Financial fraud, director reputation, and shareholder wealth

ARTICLE IN PRESS

Journal of Financial Economics 86 (2007) 306–336

0304-405X/$

doi:10.1016/j

$We than

Columbia, U

Vegas, Temp

University, t

Lemmon, Cl

helpful comm�CorrespoE-mail ad

www.elsevier.com/locate/jfec

Financial fraud, director reputation,and shareholder wealth$

Eliezer M. Ficha, Anil Shivdasanib,�

aLeBow College of Business, Drexel University, Philadelphia, PA 19104, USAbKenan-Flagler Business School, University of North Carolina, Chapel Hill, NC 27599, USA

Received 21 April 2005; received in revised form 18 April 2006; accepted 16 May 2006

Available online 28 June 2007

Abstract

We investigate the reputational impact of financial fraud for outside directors based on a sample of

firms facing shareholder class action lawsuits. Following a financial fraud lawsuit, outside directors

do not face abnormal turnover on the board of the sued firm but experience a significant decline in

other board seats held. This decline in other directorships is greater for more severe allegations of

fraud and when the outside director bears greater responsibility for monitoring fraud. Interlocked

firms that share directors with the sued firm also exhibit valuation declines at the lawsuit filing.

Fraud-affiliated directors are more likely to lose directorships at firms with stronger corporate

governance and their departure is associated with valuation increases for these firms.

r 2007 Elsevier B.V. All rights reserved.

JEL classification: G30; G34; J33: K22; M41

Keywords: Director reputation; Financial fraud; Interlocking directorships; Class action lawsuits

- see front matter r 2007 Elsevier B.V. All rights reserved.

.jfineco.2006.05.012

k seminar participants at London Business School, Vanderbilt University, University of British

niversity of Calgary, University of North Carolina, McGill University, University of Nevada-Las

le University, University of Tennessee, the 2005 Corporate Governance Conference at Washington

he 2006 American Finance Association, and Jeff Coles, Jonathan Karpoff (a referee), Michael

audine Mangen, John McConnell, Bill Schwert, David Yermack, and an anonymous referee for

ents. The authors acknowledge support from the Wachovia Center in Corporate Finance.

nding author. Fax: +1919 962 2068.

dress: [email protected] (A. Shivdasani).

www.elsevier.com/locate/jfec

dx.doi.org/10.1016/j.jfineco.2006.05.012

mailto:[email protected]

ARTICLE IN PRESSE.M. Fich, A. Shivdasani / Journal of Financial Economics 86 (2007) 306–336 307

1. Introduction

The recent wave of corporate financial scandals has raised substantial concerns aboutthe effectiveness of corporate governance in the US. A commonly held view is thatfinancial scandals are symptomatic of, and result from, massive deficiencies in corporategovernance. This view has prompted widespread changes, including the Sarbanes-OxleyAct, new Securities and Exchange Commission (SEC) regulations, and governancerequirements adopted by the NYSE and NASDAQ. Additionally, organizations such asInstitutional Shareholder Services, the Council of Institutional Investors, and TheCorporate Library, among many others, have been actively promoting extensive agendasaimed at reforming the quality of corporate governance in public corporations.

There is broad agreement that financial fraud leads to significant valuation losses forinvestors, as has been apparent in numerous recent governance failures. These lossesappear to result primarily from reputational costs borne by firms as a result of the financialfraud. However, relatively little is known about the reputational costs for outside directorsof firms that are involved in fraud and the penalties suffered by these directors. In thispaper, we examine the role of reputation in the market for directorships as an incentivemechanism for monitoring fraudulent behavior. Until the recent settlements in whichoutside directors at Enron and WorldCom personally paid over $10 million, outsidedirectors appear to face few consequences for preventing financial misconduct. In thispaper, we study whether outside directors suffer reputational penalties if the firms theyserve on were accused of financial fraud.

We study a sample of firms facing shareholder class action lawsuits alleging financialmisrepresentation under rule 10(b)-5 of the 1934 Securities and Exchange CommissionAct. We find no evidence of abnormal turnover of outside directors on the boards of suedfirms following such lawsuits. However, there is a dramatic decline in the otherdirectorships held by these outside directors. On average, outside directors of sued firmsexperience a reduction of about 50% in the number of other directorships held, and 96%of outside directors who sit on another board lose at least one directorship within threeyears following the lawsuit. The reductions in directorships are greater for more seriouscases of alleged fraud, as measured by subsequent SEC enforcement actions or settlementamounts following the lawsuit.

We study three hypotheses to understand the reduction in directorships following classaction lawsuits. The reputation hypothesis holds that outside directors bear personal costsin the form of fewer board seats if their firms are involved in financial misconduct. Analternative (but not mutually exclusive) hypothesis is that the structure and composition ofboards is determined endogenously to reflect the monitoring and operating environment.According to this endogenous board hypothesis, a lawsuit can signal that a firm is in afraud-prone environment, leading its outside directors to spend more time monitoring thisfirm and to cut back on their other board seats. Finally, we consider the legal liabilityhypothesis which posits that outside directors cut back on board seats following a lawsuitin an attempt to minimize their future legal exposure.

We conduct several tests to explore these hypotheses. We examine how news of a lawsuitaffects other firms that are linked to the outside directors of sued firms (i.e., director-interlocked firms). We find that interlocked firms experience significant negative abnormalreturns at the time of the lawsuit filing, a finding consistent with all three of the hypothesesthat we study. We uncover support for the endogenous board hypothesis when we examine

ARTICLE IN PRESSE.M. Fich, A. Shivdasani / Journal of Financial Economics 86 (2007) 306–336308

the probability that a firm faces a financial fraud lawsuit. In a simultaneous equationsframework, we find that firms most prone to a fraud lawsuit are also more likely to have anoutside director who has been previously sued for fraud. This result points to theendogeneity of corporate board composition with respect to fraud likelihood. However, wealso find that the presence of sued directors increases the probability of a fraud lawsuit,a finding consistent with the reputation hypothesis which holds that such directors haveweaker reputations for monitoring.To understand why directorships are lost by sued directors, we study patterns of

departures of sued directors from boards of interlocked firms. Sued outside directors aremore likely to lose directorships at interlocked firms that face a higher probability of beingsued for financial fraud. This suggests that the decline in directorships is partially driven bythe desire of outside directors to reduce their future legal exposure. Sued outside directorsare also more likely to depart firms with strong corporate governance, as measured by theGompers, Ishii, and Metrick (2003) score. In an event-study analysis, we find thatannouncements of departures of fraud-associated outside directors from boards ofdirector-interlocked firms are viewed positively by investors.Overall, our evidence is most supportive of the reputation hypothesis, but we also

uncover evidence in support of the endogeneity and legal liability hypotheses. Irrespectiveof which hypothesis best explains the decline in outside directorships, our results illustratethat outside directors of firms accused of fraud bear a financial penalty. To the extent thatdirectors optimize the cost-benefit tradeoff of serving on a corporate board prior to thelawsuit, the reduction in directorships indicates that the welfare of these outside directors isreduced following the lawsuit. We estimate the direct financial value of a lost directorshipto be approximately $1 million.Our findings are relevant to the current debate on the role of directors in financial fraud.

For example, Section 305 of the Sarbanes-Oxley Act grants the SEC with powers to bandirectors from sitting on corporate boards if they violate the fraud rules of Section 10(b) ofthe 1934 SEC Act. Our results suggest that the benefits of such regulation might be morelimited than expected since outside director turnover on other boards displays a highsensitivity to the reputational capital of directors.The paper proceeds as follows. Section 2 reviews the relevant literature and presents our

research questions. Section 3 describes our data, explains the sample selection process, andpresents event-study results for firms facing financial fraud lawsuits. Section 4 examines theeffect of financial fraud on director reputation.1 Section 5 studies the effect of the class actionlawsuit on other firms interlocked through shared directors. Section 6 investigates theprobability of retaining directorships following the lawsuit and examines investor reactionsto departures of outside directors who sit on boards accused of fraud. Section 7 concludes.

2. Financial fraud and director reputation

2.1. Valuation consequences of financial fraud

Financial fraud typically has a substantial adverse valuation effect on companies. Tointerpret this effect, it is helpful to understand the typical sequence of information released

1A 10(b)-5 class action filing is an allegation of financial misconduct that does not automatically imply guilt.

For expositional convenience, we occasionally use the term ‘‘fraud’’ to designate those firms facing a lawsuit.


to the market. Karpoff, Lee, and Martin (2005) provide a comprehensive review of theenforcement process by the SEC and Department of Justice (DOJ). They note that indica-tions of fraud usually surface with the release of a trigger event, commonly a self-initiatedpress release by a firm alerting investors to the possibility of financial irregularities. Triggerevents are frequently followed by private litigation, such as shareholder class actionlawsuits that are filed quickly thereafter. Trigger events could also be accompanied by aninformal SEC inquiry, which determines whether further SEC action in the form of aformal investigation is warranted. Formal actions by the SEC sometimes result in anAccounting and Auditing Enforcement Release (AAER).

Karpoff, Lee, and Martin (2005) find significant negative valuation effects at all stages ofthe disclosure and enforcement process. They show that announcements of trigger eventslead to an average one-day abnormal return of �25.24%, class action lawsuits lead toabnormal returns of �7%, and company announcements of investigation events areassociated with abnormal returns averaging �14.4%. Cumulatively, their estimatesindicate an average valuation effect of �41% for accounting and enforcement actionsrelated to financial fraud. They show that direct penalties account for only a small portionof the value loss and that the bulk of the decline in value results from damaged reputationsof the firms. Reputational losses can arise because of managers devoting substantial timeto the investigation process at the expense of company business, reduced credibility incontracting with suppliers of capital and customers, and a higher cost of capital (Klein andLeffler, 1981; Jarrell and Peltzman, 1985; Karpoff and Lott, 1993). Thus, the evidenceshows that reputational effects such as impaired operations or higher capital and financingcosts impose firm-level costs.

2.2. Reputational incentives and director reputation

According to the reputation hypothesis (Fama, 1980; Fama and Jensen, 1983), vigilantdirectors establish reputations as good monitors and are rewarded with additional boardseats. Lax directors suffer a decline in reputation and bear a personal cost in the form offewer opportunities to serve on other boards. Additional incentives to monitor can arisefrom equity holdings, restricted stock and option awards, and turnover of outside directors(Yermack, 2004). The reputation hypothesis holds that outside directors of firms accusedof financial misconduct suffer personal losses in the form of damaged reputations.According to this hypothesis, the loss in reputation should be greater in more severe casesof fraud. In our empirical analysis, we measure how reputation is affected by examiningthe changes in the number of other directorships held.

Evidence on reputational penalties for directors in the context of financial fraud ismixed. Agrawal, Jaffe, and Karpoff (1999) find that turnover of CEOs and outsidedirectors is unchanged after fraud. Helland (2006) finds that outside directors of firmsfacing class action lawsuits actually experience an increase in the number of other boardseats held following the lawsuit. Srinivasan (2005) finds that outside directors, particularlythose serving on the audit committee, experience substantial turnover on the boards offirms that restate earnings. However, he finds only a small decline in other board seats heldby outside directors of firms that restate earnings.

Reputational effects can actually have the perverse impact of increasing the directorshipopportunities for sued outside directors. Given the struggle for power between the boardand the CEO described by Hermalin and Weisbach (1998), some CEOs might prefer


outside directors with a reputation for lax monitoring if their presence on the board makesit easier to perpetrate fraud, consume perquisites, or carry out other value-destroyingactivities. Alternatively, it is possible that outside directors who have served on the boardsof sued firms are desirable candidates for other boards because of their experience inresponding to such litigation. Helland (2006) suggests that the friendly director and legalexperience views underlie the increase in directorships of outside directors accused of fraudin his sample. Finally, liability or reputational concerns can also prompt outside directorsof sued firms to forgo directorships if they become averse to bearing the costs of fraud atanother firm on whose board they sit.Reputational effects for directors have been found to be important in other settings.

Outside directors hold fewer board seats after serving on boards of companies thatexperience financial distress (Gilson, 1990), that are liquidated (Harford, 2003), and thatperform poorly (Yermack, 2004). Active and retired CEOs of well-performing firms getadditional directorships (Brickley, Coles, and Linck, 1999; Fich, 2005), but thoseof dividend-reducing firms lose board seats (Kaplan and Reishus, 1990). Coles andHoi (2003) show that directors of companies rejecting antitakeover provisions ofPennsylvania Senate Bill 1310 are more likely to add directorships. However, most ofthese events are probably different from financial fraud, because boards might not beaware of the misconduct. Since management has the ability to control the flow of financialinformation to the board, fraud could be inherently difficult for outside directors touncover and might therefore have few, if any, reputational consequences for outsidedirectors.

2.3. Endogeneity of corporate board structure

An important consideration in studying reputational effects is the potential endogeneityof board structure. The optimal board composition varies across firms due to differences inindustry attributes. Some industries might be more vulnerable to fraud because of thecharacteristics of optimal customer, supplier, and employment contracts. As we describebelow, our sample of sued firms is unevenly concentrated across industries.According to the endogenous board hypothesis, firms facing a higher vulnerability to

fraud should employ directors with expertise and reputation for detecting fraud.Therefore, boards of sued firms (where ex-post, fraud probability is revealed to be high)are more likely to have outside directors with reputations for monitoring and frauddetection. If the filing of a lawsuit signals that a firm is more prone to fraud than previouslythought, the endogenous board hypothesis suggests that a lawsuit should trigger anendogenous adjustment in the number of board seats held by outside directors.Specifically, outside directors in firms accused of fraud could choose to cut back onother board seats in order to increase their monitoring of firms accused of fraud becausethese accused firms require more monitoring than previously thought. Thus, theendogenous board hypothesis offers predictions that are similar to those of the reputationhypothesis with regard to the number of other board seats held by outside directors.

2.4. Research questions and experimental design

We conduct four sets of tests. First, we examine the turnover of outside directors on theboards of firms accused of fraud and investigate the changes in their directorships held at


other firms. Both the reputation and endogenous board hypotheses predict a drop indirectorships held at other firms, as do legal liability considerations for outside directors.

Second, we examine how a fraud lawsuit affects the valuation of director-interlockedfirms. According to the reputation hypothesis, fraud lawsuits damage the reputation ofoutside directors, leading to a reduction in value of the other firms where they hold boardseats. If a lawsuit signals that other firms with similar attributes are more fraud-prone thanexpected, the endogeneity hypothesis suggests that other firms that are linked throughshared directors would also suffer a valuation decline if they share common attributes withthe sued firm. In an effort to distinguish between these two hypotheses, we consider theseverity of the fraud allegations in our tests. The reputation hypothesis predicts that moreegregious cases of fraud should result in greater penalties for outside directors.

Our third set of tests examines the reputation and endogeneity hypotheses byconsidering the probability of fraud lawsuits and the presence of sued directors on theboard. We employ a simultaneous equations framework in which both the presence of sueddirectors and the probability of fraud lawsuits are endogenously determined.

Finally, our fourth set of tests analyzes and compares the characteristics of firms inwhich directors accused of fraud lose their board seats relative to firms in which directorsretain their board seats. We test predictions of our hypotheses to understand thedeterminants of directorship losses following fraud lawsuits and we study the valuationimplications of these director departures from boards of interlocked firms.

3. Data

3.1. Shareholder class action lawsuits

We identify firms accused of financial fraud by the incidence of a shareholder classaction lawsuit alleging violation of rule 10(b)-5 of the SEC Act of 1934. This ruleproscribes, among other things, ‘‘the intent to deceive, manipulate, or defraud withmisstatements of material fact made in connection to financial condition, solvency andprofitability.’’ We exclude complaints that allege insider trading, which is an activity thatmay or may not be costly to shareholders.

Using class action lawsuits to identify financial misconduct offers some advantages overother approaches. Two other common approaches are announcement of an earningsrestatement and an Accounting and Auditing Enforcement Release (AAER). While manyof the sued firms eventually restate earnings, not all firms do so prior to the lawsuit filing.Further, as Karpoff, Lee, and Martin (2005) note, using press reports or an AAER canmiss many instances of reporting violations, and many targeted companies are delistedbefore regulators formally file charges. Class action lawsuits, on the other hand, tend to befiled quickly after the disclosure of reporting violations, on average one day after thetrigger event announcement. Further, Agrawal and Chadha (2005) note that SECenforcements are probably tilted towards the more egregious and high-profile cases offinancial fraud since limited resources prevent the SEC from pursuing all cases.

A drawback to using class action lawsuits to identify financial fraud is that the discoveryphase in lawsuits follows the filing of the suit, whereas it precedes enforcement actions suchas AAERs. Therefore, our lawsuit sample contains events where fraud is alleged, but notactually proven. Lawsuits usually precede federal enforcements, however, and aretherefore more suitable to study shareholder wealth effects, even though some lawsuits


might be eventually dismissed. For expositional convenience, we refer to lawsuit filings asfinancial fraud, but readers should note that our sample includes only cases where fraud isalleged to have occurred, resulting in a bias against uncovering a significant role forreputational incentives and associated valuation effects.In an attempt to address this limitation, we examine whether SEC enforcement actions

occur within our sample and also collect data on settlement amounts. We consider thepresence of an AAER or a high subsequent settlement as evidence of the severity of thefraud allegation and test whether changes in directorships are more pronounced for suchfirms. Karpoff, Lee, and Martin (2006) show that legal penalties tend to be driven by theextent of the harm caused by the misconduct, but that settlement amounts are linked to themagnitude of shareholder losses. Cox and Thomas (2004) show that settlement amountsare higher in more severe cases of alleged fraud.Prior evidence shows that class action lawsuits have material consequences for firms

named as defendants. Skinner (1997) finds that from 1988 to 1994, 20% of the casesinvolve settlements that exceed 10% of the firm’s annual sales. Ferris and Pritchard (2001)also argue that the Private Securities Litigation Reform Act of 1995 makes it more difficultto bring financial fraud class action suits by requiring plaintiffs to ‘‘stay’’ discovery whilemotions to dismiss are pending, thereby depriving plaintiffs of the sources most likelyto provide material facts needed to demonstrate fraud. They note that this Act hassignificantly lowered the incidence of frivolous shareholder lawsuits. Karpoff, Lee, andMartin (2006) show that this Act has not resulted in lower financial penalties, however.

3.2. Sample construction

Our sample period is from 1998 to 2002, and observations are drawn from thePricewaterhouseCoopers class action database and the Stanford University andCornerstone Research litigation database. The initial sample includes 685 litigation filingsfor 580 different companies during the five-year period. We use a two-step selection rule todetermine our sample. First, because offenses to rule 10(b)-5 also include insider trading,we read the complaints and focus only on those pertaining to financial misrepresentationswhere no insider trading is alleged. Second, we require firms to have stock market,accounting, and compensation data for the year of the lawsuit filing and the preceding yearon CRSP, Compustat, and ExecuComp, respectively. We also exclude a few cases wherethe filing date for the lawsuit cannot be identified reliably or where the firm is not theprimary defendant in the lawsuit. These criteria produce a final sample of 216 observationsinvolving 216 different firms.Table 1 displays summary statistics on the lawsuits in the sample. Panel A shows that

58% of the lawsuits allege use of improper accounting practices (including revenue,expense, and/or earnings management), 47% allege that management provided false ormisleading statements and/or failed to disclose material information, and 40% allegemisleading financial projections including actions to inflate the stock price. These numbersadd up to more than 100% since the categories are not mutually exclusive.The duration of the alleged violations (i.e., the class period) is, on average, over one year

(376 days) with a median of 260 days, as shown in Panel B. Panel B also shows thatlawsuits are typically filed about three months after the end of the class period. Nearly all(91%) of the lawsuits result in out-of-court settlements and such settlements take placeabout two and one-half years after the lawsuit is filed. The sample displays some clustering

ARTICLE IN PRESS

Table

1

Summary

offinancialfraudclass

actionlawsuits

Summary

ofclass

actionsuitfilingsforalleged

violationofrule

10(b)-5oftheSEC

Act

of1934.A

firm

isincluded

inthesample

ifitis

featuredin

either

the

PricewaterhouseCoopersLitigationdatabase

ortheStanford

Law

SchoolSecurities

Clearinghouse

database

foralleged

violationofrule

10b(5)involvingfinancial

misrepresentation.Tobeincluded

inthesample

afirm

must

haveatleast

$100millionin

assetsin

theyearprecedingthesuitanddata

available

inExecuComp,

CRSP,andCompustat.Panel

Adescribes

thenature

oftheallegationandPanel

Bprovides

inform

ationonthetimingofthelawsuits.Panel

Cpresents

settlement

amountinform

ation.In

Panel

Dwereport

market-adjusted

(equallyweightedindex)abnorm

alreturns(A

Rs)

generateddueto

class

actionsuitfilings.Foreach

of

thesetests,wereportaverage

AR

sforthe(�

1,0)and(0,0)intervals,respectively.Wereporttw

o-tailed

p-values

beloweach

ARestimate.Toevaluate

thedifference

in

themean

AR

s,wereportatw

o-tailed

t-statistic.Accountingandauditingenforcem

entreleases(A

AERs)are

releasesfrom

theSecurities

andExchangeCommission

(SEC)thatare

enforcem

entactionstaken

forviolationsofSECandfederalrules.Theclass

periodisthetimeframe,asmentioned

inthecomplaint,duringwhichitis

believed

thealleged

fraudoccurs.

Pan

elA

:A

lleg

ati

on

s

Incidence

N(%

sample)

False/misleadingstatements

includingfailure

todisclose

materialinform

ation.

False/misleadingrevenueor

earningprojectionsincluding

actionsto

inflate

stock

price.

Improper

accountingpractices,

includingrevenue,

expense,and/or

earningsmanagem

ent.

All

102(47.22%

)86(39.82%

)125(57.87%

)

Subsample

withAAERs

26(12.04%

)21(10.28%

)30(13.88%

)

Pan

elB

:T

imin

g

Nu

mb

ero

fd

ay

sN

Mean

Median

Intheclass

period

216

376

260

Betweentheendoftheclass

periodandthelawsuitfiling

216

98

77

Betweenthelawsuitfilingandsettlementdate

197

892

910

E.M. Fich, A. Shivdasani / Journal of Financial Economics 86 (2007) 306–336 313

ARTICLE IN PRESS

Pan

elC

:S

ettl

emen

ta

mo

un

ts

Mean

Median

25th

Percentile

75th

Percentile

In$millions

21.67

6.67

3.35

40.5

Panel

D:

Inve

stor

react

ions

Interval

NAverageAR

t-st

ati

stic

Wil

cox

on

Zst

ati

stic

(�1,0)

200

�5.95%

�13.656

�2.547

(0,0)

200

�3.25%

�10.539

�3.021

AAER

Issued?

NInterval

AverageAR

t-st

ati

stic

Interval

Ave

rag

eA

Rt-

sta

tist

ic

Yes

43

(�1,0)

�7.20%

3.61

(0,0)

�4.02%

2.05

(0.00)

(0.00)

No

157

�5.61%

�3.04%

(0.01)

(0.02)

Settlem

entin

the

topquartile?

NInterval

AverageAR

t-statistic

Interval

AverageAR

t-statistic

Yes

50

(�1,0)

�7.81%

3.75

(0,0)

�5.23%

2.66

(0.00)

(0.00)

No

147

�5.29%

�2.48%

(0.01)

(0.04)

Table

1(c

onti

nu

ed)

E.M. Fich, A. Shivdasani / Journal of Financial Economics 86 (2007) 306–336314


in three industries – Chemicals, Energy, and Business Services – which together represent46% of the events. The remaining observations are distributed relatively evenly across awide number of industries.

Our sample period of 1998–2002 overlaps with the Internet boom and bust period,raising the question of whether the fraud lawsuits are linked to the collapse of Internetfirms. We examine how many of the sued firms are Internet firms by following theapproach in Schultz and Zaman (2001). As they discuss, SIC codes cannot identify Internetfirms because these companies span several different industry classifications. We thereforeidentify Internet firms as those in the ‘‘Internet Stock List’’ maintained at www.Internet-news.com. and in USA Today’s ‘‘Internet 50’’ list during our sample period. Only three ofour sample firms are classified as Internet firms, suggesting that most of our samplelawsuits are not an outcome of the Internet collapse.

As discussed earlier, class action lawsuits are typically accompanied by a decline in thestock price, and this pattern is evident in our sample as well. In untabulated tests, wecalculate the cumulative abnormal returns (CARs) for the sample starting 100 days beforethe filing date, using standard event-study methodology (Dodd and Warner, 1983) withmarket model parameters estimated over the prior one-year interval. The CARs are closeto zero until approximately 20 trading days, or one calendar month, prior to the filing andthen become sharply negative, averaging �16% during the (�20, �3) window relative tothe lawsuit filing date. Similar patterns of negative returns prior to class action filings aredocumented by Karpoff, Lee, and Martin (2005) and indicate the revelation of significantadverse information, or ‘‘trigger events,’’ prior to the lawsuit filing.

We search news articles in the month prior to the lawsuit to identify the trigger events.The most frequent is self-disclosure, when the firm voluntarily admits some wrongdoing infinancial reporting, such as improper recording of expenses or revenues, inconsistentaccounting treatment, etc. Self-disclosure occurs in 113 cases, or 52% of the sample.Internal investigations, events in which the firm announces a review of its accountingpractices and/or the accuracy of information disclosed previously to investors, occur in 62cases (29%). Earnings restatements occur in 36 firms (17%), and 22 firms (10%) announcea change in the auditor. Eleven firms in the sample delay SEC filings, five acknowledgethe review of accounting practices without an admission of wrongdoing, and two firmsare subject to allegations of financial misrepresentation by whistleblowers. In 15 cases, weidentify a miscellaneous category of trigger events that includes bribery, shareholderinquiries, federal investigations, and managerial turnover.

In an attempt to identify lawsuits that represent serious lapses in corporate governance,we collect data on AAERs for our sample. We find that 45 of our 216 sample firms aresubject to an AAER and that the alleged violations in the AAERs match those in the classaction lawsuits.2 As an additional measure of the severity of misconduct, we collectsettlement amounts from PricewaterhouseCoopers. For each settlement, we read thesummary notice of the proposed class action settlement in order to ensure that thesettlement corresponds to the class action in our dataset. We note that since settlements

2We use the presence of an AAER as a proxy for SEC enforcement, recognizing that not all enforcements result

in an AAER. We also check the AAERs and find that in many instances the trigger events identified by the SEC

coincide with those we identify using press releases. For example, the trigger events in the AAERs include

30 instances of self disclosure, 5 cases of delayed filings, 13 occurrences of restatements, 9 events of internal

investigations, and 1 episode involving a whistleblower.

http://www.Internetnews.com

http://www.Internetnews.com


and AAERs often occur after considerable time has lapsed, our process has the potential tomiss some serious instances of fraud that have yet to be resolved.Panel C of Table 1 presents information on the settlement amounts for the 197 firms for

which settlement data are available. The average settlement is $21.67 million and the 25th and75th quartiles have values of $3.35 million and $40.5 million, respectively. As shown by Coxand Thomas (2004), SEC actions and high settlement awards are highly correlated, but thiseffect appears to be driven by the size of shareholder losses (Karpoff, Lee andMartin, 2006). Inour sample, 36 of the 43 firms subject to an AAER are also in the top quartile of settlements.

3.3. Valuation impact of fraud litigation

Like earlier studies, our sample shows a significant negative market reaction around thefraud lawsuit filing date. In Panel D of Table 1, we report event-study results for the 216firms in the sample, dropping 16 cases where other major news is released on the lawsuitfiling date (to control for possible bias on the market model parameters, we use simple net-of-market returns in place of market model returns and obtain similar results). Over the( 1,0) interval, the abnormal returns average �5.95%. This estimate is comparable toKarpoff and Lott (1993), who find two-day excess returns of �4.56% for a sample between1978 and 1987, as well as Karpoff, Lee, and Martin (2005), who show a market reaction of�7.00% during 1978–2002. Thus, the filing of a financial fraud class action suit isassociated with a significant negative revaluation of the firm’s equity. For comparison,Agrawal and Chadha (2005) document an average two-day excess return of �4.2% for asample of 159 earnings restatements from 2000 to 2001.Investor reactions are significantly more negative for more serious allegations of

financial fraud. The subsample of class action suits that are accompanied by an AAERdisplays an average two-day excess return of �7.2% compared to �5.61% when no AAERis present. Similarly, filings that result in high settlement awards (measured as the topquartile of settlements) result in an average excess return of �7.81%, compared to anaverage excess return of �5.29% for other lawsuits. In a broader study of corporatelawsuits, Bhagat, Bizjak, and Coles (1998) also find that fraud related lawsuits have moreadverse valuation effects compared to other types of litigation.

4. Lawsuits and directors’ reputation

We record changes in directorships held by outside directors of sued firms for three yearsfollowing the lawsuit. Since many firms appoint directors with staggered three-year terms,the three-year period ensures that a reappointment decision has occurred for each directorfollowing the fraud litigation. We are able to track the post-lawsuit directorships for the113 firms that face class action suits between 1998 and 2000. We track directorship changesby reading proxy statements and bibliographies that appear for each director in the firms’annual reports. When necessary, we supplement with records from The Million Dollar

Directory and Who’s Who in Finance and Industry.

4.1. Changes in directorships

Table 2 shows the frequency of outside directors who remain on the board of the firmfacing fraud litigation. Despite significant firm valuation loss associated with the lawsuit

ARTICLE IN PRESS

Table 2

Directorships held by directors of sued firms

This table provides information on outside directors of 113 sued firms from 1998-2000 as well as on the

directorships they hold during the three years after the lawsuit. Year 0 is the year in which the lawsuit is filed.

Year relative to lawsuit filing

0 1 2 3

Percentage of original outside directors remaining on

board of sued firm

– 94.41% 89.12% 83.24%

Average number of other directorships held by outside

directors of sued firms

0.95 0.85 0.54 0.47

Percentage of outside directors of sued firms sitting on

other boards

49.56% 46.18% 31.32% 29.12%

Average number of other directorships held by outside

directors who hold at least one other board seat in addition

to seat on sued firm’s board

1.92 1.84 1.71 1.62

Percentage of original outside directors who hold at least

one other board seat in addition to seat on sued firm’s

board and lose at least one directorship

– 84.21% 89.47% 95.72%

Percentage of outside directors of sued firms who lose all

other directorships:

Number of other board seats held when fraud lawsuit is

filed:

1 – 83.80% 88.89% 96.30%

2 – 55.56% 63.42% 79.17%

3+ – 31.94% 47.69% 48.61%


filing, almost all outside directors retain their board positions on the firms involved in thelawsuit. Three years after the lawsuit filing, 83% of outside directors retain their board seatat the sued firm. Yermack (2004) finds a 4.6% unconditional annual frequency of outsidedirector turnover (exclusive of delistings) and Fich and Shivdasani (2006) document a6.7% annual departure frequency among the directors of large firms. Therefore, the sampleof fraud firms appears to display no abnormal turnover among outside directors. Thisfinding is consistent with Agrawal, Jaffe, and Karpoff (1999) who also find no evidence ofabnormal turnover among outside directors in a sample of fraud firms.

In sharp contrast, outside directors experience a substantial decline in the number ofother board appointments held. The average number of board seats held at other firms(i.e., interlocking firms) declines from 0.95 in the year prior to the lawsuit to 0.85 in year+1 and further to 0.47 by year +3. The reduction in the average number of directorshipsoccurs both because of a decline in the number of directorships held as well as a lowerfrequency of outside directors who sit on other boards. Table 2 shows that in the year priorto the lawsuit filing, 49.6% of outside directors hold at least one other board appointment.This percentage drops to 46% in year +1 and to 29% by year +3 following the lawsuit.Of the subset of outside directors who hold at least one other board seat prior to the fraudlawsuit filing, 96% experience a reduction of at least one board appointment. The lastthree rows of Table 2 display the frequency with which outside directors lose all otherboard appointments. Of the directors who hold one other board seat, over 96% lose thatseat by year +3. For outside directors who hold three or more other board seats prior tothe lawsuit, almost 49% lose all other directorships by year +3. Thus, the data indicate


a drastic reduction in the number of other board seats held by outside directors of firmsfacing financial fraud lawsuits.The large decline in outside directorships is consistent with several hypotheses. Under

the reputation hypothesis, the decline reflects the damaged reputation of outside directorsof fraud-accused firms. If the lawsuit signals that the firm is in a fraud-prone environment,the endogenous board hypothesis suggests that outside directors might leave other boardsand reallocate their efforts to spend more time monitoring the sued firm. The decline inboard seats could also result from an increased focus by outside directors on the legalissues confronting the fraud-accused firm and an attempt to manage their personal legalexposure. To the extent that directorships are valuable to outside directors, the dataindicate that outside directors of firms accused of fraud bear a personal cost irrespective ofwhich hypothesis underlies the reduction in board seats.Our results differ from Helland (2006), who finds that the number of board positions

rises for outside directors following lawsuits alleging fraud. However, several methodo-logical issues complicate his analysis. First, he considers lawsuits filed during 1985–2002, aperiod that includes a large number of cases initiated prior to the 1995 Private SecuritiesLitigation Reform Act. Second, he does not distinguish between independent and ‘‘gray’’directors. If gray directors sit on the board for reasons other than monitoring, fraudallegations should not have reputational consequences for these directors. Finally, whilewe evaluate reputational effects over a three-year period, Helland uses an eight-yearhorizon, possibly obscuring the loss in directorships that occurs over a shorter time frame.Our results also differ from Srinivasan (2005), who studies firms that restate earnings

and finds small declines in other directorships, concentrated primarily among members ofthe audit committee of the restating firm. A limitation of his approach is that it treats anearnings restatement as an unambiguous signal of board failure. This assumption can beproblematic if the board was responsible for uncovering a reporting failure, in which casedirectors might be rewarded for their discovery. Srinivasan (2005) also appears to includedirectors who hold no other directorships in the analysis. By definition, these directorsdo not have any board seats to lose, causing the overall losses in directorships to bemeasured less precisely.

4.2. Magnitude of personal losses

Do lost directorships represent a financial penalty to outside directors? To provide someperspective on the financial magnitude of the reputational effects, we estimate the value ofa lost directorship for directors of sued firms. We make four assumptions that are largelyconsistent with those in Yermack (2004), who studies the incentives and compensationreceived by outside directors. First, we assume that all directors retire by age 72. Second,we assume that the probability of an outside director departure increases by 4% each yeardue to acquisitions, delistings, or CEO turnover. Third, for each director we assume thatboard compensation would have remained the same as that received during the turnoveryear. For compensation, we include the annual retainer, meeting fees, committee fees, andthe Black-Scholes value of option awards. Finally, we assume that the probability ofdeparture increases by 2% when firm performance is below the median industryperformance in a given year based on three-digit SIC industry-adjusted stock returns.Using return on assets or adjusting the annual stock return by the CRSP value-weightedindex generates similar results.


For each outside director among interlocked firms, we estimate the expected remainingboard tenure. For example, an individual aged 54 would have an expected remainingtenure of 11.7 years. Using a real discount rate of 3%, we estimate that the present value(PV) of lost compensation for directors of fraud firms is $990,155.

Judging whether this PV represents a meaningful loss for sued directors requires anestimate of their wealth, which cannot be readily calculated. We provide two benchmarksagainst which this loss could be compared. The first is the estimated wages of CEOs, sinceapproximately 30% of all outside directors are CEOs of other firms. From proxystatements and ExecuComp, we estimate average CEO total annual compensation to be$7.3 million for outside directors who are CEOs of other firms. For these individuals, thePV of their compensation, assuming a CEO tenure of seven years, is estimated at $45.5million. Hence, a single directorship loss represents approximately 2.13% of theindividual’s wage-related net worth. Alternatively, the lost PV could be compared to anestimate of total wealth. According to Liebowitz (2003), the wealthiest 1% of UShouseholds have an average wealth of $10.2 million.3 Thus, if this is representative ofoutside directors in our sample, a loss of a directorship represents a loss of 9.7% ofnet worth.

These calculations suggest that outside directors suffer a non-trivial loss, but thisinterpretation is subject to two important caveats. First, the NPV of a board appointment,i.e., the excess of the PV of compensation to the PV of effort, is probably much less than $1million. Second, non-monetary rewards such as visibility, prestige, and the business andsocial networking benefits associated with a directorship are also likely to be veryimportant for some directors. Though it is impossible to accurately value these effects, ifoutside directors are optimizing the cost-benefit tradeoff of a directorship before thelawsuit, the subsequent reduction in directorships implies a net reduction in the welfare ofthese directors after the lawsuit.

5. Effects of class action filings on interlocked firms

To understand why outside directors lose board seats at other firms, we study the impactof the lawsuit on the other firms that also employ these outside directors (henceforth,director-interlocked firms). We first examine whether the fraud lawsuit has a valuationimpact on director-interlocked firms. We then study how the presence of directors whohave served on sued firms affects the probability that a firm is sued.

5.1. Wealth effects for director-interlocked firms

Our sample of sued firms contains 1,241 outside directors, 396 of which holddirectorships in firms other than the firm sued for fraud. These 396 directors hold 1,085board seats in 1,022 different publicly traded firms for which CRSP data are available.Thus, we have 1,022 director-interlocked firms that are tied to sued firms through at leastone outside director.

Investor reactions for the director-interlocked sample are negative and statisticallysignificant as shown in Panel A of Table 3. Over the (�1,0) interval, director-interlockedfirms experience an average two-day CAR of -0.98% when a firm is subject to fraud

3See /http://faireconomy.org/econ/state/Talking_Taxes/short%20answers%20only.htmlS.

http://faireconomy.org/econ/state/Talking_Taxes/short%20answers%20only.html

ARTICLE IN PRESSTable

3

Abnorm

alreturnsfordirector-interlocked

firm

saroundclass

actionsuitfilings

Panel

Areports

CA

Rsforasample

offirm

sthatshare

atleast

oneoutsidedirectorwithasued

companyfrom

Table

1,butwherethefirm

isnotnamed

inthe

complaint.Toassessthestatisticalsignificance

ofthe

AR

swereport

WilcoxonrankZ

and

tstatisticsfrom

two-tailed

tests.

InPanel

B,wesort

the

AR

sinto

two

subsamplesbywhether

theshareddirectorsits

onafirm

thatreceived

anAccountingandAuditingEnforcem

entRelease

(AAER)in

additionto

theclass

action

lawsuit.In

PanelC,wesplittheinvestorreactionsbywhether

thesettlementamountisin

thetopquartile.In

PanelD,wesplitreactionsbywhether

theshareddirector

withthesued

companyis

inthesamethree-digitSIC

codeindustry

astheinterlocked

firm

.Foreach

subsample,wereport

average

AR

sforthe(�

1,0)and(0,0)

intervals,respectively.Wereporttw

o-tailed

p-values

beloweach

ARestimate.Toevaluate

thedifference

inthemean

AR

sin

thetw

osubsamples,wereportatw

o-tailed

t-statistic.

Pan

elA

Interval

NAverageAR

t-statistic

Wilcoxon

Zstatistic

(�1,0)

1,022

�0.98%

�4.873

�3.282

(0,0)

1,022

�0.92%

�6.504

�5.464

Pan

elB

AAER

issued?

NInterval

AverageAR

t-st

ati

stic

Interval

Ave

rag

eA

Rt-

stati

stic

Yes

234

(�1,0)

�1.25%

1.77

(0,0)

�1.20%

1.68

(0.00)

(0.00)

No

788

�0.90%

�0.84%

(0.01)

(0.02)

Pan

elC

Settlem

entin

thetopquartile?

NInterval

AverageAR

t-statistic

Interval

Ave

rag

eA

Rt-

stati

stic

Yes

255

(�1,0)

�1.39%

1.98

(0,0)

�1.36%

1.73

(0.00)

(0.00)

No

767

�0.84%

�0.77%

(0.02)

(0.06)

Pan

elD

Interlocked

directorin

thesameindustry?

NInterval

AverageAR

t-statistic

Interval

Ave

rag

eA

Rt-

stati

stic

Yes

58

(�1,0)

�1.17%

1.66

(0,0)

�1.12%

1.67

(0.00)

(0.00)

No

964

�0.97%

�0.91%

(0.00)

(0.00)



litigation. Investors appear to assess the potential severity of the fraud allegation at thetime of lawsuit filing and react more negatively to lawsuits that involve more serious casesof fraud. Panel B shows that announcement excess returns are more negative for lawsuitswhere an AAER occurs, and Panel C illustrates that excess returns are lower for lawsuitsthat result in high settlement amounts, defined as those in the top quartile. Both effects aresignificant at the 10% level. The negative returns to director-interlocked firms cannot befully explained by industry effects. For each outside director in the fraud sample, weexamine whether any of their other directorships are held in firms with the same three-digitSIC code. Panel D shows that excess returns are slightly lower for interlocked firms thatare in the same industry as the fraud firm and that the difference between these returns issignificant at the 10% level. However, both same- and across-industry interlocked firmsdisplay significant negative announcement returns, implying that the entire effect cannotbe explained by industry factors.4

The negative valuation effect for interlocked firms is consistent with severalexplanations. A fraud lawsuit can serve as a negative signal of the outside director’sreputation, leading investors to revise downward the valuation of the director-interlockedfirms. Alternatively, fraud litigation can be a time-consuming endeavor for outsidedirectors, preventing them from devoting sufficient attention to the interlocked boards,thereby eliciting a negative reaction by shareholders of interlocked firms. A thirdpossibility is that the negative returns reflect investors’ concerns about an increasedprobability that the director-interlocked firm is susceptible to fraud. However, the negativereturns cannot be explained solely by industry effects, implying that a signaling effect, ifpresent, operates through common characteristics other than industry. Such commoncharacteristics could include the presence of the same auditor, or customer-supplierrelations and/or strategic alliances that are affected as a result of the class action lawsuit.

To explore these explanations further, we estimate multivariate regressions in Table 4using the abnormal returns (CARs) for the director-interlocked firms as the dependentvariable.5 These models include several variables intended to capture common industryand operating characteristics between sued and interlocked firms. We include a same-industry indicator to measure when the sued and interlocked firm have the same three-digitSIC code. Balsam, Krishnan, and Yang (2003) find that auditors tend to specializeaccording to industries and hence we include a ‘‘same auditor’’ indicator to control for thequality of auditing.6 In addition, we add a ‘‘strategic alliance’’ indicator to measurebusiness relationships between the fraud and director-interlocked firms that could beaffected by the fraud. We collect data on strategic alliances from the Securities DataCorporation’s Joint Venture/Strategic Alliance database and supplement it with searchesof Lexis/Nexis. We consider strategic alliances to occur when firms enter into contracts tosupply products or services, manufacture products, distribute or market products, licensemanufacturing or distribution rights, undertake joint research and development activities,

4As an alternative classification, we also define industries using the Fama and French (1997) industry groupings.

With this approach, we are unable to detect any significant differences in announcement returns for interlocked

firms based on whether the sued and interlocked firms are in the same industry.5The sample consists of 891 observations for which all independent variables are available.6Each firm’s auditing company is extracted from proxy statements, and supplemented when necessary with

information from Who Audits America, A Directory of Publicly-Traded Companies & the Accounting Firms Who

Audit Them. Menlo Park, CA: Data Financial Press.

ARTICLE IN PRESS

Table 4

Multivariate analysis of abnormal returns

OLS estimates of cumulative abnormal returns on firms that share outside directors with companies subject to

class action suits for alleged infringement of rule 10(b)-5 of the SEC Act of 1934. Busy board is a dummy variable

that equals one if at least 50% of outside directors hold three or more directorships. Board size is the natural log

of the number of directors. All director characteristics are recorded for the director shared with the sued firm. If

more than one director is shared then averages are taken. The CEO indicator equals one if the firm’s CEO is a

director in the sued firm. The Lawyer and MBA variables are (0,1) indicators. Standard errors are in parenthesis

below each estimate. All models control for industry effects coded using three-digit SIC codes.

Dependent variable CAR (-1,0) Model (1) Model (2) Model (3)

Estimate p-

value

Estimate p-

value

Estimate p-

value

Firm Characteristics

Interlocked firm is in the same industry

as sued (0,1)

�0.00179 0.1941 �0.0021 0.1262 �0.00203 0.1355

(0.00138) (0.00137) (0.00136)

Same auditor as sued firm (0,1) �0.00189 0.1519 �0.00199 0.1287 �0.00204 0.1154

(0.00132) (0.00131) (0.00129)

Strategic alliance with sued firm (0,1) �0.00124 0.6775 �0.00038 0.8968 �0.00064829 0.825

(0.00298) (0.00296) (0.00293)

Internet firm (0,1) �0.00358 0.6235 �0.00249 0.7311 �0.00355 0.6204

(0.00729) (0.00724) (0.00717)

AAER issued on sued firm (0,1) �0.01374 0.0001 �0.01183 0.0001

(0.00216) (0.00217)

Settlement amount paid by sued firm �0.00866 0.0001 �0.00707 0.0001

(0.00158) (0.00158)

Number of previous lawsuits (0,1) 0.0293 0.1909 0.03458 0.1202 0.03126 0.1557

(0.02238) (0.02223) (0.022)

Firm size �0.00022 0.7648 �0.00015 0.8351 �0.00001234 0.9862

(0.000723) (0.000717) (0.000711)

Return on assets �0.01836 0.1664 �0.01405 0.2871 �0.01437 0.2712

(0.01326) (0.01319) (0.01305)

Director Characteristics

Audit committee (0,1) in interlocked

firm

�0.00371 0.0455 �0.00451 0.0141 �0.00382 0.0362

(0.00185) (0.00183) (0.00182)

Audit committee (0,1) in sued firm �0.00318 0.0871 �0.00324 0.08 �0.00328 0.0724

(0.00186) (0.00185) (0.00182)

Multiple directors shared with sued

firm (0,1)

�0.00725 0.0013 �0.00726 0.0012 �0.00684 0.002

(0.00224) (0.00223) (0.0022)

CEO (0,1) �0.00561 0.0215 �0.0053 0.0293 �0.00541 0.0243

(0.00244) (0.00243) (0.0024)

Number of committees 0.000689 0.4238 0.000848 0.3212 0.00067781 0.4232

(0.000861) (0.000854) (0.000846)

Years as director �0.00161 0.125 �0.00212 0.0435 �0.00212 0.041

(0.00105) (0.00105) (0.00104)

Age 0.00281 0.5545 0.00203 0.6688 0.00238 0.6104

(0.00476) (0.00473) (0.00468)

Lawyer �0.00378 0.1242 �0.00372 0.1282 �0.0031 0.2

(0.00246) (0.00244) (0.00242)

MBA 0.00202 0.3815 0.00242 0.2896 0.00229 0.3127

(0.0023) (0.00229) (0.00226)


ARTICLE IN PRESS

Table 4 (continued )

Dependent variable CAR (-1,0) Model (1) Model (2) Model (3)

Estimate p-

value

Estimate p-

value

Estimate p-

value

Board Characteristics

Governance index �0.00063 0.0106 �0.00047 0.0552 �0.00045329 0.0635

(0.000246) (0.000247) (0.000244)

Busy board (0,1) �0.00416 0.0016 �0.00387 0.0031 �0.00403 0.0018

(0.00131) (0.0013) (0.00129)

Independent board 0.00313 0.0162 0.00325 0.012 0.0033 0.0099

(0.0013) (0.00129) (0.00128)

Board size �0.00119 0.6455 �0.00256 0.3259 �0.00107 0.6735

(0.00259) (0.0026) (0.00255)

Ownership by directors (% of

common)

0.00052 0.0408 0.000494 0.0507 0.00050681 0.0425

(0.000254) (0.000252) (0.000249)

Director stock option plan available

(0,1)

0.00246 0.0851 0.00215 0.1307 0.00249 0.0765

(0.00143) (0.00142) (0.00141)

Industry Effects Yes Yes Yes

F-value 26.05 0.0001 26.37 0.0001 28.43 0.0001

Adjusted R2 0.6355 0.6401 0.666

N 891 891 891


and/or share existing technology or methods. All models also include a vector of industryeffects using three-digit SIC codes.

We also include a control variable to denote whether the interlocked firm is consideredto be an Internet stock. The Internet indicator is intended to control for potential internetbubble effects that could affect stock prices of interlocked firms. We find that only six ofthe interlocked firms are classified as Internet firms.

We examine whether market reactions for interlocked firms vary by the severity of thefraud allegation as measured by an AAER and the settlement amount. We include anindicator variable if the director serves on the audit committee of the sued firm. Since auditcommittee members are likely to develop reputations in fraud detection, fraud incidencewould be more damaging to their reputation, and hence the excess returns for theirinterlocked firms should be lower. We also add an indicator if multiple directors of a suedfirm sit on the board of a director-interlocked firm, as is observed in 17 cases in our sample.Finally, we add two indicator variables to measure the status of the sued director on theinterlocked board: one for whether the director is the CEO of the interlocked firm and ananother if the director sits on the audit committee of the interlocked firm. Reputationalconsiderations suggest that wealth effects should be more pronounced for such directors.Finally, the models include a series of variables designed to measure the strength ofcorporate governance as well as controls for director and firm characteristics.

Industry or other shared attributes between the sued and interlocked firm do not have alarge impact on the announcement returns. The coefficient on the shared industry indicatoris low and either marginally significant or insignificant across specifications. The Internetfirm indicator is also insignificant. Further, neither the strategic alliance indicator nor thesame-auditor indicator yield significant coefficients. However, variables that measure fraud


severity and director responsibility are highly significant in both models. In model (1),CARs are lower if the lawsuit is followed by an AAER (�0.013, p-value ¼ 0.0001), andin model (2) the settlement amount is also negative and significant (�0.002, p-value ¼0.0001). The presence of multiple directors from fraud-afflicted firms is also associatedwith significantly lower excess returns for director-interlocked firms.Investor reactions at interlocked firms are more negative if the sued director serves on

the interlocked firm’s audit committee. Since the audit committee has primary responsi-bility for the accuracy of financial reporting, this suggests that the reputational impact of afraud lawsuit is more severe for directors who are closely involved in monitoring financialfraud. Similarly, serving on the audit committee of the sued firm indicates a potentiallyhigher degree of oversight failure on the part of an outside director. Consistent with thereputation hypothesis, investor reactions at interlocked firms are more negative if thedirector is a member of the sued firm’s audit committee.When the director is the CEO of the interlocked firm, investor reactions are also

significantly more negative. Since incentives of CEOs to engage in fraud are arguablylarger, and their role in either perpetuating or preventing fraud more instrumental thanthat of other board members, the association with fraud for a CEO appears moredamaging to shareholder wealth.We examine the effect of corporate governance on the interlocked firm returns, but a

priori, the expected relation is not obvious. Strong governance at the interlocked firmcould mitigate investors’ concerns about the increased potential for fraud. However, as wedescribe below, strongly governed firms are less likely to face fraud lawsuits. Therefore,a lawsuit could be more of a surprise at well-governed firms, eliciting more negativeannouncement returns. The results in Table 4 show that the independent board indicatorand equity ownership by directors are positively related to announcement returns and thegovernance index displays a negative coefficient. Thus, measures of strong governanceappear to mitigate the negative stock price reaction.7

In summary, the reputation hypothesis at least partially explains the negative excessreturns for director-interlocked firms. Excess returns are lower when the fraud allegation ismore severe, when the shared director sits on the audit committee of the sued firm, andwhen multiple directors of the sued firm sit on the interlocked board. Excess returns arealso lower when the director has important responsibility for fraud prevention and controlat the interlocked firm, as proxied by status as CEO or audit committee member. Theseresults obtain after controlling for industry effects and for the presence of strategicalliances or common auditors.

5.2. Probability of financial fraud

Negative excess returns for director-interlocked firms can arise both because lawsuitscould serve as an adverse signal of the board’s reputation and because lawsuits can signal ahigher likelihood of fraud at the director-interlocked firms. To explore whether such acontagion effect exists, we examine the probability of a financial fraud lawsuit. Specifically,we test whether the presence of outside directors who have been previously associated with

7A board is considered to be independent if 50% or more of directors are independent. We define directors to be

independent if they are not current or former employees of the firm, and do not share any family or material

business relation with the firm or its management.


fraud litigation while serving on another board increases the probability that a firm is suedfor fraud. Such directors, whom we refer to as tainted directors, could possess damagedreputations as fraud monitors and lead to a higher probability of fraud on other firms thatthey serve on. However, their experience with fraud litigation could also make themdesirable candidates for other firms prone to similar litigation.

Because corporate governance attributes vary systematically by industry, we constructan industry-matched control sample of firms that are not subject to fraud litigation.Matching firms are identified by controlling for industry, pre-event performance, and firmsize as suggested in Barber and Lyon (1996). We use total assets in the year prior to thefiling to proxy for firm size, industry-adjusted return on assets (ROA) to proxy for firmperformance,8 and SIC codes to control for industry classification. We are able to matchall firms at the two-digit SIC code level, 72 firms (33% percent) at the three-digit level and31 firms (14% percent) at the four-digit SIC code level.

Table 5 compares the 216 sued firms and their corresponding control firms. Boards ofsued firms tend to be larger than those of control firms by almost one director. Sued firmsalso tend to have fewer independent outside directors, consistent with Dechow, Sloan, andSweeney (1996) and Beasley (1996). The board is classified as ‘‘busy’’ (i.e., at least 50% ofoutside directors hold three or more directorships) in 18% of sued firms and 15% ofcontrol firms. Sued firms are more likely to have a combined CEO-chairman leadershipstructure than control firms. Following Agrawal and Chadha (2005), we examine thepresence of firm-founders and whether a director has financial expertise (i.e., a CPA, CFA,or experience in corporate financial management). We observe a lower percentage ofoutside directors with financial expertise in sued firms. Tainted directors are present on29% of the sued firm boards, but on only 18% of the control firm boards. In addition, theGompers, Ishii, and Metrick (2003) index is significantly higher for sued firms suggesting aweaker governance structure for these firms.9

Model (1) of Table 6 presents estimates from a logit model where the dependent variabletakes the value of one if the firm is sued for fraud and zero if the firm is a matching non-fraud firm. The results indicate significant differences in corporate governance. A highgovernance index (indicating weak governance) significantly increases the probability of afraud lawsuit. In addition, measures of weak board monitoring, such as a non-independent, busy, or staggered board, all significantly increase the probability of fraudlitigation. In economic terms, a one-point increase in the governance index increases theprobability of a fraud lawsuit by 0.86%, an independent board decreases the probabilityby 4.8%, while a busy board increases the probability by 5.9%.10

The presence of a tainted outside director significantly raises the probability of a fraudclass action suit. This result is consistent with both the reputation and the endogenous

8We calculate ROA as follows. The numerator is operating income before depreciation (Compustat item 13)

plus the decrease in receivables (Compustat item 2), the decrease in inventory (Compustat item 3), the increase in

current liabilities (Compustat item 72) and the decrease in other current assets (Compustat item 68). The

denominator is the average of beginning- and ending-year book value of total assets (Compustat item 6). This

ratio is then adjusted by the median ROA for firms in the same 2-digit SIC code.9Since the Gompers, Ishii, and Metrick (2003) index is also viewed as a measure of antitakeover protection, this

result might alternatively reflect that sued firms are more insulated from the discipline of external takeovers.10These estimates correct for the state based sample approach employed using the procedure described in

Manski and McFadden (1981). In our sample, the unconditional probability of experiencing a rule 10(b)-5 class

action lawsuit is 16.9%.

ARTICLE IN PRESS

Table 5

Univariate comparison of fraud firms and control sample

This table reports sample means for 216 firms subject to a class action lawsuit and 216 matching firms. We

match firms following Barber and Lyon (1996) by including companies that are not sued with pre-event

performance and size similar to those of the sued firms. Busy board is ‘‘one’’ if 50% of outside directors hold three

or more total directorships in for-profit firms. The governance index is the Gompers, Ishii, and Metrick (2003)

index. We calculate the market-to-book ratio as the market value of the firm’s equity at the end of the year plus

the difference between the book value of the firm’s assets and the book value of the firm’s equity at the end of the

year, divided by the book value of the firm’s assets at the end of the year. Size is calculated as the natural

logarithm of a firm’s asset (Compustat item 6). Return on assets (ROA) is calculated as operating income before

depreciation (Compustat item 13) plus the decrease in receivables (Compustat item 2), the decrease in inventory

(Compustat item 3), the increase in current liabilities (Compustat item 72), and the decrease in other current assets

(Compustat item 68), divided by the average of beginning- and ending-year book value of total assets (Compustat

item 6). This ratio is then adjusted by the median ROA for firms with the same two-digit SIC. All characteristics

are calculated for the year preceding the class action suit. For each characteristic, we provide two-tailed t- and

Z-statistics to test for equality of mean and median values, respectively.

Characteristic Fraud Sample Control Sample t Z

Board size 13.431 12.602 2.3 0.86

Percentage of outside directors 52.1% 55.3% 1.77 1.59

Percentage of firms with busy board (0,1) 17.69% 14.72% 1.61 2.73

Percentage of firms with a staggered board 53% 41.13% 2.46 2.44

Tainted outside director (0,1) 0.2917 0.1806 2.73 2.7

CEO is chairman (0,1) 0.8287 0.7037 3.1 3.06

CEO belongs to the founding family 0.275 0.195 2.23 1.98

Number of board meetings 7.6343 7.5556 0.24 0.84

Percentage of outside directors with financial expertise 23.4% 42.78% 3.02 2.5

CEO salary+bonus (in $MM) 1.4248 1.5871 1.44 2.3

CEO tenure (in years) 7.7553 7.625 0.22 2.85

Governance index 9.3611 8.537 2.98 2.5

Market-to-book ratio 1.3427 1.2991 0.84 3.51

Firm size 8.8021 8.6383 1.55 0.76

ROA 18.53% 18.92 0.30 1.35

N ¼ 216 N ¼ 216


board hypotheses. According to the reputation hypothesis, directors who have taintedreputations as weak monitors due to prior lawsuits tend to allow financial fraud to occuron other boards. The endogenous board hypothesis suggests that sued directors are thosewho have expertise in dealing with fraud lawsuits and are therefore sought for service byfirms at risk of being sued. Therefore, the coefficient on sued director may simply reflectthe pattern that such directors self-selected to sit on boards of fraud-prone firms.To distinguish between the reputation and the endogenous board hypotheses, we

estimate a simultaneous-equations framework to account for the endogeneity betweentainted directors and fraud lawsuits. Specifically, we estimate the following system:

Prob: Tainted Director ¼ d1 Prob: Fraud þ b1X 1 þ �1,

Prob: Fraud ¼ d2 Prob: Tainted Directorþ b2X2þ �2,

where X1 is a vector of exogenous determinants of tainted director probability and X2 is acorresponding vector for fraud probability.

ARTICLE IN PRESS

Table 6

Probability of Fraud Lawsuit

Estimation of fraud using ordinary logit and simultaneous logit equation approaches. Fraud is a dummy

variable equal to one if the firm is sued for allegedly violating rule 10(b)-5 of the 1934 SEC Act. We analyze the

matched sample of fraud firms and non-fraud firms described in Table 5. In models (1) and (3), the dependent

variable takes the value of one if the firm is a fraud firm (defendant in class action lawsuit). The key independent

variable in (1) is ‘‘Tainted director’’ an indicator that is one if any current outside director also sits on the board of

a sued firm. Fraud instrument in model (2) equals the fitted value from an untabulated first stage regression.

Tainted director instrument in (3) equals the fitted value from a different untabulated first stage regression in

model (3). Delaware firm is a dummy variable equal to one for firms incorporated in the state of Delaware.

Standard errors are reported in parentheses.

Regular Logit 2nd Stage 2nd Stage

Dependent ¼ Fraud Dependent ¼ Tainted Dependent ¼ Fraud

Model (1) Model (2) Model (3)

Variable Estimate p-value Estimate p-value Estimate p-value

Constant �7.3855 0.0001 �1.8789 0.3103 �7.9268 0.0001

(1.6459) (1.852) (1.8326)

Fraud instrument 1.5333 0.0397

(0.7454)

Tainted director 0.5452 0.0044

(0.1914)

Tainted director instrument 3.9296 0.0359

(1.8732)

Governance Characteristics

Governance index 0.1075 0.005 �0.00971 0.8264 0.0674 0.1376

(0.0383) (0.0443) (0.0454)

Board size 0.121 0.0026 0.0774 0.0943 0.0524 0.3781

(0.0402) (0.0462) (0.0594)

Independent board (0,1) �0.5998 0.0085 0.533 0.0412 �0.9613 0.0006

(0.2278) (0.2611) (0.2803)

Staggered board (0,1) 0.5157 0.0221 �0.3531 0.1666 0.7109 0.0058

(0.2253) (0.2553) (0.2576)

Busy board (0,1) 0.7419 0.0062 �0.0873 0.7824 0.5715 0.0589

(0.2713) (0.3162) (0.3026)

Number of meetings 0.0227 0.5138 �0.0559 0.1429 0.0453 0.2357

(0.0348) (0.0382) (0.0382)

Director w/accounting expertise (0,1) �0.4471 0.0417 �0.0326 0.8953 �0.3199 0.2077

(0.2196) (0.2475) (0.2539)

CEO tenure 0.0114 0.5309 �0.00508 0.7995 0.0101 0.6085

(0.0183) (0.02) (0.0196)

CEO ownership (% of common) �0.0464 0.161 �0.0586 0.2159 �0.00264 0.9454

(0.0331) (0.0474) (0.0386)

CEO is chairman (0,1) 0.8245 0.0023 �0.2765 0.3883 0.9116 0.0029

(0.2703) (0.3206) (0.3065)

CEO from founding family (0,1) 0.4748 0.039 0.1219 0.6321 0.4302 0.109

(0.23) (0.2547) (0.2684)

Salary+bonus �0.3122 0.004 0.0473 0.7064 �0.3459 0.0049

(0.1085) (0.1257) (0.1231)

Black-Scholes value of options granted 0.0582 0.0205 �0.0304 0.2685 0.0802 0.0034

(0.0251) (0.0275) (0.0274)


ARTICLE IN PRESS

Table 6 (continued )

Regular Logit 2nd Stage 2nd Stage

Dependent ¼ Fraud Dependent ¼ Tainted Dependent ¼ Fraud


Variable Estimate p-value Estimate p-value Estimate p-value

Firm Characteristics

Market-to-book 0.4106 0.1162 0.1884 0.511 0.1826 0.5538

(0.2614) (0.2866) (0.3085)

Return on assets (ROA) 0.0267 0.9742 0.7114 0.4068 �0.3621 0.712

(0.8244) (0.8575) (0.9807)

Firm size (log of sales) 0.3596 0.0068 �0.0749 0.6189 0.4138 0.0042

(0.1328) (0.1505) (0.1447)

Identifying Variables

Previously sued firm 1.5253 0.0001

(0.2768)

Delaware firm 0.6489 0.033

(0.3043)

Pr 4 w2 o0.001 o0.001 o0.001


Identification of this system requires a variable in X1 that is related to fraud probabilitybut unrelated to the probability of a tainted director on the board. To meet thisrequirement, we use an indicator variable for Delaware incorporation. This choice is basedon existing literature. For example, Romano (1985) notes that Delaware firms are moreexposed to shareholder litigation. According to Iacobucci (2004), shareholder class actionsare particularly costly for defendant firms incorporated in Delaware. Weiss and White(2004) argue that Delaware law makes shareholder class actions attractive, which inducesDelaware firms to settle quickly to avoid the higher litigation expenses.We also require a variable in X2 that is related to the probability of a tainted director on

the board but unrelated to the probability of fraud litigation. For this purpose, we use anindicator variable for whether the firm has been sued previously in a fraud class action suit.According to the endogenous board hypothesis, a prior lawsuit would signal that the firmis in a fraud-prone environment, prompting firms to add fraud monitoring specialists tothe board. Therefore, a prior lawsuit should be correlated with the presence of a tainteddirector on the board, but is unlikely to influence the probability of another fraud-relatedlawsuit.Since probabilities of lawsuits and tainted directors are not observed directly, estimation

of the simultaneous equations system occurs in two stages. In the first stage, theprobabilities of a fraud lawsuit and of a tainted director are estimated including allexogenous variables using logit. We then use the predicted values from the first stage asexplanatory variables in the second-stage models. To conserve space, Table 6 reports onlythe results of the second-stage models.11

11Though not tabulated, the results from the first stage models show that the Delaware indicator and the

previous lawsuit indicator satisfy the properties required for identification of the system. The Delaware indicator


We find support for the endogenous board hypothesis. In model (2), the probability of atainted director is positively related to the probability of a fraud lawsuit. The coefficient onthe fraud instrument is positive and significant at the 5% level. However, the endogeneityof tainted directors does not fully explain the relation between fraud likelihood and tainteddirectors that is documented in model (1). The second-stage estimates in model (3) showthat the instrument for a tainted director is positively related to fraud probability,consistent with the reputation hypothesis. Thus, accounting for the endogeneity of tainteddirectors and fraud likelihood using simultaneous equations models reveals support forboth the reputation and endogenous board hypotheses.12

Other results in Table 6 are consistent with existing literature. Consistent with Agrawaland Chadha (2005), both the presence of an outside director with financial expertise andthe absence of a firm founder lower the likelihood of financial fraud, but these results losestatistical significance in the simultaneous equations framework. Firm size also yields apositive and significant coefficient. As in Jones and Weingram (1996) and Field, Lowry,and Shu (2005), this probably reflects the tendency of larger firms to be more attractivelawsuit targets because of their ability to pay larger damages to plaintiffs. As in Beasley(1996), we find a positive and significant coefficient for board size, but this effect losessignificance in the simultaneous equations framework in model (3). We also find thatoption-based CEO compensation increases fraud probability, similar to results in Burnsand Kedia (2006).

6. Directorship losses and valuation effects for interlocking firms

6.1. Probability of retaining directorships

To understand the reduction in directorships, we examine why outside directors losesome board seats while continuing to serve on other boards. If directorships are lost as apenalty for damaged reputations, we expect the lost directorships to be among well-governed firms. The incentive to preserve the board’s reputation will be greater for firmswith strong governance, and hence well-governed firms are likely to be more proactive inreplacing directors with damaged reputations. The reputation hypothesis further predictsthat directorships are more likely to be lost following severe instances of fraud as well aswhen directors with greater reputation are associated with fraud. In addition, it suggeststhat directors who serve in board positions that bear significant responsibility for frauddetection are more likely to lose these positions.

We also examine whether the likelihood that the interlocked firm will be sued affectswhether a sued director retains that board seat. Under the endogenous board hypothesis,outside directors specializing in monitoring fraud could choose to spend more time onfirms that are likely to be sued. According to the legal liability hypothesis, however, weexpect that sued directors will forgo directorships in interlocked firms that are more likely

(footnote continued)

is positively related to the probability of a fraud lawsuit at the 1% level of significance, but not significantly

related to the probability of a tainted director. The previously sued indicator is positively related to the probability

of a tainted director at the 1% level but not significantly related to the probability of a fraud lawsuit.12Since the dependent variables in the simultaneous equations framework can only be observed as binary

outcomes, we can only test whether the coefficients on d1 and d2 are statistically significant but cannot evaluate theeconomic importance of these effects (Maddala, 1983).


to be sued for fraud. Thus, the probability of retaining a board seat should be negativelyrelated to the likelihood that the firm will be sued if litigation avoidance is a primaryconsideration.We analyze the directorships held by outside directors of the 113 firms that are sued for

fraud during the first three years of our sample period. We tally 678 outside directors forthese firms, 336 of which hold at least one other board seat in addition to their directorshipat the sued company. These 336 directors hold a total of 645 different outsidedirectorships. We analyze 487 of the 645 directorships for which we are able to getcomplete data from Compustat and proxy statements on interlocked firms during the yearthe class action suit is filed. We construct an indicator variable that is one if an outsidedirectorship is retained and is zero if it is lost within the first three years following the classaction suit and use it as the dependent variable in a logistic regression.13

The first set of explanatory variables in Table 7 relates to lawsuit characteristics. Weinclude indicators for whether the lawsuit is followed by an AAER and whether thesettlement is in the top quartile. We also include the excess returns for the sued firm aroundthe lawsuit filing as an additional proxy for the severity of the fraud allegation.Firm-specific variables include the market-adjusted stock return of the interlocked firm,

because prior firm performance can affect whether outside directors are added to ordropped from the board (Hermalin and Weisbach, 1988). We include firm size sinceappointments at large firms provide greater visibility to directors and hence directors mightbe reluctant to voluntarily forgo directorships at large firms. The model controls for thenumber of fraud lawsuits filed against the interlocked firm over the previous five-yearperiod. We also include an Internet dummy variable in the regressions to control fordirectorships lost due to the Internet collapse (and related lawsuits) rather than fromreputational effects.The models include governance characteristics of the interlocked firm. We use the

Gompers, Ishii, and Metrick (2003) governance index along with indicator variables forwhether the board is independent, whether the board is busy, whether the positions ofCEO and chairman are combined, and whether the CEO at the time of the director’s initialappointment to the board is no longer in office. This last variable is intended to capturethe possible reluctance of CEOs to remove directors that they originally appointed tothe board.We examine several director-specific attributes. We include indicator variables that

equal one if the director is on the audit committee of the sued firm or the audit committeeof the interlocked firm. We also use the reputation of the outside director, measured bythe number of other board seats held. Other variables include indicators for whether thedirector is a CEO of a public company, a lawyer, or a gray director on the board ofthe director-interlocked firm, as well as the director’s age and the director’s tenure on theboard of the interlocked firm.The results in Table 7 indicate that directors of sued firms are more likely to re-

tain directorships in interlocked firms with a high governance index. In economic terms,

13In unreported tests, we estimate clustered logit models to address the concern that the observations might not

be independent if the firms share certain common attributes. We estimate models clustered by the law firm and by

the auditing firm and obtain inferences that are virtually identical to those described here. The clustered logit

model provides robust standard errors in the event of clustering by relaxing the independence assumption to

independence between the clusters only.

ARTICLE IN PRESS

Table 7

Probability of retaining a board seat

This table presents logistic regressions of the probability of retaining a board seat for outside directors of fraud

firms. The sample consists of 487 directorships held in non-sued firms by outside directors of sued firms. We check

whether a directorship is lost anytime three years after the lawsuit is filed. With this information, we construct a

dependent variable that equals one if the director keeps a directorship and equals zero otherwise. Standard errors

are reported in parentheses below each coefficient estimate. The symbols ***,**, and * denote statistical

significance at the 1%, 5%, and 10% levels, respectively.

Model (1) Model (2)

Constant

Firm characteristics 6.5918 *** 7.3919 ***

(1.6114) (1.662)

Market reaction of sued firm (day 0 AR) 47.7327 *** 48.0851 ***

(8.1026) (8.204)

Market-adjusted stock return (Ri–Rmkt)t�1 0.0288 0.0258

(0.0216) (0.0232)

AAER issued (0,1) �0.5931 ** �0.5972 **

(0.2658) (0.2667)

Settlement in the top quartile (0,1) �0.038 * �0.0414 *

(0.0228) (0.0231)

Governance index 1.4874 *** 1.4777 ***

(0.3448) (0.345)

Probability of being sued �12.8964 **

(5.8952)

Number of previous lawsuits in the past five years �0.3766 �0.3582

(0.3341) (0.3338)

Firm size (natural log of sales) 0.1405 ** 0.1328 **

(0.062) (0.063)

Internet firm (0,1) 0.1186 0.2001

Board characteristics (0.4954) (0.5049)

Busy board (0,1) �0.1188 �0.2043

(0.3775) (0.3832)

CEO is also chairman of the board (0,1) �0.409 �0.318

(0.3813) (0.3865)

Board is independent (0,1) �0.2243 �0.2195

(0.6234) (0.6217)

Appointing CEO no longer in office (0,1) �0.3824 �0.3262

Director characteristics (0.2634) (0.266)

In audit committee of sued firm (0,1) �0.4504 * �0.4581 *

(0.2529) (0.2547)

In audit committee of interlocked firm (0,1) �2.4118 *** �2.4482 ***

(0.3488) (0.3511)

Number of directorships held �0.3181 *** �0.3383 ***

(0.0887) (0.091)

Gray (0,1) �0.2908 �0.3338

(0.5446) (0.5892)

Director’s age (in years) �0.00494 �0.00387

(0.018) (0.0181)

Lawyer (0,1) �0.1736 �0.1985

(0.2878) (0.2897)

CEO of public company (0,1) �0.3238 �0.3015

(0.2505) (0.2528)

Director’s ownership (% of common) 3.4731 *** 3.5546 ***

(1.3481) (1.3479)

Pr 4 w2 o0.0001 o0.0001



a one-point increase in the governance index increases the probability of seat retention by3.17%. Measures of the severity of the fraud allegation are also important. Directorshipsare less likely to be retained if there is an AEER issued, if the lawsuit results in a highsettlement, or if the market reaction to the initial lawsuit filing is more negative. Thus,directors involved in more serious allegations of fraud experience a greater decline indirectorships, a pattern consistent with the reputation hypothesis. From an economicperspective, the issue of an AAER appears to have a larger impact on board seat retention.An AAER decreases the probability of retaining a board seat by 8.74% but a settlement inthe top quartile decreases the probability of retaining a board seat by only 0.65%.Several director-specific attributes are also important in determining whether board

seats are retained. Directors who serve on the audit committee of the interlocked firmface an 8.26% greater probability of losing their board seat at the interlocked firm. It ispossible that the class action lawsuit has a more damaging impact on the reputation ofaudit committee members or that they are more likely to reduce directorships because oftime constraints imposed by the lawsuit. We also find weak evidence that sitting on theaudit committee of the sued firm is more likely to result in lost directorships. Thecoefficient on this variable is significant only at the 10% level in models (1) and (3), but itseconomic effect appears to be large—audit committee members have a 33.16% higherprobability of losing their seat on the interlocked board. We also find that higher equityownership by directors increases the likelihood of retaining their outside directorship,while directors who sit on more boards are more likely to experience a reduction inboard seats.Overall, these results suggest that at least part of the reduction in board seats can be

attributed to a disciplinary effect in the market for outside directorships. Specifically, thepattern of directorships being retained at firms with a high governance index suggests thatwell-governed firms are more likely to remove sued directors from their boards. Thispattern is also consistent with the endogenous board hypothesis where outside directorschoose to focus on other boards that face a greater need for monitoring while forgoingdirectorships at firms that have strong governance. However, this result is inconsistentwith outside directors voluntarily resigning outside directorships due to concerns aboutlegal liability.Several results suggest that reputational penalties are higher at the margin for outside

directors. We find that directors are more likely to lose board seats when fraud severityis high and when their status as an audit committee member confers a greater respon-sibility for monitoring fraud. In addition, directors who serve on many boards aremore likely to lose directorships, a pattern that is also consistent with the reputationhypothesis.

6.2. Investors’ reactions to director departures

We conduct an event study around the loss of directorships. If directors associated withfinancial fraud are replaced on interlocked boards because of damaged reputation, theirdeparture should be welcome news for shareholders of interlocked firms. In thesecircumstances, a replacement of a fraud-affiliated director can serve as a positive signal ofthe board’s monitoring intensity. However, if fraud-affiliated directors voluntarily forgodirectorships at interlocked firms, their departure might be bad news for shareholders if theinterlocked firm loses access to a valuable director.


We track director departures from boards of interlocked firms over a two-year periodfollowing the filing of the class action lawsuit. To determine which directors remain on theboard and which subsequently depart the board, we review both the annual report and theinterlocked firm’s proxy statements. We search the Wall Street Journal Index and Lexis/Nexis when we are able to identify a departure, and read newspaper stories and companypress releases in order to ascertain the reason for the departure. We set aside all departuresfor which an announcement date is not available. We also exclude departures related tonormal retirements and to board term limits. We discard six departures related to death orillness and two that are publicized to be forced. Finally, we remove a handful ofannouncements that are contaminated with other major news on the date of the departure.We are able to identify 144 announcements related to voluntary director departures. Weclassify exits to be voluntary if a director leaves either to pursue other interests or to take aposition elsewhere. Of the 144 announcements, 94 departures (65.28%) are related todirectors of companies sued for fraud.

Panel A of Table 8 presents excess returns (CARs) for announcements of departures ofoutside directors from boards of interlocked firms. For the entire sample, the (-1,0) CARs

are positive and significant with a mean (median) of 1.46% (1.31%). Dividing the sampleof departures between directors involved in a fraud lawsuit and other directors notinvolved in fraud allegations reveals significant differences. The average CAR is positiveand significant when the departing directors are related to sued companies, but is notsignificantly different from zero for announcements of departures of outside directors whoare unaffiliated with sued firms.

In Panel B of Table 8, we estimate three multivariate regression models using thedeparture announcement returns where we control for changes in board size resultingfrom the departure and the identity of the replacing director. We include an indicatorfor whether the departing director sits on the board of the firm accused of fraud. Wealso include variables to indicate whether a departing director sits on a board accusedof a more severe allegation of fraud as indicated by an AAER or a high settlementamount.

The estimates indicate that the announcement of a departure of an outside director whosits on the board of a sued firm is positive news for shareholders of interlocked firms. Inmodel (1), the coefficient estimate on the sued-firm director variable is positive andsuggests that the announcement return is 2.6 percentage points higher for such departures.The magnitude of the investors’ reaction suggests that the economic effect of sued-firmdirector departures is large. Further, the abnormal returns for sued-firm director de-partures are higher when these directors are associated with more severe allegations offraud, as measured by an AAER or a high settlement amount.

These event-study results are consistent with several explanations. First, the departure ofa sued-firm director would be welcome news if investors expect a replacement director ofhigher quality. Second, our earlier results show that firms with sued directors are morelikely to face fraud lawsuits themselves. Thus, the departure of a fraud-affiliated directormight signal a reduced probability that the interlocked firm will face a similar fraud classaction suit. Third, the departure could signal that the board is more vigilant than investorsexpected, leading to a positive stock price reaction. However, the event-study results donot support the view that voluntary cutbacks on board commitments by sued directorstend to deprive interlocked firms of valuable expertise or monitoring services by thesedirectors.

ARTICLE IN PRESS

Table 8

Investor reactions to voluntary departures of outside directors

Panel A presents day 0 abnormal returns (ARs) and day (�1,0) cumulative abnormal returns (CARs) associated

with announcements of the departure of outside directors from the board. Reported below each estimate are p-

values from two-tailed tests using a t-test for means, and a Wilcoxon Z signed-rank test for medians. Panel B

presents three ordinary least squares (OLS) regressions where the dependent variable is the cumulative abnormal

return (CAR) over the (�1,0) interval. The key independent variable in all regressions is an indicator for whether

the departing director is a director of a sued firm.

Panel A: event study

Is the director related to a sued firm?

All departures N ¼ 144 YES N ¼ 94 NO N ¼ 50 t-statistic Z–statistic

Mean AR

(0,0) 0.0081 0.0117 0.00133 3.29

(0.08) (0.01) (0.36)

(�1, 0) 0.0146 0.0203 0.004 2.97

(0.07) (0.01) (0.29)

Median AR

(0,0) 0.0073 0.0094 0.0009 3.76

(0.05) (0.01) (0.28)

(�1, 0) 0.0131 0.018 0.0036 3.01

(0.03) (0.01) (0.19)

Panel B: Multivariate Analyses


Dependent variable Day (�1,0) CAR Day (�1,0) CAR Day (�1,0) CAR

Estimate p-value Estimate p-value Estimate p-value

Intercept 0.016 0.0001 0.021 0.0001 0.028 0.0001

Departing director is related to a sued firm (0,1) 0.026 0.0001 0.028 0.0001 0.0303 0.0001

Number of directorships of departing director 0.028 0.044 0.016 0.02 0.013 0.047

Board reduction (0,1) 0.005 0.079 0.003 0.12 0.004 0.112

Independent board 0.004 0.444 0.002 0.389 0.004 0.503

Tenure of departing director 0.011 0.075 0.017 0.13 0.014 0.14

Independent replacement 0.008 0.61 0.003 0.59 0.007 0.66

Gray replacement �0.011 0.112 �0.010 0.081 �0.009 0.091

Equity ownership of replacement 0.000 0.596 0.000 0.472 0.000 0.481

Firm size (natural log of sales) 0.008 0.59 0.011 0.421 0.010 0.471

Settlement in the top quartile (0,1) 0.011 0.077 0.009 0.052

AAER Issued (0,1) 0.016 0.022 0.021 0.038

R2 23.05 25.10 26.89

N 144 144 144


7. Conclusions

We investigate reputational effects of financial fraud for outside directors of firmsaccused of fraud using a sample of firms facing class action lawsuits alleging violation ofSEC rule 10(b)-5. Despite almost no evidence of abnormal turnover from the board of


firms accused of fraud, we find that fraud is followed by a large and significant decline inthe number of other board appointments held by outside directors. This decline isconsistent with both a reputational penalty being borne by outside directors as well as anendogenous adjustment of monitoring expertise, where the expertise is reallocated to firmsthat are revealed to be more fraud-prone than previously expected.

We show that a contagion effect exists for financial fraud through the board of directors.Firms that share directors with other boards accused of fraud are more likely to face fraudallegations themselves. When a firm faces a fraud class action lawsuit, other firms thatemploy outside directors of the defendant firm also experience a significant decline invaluation. Both these effects increase with the severity of the fraud allegation as measuredby AAER releases and settlement amounts. The valuation loss for the interlocking firms ismagnified if these firms possess weak governance characteristics and if the interlockeddirectors play a potentially critical role in influencing the likelihood of fraud due to theirstatus as a member of the audit committee of either the sued or the interlocked firm, or asthe CEO of the interlocked firm.

Our findings show that outside directors are more likely to lose other boardappointments when the severity of the fraud allegation is high, and when the outsidedirector sits on the audit committee of the interlocked firm. Directorships are also morelikely to be lost at firms with strong corporate governance, as measured by the Gompers,Ishii, and Metrick (2003) index. We also find that investors of interlocked firms experiencepositive announcement returns when a sued director departs the board. Overall, our resultsillustrate that fraud allegations have significant reputational consequences for outsidedirectors of these firms and valuation effects for the other firms that are linked to suedfirms through interlocked directorships.

References

Agrawal, A., Chadha, S., 2005. Corporate governance and accounting scandals. Journal of Law and Economics

48, 371–406.

Agrawal, A., Jaffe, J.F., Karpoff, J.M., 1999. Management turnover and governance changes following the

revelation of fraud. Journal of Law and Economics 42, 309–342.

Balsam, S., Krishnan, J., Yang, J.S., 2003. Auditor industry specialization and earnings quality. Auditing:

A Journal of Practice and Theory 22, 71–97.

Barber, B.M., Lyon, J.D., 1996. Detecting abnormal operating performance: the empirical power and

specification of test statistics. Journal of Financial Economics 41, 359–399.

Beasley, M.S., 1996. An empirical analysis of the relation between the board of director composition and financial

statement fraud. The Accounting Review 71, 443–465.

Bhagat, S., Coles, J.L., Bizjak, J.M., 1998. The shareholder wealth implications of corporate lawsuits. Financial

Management 27, 5–27.

Brickley, J.A., Coles, J.L., Linck, J., 1999. What happens to CEOs after they retire? New evidence on career

concerns, horizon problems, and CEO incentives. Journal of Financial Economics 52, 341–377.

Burns, N., Kedia, S., 2006. The impact of performance-based compensation on misreporting. Journal of Financial

Economics 76, 35–67.

Coles, J.L., Hoi, C-K., 2003. New evidence on the market for directors: Board membership and Pennsylvania

senate bill 1310. Journal of Finance 58, 197–230.

Cox, J.D., Thomas, R.S., 2004. Public and private enforcement of the securities laws: have things changed since

Enron? Notre Dame Law Review 80, 893–907.

Dechow, P.M., Sloan, R.G., Sweeney, A., 1996. Causes and consequences of earnings manipulation: An analysis

of firms subject to enforcement actions by the SEC. Contemporary Accounting Research 13, 1–36.

Dodd, P., Warner, J.B., 1983. On corporate governance: a study of proxy contests. Journal of Financial



Fama, E., 1980. Agency problems and the theory of the firm. Journal of Political Economy 88, 288–303.

Fama, E., French, K.R., 1997. Industry costs of equity. Journal of Financial Economics 43, 153–193.

Fama, E., Jensen, M., 1983. The separation of ownership and control. Journal of Law and Economics 26,

301–325.

Ferris, S.P., Pritchard, A. C., 2001. Stock price reactions to securities fraud class actions under the private

securities litigation reform act. The University of Michigan Law and Economics Research Paper No. 01-009.

Fich, E.M., 2005. Are some outside directors better than others? Evidence from director appointments by Fortune

1000 firms. Journal of Business 78, 1943–1972.

Fich, E.M., Shivdasani, A., 2006. Are busy boards effective monitors? Journal of Finance 61, 689–724.

Field, L., Lowry, M., Shu, S., 2005. Does disclosure deter or trigger litigation? Journal of Accounting and


Gilson, S., 1990. Bankruptcy, boards, banks and blockholders: Evidence on changes in corporate ownership and

control when firms default. Journal of Financial Economics 27, 355–387.

Gompers, P.A., Ishii, J.L., Metrick, A., 2003. Corporate governance and equity prices. Quarterly Journal of

Economics 118 (1), 107–155.

Harford, J., 2003. Takeover bids and target directors’ incentives: The impact of a bid on directors’ wealth and

board seats. Journal of Financial Economics 69, 51–83.

Helland, E., 2006. Reputational penalties and the merits of class action securities litigation. Journal of Law and


Hermalin, B., Weisbach, M., 1988. The determinants of board composition. Rand Journal of Economics 19,

589–606.

Hermalin, B., Weisbach, M., 1998. Endogenously chosen boards of directors and their monitoring of the CEO.

American Economic Review 88, 96–118.

Iacobucci, E.M., 2004. Toward a signaling explanation of the private choice of corporate law. American Law and

Economics Review 6, 319–344.

Jarrell, G., Peltzman, S., 1985. The impact of product recalls on the wealth of sellers. Journal of Political

Economy 92, 512–536.

Jones, C.L., Weingram, S.E., 1996. The determinants of 10b-5 litigation risk. Working paper No. 118. John M.

Olin Program in Law and Economics, Stanford University.

Kaplan, S., Reishus, D., 1990. Outside directorships and corporate performance. Journal of Financial Economics

27, 389–410.

Karpoff, J.M., Lott, J.R., 1993. The reputational penalty firms bear from committing criminal fraud. Journal of

Law and Economics 36, 757–802.

Karpoff, J.M., Lee, D.S., Martin, G.S., 2005. The cost to firms of cooking the books. Working paper.

The University of Washington.

Karpoff, J.M., Lee, D.S., Martin, G.S., 2006. Legal penalties for financial misrepresentation. Working paper.

The University of Washington.

Klein, B., Leffler, K.B., 1981. The role of market forces in assuring contractual performance. Journal of Political

Economy 89, 615–641.

Liebowitz, M., 2003. Winning Responses to Tough Tax Questions, published at www.ufenet.org, May 12.

Maddala, G., 1983. Limited dependent and qualitative variables in econometrics. Econometric Society

Monographs in Quantitative Economics No. 3, Cambridge University Press, Cambridge.

Manski, C.F., McFadden, D.L., 1981. Structural Analysis of Discrete Data and Econometric Analysis. The MIT

Press, Cambridge.

Romano, R., 1985. Law as a product: Some pieces of the incorporation puzzle. Journal of Law, Economics, and

Organization 1, 225–283.

Schultz, P., Zaman, M., 2001. Do the individuals closest to Internet firms believe they are overvalued? Journal of

Financial Economics 59, 347–381.

Skinner, D., 1997. Earnings disclosures and stockholders lawsuits. Journal of Accounting and Economics 23,

249–282.

Srinivasan, S., 2005. Consequences of financial reporting failure for outside directors: evidence from accounting

restatements. Journal of Accounting Research 43, 291–334.

Weiss, E.J., White, L.J., 2004. File early, then free ride: how Delaware law (mis)shapes shareholder class actions.

Vanderbilt Law Review 57, 1797–1881.

Yermack, D., 2004. Remuneration, retention, and reputation incentives for outside directors. The Journal of

Finance 59, 2281–2308.

http://www.ufenet.org

Documents

Financial fraud, director reputation, and shareholder wealth