Corporate Scandals, Capital Structure and … Scandals...Corporate Scandals, Capital Structure and...

Corporate Scandals, Capital Structure andContagion E¤ect

Stefano Bonini*New York University, Stern School of Business44W 4th St., 10012, New York, NY, USA

[email protected]

Diana BoraschiBocconi University,

Via Roentgen 1, 20122, Milan, [email protected]

This draft: 1st May 2009

JEL Codes: G32, G33, K41Keywords: corporate scandals, security o¤erings, capital structure, contagion ef-

fect, market timing.

Acknowledgement 1 *Corresponding author. We thank Julian Franks, Matthew Pritskerand seminar participants at the NYU Pollack Center for Law and Business for helpful com-ments and suggestions. We are particularly indebted with Je¤rey Wurgler for invaluable sug-gestions and support. We gratefully acknowledge Cornerstone Research support for providingsecurity class action data The ideas expressed in the paper do not necessarily re�ect those ofthe authors�a¢ liation. Any errors remain our own.

Abstract

A wave of corporate scandals has recently hit the market reviving attention on the e¤ectsof these events on shareholders�value, corporate governance and stock market reaction. Thedocumented far-reaching e¤ects of corporate scandals on security prices may have a markettiming value that managers may be willing to exploit. In this paper we analyze whethercompanies involved in a security class action do exhibit di¤erential capital structure decisionsand if the information revealed by a corporate scandal a¤ects security issuances by industrypeers. Our �ndings show that before a SCAS is �led, companies engaged in a scandal had ahigher number of security o¤erings and due equity mispricing they were more likely to useequity as a �nancing mechanism. Following the SCAS �ling they also exhibit decreasing bookand market leverage. �nally we observe signi�cant contagion e¤ects on industry peers. Theseresult suggest that investors tend to process company-speci�c news as a more industry-wideinformation.

Introduction

A wave of corporate scandals has recently hit the market reviving attention on the

e¤ects of these events on shareholders�value, corporate governance and stock market

reaction. Academic research has shown that companies su¤er a considerable decline in

both stock prices and debt ratings upon Chapter 11 �ling announcements, �nancial re-

ports re-stating or �nancial distress announcements (Palmrose, Richardson and Scholz,

2004; Lang and Stulz, 1992; Brewer and Jackson, 1997). Scandals early detection,

if not prevention, is therefore valuable to stakeholders. Agrawal and Chadha, (2005)

document that appropriate corporate governance mechanisms may positively in�uence

the probability of earnings restatments. Agrawal and Cooper (2007) support this ev-

idence highlighting the higher turnover of top management and top �nancial o¢ cers

soon before and immediately after an accounting scandals. Dyck, Morse and Zingales,

(2007) show that non-traditional mechanisms and stakeholders-at-large play a consid-

erable role in triggering fraud-detection. Given the documented far-reaching e¤ects of

corporate scandals, it is interesting to ask whether managerial behavior in companies

engaged in a corporate scandal a¤ects also �nancial decision on capital raising, and

in particular whether managers, anticipating the risks of a corporate scandal exhibit

di¤erent capital structure policies than those of their peers. Surprisingly though this

question is still unanswered. In this paper we try to �ll this gap by looking at the

security issuance patterns of companies engaged in Security Class Actions Suit between

1996 and 2005. In particular, we address three main research questions:

(a) What is the ex-ante and ex-post �nancing pattern of �rms engaged in a corporate

scandal?;

(b) Is there a contagion e¤ect in the �nancing pattern of the industry after a cor-

porate scandal was revealed?;

(c) Is there a contagion e¤ect in the stock prices of the industry after a corporate

scandal was revealed?

Previous literature has addressed corporate scandals by studying cases of bank-

ruptcy announcement, public announcement of frauds in the press and earnings re-

statements. In this paper we adopt the engagement in a security class action suit as

a proxy of a corporate scandal. We collect data from the Stanford Securities Class

Action Clearinghouse (SSCASC) database1. This measure of corporate scandals allow

to generalize the results to a broader set of cases because it deals with actions that: a)

1Database is mantained in cooperation with Cornerstone Research.

3

are important enough to have meaningful e¤ects on security-holders value and b) leave

the company as a going concern allowing menaningful ex-ante and ex-post di¤erential

analysis. In fact less than 10% of cases included in the SSCAC database end up in

bankruptcy announcements.

Our �ndings show that before a SCAS is �led, companies engaged in a scandal had

a higher number of security o¤erings compared to their industry average. At the same

time, we document that since �rms before the scandals experienced overvaluation in

stock prices they were more likely to use equity as a �nancing mechanism. Compared

to their peers, �rms involved in a security class action consistently issue more equity in

the two-year period preceding the �ling of the suit. Consistently with Market Timing

Hypotheses, we �nd that SCAS �rms exhibit decreasing book and market leverage

before the �ling due to abnormal volumes of equity o¤erings. Soon after the �ling

though, leverage increases sharply and signi�cnatly due to the readjustment in equity

market value.

Looking at contagion e¤ects on the �nancing pattern of the industry, we �nd that

equity issuances decrease for both peers and SCAS �rms over time, and this decrease

is more pronounced for SCAS �rms. We observe that close to the SCAS �ling there is

a decrease in debt and equity issuances for both samples. The existence of a signi�cant

negative equity and debt issuance trends can be interpreted as a contagion e¤ect in the

�nancing pattern, i.e. a SCAS �ling generates a decrease in equity and debt o¤erings

in the industry.

Finally, we investigate the e¤ect of corporate scandals on the �rm�s competitors�

stock prices. We test the presence of a negative contagion e¤ect on stock prices of the

industry of the �rm involved in a corporate scandal. For the [-1,0] and [-5,5] event

windows we �nd that peers su¤er a cumulative abnormal return of -.20% and -.65%

respectively. These results shed light to the fact that corporate scandals do have a

negative impact on their industry. Furthermore we study this contagion e¤ect dividing

the sample into accounting and non-accounting allegations and we �nd that the negative

stock price reaction of focal �rms with non-accounting allegations is strongly signi�cant

for the [-15,+15] and [-20,+20] event windows, while this is not the case for accounting

allegations. Cases with accounting allegations do not show a statistically signi�cant

contagion e¤ect in their industry. This result is aligned with Gande and Lewis (2009)

who provided evidence on the price reaction to SCAS �lings.

These results allow to shed light on the �nancing and security issuance behavior of

�rms engaged in corporate scandals. Our results also allow to conclude that indepen-

4

dently from their intensity, corporate scandals do generate e¤ects at the industry level

by leading to a contraction in security o¤erings and a decrease in stock returns for all

the industry constituents.

The remainder of this paper is organized as follows. Section I summarize previous

work on corporate scandals and present the hypotheses that we test in our study.

Section II presents the data and summary statistics. Section III presents the results of

our empirical analysis of the �nancing pattern of �rms engaged in corporate scandals.

Section IV presents the results of our empirical analysis of the contagion e¤ect in stock

prices and �nancing pattern. Section V presents the robustness tests performed and

Section VI concludes the paper.

1 Motivation and Hypotheses

1.1 Corporate Scandals and Security O¤erings

Corporate scandals can be de�ned as widely publicized incidents involving allegations

of managerial wrongdoing, disgrace, or moral outrage of one or more members of a com-

pany. Typical schemes of fraudulent behavior include misstatements of �nancial �gures

on current, past or future investments or operations, failure to disclose information or

delay information retention, bribery, insider trading, and any other illegal activities

that hurt the shareholders of the �rm (Dyck, Morse and Zingales, 2007). Companies�

misconduct may have two separate but important e¤ects on managerial actions: �rst,

it may signi�cantly reduce stock prices making secondary equity o¤erings increasingly

diluting and costly; second it will reasonably reduce, or cancel altogether, manager-

ial independence in taking capital structure-related decisions. Managers, arguably, are

aware of the risks associated with malpractices and therefore have strong incentives

to �make the best out of it while it lasts�. It is then reasonable to expect that they

will exploit this information asymmetry to increase the amount of funds they collect

to anticipate a potential capital constraint after the scandal eruption. Funds then may

be used in several ways: to deliver a steady stream of cash �ows, dividend payments

and investments, to pursue buyback plans, to rebalance at a lower cost the �nancial

structure of the company or simply to enhance the liquidity stock in a similar spirit to

Ivashina and Scharfstein (2009). SCAS �ling documents provide meaningful examples

of these behaviors. In Cisco (2001) the plainti¤ alleges that: "[...] After completing

5

more than 20 major acquisitions between 9/99 and 2/01, by issuing more than 400 mil-

lion shares of Cisco stock, [...] , on 2/6/01, Cisco announced extremely disappointing

2ndQ F01 results"; similarly in Bay Networks (1997) it is alleged that: "[...] materially

false or misleading statements enabled Bay Networks to Complete stock-for-stock acqui-

sitions during the Class Period". Working capital �nancing is claimed by the plainti¤

in Supergen 2003: "[...] SuperGen sold millions of shares and notes [...] so as to pro-

vide it with ample monies to fund its operations. However, this all took place prior to

revelations concerning the veracity of the Company�s statements regarding Mitozytrex".

In this spirit we conjecture our �rst hypothesis:

Hypothesis 1: Ex-ante, �rms engaged in a corporate scandal have a greater amountof security o¤erings compared to their industry average.

The Market Timing Hypothesis of Capital Structure states that when making deci-

sions about funding, managers take into account the current conditions of the debt and

equity markets. Managers will choose the funding scheme that looks more favorable

pro-tempore, and if none seems to be favorable then fund raising might be deferred.

Support to the market timing theory comes from the empirical evidence that shows the

existence of managers�opportunism when it comes to the �rm�s �nancing needs (Gra-

ham and Harvey, 2001). Although this theory is short in explaining many of the factors

that have been traditionally considered in the studies of corporate capital structure, it

has strong empirical evidence that supports the existence of a behavioral component

in managers when it comes to �nancing their �rms. Baker and Wurgler (2002) build

their capital structure predictions on the historical stock prices of �rms and further

evidence con�rms that indeed stock prices play an important role in explaining capital

structure and capital structure changes (Welch, 2004). As for stock prices, the market

timing hypothesis argues that �rms tend to issue equity after the value of their stock

has increased (Hovakimian, Opler and Titman, 2001) and that corporate leverage is

best understood as the cumulative e¤ect of past attempts to time the market (Baker

and Wurgler, 2002). One important assumption underlying the market timing hypoth-

esis is the possible existence of stock price misvaluation. If this is the case, managers

of a �rm that has overvalued (undervalued) stock price will opportunistically exploit

this mispricing by issuing equity (debt). This later fact was con�rmed by Graham and

Harvey (2001). In a interview survey to 392 U.S. and Canadian CFOs, 76% of the

sample reported that the amount by which their stock was over or undervalued was an

"important" or "very important" factor when taking decisions about equity issuance.

6

Corporate scandals act as information revelation mechanisms to equity market par-

ticipants. A scandal shed new light on the actual managerial and accounting practices

of the �rm, revealing information that was previously unavailable to investors. Evidence

show that in extreme cases ending in bankruptcy �ling, investors reaction is strong and

signi�cant with sharp declines in stock prices of the �rms involved in the scandal (Lang

and Stulz, 1992; Rao and Hamilton, 1996; Agrawal and Chadha, 2005). The stock price

drop following such events can be interpreted as evidence of a previous stock overvalua-

tion due to either an accounting phenomenon (such as a misrepresentation of earnings)

or also because some information regarding the company�s investments or risk exposure

was not fully available to the market. Accordingly, we expect that:

Hypothesis 2: Ex-ante, �rms engaged in a corporate scandal made greater use ofequity �nancing compared to their industry averages.

If managers -due to the information asymmetry which eventually lead to a scandal

- time the market issuing more equity when the stock is overvalued, then we can con-

jecture two ancillary predictions. First, if equity issuance is higher than their peers,

leverage by construction should be lower Similarly, once a scandal erupts, the abnormal

security issuance pattern should revert towards the industry mean. Accordingly we

de�ne the following two hypotheses:

Hypothesis 3: Ex-ante, �rms engaged in a corporate scandal had lower levels ofleverage compared to their industry average.

Hypothesis 4: After the corporate scandal is unveiled, the stock price of these �rmsadjusts to its �fair�price and thus �rms start �nancing themselves as the mean of their

industry.

1.2 Corporate Scandals and Contagion E¤ect on �nancing pat-

tern

Academic research on contagion e¤ects at the corporate level, has focused on the

spillovers of shocks occurring in one entity to other entities. Previous literature has

explored the contagion e¤ects on stock returns following a bankruptcy (Lang and Stulz

1992), earning restatment (Gleason et al. 2008) and managerial forecasts announce-

ments (Ramnath 2002). Similarly, Giesecke (2003) and Theocarides (2007) have ex-

plored contagion in the corporate bond market showing that bond prices, yields and

7

spreads react to �rm-speci�c information. Yet, no previous study has investigated the

existence of a contagion e¤ect on capital structure decisions by companies. Since listed

companies raise capital in the market, they are exposed to investor sentiments and

market momentum and, possibly, to information concerning contiguous companies that

investors may transfer to the entire industry. The recent �nancial crisis has provided

an illuminating example of this phenomenon where inherently sound companies have

experienced the same dry-up in capital as weaker peers in their industry. Despite their

managers�e¤orts, "the capital market window [was] just closed" for both high and low

quality companies (Federal Reserve Board (2008)).

In this spirit, a SCAS �ling is a signal that a meaningful mismanagement has oc-

curred in a company. Investor may infer that this behavior can be common practice

across the industry and therefore increase the capital constraints on peer companies. A

highly constrained �nancing environment will lead to increased cost of external �nanc-

ing and ultimately to a contraction of the total security o¤erings of the industry. Thus,

we generate the following hypothesis:

Hypothesis 5: The eruption of a corporate scandal will cause a contagion e¤ect onthe �nancing pattern of the industry peers, generating a contraction in both debt and

equity issuances.

There are several characteristics in an industry that can a¤ect the existence and

magnitude of a contagion e¤ect. We expect the degree of similarity among the �rms�

cash �ows to intensify the extent of the contagion e¤ect on the �nancing pattern of

one industry. This intensi�cation of the contagion e¤ect is due to the fact that highly

similar �rms are likely to have investments with similar cash �ow characteristics and

similar risk exposures (Lang and Stulz, 1992). If a security class action suit conveys

bad news about the projection of cash �ows or �rm risk, then investors will be more

likely to reassess also the �rm�s competitors�cost of �nancing. Measuring similarity as

the correlation of returns between the competitors and the �rm engaged in a corporate

scandal2, we expect that:

Hypothesis 6: Coeteris paribus, the contagion e¤ect on the �nancing pattern ofthe peers group is larger for industries in which competitors show a higher degree of

cash �ows similarity with �rms involved in the corporate scandal.

2As measured by the correlation of returns between the industry portfolio and the �rm engaged inthe corporate scandal in the year preceding the �ling of the class action suit.

8

1.3 Corporate Scandals and Contagion E¤ect on stock prices

A natural second step would be to evaluate if these corporate scandals a¤ected also

the competitors� returns. Most studies on contagion e¤ects have focused mainly on

US bank failures (Kannas, 2004). These studies state that the failure of a large bank

can undermine public con�dence in the banking system as a whole, which may in turn

threaten the stability of the �nancial system by causing runs on other banks (Diamond

and Dybvig, 1983; Aharony and Swary, 1983; Swary 1986). One seminal study in the

topic of the contagion e¤ect that departs from the banking industry, investigates the

e¤ect of bankruptcy announcements on the equity value of the �rm�s competitors (Lang

and Stulz, 1992). The authors �nd that on average the market value of a weighted port-

folio of the common stock of the bankrupt �rm�s competitor�s decreases by 1% at the

time of the bankruptcy announcement and that this decline is statistically signi�cant.

Lang and Stulz (1992) tested the existence of a contagion e¤ect in non-�nancial �rms

at an intra-industry level and later Brewer and Jackson (2002) extended these results at

the inter-industry level working on a database of commercial banks and life insurance

companies. Ferris et al. (1997) demonstrated that large �rm bankruptcies generate a

dominant contagion e¤ect. In their study, competitors experience a signi�cant loss of

0.56% in the three-day window around Chapter 11 announcement. Small �rm bank-

ruptcies also generate a dominant contagion e¤ect among smaller sized competitors.

This research question is closely related with Gande and Lewis (2009) who documented

statistically signi�cant market price e¤ects following a corporate scandal. Looking at

security class actions they use stock price returns, legal environment and the expected

e¤ects of a class action to develop a probabilistic model to predict the initiation of a

SCAS. The corporate �nance-related variables they use in their model are unexpected

earnings and managerial compensation but there is no metric addressing capital struc-

ture phenomena. Yet, it is reasonable to expect that corporate scandal have a di¤erent

impact on stock prices of industry peers of a company involved in a SCAS conditional

on previous capital structure decisions such as leverage and cash �ow level. To test this

intuition we generate the following hypotheses

Hypothesis 7: Ex-post, �rms engaged in a corporate scandal will cause a negativecontagion e¤ect on stock prices within their industry.

Standard corporate �nance show that leverage increases a company�s riskiness and

therefore its stock volatility. Thus, when a scandal is unveiled, stock price reaction by

9

peers should be positive and increasing in leverage due to greater elasticity of equity

value to the total value of the �rms.

Hypothesis 8: The contagion e¤ect on stock prices of peer companies is larger forhighly leveraged industries.

Firms�cash �ows similarity may as well result in a higher response of peers stock

prices. Firms with comparable investment structures generating similar cash �ows are

arguably exposed to the same risk factors (Lang and Stulz, 1992). Since a security

class action suit generally conveys bad news about future cash �ows and the �rm�s risk,

investors will be more likely to reassess the value of peers�equity the higher the degree

of similarity in cash �ows. Measuring similarity as the correlation of returns between

the competitors and the �rm engaged in a corporate scandal, we expect that:

Hypothesis 9: The contagion e¤ect on stock prices of peer companies is larger thehigher the degree of cash �ow similarity of the competitors of the �rm involved in the

corporate scandal.

2 Data and summary statistics

2.1 Data

Previous literature on corporate scandals has adopted earnings restatements, bank-

ruptcy announcement and announcement of frauds in the press as measures of a scan-

dal. In this paper we depart from these approaches and we proxy a corporate scandal

by the the �ling of a security class action suit in the United States, as emerging from

the Stanford Securities Class Action Clearinghouse database. This de�nition of corpo-

rate scandal helps us generalize the results to a broader set of corporate events because

it deals with less severe cases than a �nancial default, as less than 10% of our cases

end up in bankruptcy announcements. By adopting data at the Security Class Action

level we can test whether scandals do a¤ect �rms and their peers behavior condition-

ally and unconditionally on the scandal intensity. Our database includes several types

of corporate scandals such as self-dealing frauds, disclosure failure, misrepresentation

of accounting data, etc. One important concern, as highlighted by Dyck, Morse and

Zingales (2007), is the possible inclusion of cases that were just frivolous allegations; to

deal with this potentially severe sample bias issue we also exclude actions �led before

10

the passing of the Private Securities Litigation Reform Act of 1995 (PSLRA) that was

designed, among others, with the goal of reducing courts�workload on trivial suits.

The original Class Action Suits database has 2,479 cases from January 1996 to

December 2006. We keep only cases �led between January 1996 and December 2005

to allow for the availability of at least two years of �nancial statement data after the

suit �ling. We then dropped highly speci�c SCAS classi�ed as �Analyst related�, �IPO

Allocation�, �Mutual Fund�and �Option Backdating� (thus we only leave �Classic�

SCAS cases)3. The rationale is that these cases are generally related to one isolated

event (listings or managerial compensation) that is less likely to have an impact on

a broader cross-section of security holders. We dropped private holdings, �rms in the

�nancial and utilities sectors (sic codes 6000-6999 and 4900-4999), and cases that didn�t

have Compustat and CRPS information for the period required. The �nal sample

reduces to 793 security class action suit cases. Fifty four percent (432) of the cases

involve accounting allegations, and the remaining 46 percent (361) are classi�ed as

cases involving non-accounting allegations. At the time of data collection, 16 percent

(127) of the cases were still ongoing, while the remaining 84 percent (666) of the cases

were already settled. We matched the �rms from the SCAS database with Compustat

and CRPS using the �rm�s cusip. In the �nal sample of SCAS cases we have 765 cusips,

meaning that several �rms might have more than one security class action suit �ling.

Mean total assets in the �ling year of these �rms were $4,642.62 millions. The sample

contains a total of 204 di¤erent 4-digit sic codes, that we use to generate peer-groups

comparisons. To identify the dispersion of cases by industry we classi�ed each case

according to the Fama & French industry classi�cation (1997): on average we have 21

di¤erent Fama & French industries in each �ling year (see Table I).

INSERT TABLE 1 HERE

Finally, to control that security class action suits are not a proxy of bankruptcy -

3The majority of the cases in the database are classi�ed as Classic. �IPO Allocation cases� arecases �led from 2001 to 2002 alleging that underwriters engaged in undisclosed practices in connectionwith the distribution of certain IPO shares. �Analyst related cases�are cases �led from 2001 to 2004alleging that the brokerage �rm analysts falsely provided favorable coverage for certain issuers. TheseAnalyst cases involve securities directly a¤ected by allegedly false analyst research reports. �MutualFund cases�are cases �led from 2003 to 2004 alleging wrongful acts in the management of the funds.All Mutual Fund cases involve funds alleged to have engaged in market timing practices. When themutual fund parent company is sued only for misstatements or omissions of material information, thecase is classi�ed as Classic rather than Mutual Fund. Classic cases are cases involving 10(b) claims(misstatements or omissions) and/or other common securities law violations.Classic cases are all cases that are not IPO Allocation, Analyst and Mutual Fund cases.

11

more speci�cally of Chapter 11 �lings - we matched our data with LoPucki�s UCLA

Bankruptcy Research Database. We manually merged information from the two data-

bases and observed that on average only 6% of the �rms in our sample �led for chapter

11 in the period 2 years before or after the �ling of the suit. This result allows us

to argue that, since SCAS are not a proxy for bankruptcy, capital structure changes

are not a result of bankruptcy driven corporate restructuring . Table II provides the

distribution of cases included in our sample by event year, type of allegations and by

amount of copanies that eventually �led for Chapter 11 in the two years before or after

the �ling of the security class action suit.

INSERT TABLE 2 HERE

To allow comparisons with the average �nancing behavior industry peers, we con-

struct a measure given by the value-weighted portfolio of �rms not involved in the SCAS

in the event year that are classi�ed with the same 4-digit sic code.

2.2 Variables de�nition

Capital structure variables were collected following Baker & Wurgler (2002). Book

equity is measured as total assets minus total liabilities and preferred stock plus deferred

taxes and convertible debt. Market equity is measured as the number of common shares

outstanding multiplied by the stock price. Book debt is measured as total assets minus

book equity. Book leverage is measured as book debt divided by total assets. Market

leverage is measured as book debt divided by the sum of total assets minus book equity

plus market equity. The amount of total �yearly- security o¤erings is measured as

the sum of debt issuances and book equity issuances. Debt issuances are measured as

the change in total assets minus change in book equity divided by total assets. Book

equity issuances are measured as the change in book equity minus the change in balance

sheet retained earnings, divided by total assets4. Additionally, since debt and equity

issuance are sometimes negative, indicating repurchases or debt and equity voluntary

cancellations,we constructed a dummy variable equal to one when wither equity or debt

issuances are smaller than zero, and zero otherwise.

4Debt and equity issues could also be measured using cash �ow data. We used balance sheet databecause there was more data available, and thus the amount of cases under analysis was greater.

12

3 Corporate scandals and Capital Structure

3.1 Security o¤erings

We conjectured that since fraud detection may a¤ect the availability and cost of future

�nancing, managers have incentives to take advantage of this information asymmetry

with the market to increase the amount of funds they raise. Similarly, we expected a

�rm engaged in a fraudulent behavior -such as lack of disclosure of information and/or

misstatement of accounts- to have greater needs of cash and liquidity, which would

translate in a greater amount of capital raise. Following this intuition, we compared

the weighted average amount of security o¤erings made by the sample of �rms engaged

in a SCAS with the average amount of o¤erings made by their peers (the value-weighted

portfolio of the remaining �rms with the same 4-digit sic code). The comparison was

performed using data for the 6 years window f�2;+3g around the �ling of the SCAS.Results reported in Table 3 o¤er support to our hypotheses.

INSERT TABLE 3 HERE

Ex-ante, �rms engaged in a corporate scandal issue signi�cantly more securities

than their peers. Yet, this issuance pattern is abnormal and disappears after the SCAS

�ling. On average, two years before the event, �rms engaged in a corporate scandal

issue 5.35 times more securities than their peer sample. One year before the �ling,

abnormal security issuance starts decreasing but is still 2.52 times higher than that

of the industry peers. In the event year, i.e. when, the SCAS is �led, the abnormal

issuance is 101% higher than the peer group All di¤erences are statistically signi�cant

at the 1% level on both one and two-tailed tests

Hypothesis 4 predicted that once the information gap with the market that allowed

abnormal security issuance is eliminated, the issuance pattern should revert towards

the market mean. Results reported in table 3 con�rm this intuition: On the three

years after the SCAS �lings, sued �rms decrease considerably their security o¤erings

and their issuance pattern is not statisticaly di¤erent from that of their peers. In fact

there is a mild evidence, although insigni�cant, that issuances are less than the industry

average. This result is not surprising and can be interpreted as an overshooting e¤ect:

market reacts sharply to the SCAS and prices drop below their "fair" value reducing

the chances for capital raising.

13

3.2 Financial mix: Equity and Debt o¤erings

The previous analysis showed that there is robust evidence of greater security issuance

before a scandal erupts, which supports the idea that �rms and managers exploited a

temporary overpricing due to undisclosed information. Yet, this information gap should

e¤ect more havily equity than debt. According to the Market Timing Hypothesis, �rms

with higher current stock prices -relative to their past stock prices, book values or

earnings- are more likely to issue equity rather than debt and repurchase debt rather

than equity (Hovakimian, Opler and Titman, 2001). In this light we argue that the

retained information allows �rms to mantain overvalued stocks, leading to higher equity

issuances. Accordingly, we expect these �rms to show smaller evidence of a di¤erential

issuance of public debt.

INSERT TABLE 4 HERE

Results reported in Table IV con�rm our predictions. Ex-ante SCAS �rms issue

far more equity than their comparable weighted average portfolio of peers, and the

di¤erence is statistically signi�cant for all dates. Two years before the event, �rms

engaged in a corporate scandal issue 7.7 times more equity than their peer sample.

Similarly to results observed for the security issuances test, this pattern is decreasing in

time although signi�cance is consistently high at the 1% level. In particular, one year

before the event the event (t = �1) SCAS �rms issued 4.26 times more than their peers;during the year of the �ling of the security class action, the abnormal equity issuance

drops to 2.39 times than the peers�sample. As predicted, after the event, SCAS �rms

reduce considerably their equity issuances which are never signi�cantly di¤erent from

the industry average.

Debt issuance evidence provides additional support to the hypotheses. Before the

scandal is unveiled, SCAS �rms make a remarkably smaller use of debt as opposed to

equity. Cross-sectionally, debt o¤erings are aligned with those of the industry peers

with the exception of one year before the �ling. Yet, �nancing decisions after the SCAS

�ling change sharply: equity issuances shrink and debt issuances turn negative and

signi�cant for the �rst two years of the event window. At t = 3 , debt issuance is

still negative although not signi�cant. Firms in the peers�sample show a signi�cantly

di¤erent behavior with both debt and equity o¤erings being relatively stable in the two

windows before and after the SCAS �ling. Interestingly , issuance �gures show a strong

14

evidence of discrete, one-time downward changes around the event date. Since �gures

are estimated over event windows distributed over a 10 years time horizon, it is not

likely that this change is correlated with market conditions. Di¤erently, we interpret

this change as a possible consequence of a contagion e¤ect on peers: when a SCAS is

�led, investor may increase their risk estimates that other companies have engaged in

similar practices thus reducing stock prices and incresing debt required yields, which

ultimately result in more costly capital and deferred or reduced capital raising. We

address this issue in section 4.

3.3 Leverage

The previous analyses show remarkable di¤erences in the security issuance patterns of

companies targeted by a SCAS. Yet these �gures may not fully capture the complete

set of �nancing decisions by companies. In fact privately negotiated �nancing like bank

loans are by construction excluded from the data. This source of capital is largely used

in addition to publicly placed securities to shape up companies��nancial structures.

In particular, following hypothesis 3 and previous results, we should expect market

leverage to be not signi�cantly di¤erent or decreasing from that of the industry due to

overpriced equity before the SCAS, and to increase soon therafter due to the strong

adjustment in prices following the SCAS announcement. Similarly, book leverage should

decrease before the �ling as an e¤ect of incremental equity increase and rise in the

following years as evidence of a greater use of non-public debt by the company due to

too costly or closed market conditions.

We tested these intuitions by analyzing the market and book leverage �gures for

companies sued by security-holders and the control peers�group.around the event date.

Results reported in Table 5 con�rm these predictions.

INSERT TABLE 5 HERE

Firms engaged in SCAS show decreasing levels of book leverage, although di¤erenes

are not signi�cant except for year �2. Di¤erently, book leverage di¤erences increasesigni�cantly from the �ling date. Furthermore, this result is fully generated by SCAS

�rms�changes since the peer group doesn�t show any signi�cant change in the average

book leverage over the 5 years event window.

Market leverage �gures are not largely di¤erent between the two groups before the

�ling date. Yet, we document a strongly signi�cant increase in market leverage at

15

the event date and for all the following years. Similarly to book leverage, market

leverage �gures for the peer group are constant over time suggesting that di¤erences

are determined by drops in the market value of equity of SCAS �rms.

4 Contagion e¤ect on external �nancing decisions

and stock prices.

4.1 Contagion e¤ect on external �nancing decisions

Lang and Stulz (1995) and Ferris et al (1997) document the existence of signi�cant con-

tagion e¤ects on stock market prices of competitors of �rms �ling for Chapter 11. In a

study focused on the Telecom industry, Akhigbe et al. (2005) show that these e¤ects

are signi�cant the higher the degree of similarity in size and cash �ows of the competi-

tors. These results support the idea that existing stakeholders react to the bankruptcy

�ling news since it reveals adverse information about asset values, practices and future

prospects of the industry as a whole While, price reactions to a bankruptcy �ling are

not surprisingly associated with large price drops, reactions to SCAS initiation on stock

prices of �ling companies and their peers may be less intuitive since less than 7% of the

SCAS eventually evolve in a bankruptcy �ling. Romano (1991) and Francis, Philbrick,

and Schipper (1994) documented negative stock price reactions upon the initiation of

a security lawsuit using two small sets of cases. Gande and Lewis (2009) provide a

�rst comprehensive analysis of the e¤ect on stock prices upon the �ling of a SCAS. On

average, stock market prices drop by more than 14% in the [-10;+1] window around the

�ling. Additionally, they provide preliminary evidence that stock market prices exhibit

contagion e¤ects similar to those observed by Lang and Stulz (1995) and Ferris et al

(1997). If market prices react to a peer�s bankruptcy �ling and a SCAS, arguably equity

and debt �nancing should become relatively more costly changing the external �nanc-

ing opportunity cost. Companies in fact continuously manage their capital structure by

issuing or buying back equity, raising and repaying debt conditional to market condi-

tions. Surprisingly though, to the best of our knowledge, there is no previous study on

contagion e¤ect on capital structure decisions of companies following a bankruptcy or

corporate scandal. Contagion in security issuances would thus be a reaction of investors

to readjustments in their risk evaluation of the overall industry which imposes greater

costs of �nancing.

16

In this section we provide novel evidence on this phenomen by exploring external

�nancing decisions by competitors of �rms involved in a security class-action lawsuit.

Following hypothesis 5, we model a trend variable T aimed at capturing the evolution

of external capital rasing by the aggregate of competitors in the same industry. The

values of the trend variable range from f1; 6g and are linked to the event years so thatT is equal to one when the event year is �2, T takes a value of two when the event year

is �1 and so forth.

To explore these trends in security o¤erings we performed the following cross-

sectional random-e¤ects regression:

Yit = �i + �i(T ) + "it (1)

where, Yit is the dependent variable capturing the aggregate i� th industry equity,debt or total security o¤erings, T is the trend variable, and eit is the error term of the

regression. Regression results are robust to exogenous factors like market momentum,

business cycles and sentiment since we are working with event years and not calendar

years. Business cycles, market trends, sentiment and other variables should not a¤ect

interpretation of our results. Additional robustness tests are presented in Section 5.

Figure 1 and Table 8 show regression results for SCAS �rms and their peers Our

results support the intuitions in hypothesis 5 as overall issuances decrease at an in-

creasing rate for both subsample over time. The trend coe¢ cient for both subsamples

is negative, statistically signi�cant and, not surprisingly, greater for the SCAS subsam-

ple. Intercept are large and positive, indicating a positive net security issuance over

time. Regression signi�cance as captured by Wald statistic�s �2 is robustly signi�cant

at the 1% level.:

INSERT FIGURE 1 HERE

Previous results on security issuance by SCAS �rms suggested the existence of a

di¤erent e¤ect on debt and equity deals. Following this evidence and the prediction in

Hypothesis 6 we break down the security issuance trend analysis by type of security. As

reported in Figure 2 and Table 8, we �nd that equity issuances decrease for both peers

and SCAS �rms. The trend coe¢ cient of the troubled �rms is over 13 time larger than

the one of their peers. Still, peers experience a negative, strongly signi�cant coe¢ cient

which indicates a contraction in capital raising in public markets. Results for debt

17

issuance are somwhat di¤erent. Not surprisingly, regression esitimates for SCAS �rms

are not signi�cant. This result can be explained recalling the evidence on debt issuance

and book leverage of SCAS �rms which showed a strong decrease in debt issuance

after the �ling followed at t = +2 by a recovery. On the other hand, results for the

peers group are strongly signi�cant with a negative coe¢ cient for the Trend variable

which indicates that a Security Class Action lawsuit on one competitor a¤ects the debt

capacity of the entire industry. In summary, we �nd that in the vicinity of the event

there is a decrease of both debt and equity issuances for both samples.

INSERT FIGURE 2 HERE

INSERT TABLE 6 HERE

Contagion e¤ect and cash �ow similarity

Hypothesis 6 argued that if a contagion on capital structure decisions exists, it should

be larger the closer the similarity of companies�cash �ows. To test this hypothesis we

model similarity as the correlation of returns between the industry portfolio and the

�rms engaged in the corporate scandal for the year preceding the �le of the class action

suit We then de�ne a dummy variable equal to one if the correlation of returns between

the industry portfolio and the �rms engaged in the corporate scandal for the year

preceding the �le of the class action suit falls within the 51st and 100th percentile (High

correlation), and zero otherwise (low correlation).Table 9 presents results for the peers

group sorted by the degree of cash-�ows correlation with the relevant SCAS company.

Results are statistically strong and signi�cant at all levels and indicate that security

issuance opportunities are positively a¤ected by corporate events in the industry the

higher the degree of cash �ows similarity between the sued company and its peers. This

result is twice as strong for equity rather than debt suggesting that shareholders react

signi�cantly, reducing �nancing opportunities or increasing their cost for any company

in the same industry niche. .

INSERT TABLE 7 HERE

4.1.1 Negative issuance

Previous results have shown that both SCAS �rms and their peers have a lower level of

security issuance after the security class action �ling. Interestingly, this phenomenon

generates also cases of "negative issuance". Negative debt issuance can be often the

18

simple repayment of outstanding debt without any rollover. In such a case, assuming

that companies have a fairly stable short term �nancial structure, the negative issuance

pattern should be rather stable over the event window. Yet, if some extraordinary event

occurs, a¤ecting the company current and expected cash-�ows an abnormal negative

issuance pattern would be a signal of a debt restructuring process involving some degree

of debt cutting. Negative equity interpretation is less intuitive since book equity is a

permanent liability in a company�s balance sheet.

In table 8 we report �gures for a simple discrete analysis of the number of �rms for

which debt and equity issuances �gures were less or equal to zero during the [-2,+3]

years surrounding the event.

INSERT TABLE 8 HERE

Results show that after the �ling SCAS �rms retire and/or repurchase about 88%

more equity and 74% more debt. In the SCAS subsample, negative debt issuance may

be the result of debt repayment and cancellation due to restructuring taking place after

the suit has been �led. Agrawal and Cooper (2007) show in fact that immediately after

the scandal, most of the company change their top management and initiate profound

restructuring processes encompassing also debt renegotiation. This same interpretation

may apply to the equity �gures: most of the restructuring plans imply that large

dilutions for existing shareholders which result in negative changes in book equity and

retained earnings.

Surprisingly though, also company�s in the peer group show an increasing amount

of negative issuances. The di¤erences are strong and signi�cant both across samples

and time. This could also be interpreted as a contagion e¤ect, meaning that the �ling

of a SCAS in the industry decreased the opportunities for security o¤erings of its peers

around the event.

4.2 Contagion e¤ect on stock prices

Previous results build on the argument that corporate scandals convey information

about a �rm�s cash �ows, accounting or management practices that investors may con-

sider more an industry-wide phenomenon rather than a company-speci�c, isolated event.

This inference generate a negative e¤ect on stock prices of both SCAS �rms and their

peers. First evidence of this e¤ect and the spillovers on competitors has been provided

by Gande and Lewis (2009). Yet, in their study there is no evidence of any di¤erential

19

e¤ect on stock prices conditional on capital structure and �nancial characteristics of

the industry which may arguably impact on the magnitude of investors�response to

scandals at the intra-industry level. In this section we begin by testing general conta-

gion e¤ects on stock prices following a SCAS announcement and control for correlation

of returns, leverage and type of allegation. Adopting an event study methodology, we

examine abnormal returns on a a set of short-term windows (2-day, 3-days, 5-days,

11-days, 31-days and 41 days bracketing the event). We choose to restrict our study to

short-term windows as working with longer horizon could introduce noise in our results.

The speci�c bracketings are constructed to capture quasi-instantaneous and anticipated

or delayed stock price reactions to the �ling announcement.

Following MacKinley (1997) and Khotari and Warner (2006) we estimate a standard

market model for every company sued in a security class action i 2 I where I is the setof SCAS �rms through the following equation:

Rit = �i + �iRmt + "it (2)

where Rit is the predicted normal rate of return of the security i at time � , Rmt is the

value-weighted return of the S&P500 index, �iand �iare the parameters to be estimated,

and "itis the error term of the regression. The distributions of stock returns were

assumed to be jointly multivariate normal and independently and identically distributed

over time, thus E("it) = 0 and var("it) = �2"i. Equation (2) is estimated with daily

observations over the period (� � 250; � � 50) preceding the �ling of the class action suitat � = 0. Using the market model estimated parameters, we compute daily abnormal

returns for both �rms being sued and their peers. The daily abnormal return of a

security is computed by subtracting the predicted normal return from the actual return

for each day in the event window. LettingdARi� be the abnormal returns for �rm i at

time � the sample abnormal return is:

dARi� = Ri� � (�̂i + �̂iRm� ) (3)

wheredARi� is the abnormal rate of return of the security i in the event window, Ritis the actual rate of return of the security i in the event window, and (�̂i+ �̂iRm� ) is the

expected normal rate of return of the security i in the event window calculated using the

market model. The aggregation of abnormal returns is bi-dimensional: through time

and across securities and follows this process. We �rst compute the average abnormal

returns for all i as:

20

AR� =1

I

IXi=1

dARi� (4)

For any security i, we then compute the cumulative abnormal return from � 1to � 2as the sum of the abnormal returns within that event window:

[CARi(� 1; � 2) =�2X�=�1

dARi� (5)

The average abnormal returns, across the I SCAS companies, are aggregated over

the event window as follows:

CAR(� 1; � 2) =

�2X�=�1

AR� (6)

Finally, we tested whether the cumulative abnormal returns were statistically dif-

ferent from zero using:

�1 =CAR(� 1; � 2)

var(CAR(� 1; � 2))1=2� N(0; 1) (7)

This distributional result is asymptotic with respect to the number of securities N

and the length of the estimation window (200 days in this study).

We follow the same procedure for calculating AR and CAR for the 4-digit SIC code

peer group of the sued company, excluding the latter from the estimations.

4.2.1 Event study results

Our results show that �rms engaged in a security class action suit su¤er a negative

�and somewhat signi�cant- cumulative abnormal return of -.0168 in the [-10,+10] event

window. In the [-1, 0] and [-5,+5] event windows we obtain a -.0027 and -.0093 CAR

respectively but with no statistical signi�cance. When comparing our results with

those of previous literature,5 ours seem to be much milder and less signi�cant for the

corporate scandal �rms. The di¤erence in our results could be explained by the fact

5Lang and Stulz (1992) obtain a -.2166 and a -.2825 CAR in the [-1,0] and [-5,+5] event windowsrespectively.Ferris et al. (1997) obtain a -.1755 and a -.2680 CAR in the [-1, 0] and [-5,+5] event windowsrespectively.

21

that security class action suits are not a proxy for chapter 11, and that the other two

papers mentioned here are related to bankruptcy announcements. Thus, while our

results of negative stock price reaction to corporate scandals are not as strong as those

presented in previous somewhat related literature, this is to be expected just by the

de�nition of what is a corporate scandal (bankruptcy announcements are more prone

to stock price declines because �rms have greater probabilities to close operations and

thus their equity value sharply declines).

When evaluating the stock price contagion generated by corporate scandals, our

results show a negative �and statistically signi�cant- cumulative abnormal return of

-.20% (p value=0.057) and of -.65% (p-value=0.049) in the [-1,0] and [-5,5] event win-

dows respectively. These results allow us to test our hypothesis that �rms engaged in

corporate scandals do a¤ect their industry in terms of stock returns. The engagement of

a participant of the industry in a security class action suit negatively a¤ects the returns

of its peers. Although we are not studying bankruptcy announcements, our results are

also consistent with those of Ferris at al. in windows [-1,0] and [-5,+5] (-.23% and -.06%

respectively). At the same time, for the event window [-5,5] our results are bigger in

term of the contagion e¤ect, meaning that in this window a security class action suit

a¤ect more the industry than a bankruptcy announcement of a large �rm.

Overall, our results indicate that SCAS �rms su¤er a negative stock price decline

in a wider event window than their peers do, which could be interpreted as a type of

anticipation e¤ect of the scandal. Peers seem to be a¤ected in a smaller window (nearer

the event). This could mean that investors do not anticipate the consequences of the

SCAS �ling in the industry, or just that there is a lag in the incorporation of the data

at the industry level. Table 9 shows the results of the event study in the di¤erent event

windows.

INSERT TABLE 9 HERE

Interaction e¤ect with industry characteristics

The degree of similarity among the �rms�cash �ows (measured as the correlation of

returns between the industry portfolio and the �rms engaged in the corporate scandal

for the year preceding the �le of the class action suit) should intensify the extent of the

contagion e¤ect. This intensi�cation of the contagion e¤ect is due to the fact that highly

similar �rms are likely to have investments with similar cash �ow characteristics and

similar risk exposures. We hypothesized that ceteris paribus, the �negativecontagion

22

e¤ect on stock prices is greater for industries in which competitors have similar cash

�ows to those of the �rm involved in the corporate scandal. To test our hypothesis

we constructed a dummy variable for returns correlation as in the previous section (to

split the sample among �rms that had high and low correlation of returns with their

industry). Our results indicate this control variable do intensify the contagion e¤ect

on stock returns. We �nd that �rms with high correlation of stock returns generate a

negative �and statistically signi�cant- contagion e¤ect of -1.03% (p value=0.035) in the

[-5,5] event windows. In the overall sample the contagion e¤ect for the same window was

only -0.65% (p value=0.049). As Table 10 depicts, we do not �nd any other signi�cant

changes in the remaining event windows.

INSERT TABLE 10 HERE

Furthermore, we also stated that if the corporate scandal conveys negative infor-

mation about the industry, it is expected that the percentage fall in equity of the

�rm�s competitors increase with their leverage. We expected that ceteris paribus, the

�negativecontagion e¤ect on stock prices is greater for highly leveraged industries. To

test our hypothesis we created a dummy variable equal to one if the industry leverage

mean was within the 51th and 100th percentile of the sample in the year of the �ling

(for both book and market leverage). For us, a dummy equal to one meant that the �rm

was in a highly leveraged industry. We then replicated the contagion e¤ect analysis as

with the �rst control.

Our results indicate that high book leverage has a considerable e¤ect on the [-5,5]

and [-10,10] windows for SCAS stock price reaction. In the overall sample, the stock

price reaction of troubled �rms was of 0.93% and �1.68% (and not signi�cant) on the

[-5,5] and [-10,10] event windows respectively. Instead, studying only highly leveraged

industries, we �nd a statistically signi�cant stock price reaction of -3.04% and �4.50% on

the [-5,5] and [-10,10] event windows respectively. The contagion e¤ect in this analysis

results not signi�cant. Thus, by splitting the sample according to leverage (high and

low) we loose the signi�cance of the contagion e¤ect on stock price. Table 11 shows

the results of the contagion e¤ect for the high and low leverage samples in two event

windows.


23

Accounting vs. non-accounting allegations

Given that the SCAS database provided us with the classi�cation of each �ling into

accounting and non-accounting allegations we divided the sample and explored if the

stock return decline and the contagion e¤ect in these two subsamples were di¤erent. We

expected both phenomenon (stock return decline and contagion e¤ect) to remain con-

stant but to our surprise this was not the case. We found that the negative stock return

reaction of �rms engaged in a security class action suit with non-accounting allegations

is strongly signi�cant for the [-15,+15] and [-20,+20] event windows (-3.65% and -3.69%

respectively). At the same time, non-accounting SCAS generate a statistically signif-

icant -1.09% contagion e¤ect in the [-5,+5] window. Again we see that the contagion

e¤ect exists in the days near the �ling, while the decrease in returns of the SCAS �rms

are visible in wider windows. When studying the contagion e¤ect of this subsample the

results are stronger and more signi�cant than those obtained with the combined sample.

We also found that SCAS with accounting allegations do not su¤er �statistically signi�cant-

stock price reactions and that they do not generate a statistically signi�cant contagion

e¤ect in their industry. This result is surprising because one would expect any type of

SCAS �ling to at least generate a negative return after the scandal is unveiled.

To justify the existence of the contagion e¤ect uniquely in the non-accounting allega-

tions subsample, it is possible to argue that these type of allegations have an unknown

outcome (predicting what will happen after the lawsuit is di¢ cult for investors), while

the accounting allegations are easily interpreted and investors can foresee what will

happen after the suit. Table 12 depicts the results obtained in each subsample.


5 Robustness tests

To test the validity of our results we performed a series of robustness checks. We split

our original database according to di¤erent criteria and then replicate our analyses.

� Type of allegations

We split our sample according to the type of allegation of the security class action

suit. The database had an accounting and non-accounting allegations classi�cation.

24

The results of the analysis of leverage, �nancing pattern and issuing trend of SCAS and

peers remained unchanged.

� Chapter 11 �lings

We matched our database with LoPucki�s Bankruptcy Research Database at UCLA.

We created a dummy variable equal to one if the troubled �rm �led for chapter 11 in

the period 2 years before or after the security class action suit �ling. The results of

the analysis of leverage, �nancing pattern and issuing trend of SCAS and peers remain

unchanged. Overall, �rms that eventually �led for chapter 11 do not di¤er from those

who didn�t.

� Sentiment of the �ling year

Using Baker and Wurgler�s (2006) sentiment index we created a dummy equal to one

if the sentiment of the SCAS �ling year was greater than zero, zero otherwise. Again,

the results of the analysis of leverage, �nancing pattern and issuing trend of SCAS and

peers remain unchanged. The market sentiment of the �ling year doesn�t a¤ect the

results of our analysis.

� Size

To control for size we used two di¤erent measures. First, using total assets, we

created a dummy variable equal to one if the SCAS was within the 51th and 100th

percentile of the SCAS �rm sample (zero otherwise). This �rst measure attempted to

catch a size e¤ect within the SCAS sample. Second, and again using total assets, we

created a dummy variable equal to one if the SCAS was within the 51th and 100th

percentile of its industry (zero otherwise). This second measure attempted to catch a

size e¤ect within the industry. We repeated all the analyses splitting the sample by size

and obtained basically the same results as before.

6 Conclusions

This paper presented the analysis of the ex-ante and ex-post �nancing pattern of �rms

engaged in security class action suits as well as the contagion e¤ect that these scandals

generate in the industry in terms of stock returns and �nancing pattern as well. In the

�rst part of the paper we tested that ex-ante, �rms engaged in a corporate scandal have

25

a greater amount of security o¤erings compared to its industry average. On average, two

years before the event, �rms engaged in a corporate scandal issue 435% more securities

than their peer sample. The year before the event they issue 252% more securities

than their peer sample. And the year were the event happens (the �ling of a SCAS)

the di¤erence of means is 101% more and still statistically signi�cant. In line with

the predictions of the market timing hypothesis, we tested that ex-ante, �rms engaged

in a corporate scandal made greater use of equity �nancing compared to its industry

average. Two years before the event, �rms engaged in a corporate scandal issue on

average 670% more equity than their peer sample. This di¤erence alone is more than 6

times higher than the average amount of equity o¤erings that their peers�sample issued.

This behavior is persistent in the year preceding the event and in the year of the event

itself. Contrary to what the theory would predict, ex-ante �rms engaged in a corporate

scandal had higher levels of leverage. We �nd book leverage of corporate scandal �rms

to be higher in all the periods (and this di¤erence is statistically signi�cant in all years).

On average, during the �ve periods that we studied, �rms engaged in a SCAS had 1.42

times the book leverage that their peers did. For the market leverage analysis we �nd

the same results. Ex-post, we found that �rms engaged in a security class action suit

�nance themselves in the same manner as the mean of their industry. Ex-post, debt

issuances are lower than equity issuances for both samples, and this e¤ect is even more

pronounced for SCAS �rms. The weighted average peers�sample show a more steady

issuing behavior, and both equity and debt issues decrease but not as sharply as in

the SCAS sample. Ex-post the di¤erence in the means of debt and equity issuances of

SCAS �rms and their peers is not statistically signi�cant, which lead us to conclude

that both means are the same for both samples.

In the second part of the paper we analyzed the existence of a trend in the security

o¤ering pattern that can be interpreted as a contagion e¤ect in the �nancing pattern of

the industry. The results show that surrounding the event, both the corporate scandals

�rms and the weighted average matched sample of competitors have a signi�cant and

decreasing trend in both equity and debt issuances. We conclude that the �ling of a

security class action suit do a¤ect the opportunity that �rms within an industry have

to o¤er securities to the market. We control for returns correlation and �nd that the

more similar the �rm is to its industry, the lower the contagion e¤ect on their issuing

pattern.

Finally we tested the existence of a contagion e¤ect in the industry�s stock returns due

to the �ling of a security class action suit. Our results show a negative �and statistically

26

signi�cant- cumulative abnormal return of -0.20% (p value=0.057) and of -0.65% (p-

value=0.049) in the [-1,0] and [-5,5] event windows respectively. We also �nd that

the contagion e¤ect is mainly generated by �rms engaged in security class action suits

with the non-accounting allegations. SCAS with accounting allegations do not su¤er

�statistically signi�cant- cumulative abnormal returns and that they do not generate a

statistically signi�cant contagion e¤ect in their industry.

27

References

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

of Law and Economics 48, 371-406.

Agrawal, A, Cooper T., 2007, Corporate governance consequences of accounting

scandals: Evidence from top management, CFO and Auditor Turnover, 2nd Annual

Conference on Empirical Legal Studies Paper. AFA 2009 San Francisco Meetings Paper.

Akhigbea A., Martin A. D., Whytec M., 2005, Contagion e¤ects of the world�s

largest bankruptcy: the case of WorldCom The Quarterly Review of Economics and

Finance

Volume 45, Issue 1, February 2005, Pages 48-64

Baker, M,WurglerJ., 2002, Market Timing and Capital Structure, The Journal of

Finance 57, 1-32.

Baker, M,WurglerJ., 2006, Investor sentiment and the cross-section of stock returns,

The Journal of Finance 61, 1645-1680.

Bay Networks (1997) Case docket number: 97-CV-728, Court: N.D. California,

Filing date 02/28/1997.

Brewer E, Jackson W.E., 2002, Inter-Industry Contagion and the Competitive Ef-

fects of Financial Distress Announcements: Evidence from Commercial Banks and Life

Insurance Companies, Federal Reserve Bank of Chicago, Working paper No. 2002-23.

Cisco (2001), Case docket number: 01-CV-20418, Court: N.D. California, Filing

date 4/20/2001.

De Bondt W. F. M., Thaler R., 1985, Does the Stock Market Overreact?, The

Journal of Finance 40, 3, 793-805.

De Bondt W. F. M., Thaler R., 1990, Do Security Analysts Overreact?, The Amer-

ican Economic Review, 80, 2, 52-57.

Dyck, I. J. A., Morse, A., Zingales L., 2007, Who blows the whistle on corporate

fraud?, Chicago GSB Research Paper No. 08-22. CRSP Working Paper No. 618.

Fama, E. F., French K. R., 1997, Industry Costs of Equity, Journal of Financial

Economics 43, 153-194.

Fama, E. F., French K. R., 2002, Testing Trade-O¤ and Pecking Order predictions

about Dividends and Debt, Review of Financial Studies 15, 1-33.

Federal reserve Board (2008), The October 2008 Senior Loan O¢ cer Opinion Survey

on Bank Lending Practices, October, Board of Governors of the Federal Reserve System,

Washingto, USA, http://www.federalreserve.gov/boarddocs/surveys.

28

Ferris, S. P., Jayaraman, N., Makhija, A., 1997, The response of competitors to

announcements of bankruptcy: An empirical examination of contagion and competitive

e¤ects, Journal of Corporate �nance 3, 367-395.

Gande, A. and Lewis, C. M.,Shareholder Initiated Class Action Lawsuits: Share-

holder Wealth E¤ects and Industry Spillovers. Journal of Financial and Quantitative

Analysis (JFQA), Forthcoming.

Gieseke K., Correlated default with incomplete information Journal of Banking &

Finance Volume 28, Issue 7, July 2004.

Gleason, C., W.B. Johnson and N.T. Jenkins. 2008. Financial statement credibility:

the contagion e¤ects of accounting restatements. The Accounting Review 81 (83-110).

Graham, J.R., 2000, How big are tax bene�ts of Debt?, Journal of Finance 55,

1901-1941.

Graham, J. R., Harvey C., 2001, How do CFOs make capital budgeting and capital

structure decisions?, Journal of Financial Economics 60, 187-243.

Hovakimian, A., Opler, T., Titman S., 2001, The Debt-Equity choice, The Journal

of Financial and Quantitative Analysis 36, 1-24.

Ivashina, V. and Scharfstein, D. S., 2009, Bank Lending During the Financial Crisis

of 2008. Available at SSRN: http://ssrn.com/abstract=1297337

Kanas, A., 2005, Pure Contagion E¤ects in International Banking: The case of

BCCI�s Failure. Journal of Applied Economics 8, 101-123.

Khotari, S.P., Warner J. P.�2006, Econometrics of event studies, in B. Espen Eckbo,

ed: Handbook of Corporate Finance: Empirical Corporate Finance, Handbooks in

Finance Series (Elsevier/North-Holland).

Lang, L., Stulz, R., 1992, Contagion and competitive intra-industry e¤ects of bank-

ruptcy announcements: An empirical analysis, Journal of Financial Economics 32, 45-

60.

MacKinlay, A. C., 1997, Event studies in economics and �nance, Journal of Eco-

nomic Literature, 35, 13-39.

Markham, J., 2006, A Financial History of Modern placecountry-regionU.S. Corpo-

rate Scandals: from Enron to Reform (M.E. Sharpe.London, England)

Mikkelson, W. H., Partch M. M., 1985, Stock price e¤ects and costs of secondary

distributions, Journal of Financial Economics 14, 165-194.

Miller, M. H., 1977, Debt and taxes, Journal of Finance 32, 261-275.

Myers, S. C., 1984, The Capital Structure Puzzle, The Journal of Finance 39, 575-

592.

29

Myers S. C., 2001, Capital Structure, The Journal of Economic Perspectives 15,

81-102.

Palmrose, Z., Richardson, V.J., Scholz S., 2004, Determinants of market reactions

to restatement announcements, Journal of Accounting and Economics 37, 59-89.

Rao, S., Brooke Hamilton J., 1996, The E¤ect of Published Reports of Unethical

Conduct on Stock Prices, Journal of Business Ethics 15, 1321-1330.

Ramnath, S. 2002. Investor and analyst reactions to earnings announcements of

related �rms: an empirical analysis. Journal of Accounting Research 40(5): 1351-1376.

SuperGen (2003), Case docket number: 03-CV-1576, Court: N.D. California, Filing

date 04/14/2003.

Theocharides, G., 2007, Contagion: Evidence from the Bond Market. Available at

SSRN: http://ssrn.com/abstract=811548

Welch, I., 2004, Capital Structure and Stock returns, Journal of Political Economy

112, 106-131.

30

Table IYearly distribution of events and

Fama & French industriesThis table reports the distribution of security class action suit cases by filingyear, from January 1996 to December 2006. Fama & French industries wereassigned using 4digit sic codes and the classification provided in their paperof 1997.

Filing year (SCAS) N Fama French industries1996 47 191997 66 221998 88 241999 75 222000 87 212001 81 202002 90 252003 63 212004 82 212005 67 242006 47 18Total 793

Table IIAmount of cases studies by event year, type of allegation

chapter 11 filingThis table reports the distribution of security class action suit cases by event year. The event year(t=0) is defined as the year in which the security class action suit was filed against the firm. Thepercentages of cases according to the type of allegation (accounting and nonaccounting), and tothe filing of chapter 11 (2 years after or before the filing) are also presented.

Year (event) NAccountingallegations

Nonaccountingallegations

Filed forChapter 11in t=[2,2]

Didn't file forChapter 11 in

t=[2,2]t=3 735 55.5% 44.5% 8.5% 91.5%t=2 754 55.0% 45.0% 8.8% 91.2%t=1 717 54.3% 45.7% 7.6% 92.4%t=0 627 53.4% 46.6% 5.4% 94.6%t=1 551 53.4% 46.6% 4.4% 95.6%t=2 458 54.8% 45.2% 4.2% 95.8%t=3 366 54.8% 45.2% 4.0% 96.0%

31

Table IIIMean security offerings by event year

This table reports the –weighted mean security offerings of firms engaged in a corporate scandal (proxied bythe filing of a security class action suit), and that of a valueweighted portfolio of the remaining firms with thesame 4digit sic code (by event year). The event year (t=0) is defined as the year in which the security classaction suit was filed against the firm. The amount of total –yearly security offerings is measured as the sum ofdebt issuances and book equity issuances. Debt issuances are measured as the change in total assets minuschange in book equity divided by total assets. Book equity issuances are measured as the change in book equityminus the change in balance sheet retained earnings, divided by total assets. The last two columns of the tablepresent the results of the one and twotailed meandifference tests.

t Variable Obs Mean Mean(diff) Pr(|T|>|t|)(1) Pr(T>t) (2)

2 Security offerings SCAS 629 0.576302 Security offerings PEERS 629 0.10765 0.46865 0.0002 (***) 0.0001 (***)1 Security offerings SCAS 638 0.389631 Security offerings PEERS 638 0.11058 0.27905 0.0000 (***) 0.0000 (***)0 Security offerings SCAS 553 0.184030 Security offerings PEERS 553 0.09158 0.09245 0.0002 (***) 0.0001 (***)1 Security offerings SCAS 483 0.042061 Security offerings PEERS 483 0.07192 0.02986 0.4090 0.79552 Security offerings SCAS 403 0.064282 Security offerings PEERS 403 0.06875 0.00447 0.8840 0.55803 Security offerings SCAS 322 0.074223 Security offerings PEERS 322 0.06732 0.00690 0.9275 0.4637

(1)Ha: mean(diff) ≠ 0(2)Ha: mean(diff) > 0

32

Table IVMean debt and equity issuances by event year

This table reports mean debt and equity issuances of firms engaged in a corporate scandal (proxied bythe filing of a security class action suit), and a valueweighted portfolio of the remaining firms with thesame 4digit sic code (by event year). The event year (t=0) is defined as the year in which the securityclass action suit was filed against the firm. Debt issuances are measured as the change in total assetsminus change in book equity divided by total assets. Book equity issuances are measured as the changein book equity minus the change in balance sheet retained earnings, divided by total assets. The lastthree columns of the table present the results of the one and twotailed meandifference tests.

t Variable Obs MeanMean(diff) Pr(|T|>|t|)(1) Pr(T>t)(2) Pr(T<t)(3)

2 Equity issuances SCAS 629 0.538372 Equity issuances PEERS 629 0.06988 0.46849 0.0180(*) 0.0090 (**) 0.99101 Equity issuances SCAS 638 0.308941 Equity issuances PEERS 638 0.07248 0.23647 0.0000(***) 0.0000(***) 1.00000 Equity issuances SCAS 553 0.147920 Equity issuances PEERS 553 0.06199 0.08593 0.0000(***) 0.0000(***) 1.00001 Equity issuances SCAS 483 0.074331 Equity issuances PEERS 483 0.04548 0.02885 0.2560 0.1280 0.87202 Equity issuances SCAS 403 0.089342 Equity issuances PEERS 403 0.04590 0.04344 0.0886 0.0443 (*) 0.95573 Equity issuances SCAS 322 0.081923 Equity issuances PEERS 322 0.04269 0.03923 0.0911 0.0455 (*) 0.9545

2 Debt issuances SCAS 632 0.037732 Debt issuances PEERS 632 0.039880.00215 0.9901 0.5050 0.49501 Debt issuances SCAS 640 0.081221 Debt issuances PEERS 640 0.03920 0.04202 0.0077(**) 0.0038 (**) 0.99620 Debt issuances SCAS 555 0.036060 Debt issuances PEERS 555 0.03158 0.00448 0.7657 0.3829 0.61711 Debt issuances SCAS 485 0.032591 Debt issuances PEERS 485 0.029340.06193 0.0026(**) 0.9987 0.0013 (***)2 Debt issuances SCAS 406 0.025192 Debt issuances PEERS 406 0.023680.04887 0.0688 0.9656 0.0344 (*)3 Debt issuances SCAS 325 0.010693 Debt issuances PEERS 325 0.024720.03541 0.5677 0.7161 0.2839

(1)Ha: mean(diff) ≠ 0(2)Ha: mean(diff) > 0(3)Ha: mean(diff) < 0

33

Table VMean market and book leverage by event year

This table reports the mean market book leverage of firms engaged in a corporate scandal (proxied by the filing of asecurity class action suit), and for a the valueweighted portfolio of firms with the same 4digit sic code by event year,excluding the SCAS firm. The event year (t=0) is defined as the year in which the security class action suit was filedagainst the firm. Market leverage is measured as book debt divided by the sum of total assets minus book equity plusmarket equity. Book leverage is measured as book debt divided by total assets. The last two columns of the tablepresent the results of the one and twotailed meandifference tests.

PANEL A: MARKET LEVERAGE

t Variable Obs Mean Mean(diff) Pr(|T|>|t|)(1) Pr(T>t)(2)

2 Market leverage SCAS 607 0.236362 Market leverage PEERS 607 0.23316 0.00320 0.7186 0.35931 Market leverage SCAS 633 0.251901 Market leverage PEERS 633 0.23412 0.01778 0.0496 (*) 0.0248 (*)0 Market leverage SCAS 570 0.372130 Market leverage PEERS 570 0.23748 0.13465 0.0000 (***) 0.0000 (***)1 Market leverage SCAS 498 0.381131 Market leverage PEERS 498 0.23873 0.14240 0.0000 (***) 0.0000 (***)2 Market leverage SCAS 417 0.359922 Market leverage PEERS 417 0.23091 0.12901 0.0000 (***) 0.0000 (***)3 Market leverage SCAS 327 0.364733 Market leverage PEERS 327 0.23017 0.13456 0.0000 (***) 0.0000 (***)

PANEL B: BOOK LEVERAGE

t Variable Obs Mean Mean(diff) Pr(|T|>|t|)(1) Pr(T>t)(2)

2 Book leverage SCAS 706 0.653272 Book leverage PEERS 706 0.42999 0.22328 0.0265 (*) 0.0132 (*)1 Book leverage SCAS 660 0.483291 Book leverage PEERS 660 0.42294 0.06035 0.1058 0.0529 (*)0 Book leverage SCAS 572 0.526150 Book leverage PEERS 572 0.42488 0.10127 0.0000 (***) 0.0000 (***)1 Book leverage SCAS 501 0.625561 Book leverage PEERS 501 0.43461 0.19096 0.0068 (**) 0.0034 (**)2 Book leverage SCAS 420 0.581462 Book leverage PEERS 420 0.42081 0.16065 0.0001 (***) 0.0001 (***)3 Book leverage SCAS 330 0.754973 Book leverage PEERS 330 0.41863 0.33634 0.0369 (*) 0.0185 (*)

(1)Ha: mean(diff) ≠ 0(2)Ha: mean(diff) > 0

34

Figure ISecurity offering trend analysis

This graph reports the results of the regression: Yjt=αj+βj(T)+εjt; where, Yjt are totalsecurity issuances, T is a trend variable that ranges from {1,5}, and eit is the errorterm of the regression. The amount of total –yearly security offerings is measured asthe sum of debt issuances and book equity issuances.

0.1

.2.3

.4.5

Xb

4 2 0 2 4t

SCAS PEERS

35

0.1

.2.3

.4X

b

4 2 0 2 4t

Avg Equity issuances SCAS Avg Debt issuances SCASAvg Equity issuances PEERS Avg Debt issuances PEERS

Figure IIEquity and debt issuance trend analysis

This graph reports the results of the regression: Yjt=αj+βj(T)+εjt; where, Yjt are totaleither equity or debt issuances, T is a trend variable that ranges from {1,5}, and eit is theerror term of the regression.

36

Table VISecurity offering trend analysis

This table reports the results of the regression: Yjt=αj+βj(T)+εjt; where, Yjt are eitherequity, debt or total security issuances, T is a trend variable that ranges from {1,6}, and eit

is the error term of the regression. The amount of total –yearly security offerings ismeasured as the sum of debt issuances and book equity issuances. Debt issuances aremeasured as the change in total assets minus change in book equity divided by total assets.Book equity issuances are measured as the change in book equity minus the change inbalance sheet retained earnings, divided by total assets.

Sample: SCAS firms

Dependent variableEquity

issuancesDebt

issuances

Totalsecurityofferings

Intercept 0.509521 (***) 0.082482 0.586441 (***)P>|z| 0 0.313 0

Trend coeff. 0.0783101 (***) 0.018758 0.0927902 (***)P>|z| 0.001 0.392 0

N 721 724 721Wald chisquare 10.67 (***) 0.73 32.05 (***)P>chisquare (0.001) (0.392) (0.000)

Sample: PEERS

Dependent variableEquity

issuancesDebt

issuances

Totalsecurityofferings

Intercept 0.078071 (***) 0.044634 (***) 0.120455 (***)P>|z| 0 0 0

Trend coeff. 0.0057357 (***) 0.0038771 (***) 0.0092945 (***)P>|z| 0 0 0

N 782 782 782Wald chisquare 47.75 (***) 21.96 (***) 57.18 (***)P>chisquare (0.000) (0.000) (0.000)

37

Table VIIContagion effect analysis according to correlation of stock returnsThis table reports the results of the regression: Yjt=αj+βj(T)+εjt; where, Yjt are eitherequity, debt or total security issuances, T is a trend variable that ranges from {1,6},and eit is the error term of the regression. The amount of total –yearly securityofferings is measured as the sum of debt issuances and book equity issuances. Debtissuances are measured as the change in total assets minus change in book equitydivided by total assets. Book equity issuances are measured as the change in bookequity minus the change in balance sheet retained earnings, divided by total assets.The high/low correlation of returns dummy is defined as: 0 if correlations of returns(between SCAS and PEERS in the year preceding the filing) lies within the [150th]percentile and 1 if it lies within the [51100]th percentile in the year before the filing ofthe SCAS.

PEERSCorrelation of returns High LowDependent variable: Total security offeringsIntercept 0.138329 (***) 0.100738 (***)

P>|z| 0.000 0.000Trend 0.0115997 (***) 0.0045924 (***)

P>|z| 0.000 0.000N 197 198Wald chisquare 3746.40 705.71P>chisquare 0.000 0.000

Correlation of returns High LowDependent variable: Equity issuancesIntercept 0.096095 (***) 0.066687 (***)

P>|z| 0.000 0.000Trend 0.0090866 (***) 0.0028100 (***)


Correlation of returns High LowDependent variable: Debt issuancesIntercept 0.049363 (***) 0.035994 (***)

P>|z| 0.000 0.000Trend 0.0045173 (***) 0.0012588 (***)


38

Table VIIISecurity offering trend analysis –Discrete analysis

This table reports the results of negative debt and equity issuances in the different event years. Wecounted the amount of cases where debt/equity issuances were less or equal than zero.

Equity issuances (discrete analysis)SCAS PEERS

t Obs Eq_iss<=0 % Eq_iss<=0 Obs Eq_iss<=0 % Eq_iss<=02 629 90 14.31% 754 91 12.1%1 638 95 14.89% 717 103 14.4%0 553 145 26.22% 627 99 15.8%1 483 135 27.95% 551 108 19.6%2 403 105 26.05% 458 97 21.2%3 322 96 29.81% 366 73 19.9%

Debt issuances (discrete analysis)SCAS PEERS

t Obs Debt_iss<=0 % Debt_iss<=0 Obs Debt_iss<=0 % Debt_iss<=02 632 175 27.69% 754 154 20.4%1 640 174 27.19% 717 135 18.8%0 555 211 38.02% 627 153 24.4%1 485 244 50.31% 551 128 23.2%2 406 217 53.45% 458 109 23.8%3 325 159 48.92% 366 92 25.1%

39

Table XContagion effect analysis by event window

This table reports the cumulative abnormal returns of firms engaged in a corporate scandal(proxied by the filing of a security class action suit), and a valueweighted portfolio of theremaining firms with the same 4digit sic code (by event year). The event year (t=0) is definedas the year in which the security class action suit was filed against the firm. The dailyabnormal return of a security is computed by subtracting the predicted normal return(estimated using the market model) from the actual return for each day in the event window.

Day/windowrelative toSCAS filing Reaction of SCAS firms Reaction of PEERS

N AR/CAR t P>|t| N AR/CAR t P>|t|

10 522 0.0006784 0.33 0.741 527 0.000399 0.46 0.6479 522 0.0009623 0.44 0.662 527 0.0001764 0.21 0.838 522 0.0010429 0.42 0.676 527 0.0001714 0.19 0.8487 522 0.0037979 1.78 0.076 527 0.0004275 0.51 0.6126 522 0.0003093 0.14 0.892 527 0.0009647 1.13 0.2595 522 0.0001869 0.07 0.945 527 0.0018305 2.15 0.0324 522 0.0011983 0.62 0.535 527 0.0000035 0 0.9973 522 0.0014929 0.69 0.491 527 0.0008165 0.95 0.3452 522 0.0016338 0.68 0.499 527 0.0012778 1.54 0.1241 522 0.0007261 0.34 0.735 527 0.0023747 3.28 0.001(**)0 522 0.0034582 1.22 0.223 527 0.0002952 0.37 0.7131 522 0.0017424 0.73 0.465 527 0.0001036 0.13 0.8962 522 0.0007818 0.36 0.718 527 0.0006204 0.75 0.4563 522 0.0027343 1.16 0.246 527 0.0018219 2.06 0.04 (*)4 522 0.0036303 1.66 0.098 527 0.0006182 0.66 0.5125 522 0.00227 0.9 0.369 527 0.0016159 1.45 0.1496 521 0.0018141 0.74 0.458 527 0.0015726 1.82 0.077 522 0.0020588 0.75 0.454 527 0.0006809 0.79 0.4328 522 0.0032792 1.45 0.149 527 0.0000998 0.08 0.9349 523 0.0018022 0.73 0.466 527 0.0003127 0.33 0.73810 523 0.0005527 0.29 0.769 527 0.0016415 1.22 0.224

[1,0] 527 0.0027062 0.81 0.42 527 0.0020795 1.91 0.057(*)[0,+1] 527 0.0051513 1.68 0.094 527 0.0001916 0.18 0.861[1,+1] 527 0.004432 1.21 0.225 527 0.0021831 1.67 0.095[5,+5] 527 0.0093642 1.28 0.202 527 0.0065921 1.98 0.049(*)

[10,+10] 527 0.0168449 1.72 0.086 527 0.0023214 0.48 0.632

40

Table XIContagion effect analysis according to correlation of stock returns

This table reports the cumulative abnormal returns of firms engaged in a corporate scandal(proxied by the filing of a security class action suit), and a valueweighted portfolio of theremaining firms with the same 4digit sic code (by event year). The event year (t=0) isdefined as the year in which the security class action suit was filed against the firm. The dailyabnormal return of a security is computed by subtracting the predicted normal return(estimated using the market model) from the actual return for each day in the event window.The high/low correlation of returns dummy is defined as: 0 if correlations of returns (betweenSCAS and PEERS in the year preceding the filing) lies within the [150th] percentile and 1 ifit lies within the [51100]th percentile in the year before the filing of the SCAS.

Sample A: HIGH correlation of returns

Day/windowrelative toSCAS filing Reaction of SCAS firms Reaction of PEERS

N AR/CAR t pvalue N AR/CAR t pvalue

[1,0] 264 0.0004374 0.11 0.916 264 0.001454 0.82 0.41[0,+1] 264 0.0019944 0.55 0.586 264 0.001227 0.72 0.472[1,+1] 264 0.0023665 0.51 0.612 264 0.0024984 1.19 0.234[5,+5] 264 0.0178436 1.85 0.065 264 0.0103162 2.12 0.035(*)

[10,+10] 264 0.0222732 1.92 0.056(*) 264 0.0024595 0.38 0.707

Sample B: LOW correlation of returns

Day/windowrelative toSCAS filing

Reaction of SCAS firms LOW correlation of returns

Reaction of PEERS LOWcorrelation of returns

N AR/CAR t pvalue N AR/CAR t pvalue

[1,0] 263 0.0049835 0.94 0.348 263 0.0027074 2.11 0.036[0,+1] 263 0.0083201 1.69 0.093 263 0.0016156 1.18 0.237[1,+1] 263 0.0065054 1.16 0.249 263 0.0018666 1.2 0.232[5,+5] 263 0.0008525 0.08 0.939 263 0.0028539 0.63 0.532

[10,+10] 263 0.011396 0.72 0.472 263 0.0021829 0.31 0.76

41

Table XIIContagion effect according to leverage

This table reports the cumulative abnormal returns of firms engaged in a corporatescandal (proxied by the filing of a security class action suit or a bankruptcyannouncement), and a valueweighted portfolio of the remaining firms with the same 4digit sic code. The sample is divided using a dummy variable equal to one if the SCASfirm was within the 51100 percentile of book leverage. Results of the market leverageanalysis are not presented but remain unchanged.

Day/windowrelative toSCAS filing

Reaction of SCAS firms HIGH leverage

Reaction of PEERS HIGHLeverage

N CAR t P>|t| N CAR t P>|t|

[1,0] 183 0.0002762 0.04 0.965 183 0.0012991 0.67 0.503[0,+1] 183 0.0014779 0.27 0.786 183 0.0005994 0.33 0.74[1,+1] 183 0.0028338 0.45 0.65 183 0.0018701 0.85 0.399[5,+5] 183 0.0304444 2.21 0.028(*) 183 0.0082169 1.53 0.128

[10,+10] 183 0.0450035 2.5 0.013(*) 183 0.0079747 1 0.319Day/windowrelative toSCAS filing

Reaction of SCAS firms LOWleverage

Reaction of PEERS LOWLeverage

N CAR t P>|t| N CAR t P>|t|

[1,0] 184 0.005416 0.97 0.336 184 0.0024428 1.63 0.105[0,+1] 184 0.0081828 1.67 0.096 184 0.0007883 0.47 0.639[1,+1] 184 0.0126792 2.05 0.042(*) 184 0.0015167 0.81 0.418[5,+5] 184 0.0119834 1.06 0.288 184 0.0042263 0.82 0.413

[10,+10] 184 0.0013997 0.09 0.925 184 0.0084321 1.05 0.293

42

Table XIIICumulative abnormal returns and contagion

effect by type of allegationThis table reports the cumulative abnormal returns of firms engaged in a corporatescandal (proxied by the filing of a security class action suit), and a valueweightedportfolio of the remaining firms with the same 4digit sic code. The results are dividedin two subsamples according to the type of allegations related to the security classaction suit (accounting and nonaccounting).

Accounting allegations

Reaction of SCAS stock Reaction of peers' stock

Event window N CAR t P>|t| N CAR t P>|t|

[1,0] 262 0.0071 1.27 0.207 262 0 1.19 0.234

[0,+1] 262 0.0074 1.53 0.126 262 0 0.2 0.845

[1,+1] 262 0.008 1.43 0.153 262 0 1.41 0.160

[2,+2] 262 0.0039 0.49 0.623 262 0 0.06 0.951

[5,+5] 262 0.0064 0.57 0.569 262 0 0.41 0.684

[10,+10] 261 0.0088 0.62 0.535 261 0 0.71 0.476

[15,+15] 261 0.0008 0.04 0.964 261 0 0.4 0.692[20,+20] 260 0.0031 0.16 0.872 260 0 0.1 0.916

NonAccounting allegations

Reaction of SCAS stock Reaction of peers' stock

Event window N CAR t P>|t| N CAR t P>|t|

[1,0] 268 0.0011 0.3 0.768 268 0 1.36 0.176

[0,+1] 267 0.0021 0.54 0.593 267 0 0.31 0.759

[1,+1] 267 0.0012 0.26 0.798 267 0 0.99 0.324

[2,+2] 267 0.0014 0.21 0.834 267 0 1.42 0.158

[5,+5] 266 0.012 1.25 0.211 266 0 2.51 0.013(*)

[10,+10] 266 0.0247 1.83 0.068 266 0 1.29 0.200[15,+15] 266 0.0365 2.26 0.025(*) 266 0 1.61 0.108[20,+20] 266 0.0369 1.92 0.057(*) 266 0 1.04 0.299

43

Date post:	20-Apr-2018
Category:	Documents
Upload:	trantram
View:	223 times
Download:	4 times

Corporate Scandals, Capital Structure and … Scandals...Corporate Scandals, Capital Structure and...

Documents