing the positive relationship between IPO initial return and stock
turnover/trading volume.26 Based on these ﬁndings, we believe there exists an omitted
variable “investor interest level” in the error term, II], of equation (1.5). While
representing the similar ﬁrms’ “investor interest level.” matched turnover should be
strongly correlated with Lil. Therefore, as pointed out by Ljungqvist (2004), matched

turnover appears to fail the rank condition for instrument validity. In addition, matched
turnover is partially correlated with litigation risk at only 10% signiﬁcance level (Lowry
and Shu (2002), p. 329). Due to these two reasons, we think matched turnover is a weak

instrument for litigation risk.

 

26 Miller and Reilly (1987) ﬁnd that lPOs with a positive initial return have a higher trading volume (scaled
by shares issued) for each of the ﬁrst ﬁve days and the 215‘ day of trading. Schultz and Zaman ( 1994) ﬁnd a
higher trading volume (scaled by shares issued) for underpriced lPOs during the ﬁrst day of trading. Ellis.
Michaely, and O’Hara (2000) ﬁnd the trading volume of the “hot” lPOs is more than double that of the
other lPOs.

26

ins:

Psi

QC.

1P1

'.1 i
\\\L

116).}

”LT.

1173:.

In this paper. we propose an exogenous policy change dummy (PSLRA) as the
instrumental variable for litigation risk. PSLRA = I if an IPO is issued in the post-
PSLRA period (year 1996 to 1998), and 0 if an IPO is issued in the pre-PSLRA period
(year 1990 to 1993). We have shown that the passage of the PSLRA in 1995 decreases
IPO ﬁrms’ litigation risk. Furthermore, this legal policy change dummy is exogenous in
estimating IPO underpricing because it just adopted the procedural hurdles to ﬁling
securities fraud class actions and never dealt with IPO pricing procedure. As showed in
next section, it is also partially correlated with litigation risk at 1% signiﬁcance level.
Hence, we believe this instrumental variable is completely exogenous and stronger than
matched turnover.

To estimate the deterrence effect, we follow Lowry and Shu (2002) using prior
market return as the exogenous instrumental variable for IPO initial return. Prior market
return is equal to the compounded value-weighted NYSE/AMEX/NASDAQ daily returns
(including dividend) within 15 days prior to the IPO offering date. As earlier research has
documented, prior market return is signiﬁcantly correlated with IPO underpricing. In
addition, we agree that it is exogenous in estimating IPO ﬁrms’ litigation risk. Panel A of
Table 1.4 shows that prior market return is signiﬁcantly higher in the post-PSLRA
period, which might reﬂect issuers’ tendency to time their lPOs more or the booming
market of the late 19905. Panel B shows that sued ﬁrms have a slightly lower prior
market return than non-sued ﬁrms. Both difference tests are at the 10% signiﬁcance level.

Control variables are the common exogenous variables in both insurance and

deterrence effects, i.e.. they are related to both initial return and litigation risk. Most of

27

[C\\
llt‘k'

1‘1»;

cap
tin:
are
dis.
ran}
less

and

 

 

these variables have been introduced in the Data section; in this section. we summarize
their relationship with IPO ﬁrms’ litigation risk and IPO underpricing.

Ln(Market Cap) = the natural logarithm of market capitalization. This size
measure should be positively related to litigation risk based on the “deep pocket” theory.
The relationship between market capitalization and initial returns could be mixed. With
less information asymmetry, larger ﬁrms should have lower initial returns. Underwriters,
however, may underprice bigger issues by a larger amount because it is harder to sell
bigger issues (Michaely and Shaw (1994)).

UW rank = IPO lead underwriter rank. VC backed = 1 if an IPO is venture
capitalist backed and 0 otherwise. The “deep pocket” theory suggests that the lPOs
underwritten by more prestigious investment banks and/or backed by venture capitalists
are subject to higher litigation risk. Higher rank underwriters and venture capitalists may
disclose more of IPO ﬁrms’ information to protect their reputation: therefore, underwriter
rank and VC backed dummy should be negatively related to IPO underpricing because of
less information asymmetry (Ritter (1984). Carter and Manaster (1990), and Michaely
and Shaw (1994)).

Ln(l+Age) = the natural logarithm of (1+Age). High-tech = I if an IPO ﬁrm is
classiﬁed as a high-tech ﬁrm in SDC and 0 otherwise. NYSE-AMEX = 1 if an IPO ﬁrm
is listed on the NYSE or AMEX and 0 otherwise. With size controlled in a multivariate
regression, older ﬁrms and ﬁrms listed on NYSE or AMEX should have less litigation
risk and less IPO underpricing due to their lower uncertainty levels and more publicly

available information (Carter and Manaster (1990) and Megginson and Weiss (1991)).

28

I171

l‘CI

1112

Hu

and

II;-

V"
[it I

UTIL'

Sir?

SCI

High-tech ﬁrms should have higher litigation risk and higher IPO underpricing due to
their higher uncertainty levels.

Insider selling = 1 if an IPO ﬁrm has a secondary offering in IPO and 0 otherwise.
This dummy should be positively related to the IPO ﬁrm’s litigation risk because insider
selling makes it easier for plaintiffs to establish motive or scienter in lawsuits. It should
be negatively related to IPO underpricing because insider selling signals poor quality of
the issuing ﬁrm.

Price update should be positively correlated with litigation risk because issuers
setting offer price more aggressively are more likely to be sued. According to Benveniste
and Spindt (1989), underwriters only partially update the offer price from the midpoint of
price ﬁling range to compensate institutional investors through IPO underpricing. Earlier
research also ﬁnds a strong positive relationship between price update and initial return.
We predict the same positive relationship in our regression.

Stock volatility should be positively related to litigation risk because lawsuits are
typically ﬁled after a steep drop in stock price. It should also be positively related to
underpricing because high volatility is associated with high information asymmetry.
Similar to stock turnover, IPO ﬁrms’ own stock volatility is not in managers’ information
set at the time of offering because it is based on post-IPO daily returns. Again, matched
sample standard deviation (matched SD) is used following Lowry and Shu (2002).27
Interestingly, Table 1.4 shows this matched SD is not a good proxy for IPO ﬁrms’ own

standard deviation. As shown in Panel A, the difference of average matched SDs between

 

27 The matched sample is the same one used to construct matched turnover. For each matched ﬁrm in a
particular IPO ﬁrm’s matched sample, standard deviation is calculated based on its daily returns over a one-
year interval prior to this IPO’s offering date. An lPO’s matched SD is the average standard deviation of all
matched ﬁrms in its matched sample.

29

pre- and post-PSLRA periods is not as strong as that of IPO ﬁrms’ own standard
deviations. The t-test is insigniﬁcant. More important. Panel B shows that the average
matched SD of non-sued ﬁrms is actually higher than that of sued ﬁrms, which reverses
the relationship of sued and non-sued IPO ﬁrms’ own standard deviations. We think
taking the average of regular ﬁrms’ stock volatilities cannot capture IPO ﬁrms’ own
volatility because lPOs usually have higher stock volatility than other publicly traded
ﬁrms.

In this paper, we propose a new proxy (matched IPO SD), which focuses on IPO
ﬁrms’ own stock volatilities only. When issuers try to predict future stock volatility
during the IPO process. it makes sense for them to refer to the post-offering stock
volatility of lPOs issued one or two years earlier. Therefore, for each particular IPO. we
construct its matched IPO sample by selecting all lPOs with the same I-digit SIC code
and issued between day -388 and day -753, with day 0 as the offering date of this
particular IPO. Its matched IPO SD (one year) is equal to the average of standard
deviations (one year) of ﬁrms in its matched IPO sample.28 Table 1.4 shows matched IPO
SD has very similar characteristics to [PO ﬁrms’ own standard deviation (25 days). It is
signiﬁcantly higher following the enactment than prior to it. Sued ﬁrms have
signiﬁcantly higher matched IPO SD than non-sued ﬁrms do. Overall, in terms of both

difference tests’ sign and statistical signiﬁcance, it is a much better proxy for IPO ﬁrms’

own standard deviation (25 days) compared to matched SD.29

 

28 Standard deviation (one year) = the standard deviation of daily returns over a one-year interval
beginning one-month after the offering date. Day -388 to -753 is selected as the time interval for matched
lPOs because we want to make sure all matched IPOs’ standard deviations (one year) coverage is within
issuers’ information set on the offering date.

“9 For comparison purposes, we construct a half-year measure as well. Standard deviation (half year) == the
standard deviation of IPO ﬁrms’ daily returns over a half-year interval beginning one month after the

30

 

e.

In summary. variables used in regressions are as follows. X is a vector of all
exogenous variables in the simultaneous equations system, which includes all control

variables discussed above and instrumental variables for both litigation risk and initial
return. X. includes all control variables discussed above plus prior market return. which
is the instrumental variable for initial return. X2 includes all control variables discussed

above plus PSLRA or matched turnover, which are the instrumental variables for

litigation risk. Therefore. X is in fact the union ole and X2.

 

offering date. Matched IPO SD (half year) = the average of standard deviations (half year) of these matched
ﬁrms in matched IPO sample, which includes all lPOs with the same I-digit SIC code and issued between
day -205 and day -387. We have also constructed matched samples by matching listed exchanges and size;
however, the cross-sectional comparisons show that matched IPO SD (one year) and matched IPO SD (half
year) are the two best choices. Between these two measures, the correlation coefﬁcient between matched
IPO SD (one year) and standard deviation (25 days) is 0.18, which is higher than the 0.14 of matched IPO
SD (half year). Therefore, we present matched IPO SD (one year) in the main empirical test results.
Matched IPO SD (halfyear) is discussed in the sensitivity analysis.

31

1.5 Empirical results

This section addresses the multivariate regression results of the LRH of IPO
underpricing. Section 1.5.1.1 examines the insurance and deterrence effects in a
simultaneous equations system with PSLRA and prior market return as instrumental

variables. Section 1.5.1.2 conducts various sensitivity analyses.

1.5.1 Regression results using the simultaneous equations system
1.5.1.1 Insurance effect results

Table 1.5 shows the regression results of estimating equations (1.5) and (1.1)
using the two-stage estimation method in Maddala (1983) and the generated instrument
approach in Wooldridge (2001) with a policy change dummy (PSLRA) as the external
instrumental variable for litigation risk. Column 1 shows the ﬁrst-stage probit ML results
of equation (1.4), which is shared by both methods. Columns 2 and 3 show the second-
stage regression results of the insurance effect using the two-stage estimation method and
the generated instrument approach.

The ﬁrst-stage regression is also a multivariate test of the impact of the Private
Security Litigation Reform Act on IPO ﬁrms’ litigation risk. Column 1 shows that the
PSLRA does reduce IPO ﬁrms’ litigation risk given that the PSLRA dummy is
signiﬁcantly negatively correlated with the lawsuit dummy. With other characteristics
remaining the same, the probability of IPO ﬁrms being sued decreases by 4% after the

passage ofthe PSLRA at the 1% signiﬁcance level, which is consistent with the ﬁndings

32

CIT.

(7.4;-

a:

“Q
51.;

re-

CI.

It‘ ‘.

ftl

 

 

in Beatty, Drake, and Hogan (2002). The Chi-square statistic of PSLRA is 12.21. which
further suggests that it is a strong instrument for litigation risk.

The second-stage regression results of both methods suggest that IPO ﬁrms’
expected litigation risk have no signiﬁcant effect on IPO initial return. Although the signs
are in the right direction, neither the coefﬁcient of lawsuit index function in Column 2
nor the coefﬁcient of lawsuit in Column 3 shows any statistical signiﬁcance. Contrary to
the earlier research, this ﬁnding is against the insurance effect ofthe LRH, i.e., IPO ﬁrms
do not use expensive IPO underpricing as insurance against possible future class action
lawsuits. In addition, the coefﬁcient oflawsuit index function in Column 2 is 0.01, which

is much smaller than the 0.28 of lawsuit in Column 3. This difference is reasonable

because the two-stage estimation method identiﬁes 7102 while the generated instrument

approach identiﬁes y] itself.

Other regression results of the two methods are very close in terms of sign,
statistical, and economic signiﬁcances because both methods identify Bl directly. These
results are similar to the ﬁndings in prior literature. The coefﬁcient of our new stock
volatility measure (matched IPO SD (one year)) is signiﬁcantly positively related to
initial return, which shows that volatile lPOs have high underpricing. Interestingly, the
coefﬁcient of matched SD is negative, which shows no economic rationale. This suggests
that matched IPO SD (one year) is a better proxy for IPO ﬁrms’ stock volatility.3O Initial
return and prior market return are signiﬁcantly positively related. Supporting Michaely

and Shaw (1994) that the underwriter might offer underpricing to institutional investors

 

30 . . .
We also test the hypotheSIs usmg either matched SD or matched IPO SD (one year) as the only proxy for
stock volatility. Results presented in Tables 1.5 and 1.6 are qualitatively unchanged.

33

for

>0.

1m

 

~
:1!

for selling a large bulk of shares. initial return is signiﬁcantly positively related to
Ln(market cap). Consistent with the information asymmetry hypothesis, initial return is
negatively related to UW rank, Ln(l+age) and VC backed, where the coefﬁcients of UW
rank and Ln(l+age) are signiﬁcant. The uncertainty level makes the coefﬁcient of NYSE-
AMEX signiﬁcantly negative and the coefﬁcient of high-tech signiﬁcantly positive.
Consistent with the signaling hypothesis, the coefﬁcient of insider selling is negative but
not signiﬁcant. Price update is signiﬁcantly positively related to initial return, which
suggests that underwriters only partially update the offer price to compensate institutional

investors for their information revealing.

1.5.1.2 Deterrence effect results

Table 1.6 shows the regression results of estimating equation (1.6) using the two-
stage estimation method in Maddala (1983) with prior market return as the external
instrumental variable for initial return. The ﬁrst two columns show the regression results
of the full sample. The second two columns show the regression results after dropping
dismissed/terminated cases.3| The regression results of both samples are similar in terms
of both economic and statistical signiﬁcance as shown in Table 1.6. Columns 1 and 3 are
the ﬁrst-stage OLS regression results of equation (1.3). Columns 2 and 4 are the second-
stage regression results of the deterrence effect.

Columns 1 and 3 show that lPOs issued following high market returns tend to

have a high initial retum. With other characteristics remaining the same. IPOs’ initial

 

3 . . . .
I Lowry and Shu (2002) argue that these dismissed/terminated cases should not be brought to court in the
ﬁrst place. They ﬁnd a signiﬁcant deterrence effect after dropping these cases.

34

U71;

llIt’

(I

I‘ll. .

1111.- -

ll’lt

fill-Q

 

ill:~

 

return increases by 0.86% when their prior market return increases by 1%. The
coefﬁcient of prior market return is signiﬁcant at the 1% level. which suggests that prior
market return can be used as an instrumental variable for initial return.

The results of the second-stage regressions for both samples suggest that IPO
underpricing has no signiﬁcant effect on reducing a ﬁrm’s future litigation risk. Although
the signs are in the right direction, neither coefﬁcient of predicted initial return in either
Column 2 or Column 4 shows any statistical signiﬁcance. Contrary to the earlier ﬁnding
once again, our result is against the deterrence effect of the LRH, i.e., issuers
underpricing their I POs more have no advantage in deterring future class action lawsuits.

The second-stage regression of the deterrence effect is also a test for the impact of
the PSLRA on IPO ﬁrms’ litigation risk after controlling for the IPO underpricing level
and other ﬁrm characteristics. The coefﬁcient of the PSLRA dummy is negative and
signiﬁcant at the 1% level, constituting strong evidence that the PSLRA does reduce IPO
ﬁrms’ litigation risk. Stock volatility has a strong impact on IPO ﬁrms’ litigation risk too.
The coefﬁcient of matched IPO SD (one year) is positive and signiﬁcant at the 1% level.
The other volatility measure matched SD shows no signiﬁcance at all, which further
supports that matched IPO SD (one year) works better as a proxy for IPO stock volatility
than does matched SD. The evidence of the “deep pocket” theory seems mixed. Market
capitalization and UW rank are positively related to litigation risk; however, the VC
backed dummy is negatively related to litigation risk. The coefﬁcients of both Ln(l+age)
and NYSE—AMEX dummy are negative, and the coefﬁcients of the high-tech dummy are
positive, which shows that IPO ﬁrms with a higher uncertainty level tend to have a higher

litigation risk after controlling for ﬁrm size. The coefﬁcients of insider selling and price

35

update are positive, which is reasonable because both of them help plaintiffs and lawyers
making scienter statements.

Our results provide strong evidence against the LRH of IPO underpricing. Neither
the insurance effect nor the deterrence effect is signiﬁcant. IPO ﬁrms do not use
underpricing as insurance against class action lawsuits, nor do lPOs with high

underpricing have any advantage in deterring lawsuits.

1.5.2 Sensitivity analysis

Our results are opposite to those of Lowry and Shu (2002), who ﬁnd strong
evidence supporting the LRH. To examine the robustness of our results. we try various
sensitivity tests in this subsection.

Because their sample period only covers the pre-PSLRA period, we ﬁrst compare
the regression results between two separate sub-sample periods. Table 1.7 shows the
regression results of the deterrence effect for both pre- and post-PSLRA periods using the

same variables as theirs, because we would like to compare the regression results under

their model set-up.32 The ﬁrst (second) two columns show regressions based on the pre-
(post-) PSLRA sub-sample. Similar to the results in Table 1.6, prior market return is
signiﬁcantly positively related to IPO initial return; therefore, it serves as a good
instrument for initial return. We further ﬁnd the insigniﬁcant relationship between IPO
initial return and litigation risk in both sub-sample periods. These testing results remain

qualitatively unchanged after dropping dismissed/terminated cases. The deterrence effect

 

‘2 . . . .

J Because PSLRA is dichotomous. we are not able to test the insurance effect for both we and post-
PSLRA periods using PSLRA as the instrument. For matched turnover. we argue that it is not an
appropriate instrument to be used. It will be further discussed in Table 1.8.

36

found in the earlier research does not show up in our sample regardless ofthe choice of
sample period.

While PSLRA is never signiﬁcantly related to initial return in Table 1.6, an
interesting result we ﬁnd in Columns 1 and 3 of Table 1.7 is that matched turnover turns
out to be signiﬁcantly related to IPO initial return, which certainly casts doubt on the

exogeneity of matched turnover. As we have discussed in section 1.4, matched turnover
might be correlated with the error term p. of equation (1.5) because it could be a good
proxy for the omitted variable “investor interest level”. We cannot test this correlation
statistically since pl is unobservable: however, with the other instrument PSLRA

available, we can at least compare their overall performance as either instruments or
control variables.

The comparison results are shown in Table 1.8. We test the insurance effect in the
whole sample period using two types of model set-ups. In the type (1) model, we treat
matched turnover as a control variable and PSLRA as an instrument. In the type (2)
model, we treat PSLRA as a control variable and matched turnover as an instrument. We

test these two models under both the two-stage estimation method and the generated

instrument approach. 33 Both model set-ups share the same ﬁrst-stage probit ML
regression, which is shown in Column 1. The coefﬁcient of PSLRA is negatively related
with litigation risk at the 1% level and matched turnover is positively related with
litigation risk at the 10% level. The Chi-square statistics are 13.38 for PSLRA and 3.48
for matched turnover. This further shows that matched turnover is a weaker instrument

than PSLRA is. Columns 2 and 4 show the second-stage results of the type (1) model

 

33 . . . . . .
We Include matched IPO SD in regreSSIons because earlier tests suggest it is a better proxy for stock
volatility than matched SD is. The results remain qualitatively unchanged with matched IPO SD excluded.

37

under two methods using PSLRA as the instrument. Both columns show no signiﬁcant
relation between expected litigation risk and IPO initial return, which is consistent with
the ﬁndings in Table 1.5. Opposite results are shown in Columns 3 and 5, which present
the second-stage results of the type (2) model under two methods using matched turnover
as the instrument. Both columns show that the relationship between expected litigation
risk and IPO initial return is signiﬁcant at the 1% level. In summary, the insurance effect
becomes signiﬁcant whenever we treat matched turnover as the instrument. The
insurance effect shows no signiﬁcance at all whenever we treat matched turnover as the
control variable. F roim both empirical evidence and economic common sense, we believe
our policy change dummy is a stronger instrumental variable than matched turnover is.
Since weak instruments may worsen the effect of simultaneity bias, instead of solving it,
our results raise the concern that the signiﬁcant insurance effect found in the earlier
research might be driven by the choice of a potentially inappropriate instrumental
variable (matched turnover).

We further test the LRH under a different sample selection standard. First, we use
cases ﬁled under Section 1 1 of the 1933 Act only. We, however, are only able to test the
deterrence effect in the post-PSLRA period due to limited data availability.34 We then
use IPO-related cases with class period starting from the offering date. Here, both
insurance effect and deterrence effect are tested with PSLRA and prior market return
used as instrumental variables. We also rerun tests using cases ﬁled within 2 years, 1.5

years and 1 year after the IPO offering date.35 No signiﬁcantly different results are found.

 

34 We are able to collect Section 11 information for almost the entire sub-sample in the post-PSLRA
period, but it is missing for half ofthe sub—sample in the pre-PSLRA period.

3 Lowry and Shu (2002) used this practice because they think “a lawsuit that occurs long after an IPO is
more likely prompted by random stock price ﬂuctuations not in the managers‘ information set at the time of

We also test the LRH using some alternative control variables. Loughran and
Ritter (2004) reveal that SDC does not capture all dual-class shares and CRSP only

reports publicly traded shares. They also point out that SDC does not identify all the
Internet ﬁrms. Their tech stock deﬁnition is also different from SDC’s. 36 We construct

our market capitalization and high-tech dummy following them37 and rerun the tests
using these two new variables along with the alternative stock volatility measure matched
IPO SD (half year). Results remain qualitatively unchanged from those presented.

In 2001, there were more than 300 “IPO allocation” class action lawsuits ﬁled in
the Southern District Court of New York. These cases are targeted at underwriters’ and
certain institutional owners’ “laddering” activities, which inﬂate IPO aftennarket price.
Although issuers themselves may not directly be involved in these activities, they still
become defendants along with the underwriters accused by omitting this “Iaddering”
information from the IPO registration statement. Therefore, lPOs with higher
underpricing would have higher chances to trigger these “IPO allocation” lawsuits.38
This effect is exactly opposite to the LRH; therefore, including these cases in the sample
may introduce noise and decrease testing power. Since our sample stops in the year 1998

and most of these “IPO allocation” cases are related to lPOs offered in the year 1999 and

 

IPO, and this greater noise might mask the relation between underpricing and litigation risk”. We actually
believe this is exactly one of the reasons why the LRH is not true.

36 They use the number of post-issue shares outstanding from CRSP for single-class lPOs. For dual-class
lPOs, they use data from a variety of sources including Smart and Zutter (2002), SDC. CRSP, Dealogic,
and the prospectus on EDGAR. They merge the Internet identiﬁcations from SDC, Dealogic, and
IPOMonitor.com. Their tech stock includes the following SIC codes: 3571, 3572, 3575. 3577, 3578
(computer hardware), 3661, 3663. 3669 (communications equipment), 3671. 3672, 3674, 3675, 3677-3679
(electronics), 3812 (navigation equipment), 3823, 3825-3827, 3829 (measuring and controlling devices).
3841, 3845 (medical instruments), 4812, 4813 (telephone equipment), 4899 (communications services).
and 7371-7375, 7378, 7379 (software).

37 The dual-class number of shares and Internet stock identiﬁcation are available on Dr. Jay Ritter’s web
page [http://bear.cba.uﬂ.edu/ritter/ipodata.htm].

‘ Hao (2004) ﬁnds that higher IPO underpricing increases litigation risk.

39

after, our results should not be affected signiﬁcantly by them. Still. we do have eight
“IPO allocation” cases related to lPOs issued in the year 1998; therefore. we rerun the
tests after dropping the year 1998. Again, we get qualitatively unchanged results.

One concern of our instrumental variable policy change dummy (PSLRA) is that
it may also be a proxy for time trend, which may be positively related to IPO initial
return. Loughran and Ritter (2004), however, show that the IPO initial return did not
increase monotonically over time in the 19905, but experienced a sudden jump up in
1995.39 We believe this jump was a result of changed IPO ﬁrms’ characteristics. When
the PSLRA decreased of ﬁrms’ overall litigation risk. underwriters may agree to take
public more risky ﬁrms that they may otherwise refuse without the new law; therefore,
the increase of initial return should be caused by the characteristic change of IPO ﬁrms,
not by time trend.40 After controlling for ﬁrms’ characteristics including litigation risk.
we ﬁnd initial return does not correlate with the PSLRA dummy. Therefore, we believe
our policy change dummy is a strong instrumental variable. We further include GDP
annual growth rate in the explanatory variables to control for economic condition. Again,

results remain qualitatively unchanged.

 

39 IPO initial return ranged from 10.2% to 12.8% between 1990 and 1993. It dropped to 9.8% in 1994.
Then there was a big jump up to 21.5% in 1995, after which it ranged from 14% to 22.2% between 1996
and 1998.

40 Although the PSLRA was enacted in Dec. 1995, it was introduced by House Republicans on January 4.
1995, and the passage of it was anticipated shortly aﬂer its ﬁrst introduction (WSJ ( 1995)). Therefore, the
stock market may start showing its impact in 1995.

40

1.6 Conclusion

The litigation risk hypothesis (LRH) of IPO underpricing predicts that IPO ﬁrms
exposed to higher litigation risk should have higher IPO underpricing in equilibrium.
Though the Private Securities Litigation Reform Act of 1995 was designed to reduce
ﬁrms’ litigation risk, IPO underpricing actually increased in 1995 and after. This
discrepancy casts doubt on either the effectiveness of the PSLRA on IPO ﬁrms or the
correctness of the LRH. This paper tests the impact of the PSLRA on IPO ﬁrms’
litigation risk and investigates LRH using a simultaneous equations system.

We ﬁnd the PSLRA of 1995 does reduce IPO ﬁrms’ litigation risk in our sample
period (1990 to 1998). We ﬁnd less “deep pocket” digging and less “racing to the
courthouse” in the post-PSLRA period. Our empirical results counter the LRH of IPO
underpricing in both insurance and deterrence effects. On one hand, issuers expecting
higher litigation risk do not underprice their lPOs more. On the other hand, issuers with
higher IPO underpricing do not have an advantage in deterring lawsuits. We ﬁnd no
signiﬁcant deterrence effect either prior to or following the enactment of the PSLRA.

It is important to note that the results of this paper speciﬁcally concern the
litigation risk hypothesis of IPO underpricing, not the importance of litigation risk in
general. We do believe litigation risk affects corporate decisions; however, both legal
perspectives and empirical evidence suggest that litigation risk is likely not one of the

main reasons explaining the IPO underpricing puzzle.

41

Chapter 2

A MICRO EXAMPLE OF LAW AND FINANCE: A STUDY OF THE PRIVATE
SECURITIES LITIGATION REFORM ACT OF 1995 AND THE SARBANES-

OXLEY ACT OF 2002

2.1 Introduction

Since Enron’s implosion, a seemingly ever-growing list of corporate frauds and
accounting scandals has staggered the securities markets. Global Crossing, WorldCom,
Adelphia, and many others have been sinking in share price and triggering SEC and
criminal investigations. Some legislators believe the Private Securities Litigation Reform
Act of 1995 (PSLRA) should be blamed for these scandals (Labaton (2002), Avery
(1996), and Prichard (1999) etc.). This law adopted a series of procedural obstacles to
securities fraud class actions to discourage abusive lawsuits, thereby reducing ﬁrms’
litigation risk. Instead of deterring frivolous lawsuits as it was intended to, opponents
believe it made it too easy for ﬁrms to escape liability for securities fraud and thus
created a climate in which frauds are more likely to occur. In the hope of restoring
investor conﬁdence and reinforcing corporate accountability and professional
responsibility, Congress passed the Sarbanes-Oxley Act (SOA) in July of 2002. The SOA
takes a series of “sweeping” reform provisions to improve the quality of corporate
governance and the auditing profession. Many legislators believe the SOA could serve as
an important counter to the PSLRA by providing tools to help plaintiffs meet the

PSLRA’s heightened pleading standard. reversing the ﬁrm-friendly legal atmosphere

42

under the PSLRA and restoring shareholders’ ability to monitor ﬁrms through legal
activities (Kaplan and Greene (2003), Brickey (2003) etc.).

The purpose of this study is to investigate shareholder reaction to a series of
legislative events related to the passage of the PSLRA and the SOA. From the
shareholders’ point of view, we would like to answer two questions. First, do
shareholders think the PSLRA will deter meritorious lawsuits and thus reduce their legal
protection from corporate frauds? Second, do shareholders think the SOA will counter
the effect of the PSLRA and restore their legal protection? We concentrate our analysis
on high litigation risk ﬁrms, which are most likely to be affected by the passage of these
two laws. We ﬁIrther divide our sample into sub-samples that are subject to meritorious
lawsuits or frivolous lawsuits. Using three event study approaches, we ﬁnd both
statistically and economically signiﬁcant negative abnormal returns during the passage of
the PSLRA in the group of high litigation risk ﬁrms that are more likely to face
meritorious lawsuits; there is negative but insigniﬁcant shareholder reaction to the
passage of the PSLRA in the group of high litigation risk ﬁrms which are likely to face
frivolous lawsuits. Furthermore, shareholders of both groups react negatively and
signiﬁcantly to the passage of the SOA. Our results suggest that shareholders believed the
PSLRA would deter meritorious lawsuits instead of frivolous lawsuits as it was intended
to. Furthermore, the PSLRA reduces the important legal monitoring achieved by
meritorious lawsuits and increases the incidence of fraud. More disappointing,
shareholders did not believe the recently passed SOA would counter the effect of the

PSLRA and bring better protections for investors against corporate fraud.

43

This paper contributes to the existing literature in ﬁve ways. First, this is the ﬁrst
empirical paper that connects the PSLRA and the SOA and tests them together to see the
interaction between these two laws. Second. by separating samples based on their
litigation risk level and chances of facing meritorious lawsuits or frivolous lawsuits, we
empirically show the adverse impact of the PSLRA on necessary legal protections for
shareholders against corporate fraud. Third, for the passage of the SOA, we ﬁnd results
opposite to earlier empirical ﬁndings, which show that shareholders think the SOA is
beneﬁcial in general. Fourth, the results of this paper provide us with a better
understanding of the effects of the legal environment on the corporate scandals in the late
Nineties. Filth, this paper provides a micro example of the interaction of law and ﬁnance
and helps us better understand the impact of securities law on corporate decisions.

The rest of this chapter is organized as follows. Section 2.2 further discusses the
literature related to the PSLRA and the SOA. Section 2.3 describes the empirical research
design. Section 2.4 discusses three event study methods that are used and their empirical

results. Section 2.5 draws the conclusion.

44

2.2 Related literature
2.2.1 The Private Securities Litigation Reform Act of 1995

On December 22. 1995, Congress enacted the Private Securities Litigation
Reform Act of 1995. which amended the Securities Act of 1933 and the Securities
Exchange Act of 1934. This law adopted a series of procedural obstacles to securities
fraud class actions to discourage abusive lawsuits, thereby reducing ﬁrms’ litigation risk.
These obstacles include the following:

o The replacement of “joint and several liabilities” by “proportionate liability”
reduces the plaintiffs’ incentive of looking for “deep pockets”;

0 Plaintiffs’ stay of discovery in a pending motion to dismiss reduces ﬁrms’
defending costs:

a A new lead plaintiff appointment process and the prohibition of bonus payment to
lead plaintiffs stop lawyers from hiring “professional plaintiffs”;

o A heightened pleading standard increases the cost of ﬁling lawsuits;

o A new 90-day bounce back rule stops cases from rushing into courts;

0 The mandatory sanction of Rule 1 l ofthe Federal Rules of Civil Procedure deters
any party from making frivolous lawsuits;

o The comprehensive disclosure of the provisions of settlement and attorney fees
and the limitation on attorney fees to a “reasonable percentage” of any damage
award reduce attorneys’ proﬁts from lawsuits; and

o A “safe harbor” for forward-looking statements protects ﬁrms from lawsuits

aiming at forward-looking statements.

There have been debates on the qualities ofthe PSLRA from the very beginning.
Although it was hoped that the PSLRA would deter frivolous lawsuits. critics of the Act.
including President Clinton, who initially vetoed the legislation, thought that the PSLRA
would set too high a barrier for plaintiffs ﬁling legitimate lawsuits. More importantly. it
might make corporations immune from the legal monitor that had previously helped keep
them honest. In view of scandals such as Enron and WorldCom, some legislators blamed
the PSLRA, which created an atmosphere in which frauds became more likely to occur
(Labaton (2002)). As Loomis(2002) reported: “John Coffee, a Columbia Law School
professor, who testiﬁed before Congress that the diminished legal threat posed to auditors
in the wake of the act. coupled with the rise in non-audit work done by accounting ﬁrms,
makes a ‘credible story’ that auditors today ‘are subject to a greater temptation to defer to
management with regard to questionable accounting principles.”’ Avery (1996) and
Prichard (1999) also believe the PSLRA increases the incidence of fraud by effectively
preventing meritorious claims from being heard in federal courts.-

Defenders of the PSLRA argue that it does not deter meritorious cases. Johnson,
Kasznik, and Nelson (2001) ﬁnd that earnings and sales forecasts increased following the
passage of the PSLRA and that the safe harbor had no adverse impact on the quality of
forward-looking information. Pritchard (2003) thinks the combination of higher
settlements and a higher dismissal rate suggests that the PSLRA is doing a cost-effective
job of deterring corporate fraud. Johnson, Kasznik and Nelson (2000) ﬁnd the PSLRA
may deter some meritorious lawsuits, but its beneﬁts dominate. Furthermore, Johnson,
Nelson and Pritchard (2002) ﬁnd both accounting and insider trading are signiﬁcantly

correlated with lawsuits in the post-PSLRA period and conclude that it discourages

46

frivolous securities lawsuits. However, they agree that their study design cannot measure
the potential costs of the PSLRA in discouraging meritorious lawsuits.

In addition, some studies claim that the PSLRA did not work at all because it did
little to prevent plaintiffs lawyers from ﬁling lawsuits. Grundfest and Perino (1997)
think the PSLRA has no effect since plaintiffs can simply shift lawsuits to state courts.
Bajaj, Mazumdar, and Sarin (2003) ﬁnd an immediate drop in the number of federal
cases, but this shift tendency diminishes very quickly. The number of federal cases
promptly rises in both 1997 and 1998 while the number of state cases continues to drop.
Perino (2003) uses data from 1996 to 2001 and draws a similar conclusion. These
univariate tests, however, cannot control the frequency of corporate fraud. The increased
number of lawsuits may be a result of more fraud.

From a shareholder’s point of view, Spiess and Tkac (1997) and Johnson, Kasznik
and Nelson (2000) ﬁnd the stock market reacts more positively to the enactment of the
PSLRA for ﬁrms at higher litigation risk. “This positive effect, however, diminishes as
the probability of being sued for committing fraud increases.” Johnson, Nelson and
Pritchard (2000) arrive at a similar conclusion by studying the stock market’s reaction to
the Ninth Circuit’s decision, “In re Silicon Graphics Securities Litigation,” which
adopted the most stringent interpretation of the PSLRA’s “strong inference” standard for
pleading scienter in securities fraud cases. Ali and Kallapur (2001), however, overthrow
earlier results by ﬁnding that shareholders in four high-litigation-risk industries react
negatively on average to PSLRA’s restrictions on their ability to bring securities-related

lawsuits. Our results agree with Ali and Kallapur (2001); however. we further show that

47

ﬁrms subject to meritorious lawsuits dislike the PSLRA much more than ﬁrms subject to

frivolous lawsuits do.

2.2.2 The Sarbanes-Oxley Act of 2002

Urged by the list of corporate scandals at the time, President Bush signed the
Sarbanes-Oxley Act into law on July 30, 2002. The purpose of this law is “to protect
investors by improving the accuracy and reliability of corporate disclosures.” The main
provisions include:

0 Accounting changes. such as the establishment of an accounting oversight board,
the requirement of independent audit committee members, the prohibition of
accounting ﬁrms from performing certain non-audit services for an audit client,
required rotation of audit partners every 5 years. and the enforced SEC review of
ﬁrms’ ﬁnancial statement ﬁlings at least once every 3 years;

0 Corporate responsibilities, such as CEO and CFO certiﬁcation of ﬁnancial
statements;

0 Enhanced Financial Disclosures, especially about the pro forma statements: any
report ﬁled with SEC or any public disclosure or press or other release shall not
contain any untrue statement or omitted material facts;

0 Solving analyst conﬂicts of interest, through such means as the separation of
investment banking and securities advice functions of securities ﬁrms; and

0 Strengthening corporate and criminal fraud accountability, through actions such
as imposing criminal penalties for knowingly certifying ﬁnancial reports that fail

to comport with the requirements of the SOA. prohibiting persons involved in

48

fraud from serving as ofﬁcers or directors, and the extension of the Statute of
Limitation to 5 years for cases related to corporate fraud.

Although the SOA does not directly address the PSLRA, Kaplan and Greene

(2003) and Brickey (2003) believe that the SOA may change private securities litigation

in several noteworthy ways:

It provides tools to help plaintiffs meet the PSLRA’s heightened pleading
standards. These tools include disclosures of internal corporate information,
protections for current employees who cooperate with plaintiffs’ counsel, and
requirement of directors and ofﬁcers to know more about ﬁrms’ disclosures and
disclosure problems;

It may cause more shareholder derivative claims under a “corporate malpractice”
theory of liability due to its requirements of CEO and CFO certiﬁcation of
ﬁnancial statements:

It reverses the culture under the PSLRA that there are many frivolous lawsuits
and that companies are victims of these lawsuits. Under this new culture, people
realize that securities fraud does exist and has been committed by respected

business leaders and professional advisors. Therefore, private securities litigation

' is one of the effective ways to prevent fraud, monitor ﬁrms, and protect investors.

Critics, however, complain that the SOA provides little reformation, but rather

adds extra costs to ﬁrms. Perino (2003) thinks the SOA is unlikely to have much of a

deterrent effect on corporate fraud due to its disorganization and inadequate reformation.

Cunningham (2003) sees the SOA as more sweep than reform. He thinks the law is

needlessly redundant and “reenacts in a new federal guise more than a dozen existing

49

federal regulations. state laws. stock exchange and securities industry rules. accounting or
auditing practices. and corporate governance norms.” In addition, Ribstein (2002) thinks
the SOA may increase agency costs by distorting executives’ incentives to take on value-
maximizing transactions, encouraging executives to move to less monitored ﬁrms and
activities, increasing ﬁrms’ costs of obtaining information about executives’ fraudulent
activities, and increasing friction in the ﬁrm as a result of reduced trust.

From a shareholder’s point of view, Rezaee and Jain (2003) examine the market
reaction to the passage of the SOA and ﬁnd a positive (negative) abnormal return at the
time of several legislative events that increased (decreased) the likelihood of the passage.
They further ﬁnd that the market reaction is more positive for ﬁrms with more effective
corporate governance, reliable ﬁnancial reporting. and credible audit functions. Li, Pincus
and Rego (2004) also ﬁnd signiﬁcantly positive stock returns associated with events that
resolved uncertainty about the Act’s ﬁnal provisions or were informative about its
enforcement. Using several event study methods and different grouping standards, we

ﬁnd evidence opposite to their ﬁndings.

50

2.3 Research Design
To test shareholder reaction to the passage of PSLRA and SOA. we apply three

event study methods: Cumulative Abnormal Return (CAR) approach. equal-weighted

portfolio Ordinary Least Square (OLS) approach, and Generalized Least Square (GLS)

approach.“ Our sample consists of US. common stocks for which daily return data are
available on the ﬁles of the Center for Research in Security Prices at the University of
Chicago (CRSP). We further divide these ﬁrms into different groups based on their

litigation risk level, earning management level, and shareholder rights level.

2.3.1 Measure of Litigation Risk Level

Since ﬁrms subject to different litigation risks are likely to be affected differently
by the passage of PSLRA and SOA, we separate our sample into high and low litigation
risk groups. Following Francis, Philbrick and Schipper (1994), Johnson. Kasznik and
Nelson (2000), and Ali and Kallapur (2001), the high litigation risk group includes all
ﬁrms in four industries: computer (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). Due to recent massive accounting
scandals, we further include accounting ﬁrms (SIC codes 8720 and 8721) in the high
litigation risk group. For data from the year 1995, there is no accounting ﬁrm in our

sample. Firms that are not in these ﬁve industries are included in the low litigation risk

group.

 

41 . . . . . .
We Will discuss these three methods in detail in Section 2.4.

51

2.3.2 Measure of Earnings Management

In order to test whether the PSLRA deters shareholders from ﬁling meritorious
lawsuits and whether the SOA ﬁxes this problem, we need to identify ﬁrms subject to
meritorious or frivolous lawsuits. Firms exposed to meritorious lawsuits are those who
committed fraud, while ﬁrms exposed to frivolous lawsuits are those who did not commit
fraud. We use ﬁrms’ earning management measure in 1994 (2001) to identify ﬁrms that
tend or tend not to commit fraud before the passage of PSLRA (SOA). Following
Kothari, Leone and Wasley (2005), we employ a modiﬁed-Jones model developed in
Dechow, Sloan and Sweeny (1995) to construct two types of discretionary accrual:
abnormal accruals and performance-matched abnormal accruals.

First, we calculate total accruals (TA) as deﬁned in equation (2.1) using

COMPUSTAT data items.

TAU = (ANCC’AL), - ACLW — DAL),)/ASSETI.,),_l (2_1)
where:
ANCCA. : Change in ﬁrm i’s non-cash current assets from year y-l to y
W" (AData4 — ADatal );

ACL- = Change in ﬁrm i’s current liabilities excluding the current portion

”y of long-tenn debt from year y-l to y (ADataS — AData34 );

DAL}, = Firm i’s depreciation and amortization in year y ( Datal4 );

ASSETLyq = Firm i’s total assets in year y-l (Data6).

Second, we run OLS regression cross-sectionally on equation (2.2) to estimate

¢0.¢‘.¢2.¢3 and “by using all ﬁrms with the same two-digit SIC code in year 1994 or

2001. lfthe number of ﬁrms with the same two-digit SIC code is less than 10, we drop
this industrial group. To adjust for outliers of each variable. we make any number larger
than the top 1% equal to the level at the top 1% and any number smaller than the bottom

1% equal to the level at the bottom 1%.

TAL'V = ¢O + ¢l (l/AS‘SIETl-LV—l ) + ¢2 (ASALEKS in", - AARiqy )/‘A‘SVSETig'V—l

(2.2)
+ p3 PPE 11y / ASSE T W4 + Vt},
where:
ASALESL}, = Change in ﬁrm i’s net sales from year y-l to y (ADatalZ );
AAR- ___ Change in ﬁrm i’s accounts receivable from year y-l to y
”y (AData302 );
PPEW = Firm i’s gross property, plant, and equipment in year y (Data7 );

Vi,y = i.i.d. normal distributed error term.

The abnormal accruals (AA) are the residuals from the OLS regression results of
equation (2.2). The performance-matched abnormal accruals (PMAA) for ﬁrm i in year y

is its AA in year y minus its matched ﬁrm’s AA in year y, while its matched ﬁrm is

deﬁned as the ﬁrm with the closest return on assets in year y (ROAU) and the same two-

digit SIC code. Here, ROAM. is deﬁned as the following:

R0Ai.y = Firm i’s income before extraordinary items in year y (Data!8)/ASSET,,,.-I

53

To separate earning managers from non-earning managers, we follow Li, Pincus.
and Rego (2004) to sort ﬁrms based on their AA or PMAA in year 1994 or 2001, and
place ﬁrms into positive, zero, or negative earnings management (EM) portfolios. The
positive (negative) EM portfolio is composed of ﬁrms whose. AA or PMAA is included in
the top (bottom) one-third distribution and are therefore more likely to commit fraud. The
zero EM portfolio is composed of ﬁrms whose AA or PMAA is included in the middle
one-third distribution and are therefore less likely to commit fraud.

Table 2.1 shows the distribution of abnormal accruals and performance-matched
abnormal accruals of different portfolios based on their litigation risk level and earning
management level. By construction, ﬁrms within a positive or negative earning
management group have both economically and statistically signiﬁcant AA or PMAA.
Firms within a zero earning management group have both economically and statistically
insigniﬁcant PMAA and economically insigniﬁcant AA. Overall, high litigation risk
ﬁrms have a slightly higher earning management than low litigation risk ﬁrms have. We
consider high litigation risk ﬁrms with positive or negative EM as ﬁrms who are likely
subject to meritorious lawsuits and high litigation risk ﬁrms with zero EM as ﬁrms who

are likely subject to frivolous lawsuits.

2.3.3 Measure of Shareholder Rights

We further employ ﬁrm’s “Shareholder Rights Index” in 1995 (2002) to identify
ﬁrms who are subject to meritorious or frivolous lawsuits. Shareholders with strong

rights monitor ﬁrms using those rights instead of legal actions since lawsuits are time

54

consuming and ﬁnancially costly. Consequently, if they ﬁle lawsuits, those tend to be
meritorious. Shareholders with weak rights do not have many rights other than legal
actions to monitor ﬁrms. Consequently, if they ﬁle lawsuits, those tend to be frivolous.

Based on each ﬁrm’s corporate-govemance provisions in Corporate Takeover
Defenses42 from publications of the Investor Responsibility Research Center (IRRC),43
Gompers, Ishii, and Metrick (2003) construct a “Shareholder Rights lndex”44 (G) for
each ﬁrm. with a lower G representing stronger shareholder rights and a higher G
representing weaker shareholder rights.45 While they break up G into 10 groups, we only
divide G into 3 groups to simplify the analysis. The low G group includes their groups
#1, 2, and 3 (G37), who are the ﬁrms with strong shareholder rights and subject to
meritorious cases. The high G group includes their groups #8, 9, and 10 (G212), who are
the ﬁrms with weak shareholder rights and subject to frivolous cases. The mid G group
includes their. groups #4, 5. 6 and 7 (SSGSII), who are the ﬁrms with average
shareholder rights.

Table 2.2 shows the distribution of “Shareholder Rights lndex” of different

portfolios based on their litigation risk level and G level. By construction, ﬁrms within

 

42 These provisions can be categorized into ﬁve thematic groups: tactics for delaying hostile bidders,
voting rights, director/ofﬁcer protection, other takeover defenses, and state laws.

3 These data are derived from a variety of public sources including corporate by-laws and charters, proxy
statements, annual reports, and 10-K and IO-Q documents ﬁled with the SEC. The IRRC’s universe is
drawn from the Standard & Poor’s (S&P) 500 as well as the annual lists of the largest corporations in the
publications of Fortune, Forbes, and Businessweek. The IRRC’s sample expanded by several hundred ﬁrms
in 1998 through the addition of some smaller ﬁrms and ﬁrms with high institutional-ownership levels.
Their analysis uses all ﬁrms in the IRRC universe except those with dual-class common stock (less than 10
percent of the total). The IRRC universe covers most of the value-weighted market: even in 1990 the IRRC
tracked more than 93 percent of the total capitalization of the combined New York Stock Exchange
(NYSE). American Stock Exchange (AMEX), and NASDAQ markets.

44 - .. .. - -
It is also called Governance Index in their paper.

45 . . . . . . . .
Their index construction IS straightforward: for every ﬁrm, they add one pomt for every provmon that
reduces shareholder rights.

55

the low G group have an average G around 5.8; ﬁrms within the mid G group have an
average G around 9.5: and ﬁrms within the high G group have an average G around 13.
There is no signiﬁcant difference of G level between high litigation risk and low
litigation risk ﬁrms. We consider high litigation risk ﬁrms with low G as ﬁrms who are
likely subject to meritorious lawsuits and high litigation risk ﬁrms with high G as ﬁrms .

who are likely subject to frivolous lawsuits.

2.3.4 Legislative Event Dates

For conducting tests of shareholder reaction to the passage of PSLRA and SOA,
we need to identify a series of legislative event dates for each law. We follow Ali and
Kallapur (2001) to decide the legislative event dates for the passage of PSLRA. Table 2.3
Panel A refers to part of their Table l, which lists these events and indicates whether they
increased or decreased the likelihood of the passage of PSLRA. There are 22 event dates
in total with four critical dates. On 12/5/05, the Senate approved the bill with a vote of
65-30. On 12/6/95, the House approved the bill with a vote of 320-102. 12/20/95 was the
date that news of both President Clinton’s veto and the House override of the veto
reached the market. Since these two events have opposite impacts on the likelihood of
PSLRA’s passage, there is debate in the literature as to their effects. Ali and Kallapur
(2001) believe the veto on the night of December 19 was more surprising than was the
House override on December 20, while Spiess and Tkac (1997) and Johnson, Kasznik
and Nelson (2000) argue the opposite. Thus, whether the stock price reaction of high-

litigation-risk ﬁrms on December 20 should be attributable to the veto or to the House

56

vote is questionable. The last critical date is 12/22/95. when the Senate overrode
President Clinton’s veto with a vote of 68-30.

We refer to Rezaee and Jain (2003) and Li, Pincus, and Rego (2004) to decide the
legislative event dates for the passage of SOA. Table 2.3 Panel B lists these events and
indicates whether they increased or decreased the likelihood of the passage of SOA.
There are 18 event dates in total with ﬁve critical dates. On 7/15/02, the Senate approved
S. 2673, which was considered too harsh on ﬁnancial ﬁrms and corporations. On 7/ 16/02,
the House approved HR. 51 18, which was viewed as too weak to be effective. As Rezaee
and Jain (2003) pointed out, “The extent of disagreements between these versions,
coupled with the limited time to compromise these differences in the ﬁrst part of July
2002, created signiﬁcant doubt that any reform bill would pass before Congress departed
Washington for the August recess (Geewax (2002) and Oppel (2002))” We follow their
opinion that investors saw these differences as a signal of a decreasing likelihood of the
passage of the SOA. On 7/24/02, the House-Senate Conference Committee reconciled the
House and Senate bills and ﬁnally issued the Sarbanes-Oxley Act of 2002. On 7/25/02,
Congress passed the SOA by a vote of 423-3 in the House and 99-0 in the Senate. On
7/30/02, President Bush signed the SOA into law. Following Rezaee and Jain (2003), we

think that investors considered these three events as favorable to the passage of the SOA.

57

2.4 Event Study Methods and Empirical Results
To test shareholder’s reaction to the passage of PSLRA and SOA, we divide our

sample into several groups based on their litigation risk level, earning management level,
and shareholder rights level. Then we apply three methods: CAR approach, equal-

weighted portfolio OLS approach. and GLS approach.

2.4.1 Cumulative Abnormal Return Approach

We ﬁrst examine the cumulative abnormal returns of each portfolio during the
periods around the passage of PSLRA and SOA. The estimation window is [1/1/1994,
12/31/1994] for PSLRA (252 trading days) and [Ill/2001, 12/31/2001] for SOA (248
trading days). The event window is [12/4/1995, 12/26/95] for PSLRA (16 trading days)
and. [7/23/2002, 7/31/2002] for SOA (7 trading days).46 Firms with less than 252/248
trading days data in the estimation window or 16/7 trading days data in the event window
are excluded from the ﬁnal sample. Since event dates are clustered for all securities in the
sample, we use ‘portfolio’ abnormal returns to take into account cross-sectional
dependence in the security-speciﬁc abnormal returns (Collins and Dent (1984), Brown
and Warner ( 1985)).

We follow the methods used in Brown and Warner (1985) to calculate CARS.

Deﬁne A” as the abnormal return for security i at day t. For every security i, A” is

estimated using the OLS market model:

 

~16 . . . . .
We include one day before and after the ﬁnal passage to take into account possrble Information leakage
and lags.

58

A A
ALI : RIJ Tat" 781' Rm.t (2‘3)

where:
R” = Daily return for ﬁrm i at day t:
RmJ = Value-weighted index return for day t;
d, and '3’, = OLS values from the estimation period.

For each event window [T1, T3], the test statistic is the ratio of the cumulative
portfolio abnormal return (CARp) to its estimated standard deviation. Portfolio abnormal

return is the simple average of N, securities’ abnormal returns in each portfolio. The
standard deviation is estimated from the time-series of portfolio abnormal returns in the

estimation period [T3, T4].

 

T2 A2
CARP ZS (Ant) (2.4)
t=T1
where:
T2
CARp 2 AN (2.5)
(:71
1 “I
A,” = "TX/lit (2-6)
N! i=1

59

 

7

A l ’29 -
SKA/2.!) : 2 (ApJ _ Ap) (2-7)

 

T4 ‘T3 1:73
__ 1 ’:T4
A : ———-—-— A
I) T4 _ T3 +1 1:27“ [)J (2.8)

The test statistic is unit normal under the null hypothesis that CARp is equal to
zero. During these event windows, the probability that PSLRA and SOA would pass
increased’from less than 1.0 to precisely 1.0. Positive (negative) CARp would indicate
that shareholders viewed PSLRA and SOA as beneﬁcial (not beneﬁcial). Table 2.4
compares the CARI, during the passage of PSLRA and SOA.

Panel A shows the results by dividing the sample into groups based on litigation
risk level and abnormal accrual level. Panel B shows the results by dividing the sample

into groups based on litigation risk level and performance-matched abnormal accrual

level. Shareholders of high litigation risk ﬁrms dislike the passage of PSLRA, especially
the shareholders of earning managers. The CARP; of high litigation risk portfolios are

negative. Regardless of the two different choices of earning management measure,

shareholders of earning managers and high litigation risk portfolios (ﬁrms subject to
meritorious lawsuits) dislike the PSLRA the most. The C ARp of both NEG EM and POS

EM high litigation risk portfolios are economically and statistically signiﬁcant.

Shareholders of zero EM and high litigation risk portfolio (ﬁrms subject to frivolous
lawsuits) also dislike the PSLRA, but the CARp is not statistically signiﬁcant. This shows

that shareholders do not believe the PSLRA will deter frivolous lawsuits as it was

intended to, and they are actually afraid that the PSLRA will deter meritorious lawsuits.

60

Shareholders of earning managers and low litigation risk portfolios reacted

insigniﬁcantly. The ZERO EM and low litigation risk portfolio actually has positive
CARI, during the passage of PSLRA.

Regardless of litigation risk level and earning management level. shareholders
dislike the passage of SOA in general. Shareholders of earning managers and high
litigation risk portfolios (subject to meritorious cases) dislike the SOA the most, then the
shareholders of non-earning managers (subject to frivolous cases) and high litigation risk
ﬁrms, then the low litigation risk ﬁrms. These results show that shareholders generally do
not think the SOA will bring them more protection. This is especially true for
shareholders of ﬁrms subject to lawsuits, who actually think SOA does not deter ﬁrms
from committing fraud, but brings extra costs instead.

Panel C shows the results achieved by dividing the sample into groups based on
litigation risk level and shareholder rights level. Shareholders of low G and high litigation

risk portfolios (ﬁrms with the strongest shareholder rights and those subject to
meritorious lawsuits) dislike the PSLRA the most. The C’ARps are both economically and

statistically signiﬁcant. Shareholders of mid G and high litigation risk portfolio also
dislike the PSLRA, but much less signiﬁcantly. Shareholders of high G and high
litigation risk portfolio (ﬁrms with the weakest shareholder rights and those subject to
frivolous lawsuits) react negatively but insigniﬁcantly to the passage of PSLRA.
Therefore, the signiﬁcance of CARp increases as shareholder rights increase, i.e., as
ﬁrms’ likelihood of being exposed to meritorious lawsuits increases. This again shows
that shareholders do think the PSLRA will deter their ability to monitor ﬁrms with

meritorious lawsuits. Shareholders of low litigation risk portfolios reacted insigniﬁcantly.

6|

Shareholders of high litigation risk ﬁrms also dislike the passage of SOA. The
CARps of high litigation risk portfolios are negative and signiﬁcant. Shareholders of low
G and high litigation risk portfolio (subject to meritorious lawsuits) dislike the SOA the

most. The CARp is both economically and statistically signiﬁcant. Shareholders of mid G

and high litigation risk portfolio also dislike the PSLRA, but the CARp is economically

less signiﬁcant. Shareholders of high G and high litigation risk portfolio (subject to
frivolous lawsuits) react the least signiﬁcantly to the passage of SOA. This again shows
that shareholders of high litigation risk ﬁrms do not think the SOA will bring them more
protection, especially ﬁrms subject to meritorious lawsuits. The low litigation risk
portfolios reacted less signiﬁcantly. Excepting the low G group, the mid and high G
portfolios do not react signiﬁcantly to the passage of SOA. These results show that
shareholders of high litigation risk and strong shareholder rights ﬁrms dislike the SOA

the most.

2.4.2 Equal-Weighted Portfolio OLS Approach

The CAR approach cannot resolve two concerns with efﬁciency that are typically
associated with policy event studies: (1) multiple announcement events for a given policy
change make it difﬁcult to identify when the market ﬁrst anticipates the effects of the
events; and (2) the estimation period variance used in the CAR approach may change in
the event period. Therefore, we further apply the equal-weighted portfolio OLS approach
for sensitivity analysis. This method is used in Ryngaert and Netter (1998), Karpoff and

Malatesta (1989), and Ali and Kallipur (2001). The model is constructed as follows:

62

./
RpJ : ap + IBpRmJ + ZIU_/'D_j + zIn! (2-9)
j=l

where:

\'
I "

Rm = 72R), (2.10)
1 i=1

D- _ A dummy variable that equals to I when the day corresponds to the

j _ event datej and 0 otherwise.

The event dates have been introduced in Section 2.3.4. There are 22 event dates
(J=22) for the passage of PSLRA and 18 event dates (J=I8) for the passage of SOA. We

estimate equation (2.9) by OLS using 1995 daily stock return data for PSLRA and 2002

data for SOA. The coefﬁcient ,ul- represents the average abnormal return of different

portfolios on event datej. Each portfolio consists of ﬁrms with 252 trading days in either
1995 or 2002. Table 2.4 shows that the shareholders of high litigation risk ﬁrms react
much more strongly to the passage of laws than do the shareholders of low litigation risk
ﬁrms. Therefore, we concentrate on high litigation risk ﬁrms from now on. We further
divide high litigation risk ﬁrms into three groups based on their earning management
level and shareholder rights level.

Table 2.5 reports OLS estimates of equation (2.8) for shareholders of different
earning management portfolios. Using two types of EM measure, we concentrate on
positive, zero and negative earning managers. Panel A reports results for the reaction to
the passage of PSLRA. Shareholders of high litigation risk ﬁmis dislike the passage of

PSLRA. The abnormal return on 12/5/95 (Senate approves the conference committee bill)

63

is negative and statistically signiﬁcant. From both earning management measures.
shareholders of negative earning management ﬁrms (ﬁrms subject to meritorious
lawsuits) dislike the PSLRA the most. The abnormal returns of NEG EM portfolio on
12/5/95 (Senate approval) and 12/6/95 (House approves the conference committee bill)
are both economically and statistically signiﬁcant. Shareholders of zero earning
management ﬁrms (ﬁrms subject to frivolous lawsuits) also dislike the PSLRA, but the
reaction is much less signiﬁcant. For each group, the abnormal return on 12/20/95
(President vetoes bill and House overrides veto) is positive. The interpretation of this date

is inconclusive because of the two confounding events. but Ali and Kallapur (2001) think

1:22
of it as decreasing the probability of the passage. The F test of 221] =0 shows the
j:16

signiﬁcance ofthe cumulative abnormal returns on these critical dates. The CARs of both
NEG EM and POS EM high litigation risk portfolios (subject to meritorious lawsuits) are
both economically and statistically signiﬁcant. Shareholders of zero EM and high
litigation risk portfolio (subject to frivolous lawsuits) also dislike the PSLRA, but much
less signiﬁcantly. This again shows that shareholders do think the PSLRA will deter the
meritorious lawsuits and make ﬁrms immune from legal monitoring.

Panel B reports results for the reaction to the passage of SOA. Shareholders of
high litigation risk ﬁrms also dislike the passage of SOA. The abnormal return on 7/16/02
(unfavorable to the passage of SOA) is positive and statistically signiﬁcant. The
abnormal returns on 7/24/02 and 7/25/02 (favorable to the passage of SOA) are negative
and statistically signiﬁcant. There is no clear difference across the three earning

management groups based on a AA measure. From a PMAA measure. however.

64

shareholders of earning managers (subject to meritorious lawsuits) dislike the SOA more
than shareholders of non-earning managers (subject to frivolous lawsuits) do. The F test
j=18
of Eu]- 2 0 shows the signiﬁcance of the CARs on the last three critical dates. The
j=lo
CARS of high litigation risk ﬁrms are both economically and statistically signiﬁcant
regardless of their earning management level. Nevertheless, from a PMAA measure,
shareholders of earning managers (subject to meritorious lawsuits) dislike the SOA more
than the shareholders of non-earning managers (subject to frivolous lawsuits) do. These
results again show that shareholders of ﬁrms subject to lawsuits generally do not think
the SOA will bring them more protection, especially the shareholders of ﬁrms subject to
meritorious cases, who actually think the SOA does not monitor ﬁrms from fraud, but
brings extra costs instead.

Table 2.6 reports OLS estimates of equation (2.8) for shareholders of different
shareholder rights portfolios. We concentrate on the low G, mid G, and high G groups.
Panel A reports results for the reaction to the passage of PSLRA. Shareholders of low G
ﬁrms (ﬁrms with the strongest shareholder rights and those subject to meritorious
lawsuits) dislike the PSLRA the most. The abnormal return of low G portfolio on 12/5/95
(Senate approves) is negative and the abnormal return on 12/6/95 (House approves) is
negative and statistically signiﬁcant. Shareholders of mid G ﬁrms (ﬁrms with average
shareholder rights) also dislike the PSLRA but much less signiﬁcantly. Shareholders of

high G ﬁrms (ﬁrms with the weakest shareholder rights and those subject to frivolous

j=22
lawsuit) have the least reaction to the passage of PSLRA. The F test of 2,21]- : 0 shows
jzlf)

65

the signiﬁcance of the cumulative abnormal returns on these critical dates. Low G ﬁrms
(subject to meritorious lawsuits) have negative CAR with the biggest absolute value. Mid
G ﬁrms have a relatively weaker but still negative response to the passage of PSLRA.
High G ﬁrms (subject to frivolous lawsuits) actually have positive CAR, but it is not
statistically signiﬁcant. This shows PSLRA may help deter some frivolous lawsuits, but
more importantly, it deters the meritorious lawsuits and reduces shareholders’ power to
monitor ﬁrms.
Panel B reports results for the reaction to the passage of SOA. Shareholders of
high litigation risk ﬁrms dislike the passage of SOA. The abnormal return on 7/16/02
(unfavorable to the passage of the law) is positive and signiﬁcant. The abnormal returns
on 7/24/02 and 7/25/02 (favorable to the passage of the law) are negative and signiﬁcant.
j=18
The F test of Zn]- :0 tells a similar story. Shareholders of low G ﬁrms (subject to
j=16
meritorious lawsuits) dislike the SOA the most, then the shareholders of mid G ﬁrms
(ﬁrms with average shareholder rights). Shareholders of high G ﬁrms (subject to
frivolous lawsuits) react insigniﬁcantly to the passage of SOA. These results again show
that shareholders of ﬁrms with high litigation risk generally do not think the SOA will

bring them more protection, especially shareholders of ﬁrms subject to meritorious

lawsuits, who actually think SOA will bring additional costs to the companies.

2.4.3 Generalized Least Square Approach

While the previous two methods concentrate on portfolio returns, we also employ

the third approach of generalized least squares estimation, which uses individual security

66

return data and a full covariance matrix of residuals estimated from the ﬁrst step OLS
regressions. This method was ﬁrst introduced into event studies by Schipper and
Thompson (1983). Compared to the equal-weighted portfolio OLS approach, this method

A
has two advantages: (1) it allows each ﬁrm to have their own beta coefﬁcient ﬂ,- ; and (2)

using a full covariance matrix allows us to account for both heteroscedasticity and cross-
sectional correlation in security return residuals, which is serious in regulatory event
studies since each announcement event occurs on the same calendar date for all affected
ﬁrms, and affected ﬁrms usually have industry or other factors in common (Schipper and
Thompson (1983)). In this section. we still concentrate on high-litigation risk ﬁrms. The

model is constructed as follows:

R=XF+E (2.11)

where:

 

 

Ri
R = R,‘
.R1 .
R,- = T X! time-series vector of daily returns to ﬁrm i
T 0 a)
X = T '
0 T a)
Y = [1 Rm]
Rm = T><1 time-series vector of value-weighted index return

67

TXJ matrix of event date variables, with one column for each
event date considered. With uniform treatment of

0,1 = announcements, each column contains ones and zeroes to
identify days on which events were not announced (zeroes) and
days on which events were announced (ones).

.- —

a1

ﬂi

0!

ﬂ/
J’J I

 

 

t9]j

 

 

T X] vector of error terms assumed to be serially independent.
8,- : independent of the index return and the event date variables, and
identically distributed normally.

Within this system, the error terms are assumed to be serially independent with a
stationary cross-sectional covariance structure represented by the contemporaneous
residual covariance matrix 21x1 . Thus, E has a covariance matrix of 2691 . The
parameters in F are estimated with joint GLS exploiting an estimate of XIX, , which is
the full covariance matrix of residuals from the ﬁrst step OLS regressions of (2.1 1). One
assumption we make is that all ﬁrms have identical underlying event date parameters.

Therefore. only one event date coefﬁcient ,u j is estimated for each event. Since we

estimate each portfolio within the similar industry and earning management or

shareholder rights level, this assumption is reasonable and technically more feasible.

68

Schipper and Thompson (1983) pointed out that the number of time-series observations
should be greater than twice one plus the number of ﬁrms (T<=2*(1+1)) for results to be
reasonable. Therefore. we further divide the earning management measure into ten
deciles and use the original G index constructed by Gompers, Ishii, and Metrick (2003).

We test two hypotheses concerning the uj parameters. The ﬁrst is that for a
particular event date, all of the individual 211- parameters across the sample of ﬁrms are

equal to zero.
ilj=0Vj- (2.12)

The second hypothesis we test is that the sum of critical event parameters for a

particular law is equal to zero.

.lz
ij =0 . (2.13)
i=JI

where J; is 16 for both PSLRA and SOA; and J2 is 22 for PSLRA and 18 for SOA.

Table 2.7 presents the GLS results of shareholders of high litigation risk and
different earning management portfolios. We concentrate on 10 different earning
management levels based on either abnormal accrual or performance-matched abnormal
accrual. We ﬁrst rank each ﬁrm’s AA/PMAA from high to low, and then divide them
into 10 deciles. Positive earning managers have lower group numbers; negative earning
managers have higher group numbers; and non-earning managers have middle group
numbers. Panel A reports estimation results of AA groups for the passage of PSLRA.

Panel B reports estimation results of PMAA groups for the passage of PSLRA. In order

69

to save space, we only show the regression results of the last seven event dates, which are
critical to the passage ofthe PSLRA. Shareholders of high litigation risk ﬁrms dislike the
passage of PSLRA. From both individual event date results and CAR results,
shareholders of earning managers (ﬁrms subject to meritorious lawsuits) dislike the
PSLRA more than the shareholders of non-earning managers (ﬁrms subject to frivolous
lawsuits) do. When the group number reaches to both ends (higher or lower), CARS are
both economically and statistically signiﬁcant. When the group number reaches to the
middle, CARS are insigniﬁcant and even become positive. This shows that shareholders
do think the PSLRA will deter meritorious lawsuits.

Panel C reports the estimation results ofAA groups for the passage of SOA. Panel
D reports the estimation results of PMAA groups for the passage of SOA. Shareholders
of high litigation risk ﬁrms generally dislike the passage of SOA regardless of their
earning management level. The abnormal return on 7/16/02 (unfavorable to the passage
of the law) is positive and signiﬁcant. The abnormal returns on 7/24/02 and 7/25/02
(favorable to the passage of the law) are negative and signiﬁcant. There is no clear

j=18
difference across the ten earning management groups. The F test of 2 ,u j = 0 shows the
j=l6

signiﬁcance of the cumulative abnormal returns on the last three critical dates. The CARS
of high litigation risk ﬁrms are both economically and statistically signiﬁcant regardless
of their earning management level. These results again show that shareholders of ﬁrms
subject to lawsuits generally do not think the SOA will bring them more protection; this
is especially true for shareholders of ﬁrms that manage their earnings, who actually

believe SOA does not monitor corporate fraud but brings extra costs instead.

70

Table 2.8 presents the GLS results of shareholders of high litigation risk and
different shareholder rights portfolios. We concentrate on 10 groups based on their
"shareholder rights index” (G) level as introduced in Gompers, Ishii, and Metrick (2003).
Panel A reports the estimation results of G groups for the passage of PSLRA. The
regression results are not very signiﬁcant. One interesting result is that the F test of
1:22

Z ,uj = 0 shows that shareholders of high G ﬁrms (ﬁrms with the weakest shareholder
j=16
rights and those subject to frivolous lawsuits) have a signiﬁcantly positive reaction to the
passage of PSLRA.

Panel B reports the estimation results of G groups for the passage of SOA.
Shareholders of high litigation risk ﬁrms dislike the passage of SOA. The abnormal
return on 7/16/02 (unfavorable to the passage of the law) is positive and signiﬁcant. The
abnormal returns on 7/24/02 and 7/25/02 (favorable to the passage of the law) are

j=18
negative and signiﬁcant. The F test of ij =0 tells a similar story. Shareholders of
j=16
lower G ﬁrms (ﬁrms with the strongest shareholder rights and those subject to
meritorious lawsuits) dislike the SOA more than the shareholders of mid G ﬁrms (ﬁrms
with average shareholder rights) do. Shareholders of high G ﬁrms (ﬁrms with the weakest
shareholder rights and those subject to frivolous lawsuits) react insigniﬁcantly to the
passage of SOA. These results again show that shareholders of ﬁrms subject to lawsuits
generally do not think the SOA will bring them more protection, especially shareholders

of ﬁrms subject to meritorious lawsuits. who actually think the SOA will bring extra

costs to the companies.

71

2.5 Conclusion
While the Private Securities Litigation Reform Act of 1995 was blamed for the

massive corporate and accounting scandals that occurred in the late Nineties, legislators
hope the Sarbanes-Oxley Act of 2002 could serve to counter the PSLRA. Using event
study approaches, we try to ﬁnd out whether the PSLRA diminished the legal protection
for shareholders against corporate fraud and whether the SOA ﬁxed this problem. While
we ﬁnd evidence showing that the PSLRA did deter meritorious lawsuits from being ﬁled
and created a friendly legal atmosphere for ﬁrms, we cannot ﬁnd any evidence showing
that the SOA can reverse this damage. Instead, we ﬁnd that Shareholders generally dislike
the passage of SOA and think it will bring extra costs to ﬁrms. Our results support the
view that the PSLRA serves an important role in causing massive corporate scandals:
however, the SOA is not effective in ﬁxing this problem. These ﬁndings help us better
understand the cause of recent corporate scandals and further suggest that legislators

should ﬁgure out a more effective law to restore legal protection for investors.

72

 

APPENDIX 1

TABLES OF CHAPTER 1

73

Table 1.1

Distribution of Sample Firms Across Years

The sample consists of initial public offerings (lPOs) issued between 1990 and 1998 that
are available in the Thomson Financial Securities Data database. C losed-end funds, unit
offerings, REITS, ﬁnancial ﬁrms, reverse LBOS, ADRs, Spin-offs, lPOs with an offer
price below $5.00 per share, ﬁrms without exchange information and ﬁrms listed in the
small capital market or the OTC market are excluded. An IPO ﬁrm is identiﬁed as sued
if the federal class action lawsuits related to its IPO are listed in the Security Class
Action Alert newsletter between January 1990 and August 2002. Average months
between IPO and ﬁling of lawsuits is calculated as the average of months between IPO
offering month and lawsuit ﬁling month.

 

Average months between

 

 

 

 

 

IPO issue year # lPOs # lPOs sued % lPOs sued IPO and ﬁling of lawsuits
I990 79 4 5.06% 14.25
1991 219 14 6.39% 11.21
1992 283 18 6.36% 14.83
1993 365 28 7.67% 13.07
Sub Total 946 64 6.77% 13.23
1994 308 10 3.25% 13.50
1995 351 18 5.13% 11.89
Sub Total 659 28 4.25% 12.46
1996 527 15 2.85% 13.07
1997 348 15 4.31% 12.40
1998 201 20 9.95% 20.70
Sub Total 1076 50 4.65% 15.92
Total 2681 142 5.30% 14.03
Total w/o 1994 &
1995 2022 114 5.64% 14.41

 

74

 

Table 1.2

Lawsuit Status and Resolution Characteristics

An IPO ﬁrm is identiﬁed as sued if the federal class action lawsuits related to its IPO are
listed in the Security Class Action Alert newsletter between January 1990 and August
2002. Corresponding case decision information is collected by searching the Public
Access to Court Electronic Records (PACER) service {http://pacer.psc.uscourts.gov],
the Stanford Securities Class Action Clearinghouse [http://securities.stanford.edu]. the
Gilardi and Co. class action administration web site
[http://www.gilardi.com/allcases.html], company SEC ﬁlings, Lexis-Nexis newswire
and Lexis-Nexis legal research. Cases are categorized by IPO issue year from 1990 to
1998 (with 1994 and 1995 excluded). Average decision time length is the average of
months between lawsuit ﬁling month and case decision month. Average settlement
amount is calculated as the average of total settlement fund amount. Average
settlement/proceed ratio is calculated as the average of settlement amount to IPO
proceeds raised ratio.

 

Dismissed or Pending

 

 

 

 

Settled cases Terminated cases
Average Average Average
# # decision settlement Average # decision
IPO Cases time amount settlement/proceed Cases time # Cases
Issue year sued settled length ($millions ratio D/T length pending
1990 4 2 40.00 6.70 30.24% 2 12.00 0
1991 14 12 28.08 4.43 13.30% 2 7.50 0
1992 18 11 24.09 5.19 19.53% 6 18.33 1
1993 28 18 27.89 3.16 17.20% 9 16.67 1
Sub Total 64 43 27.53 4.19 17.47% 19 15.74 2
I996 15 8 24.75 3.70 12.22% 7 16.14 0
1997 15 10 30.70 3.93 19.43% 5 21.50 0
1998 20 14 20.83 15.20 51.84% 6 16.11 0
Sub Total 50 32 26.25 6.92 25.97% 18 17.32 0
Total 114 75 27.07 5.17 20.53% 37 16.45 2

 

75

Table 1.3

Descriptive Statistics of Sued vs. Non-sued lPOs before and after the Enactment of
the Private Security Litigation Reform Act of 1995

The ﬁnal sample consists of 946 lPOs from 1990 to 1993 and 1.076 lPOs from 1996 to
1998 that are available in the Thomson Financial Securities Data (SDC) database.
Closed-end funds, unit offerings, RElTs, ﬁnancial ﬁrms. reverse LBOS. ADRs, spin-offs,
lPOs with an offer price below $5.00 per share, ﬁrms without exchange information and
ﬁrms listed in the small capital market or the OTC market are excluded. An IPO ﬁrm is
identiﬁed as sued if the federal class action lawsuits related to its IPO are listed in the
Security Class Action Alert newsletter between January 1990 and August 2002. There are
64 sued lPOs from 1990 to 1993 and 50 sued lPOs from 1996 to 1998. Some variables
are based on fewer observations due to missing data. Initial retum is the percentage
change between the offer price and the ﬁrst closing price. Price update is the percentage
change between the midpoint of ﬁling price range in prospectus and the offer price.
Market capitalization is equal to the ﬁrst closing price multiplied by the number of Shares
outstanding after offering in all markets. Proceeds raised are equal to the offer price
multiplied by the shares offered in all markets. Both market capitalization and proceeds
raised are measured in millions of 1983 dollars using the Consumer Price Index.
Underwriter rank is equal to the lead underwriter rank measure used in Loughran and
Ritter (2004), with higher ranks representing higher quality underwriters. Standard
deviation (25 days) equals the standard deviation of daily returns during the ﬁrst 25 days
starting with the offering date (day 0 to day 24). Age is equal to the difference between
ﬁrm’s IPO offering year and founding year. IPO ﬁrms are labeled as high-tech, venture
capitalist backed, traded on the NYSE/AMEX and with secondary offering in all markets
according to the SDC classiﬁcations. Panel A compares IPO ﬁrms’ characteristics in the
pre-PSLRA period to that in the post-PSLRA period. Panel B compares sued and non-
sued ﬁrms in the pre-PSLRA period vs. post-PSLRA period. The pre-PSLRA period
represents the years 1990 to 1993. The post-PSLRA period represents the years 1996 to
1998.

76

Panel A: Pre-PSLRA Period vs. Post-PSLRA Period

 

 

 

Pre-PSLRA Post-PSLRA
Period Period
Median Median Wilcoxon test
Mean Mean T-test
(SE) (SE)
Initial Return 0.0625 0.1000 -4.5510 ***
0.1215 0.1736 -4.4524 ***
(0.0060) (0.0096)
Price Update 0.0000 0.0000 0.8460
-0.0018 -0.0024 0.0657
(0.0062) (0.0064)
Market Capitalization 46.50 59.20 -5.5310 ***
159.00 179.00 -0.3878
(41.80) (31.60)
Proceeds Raised 18.50 22.40 -5.7330 ***
33.90 40.20 -1.5755
(2.02) (3.35)
Underwriter Rank 8.0000 8.0000 -3.2220 ***
7.0735 7.3138 -2.5971 ***
(0.0693) (0.0618)
Standard Deviation (25 days) 0.0347 0.0368 -3.3910 ***
0.0365 0.0408 -5.1939 ***
(0.0005) (0.0007)
Age 9.0000 7.0000 5.9970 ***
16.9420 13.4990 3.8367 ***
(0.6742) (0.5972)
% high-tech lPOs 43.66% 52.60%
% lPOs on NYSE/AMEX 16.28% 17.10%
% venture capitalist backed IPOs 50.48% 36.37%
43.76% 35.78%

% lPOs with secondary offering

 

77

Panel B: Non-sued IPO Firms vs. Sued IPO Firms in the Pre-PSLRA Period vs. Post-

 

 

 

 

 

PSLRA Period
Pre-PSLRA Period Post-PSLRA Period
Non-sued Sued Non-sued Sued
lPOs lPOs lPOs lPOs
Median Median Wilcoxon Median Median Wilcoxon
test test
Mean Mean T-test Mean Mean T-test
(SE) (SE) (SE) (SE)
Initial Return 0.0625 0.0682 -0.4780 0.0957 0.1379 -1.6450 *
0.1209 0.1304 -0.4013 0.1635 0.3805 -4.7943 ***
(0.0062) (0.0225) (0.0077) (0.1322)
Price Update 0.0000 0.0690 -2.7910 *** 0.0000 0.0667 -1.8650 *
-0.0056 0.0508 -2.3081 ** -0.0045 0.0421 -1.5114
(0.0064) (0.0217) (0.0066) (0.0273)
Market 42.40 62.00 -3.1380 *** 74.70 78.70 -1.0510
Capitalization 185.00 152.00 0.1468 261.00 192.00 0.2867
(60.10) (41.30) (52.10) (44.70)
Proceeds 17.90 21.80 -3.5360 *** 22.40 23.50 -l.2810
Raised 33.10 44.10 -1.3639 40.20 40.70 -0.0301
(2.07) (8.79) (3.50) (6.64)
Underwriter 8.0000 8.0000 -2.9370 *** 8.0000 8.0000 -l.6580 *
Rank 7.0119 7.9219 -3.3l45 *** 7.2916 7.7700 -l.6315
(0.0732) (0.1449) (0.0638) (0.2246)
Standard 0.0345 0.0371 -l.2530 0.0366 0.0403 -2.5460 **
Deviation 0.0364 0.0379 -0.8154 0.0401 0.0556 -4.9204 ***
(25 days) (0.0005) (0.0015) (0.0006) (0.0054)
Age 9.0000 7.0000 2.0610 ** 7.0000 7.0000 0.5050
17.3341 11.7344 2.1236 ** 13.5912 11.6531 0.6906
(0.7127) (1.6388) (0.6161) (2.3453)
% high-tech 43.20% 50.00% 52.24% 60.00%
lPOs
% lPOs on 16.21% 17.19% 17.06% 18.00%
NYSE/AMEX
% venture 49.83% 59.38% 36.49% 34.00%
capitalist
backed lPOs
% lPOs with 43.20% 51.56% 35.28% 46.00%
secondary
offering

 

*, **, *** indicate signiﬁcance at the 10%. 5%. 1% levels, respectively.
Using two-tailed t-test and Wilcoxon test.

78

Table 1.4

Descriptive Statistics of Instrumental Variables and Matched Sample Proxies

The ﬁnal sample consists of 946 lPOs from 1990 to 1993 and 1.076 IPOS from 1996 to
1998 that are available in the Thomson Financial Securities Data (SDC) database.
Closed-end funds, unit offerings. RElTs. ﬁnancial ﬁrms. reverse LBOS. ADRs, spin-offs,
lPOs with an offer price below $5.00 per share, ﬁrms without exchange information and
ﬁrms listed in the small capital market or the OTC market are excluded. An IPO ﬁrm is
identiﬁed as sued if the federal class action lawsuits related to its IPO are listed in the
Security Class Action Alert newsletter between January 1990 and August 2002. There are

64 sued lPOs from 1990 to 1993 and 50 sued lPOs from 1996 to 1998. Some variables

. . . . 387
are based on fewer observations due to missmg data. Turnover 15 equal to [1- “1:22 (1-

volume tradedt/total shares] with the IPO offering date as t = 0. Matched sample consists
of ﬁrms in the same three-digit, two-digit. one-digit or zero-digit SIC code with market
capitalization within 80—120% of IPO ﬁrm’s market capitalization at the close of the ﬁrst
trading day. Matched turnover is the average of matched ﬁrms’ turnovers, which are
based on daily returns over a one-year interval prior to the IPO offering date. Prior
market return is equal to the compounded value-weighted NYSE/AMEX/NASDAQ daily
returns (including dividend) within 15 days prior to the IPO offering date. Matched SD is
the average of matched ﬁrms’ standard deviations, which are based on daily returns over
a one-year prior interval to the IPO offering date. Matched IPO SD (one year) is the
average standard deviation (one year) of all matched lPOs in the matched IPO sample,
which includes all lPOs with the same l-digit SIC code and issued between day -388 and
day -753. Matched IPO SD (half year) is the average standard deviation (half year) of all
matched IPOs in the matched IPO sample, which includes all lPOs with the same l-digit
SIC code and issued between day -205 and day -387. All other variables are deﬁned in
Table 1.3. Panel A compares the IPO ﬁrms’ characteristics in the pre-PSLRA period to
that in the post-PSLRA period. Panel B compares sued and non-sued ﬁrms in the whole
sample period. The pre-PSLRA period represents the years 1990 to 1993. The post-
PSLRA period represents the years 1996 to 1998.

79

71% it
this.
{iii
n and
rm ii
n 111:
i are
tilts

5155
iil‘ii
IliSI

Panel A: Pre—PSLRA Period vs. Post-PSLRA Period

 

 

 

Pre-PSLRA Post-PSLRA
Period Period
Median Median Wilcoxon test
Mean Mean T-test
(SE) (SE)
Turnover 0.6399 0.6484 -1.7860 *
0.6303 0.6496 -2.1097 **
(0.0066) (0.0063)
Matched Turnover 0.4937 0.5722 -10.3210 ***
0.4838 0.5566 -9.7l72 ***
(0.0058) (0.0049)
Prior Market Return 0.0065 0.0215 -9.7260 ***
0.0069 0.0166 -7.7586 ***
(0.0007) (0.0010)
Standard Deviation (25 days) 0.0347 0.0368 -3.3910 ***
0.0365 0.0408 -5.l939 ***
(0.0005) (0.0007)
Matched SD 0.0447 0.0474 -3.3640 ***
0.0466 0.0475 -1 .2481
(0.0006) (0.0004)
Matched IPO SD (one year) 0.0457 0.0471 -12.1360 ***
0.0450 0.0476 -l3.3214 ***
(0.0001) (0.0001)
Matched IPO SD (half year) 0.0417 0.0465 -14.8840 ***
0.0418 0.0458 -16.7905 ***
(0.0002) (0.0001 )

 

80

Panel B: Non-sued IPO Firms vs. Sued IPO Firms
Non-sued IPO

 

 

Firms Sued IPO Firms
Median Median Wilcoxon test
Mean Mean T-test
(SE) (SE)
Turnover 0.6376 0.7550 -5.2850 ***
0.6347 0.7392 -5.3066 ***
(0.0047) (0.0180)
Matched Turnover 0.5338 0.5876 -3.3450 ***
0.5198 0.5684 -2.9396 ***
(0.0040) (0.0141)
Prior Market Return 0.0116 0.0066 1.8500 *
0.0124 0.0072 1.8502 *
(0.0007) (0.0029)
Standard Deviation (25 days) 0.0358 0.0383 -2.3930 **
0.0384 0.0456 -4.0200 ***
(0.0004) (0.0027)
Matched SD 0.0462 0.0458 0.7820
' 0.0472 0.0449 1.4947
(0.0004) (0.0013)
Matched IPO SD (one year) 0.0464 0.0492 -3.7150 ***
0.0463 0.0478 -3.4522 ***
(0.0001) (0.0004)
Matched IPO SD (half year) 0.0448 0.0459 -2.0350 **
0.0439 0.0452 -2.3679 **
(0.0001) (0.0006)

 

*, **, *** indicate signiﬁcance at the 10%, 5%, 1% levels, respectively.

Using two-tailed t-test and Wilcoxon test.

81

Table 1.5

Insurance Effect Regression Results

These regressions test the insurance effect of the litigation risk hypothesis using both the
two-stage estimation method and the generated instrument approach. where litigation risk
is treated as the endogenous variable. The ﬁnal sample size consists of 1,929 IPOS issued
between the years 1990 and 1998 (with 1994 and 1995 excluded). Column 1 is the ﬁrst-
stage probit ML regression. Column 2 is the second-stage OLS under the two-stage
estimation method. Column 3 is the second-stage ZSLS under the generated instrument
approach. Lawsuit = I if an IPO is sued under the Securities Acts of 1933 and/or the
Securities Exchange Act of 1934 between January 1990 and August 2002, and 0
otherwise. Lawsuit Index = predicted index function from the ﬁrst-stage probit ML
regression. PSLRA = I if an IPO is issued in the post-PSLRA period (years 1996 to
1998), and 0 if an IPO is issued in the pre-PSLRA period (years 1990 to 1993).
Ln(Market Cap) = the natural logarithm of market capitalization. UW rank = IPO lead
underwriter rank. VC backed = I if an IPO is venture capitalist backed and 0 otherwise.
Ln(l+Age) = the natural logarithm of (1+ Age). High-tech = I if an IPO ﬁrm is in the
high-tech industry and 0 otherwise. NYSE-AMEX = 1 if an IPO ﬁrm is listed on the
NYSE or AMEX and 0 otherwise. Insider selling = 1 if an IPO ﬁrm has a secondary
offering in the IPO process and 0 otherwise. All other variables are deﬁned in earlier
tables. The standard errors of the two-stage estimation method are calculated using the
non-parametric bootstrap. All standard errors are reported in parentheses.

ndl’i

If to
831.
la.
1 1513.
the
III:
his
list
the

 

 

 

 

Variable F irst-stage Second-stage (Dep.=lnitial Return)
(Dep.=Lawsuit) Two-stage Generated IV
Estimation ﬂiproach
Lawsuit Index 0.0140
(0.0300)
Lawsuit 0.2828
(0.2104)
PSLRA -0.3843 ***
(0.1100)
Prior Market Return -2.3124 0.8909 *** 0.9465 ***
(1.6812) (0.2741) (0.2034)
Ln(Market Cap) 0.0308 0.0413 *** 0.0415 ***
(0.0507) (0.0096) (0.0058)
UW rank 0.0794 ** -0.0102 ** -0.0112 ***
(0.0348) (0.0049) (0.0037)
VC backed -0.0494 -0.0187 -0.0193
(0.1090) (0.0118) (0.0121)
Ln(l+Age) -0.1117 ** -0.0246 *** -0.0231 ***
(0.0527) (0.0037) (0.0061)
High-tech 0.0528 0.0244 *** 0.0238 *
(0.1 100) (0.0088) (0.0123)
NYSE-AMEX 0.0343 -0.0713 *** -0.0718 ***
(0.1410) (0.0146) (0.0160)
Insider Selling 0.1 139 -0.0080 -0.0100
(0.0992) (0.0106) (0.01 18)
Price Update 0.4342 * 0.4600 *** 0.4494 ***
(0.2547) (0.0550) (0.0325)
Match SD 1.4835 -0.4904 ** -0.4741
(3.9089) (0.2478) (0.4257)
Matched IPO SD (one year) 43.2009 *** 3.7577 *** 3.2513 **
(12.3117) (1.3065) (1.4370)
Intercept -4.4336 *** -0.5873 *** -0.6040 ***
(1.0792) (0.1574) (0.1 170)
Pseudo R2 (Adj R2) 0.0537 02414 03069

 

*, **, *** indicate signiﬁcance at the 10%, 5%, 1% levels, repectively.
Using two-tailed tests.

83

Table 1.6

Deterrence Effect Regression Results

These regressions test the deterrence effect ofthe litigation risk hypothesis using the two-
stage estimation method, where initial return is treated as the endogenous variable. The
ﬁrst two columns show the regression results under full sample, which consists of 1,929
lPOs issued between the years 1990 and 1998 (with 1994 and 1995 excluded). The
second two columns show the regression results under the subsample after dropping
dismissed/terminated (DfF) cases. which consists of 1.892 observations. Columns 1 and 3
show the ﬁrst-stage OLS regression results. Columns 2 and 4 show the second-stage
probit ML regression results. Predicted initial return = predicted value from ﬁrst-stage
OLS regression. All other variables are deﬁned in earlier tables. The standard errors of
the two-stage estimation method are calculated using the non-parametric bootstrap. All
standard errors are reported in parentheses.

84

{.1

 

Variable

Full sample

Drop D/T cases

 

 

 

F irst-stage Second-stage F irst-stage Second-stage
(Dep.=lnitial (Dep.=lnitial
Return) (Dep.=Lawsuit) Return) (Dep.=Lawsuit)
Predicted Initial -2.6937 -2.6404
Return (1.7502) (2.2786)
Prior Market 0.8584 *** 0.8665 ***
Return (0.1871) (0.1897)
PSLRA -0.0054 -0.3989 *** -0.0067 -0.4042 ***
(0.0119) (0.1008) (0.0121) (0.1282)
Ln(Market Cap) 0.0417 *** 0.1432 0.0429 *** 0.0973
(0.005 8) (0.0879) (0.0059) (0.1074)
UW rank -0.0091 *** 0.0548 -0.0093 *** 0.0633 *
(0.0033) (0.0347) (0.0034) (0.0372)
VC backed -0.0194 -0.1017 -0.0197 -0.2577 **
(0.0120) (0.1038) (0.0122) (0.1286)
Ln(l+Age) -0.0261 *** -O.1821 ** -0.0267 *** -0.1753 **
(0.0056) (0.0720) (0.0057) (0.0852)
High-tech 0.0251 ** 0.1205 0.0261 ** 0.0741
(0.0120) (0.1192) (0.0122) (0.1322)
NYSE-AMEX -0.0708 *** -0.1565 -0.0720 *** -0.2467
(0.0157) (0.1854) (0.0159) (0.2425)
Insider Selling -0.0064 0.0967 -0.0049 0.0727
(0.01 12) (0.0902) (0.01 14) (0.11 17)
Price Update 0.4661 *** 1.6897 * 0.4650 *** 1.6277
(0.0291) (0.8668) (0.0294) (1.0987)
Match SD -O.4696 0.2185 -0.4843 3.4519
(0.4182) (3.5265) (0.4222) (3.801 1)
Matched IPO SD 4.3634 *** 54.9547 *** 4.5206 *** 55.5730 ***
(one year) (1.2532) (15.2009) (1.2696) (21.5055)
Intercept -0.6495 *** -6.1833 *** -0.6731 *** -5.6795 ***
(0.1173) (1.5233) (0.1192) (2.0393)
Pseudo R2 (Adj R2) 0.2414 0.0537 0.2433 0.0503

 

*, **, *** indicate signiﬁcance at the 10%, 5%, 1% levels, repectively.
Using two-tailed tests.

85

Table 1.7

Regression Results of the Deterrence Effect in the pre- and post-PSLRA Periods

These regressions test the deterrence effect of the litigation risk hypothesis for both the
pre- and post-PSLRA periods. Initial return is the endogenous variable with prior market
return as the instrument for it. The regressions in the ﬁrst two columns are based on the
pre-PSLRA period sub-sample, which consists of 907 lPOs issued between the years
1990 and 1993. The regressions in the second two columns are based on the post-PSLRA
period sub-sample, which consists of 1,022 lPOs issued between the years 1996 and
1998. Columns 1 and 3 are the ﬁrst-stage OLS regression. Columns 2 and 4 are the
second-stage probit ML under the two-stage estimation method. All variables are deﬁned
in earlier tables. The standard errors of the two-stage estimation method are calculated
using the non-parametric bootstrap. All standard errors are reported in parentheses.

86

eriods

llt‘llt lllf
r market
0 on the
he )tm
PSLRI
0% and
are the
defined
lculttetl

1.

 

Variable

Pre-PSLRA Period

Post-PSLRA Period

 

 

 

First-stage Second-stage F irst-stage Second-stage
(Dep.=1nitial (Dep.=lnitial
Return) (Dep.=Lawsuit) Return) (Dep.=Lawsuit)
Predicted Initial -2.3811 -3.6285
Return (2.5266) (2.7106)
Prior Market 1.0740 *** 0.7003 ***
Return (0.2345) (0.2688)
Matched Turnover 0.0898 *** 0.5034 0.1117 * 1.5091 ***
(0.0330) (0.4843) (0.0640) (0.4843)
Ln(Market Cap) 0.0144 ** 0.0925 0.0629 *** 0.1922
(0.0057) (0.0718) (0.0110) (0.1951)
UW rank -0.0137 *** 0.0739 -0.0060 0.0297
(0.0033) (0.0532) (0.0056) (0.0536)
VC backed -0.0069 0.0160 -0.0396 ** -0.3302
(0.0122) (0.1740) (0.0199) (0.2149)
Ln(l+Age) -0.0301 *** -0.2446 ** -0.0242 *** -0.1594
(0.0060) (0.0952) (0.0090) (0.1019)
High-tech 0.0085 0.0328 0.0200 0.1385
(0.0123) (0.1646) (0.0202) (0.1395)
NYSE-AMEX -0.0218 -0.0363 -0.1127 *** -0.3015
(0.0153) (0.2009) (0.0266) (0.3585)
Insider Selling 0.0156 0.1115 -0.0195 0.0926
(0.01 11) (0.1487) (0.0186) (0.1578)
Price Update 0.4054 *** 1.5067 0.4694 *** 2.1614
(0.0307) (1.1669) (0.0469) (1.3362)
Match SD -1.0662 *** -4.8507 0.4654 11.0817 *
(0.3479) (5.8115) (0.8981) (6.2036)
Intercept 0.0378 -2.9187 ** -0.9423 *** -5.7905 *
(0.0981) (1.1462) (0.2068) (3.0495)
Pseudo R2 (Adj R2) 0.2858 0.0571 0.2325 0.0420

 

*, **, *** indicate signiﬁcance at the 10%, 5%, 1% levels, repectively.
Using two-tailed tests.

87

Table 1.8

Performance Comparison of PSLRA and Matched Turnover

These regressions compare the performance of PSLRA and matched turnover by testing
the insurance effect using two types of model set-ups, where litigation risk is the
endogenous variable. In the type (1) model. matched turnover is a control variable and
PSLRA is an instrument. In type (2) model. PSLRA is a control variable and matched
turnover is an instrument. We run tests under both two-stage estimation method and
generated instrument approach. The ﬁnal sample size consists of 1.929 lPOs issued
between the years 1990 and 1998 (with 1994 and 1995 excluded). Column 1 is the ﬁrst-
stage probit ML regression. Columns 2 and 3 are the second-stage OLS under the two-
stage estimation method. Columns 4 and 5 are the second-stage 2SLS under the generated
instrument approach. All variables are deﬁned in earlier tables. The standard errors of the
two-stage estimation method are calculated using the non-parametric bootstrap. All
standard errors are reported in parentheses.

88

 

Variable

Second-stagg(Dcp.=lnitial Return)

 

 

 

 

First-stage Two-stage Estimation Generated 1V approach
(I)cp.=La\\'suit) (1) (2) (1 ) (2)
Lawsuit Index 0.0208 0.1543 ***
(0.0291) (0.0437)
Lawsuit 0.3449 0.8281 ***
(0.2103) (0.3229)
PSLRA -0.4069 *** 0.0543 *** 0.0281
(0.1112) (0.0169) (0.0199)
Matched Turnover 0.6271 * 0.0837 *** 0.0763 **
(0.3363) (0.0261) (0.0384)
Prior Market Return -2.4070 0.9021 *** 1.2233 *** 0.9577 *** 1.0730 ***
(1.6888) (0.2750) (0.2857) (0.2066) (0.2502)
Ln(Market Cap) 0.0055 0.0374 *** 0.0366 *** 0.0380 *** 0.0390 ***“
(0.0524) (0.0096) (0.0096) (0.0062) (0.0074)
UW rank 0.0711 ** -0.0115 ** ~0.0210 *** -0.0123 *** -0.0152 ***
(0.0350) (0.0047) (0.0052) (0.0037) (0.0048)
VC backed -0.0538 -0.0197 * -0.0125 -0.0202 -0.0159
(0.1091) (0.0119) (0.0120) (0.0123) (0.0152)
Ln(l+Age) -0.1088 ** —0.0235 *** -0.0090 * -0.0221 *** -0.0160 **
(0.0527) (0.0037) (0.0054) (0.0061) (0.0081)
High-tech 0.0117 0.0190 ** 0.0174 * 0.0187 0.0202
(0.1119) (0.0089) (0.0090) (0.0126) (0.0152)
NYSE-AMEX 0.0342 -0.0704 *** -0.0750 *** -0.0710 *** -0.0725 ***
(0.1418) (0.0146) (0.0145) (0.0163) (0.0197)
Insider Selling 0.1078 -0.0093 -0.0237 ** -0.0113 -0.0162
(0.0996) (0.0105) (0.0106) (0.0119) (0.0146)
Price Update 0.3883 0.4505 *** 0.3987 *** 0.4401 *** 0.4213 ***
(0.2571) (0.0554) (0.0596) (0.0326) (0.0406)
Match SD 0.6369 -0.6156 ** -0.7005 *** -O.5840 -0.5426
(4.0572) (0.2390) (0.2392) (0.4358) (0.5277)
Matched IPO SD 42.5245 *** 3.2538 ** -2.4212 2.7981 * 0.6483
(one year) (12.3586) (1.2835) (2.0693) (1.4398) (2.1431)
Intercept -4.1526 *** -0.5099 **"‘ 0.0443 -0.5485 *** -0.4659 "*
(1.0929) (0.1522) (0.2171) (0.1198) (0.1643)
F 48.8300 48.8300 45.4400 30.8000
Prob > F 0.0000 0.0000 0.0000 0.0000
Pseudo R2 (Adj R2) 0.0578 0.2439 0.2439 0.1843

 

*Signiﬁcant at the 10% level, “Signiﬁcant at the 5% level, ***Signiﬁcant at the 1% level.
Using two-tailed tests.

89

APPENDIX 2

TABLES OF CHAPTER 2

90

Table 2.1

Distribution of Earning Management Level

The sample consists of US. common stocks for which daily return data are available on
the ﬁles of the Center for Research in Security Prices at the University of Chicago. This
table shows the distribution of modiﬁed-Jones model abnormal accruals (AA) and
performance-matched abnormal accruals (PMAA) of different portfolios based on their
litigation risk level and earning management level. The positive earning management
portfolio (POS EM) is composed of ﬁrms with AA or PMAA included in the top one-
third distribution; the zero earning management portfolio (ZERO EM) is composed of
ﬁrms with AA or PMAA included in the middle one -third distribution; the negative
earning management portfolio (NEG EM) is composed of ﬁrms with AA or PMAA
included in the bottom one-third distribution. The high litigation risk group includes all
ﬁrms in ﬁve industries: computer (SIC codes 3570 — 3577 and 7370 — 7374), electronics
(SIC codes 3600 — 3674), pharmaceuticals/biotechnology (SIC codes 2833 — 2836 and
8731 — 8734), retailing (SIC codes 5200 — 5961), and accounting (SIC codes 8720 and
8721). Firms that are not in these ﬁve industries are included in the low litigation risk
group. Panel A presents the distribution of modiﬁed-Jones model abnormal accruals.
Panel B presents the distribution of performance-matched modiﬁed-Jones model
abnormal accruals. All standard errors (S.E.) are reported in parentheses.

91

Panel A: Distribution of Modiﬁed-Jones Model Abnormal Accruals

 

 

 

 

 

 

 

1994 before PSLRA
EM Group Variables Full sample High Litigation Low Litigation risk
POS EM Mean AA 0.112 *** 0.127 *** 0.107 ***
(S.E.) ‘ (0.004) (0.009) (0.005)
Median AA 0.076 0.090 0.074
N 704 156 548
ZERO EM Mean AA -0.001 ** -0.002 ** 0001
(SE) (0.001) (0.001) (0.001)
Median AA -0.001 -0.002 -0.001
N 660 156 504
NEG EM Mean AA -0.111 *** -0.122 *** —0.106 ***
(S.E.) (0.004) (0.008) (0.005)
Median AA 0077 -0.090 -0.074
N 630 187 443
2001 before SOA
EM Group Variables Full sample High Litigation Low Litigation risk
POS EM Mean AA 0.105 *** 0.106 *** 0.105 ***
(S.E.) (0.003) (0.004) (0.003)
Median AA 0.078 0.082 0.075
N 1255 421 834
ZERO EM Mean AA 0.009 *** 0.007 *** 0.009 ***
(S.E.) (0.000) (0.001) (0.001)
Median AA 0.010 0.007 0.011
N 1388 399 989
NEG EM Mean AA -0.111 *** -0.123 *** -0.104 ***
(S.E.) (0.003) (0.006) (0.004)
Median AA 0075 -0.079 -0.073
N 1182 465 717

 

Panel B: Distribution of Performance-Matched Modiﬁed-Jones Model Abnormal

 

 

 

 

 

 

 

Accruals
1994 before PSLRA
EM Group Variables Full sample High Litigation Low Litigation risk
POS EM Mean PMAA 0.105 *** 0.108 *** 0.104 ***
(S.E.) (0.004) (0.007) (0.005)
Median PMAA 0.070 0.080 0.067
N 683 163 520
ZERO EM Mean PMAA 0.000 0.001 0.000
(S.E.) (0.001) (0.001) (0.001)
Median PMAA 0.000 0.001 0.000
N 651 153 498
NEG EM Mean PMAA -0.101 *** -0.109 *** -0.098 ***
(S.E.) (0.004) (0.008) (0.004)
Median PMAA -0.070 -0.079 -0.067
N 660 183 477
2001 before SOA
EM Group Variables Full sample High Litigation Low Litigation risk
POS EM Mean PMAA 0.097 *** 0.101 *** 0.095 ***
(S.E.) (0.003) (0.004) (0.003)
Median PMAA 0.068 0.074 0.065
N 1236 431 805
ZERO EM Mean PMAA -0.001 0.000 -0.001
(S.E.) (0.000) (0.001) (0.000)
Median PMAA 0.000 0.001 -0.001
N 1333 401 932
NEG EM Mean PMAA -0.102 *** -0.115 *** -0.094 ***
(S.E.) (0.003) (0.006) (0.003)
Median PMAA -0.070 -0.077 -0.067
N 1256 453 803

 

*, **, *** indicate signiﬁcance at the 10%, 5%, 1% levels, respectively.

Two-tailed tests are used.

93

Table 2.2

Distribution of Shareholder Rights Level

This table shows the distribution of Gompers, Ishii, and Metrick (2003)'s “Shareholder
Rights Index” of different portfolios based on their litigation risk level and G level. The
low G group includes their groups #1, 2, and 3 (G37), who are the ﬁrms with strong
shareholder rights. The mid G group. includes their groups #4. 5, 6 and 7 (8SGSI 1), who
are the ﬁrms with average shareholder rights. The high G group includes their groups #8,
9, and 10 (G212), who are the ﬁrms with low shareholder rights. Litigation risk level has
been deﬁned in Table 2.1. All standard deviations (S.D.) are reported in parentheses.

94

 

1995

 

 

 

 

 

 

G Group Variables Full sample High Litigation Low Litigation risk
Low G Mean G 5.793 5.796 5.792
(S.D.) (1.205) (1.194) (1.211)
Median G 6.000 6.000 6.000
N 386 93 293
Mid G Mean G 9.509 9.449 9.522
(S.D.) (1.101) (1.118) (1.097)
Median G 10.000 9.000 10.000
N 690 127 563
High G Mean G 12.970 13.000 12.965
(S.D.) (1.089) (1.054) (1.097)
Median G 13.000 13.000 13.000
N 338 55 283
2002
G Group , Variables Full sample High Litigation low Litigation risk
[ow G Mean G 5.947 6.059 5.893
(S.D.) (1.139) (1.111) (1.150)
Median G 6.000 6.000 6.000
N 514 169 345
Mid G Mean G 9.413 9.064 9.548
(S.D.) (1.100) (1.052) (1.089)
Median G 9.000 9.000 10.000
N 953 265 688
High G Mean G 13.083 13.000 13.100
(S.D.) (1.178) (1.160) (1.184)
Median G 13.000 13.000 13.000
N 312 53 259

 

Table 2.3

Legislative Event Dates

This table lists the legislative event dates for the passage of the Private Securities
Litigation Reform Act of 1995 (PSLRA) and the Sarbanes-Oxley Act of 2002 (SOA).
Panel A refers to part of Ali and Kallapur (2001)'s Table 1 and lists events, event
numbers, event dates, and predictions of whether they increased (1) or decreased (D) the
likelihood of the passage of PSLRA. Panel B refers to both Rezaee and Jain (2003) and
Li, Pincus, and Rego (2004) and lists events, event numbers, event dates, and predictions
01“ Whether they increased (1) or decreased (D) the likelihood of the passage of SOA.

96

Panel A: Legislative Event Dates Leading to the Passage of the Private Securities
Litigation Reform Act of 1995

 

 

 

Event # Date Event

1 19950104 Common Sense Legal Reform Act introduced in the
House.

2 199501 18 Private Securities Litigation Reform Act of 1995
introduced in the Senate.

3 19950125 SEC Chief warns legislators that changes envisaged by
Republicans would impair the nation's securities
regulation system.

4 19950209 Legislators modify bill to meet some of SEC's

5 19950214 Finance Subcommittee passes the bill, 16-10.

6 19950215 SEC rejects revised bill.

7 19950216 House Commerce Committee passes the bill, 32-10.

8 19950308 House passes its bill, 325-99.

9 19950519 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.

10 19950524 Senator D'Amato complies.

1 1 19950525 Senate Banking Committee approves the bill, 1 1-4.

12 19950628 Senate passes its bill, 70-29.

13 19951024 House and Senate staffers hammer out a compromise.

14 19951 1 16 SEC approves safe harbor provisions, making a
presidential veto less likely.

15 19951128 Conference Committee ﬁnalizes the bill.

16 19951205 Senate approves the bill, 65-30.

17 19951206 House approves the bill, 320—102.

18 1995121 1 President indicates willingness to sign the bill.

19 19951218 Clinton is ofﬁcially undecided. Rumors are that he will
veto the bill.

20 19951220 President vetoes bill (on 12/19/1995 after market
close), and House overrides veto, 319-100.

21 19951221 Senate poised to override the veto.

22 19951222 Senate overrides veto, 68-30.

 

97

Panel B: Legislative Event Events Leading to the Passage of the Sarbanes-Oxley

 

 

Act of 2002
Event # Date Prediction Event

1 200201 17 SEC. Chairman Harvey Pitt proposes oversight board.

2 20020212 Legislation to be introduced in the House.

3 20020214 The House of Representatives (Oxley, the Committee on
Financial Services) introduced HR. 3763 to protect
investors by improving the accuracy and reliability of
corporate disclosures.

4 20020308 Introduction of S. 2004 in the Senate.

5 20020424 House considers and passes HR. 3763.

6 20020425 Senate Judiciary approves legislation.

7 20020618 The Senate Banking Committee approves (17-4) a bill to
tighten regulation of the accounting profession,
corporate executives, and ﬁnancial analysts.

8 20020620 The SEC proposed the creation of a nine-member Public
Accountability Board to oversee the accounting
profession.

9 20020625 Senator Sarbanes introduced $2673 to ( 1) improve the
quality and transparency in ﬁnancial reporting; (2)
designate an independent Public Accounting Board; (3)
enhance the standard setting process for accounting
practices; and (4) improve SEC resources and oversight.

10 20020703 Committee Report on S. 2673.

1 1 20020708 Senate considers S. 2673.

12 20020709 Bush’s Wall Street speech.

13 20020715 The Senate passed S. 2673.

14 20020716 The House passed HR. 5118.

15 20020719 There were some uncertainties regarding the form,
content, and the possibility of passage of the Act.

16 20020724 The Congressional Conference Committee reached
an agreement on comprehensive reform legislation
(the Sarbanes-Oxley Act of 2002).

17 20020725 Congress passed the Sarbanes-Oxley Act of 2002 by
a vote of 423-3 in the House and 99-0 in Senate.

18 20020730 President Bush signed into law the Sarbanes-Oxley

Act of 2002.

98

Table 2.4

Comparisons of Portfolio CARs around PSLRA and SOA Event Dates

This table compares the Cumulative Portfolio Abnormal Returns (CARp) during the
passage of PSLRA and SOA. CARp are calculated based on methods developed in
Brown and Warner (1985). For every security. abnormal return is estimated using the
OLS market model. The estimation window is [1/1/1994, 12/31/1994] for PSLRA (252
trading days) and [1/1/2001. 12/31/2001] for SOA (248 trading days). The event window
is [12/4/1995, 12/26/95] for PSLRA (16 trading days) and [7/23/2002, 7/31/2002] for
SOA (7 trading days) . Firms with less than 252/248 trading days data in the estimation
window or 16/7 trading days data in the event window are excluded from the ﬁnal
sample. Panel A shows the results by dividing the sample into portfolios based on
litigation risk level and abnormal accrual level. Panel B shows the results by dividing the
sample into portfolios based on litigation risk level and performance-matched abnormal
accrual level. Panel C shows the results by dividing the sample into portfolios based on
litigation risk level and shareholder rights level. All standard errors (S.E.) are reported in
parentheses.

Panel A: CARp of Portfolios Based on Litigation Risk Level and Abnormal
Accrual Level

 

 

 

PSLRA SOA

AA EM High Low Litigation High Litigation Low Litigation
group Variables Litigation Risk Risk Risk Risk
POS EM CARp -5.629 ** -0.763 -10.488 *** -3.790 ***

(S.E.) (2.285) (1.437) (2.470) (1.444)

N 137 477 383 754
ZERO EM CARp -2.603 1.022 -9.873 *** —3.1 19 **

(S.E.) (1.856) (1.317) (2.266) (1.296)

N 152 476 382 935
NEG EM CARp -5.633 *** -0.231 -14.064 *** -5.119 ***

(S.E.) (2.087) (1.553) (3.027) (1.686)

N 166 384 404 629

 

99

\\

Panel B: CARp of Portfolios Based on Litigation Risk Level and Performance

Matched Abnormal Accrual Level

 

 

 

PSLRA SOA

PMAA High Low Litigation High Litigation Low Litigation
EM group Variables Litigation Risk Risk Risk Risk
POS EM CARp -5.333 *** -0.355 -12.443 *** -4.326 ***

(S.E.) (2.037) (1.500) (2.755) (1.525)

N 145 450 391 712
ZERO EM CARp -2.202 0.701 -9.778 *** -2.639 **

(S.E.) (1.958) (1.217) (2.357) (1.306)

N 147 464 369 876
NEG EM CARp -6.165 *** -0.315 -12.221 *** -4.934 ***

(S.E.) (2.095) (1.431) (2.669) (1.543)

N 163 423 409 730

 

Panel C: CARp of Portfolios Based on Litigation Risk Level and Shareholder

Rights Level

PSLRA

SOA

 

 

High Low Litigation High Litigation Low Litigation
C group Variables Litigation Risk Risk Risk Risk
Low G CARp -5.506 *** -0.264 -18.852 *** -4.643 ***
(S.E.) (1.934) (0.836) (3.751) (1.429)
N ' 94 299 165 343
Mid G CARp ~3.948 ** -0.004 -13.457 *** -1.047
(S.E.) (1.615) (0.695) (2.674) (1.271)
N 122 558 261 666
High G CARp -1.988 0.212 -5.906 *** 0.580
(S.E.) (1.399) (0.703) (1.973) (1.325)
¥ N 52 277 56 251

 

=0:

TWO-tailed tests are used.

100

, **, *** indicate signiﬁcance at the 10%, 5%, 1% levels, respectively.

Table 2.5

Results of Equal-Weighted Portfolio OLS Approach for Different EM and High
Litigation Risk Groups

This table presents the equal-weighted portfolio OLS estimate results for shareholders of
high litigation risk and different earning management portfolios. There are 22 event dates
for the passage of PSLRA and 18 event dates for the passage of SOA. 1995 daily stock
return data is used for PSLRA and 2002 data is used for SOA. Each portfolio consists of
ﬁrms with 252 trading days in either 1995 or 2002. High litigation risk portfolios are
further divided into three groups based on their abnormal accruals (AA) or performance-
matched abnormal accruals (PMAA) level. A dummy variable is used that equals to 1
when the day corresponds to the event date as deﬁned in Table 2.3 and 0 otherwise. Panel
A reports the estimation results for the passage of PSLRA. Panel B reports the estimation
results for the passage of SOA. All standard errors (S.E.) are reported in parentheses.

101

Panel A: Equal-Weighted Portfolio OLS Estimation Results for the Passage of PSLRA
AA Portfolios PMAA Portfolios

 

 

 

Dummy 0"" POS EM ZERO EM NEG EM POS EM ZERO EM NEG EM
D1 19950104 0.220 0.201 -0.371 -0.240 0.101 0.117
(0.522) (0.471) (0.605) (0.525) (0.527) (0.572)
D2 19950118 0.191 0.445 0.333 0.017 -0002 0.879
(0.522) (0.471) (0.605) (0.525) (0.527) (0.572)
D3 19950125 0504 -0575 4.198 ** -0.396 -0.852 4.063 *
(0.522) (0.471) (0.605) (0.524) (0.527) (0.571)
D4 19950209 0.197 0.799 * 0.620 0.366 0.537 0.714
(0.522) (0.471) (0.605) (0.524) (0.527) (0.572)
D5 19950214 -0159 0.124 0.318 0.125 -0473 0.589
(0.522) (0.471) (0.605) (0.524) (0.527) (0.571)
D6 19950215 -0.075 -0.360 0.065 -0.208 0.158 -0.269
(0.523) (0.471) (0.605) (0.525) (0.528) (0.572)
D7 19950216 -0.065 0.349 0.092 0.181 -0.136 0.302
(0.522) (0.471) (0.605) (0.524) (0.527) (0.572)
D8 19950308 0.406 0.097 0.047 0.432 0.631 -0444
(0.522) (0.471) (0.605) (0.524) (0.527) (0.571)
D9 19950519 0.183 -0.036 0.574 0.193 0.294 0.280
(0.522) (0.471) (0.605) (0.524) (0.527) (0.572)
D10 19950524 0158 -0.031 -0.780 -0.254 -0.167 -0.578
(0.522) (0.471) (0.605) (0.524) (0.527) (0.571)
D11 19950525 -O.408 -0.037 -0.169 -0.431 -0.174 -0020
(0.522) (0.471) (0.605) (0.524) (0.527) (0.571)
D12 19950628 0535 -0.603 -0.637 -0.761 -0725 -0.335
(0.522 (0.471) (0.605) (0.524) (0.527) (0.572)
D13 19951024 0674 -0.028 -0004 -0.344 -0.280 -0.058
(0.522) (0.471) (0.605) (0.524) (0.527) (0.571)
D14 19951116 -0.370 -0.758 -0.194 -0.228 -0.496 -0.555
(0.523) (0.472) (0.606) (0.525) (0.528) (0.573)
D15 19951128 0.038 0.042 0.283 0.386 -0.233 0.212
(0.524) (0.472) (0.607) (0.527) (0.529) (0.574)
D16 19951205 -0.627 -0.855 * 4.145 * -0.486 -0.795 4.330 **
(0.523) (0.471) (0.605) (0.525) (0.528) (0.572)
D17 19951206 -0.408 0.029 4.044 * -0.435 0.458 -1.378 **
(0.522) (0.471) (0.605) (0.524) (0.527) (0.571)
D18 19951211 4.046 ** -0.640 -0.440 -0.704 -0.724 -0.654
(0.522) (0.471) (0.605) (0.524) (0.527) (0.571 )
D19 19951218 0778 -0.098 -0.987 4.194 ** 0.036 -0715
(0.540) (0.486) (0.625) (0.542) (0.545) (0.590)
020 19951220 0.511 0.723 1.331 ** 0.576 0.886 1.144 **
(0.524) (0.472) (0.607) (0.526) (0.529) (0.573)
D21 19951221 —0.639 0.179 -0.166 -0.762 0.065 0.066
(0.524) (0.472) (0.607) (0.526) (0.529) (0.573)
[)22 19951222 -0.470 -0.216 -0.278 -0301 -0.206 -0.427
_ (0.522) (0.471) (0.605) (0.525) (0.527) (0.572)
N 148 154 175 154 149 174
_/ = 22
D j
j=16
-3.458 ** -0.877 -2.729 * -3.306 H -0.282 -3295 **

 

102

Panel B: Equal-Weighted Portfolio OLS Estimation Results for the Passage of SOA
AA Portfolios

PM 3.4 Portfolios

 

 

 

Dmmy Date POS EM ZERO EM NEG EM POS EM ZERO EM NEG EM
D1 20020117 0.056 0.302 0.448 0.097 0.187 0.516
(0.776) (0.731) (0.928) (0.870) (0.755) (0.827)
D2 20020212 0.059 0.204 0.040 0.094 -0173 0.353
(0.775) (0.731) (0.927) (0.869) (0.754) (0.827)
D3 20020214 -0727 -0.666 -0.882 -0.374 -0797 4.094
(0.775) (0.731) (0.927) (0.869) (0.754) (0.827)
D4 20020308 1.002 1.203 1.686 * 1.022 1.228 1.644 **
(0.776) (0.731) (0.927) (0.870) (0.754) (0.827)
D5 20020424 0.233 -0.230 -0295 -0.061 0.177 -0.394
(0.776) (0.731) (0.927) (0.870) (0.754) (0.827)
D6 20020425 -0543 -0.260 -0313 -0.190 -0331 -0.576
(0.775) (0.731) (0.927) (0.869) (0.754) (0.827)
D7 20020618 0754 -0453 -0433 -0.531 -0.615 -0.492
(0.775) (0.731) (0.927) (0.869) (0.754) (0.827)
D8 20020620 0.205 -0.328 0.018 0.300 -0.026 -0.361
(0.776) (0.732) (0.928) (0.871) (0.755) (0.828)
D9 20020625 .0521 -0100 0.149 -0135 -0.339 0.006
(0.777) (0.732) (0.928) (0.871) (0.755) (0.828)
D10 20020703 -0394 -0539 4.658 r -0.634 -0.798 4.184
(0.776) (0.731) (0.927) (0.870) (0.754) (0.827)
Dll 20020708 -O.561 4.073 -0.656 -0.924 -0.518 -0.831
(0.776) (0.731) (0.928) (0.870) (0.755) (0.827)
D12 20020709 0.646 0.549 0.972 0.807 0.870 0.522
(0.779) (0.734) (0.930) (0.873) (0.757) (0.830)
D13 20020715 -0.251 -0.403 -0.031 -0324 -0.218 -0.l38
(0.776) (0.731) (0.927) (0.870) (0.754) (0.827)
014 20020716 1.538 ** 1.780 ** 1.682 * 2.269 *** 0.951 1.747 **
(0.777) (0.732) (0.928) (0.871) (0.755) (0.828)
D15 20020719 0.547 0.723 1.111 0.799 0.837 0.768
(0.783) (0.738) (0.935) (0.878) (0.761) (0.834)
016 20020724 -2.446 *** -2.097 *** -3457 *** -3.103 *** 4.858 ** -3.026 ***
(0.795) (0.749) (0.950) (0.891) (0.773) (0.847)
017 20020725 4.215 4.257 * 4.860 ** 4.433 4.329 * 4.581 *
(0.776) (0.731 ) (0.927) (0.870) (0.754) (0.827)
D18 20020730 0.300 -0.055 -0.008 -0.215 0.291 0.158
(0.776) (0.731) (0.927) (0.870) (0.754) (0.827)
N 372 378 396 381 364 401
j = 18
Z 0 l
j: 16
-3.362 N -3409 *** -5.326 *** 4751 *** -2896 H -4449 m

 

*, "*, *** indicate signiﬁcance at the 10%, 5%, 1% levels, respectively. Two-tailed tests are used.

103

Table 2.6

Results of Equal-Weighted Portfolio OLS Approach for Different Shareholder
Rights and High Litigation Risk Groups

This table presents the equal portfolio OLS estimate results for shareholders of high
litigation risk and different shareholder rights portfolios. There are 22 event dates for the
passage of PSLRA and 18 event dates for the passage of SOA. 1995 daily stock return
data is used for PSLRA and 2002 data is used for SOA. Each portfolio consists of ﬁrms
with 252 trading days in either 1995 or 2002. High litigation risk portfolios are further
divided into three groups based on their shareholder rights (G) level. A dummy variable
is used that equals to 1 when the day corresponds to the event date as deﬁned in Table 2.3
and 0 otherwise. Panel A reports estimation results for the passage of PSLRA. Panel B
reports estimation results for the passage of SOA. All standard errors (S.E.) are reported
in parentheses.

104

older

01 t0.
:5 for the
:1 return
01 tints
6 11101::
1 3:13:51:
little 3.3

Panel B
{CPO-Rel

Panel A: Equal-Weighted Portfolio OLS Estimation Results for the Passage of PSLRA

 

 

 

Dummy Date Low G Mid G High G
D1 19950104 0.146 0.210 0.029
(0.560) (0.428) (0.423)
D2 19950118 0.133 0.404 0.012
(0.560) (0.428) (0.423)
D3 19950125 0855 -1.246 *** 0.163
(0.559) (0.428) (0.422
D4 19950209 1.050 * 0.866 ** 0.260
(0.560) (0.428) (0.423)
D5 19950214 -0.021 —0.181 0.113
(0.559) (0.428) (0.422)
D6 19950215 -0.113 -0.335 0.119
(0.560) (0.429) (0.423)
D7 19950216 0.001 -0.196 -0.379
(0.560) (0.428) (0.422)
D8 19950308 -0.039 0.198 0.060
(0.559) (0.428) (0.422)
D9 19950519 0371 0.367 -0.126
(0.560) (0.428) (0.423)
D10 19950524 -0.451 -1.202 *** -0.062
(0.559) (0.428) (0.422)
D11 19950525 0.100 0.062 -0.085
(0.559) (0.428) (0.422)
D12 19950628 -0.566 -0.213 -0.047
(0.560) (0.428) (0.423)
D13 19951024 -0.629 -0.259 -0.359
(0.559) (0.428) (0.422)
D14 19951116 -0.779 0.104 0.231
(0.561 ) (0.429) (0.423)
D15 19951128 1.177 ** 1.181 *" -0.247
(0.562) (0.430) (0.424)
016 19951205 -0.910 -0.535 -0.030
(0.560) (0.429) (0.423)
D17 19951206 -0.970 * -0.526 -0.050
(0.559) (0.428) (0.422)
D18 19951211 0354 -0.818 "‘ -0.615
(0.559) (0.428) (0.422)
D19 19951218 0.034 -0.172 -0.508
(0.578) (0.443) (0.437)
020 19951220 0.308 0.238 1.107 ***
(0.561) (0.430) (0.424)
D21 19951221 0.160 0.673 0.822 "
(0.561 ) (0.430) (0.424)
022 19951222 0.556 0.381 0.038
(0.560) (0.428) (0.423)
N 94 123 51
j = 22
j = 16
-1.177 -0.760 0.764

 

105

Panel B: Equal-Weighted Portfolio OLS Estimation Results for the Passaggof SOA

 

 

 

Dummy Date Low G Mid C High G
D1 200201 17 0.582 0.699 0.589
( 1 .242) (0.939) (0.703)
D2 20020212 0.176 0.253 0.236
(1.240) (0.938) (0.702)
D3 20020214 —0.804 -0.877 -0.438
(1.240) (0.938) (0.702)
D4 20020308 I .845 1.230 0.238
(1.241) (0.938) (0.702)
D5 20020424 -0.286 -0.075 -0.505
(1.241) (0.938) (0.702)
D6 20020425 0.586 0.066 0.858
(1.240) (0.938) (0.702)
D7 20020618 -1.290 -1.110 0.013
(1.240) (0.938) (0.702)
D8 20020620 -0.836 -0.571 -0.283
(1.242) (0.939) (0.703)
D9 20020625 -0.837 -0.6 l 3 -0.629
(1.243) (0.940) (0.703)
D10 20020703 0.748 -0.354 -0.324
(1.241) (0.938) (0.702)
D11 20020708 -1.824 -1.638 "' -0.l29
(1.242) (0.939) (0.703)
D12 20020709 0.458 0.668 0.853
(1.245) (0.942) (0.705)
D13 20020715 1.160 0.448 -1.185 *
(1.241) (0.938) (0.702)
D14 20020716 2.624 ** 2.066 ** 0.218
(1.242) (0.939) (0.703)
D15 20020719 2.538 ** 1.510 1.275 *
(1.252) (0.947) (0.709)
D16 20020724 -3.399 *** -2.524 *** -0.566
(1.271) (0.961) (0.720)
DI7 20020725 -3.602 *** -2.857 *** -0.976
(1.241 ) (0.938) (0.702)
D18 20020730 0.158 0.345 -0.198
(1.241) (0.938) (0.702)
N 160 259 54
j = 18
D
j = 16
6.843 *** -5.036 *** -l.741

 

*, **, *** indicate signiﬁcance at the 10%, 5%, 1% levels, respectively. Two-tailed tests are used.

106

2205:23 :_ :2892 2: 3:9 22:: ::m::m:m :< .<Om.:o
uwsmmma 2: 5.: 3:2,: <32: :0 2:22 5:252: 2: 282 Q 35:: .<Om :o owmmmma 2: :0: 3:2w <<:o 3:22 5:252:
2: 20:2 U 35:: (Jim: .:o 2:22: 2: 8.: 8:24: <32: :0 2:22 5:25:23 2: 3.82 m 3:9: (Jim: :0 owmmmma
2: 5.: 3:24: << :0 2:22 5:252: 2: 2222 < 35:: .<Om 5: 88: E26 32:8 :2 2m: 2: :5 «it—ma: :0: 22m: :53
:32; $32 2:: 2::0 2:22 :222w2 2: 32:2 3:: 25 .<Om .:o 2:222: 2: 8.: 22:: ::o>u w: :5: «9:45;? 2:22: 2:
8.: 22:: EDS mm 2: 22:. .2535: o :5: Wm 29:: :_ :u:_.:o: m: 2:: ::o>o 2: e: m::o:m2:oo .3: 2: :23, _ e: 2.13:3 :2:
:2: m_ 933:; 555:: < 22:53: 98%: 2:22: 3:: ﬂow—~52: w:_E:u-:o: ::: 22::5: 32m 5:3: :22: mLwacmE 4:553
3:26: 2:225: 95:: 530. DE: 20.“:ng wEEB ”5:20: ._u>o_ 232:: 2:28: 3:203: :uﬁEEdogESto: :0
$24: 2:28: 2:55: :2: :o :22: 2::2w :2 9:: 322: :2:::.: 2: 8:820: :3: 5:26: :wi .moom :o moo: :22: E
2.3: 43:32: mmm 5:3 25: :0 22:00 0:04.20: :22: .<Om 5.: :2: m: 8:: moom :5: <14?— 8: :2: m: 8:: ::::2 x82 36:
m2: 223:5: ::oEow::mE 4:553 222:: ::m .37.: :o::w:: :wE :0 2023222 :0 3322 WK: 2: 3:22: 28...: 2E:

3:20 2E 55:231— .:ui :5: 28:59:32 9523— E2229 .3 3.29m mAO

EN 935.

107

 

 

 

 

 

.3:- 35? a. :3. a? :2: 2:: 2:..- 93.:- t... .3:- .. mo;-
Emsuﬁ
.2 2 2 .m a. 2 z. 2 on a. z
.22: .3.2: .23: 8.2: 3mm: 39...: .22: .32: .22: 33...:
8...: Sn...- F-N... .2... 3...: .8... 3...... 2”... .. 8:.- 33. "2.22 2:
.22: :23: .21: : .2: .32: 3.6.2: .32: .22: 83...: .23:
NS... 2...:- . 2:.- E: 83. ....m.: m2: .8: .. 2:3- 3.:- EE? 2:
$2.: 35...: .22: c .2: 3.22.: .22: .32: .22: 83...: .22:
:1 m2." .. 2.... 2m... .. 2...: m3... 2.....- 22. 2...: 2a...- 2:... .523. :8
:22: .23: .23: 22...: 52...: .34.: 32...: 5.3: 35...: 3.2:
t... .5. _- 3.3- .... mt..- Sn..- .. 8...: t. 2...: _. £2:- 23- :6...- ..... 2.2- .2282 2:
.22: 35...: .23: .22: .22: ENE: .22: E32: .22: .34.:
2:.- mm...:- . 2...:- So...- 3...:- :4:- 4... m3...- : 3.2..- 23- .. :3...- :~_m§. .2:
. 83.: $2.: .22: 8.2: 82...: .22: .22: .32: .22: 33...:
t. 3....- 29? a2...- »22 m3... .2... 22:- m8... .. 2n..- .2...- 822: 2:
.34.: 35...: .23: .22: .22: .22: 32...: .22.: .22: .22:
.. 3...:- .....e. .32.- .. :2... 22.- .. :2...- .. .3:- 23. .. 2:.— 22: 222.: S:
.: a a b o m .. ..- N . 3.5 2:55

8:895.— .a:..8< 1.5.8574

 

495mm :: yuammsm 2.: .8.— mnzohc <<.:o 8.3:: m40 .< .05:—

108

 

 

 

: 8:2- 2...- ..88: .:..:- m::.: 8::- :2: i... :::.m- : 2:8- : 888-
.:Dmmf-LN

8 :8 ..m 8.. 2 3 8 8.. 8 8.. z
88:. 82:. 82:. 82:. 82.8.} - 8.2:. 82:. 83:. 83:. .. 8:.
3.8.: :2.:- 3.: .:..:- 33. r. 82:- :2: 2...? 2...: .2: 222:. NS
88:. 22:. 88:. 82:. 23:. 82:. 82:. 83:. 83:. 88:.

z ...m..- :2:- . :8: :88: :. S. 8.: :8:.:- 28:- . 88:- 83:- 822:. .8
88:. 22:. 82:. 82:. .8...:. 82:. 82:. 83:. 83:. 88:.

..z. ::.... 3:. .88.. ..2: :2: 8:: :2: ..2: 2...: 88.: 28.: 322:. :8
83:. 82:. 82:. 82:. 2.3:. 82:. 82:. 83:. 82:. 88:.

.1. .3..- 88:- 2:: 8::- :2.: :2: 8...:- i :2... 8.8.:- .. 88:- 8.8.2:. :5
88:. 82:. 88:. 82:. 83:. 82:. 82:. 83:. 83:. 2 8:.
82:- :8..:- ::8.:- 3.:- 2:.:- 88:- :8:- 8:m:. : :2. s: 2....- .8.2:. 8.2
88:. 82:. 82:. 82:. 83:. 82:. 82:. 82:. 83:. .. 8:.
2.:- :....:- 8.8:- 2.: ::.:- .2. :::.: .. 88.: 2:.:- 2::- t; 88..- ::~.2:. 8...
88:. 82:. 82:. 82:. 83:. 82:. 82:. 83:. 23:. 2 8:.

.. 8:8..- .. 28:. :2: 8:.:- :8: 2...? :8: 83:- .. 82:- ..::.:- 222:. :5
:. : 8 8 8 8 .. :- 8 . 3.... 8:5...

 

83.8.3.— _a=..:u< FEB—=7. cop—Baazéunaﬁuotom

 

(Jul—Wan MO OWWmmwa— 0.: h¢.— mn—SQLU <<Ea~ kc 3:50"— MJMV um _9=Nn—

109

 

 

 

..:.:.. 53.".- _.:.:.. home. :1. cmmﬁu vend- _.:.:.. Chmﬁ- :1. Nmo.m- ... mead- 1.... Omc.m- ..2; 000... 3..., mwméu
.50 ENN
a.

ex 3.. m: R_ o: E _: m: E ox z
:38 33.8 32.8 ASN.8 82.8 :28 32.8 62.8 :28 £38

Sm... :. .23- ~86- £3 32. 32.3- 2g N. 3- 2.. 3:5 23- 2382. in
:38 3 _ ...8 32.8 83.8 82.8 $328 83.8 32.8 25.8 83.8

3.2- 3... £2- : «5...- t; 3:... r; .52.- 8...... to... .5. 32.. t. NF..- 1: 9:... £888 :a
3.3.8 82.8 82.8 $2.8 $2.8 $2.8 $2.8 3.28 8,28 338

t... 2:- .1; ~22- .2: £3- mono 3.; £2- 3: $2- 3: 83. 3: So..- i... .21- a: 2.3. 3.888 82
8.5.8 5.2.8 A22: :38 32.8 83.8 3.2.8 2.2.8 3.38 23.8

.. 23.. t. cmod :. SS 53. _. Eco 3:. $3 2.1.- i a: 3. wood 0%.? 8888 Do
$3.8 $3.8 $2.8 $3.8 83.8 83.8 82.8 32.8 :58 $3.8

.. Z... .2: 0%.. r; 3: 1:. am»... .2; m8... 1.. 8: 1:. «3.. r... 2:.— ..1. SN.— SZ. £833 In
€908 3:..8 32.8 63.8 33.8 .2328 83.8 32.8 :58 83.8

«.8... $2.- 3. F2- 32. :3- £3. r. 2:. $2.- $2. $2.. .2883 2a

8 a a b e m w n N _ 3.5 35.5

 

8:8.qu— _a=..3< .aE..c.:—<

(om hO omrammak 0.: Law.— mﬂﬁOhU <<hO ma—Smom Mn—Q “U —028&

HO

63: En Emu. 3:35.58- ..A_u>_..uuamu.. .232 Q... asm xxx: 2: a 8:3..._cw_m 2355 I... ..1 ....

 

 

 

 

8:82.:— _a=..3< 35.85:. avian—2-358329“—

...I 33. t: 32- :- 25? 3:. .:..N- 1.. $3- _.. 8.3- t... @2- .: ES. 3: mm:- 3; 3.3.-
“Ox—ELM

m8 3. 8. am. 3 c: N: 8... mm. S 2
$28 $.38 $2.8 $2.8 $3.8 $28 :38 $3.8 $28 $38
:5... NE... SN..- 02.... 22.. $3- .. ...S- Nood- 85.8 . ~88 283% $5
$3.8 $3.8 $5.8 $2.8 $3.8 .2 n8 . .38 $3.8 $3.8 $5.8
8:.- i 2:..- r. 85.... 2. 3.... r... ..S..- at...- .:. $.3- até- .. 8.....- 3.....- £58.: to
$3.8 $3.8 $.38 $3.8 $28 :28 :88 $3.8 $28 .398

3:. MEN- .1 25. t. .33- i... $.1- 2: :5..- .38 i 82. ..- 32- : $3..- .:.- 83- 338...“. .:a
$2.8 $3.8 $2.8 $2.8 $3.8 $2.8 $3.8 $28 :58 .338
2 m... 2. $2 3.3- : No.3 83 i. .3... .:N... - 33 . 2... NR..- 8383 22
$3.8 $3.8 $2.8 . _ 2.8 $3.8 $3.8 $3.8 $3.8 $3.8 $5.8
$3. .2. .3... r. mam. r; :3... SN... 2.. m3... .2. an... r. we... .. .:5 .. mm: 2588 E:
$3.8 $2.8 $2.8 $2.8 $3.8 $2.8 :38 $3.8 $3.8 $5.8

1:. 2:..- 2......- .. 22.. SN... 5.....- .2. 3.....- 3m... 23...- .2... 3a...- 2223 an
.: a a h e m a m N _ 3.5 3:55

 

<8 .9 amalgam 2.. .2 2:85 <55. 3 2.3:. Bu 5 .23..

Ill

$0355.00 E 002000.
0.0 :.w. 0.0..0 0.00:0.m :< .<Om .0 3.02.00 0:. .8 00:0... 0 :0. .0 0.050. :2.0:...m0 3.000. m 3:00 .68ng .0 0.00200
0:. .8 09.0% D :0. .0 0.0.00. :2...E_.m0 8.000. < 00:0: .<Om .0. 00.00 .:0>0 .02.... .20 .02 0:0 «Jim. .8 00.00 .:0>0 0.22.0
:0>0m $0. .0 2:00. :2mm0.w0. 0:. 305 5:0 03 .<Om .0 09.0000 0... .8 0.0.00 .:0>0 .2 0:0 xxx-5n. .0 0.000000 0:. .8 3.00
.:0>0 mm 0.0 0.0:.- .0m_.$.0:.0 0 0:0 m.N 0.00% 2 002.00 00 0.00 .:0>0 0:. 0. 00:83:00 >00 0:. :00? _ 0. 20:00 .0... 022.0.)
...EE..0 < $30.. .00_0:0.0:m .0200...» 90:30.00. 0 5sz 0:0 30.... .00_0:0.0..m .0w:0..m w:_.:0m0.00. O .032 5.3 2:...
.32 Saw: m.00.0:0.0:m 20:. :0 002... 00:0... :0. 0.2 0003.0 .08.... 0.0 828.50 22. 5:03.: :3: .moom .0 moo. .0020
2 0.200 .0202. Nmm .005 0E... .0 0.20:0. 28:00 :05 .<Om .8 000.. m. 0.00 Noom 0:0 «Em-.mn. .8 00m: 2 0.00 53.0. ..00.m
>000 moo. 20:8..00 me: .00—000.000 .:0.0.._0 0:0 22. :2.0w_.: :3: .0 0.020520% .0 0.0.0.0. mam. 0:. 02000.0 030. 2.;

00.5.0 221 5:02.: .35 0:0 E05035: win-Em 2.0.8.5 .0 0.300% mx—O

”N 030,—.

 

 

 

:2 mm... .21..- 23. mm... .22.- ONE. :5... 82. $2.-

.HDNNSLN
2 2 .N om 3 mm mm x. 2 cm 2
5...... $3.8 $3.8 $3.8 :28 5.8 $3.8 82.8 :38 33.8
5m... 8m... 8.....- can... 32.. .2. as... 3.....- :2. 8...... 2.... «2.22 NS
2.2.... $3.8 $8.8 2.2.8 9.3.8 3.38 338 3:8 32.8 :88
$2. 8.2. Mam... 1&2. .18.”... N8... :25... $2.- :33... 2.2. :23... .8
$3.8 $3.8 83.8 $8.8 $3.8 $3.8 63.8 83.8 32.8 :38
3.....- 82.- :3 3... mg... .82. #2. n3... 2.....- 3.2.- 9.....- 32mg. :8
:38 28.8 :38 3.3.8 2.3.8 :38 3.3.8 :mw8 $88 33.8
2:. 32.- 23.- N2... 8.....- 22.- oz... Sm... 3.2.- 22.- 2238 2a
E18 33.8 3.3.8 :28 :38 53.. 39.8 32.8 :28 $3.8
so... 9...... ...t..%... 11.2.8 83.- $3.- ﬁn..- 2.2.- 8.....- 2.3.- :23... 82
:38 $3.8 3.8.8 5.2.. .32.. 3.9.8 $9.8 88.8 :38 $38
3m... ~82. 5.... 2...... 32.. r. 3.....- :2... $2.- SN... 2.....- .233. to
5...... 33.8 $8.8 $2.8 $3.8 3:..8 39.8 83.8 :38 338
Re... 32. 22.- 2.....- 3.2. r. 3...... 2.....- 22.. :5... 23.. 38mg. 2:
8 a a .. c m a n N . 33 3.55

 

8:85.. u

 

55?. 2 “mad...— 2.. S. 2:55 u... 3.3:. 30 ”< .23..

. ..m: 2.. 3m... 0 3.-....» .>. 3.833. ,m u>u a a .o u 2.. 8:3. Em;- 33_.:_ . .
. . .. .. \.. -. . . . 3.2.. .I. ...

 

 

 

2......- _....... .1... .3..- . 2....- ....2.- .1... woo...- :; 5....- ......... 8.....- :. 2......- ...:. 9.....-
..QEELN
: 2 E ..m a. z. 3 m. s. 8. .z
.32.. 5.2.. .82.. .22.. ........8 5...... .3.2: .08.... 5%.... 6%....
mm... .25...- 22.- ......... .. 3.... $2.- _. .2... o. 2.. 2.....- .K2.- 3.888 x...
.32.. .32.. .898 .22.. .32.. .53.... .33.... 23...... 53.8 38.8
2......- 32. 3. NS..- ,. 22.- $2.-. .2. 2.3- .. ..~...- .2; gm..- nmo..- t; R...”- mﬁ..~....~ 2..
5.2.. 5.2.. .w . 2.. .82.. 6.2.. 25...... .82.. .. .2.. 2.8.8 .32..
9.....- EN...- owoé .32.- .........- i 8....- .. .3...- t: .3...- .. 2...- .1... 2...- 388... o...
.22.. .32.. 5...... 8.2.. 3...... .33.. 5...... .82.. 3.2.... .82..
.1. 9.2 ..N2. ...2. 3...... .. S... .. .9... 1. 8...... 1... 2... «m2. .1 8.2 28...... m...
.32.. .32.. .32.. .22.. .82.. 8.3.8 .22.. .82.. 82...... .32..
8...- =3... 2.... 9.2. .. 3.»... r. S... .. 8...... t. 8.... m2... :5 mo...— 25...... 2..
3...... 8mm... .82.. .22.. .82.. .23.... .3328 .23.... .82.. 38.8
$2. .. .3...- .. 8.....- mﬁ... .. 8...- 2...... m3... :2. 5.... NE... 288... 2..
8 a a .- o m .. n N . 3.... .52....

L

8:825— U

 

<Om ._.. owummam o... .:..— m........0 O .2. 3.581 mam. "m .95...

ll4

REFERENCES

Reference

Alexander, Janet, 1993, The lawsuit avoidance theory of why initial public offerings are
underpriced, UCLA Law Review 4|, l7-73.

Ali, Ashiq, and Sanjay Kallapur, 2001, Securities price consequences of the Private
Securities Litigation Reform Act of I995 and related events, The Accounting
Review 76, 431-460.

Allen, Franklin, and Gerald R. Faulhaber, I989, Signaling by underpricing in the IPO
market, Journal of Financial Economics 23, 303-324.

Avery, John W., I996, Securities litigation reform: The long and winding road to the
Private Securities Litigation Reform Act of 1995, Business Lawyer 51, 335-378.

Bajaj, Mukesh, Sumon C. Mazumdar, and Atulya Sarin, 2003, Securities class action
settlements: An empirical analysis, Santa Clara Law Review 43, 100l-1033.

Beatty, Randolph P., Philip D. Drake, and Chris E. Hogan, 2002, The impact of the 1995
Private Securities Litigation Reform Act on litigation risk and auditor
compensation in the IPO market, Working paper, University of Southern
California.

Beatty, Randolph P., and Jay Ritter, 1986, Investment banking, reputation, and the
underpricing of initial public offerings, Journal of Financial Economics IS, 213-
232.

Beatty, Randolph P., and Ivo Welch, I996, Issuer expenses and legal liability in initial
public offerings, Journal of Law and Economics 39, 545-602.

Benveniste, Lawrence, and Paul Spindt, 1989, How investment bankers determine the
offer price and allocation of new issues, Journal of Financial Economics 24, 343-
362.

Brickey, Kathleen, 2003, From Enron to Worldcom and beyond: Life and crime after
Sarbanes-Oxley, Washington University Law Quarterly 81, 357-401.

Brown, Stephen J., and Jerold B. Warner, 1980, Measuring security price performance,
Journal of Financial Economics 8, 205-258.

Brown, Stephen 1., and Jerold B. Warner, 1985, Using daily stock returns: The case of
event studies, Journal of Financial Economics 14, 3-3 l.

116

Busaba, Walid Y. and Chun Chang, 2002, Bookbuilding vs. ﬁxed price revisited: The
effect of aftermarket trading, Working paper, University of Arizona.

Carter, Richard 8., Frederick H. Dark. and Ajai K. Singh, 1998, Underwriter reputation.

initial returns, and the long-run underperformance of IPO stocks, Journal of
Finance 53, 285-311.

Carter, Richard B., and Steven Manaster, 1990, Initial public offerings and underwriter
reputation, Journal of Finance 45, 1045-1067.

Collins, Daniel W. and Warren T. Dent, 1984, A comparison of alternative testing

methodologies used in capital market research, Journal of Accounting Research
22, 48-84.

Cunningham, Lawrence A., 2003, The Sarbanes-Oxley yawn: Heavy rhetoric, light
reform (and it might just work), University of Connecticut Law Review 36, 915-
998.

Dechow, Patricia M., Richard G. Sloan and Amy P. Sweeney, 1995, Detecting earnings
management, The Accounting Review 70, 193-225.

Drake, Philip D., and Michael R. Vetsuypens, 1993, [PO underpricing and insurance
against legal liability, Financial Management 22, 64-73.

Dunbar, Frederick, Vinita Juneja, and Denise Martin, 1995, Shareholder litigation: theory
and evidence on deterrent value and merits, Working paper, National Economic
Research Associates.

Ellis, Katrina, Roni Michaely, and Maureen O’Hara, 2000, When the underwriter is the

market maker: An examination of trading in the lPO aﬁermarket, Journal of
Finance 55, 1039-1074.

Eth, Jordan and Daniel S. Drosman, 2001, The Private Securities Litigation Reform Act:
Five years young, Review of Securities and Commodities Regulation 34, 153-164.

Foster, Todd 8., 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, Journal of Accounting research 32, 137-164.

Geewax, Marilyn, 2002, Accounting reform faces key vote in Senate panel, Cox News
Service, May 21.

117

Gompers, Paul, Joy Ishii, and Andrew Metrick, 2003. Corporate governance and equity
prices, The Quarterly Journal of Economics 118, 107-155.

Greenberger, Robert S., 2002, Questioning the books: panel, in Enron‘s wake, to review
lawsuit curbs. Wall Street journal, Feb. 6, at A8.

Grinblatt, Mark, and Chuan-Yang Hwang, 1989, Signaling and the pricing of new issues,
Journal of Finance 44, 393-420.

Grundfest, Joseph A., and Michael A. Perino, 1997, Securities litigation reform: The ﬁrst
year's experience (a statistical and legal analysis of class action securities fraud
litigation under the Private Securities Litigation Reform Act of 1995), Working
paper No. 140, John M. Olin program in law and economics working paper series,
Stanford Law School.

Hanley, Kumar, 1993, The underpricing of initial public offerings and the partial
adjustment phenomenon, Journal of Financial Economics 34, 231-250.

Hao, Qing, 2004, Litigation risk and. IPO underpricing revisited, Working paper,
University of Florida.

Heckman, James J., 1978, Dummy endogenous variables in a simultaneous equation
system, Econometrica 46, 931-959.

Hensler, Douglas A., 1995, Litigation costs and the underpricing of initial public
offerings, Managerial and Decision Economics 16, 1 11-128.

lHughes, Patricia J., and Anjan V. Thakor, 1992, Litigation risk, intennediation, and the
underpricing of initial public offerings, Review of Financial Studies 5, 709-742.

lbbotson, Roger (3., 1975, Price performance of common stock new issues, Journal of
Financial Economics 2, 235-272.

lbbotson, Roger G., Jody L. Sindelar, and Jay R. Ritter, 1988, Initial public offerings,
Journal of Applied Corporate Finance 1, 37-45.

lbbotson, Roger G., Jody L. Sindelar, and Jay R. Ritter, 1994. The market’s problems

with the pricing of initial public offerings, Journal of Applied Corporate Finance
7, 66-74.

Johnson, Marilyn F., Ron Kasznik, and Karen K. Nelson, 2000, Shareholder wealth

effects of the Private Securities Litigation Reform Act of 1995, Review of
Accounting Studies 5, 217-233.

118

Johnson, Marilyn F., Ron Kasznik, and Karen K. Nelson, 2001, The impact of Securities
Litigation Reform on the disclosure of forward-looking information by high
technology ﬁrms. Journal of Accounting Research 39, 297-327.

Johnson, Marilyn F., Karen K. Nelson, and Adam 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, Southern California
Law Review 73, 773-810.

Johnson, Marilyn F., Karen K. Nelson, and Adam C. Pritchard, 2002, Do the merits
matter more? Class actions under the Private Securities Litigation Reform Act,
Working paper, Michigan State University.

Kaplan, Barry, and Douglas Greene, 2003, Recent trends in private securities litigation
Sarbanes-Oxley Act of 2002: The counter-reform act, Glasser Legalworks.

Karpoff, Jonathan M. and Paul H. Malatesta, 1989, The wealth effects of second-
generation state takeover legislation, Journal of Financial Economics 25, 291-
322.

Keloharju, Matti, 1993, The winner's curse, legal liability, and the long-run price
performance of initial public offerings in Finland, Journal of Financial
Economics 34, 251-277.

Koh, Francis, and Terry Walter, 1989, A direct test of Rock’s model of the pricing of
unseasoned issues, Journal of Financial Economics 23, 251-272.

Kothari, S.P., Andrew J. Leone, and Charles E. Wasley, 2005, Performance matched
discretionary accrual measures, Journal of Accounting & Economics 39, 163-197.

Labaton, Stephen, 2002, Now who, exactly, got us into this? New York Times, Feb. 3, at
Sect. 3, p.1.

Leland, Hayne E., and David H. Pyle, 1977, Information asymmetries, ﬁnancial structure
and ﬁnancial intermediation, Journal of Finance 32, 371-387.

Lerach, William S., 1998, “The Private Securities Litigation Reform Act of 1995 - 27
months later”: Securities class action litigation under the Private Securities

Litigation Reform Act’s brave new world, Washington University Law Quarterly
76, 597-644.

Li, Haidan, Morton Pincus, and Sonja Rego, 2004, Market reaction to events surrounding
the Sarbanes-Oxley Act of 2002, Working paper, University of Iowa.

Loomis, Tamara, 2002, Enron and reform; Congress is looking at securities litigation law,
New York Law Journal, March 28.

119

Loughran, Tim, and Jay R. Ritter, 2002, Why don‘t issuers get upset about leaving
money on the table in lPOs? Review of Financial Studies 15, 413-443.

Loughran, Tim, and Jay R. Ritter, 2004, Why has IPO underpricing changed over time?
Financial Management, forthcoming.

Lowry, Michelle, and G. William Schwert, 2004, Is the IPO pricing process efﬁcient?
Journal of Financial Economics 71, 3-26.

Lowry, Michelle, and Susan Shu, 2002, Litigation risk and IPO underpricing, Journal of
Financial Economics 65, 309-335.

Maddala, G. S., 1983, Limited-dependent and qualitative variables in econometrics,
Cambridge University Press, Cambridge, England.

Megginson, William C., and Kathleen A. Weiss, 1991, Venture capitalist certiﬁcation in
initial public offerings, Journal of Finance 46, 879-904.

Michaely, Roni, and Wayne H. Shaw, 1994, The pricing of initial public offerings: Tests
of adverse-selection and signaling theories, Review of Financial Studies 7, 279-
319.

Miller, Robert, and Frank Reilly, 1987, An examination of mispricing, returns, and
uncertainty for initial public offering, Financial Management (summer), 33-38.

O‘Connor, Colleen M., 2002, Class-action lawsuits target IPO “laddering”, Investor
Relations Business, New York, July 15', page 1.

O’Connor, Colleen M., 2002, The big rage in investor lawsuits: some 64% of all
securities cases ﬁled in 01 involved laddering, The Investor Dealers’ Digest:
IDD, New York, July 17th, page 1.

Oppel, Richard A., 2002, Negotiators agree on broad changes in business laws, New York
Times, July 25: AI.

Perino, Michael A., 2003, Did the Private Securities Litigation Reform Act work?,
University of Illinois Law Review, 913-978.

Pritchard, Adam, 1999, Markets as monitors: A proposal to replace class actions with
exchanges as securities fraud enforcers, Securities Law Review 33, 255-350.

(Originally published under the same title in Virginia Law Review 85, 925-1020.)

Pritchard, Adam, 2003, Should Congress repeal Securities Action Reform? Cato Policy
Analysis 471, 1-17. .

120

Ouillen, Gwyn, Scott Vick, and Mayra de Aguiar, 2000, Unintended consequences of the
Private Securities Litigation Reform Act of 1995, Working paper, Alschuler
Grossman Stein & Kahan LLP.

Ratner, David L., 1998, Securities Regulation in a nutshell, West Group, St. Paul, Minn.

Reese. William A., 1998, IPO underpricing, trading volume, and investor interest,
Working paper, Tulane University.

Rezaee, Zabihollah, and Pankaj Jain, 2003, The Sarbanes-Oxley Act of 2002 and security
market behavior: early evidence, Working paper, University of Memphis.

Ribstein, Larry, 2002, Market vs. regulatory responses to corporate fraud: a critique of
the Sarbanes-Oxley Act of 2002, Journal of Corporation law, Fall, 28, 1-67.

Ritter, Jay R., 1984, The hot issue market of 1980, Journal of Business 57, 215-240.

Ritter, Jay R., 2003, Investment banking and securities issuance, in George. M.
Constantinides, Milton Harris, and Rene’ Stulz, eds.: Handbook of the Economics
of Finance (North Holland), chapter 5.

Rock. Kevin, 1986, Why new issues are underpriced, Journal of Financial Economics 15,
187-212.

Ruud, Judith S., 1993, Underwriter price support and the IPO underpricing puzzle,
Journal of Financial Economics 34, 135-151.

Ryngaert, Michael and Jeffrey M. Netter, I988, Shareholder wealth effects of the Ohio
anti-takeover law, Journal of law, Economics, and Organization 4, 373-383.

Schipper, Katherine and Rex Thompson, 1983, The impact of merger-related regulations
on the shareholders of acquiring ﬁrms, Journal of Accounting Research 21, 184-
221.

Schultz, Paul H., and Mir Zaman, 1994, Aﬁermarket support and underpricing of initial
public offerings, Journal of Financial Economics 35, 199-219.

Senator Report No. 104-98 at 38-39 (1995), reprinted in 1995 U.S.C.C.A.N. 679, 716-
717 (Additional views of Senators Sarbanes, Bryan, and Boxer).

Smart, Scott, and Chad J. Zutter, 2003, Control as a motivation for underpricing: A

comparison of dual- and single-class lPOs, Journal of Financial Economics 69,
85-110.

121

Spiess, D. Katherine, and Richard H. Pettway, 1997, The IPO and ﬁrst seasoned equity

sale: Issue proceeds, owner/managers’ wealth, and the underpricing signal,
Journal of Banking and Finance 21, 967-988.

Spiess, D. Katherine, and Paula A. Tkac, I997, The Private Securities Litigation Reform
Act of 1995: The stock market casts its vote, Managerial and Decision
Economics 18, 545-561.

Steckman et al. v. Hart Brewing Inc. et al., Civil Case No. 96—1077-K (RBB), Order
granting defendants’ motion to dismiss, United States District Court, Southern
District of California, December 24, 1996.

Stewart, C. Evan, 2002, Caveat ‘reformers’: lessons not to be Ieamed from Enron’s
collapse, Securities Regulation and law report (BNA) 34, 310-314.

Taylor, Jeffrey, 1995, GOP-controlled House likely to limit investor suits against public
companies, The Wall Street Journal, January 20: C 1.

Tinic, Seha M., 1988, Anatomy of initial public offerings of common stock, Journal of
Finance 43, 789-822.

Wooldridge, Jeffrey M., 2001, Econometric analysis of cross section and panel data, The
MIT Press, Cambridge, Massachusetts, London, England.

Zishka et al. v. American Pad & Paper Company et al., Civil Case No. 98-660-D.

Complaint for violation of the Securities Exchange Act of 1934, United States
District Court, Northern District of Texas, March 10, 1998.

122