Cosofraudstudy2010

C o m m i t t e e o f S p o n s o r i n g O r g a n i z a t i o n s o f t h e T r e a d w a y C o m m i s s i o n

F R A U D U L e N T
F i N A N C i A L
R e P O R T i N g
1 9 9 8 - 2 0 0 7

An Analysis of U. S . P u b l i c Co m p a n i e s

By
Mark S. Beasley | Joseph V. Carcello
Dana R. Hermanson | Terry L. Neal

Authors Data Collection Team at the University of Tennessee
Mark S. Beasley Masters Students Doctoral Students
ERM Initiative Julie Anderton Allison Beck
North Carolina State University Ellen Brooks Michael Borth
Megan Copple Brian Carver
Joseph V. Carcello Miriam Franklin Matt Hoag (Co-Research Manager)
Corporate Governance Center Kristin Grieco Stacy Mastrolia (Co-Research Manager)
University of Tennessee Derick Hobby Leah Muriel
Anna Hooker Colin Reid
Dana R. Hermanson Candace Hooker
Corporate Governance Center Courtney Howell
Kennesaw State University Matthew O’Brien
Laura Quinton
Terry L. Neal Garrett Wistoff
Corporate Governance Center
University of Tennessee

COSO Representatives
David L. Landsittel Larry e. Rittenberg
COSO Chair COSO Chair Emeritus
University of Wisconsin – Madison

Mark S. Beasley Charles Landes
Deloitte Professor of Enterprise Risk Vice President Professional Standards and Services Group
Management and Professor American Institute of Certified Public Accountants
North Carolina State University
American Accounting Association

Richard Chambers Jeff Thomson
President President and CEO
The Institute of Internal Auditors Institute of Management Accountants

Marie Hollein
President and Chief Executive Officer
Financial Executives International

Preface
This project was commissioned by COSO, which is dedicated to providing thought leadership
through the development of comprehensive frameworks and guidance on enterprise risk
management, internal control, and fraud deterrence designed to improve organizational
performance and governance and to reduce the extent of fraud in organizations. COSO is a
private sector initiative, jointly sponsored and funded by the following organizations:

American Accounting Association (AAA)

American institute of Certified Public Accountants (AICPA)

Financial executives international (FEI)

institute of Management Accountants (IMA)
Committee of Sponsoring Organizations
of the Treadway Commission

The institute of internal Auditors (IIA) www.co s o.o rg

F R A U D U L e N T
F i N A N C i A L
R e P O R T i N g:
1 9 9 8 - 2 0 0 7

An Analysis o f U. S . P u b l i c Co m p a n i e s

Research Commissioned by

Co m m i t te e o f S p o n s o r i n g Organizations of the Treadway Commission

May 2010

Copyright © 2010, The Committee of Sponsoring Organizations of the Treadway Commission (COSO).
1 2 3 4 5 6 7 8 9 0 PIP 19876543210

All Rights Reserved. No part of this publication may be reproduced, redistributed, transmitted or displayed in any form or
by any means without written permission. For information regarding licensing and reprint permissions please contact the
American Institute of Certified Public Accountants, licensing and permissions agent for COSO copyrighted materials.
Direct all inquiries to copyright@aicpa.org or to AICPA, Attn: Manager, Rights and Permissions, 220 Leigh Farm Rd.,
Durham, NC 27707. Telephone inquiries may be directed to 888-777-7707.

www.co s o.o rg

FRAUDULeNT FiNANCiAL RePORTiNg: 1998-2007, An Analysis of U.S. Public Companies | iii

executive Summary
COSO sponsored this study, Fraudulent Financial Reporting: 1998-2007, to provide a comprehensive
analysis of fraudulent financial reporting occurrences investigated by the U.S. Securities and Exchange
Commission (SEC) between January 1998 and December 2007. This study updates our understanding
of fraud since COSO’s 1999 issuance of Fraudulent Financial Reporting: 1987-1997. Some of the more
critical findings of the present study are:
• There were 347 alleged cases of public company fraudulent financial reporting from 1998 to 2007, versus 294 cases from
1987 to 1997. Consistent with the high-profile frauds at Enron, WorldCom, etc., the dollar magnitude of fraudulent financial
reporting soared in the last decade, with total cumulative misstatement or misappropriation of nearly $120 billion across
300 fraud cases with available information (mean of nearly $400 million per case). This compares to a mean of $25 million
per sample fraud in COSO’s 1999 study. While the largest frauds of the early 2000s skewed the 1998-2007 total and mean
cumulative misstatement or misappropriation upward, the median fraud of $12.05 million in the present study also was
nearly three times larger than the median fraud of $4.1 million in the 1999 COSO study.

• The companies allegedly engaging in financial statement fraud had median assets and revenues just under $100 million.
These companies were much larger than fraud companies in the 1999 COSO study, which had median assets and
revenues under $16 million.

• The SEC named the CEO and/or CFO for some level of involvement in 89 percent of the fraud cases, up from 83 percent of
cases in 1987-1997. Within two years of the completion of the SEC’s investigation, about 20 percent of CEOs/CFOs had
been indicted and over 60 percent of those indicted were convicted.

• The most common fraud technique involved improper revenue recognition, followed by the overstatement of existing
assets or capitalization of expenses. Revenue frauds accounted for over 60 percent of the cases, versus 50 percent in
1987-1997.

• Relatively few differences in board of director characteristics existed between firms engaging in fraud and similar firms
not engaging in fraud. Also, in some instances, noted differences were in directions opposite of what might be expected.
These results suggest the importance of research on governance processes and the interaction of various governance
mechanisms.

• Twenty-six percent of the fraud firms changed auditors between the last clean financial statements and the last
fraudulent financial statements, whereas only 12 percent of no-fraud firms switched auditors during that same time. Sixty
percent of the fraud firms that changed auditors did so during the fraud period, while the remaining 40 percent changed
in the fiscal period just before the fraud began.

• Initial news in the press of an alleged fraud resulted in an average 16.7 percent abnormal stock price decline in the two
days surrounding the news announcement. In addition, news of an SEC or Department of Justice investigation resulted in
an average 7.3 percent abnormal stock price decline.

• Long-term negative consequences of fraud were apparent. Companies engaged in fraud often experienced bankruptcy,
delisting from a stock exchange, or material asset sales following discovery of fraud – at rates much higher than those
experienced by no-fraud firms.

Given the small number of frauds examined in this study that involve time periods subsequent to the issuance of the
Sarbanes-Oxley Act of 2002, further research will be needed once sufficient time has passed to allow for more observations
of SEC fraud investigations involving post-SOX time periods before any conclusions can be reached about the effectiveness
of that legislation in reducing instances of fraudulent financial reporting.

Our hope is that insights contained herein will encourage additional research to better understand organizational behaviors,
leadership dynamics, and other important aspects of the financial reporting process that may have an impact on fraud
prevention, deterrence, and detection.

We believe the results of this study will be useful to investors, regulators, stock exchanges, boards of directors, external
auditors, and other key stakeholders as they seek to prevent, deter, and detect fraudulent financial reporting.

www.co s o.o rg

FRAUDULeNT FiNANCiAL RePORTiNg: 1998-2007, An Analysis of U.S. Public Companies | V

Content Outline
Page
Section Description

i. introduction, Key Findings, and insights 1

ii. Description of Research Approach 7

iii. Detailed Analysis of instances of
Fraudulent Financial Reporting: 1998-2007 10

iV. Conclusion 45

V. Authors 47

www.co s o.o rg

FRAUDULeNT FiNANCiAL RePORTiNg: 1998-2007, An Analysis of U.S. Public Companies | 1

i. introduction, Key Findings, and insights
Fraudulent financial reporting can have significant consequences for the organization and its
stakeholders, as well as for public confidence in capital markets. Periodic high profile cases of
fraudulent financial reporting raise concerns about the credibility of the U.S. financial reporting process
and call into question the roles of management, auditors, regulators, and analysts, among others.

The Committee of Sponsoring Organizations of the Treadway This study builds on the previous COSO-sponsored study,
Commission (COSO) sponsored this research project to Fraudulent Financial Reporting: 1987-1997. Where possible,
provide an extensive updated analysis of financial statement we use or adapt language from the prior report, and we
fraud occurrences affecting U.S. public companies. In the compare key findings from this study to our findings in the
mid-1980s, the National Commission on Fraudulent Financial 1999 study to highlight notable differences.
Reporting, sponsored by COSO, identified numerous causal
factors believed to contribute to financial statement fraud We analyzed instances of fraudulent financial reporting
(NCFFR 1987).1 In addition, the COSO-sponsored study alleged by the SEC in AAERs issued during the ten-year
released in 1999, Fraudulent Financial Reporting: 1987- period between January 1998 and December 2007. The
1997, An Analysis of U.S. Public Companies, provided a AAERs, which contain summaries of enforcement actions
comprehensive analysis of fraudulent financial reporting by the SEC against public companies, represent one
through the late 1990s (Beasley et al. 1999). of the most comprehensive sources of alleged cases
of financial statement fraud in the U.S. We focused on
Less is known about the profile of fraudulent financial AAERs that involved an alleged violation of Rule 10(b)-5
reporting since 1997.2 While the U.S. experienced an of the 1934 Securities Exchange Act or Section 17(a) of
unprecedented spate of large company accounting frauds in the 1933 Securities Act given that these represent the
2001 and 2002, including those at Enron and WorldCom, it is primary antifraud provisions related to financial reporting
unclear to what extent the typical fraud profile has changed for U.S. public companies. Our focus was on cases clearly
in the past decade. Thus, COSO commissioned this research involving financial statement fraud. We excluded from our
project to provide COSO, and others, with recent information analysis restatements of financial statements due to errors
that can be used to guide future efforts to combat the or earnings management activities that did not result in a
problem of financial statement fraud and to provide a better violation of the federal antifraud statutes.
understanding of financial statement fraud cases.
Our search identified 347 companies involved in alleged
This research has three specific objectives: instances of fraudulent financial reporting during the
ten-year period. These 347 alleged fraud instances are
• To identify instances of alleged fraudulent financial described in 1,335 individual AAERs (1,013 AAERs directly
reporting by registrants of the U.S. Securities and relate to fraud, while the other 322 describe non-fraud
Exchange Commission (SEC) disclosed by the SEC in an allegations related to the fraud companies). Findings
Accounting and Auditing Enforcement Release (AAER) reported in this study are based on information we obtained
issued during the period 1998-2007. from our detailed analysis of (a) AAERs related to each of the
• To examine certain key company and management sample fraud companies, (b) databases containing selected
characteristics for the companies involved in instances financial statement data reported in Form 10-Ks filed before
of financial statement fraud identified in AAERs and to and during the period the alleged financial statement fraud
compare certain fraud company characteristics to those of occurred, (c) proxy statements issued during the alleged
no-fraud control firms. fraud period, and (d) databases containing business press
articles about the sample companies after the fraud was
• To provide insights related to preventing, deterring, and disclosed, as well as about the no-fraud control firms.
detecting fraudulent financial reporting.

1
We use the terms “fraudulent financial reporting” and “financial statement fraud” interchangeably
throughout this document to represent the intentional material misstatement of financial statements or
financial disclosures (in notes to the financial statements or Securities and Exchange Commission (SEC)
filings) or the perpetration of an illegal act that has a material direct effect on the financial statements or
financial disclosures.
2
Others have studied aspects of fraudulent financial reporting since COSO’s 1999 study was released.
For example, see Report Pursuant to Section 704 of the Sarbanes-Oxley Act (SEC 2003), Ten Things
About Financial Statement Fraud – Second Edition (Deloitte 2008a), Ten Things About the Consequences
of Financial Statement Fraud (Deloitte 2008b), and Ten Things About Financial Statement Fraud – Third
Edition (Deloitte 2009). www.co s o.o rg

2 | FRAUDULeNT FiNANCiAL RePORTiNg: 1998-2007, An Analysis of U.S. Public Companies

Key Findings and insights
Several key findings and insights emerge from the detailed analysis of the 347 financial statement fraud cases. COSO hopes
that close evaluation of these findings and insights will spawn ideas and further research that will help to strengthen the
prevention, deterrence, and detection of fraudulent financial reporting.

Occurrences of Financial Statement Fraud Most fraud companies’ common stock (73 percent3 of
the sample) traded in over-the-counter markets and was
The vast majority of public companies appear to provide not listed on the New York or American Stock Exchanges,
financial reports that are free from material misstatements similar to the frauds examined in COSO’s 1999 study. Further
due to fraud. However, financial statement fraud continued study about differences in exchange listing requirements
to exist during the 1998-2007 time frame, including the well- may provide insights as to whether certain requirements for
publicized frauds at Enron and WorldCom, among others. registrants of the larger exchanges are relevant to the over-
During the ten-year period 1998-2007, the SEC alleged fraud the-counter markets.
involving 347 companies as described in 1,335 AAERs.
In comparison, the 1999 COSO study spanned 11 years Financial Health of Companies involved
of SEC fraud investigations in which nearly 300 frauds
were described in nearly 700 AAERs. Despite thousands Some companies committing fraud were experiencing net
of publicly-traded companies filing apparently fairly losses or were in close to break-even positions in periods
stated financial statements over the ten-year period, the before the fraud. The lowest quartile reflected companies
existence of fraud in any one of the 347 cases is significant in a net loss position and suffering from net operating cash
to stakeholders of the affected entity. In addition, while the flow shortages. Median company net income was $875,000,
incidence of SEC fraud cases increased somewhat from while median cash flow from operations was $317,000. Such
1987-1997 to 1998-2007, the magnitude of individual fraud closeness to break-even positions is consistent with results
cases increased markedly, as discussed below. Continued in COSO’s 1999 study. Thus, pressures of financial strain
focus on finding ways to strengthen financial statement or distress may have provided incentives for fraudulent
fraud prevention, deterrence, and detection is warranted. activities for some fraud companies. Enhanced skepticism
when companies are experiencing financial stress may be
Companies involved warranted for key governance participants, including the
board of directors, auditors, and regulators.
Fraud affects companies of all sizes. The companies
committing fraud had median revenues and total assets Management’s Tone at the Top
just under $100 million in the period prior to the fraud. While
the size of companies in this study was much larger than We gathered information about the types of individuals
in COSO’s 1999 study, which had median total assets of named by the SEC in the AAERs. The SEC continues to name
approximately $15 million, the range of assets or revenues senior management in AAERs for some level of involvement
for companies experiencing fraud was large. Fraud in the fraud, with the CEO and/or CFO named in almost all
companies included startups with no assets or revenues, cases. These findings have important implications for the
as well as companies with just under $400 billion in assets control environment.
or over $100 billion in revenues. Thus, fraud is not limited to
companies of a certain size. executives Named

Similarly, fraud occurred in a variety of industries. Consistent In 72 percent of the cases, the AAERs named the CEO, and
with COSO’s 1999 study, the most frequent industries in 65 percent the AAERs named the CFO as being associated
where fraud occurred included computer hardware and with the fraud. When considered together, in 89 percent
software (20 percent of the fraud companies) and other of the cases, the AAERs named the CEO and/or CFO as
manufacturing (20 percent). These findings suggest that being associated with the financial statement fraud. In
any actions to prevent, deter, or detect fraud should not be COSO’s 1999 study, the CEO and/or CFO were named in 83
limited to any particular industry. percent of the cases. In addition, although the incidence of
enforcement actions against the CEO was the same in the
current study as in the 1999 study (72 percent of cases in

3
Fifty percent of the firms were listed on NASDAQ, and 23 percent of the firms were traded on electronic
www.co s o.o rg bulletin boards, pink sheets, or via other over-the-counter markets.


each period), enforcement actions against the CFO were those related to the attitudes and rationalizations of senior
approximately 50 percent more likely in the current study (65 management in high fraud risk environments.
percent of cases, versus 43 percent in COSO’s 1999 study).
Nature of the Frauds
More study is needed to determine if there are leading
practices that help to reduce the risk of senior management We gathered extensive information from the AAERs about
involvement in financial statement fraud. For example, the nature of the frauds, including the amounts involved, the
emerging practices may exist related to the screening fraud periods, and techniques used.
and selection of senior executive officers, how they
are compensated to avoid excessive fraud risks, and Size and Time Period of the Frauds
how boards and others oversee senior management.
Mechanisms for sharing of those practices with wider For the period 1998-2007, the total cumulative misstatement
audiences may need to be considered. In addition, CPA firms or misappropriation was nearly $120 billion across 300
may want to focus additional effort on assessing the integrity fraud cases with available information (mean of nearly $400
of top management and sharing with the profession those million per case). This compares to a mean of $25 million
approaches that prove effective. of misstatement or misappropriation per sample fraud in
COSO’s 1999 study. While the largest frauds of the early
Alleged Motivations 2000s skewed the 1998-2007 total and mean cumulative
misstatement or misappropriation upward, the median fraud
The SEC’s most commonly cited motivations for fraud of $12.05 million in the present study also was nearly three
included the need to meet internal or external earnings times larger than the median fraud of $4.1 million in the 1999
expectations, an attempt to conceal the company’s COSO study. Thus, the magnitude of the fraud problem has
deteriorating financial condition, the need to increase the increased in the past decade.
stock price, the need to bolster financial performance for
pending equity or debt financing, or the desire to increase Most frauds were not isolated to a single fiscal period. The
management compensation based on financial results. average fraud period extended 31.4 months, with the median
fraud period extending 24 months. This was slightly longer
Better understanding of the psyche of individuals who have than the average and median fraud periods of 23.7 and 21
engaged in fraud may provide insights as to factors that months, respectively, reported in COSO’s 1999 study. This
cause an individual to set aside his or her set of beliefs to finding suggests that once fraud is initiated in one financial
engage in fraud. More can be learned about behavioral period (quarterly or annual), management often continues
aspects that lead to attitudes and rationalizations that to perpetrate fraud in each quarterly and annual financial
ultimately result in an individual or group of individuals statement filing for about two years.
deciding to engage in fraudulent financial reporting (see
Ramamoorti 2008). Insights are needed as to factors that Because there is a significant time lag between the
might lead an individual known to be of high integrity and occurrence of fraudulent financial reporting and the
to possess strong ethical values to subsequently justify issuance of an AAER related to that fraud instance, most of
committing a fraudulent act. Perhaps insights from prior the underlying instances of fraudulent financial reporting
research studies about leadership and other organizational described in the AAERs examined in this study occurred
behaviors in settings not involving fraud may have before the passage of the Sarbanes-Oxley Act of 2002
insights about possible motivators of fraudulent financial (SOX). Only 61 of the 347 fraud companies examined in
reporting. The academic community may be able to provide this study issued fraudulent financial statements involving
analyses or syntheses of findings and insights from prior periods subsequent to 2002, and only a small number of
organizational behavior research that would be helpful in firms were subject to the provisions of Section 404 of SOX.
identifying organizational behavior characteristics that may Thus, future research is warranted to understand the impact
be associated with drivers of fraudulent financial reporting. of SOX on fraudulent financial reporting. It is premature
to draw conclusions about the fraud detection impact of
More guidance about how management’s philosophy, integrity, that legislation based on the frauds examined in this study.
and ethical culture interact with judgment and decision making Furthermore, the approach used in this study does not
is warranted. Insights about these interactions may serve to allow us to provide any insights about the effect of SOX in
strengthen assessments of fraud risk conditions, especially preventing or deterring fraudulent financial reporting.

www.co s o.o rg


Fraud Techniques years. For example, firms engaging in fraudulent financial
reporting had more inside directors (i.e., management) than
The two most common techniques used to fraudulently no-fraud firms during the sub-period 1991-1999.4 However,
misstate the financial statements involved improper revenue following changes in stock exchange listing requirements
recognition and asset overstatements. The majority of frauds implemented by the major U.S. exchanges, statistically
(61 percent) involved revenue recognition, while 51 percent significant differences in the composition of boards no longer
involved overstated assets primarily by overvaluing existing existed between fraud and no-fraud firms in 2001-2004.
assets or capitalizing expenses. The understatement of Furthermore, while there are some differences in certain
expenses and liabilities was much less frequent (31 percent). board characteristics between fraud and no-fraud firms that
Misappropriation of assets occurred in 14 percent of the are statistically significant, in many instances the practical
fraud cases, which was similar to the 12 percent reported in significance of those differences is not overwhelming.
COSO’s 1999 study.
Additional research and information-gathering about board
The occurrence of improper revenue recognition (61 processes may be needed to determine if there are certain
percent) was higher than the rate of occurrence (50 percent) board actions or tasks that impact fraud risk oversight,
reported in COSO’s 1999 study. Close examination of revenue including board group dynamics, process flow, and board
accounting and related fraud techniques is needed to better judgment and decision making (Beasley et al. 2009). Perhaps
understand how revenue recognition is used to distort processes related to board agenda setting, the manner
financial statement information. More detailed analysis in which information is shared and discussed among the
of revenue fraud risk may be needed within industries to board members, and interactions between the board and
strengthen understanding of how revenue is fraudulently management differ between fraud and no-fraud firms. More
misstated. To the extent that improper revenue recognition study is warranted.
involves non-financial executives, better education and
training on revenue recognition concepts and SEC reporting Audit Committee
obligations are needed.
With all the focus on audit committees in the last decade, one
Valuation issues related to recording existing assets of the important insights from this study is that meaningful
deserve more focus, given that a majority of frauds involved differences in audit committee characteristics between
asset overstatements. This concern may be heightened as fraud and no-fraud firms are generally no longer observed.
financial reporting valuations become more dependent on For example, almost all fraud and no-fraud firms had audit
fair value accounting. committees; the average audit committee size for both
groups was about three members; and on average, audit
Role of the Board of Directors committees of both groups met nearly four times per year.

One of the major contributions of this study is the While many audit committee characteristics have been the
comparative analysis of board governance characteristics focus of audit committee reform and regulation over the past
between fraud firms and a similar set of no-fraud firms. This decade, there is little evidence that these characteristics
allows us to observe whether certain board characteristics are associated with the occurrence of fraudulent financial
are more likely to be associated with fraud firms relative to reporting. Although we no longer see meaningful differences
no-fraud firms. in most audit committee characteristics between fraud and
no-fraud firms, this does not mean that all audit committees
Full Board of Directors are similarly effective with respect to preventing, deterring,
and detecting fraudulent financial reporting. Future research
The overarching insight from the analysis of differences in may be needed that focuses on the interaction of other
board characteristics between fraud and no-fraud firms is governance mechanisms (e.g., the nominating committee)
the lack of notable differences in many of the governance with the audit committee’s ability to prevent, deter, and detect
characteristics that have been the focus of regulators, fraudulent financial reporting (see Carcello et al. 2010). And,
exchanges, and governance experts in the last several future research about audit committee processes may be

4
Our sample period overlapped the widely recognized Report and Recommendations of the Blue Ribbon
Committee on Improving the Effectiveness of Corporate Audit Committees (Blue Ribbon Committee (BRC)
1999). That report resulted in several changes in stock exchange listing requirements related to board
governance made in 2000 by both the NYSE and NASDAQ. As a result, we partitioned our analysis of the
data into two sub-periods, 1991-1999 and 2001-2004, based on the first fraud year. As explained later in
this document, we excluded from this sub-analysis frauds occurring in the year 2000 because the stock
www.co s o.o rg exchanges made changes to their listing requirements in 2000.


needed to determine if other characteristics and behaviors Section 404 of the Sarbanes-Oxley Act of 2002
of audit committees have an impact on the prevention,
deterrence, or detection of fraudulent financial reporting. Because of the significant time lag between the occurrence
of fraud and the subsequent issuance by the SEC of an
Compensation Committee AAER, only a small number of the 347 instances of fraud
affected accelerated filers subject to Section 404 of SOX.
Greater focus on the roles and processes used by For those firms, the nature of the Section 404 internal
compensation committees may provide helpful insights control opinions did not foreshadow future financial
as to how boards consider the impact of compensation reporting problems. The Section 404 opinions indicated
policies on the risk of fraud. Most fraud and no-fraud firms effective internal controls unless there had already been
maintained a compensation committee, and there were few a restatement or other correction of a 10-K announced.
differences in compensation committee characteristics Therefore, adverse Section 404 opinions for the small
between fraud firms and no-fraud firms. Because sample examined were not diagnostic of future reporting
compensation arrangements for senior executives are often problems, but instead only highlighted already-announced
tied to financial statement measures, more study about the reporting problems.
effect of compensation policies and processes on fraud risk
and board oversight of that risk may be needed. The small sample size available for analysis limits our ability
to draw any significant insights about auditors’ ability to
Related Party Transactions detect internal control weaknesses that may lead to fraud in
the future. It also is important to note that we are unable to
Fraud firms disclosed significantly more related party measure the impact of Section 404 in preventing or deterring
transactions than no-fraud firms. Seventy-nine percent of management from engaging in fraudulent financial reporting.
fraud firms had disclosed a related party transaction in the
proxy statement filed during the first fraud period compared Auditor Change and Auditor implications
to 71 percent of no-fraud firms for the comparable time
period. The higher frequency of related party transactions The rate of auditor changes for fraud firms was double the
for fraud firms suggests that the presence of related party rate of auditor changes for the similar set of no-fraud firms.
transactions may reflect heightened fraud risk. Greater Twenty-six percent of the fraud firms versus 12 percent of
scrutiny of related party transactions may be warranted to the no-fraud firms changed auditors between the period
determine if the nature of those transactions has broader that the company issued the last clean financial statements
implications regarding management’s integrity, philosophy, and the period the company issued the last set of fraudulent
and ethical culture. financial statements. Sixty percent of the auditor changes
for fraud firms occurred during the fraud period, while the
Auditor Considerations remaining 40 percent of fraud firms that changed auditors
did so during the fiscal period just before the fraud began.
Fraud goes undetected by auditors of all types and sizes. Big A detailed hindsight analysis of auditor changes involving
Six/Four firms audited 79 percent of the fraud companies known instances of fraud may provide helpful insights about
during the fraud period (similar for the no-fraud firms potential relations between conditions leading to auditor
at 83 percent). The challenges of detecting fraudulent changes and conditions related to fraud occurrences.
misstatements of financial information affect auditors of
entities spanning numerous industries and different sizes. Financial statement fraud sometimes implicated the external
auditor. Auditors were named in the AAERs for 78 of the 342
Type of Auditor Opinion on the Financial Statements fraud cases (23 percent) where AAERs named individuals.
This was somewhat lower than what was reported (29
Virtually all of the fraud firms received an unqualified opinion percent) in COSO’s 1999 study. When auditors were named in
on the last set of fraudulently misstated financial statements. the AAERs, about 39 percent of those named were charged
However, the unqualified audit report of fraud firms was more with violating the anti-fraud statutes, while the remaining 61
likely (56 percent) to contain additional explanatory language percent were charged with violating non-fraud provisions
than for no-fraud firms (36 percent). More research is needed including Rule 102(e) of the 1934 Securities Exchange Act.
to examine the nature of the audit report modification and National audit firms were less likely to be named in an SEC
to determine if there is any relation between the report enforcement action than were non-national firms, even
modification and the nature of the fraud technique employed. though national firms audited most of the fraud companies.

www.co s o.o rg


Consequences for individuals press disclosures of an alleged fraud. Fraud company stock
and Firms engaged in Fraud prices also abnormally declined an average of 7.3 percent
in the two days surrounding the announcement of a fraud
We gathered extensive data about consequences affecting investigation by the SEC or Department of Justice.
both individuals serving in management roles and the
companies as a whole for a period of two years subsequent In addition to the negative stock market reactions to news
to the issuance of the last AAER about the fraud. The pairing announcements about alleged fraud or fraud investigations,
of fraud firms with no-fraud firms allowed us to analyze many fraud firms suffered long-term consequences,
whether subsequent events affecting individuals and the including bankruptcy, delisting by national exchanges,
company as a whole were significantly different for fraud and material asset sales. Twenty-eight percent of fraud
firms relative to no-fraud firms. firms were bankrupt or liquidated within two years from
the year in which the SEC issued the last AAER related to
Consequences for individuals the fraud, and 47 percent were delisted from a national
stock exchange. Material asset sales also affected about
The consequences associated with financial statement 62 percent of fraud companies. These rates of occurrence
fraud were severe for individuals allegedly involved. In were significantly higher than the experiences of no-fraud
almost half of the cases (47 percent), the SEC barred one firms during those same time periods.
or more individuals from serving as an officer or director of
a public company. Civil fines were imposed in 65 percent Conclusion
of the fraud cases, and disgorgements were imposed in
43 percent of the cases. The average fine imposed by the Detailed analyses of the findings described above are
SEC was $12.4 million, and the average disgorgement was provided in the remainder of this report. We encourage
$18.1 million. The cumulative amount of fines for all fraud parties involved in financial reporting to carefully consider
companies was $2.74 billion, while the cumulative amount the detailed information presented in this report. We also
of disgorgements was $2.65 billion. The median fine was encourage further research to better understand many
$100,000, and the median disgorgement was $195,000. of the underlying factors likely to affect the prevention,
deterrence, and detection of fraudulent financial reporting.
Most CEOs and CFOs (80 percent or more) left the company COSO hopes numerous parties will recommit their efforts
within two years of the SEC’s last AAER related to the fraud. to improve the prevention, deterrence, and detection of
Twenty-one percent of CEOs were indicted within that time fraudulent financial reporting.
period, and 64 percent of the indicted CEOs were convicted.
Similarly, 17 percent of CFOs were indicted, with 75 percent of Overview of Report
the indicted CFOs being convicted.
The remainder of this report is organized as follows. Section II
Despite the magnitude of these individual consequences, the provides a description of the approach we took to identify the
severity of the penalties may not be a sufficient deterrent. More cases of fraudulent financial reporting and contains a summary
understanding about the mindset of fraud perpetrators may be of the sources and methods used to gather data related to each
needed to understand the factors individuals take into account case. Section III presents the results from our detailed analysis
when they engage in fraudulent activity. Better understanding of the 347 cases of fraudulent financial reporting. Section IV
of their perceptions about possible long-term consequences for provides concluding comments, and Section V contains a brief
engaging in fraud may provide useful perspectives about the description of the authors who conducted this study.
deterrence effect of personal consequences.
We are confident that this report, Fraudulent Financial
Consequences for Companies Committing Fraud Reporting: 1998-2007, will prove helpful to parties concerned
with corporate financial reporting and will add to the insights
Severe consequences also awaited companies committing provided by COSO’s 1999 study, Fraudulent Financial Reporting:
fraud. Companies experienced significant abnormal stock 1987-1997. We hope the study will stimulate greater awareness
price declines as news of the alleged frauds first emerged. of new opportunities for improvements in the corporate
The average fraud company’s stock price dropped by an financial reporting process, as well as avenues for future
abnormal 16.7 percent in the two days surrounding the initial research.

www.co s o.o rg


ii. Description of Research Approach
This study builds on the previous COSO-sponsored study, Fraudulent Financial Reporting: 1987-1997, by
presenting findings related to fraudulent financial reporting for the period 1998-2007. The data collection
effort was conducted under the direction of four accounting researchers (“the authors”) who oversaw
the entire study including generation of this monograph. The authors worked with two research
managers, who monitored and reviewed the work of a data collection team (“the team”). The research
managers reported to and consulted with the authors throughout the entire research process.

The first step in this research project involved the statement fraud. In most instances, the company and/or
identification of all alleged instances of fraudulent financial individuals named neither admitted nor denied guilt. To the
reporting captured by the SEC in an AAER issued during extent that enforcement biases are present, the results of
the period 1998-2007. In order to obtain detailed publicly- this study are limited. However, given no better publicly-
available information about company-wide and management available source of alleged financial statement fraud
characteristics of companies involved, the focus of this instances, we believe that this approach was optimal under
study was on instances of fraudulent financial reporting the circumstances. Furthermore, any SEC fraud investigation
allegedly committed by SEC registrants that ultimately led to is a significant event in the life of the affected company
the issuance of an AAER.5 and individuals involved in the financial reporting process,
including boards of directors and auditors. Thus, insight as
To identify instances of fraudulent financial reporting to fraud occurrences investigated by the SEC is informative,
investigated by the SEC in the period 1998-2007, the team regardless of any inherent biases that may be present in
read all AAERs issued by the SEC between January 1998 how the SEC selects its enforcement cases.
and December 2007. From the reading, the team identified
all AAERs that involved an alleged violation of Rule 10(b)-5 of For purposes of this report, the term “fraudulent financial
the 1934 Securities Exchange Act or Section 17(a) of the 1933 reporting” represents the intentional material misstatement
Securities Act. We focused on violations of these securities of financial statements or financial disclosures (in notes to
laws given that these sections of the 1933 Securities Act the financial statements or SEC filings) or the perpetration
and 1934 Securities Exchange Act are the primary antifraud of an illegal act that has a material direct effect on the
provisions related to financial statement reporting. Because financial statements or financial disclosures. The term
violations of these securities provisions generally require financial statement fraud was distinguished from other
the intent to deceive, manipulate, or defraud, they more causes of materially misleading financial statements, such
specifically indicate alleged instances of financial statement as unintentional errors and other corporate improprieties
fraud than do other provisions of the securities laws.6 that do not necessarily cause material inaccuracies in
financial statements. Throughout this report, references
The AAERs represent one of the most comprehensive to fraudulent financial reporting are all in the context of
sources of alleged, discovered cases of financial statement material misstatements. Our study excludes restatements of
fraud in the U.S. However, such an approach does limit the financial statements due to errors or earnings management
ability to generalize the results of this study to other settings. activities that did not result in a violation of the federal
Because the identification of fraud cases was based on a antifraud securities provisions.
review of AAERs, the findings are potentially biased by the
enforcement strategies employed by the staff of the SEC. The team’s reading of AAERs during this period allowed us
Because the SEC is faced with constrained resources, to develop a comprehensive list of companies investigated
there is the possibility that not all cases of identified fraud by the SEC during 1998-2007 for alleged financial statement
occurring in the U.S. were addressed in the AAERs. There fraud. The team read 1,759 AAERs, beginning with AAER
may be a heavier concentration of companies contained #1004 and ending with AAER #2762. From this process, we
in the AAERs where the SEC assessed the probability identified 347 companies (1,335 total AAERs for these 347
of a successful finding of financial statement fraud as companies) involved in alleged instances of fraudulent
high. Also, the SEC may choose to conduct “sweeps” of financial reporting. For each of these companies, we
particular industries or types of transactions, which may accumulated information about the specific securities
impact the distribution of fraud instances reported in AAERs. law violation to ensure that the case involved an alleged
In addition, the cases contained in the AAERs represent violation of Rule 10(b)-5 of the 1934 Securities Exchange Act
instances where the SEC alleged the presence of financial or Section 17(a) of the 1933 Securities Act.

5
Publicly-traded partnerships, broker-dealers, and unit investment trusts were excluded from this study.
6
We did not include other violations of laws whose only consequence gave rise to a potential contingent
liability (e.g., an “indirect effect illegal act” such as a violation of Environmental Protection Agency regulations). www.co s o.o rg


For each of the 347 companies, the Audited Financial Statement Data
team collected extensive information SeC AAeRs issued
to create a comprehensive database from 1998-2007 We obtained selected audited financial statement
of company and management addressed data from annual financial statements filed in a Form
characteristics surrounding
instances of financial statement fraud
from (a) AAERs related to the alleged
347 instances
of fraudulent financial
10-K with the SEC. We used Standard and Poor’s
COMPUSTAT® database to obtain selected balance
sheet and income statement amounts from the audited
fraud, (b) databases containing reporting. financial statements included in the Form 10-K filed
selected financial statement data with the SEC for the fiscal period preceding the first
reported in Form 10-Ks filed before known instance of fraudulently misstated financial
and during the period the alleged statements for each of the sample companies
financial statement fraud occurred, (“last clean financial statements”). This provided
(c) proxy statements issued during the alleged fraud period, us information about the financial position and results of
and (d) databases of business press articles written about the operations in the period just before the period in which the
sample companies after the fraud was revealed, as well as fraud allegedly first occurred.
about the no-fraud control firms.
We also obtained from COMPUSTAT® the name of the audit
Data Obtained from AAeRs firm responsible for auditing the financial statements issued
during the fraud period and the nature of the auditor’s opinion
The team read all AAERs issued during 1998-2007 related to on those financial statements. If the fraud period extended
the alleged financial statement fraud for each of the sample more than one fiscal year, we obtained the name of the audit
companies. In many cases, several AAERs related to a firm and the type of audit opinion issued for the last fiscal year
single fraud at one company. From the reading, the team of the fraud period.
attempted to capture the following information:
Data Obtained from Proxy Statements
1. A list of the specific annual financial statements
(contained in Form 10-Ks) or quarterly financial We obtained copies of the first proxy statement sent to
statements (contained in Form 10-Qs) fraudulently shareholders during the period in which the alleged financial
misstated and other filings with the SEC (e.g., S-1 statement fraud was in process. We reviewed these proxy
registration statements) that incorporated fraudulently statements to gather information about the characteristics
misstated financial statements. From this, we were able to of the board of directors and its audit and compensation
determine the length of time the alleged fraud occurred. committees (composition, number of meetings, etc.) that were
in place during the fraud period.
2. A brief description of the nature of the fraud allegations
including a description of how the fraud was allegedly Data from Business Press Articles
perpetrated.
To obtain information about consequences for the company,
3. The dollar amounts of the fraud and the primary accounts senior management, and board members subsequent to the
affected. revelation of the financial statement fraud, we performed
an extensive search of the Factiva database of financial
4. Identification of types of personnel and outsiders involved press articles. Among the many news sources included in
in the fraud. Factiva are over 5,000 newspapers, journals, and magazines,
including The Wall Street Journal, The New York Times,
5. An indication of the alleged motivation for committing the fraud. The Financial Times, and The Economist, and over 500
newswires including Dow Jones, Reuters, PR Newswire,
6. The industry in which the company operated. and The Associated Press.

7. A summary of the reported outcome of the SEC’s
investigation, including disciplinary action against senior
management personnel.

www.co s o.o rg


For each fraud, we performed a search for subsequent In addition to data availability issues, readers should also
consequences to the company, senior management, and recognize that a great deal of professional judgment was
board members using a series of key word search strings. necessary when collecting, categorizing, and synthesizing
Our search began with the first day of the last fiscal year in the data. Written summaries prepared from our analysis
which the fraud occurred, and ended on the last day of the of the data obtained from the AAERs comprise several
fiscal year ending two fiscal years after the fiscal year in thousand pages of text, and the team incurred over 10,000
which the last AAER related to the fraud was issued. hours to gather and summarize the data underlying this
study. We believe that we have been reasonable and
We reviewed each instance where an article or press consistent in our judgments, but the research approach was
release was identified as a result of the application of limited by the quality of our judgments.
key word search strings. We captured information about
whether the company had experienced financial difficulty Finally, the authors and research managers performed
to the point of filing for bankruptcy, being placed in a great deal of data review to ensure the quality of the
conservatorship, or liquidating. We also determined whether team’s efforts. Much of the team’s work was subjected
the company was delisted from a national stock exchange to layers of reperformance, review, and reasonableness
or a national securities association, or engaged in a material testing to promote sound and consistent data collection and
asset sale (including a sale of the company). We also summarization.
captured information about the consequences of the alleged
fraud for senior management and members of the board Given the various limitations above, we encourage
of directors, including resignation, termination, and other readers to view the results as sound approximations of the
turnover. In addition, we captured whether members of underlying reality. With the large number of individuals on
senior management were criminally indicted and convicted. the team involved, and with the need for a large amount of
Finally, to examine abnormal stock price effects linked to professional judgment due to the nature of the underlying
public disclosures of the alleged fraud, we captured the first data, the results of the study should be viewed as providing
public disclosure that suggested that material accounting a broad profile of fraudulent financial reporting during
improprieties may have occurred, and the first public this period rather than perfectly precise dollar amounts or
disclosure of an SEC or Department of Justice investigation. percentages for all data points included in this monograph.

Data Limitations

Readers should recognize that, despite the best efforts
to collect complete data for all sample companies, the
data sources used were often incomplete, and sometimes
inconsistent. For example, AAERs were uneven in their level
of disclosure, and other sources (e.g., Form 10-Ks, proxies,
etc.) sometimes were not available. Additionally, the analysis
is limited by the accuracy and completeness of information
that is reported in these sources.

www.co s o.o rg


iii. Detailed Analysis of instances of
Fraudulent Financial Reporting: 1998-2007
We analyzed instances of fraudulent financial reporting reported by the SEC in AAERs issued between
January 1998 and December 2007. After reading 1,759 AAERs, we identified 347 companies involved
in alleged instances of fraudulent financial reporting.7 In most instances, these fraud cases represent
allegations of financial statement fraud made by the SEC without the company and/or individuals
named in the AAER admitting guilt.

This section contains the findings from our reading of (a) While total assets, total revenues, and stockholders’ equity
AAERs related to each of the 347 companies, (b) databases averaged $5.772 billion, $2.557 billion, and $1.001 billion,
containing selected financial statement data reported in respectively, the median of total assets was $93.1 million, the
Form 10-Ks filed before and during the period the alleged median of total revenues was $72.4 million, and the median
financial statement fraud occurred, (c) proxy statements of stockholders’ equity was $39.5 million in the period before
issued during the alleged fraud period, and (d) databases of the fraud began. Given third quartiles of total assets of $674
business press articles written about the sample companies million, total revenues
after the fraud was disclosed. This section contains of $466 million, and
Fraud companies’
extensive information about each of the following items: stockholders’ equity of $242
median assets and
million, most of the sample revenues were just

$100
• Nature of the companies involved companies operated under
• Characteristics of the alleged fraud perpetrators the $500 million size range.9 under
• Nature of the frauds million in the year
• Board governance characteristics, including the nature of Fraud affected companies preceding the first
the audit committee and compensation committee of all sizes. Fraud fraud period.
• Issues related to the external auditor companies ranged from
• Consequences to fraud companies and perpetrators startups with no assets or
subsequent to discovery. revenues to companies
with just under $400 billion in assets or over $100 billion
To examine whether certain board governance in revenues. Similarly, stockholders’ equity ranged from
characteristics and whether certain events affecting negative equity of over $1 billion to positive equity of over
fraud firms subsequent to the revelation of a fraud event $53 billion. However, the typical size of the fraud companies
are unique to fraud companies, we gathered a sample of noted above is substantially larger than the fraud companies
similar no-fraud firms to examine whether differences exist in COSO’s 1999 study.
between fraud firms and no-fraud firms. Our methodology
for selecting and evaluating information related to no-fraud The sample companies in the 1999 study had total assets,
firms is described later in this document in the section total revenues, and stockholders’ equity that averaged
“Board Governance Characteristics.” $533 million, $233 million, and $86 million, respectively. The
median of total assets in the 1999 study was only $15.7
million, the median of total revenues was only $13 million,
Nature of Companies involved and the median of stockholders’ equity was only $5 million
in the period before the fraud began. Given third quartiles
Financial Profile of Sample Companies of total assets of $74 million, total revenues of $53 million,
and stockholders’ equity of $17 million, most of the sample
We were able to obtain the last clean financial statements fraud companies in the 1999 study operated well under the
for 313 of the 347 sample companies.8 Table 1 highlights $100 million size range, which is substantially smaller than
selected financial statement information for these fraud the sample fraud companies from the current study, even
companies. considering the effects of inflation.

7
Generally there were multiple Accounting and Auditing Enforcement Releases (AAERs) related to the
fraud at a single company.
8
Our primary source of previously issued financial statements was the COMPUTSTAT® database. There were
slight differences in availability of certain financial statement items. Thus, we were unable to locate each data
item for all of the 313 sample companies available on COMPUTSTAT®, as shown in the last row of Table 1.
9
Because some high-profile frauds involving very large companies (e.g., Enron, WorldCom, etc.) are
included in this ten-year period, the means are inflated. Therefore, we winsorized the sample by setting all
observations above the 95th percentile to equal the value for the observation at the 95th percentile. The
winsorized means were $1.9 billion for total assets, $1.6 billion for revenues, $478 million for stockholders’
www.co s o.o rg equity, $49 million for net income, and $84 million for cash flow from operations.


Some of the sample companies were financially stressed fraud allegedly began. Similarly, cash flow from operations
in the period preceding the fraud period. The median net averaged $246 million, while median cash flow from
income was only $875,000, with the 25th percentile facing operations was only $317,000. This closeness to break-even
net losses of nearly $2.1 million. The 75th percentile had positions was consistent with what was observed in COSO’s
net income just over $18 million in the year before the 1999 study.

Table 1. Financial Profile of Sample Companies
Last Financial Statements Prior to Beginning of Fraud Period

Total Stockholders’ equity Net income Cash Flow
Assets Revenues (Deficit) (Loss) From Operations
(in $000s)
Mean $5,771,693 $2,557,298 $1,000,508 $140,097 $246,332
Median $93,112 $72,360 $39,457 $875 $317
Minimum value $0 ($23) ($1,021,747) ($2,687,000) ($1,214,000)
1st quartile $14,806 $9,468 $4,765 ($2,136) ($2,007)
3rd quartile $673,805 $465,870 $242,261 $18,090 $37,384
Maximum value $391,673,000 $128,313,000 $53,206,590 $8,897,000 $16,654,000
Companies 313 311 312 311 303

National Stock exchange Listing

We reviewed the last clean financial statements and CRSP According to the 2006 Final Report of the Advisory
database to identify the national stock exchange where Committee on Smaller Public Companies (Advisory
each company’s stock traded. We were able to identify the Committee 2006), approximately 19.5 percent of all publicly-
stock exchange listing for 313 of the 347 sample companies. traded companies are registered on the New York Stock
As indicated by the pie chart in Table 2, most (50 percent) Exchange, 5.7 percent are registered on the American Stock
were traded on the NASDAQ exchange. Twenty-three Exchange, and 24.2 percent trade in the NASDAQ National
percent of the companies’ stock traded on the New York Market or NASDAQ Capital Market. The remainder trade on
Stock Exchange, and four percent of the companies’ stock the over-the-counter bulletin boards (22.6 percent ) or pink
traded on the American Stock Exchange. Finally, 23 percent sheets (28.0 percent). Thus, the mix of fraud firms trading
of the companies’ stock traded on electronic bulletin boards, in NASDAQ markets (50 percent) is higher than the overall
pink sheets, and other over-the-counter markets. profile of public companies on NASDAQ (24 percent).

Table 2. Sample Companies’ National Stock Exchange Listing
(n = 313 with Available Information)

New York Stock Exchange
American Stock Exchange
23%
NASDAQ 23%

Electronic bulletin boards,
pink sheets, and other
over-the-counter markets 4%

50%

www.co s o.o rg


The percentage of companies (73 percent) whose stock percentage of companies in COSO’s 1999 study whose stock
traded on any over-the-counter market (NASDAQ, electronic traded on the New York Stock Exchange (15 percent) or
bulletin boards, pink sheets, etc.) was in line with the American Stock Exchange (7 percent) also was fairly similar
78 percent of companies in the 1999 COSO study whose to the present study.
stock traded on any of the over-the-counter markets. The

Table 3. Primary Industries of Sample Fraud Companies

3% Computer hardware/software
Other manufacturing
1%
1% 9% 20% Healthcare and health products
Retailers/wholesalers
6%
Other service providers
6% Telecommunications
Energy and natural resources
7%
20% Financial service providers
7% Insurance
9% Real estate
11%
Miscellaneous
Not available

industries for Companies involved geographic Location of Sample Companies

We reviewed the information included in the AAERs We reviewed the AAERs to identify the geographic location
to determine the primary industry in which the fraud of the fraud companies. Most of the frauds were committed
companies operated. Similar to our findings in the 1999 at or directed from the companies’ headquarters locations.
COSO study, the two most frequent industries cited were We were able to identify the headquarters location for 329 of
computer hardware and software (20 percent) and other the 347 fraud companies. Table 4 contains information about
manufacturing (20 percent). Other frequently-cited industries the frequency of cases for states in which at least 10 fraud
in the current study were healthcare/health products (11 companies were located. Similar to sample fraud companies
percent), retailers/wholesalers (9 percent), other service examined in COSO’s 1999 study, the highest percentages of
providers (7 percent), and telecommunications (7 percent). frauds involved companies headquartered in California and
See the pie chart in Table 3. New York. In the current study, the most fraud companies
were located in California (19 percent of the fraud cases),
New York (10 percent), Texas (7 percent), Florida (7 percent),
New Jersey (5 percent), Massachusetts (4 percent), and
Illinois (4 percent). This pattern is consistent with centers of
business activity in the U.S.

www.co s o.o rg


Table 4. Locations of Fraud Companies' Headquarters
(n = 329 with available information)

20% 19

15%

10
10%
7

5 4
5%

0%

California Texas New Jersey Illinois
New York Florida Massachusetts International

Alleged Fraud Perpetrators fraud companies, representing 72 percent of the sample
companies with available information. The second most
individuals Named in the AAeRs frequently identified senior executive was the CFO. The
CFO was named in 222 of the 342 fraud companies, which
From our reading of the AAERs, we captured information represents 65 percent of the companies involved. When
about the types of company representatives and outsiders considered together, the CEO and/or CFO were named in 305
named in an AAER related to each instance of alleged of the 342 cases (89 percent).
fraudulent financial reporting. We captured names of all
individuals listed in any of the AAERs related to an instance The company controller was named in 115 of the 342 frauds,
of fraudulent financial reporting, whether these individuals representing 34 percent of the fraud instances. The chief
were charged with fraud or charged with other lesser operating officer (COO) was named in 10 percent of the
violations. The SEC named in the AAERs individuals involved frauds (35 of 342), and other vice presidents were named in
in the alleged fraud for 342 of the 347 fraud companies. Even 129 of the 342 frauds (38 percent of the cases). Lower level
though these individuals were named in an AAER, there was personnel were named in 23 percent of the cases (80 of
no certain evidence that all the named participants violated 342 fraud instances). Recall that our classification scheme
the antifraud statutes, and other individuals not named in an tracked the highest named position for an individual. Thus,
AAER may have been involved in the fraud. In addition, most the noted percentages associated with less senior positions
of the named participants neither admitted nor denied guilt may be understated. In addition, because of the relatively
of any kind. small size of some of the fraud firms in this sample, some
of the noted positions (e.g., COO) may not have been filled.
Using the highest managerial title for an individual, we Finally, SEC enforcement actions may target top executives
summarized the typical employee positions named in the more frequently than lower level employees. These factors
AAER. For example, if one individual had may contribute to the lower percentages
the titles of chief financial officer (CFO) and The CeO and/or noted for these positions.
controller, we reported that as involving
strictly the CFO position in our reporting in
CFO were named
in an AAeR for The frequency with which the AAERs
Table 5 on the next page. As noted in Table
89%
name the CEO as being allegedly involved
5, the senior executive most frequently of the in the fraud was the same (72 percent of
named in an AAER was the chief executive fraud companies. fraud companies) for the current study
officer (CEO). The CEO was named as and the 1999 COSO study. However, the
one of the parties involved in 246 of 342 frequency with which the AAERs named

www.co s o.o rg


Table 5. Types and Frequencies of Individuals Named in AAERs

100%
Current Study
89
83 COSO’s 1999 Study
80%
72
65
60%

43
40% 34 38
27
21 23
20% 18 16 15
10 10 12
7
0%

CEO CEO and/or CFO COO Lower Level Other Titles
CFO Controller Other VPs No Titles Given

Note: In many cases the AAERs cited board members for their involvement in the fraud. The vast majority of these
individuals appeared to be company managers serving on the board, including CEOs serving as Board Chair.

the CEO and/or CFO in the current study (89 percent) is statistics about the rate of particular motivations. However,
slightly higher than in the 1999 COSO study (83 percent). In among those noted, the most commonly cited reasons
addition, the CFO was approximately 50 percent more likely summarized by the SEC in the AAERs include committing the
to be subject to an SEC enforcement action in the current fraud to –
study than in the 1999 study (named in 65 percent of cases
in the current study, versus 43 percent of cases in COSO’s • Meet external earnings expectations of analysts and others
1999 study). Finally, the frequency with which the SEC named
other individuals in the AAERs was generally higher in the • Meet internally set financial targets or make the company
current study as compared to the 1999 COSO study. look better

In addition to the results in Table 5, individuals named in the • Conceal the company’s deteriorating financial condition
AAERs extended beyond company executives. In 81 of the
342 fraud companies (24 percent of the cases), outsiders • Increase the stock price
were named, generally customers and vendors. The external
auditor was named in the AAER for 78 of the 342 fraud • Bolster financial position for pending equity or debt financing
companies (23 percent of the fraud cases with information
about perpetrators), and members of the audit committee • Increase management compensation through
were named in 7 of the 342 fraud companies (2 percent of achievement of bonus targets and through enhanced
the cases). stock appreciation

Alleged Motivation for the Fraud • Cover up assets misappropriated for personal gain.

In some instances, the SEC provided discussion in the
AAERs about the alleged motivation for the fraud. Because
the SEC did not consistently describe the alleged motivations
in each fraud instance and there were often multiple
motivations for a single fraud, we do not provide summary

www.co s o.o rg


Nature of the Frauds
Total Amount of the Fraud

In an attempt to obtain a judgmental measure of the typical misappropriation over the fraud period, while the median
size of the financial statement frauds, we accumulated fraud involved $12.05 million.10 The smallest fraud was
information from the AAERs that provided some indication $47,200, while the largest totaled $25.8 billion.11 The first
of the amounts involved. In some cases, and third quartiles of cumulative
the AAERs did not disclose the dollar The average cumulative misstatements or misappropriations
amounts involved. As a result, we were misstatement amount was were $3.65 million and $55.95 million,
only able to obtain some measure of the
dollar amounts involved for 300 of the 347 $397.68 million,
respectively.12 The wide variance
between the mean and median fraud
while the median
fraud companies. As reported in Table amounts is due to a few large high-
cumulative misstatement
6, the average fraud involved $397.68 profile frauds during the period, such as
million of cumulative misstatement or was $12.05 million. the frauds at Enron and WorldCom.

Table 6. Cumulative Dollar Amount of Fraud for a Single Company
# of Sample Mean Cumulative Misstatement Median Cumulative Misstatement
Companies or Misappropriation or Misappropriation
with information (in $ millions) (in $ millions)

Cumulative 300 $397.68 $12.05
amount of
fraud for a
single company
Minimum = $47,200; Maximum = $25.8 billion; 1st quartile = $3.65 million; 3rd quartile = $55.95 million

The size of the cumulative misstatement or misappropriation Unfortunately, the AAERs do not consistently report the
in the current study was substantially larger than the dollar amounts involved in each fraud. In some instances,
cumulative misstatement or misappropriation summarized the AAERs report the dollar amounts of the fraud by noting
in COSO’s 1999 study. For the sample fraud companies in the the extent to which assets were misstated. In other cases,
1999 COSO study, the average cumulative misstatement was the AAERs report the amounts that revenues, net income,
only $25.0 million, while the median cumulative misstatement pre-tax income, or other items were misstated. We used
was $4.1 million. The first and third quartiles of cumulative the nature of the data presented in the AAER to develop
misstatements or misappropriations for the 1999 COSO study a reasonable measure of the fraud amount; however, we
were $1.6 million and $11.76 million, respectively. caution the reader that a great deal of judgment was used.
In addition, this analysis was dependent on which figures
For the period 1998-2007, the total cumulative misstatement the SEC chose to disclose in the AAERs. Accordingly,
or misappropriation was nearly $120 billion across 300 fraud the categories and figures below should be viewed as
cases with available information. This large total is driven reasonable estimates of fraud amounts (i.e., not exact point
by the numerous large company frauds of the early 2000s, estimates). Information about the amounts involved by fraud
including Enron, WorldCom, and others. It is clear that the type is provided on the next page in Table 7.
magnitude of the fraud cases was much greater in 1998-2007
than in 1987-1997.

10
To evaluate the impact of large outliers, we winsorized the data by setting the cumulative misstatement
or misappropriation amount for those frauds above the 95th percentile to be equal to the value for the 95th
percentile. The winsorized average was $203.7 million.
11
For two high-profile frauds, Royal Ahold and WorldCom, the cumulative fraud amounts provided in the
AAERs were somewhat lower than amounts we noticed in either an SEC press release or in media descriptions
of the case. For consistency, in Table 6 we always used the amounts presented in the AAERs, rather than
including any larger fraud amounts discussed in press releases or media stories.
12
Ideally, we would report misstatement information in percentage rather than dollar terms. However, we are
unable to report percentages for most companies due to the limited amount of information provided in the
AAERs about dollar misstatements and the lack of available financial statements for all fraud periods (which
reflect misstated values anyway) for those companies with AAERs reporting misstatement information. www.co s o.o rg


Asset misstatements averaged $226.74 million, with a median $10.2 million to $21.5 million. The average misappropriation
of $7.9 million. The average misstatements of revenues, of assets (i.e., theft of assets) was $16.3 million, while the
expenses, pre-tax income, and net income ranged from median misappropriation of assets was $4.0 million.
$91.44 million to $958.98 million, with medians ranging from

Table 7. Dollar Amount of Misstatements by Fraud Type
# of Fraud Mean Cumulative Median Cumulative
Companies Misstatement Misstatement
Misstatement Type with information (in $ millions) (in $ millions)

Assets 44 $226.74 $7.9
Revenue or gain 132 $455.04 $10.3
expense 26 $91.44 $19.8
Pre-tax income 20 $958.98 $21.5
Net income 36 $525.21 $10.2
Misappropriation of assets 15 $16.30 $4.0
Note: See Table 1 for the typical size of the companies involved.

While Tables 6 and 7 provide some information about the pre-tax misstatement of $6.75 million. This was substantially
average and median cumulative effects of the fraud over larger than in COSO’s 1999 study, which reported an average
the entire fraud period, Table 8 provides an overview of pre-tax income misstatement of $7.1 million and median
the largest income misstatement in a single period. For pre-tax income misstatement of $3.2 million. For AAERs
each of the companies where the related AAERs reported reporting misstatements as a function of net income (105
misstatement information as a function of pre-tax income or companies), the average largest single period misstatement
net income, we identified the largest single-year or single- of net income was $90.4 million with a median single period
quarter misstatement over that company’s fraud period. For net income misstatement of $5.0 million.13 This was also
the AAERs providing misstatement information relative to substantially larger than the average and median largest
pre-tax income (information provided for 66 companies), single period net income misstatement of $9.9 million and
the average of the largest pre-tax misstatement in a single $2.2 million, respectively, reported in COSO’s 1999 study.
period was $101.6 million, with a median single period

Table 8. Largest Single Period income Misstatement
# of Fraud Mean Largest Single Year Median Largest Single Year
Companies with or Quarter Misstatement or Quarter Misstatement
Description information (in $ millions) (in $ millions)

information reported 66 $101.6 $6.75
as a function
of pre-tax income
information reported 105 $90.4 $5.0
as a function
of net income

Timing of Fraud Period

For the 347 instances of fraudulent financial reporting, of misstatements were 1997-2001, with over 100 companies
the related fraudulently misstated financial statements misstating their financials in each of these years. Due to
were issued in calendar years beginning before 1990 and the time lag in SEC enforcement, the vast majority of the
extending through 2006. The years with the greatest number misstated periods were before the passage of SOX in 2002.

13
The winsorized means (set equal to the 95th percentile value) were $54.3 million for pre-tax income and
www.co s o.o rg $38.0 million for net income.


Only 61 of the 347 fraud companies examined in this study Methods of Fraudulently Reporting
had fraudulently misstated financial statements involving Financial Statement information
periods subsequent to 2002. Only a small number of those
involved companies subject to the reporting provisions of Based upon information included in the AAERs, we made
Section 404 of SOX. our best attempt to identify the methods used to fraudulently
report the financial statement information. As noted in Table
Typical Length of Problem Period 9, the two most common techniques used to fraudulently
misstate financial statement information involved
The financial statement frauds generally involved multiple overstating revenues and assets. Sixty-one percent of the
fiscal periods. Information to determine the number of 347 fraud companies recorded revenues inappropriately,
months from the beginning of the first fraud period to the primarily by creating fictitious revenue transactions or by
end of the last fraud period was available for all of the 347 recording revenues prematurely. This was a higher rate of
sample companies. Fraud periods extended on average revenue misstatements than the 50 percent found in COSO’s
for 31.4 months, with the median fraud period extending 24 1999 study.
months. This was slightly longer
Fraudulent misstatement
than the average and median Fifty-one percent of the
The typical length of financial statements
fraud periods of 23.7 months 347 fraud companies
of the fraud frequently involved the
period was and 21 months, respectively, overstated assets, primarily overstatement of revenues

2 years reported in COSO’s 1999 by overvaluing existing and assets. intentional
. assets or capitalizing
study. Many of the frauds misstatement of financial
began with misstatements of items that should have statements was noted much
interim financial statements been expensed.14 Thirty- more frequently than
that were continued in annual one percent of the 347 misappropriation
financial statement filings. Only 44 of the 347 companies (13 companies’ financial of assets.
percent) issued fraudulent financial statements involving statements were misstated
a period of less than twelve months. The longest problem through the understatement
period was 180 months (and it was 168 months for two other of expenses/liabilities. That rate was higher than the 18
companies). percent found in COSO’s 1999 study.

Table 9. Common Financial Statement Fraud Techniques
Percentage of the 347 Fraud Companies
Methods Used to Misstate Financial Statements Using Fraud Method a

improper revenue recognition: 61%
Recording fictitious revenues – 48%
Recording revenues prematurely – 35%
No description/“overstated” – 2%
Overstatement of assets (excluding accounts receivable overstatements due to revenue fraud): 51%
Overstating existing assets or capitalizing expenses – 46%
Recording fictitious assets or assets not owned – 11%
Understatement of expenses/liabilities 31%
Misappropriation of assets 14%
inappropriate disclosure (with no financial statement line item effects) 1%
Other miscellaneous techniques (acquisitions, joint ventures, netting of amounts, etc.) 20%
Disguised through use of related party transactions 18%
insider trading also cited 24%

a
The subcategories such as premature revenues or fictitious revenues and assets do not sum to the category totals due to
multiple types of fraud employed at a single company. Also, because the financial statement frauds at the sample companies
often involved more than one fraud technique, the sum of the percentages reported exceeds 100 percent.

14
To avoid double-counting, the information about the overstatement of assets does not include
overstatements of accounts receivable due to the revenue recognition frauds. www.co s o.o rg


Most of the financial statement fraud instances involved sales after the goods were ordered but before they were
intentionally misstating financial statement information, with shipped to the customer.
only 14 percent of the fraud cases involving misappropriation
of company assets (i.e., theft of assets). This was consistent • Improper cutoff of sales. To increase revenues, the
with earlier findings in COSO’s 1999 study that 12 percent of accounting records were held open beyond the balance
the fraud cases involved misappropriation of assets and in sheet date to record sales of the subsequent accounting
the 1987 Report of the National Commission on Fraudulent period in the current period.
Financial Reporting that 13 percent of the cases against
public companies involved misappropriation of assets. • Improper use of the percentage of completion
method. Revenues were overstated by accelerating the
As noted in Table 9, over 60 percent of the sample estimated percentage of completion for projects in process.
companies overstated revenues. The revenue
misstatements were primarily due to recording revenues • Unauthorized shipments. Revenues were overstated
fictitiously or prematurely by employing a variety of by shipping goods never ordered by the customer or by
techniques that include the following: shipping defective products and recording revenues at
full, rather than discounted, prices.
• Sham sales. To conceal the fraud, company
representatives often falsified inventory records, shipping • Consignment sales. Revenues were recorded
records, and invoices. In some cases, the company for consignment shipments or shipments of goods for
recorded sales for goods merely shipped to another customers to consider on a trial basis.
company location. In other cases, the company pretended
to ship goods to appear as if a sale occurred and then We do not report percentages for each of the above types
hid the related inventory, which was never shipped to of fraudulent revenue schemes because the language used
customers, from company auditors. by the SEC to describe fraud techniques varied extensively,
making it difficult to classify the various types in exact ways.
• Conditional sales. These transactions were Thus, it was difficult to categorize reliably the frequency of a
recorded as revenues even though the sales involved specific revenue recognition fraud technique.15
unresolved contingencies or the terms of the sale were
amended subsequently by side letter agreements, which Also, in several instances, company representatives were
often eliminated the customer’s obligation to keep the able to falsify confirmation responses directly or indirectly
merchandise. by convincing third parties to alter the confirmation
response. In other cases, company personnel created a
• Round-tripping or recording loans as sales. Some variety of false documents.
companies recorded sales by shipping goods to alleged
customers and then providing funds to the customers Over half of the sample companies misstated the financial
to pay back to the company. In other cases, companies statement information by overstating assets. Table 10
recorded loan proceeds as revenues. highlights the typical asset accounts overstated by sample
companies. Even excluding the effects of misstating
• Bill and hold transactions. Several companies accounts receivable due to the revenue recognition frauds,
improperly recorded sales from bill and hold transactions the two most common asset accounts misstated were
that did not meet the criteria for revenue recognition. inventory (51 cases) and accounts receivable (43 cases).
Other asset accounts misstated included property, plant, and
• Premature revenues before all the terms of the equipment (24 cases); cash/marketable securities (19 cases);
sale were completed. Generally this involved recording loans/notes receivable/mortgages (13 cases); investments
(12 cases); and prepaid expenses (11 cases).

15
There are many rich examples of alleged revenue frauds using the methods listed above. Interested readers
may consult the following AAERs for illustrative examples of many of these methods. These AAERs were
haphazardly selected from numerous possible examples, and there is no intent to highlight any particular company
or individual. Rather, these AAERs simply provide interesting insights into alleged revenue fraud methods.
• AAER 1422 - www.sec.gov/litigation/admin/33-7994.htm
• AAER 1559 - www.sec.gov/litigation/litreleases/lr17522.htm
• AAER 2200 - www.sec.gov/litigation/litreleases/lr19121.htm and related complaint at
www.sec.gov/litigation/complaints/comp19121.pdf
• AAER 2126 - www.sec.gov/litigation/litreleases/lr18935.htm and related complaint at
www.sec.gov/litigation/complaints/comp18935.pdf
www.co s o.o rg • AAER 2451 - www.sec.gov/litigation/admin/2006/33-8716.pdf


Table 10. Number of Fraud Cases With Asset Accounts Misstated

51
50
43
40

30
24

20 19
13 12 11
10

0

Inventory PP&E Loans/Notes Rec/Mortgages Prepaid Expenses
A/R Cash/Marketable Securities Investments

Board governance Characteristics whether certain board governance characteristics differed
between fraud and no-fraud firms.
A large body of accounting research examines the relation
between board governance characteristics and accounting For each fraud company, we selected a similar no-fraud
outcomes (for example, see Cohen et al. 2004; DeZoort et company. First, the fraud and no-fraud pairs are traded
al. 2002). To contribute to our understanding of the relation on the same stock exchange. For example, if the fraud
between the presence of fraud and board governance company was traded on NASDAQ, the no-fraud company
characteristics, we gathered information on the board of was selected from NASDAQ to control for differences in
directors and on the audit and compensation committees governance characteristics across exchanges. Second, the
from company proxy statements filed with the SEC. Because proxy data are gathered from corresponding time periods
we were interested in the governance characteristics in (i.e., to control for differences in governance characteristics
place at the time the fraud began, we gathered governance across time). Third, the industries of the fraud and no-fraud
data based on who was on the board and on the board samples are similar (based on the Standard Industrial
committees during the first fraud year by examining the Classification (SIC) codes), so as to control for any variations
proxy statements filed with the SEC in the first year of the in governance characteristics across industries.
fraud. We were able to locate proxies for 203 of the 347 fraud
companies.16 We also gathered data on board leadership Finally, after the first three constraints, we attempted to
issues disclosed by the company in the proxy statement make the size of the fraud and no-fraud companies as
(e.g., whether the same individual served as both CEO and similar as possible, since larger companies are expected
chairman of the board, whether the company’s founder was to have more advanced governance mechanisms due to
on the board, etc.) and whether there were disclosures of their greater resources. Achieving similar size was the most
related party transactions. challenging, as the other three constraints were already in
place. If we could not identify an appropriate no-fraud firm
To analyze whether certain governance characteristics whose market value of equity was within plus or minus 30
were associated with a higher incidence of fraud, we percent of the fraud firm’s market value, we then measured
gathered a sample of 203 no-fraud companies that is size using total assets (plus or minus 30 percent). Ultimately,
similar to the 203 fraud companies with available proxy the size of the fraud and no-fraud companies is within plus
information. Our goal was to compare the board governance or minus 30 percent in over 75 percent of the cases. There
characteristics of the fraud companies with similar are no significant differences in median market value of
companies apparently not engaging in fraud to identify equity or assets between fraud and no-fraud firms. Based on

16
In some instances, the companies failed to file a proxy with the SEC. For others, the relevant proxy was not
available in electronic databases or via purchase through outside vendors. www.co s o.o rg


the procedures described above, the samples of fraud and Full Board of Director Characteristics
no-fraud companies are similar and provide a reasonable
basis for comparison. Board Size and independence

Our sample period overlapped the widely recognized Report Table 11 contains information about the size and composition
and Recommendations of the Blue Ribbon Committee on of the full board of directors. For each board characteristic
Improving the Effectiveness of Corporate Audit Committees in Table 11, we report the average for the fraud firms and
(BRC 1999), jointly issued in 1999 by the New York Stock the average for the similar set of no-fraud firms for the full
Exchange and the National Association of Securities sample and for each of the sub-periods examined (1991-1999
Dealers. That report resulted in several changes in stock and 2001-2004). We also report the difference in averages
exchange listing requirements related to board governance between the fraud and no-fraud firms and report the results
implemented in 2000 by both the NYSE and NASDAQ. As of our statistical tests by providing the p-value results when
a result, we partitioned our analysis of the data into two those differences between fraud and no-fraud firms were
sub-periods, 1991-1999 and 2001-2004, based on the first statistically significant.18 We conducted tests to determine
fraud year for these 203 fraud companies.17 This allowed whether the differences between fraud and no-fraud firms
us to examine whether linkages between certain board were statistically significant for both the full sample and the
governance characteristics and fraud occurrences two sub-periods examined. Because the sample sizes for
continued subsequent to several changes in listing each of the sub-periods examined were much smaller than
requirements related to board governance. the full sample (especially for the 2001-2004 sub-period), the
lack of statistical significance in tests of each sub-period
The overarching insight from the analysis of differences may be due to lack of statistical power due to the smaller
in board characteristics between fraud and no-fraud sample sizes. Thus, there may be differences in fraud and
firms reported in the pages that follow is the lack of no-fraud firms that we cannot statistically observe due to
notable statistical differences in many of the governance size limitations in each sub-sample.
characteristics that have been the focus of regulators,
exchanges, and governance experts in the last several For all board characteristics where we report a p-value less
years. Many board of director characteristics appear to than 0.10, the differences between fraud and no-fraud firms
no longer differ significantly between fraud and no-fraud were interpreted to be statistically significant, consistent
firms. And, in some instances, the noted differences are in with most research. If no p-value is reported for a particular
directions opposite of what might be expected. Furthermore, board characteristic, readers should conclude that fraud
while some characteristics were found to be statistically and no-fraud firms do not differ significantly in that board
significant, many of those differences may lack any practical characteristic. We use this reporting technique for all tables
significance (i.e., they may be too small to matter). While where we report a statistical test of the difference between
we report whether there are statistical differences between fraud and no-fraud firms.
fraud and no-fraud firm governance characteristics, we
leave the evaluation of practical significance to the reader. As shown in Table 11, the average fraud firm had 7.7
directors on the board as compared to 8.0 directors for no-
These collective observations raise the possibility that there fraud firms. This difference was not statistically significant.
are other more important governance characteristics or
processes that affect the board’s ability to assess the risk of A large body of academic research finds that board and
financial statement fraud and oversee the implementation of audit committee independence affects the effectiveness
procedures to prevent, deter, and detect fraud. of board and audit committee oversight. We examined
the relation between board independence and fraud. In
analyzing board member independence, the following

17
While we studied AAERs issued by the SEC between 1998 and 2007, the calendar years in which these
203 frauds began were as early as 1991 and as late as 2004. In our sub-period analyses, we excluded frauds
occurring in 2000 because the BRC Report was issued in 1999 and the stock exchanges made changes to
their listing standards in 2000. Interestingly, though, more frauds began in 2000 (n = 38) than in any other
year. Thus, we re-ran our analyses including the year 2000 in the post sub-period (i.e., we compared the
1991-1999 sub-period to the 2000-2004 sub-period as a sensitivity test). Our results were very similar to those
reported in this monograph.
18
We tested whether there was a statistical difference between the fraud sample and the no-fraud sample
for each variable. We report p-values for those differences that were statistically significant at below the 0.10
level (two-tailed).
www.co s o.o rg


definitions were used to categorize individual members of were not statistically significant, except for the difference
the board of directors into one of three categories: in the percentage of grey directors who were relatives of
management. Seven percent of fraud firm grey directors
• Inside director – A director who was also an officer or were relatives of management as compared to 18 percent
employee of the company or a subsidiary or an officer of for no-fraud firms (p-value = 0.086).
an affiliated company.
When board independence was examined for the two
• Grey director – A director who was a former officer sub-periods (1991-1999 and 2001-2004), we found that
or employee of the company, a subsidiary, or an affiliate; the results for the 1991-1999 sub-period were generally
relative of management; professional advisor to the consistent with the full sample results. That is, fraud firms
company; officer or owner of a significant supplier or had statistically more inside directors than no-fraud firms for
customer of the company; interlocking director; officer 1991-1999 (p-value = 0.069). We also found that fraud firms
or employee of another company controlled by the CEO were significantly more likely to have consultants as grey
or the company’s majority owner; owner of an affiliate directors (32 percent) than were no-fraud firms (14 percent)
company; or creditor of the company. (p-value = 0.034). However, the types of directors serving on
boards in 2001-2004 were not statistically different for fraud
• Outside director – A director who had no disclosed and no-fraud firms. Thus, differences in board composition
relationship (other than stock ownership) between the following the year 2000 may no longer be associated with
director and the company or its officers. the occurrence of fraudulent financial reporting.

The average percentage of inside directors on the board for We found a decrease in the percentage of inside and grey
fraud firms was 30 percent as compared to 25 percent for directors on boards between the two sub-periods for both
no-fraud firms. This difference was statistically significant fraud and no-fraud firms. For 1991-1999, 32 percent of the
(p-value = 0.010). There was no significant difference in the fraud firm boards were composed of inside directors as
percentage of outside directors for fraud firms (60 percent of compared to only 25 percent of the fraud firm boards in
the board) versus no-fraud firms (63 percent of the board). 2001-2004. Consistent with that trend, the percentage of
outside directors on fraud firm boards increased from 56
There was no statistical difference between the two groups percent in the 1991-1999 sub-period to 67 percent in the
in the average percentage of grey directors. We were able 2001-2004 sub-period. This was consistent with a general
to analyze the types of grey directors serving on the board shift in governance expectations over time that boards
of directors for 63 fraud and 63 no-fraud firms. The most should have a greater percentage of outside directors.19
common types of grey directors were former company
officers, consultants, and outside legal counsel. Differences
in types of grey directors serving on fraud and no-fraud firms

While fraud firms had
significantly more
inside directors than
no-fraud firms in the
1991-1999 sub-period,
this difference did not
continue in the
2001-2004
sub-period.

19
We occasionally highlight shifts in overall trends by comparing findings from the 1991-1999 sub-period and
findings from the 2001-2004 sub-period to provide insights about apparent trends over time. However, we
have not performed formal statistical tests of noted differences between the two sub-periods. www.co s o.o rg


Table 11. Board of Director Composition (Means)
Full Sample 1991-1999 Sample 2001-2004 Sample

# of Pairs

# of Pairs

# of Pairs
No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud
of Firms

of Firms

of Firms
Fraud —

Fraud —

Fraud —
Sample

Sample

Sample

Sample

Sample

Sample
p-value

p-value

p-value
Fraud

Fraud

Fraud
Number of board members 203 7.7 8.0 -0.3 113 7.4 7.8 -0.4 52 7.9 8.0 -0.1
Type of board member:
inside director 203 30% 25% 5% .010 113 32% 28% 4% .069 52 25% 23% 2%
grey director 203 10% 12% -2% 113 12% 13% -1% 52 8% 11% -3%
Outside director 203 60% 63% -3% 113 56% 59% -3% 52 67% 66% 1%
Type of grey director:
Former company officer 63 45% 57% -12% 41 38% 52% -14% 13 62% 54% 8%
Relative of management 63 7% 18% -11% .086 41 9% 21% -12% 13 4% 15% -11%
Consultant to company 63 25% 16% 9% 41 32% 14% 18% .034 13 8% 8% 0%
Outside legal counsel 63 15% 10% 5% 41 15% 10% 5% 13 19% 15% 4%
interlocking director 63 4% 4% 0% 41 5% 6% -1% 13 4% 0% 4%
Banker 63 0% 0% 0% 41 0% 0% 0% 13 0% 0% 0%
Non-bank creditor 63 0% 0% 0% 41 0% 0% 0% 13 0% 0% 0%
Officer of significant 63 2% 0% 2% 41 2% 0% 2% 13 4% 0% 4%
supplier or customer
Note: A p-value that is less than 0.10 indicates that the difference between fraud and no-fraud firms was statistically significant
(two-tailed).

Board Member Age, Tenure, and expertise percent and 51 percent,
respectively; these were While the average
We also gathered data about specific characteristics of not statistically different). tenure of fraud firm
individuals who served on the boards of the fraud and no- directors was
fraud firms. Results are reported in Table 12. The age of the significantly lower than
We also examined
for no-fraud firms, there
average board member was approximately the same for each board member’s
may be little practical
the fraud and no-fraud firms (53.9 and 54.3 years of age, director experience by significance in this
respectively). measuring how many difference.
other directorships were
Board members of fraud firms had served on the fraud held by each individual
company’s board for 6.7 years on average before the first director. The average
year of the fraud, which was statistically lower than the board member served on one other corporate board (1.1
average of 7.7 years that directors of no-fraud firms served other directorships for individuals serving on fraud firm
(p-value = 0.010). Thus, individuals serving on the boards boards, 0.9 other directorships for no-fraud firms). Also, only
of fraud firms had fewer years of experience on that board 16 percent of fraud firms and 15 percent of no-fraud firms
relative to individuals serving on no-fraud firm boards. had boards where not one director served on any other
corporate board. The difference between fraud and no-fraud
Surprisingly, on average, 11 percent of fraud firms’ board firms was not statistically significant.
members had accounting or finance expertise as compared
to 9 percent for the no-fraud firms, a difference that was The results in the two sub-periods (1991-1999 and 2001-
statistically significant (p-value = 0.052). More than half 2004) were generally consistent with those reported above.
of the firms in both the fraud and no-fraud groups had at Differences in director tenure were only statistically
least one accounting or financial expert on the board (57 significant for the 1991-1999 sub-period (p-value = 0.029).

www.co s o.o rg


The length of board tenure was not statistically different that a board would have no members who sit on the board of
between fraud and no-fraud firms for 2001-2004. Also, the another firm was lower in the 2001-2004 sub-period, for both
average percentage of directors with accounting or finance fraud and no-fraud firms.
expertise was higher for fraud firms (12 percent of the fraud
firm board) than for no-fraud firms (8 percent of the no-fraud Stock Ownership
firm board) in the 1991-1999 sub-period (p-value = 0.017).
We obtained data about the extent of company stock owned
The percentage of boards with at least one director with by directors and officers of the company. Stock ownership
accounting or financial expertise was greater in the latter information was available for 196 of the 203 pairs of fraud
period, for both fraud and no-fraud firms. Also, the chance and no-fraud firms. This information is reported in Table 13.

Table 12. individual Director Characteristics (Means)
# of Pairs

# of Pairs

# of Pairs
No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud
of Firms

of Firms

of Firms
Fraud —

Fraud —

Fraud —
Sample

Sample

Sample

Sample

Sample

Sample
p-value

p-value

p-value
Fraud

Fraud

Fraud
Director age 203 53.9 54.3 -0.4 113 54.2 53.6 0.6 52 53.6 55.3 -1.7
Director tenure on board 203 6.7 7.7 -1.0 .010 113 6.7 7.8 -1.1 .029 52 6.9 8.0 -1.1
(in years)
Percentage of board 203 11% 9% 2% .052 113 12% 8% 4% .017 52 13% 11% 2%
members with accounting
or finance expertise
Percentage of companies 203 57% 51% 6% 113 56% 47% 9% 52 67% 60% 7%
with at least one
accounting or finance
expert on board
Average number of other 203 1.1 0.9 0.2 113 1.0 0.9 0.1 52 1.1 0.9 0.2
directorships held
by board members
Percentage of companies 203 16% 15% 1% 113 19% 22% -3% 52 13% 10% 3%
where not one member
of the board held any
other directorships

(two-tailed).

Directors and officers owned a significant percentage of the 15 percent of the stock for fraud firms as compared to 13
stock of both the fraud and no-fraud firms (23 percent and percent for no-fraud firms. None of the differences was
22 percent of outstanding statistically significant. The results for the two sub-periods
common shares, There was no are consistent with the full sample results.
respectively). On average, difference in stock
the highest-ranking officer ownership held by Board Chair and CeO Age and Tenure
owned 9 percent of the officers and directors
stock for both groups, and between fraud and We gathered data about certain characteristics of the
the largest stockholder no-fraud firms. individuals serving as board chair and as CEO. The results
among the officers are reported in Table 13. We collected data about the type
and directors owned of director serving as the chairman of the board for 182 of

www.co s o.o rg


the 203 pairs of fraud and no-fraud firms. The chairman of We found that, on average, the CEO was approximately 51
the board was an inside director in at least 70 percent of years old for both fraud and no-fraud firms. CEO tenure,
both fraud and no-fraud firms (75 percent of fraud firms which reflects the number of years the individual had
and 70 percent of no-fraud firms had an inside director as served as CEO of the firm, was approximately 10 years for
chairman). This likely reflects the prevalence in the U.S. of both fraud and no-fraud firms (9.4 years for fraud firms, 10.2
assigning both the position of CEO and board chair to the years for no-fraud firms). These results were not statistically
same individual. Interestingly, the percentage of firms whose different for the full sample. However, the average age of
chairman of the board was a grey director was 11 percent CEOs in the 2001-2004 sub-period was 49.9 years old for
for fraud firms as compared to 19 percent for no-fraud firms, fraud firms as compared to 53.2 years old for no-fraud firms.
a difference that is statistically significant (p-value = 0.039). The difference was statistically significant (p-value = 0.051).
That result was also statistically significant for the 1991-1999 Similarly, the average CEO tenure was statistically lower for
sub-period (p-value = 0.046). fraud firms relative to no-fraud firms for the 2001-2004 sub-
period (p-value = 0.098).

Table 13. Stock Ownership by Directors and Officers; Board Chair and CeO Traits
(Means)
# of Pairs

# of Pairs

# of Pairs
No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud
of Firms

of Firms

of Firms
Fraud —

Fraud —

Fraud —
Sample

Sample

Sample

Sample

Sample

Sample
p-value

p-value

p-value
Fraud

Fraud

Fraud
Stock owned by 196 23% 22% 1% 107 24% 23% 1% 52 18% 21% -3%
directors and officers
Stock owned by the 196 9% 9% 0% 107 11% 10% 1% 52 7% 7% 0%
highest-ranking officer
Stock owned by the 196 15% 13% 2% 107 17% 14% 3% 52 12% 13% -1%
largest holder among
officers and directors
Type of board chair:
inside director 182 75% 70% 5% 99 80% 76% 4% 48 64% 56% 8%
grey director 182 11% 19% -8% .039 99 7% 16% -9% .046 48 17% 25% -8%
Outside director 182 14% 11% 3% 99 13% 8% 5% 48 19% 19% 0%
CeO age (in years) 203 50.7 51.4 -0.7 113 51.7 50.6 1.1 52 49.9 53.2 -3.3 .051
CeO tenure (in years) 203 9.4 10.2 -0.8 113 10.1 10.2 -0.1 52 8.3 11.1 -2.8 .098

(two-tailed).

www.co s o.o rg


Number of Board Meetings Per Year Blockholders

We gathered data Often an individual or entity owns a significant portion of a
Boards of fraud firms met about the number company’s common shares. These are generally referred to
significantly more often of board meetings as “blockholders.” Consistent with corporate governance
than boards of no-fraud held during the year. literature, we defined an outside blockholder as an individual
firms. This difference may That information is or an entity that owned five percent or more of the firm’s
reflect the fact that fraud
reported in Table 14. stock. We gathered data about the extent of blockholder
firms often experienced
Perhaps surprisingly, ownership, which also is reported in Table 14.
financial stress preceding
the fraud period, which boards of fraud firms
precipitated additional met significantly more We found that approximately two-thirds of both fraud and
board meetings. often (7.7 meetings no-fraud firms had an outside blockholder who was not a
per year) than boards director (67 percent of fraud firms and 74 percent of no-fraud
of no-fraud firms (6.6 firms). Also, 23 percent of fraud and 24 percent of no-fraud
meetings per year) firms had an outside blockholder who was a director.
(p-value = 0.001). There was no difference between fraud
and no-fraud firms in the average number of board meetings During the 1991-1999 sub-period, fraud companies were
for the 1991-1999 sub-period, but fraud firms had statistically significantly less likely to have an outside blockholder who
more board meetings than no-fraud firms for the 2001-2004 was not a director. Fifty-eight percent of fraud firms had a
sub-period (p-value = 0.005). These differences may reflect blockholder who was not a director, while 75 percent of no-
the fact that fraud firms often experienced financial stress, fraud firms had a blockholder who was not a director. That
perhaps precipitating additional board meetings. difference was statistically significant (p-value = 0.009). That
difference did not continue for the 2001-2004 sub-period.
Director Turnover
internal Audit
As shown in Table 14, the number of directors who left the
board during the first fraud year was generally quite small (an Requirements to disclose the existence of an internal audit
average of 0.2 directors left fraud firm boards as compared function did not exist for the entire period of the study. We
to an average of 0.4 directors leaving no-fraud boards), but identified disclosures (some were voluntary) of an internal
this difference was statistically significant (p-value = 0.045). audit function for approximately 30 percent of both the
Fifteen percent of fraud firms had a director leave the board fraud and no-fraud firms during the full sample time period.
during the first fraud year, while 25 percent of the no-fraud Disclosure of an internal audit group was much more likely
firms had a director leave the board during the comparable in the 2001-2004 sub-period than in the 1991-1999 sub-period
year, and this difference for both fraud and no-fraud firms. Less than 20 percent of
was statistically significant During the 1st fraud firms voluntarily disclosed having an internal audit function
(p-value = 0.018). Thus, during year, director turnover in the 1991-1999 sub-period, while about 50 percent of firms
the first fraud year, director was lower for fraud disclosed having an internal audit function in the 2001-2004
turnover was lower for fraud firms than for sub-period.
firms than for no-fraud firms. no-fraud firms.

There was no difference
between fraud and no-fraud firms in the number of directors
who left the board during the first fraud year in either of
the two sub-periods. However, during the 1991-1999 sub-
period, 13 percent of fraud and 23 percent of no-fraud firms
had a director leave the board during the first fraud year, a
difference that was statistically significant (p-value = 0.058).
During the 2001-2004 sub-period, the same percentage (25
percent) of fraud and no-fraud firms had a director leave the
board during the first fraud year.

www.co s o.o rg


Table 14. Other Full Board and governance Characteristics (Means)

# of Pairs

# of Pairs

# of Pairs
No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud
of Firms

of Firms

of Firms
Fraud —

Fraud —

Fraud —
Sample

Sample

Sample

Sample

Sample

Sample
p-value

p-value

p-value
Fraud

Fraud

Fraud
Number of board 183 7.7 6.6 1.1 .001 99 7.6 6.6 1.0 49 8.0 6.2 1.8 .005
meetings per year
Number of directors who 203 0.2 0.4 -0.2 .045 113 0.2 0.4 -0.2 52 0.4 0.3 0.1
left the board during the
first fraud year
Percentage of companies 203 15% 25% -10% .018 113 13% 23% -10% .058 52 25% 25% 0%
that had a director leave
during the first fraud year
Percentage of companies 196 67% 74% -7% 107 58% 75% -17% .009 52 77% 77% 0%
w/ an outside blockholder
who was not a director
Percentage of companies 196 23% 24% -1% 107 21% 19% 2% 52 21% 29% -8%
w/ an outside blockholder
who was a director
disclosing existence of an
internal audit function

(two-tailed).

Audit Committee Characteristics size of audit committees for both fraud and no-fraud firms
was about three members. Consistent with the Blue Ribbon
Audit committees are generally responsible for the board’s Committee (BRC) Report recommendation that audit
oversight of the financial reporting process. We gathered committees have at least three members (a recommendation
extensive information about selected audit committee subsequently adopted by the stock exchanges) 70 percent of
characteristics. Among the set of 203 pairs of fraud and the fraud firms and 79 percent of no-fraud firms maintained
no-fraud firms, 193 fraud firms had an audit committee an audit committee with at least three members. The
and 199 no-fraud firms had an audit committee. So that we difference was statistically significant (p-value = 0.044).
could continue to have a set of fraud companies similarly
paired with no-fraud companies, we reduced the size of the On average, the audit committees of fraud firms had
sample for our analysis of audit committee characteristics more inside directors (5 percent of the audit committee
to 188 pairs of fraud and no-fraud firms that both had audit membership) than the audit committees of no-fraud firms
committees. This same reasoning applies to the other (2 percent), and that difference was statistically significant
variables where the sample size was less than 203. (p-value = 0.008). Likewise, 87 percent of fraud firms had
no insiders on the audit committee, versus 94 percent of
existence, Size, independence, no-fraud firms. This difference was significant
and Meeting Frequency (p-value = 0.014). Sixty-four percent of the
Few differences fraud firms and 67 percent of no-fraud
As reported in Table 15, 95 percent of fraud existed between firms maintained an audit committee that
firms maintained an audit committee, while 98 audit committees was composed entirely (100 percent of the
of fraud firms and audit committee membership) of outside,
percent of no-fraud firms maintained an audit
no-fraud firms.
committee (the difference was statistically independent directors. This difference was not
significant (p-value = 0.066)). The average statistically significant.

www.co s o.o rg


In both sub-periods, there were no differences between percent, respectively (p-value = 0.037)). Likewise, no-fraud
fraud and no-fraud firms in audit committee existence firms in the 1991-1999 sub-period were more likely to have
or average audit committee size. Only in the 1991-1999 no insiders on the audit committee (p-value = 0.048). Audit
sub-period, the percentage of fraud firms with an audit committees were more independent in the 2001-2004 sub-
committee composed of at least three members was period than in the 1991-1999 sub-period for both fraud and
significantly lower than for no-fraud firms (p-value = 0.050). no-fraud firms.

Relating to audit committee independence, the only Finally, the average number of audit committee meetings per
statistically significant difference between fraud and no- year was 3.5 for fraud firms and 3.7 for no-fraud firms, and
fraud firms in the sub-periods was that fraud firms had about half of all audit committees met four or more times per
more inside directors than no-fraud firms, but this result year. There were no significant differences between fraud
only held for the 1991-1999 sub-period (8 percent and 3 and no-fraud firms in the full sample or in either sub-period.

Table 15. Audit Committee existence, Size, independence, and Meeting Frequency (Means)
# of Pairs

# of Pairs

# of Pairs
No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud
of Firms

of Firms

of Firms
Fraud —

Fraud —

Fraud —
Sample

Sample

Sample

Sample

Sample

Sample
p-value

p-value

p-value
Fraud

Fraud

Fraud
existence of an 203 95% 98% -3% .066 113 92% 96% -4% 52 98% 100% -2%
audit committee
Number of individuals 188 3.1 3.2 -0.1 100 3.0 3.2 -0.2 51 3.4 3.4 0
on audit committee
Percentage of companies 188 70% 79% -9% .044 100 61% 74% -13% .050 51 92% 92% 0%
with an audit committe
composed of at least
three members
Type of audit committee
board member:
inside director 188 5% 2% 3% .008 100 8% 3% 5% .037 51 3% 1% 2%
grey director 188 11% 11% 0% 100 13% 15% -2% 51 8% 7% 1%
Outside director 188 84% 87% -3% 100 79% 82% -3% 51 89% 92% -3%
Percentage of companies 188 87% 94% -7% .014 100 80% 90% -10% .048 51 96% 98% -2%
with an audit committe
consisting of no
inside directors
Percentage of companies 188 64% 67% -3% 100 53% 56% -3% 51 76% 78% -2%
whose audit committee
consisted entirely of
outside directors
Number of audit committee 170 3.5 3.7 -0.2 93 2.6 2.9 -0.3 47 5.1 4.9 0.2
meetings per year
Percentage of companies 170 45% 51% -6% 93 23% 31% -8% 47 81% 85% -4%
where audit committee
met at least four times
per year
(two-tailed).

www.co s o.o rg


Surprisingly, the percentage
of individuals on audit
Financial expertise and no-fraud firms had accounting or finance
committees with finance
governance expertise or accounting expertise expertise. That difference was statistically
significant for the full sample (p-value =
was significantly higher
We also gathered data about the 0.053) and for the 1991-1999 sub-period
for fraud firms than
expertise of individuals who served on no-fraud firms for the full (p-value = 0.006). Similarly, in the 1991-1999
the audit committee. A board member sub-period, 33 percent of the fraud firms
sample and the 1991-1999
was coded as having accounting or and only 20 percent of no-fraud firms had
sub-period.
finance expertise if he or she had current at least one audit committee member
or prior experience as a CFO, CPA, with accounting or finance expertise, a
controller, or vice president of finance. significant difference (p-value = 0.037). Both
Results are provided in Table 16. On average, 14 percent fraud and no-fraud firms were more likely to have at least one
of audit committee members for fraud and 10 percent for financial expert on the audit committee in more recent years.

Table 16. Other Audit Committee Characteristics (Means)
# of Pairs

# of Pairs

# of Pairs
No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud
of Firms

of Firms

of Firms
Fraud —

Fraud —

Fraud —
Sample

Sample

Sample

Sample

Sample

Sample
p-value

p-value

p-value
Fraud

Fraud

Fraud
Percentage of audit 188 14% 10% 4% .053 100 15% 7% 8% .006 51 16% 16% 0%
committee members with
accounting or
finance expertise
Percentage of audit 188 34% 28% 6% 100 33% 20% 13% .037 51 43% 43% 0%
committees with at least
one accounting
or finance expert
Average number of director 188 1.2 1.1 .1 100 1.2 1.0 .2 51 1.2 1.0 .2
positions held by audit
committee members
on other company boards
Type of audit
committee chair:
inside director 43 2% 0% 2% 17 6% 0% 6% 16 0% 0% 0%
grey director 43 5% 9% -4% 17 6% 6% 0% 16 6% 19% -13%
Outside director 43 93% 91% 2% 17 88% 94% -6% 16 94% 81% 13%
Percentage of audit 43 21% 14% 7% 17 18% 0% 18% .070 16 19% 31% -12%
committees whose chair
had accounting or
finance expertise
Percentage of companies 203 32% 35% -3% 113 3% 1% 2% 52 81% 94% -13% .038
that included audit
committee report
or charter in proxy
Percentage of audit 188 77% 75% 2% 100 81% 73% 8% 51 73% 75% -2%
committee members who
joined audit committee
after CeO appointed
Percentage of audit 43 67% 67% 0% 17 59% 59% 0% 16 81% 88% -7%
committees whose chair
joined board after
CeO appointed
(two-tailed).

www.co s o.o rg


Because experience serving as a director might impact an board committee. For both fraud and no-fraud firms, at least
individual’s effectiveness as a board member, we collected two-thirds of audit committee members and chairs were
data about the average number of director positions held appointed after the current CEO assumed his or her position,
on other company boards (other than the relevant fraud or with differences not statistically different in the full sample
no-fraud firms) by audit committee members. We found that or either sub-period.
average was similar for fraud and no-fraud firms (1.2 and 1.1
other directorships held by audit committee members for Compensation Committee Characteristics
fraud and no-fraud firms, respectively).
We gathered information about several characteristics
Audit Committee Chair, Charter, of the companies’ compensation committees. The
and Committee Appointment Process analysis of this information is provided in the sections that
follow. We analyzed compensation committees because
The overwhelming majority of audit committee chairs were compensation, especially executive compensation, can
outside directors (93 percent for fraud and 91 percent for affect management’s motivation to commit fraud.
no-fraud firms, respectively; difference not statistically
significant).20 Overall, relatively few audit committee chairs existence, Size, independence,
had accounting or finance expertise, with no significant and Meeting Frequency
difference between fraud and no-fraud firms. However, for
the 1991-1999 sub-period, 18 percent of the audit committee As reported in Table 17, fraud firms were significantly less
chairs of fraud firms and zero percent of audit committee likely to have maintained a
chairs of no-fraud firms had accounting or finance expertise; compensation committee
A large majority of
this difference was statistically significant (p-value = 0.070). than no-fraud firms. While
both fraud and
88 percent of fraud firms no-fraud firms had
Thirty-two percent of the fraud and 35 percent of no-fraud maintained a compensation compensation
firms in the full sample included the audit committee report committee, 94 percent of committees, and there
or charter in the proxy statement (difference not statistically no-fraud firms maintained a were relatively few
significant). Only three percent of fraud firms and one compensation committee. differences in the
percent of no-fraud firms included an audit committee report That difference was characteristics of
or charter in the proxy during 1991-1999. However, during significant (p-value = 0.058). those committees
the 2001-2004 sub-period, 81 percent of fraud firms and 94 The average compensation between fraud and
percent of no-fraud firms included an audit committee report committee size was 3.1 no-fraud firms.
or charter in the proxy statement. The difference during the members for fraud firms
2001-2004 sub-period between fraud and no-fraud firms was and 3.2 members for no-
statistically different (p-value = 0.038). Charters became fraud firms. Sixty-nine percent of the fraud and 75 percent
much more common in proxy statements as a result of a of no-fraud firms maintained a compensation committee
BRC Report recommendation that was adopted by the stock with at least three members. These differences were not
exchanges. statistically significant.

We also gathered data on As for the composition of compensation committees, 85
in the 2001-2004 whether audit committee percent of fraud firm compensation committee membership
sub-period, fraud firms members, including the and 88 percent of no-fraud firm compensation committee
were less likely than committee chair, joined membership consisted of outside directors. This difference
no-fraud firms to the board after the current was not statistically significant. Eighty-nine percent of
include an audit CEO (at the time the fraud the fraud firms and 90 percent of no-fraud firms had a
committee charter or
began) was appointed. To compensation committee with no insiders; this difference
report in the proxy.
the extent that a greater was not statistically significant. Also, 66 percent of the fraud
percentage of committee and 70 percent of no-fraud firms maintained a compensation
members joined the board committee that was composed entirely of outside,
after the current CEO was appointed, the current CEO may independent directors. This difference was not statistically
have played a greater role in their appointment and, as a significant.
result, may have had greater influence over the respective

20
Only 43 pairs of firms disclosed the name of the chair of the audit committee. Given the small sample size,
results related to the audit committee chair should be interpreted with caution. www.co s o.o rg


Table 17. Compensation Committee existence, Size, independence,
and Meeting Frequency (Means)
# of Pairs

# of Pairs

# of Pairs
No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud
of Firms

of Firms

of Firms
Fraud —

Fraud —

Fraud —
Sample

Sample

Sample

Sample

Sample

Sample
p-value

p-value

p-value
Fraud

Fraud

Fraud
existence of a 203 88% 94% -6% .058 113 84% 90% -6% 52 94% 98% -4%
compensation committee
Number of individuals on 170 3.1 3.2 -0.1 88 3.0 3.1 -0.1 48 3.3 3.1 0.2
Percentage of companies 170 69% 75% -6% 88 69% 76% -7% 48 77% 67% 10%
with a compensation
committee composed of
at least three members
Type of compensation
committee member:
inside director 170 4% 3% 1% 88 4% 4% 0% 48 6% 3% 3%
grey director 170 11% 9% 2% 88 12% 10% 2% 48 8% 8% 0%
Outside director 170 85% 88% -3% 88 84% 86% -2% 48 86% 89% -3%
Percentage of companies 170 89% 90% -1% 88 90% 88% 2% 48 85% 92% -7%
with a compensation
committee consisted of
no inside directors
with compensation
committee consisted
entirely of outside directors
Number of compensation 153 3.3 3.2 0.1 80 3.1 3.0 0.1 46 3.5 3.4 0.1
committee meetings
per year
where compensation
committee met at least
two times per year
(two-tailed).

The average number of compensation committee meetings audit committees, there was no notable improvement in
per year was 3.3 for fraud firms and 3.2 for no-fraud firms. compensation committee independence across the two
Also, 73 percent of fraud firm compensation committees and sub-periods.
74 percent of no-fraud firm compensation committees met at
least two times per year. Neither difference was significant. Financial expertise and governance expertise

While for the full sample, fraud firms were less likely to have Because components of executive compensation are
had a compensation committee, this difference between sometimes based on financial statement outcome measures
fraud and no-fraud firms for each of the two sub-periods (e.g., bonus based on earnings), we examined the extent
was not statistically significant. In both sub-periods, there to which compensation committees are composed of
were no other significant differences between fraud and individuals with accounting or finance expertise. Table 18
no-fraud firms with respect to compensation committee reports that, on average, nine percent of compensation
characteristics in Table 17. Finally, unlike the case for committee members for fraud firms and five percent of

www.co s o.o rg


no-fraud firms’ compensation committee members had Compensation Committee Chair
accounting or finance expertise. Also, 22 percent of the and Committee Appointment Process
fraud firms and 14 percent of no-fraud firms had at least
one member with accounting or finance expertise on the The overwhelming majority of compensation committee
compensation committee. Both differences were statistically chairs were outside directors (89 percent for fraud and 97
significant (p-values = 0.012 and 0.034, respectively). percent for no-fraud firms; not statistically significant).21
The average number of other director positions held by Virtually none of the compensation committee chairs had
compensation committee members was similar for fraud and accounting or finance expertise. This finding is interesting
no-fraud firms (1.3 and 1.2, respectively). given the accounting and financial implications of firm
compensation practices and the associated fraud risk that
Similar to the full sample results, the percentage (10 percent) certain compensation practices may entail.
of compensation committee members having accounting
or finance expertise was statistically higher for fraud firms Seventy-five percent of the fraud firm compensation
than the percentage (4 percent) for no-fraud firms in the committee members joined the board after the CEO was
1991-1999 sub-period (p-value = 0.018). Similarly, in the appointed as compared to 70 percent for no-fraud firms.
1991-1999 sub-period, 23 percent of fraud firms had at least That difference was not statistically significant. There also
one accounting or finance expert on the compensation was no statistical difference between fraud and no-fraud
committee, versus 11 percent of no-fraud firms (p-value firms in whether the compensation committee chair joined
= 0.045). The differences between the fraud and no-fraud the board after the CEO assumed his or her position.
firms in the 2001-2004 sub-period related to accounting and
finance expertise on the compensation committee were not There was no statistically significant difference between
significant. There was no statistically significant difference the fraud and no-fraud firms, in either sub-period, in the
between the fraud and the no-fraud firms in the average percentage of outside directors serving as chair of the
number of other directorships held by compensation compensation committee or in the percentage of committee
committee members in either sub-period. chairs with accounting or finance expertise. Compensation
committee members of fraud firms were significantly more
likely than no-fraud firms to have joined the board after the
Surprisingly, the percentage CEO assumed his or her position in the 1991-1999 sub-period
of individuals on compensation (80 percent for fraud firms compared to 67 percent for no-
committees with finance or
fraud firms (p-value = 0.031)). This result did not continue in
accounting expertise was
the 2001-2004 sub-period. There was no difference, in either
significantly higher for fraud sub-period, between fraud and no-fraud firms as it relates to
firms than no-fraud firms for the compensation committee chair joining after the CEO was
the full sample and the
appointed.
1991-1999 sub-period.

21
Only 38 pairs of firms disclosed the name of the chair of the compensation committee. Given the small
sample size, results related to the compensation committee chair should be interpreted with caution. www.co s o.o rg


Table 18. Other Compensation Committee Characteristics (Means)

# of Pairs

# of Pairs

# of Pairs
No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud
of Firms

of Firms

of Firms
Fraud —

Fraud —

Fraud —
Sample

Sample

Sample

Sample

Sample

Sample
p-value

p-value

p-value
Fraud

Fraud

Fraud
Percentage of 170 9% 5% 4% .012 88 10% 4% 6% .018 48 10% 6% 4%
members with accounting
Percentage of 170 22% 14% 8% .034 88 23% 11% 12% .045 48 27% 15% 12%
compensation committees
w/ at least one accounting
or finance expert
Average number of 170 1.3 1.2 0.1 88 1.3 1.3 0.0 48 1.3 1.0 0.3
director positions held by
members on other
company boards
Type of compensation
committee chair:
inside director 38 3% 0% 3% 15 0% 0% 0% 13 8% 0% 8%
grey director 38 8% 3% 5% 15 13% 0% 13% 13 0% 8% -8%
Outside director 38 89% 97% -8% 15 87% 100% -13% 13 92% 92% 0%
Percentage of 38 0% 3% -3% 15 0% 0% 0% 13 0% 8% -8%
whose chair had accounting
Percentage of 170 75% 70% 5% 88 80% 67% 13% .031 48 67% 70% -3%
members who joined
after CeO appointed
Percentage of 38 61% 68% -7% 15 67% 53% 14% 13 62% 77% -15%
whose chair joined board
after CeO appointed
(two-tailed).

Board Leadership issues process. The CEO was also the company’s founder for 27
and Related Party Transactions percent of fraud firms as compared to 22 percent for no-
fraud firms; however, that difference was not statistically
Board Leadership issues, Appointment Process, significant.
and Personal Relationships22
The founder was on the board of directors for 42 percent
As reported in Table 19, and consistent with the general of the fraud firms as compared to 36 percent for no-fraud
practice in the U.S., the CEO also served as chairman of firms, but the difference was not statistically significant.
the board in more than two-thirds of both fraud and no- Approximately 80 percent of directors joined the board
fraud firms. There was no statistically significant difference after the CEO assumed his or her position, and there was
between fraud and no-fraud firms. We also examined the no statistically significant difference between the fraud
role of the company’s founder in the firm’s governance and no-fraud firms. Finally, family relationships among

22
We attempted to gather data on nominating committee characteristics as well. However, we only had 28
pairs of observations with nominating committee data. Given this small sample, we chose not to present any
data. Notwithstanding this fact, fraud companies were less likely to have a nominating committee (p-value =
www.co s o.o rg 0.056), although this result only held in the 1991-1999 sub-period.


non-employee directors and company officers existed for 6 significant (p-value = 0.012) for the full sample and for the
percent of the fraud companies as compared to 13 percent 2001-2004 sub-period (p-value = 0.008).
for the no-fraud companies. That difference was statistically

Table 19. Board Leadership issues (Means)

No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud
# of Pairs

# of Pairs

# of Pairs
Fraud —

Fraud —

Fraud —
of Firms

of Firms

of Firms
Sample

Sample

Sample

Sample

Sample

Sample
p-value

p-value

p-value
Fraud

Fraud

Fraud
where CeO/President and
board chair were
the same individual
where CeO/President and
the company founder were
the same individual
Percentage of companies 203 42% 36% 6% 113 42% 36% 6% 52 35% 37% -2%
where the company
founder served
on the board
Percentage of board 203 80% 77% 3% 113 82% 77% 5% 52 77% 77% 0%
members who joined board
after CeO appointed
Percentage of companies 203 6% 13% -7% .012 113 8% 11% -3% 52 2% 17% -15% .008
where board had at least
one non-employee
director related to an
officer of the company

(two-tailed).

Related Party Transactions difference was not statistically significant. Related party
transactions involving the founder occurred less often in
As shown in Table 20, fraudulent financial reporting was recent years.
more likely when a firm disclosed related party transactions.
We found that 79 percent of fraud firms had disclosed Just over 50 percent of the related party transactions
a related party transaction in the proxy statement, as involved the CEO, although there was no difference between
compared to 71 percent for no-fraud firms. That difference the fraud and no-fraud firms on an overall basis or in either
was statistically significant (p-value = 0.065). However, of the two sub-periods examined. Related party transactions
that difference was not statistically involving other senior officers or

70%
significant for either sub-period. involving board members occurred
Although over approximately 50-60 percent of the
For fraud firms, 26 percent of the of fraud and no-fraud firms time, but there were no significant
related party transactions involved the disclosed related party differences between fraud and
founder, whereas 22 percent of the transactions, fraud firms no-fraud firms.
related party transactions involved were significantly more likely
to have disclosed a related
the founder for the no-fraud firms. This
party transaction than
no-fraud firms.

www.co s o.o rg


Table 20. Related Party Transactions (Means)

# of Pairs

# of Pairs

# of Pairs
No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud

No-Fraud
of Firms

of Firms

of Firms
Fraud —

Fraud —

Fraud —
Sample

Sample

Sample

Sample

Sample

Sample
p-value

p-value

p-value
Fraud

Fraud

Fraud
Percentage of companies 203 79% 71% 8% .065 113 75% 70% 5% 52 87% 77% 10%
with related party
transactions disclosed
in the proxy
Percentage of disclosed 117 26% 22% 4% 60 28% 27% 1% 35 14% 11% 3%
related party transactions
that involved the founder
Percentage of disclosed 117 51% 54% -3% 60 58% 53% 5% 35 31% 46% -15%
that involved the CeO
Percentage of disclosed 117 52% 50% 2% 60 47% 47% 0% 35 51% 51% 0%
that involved
other senior officers
Percentage of disclosed 117 61% 66% -5% 60 67% 67% 0% 35 51% 60% -9%
that involved members of
the board of directors
(two-tailed).

issues Related to the external Auditor
Auditors Associated With Fraud Companies

We obtained information from the COMPUSTAT® database about the auditor who issued the audit opinion on the last set of
audited financial statements issued during the fraud period to identify the auditor responsible for issuing the audit opinion
on those fraudulently misstated financial statements. We were able to obtain information about the nature of the auditor’s
opinion for the last fraudulently issued financial statements for 223 of the 347 fraud firms. We were able to obtain auditor
data for 247 of the no-fraud firms.23

As reported in the pie charts in Table 21, we found that the that were fraudulently misstated to determine whether
Big Six/Four audited 79 percent of the fraud companies (177 of the auditor’s report contained any modifications or
the 223 fraud companies with available auditor information) in qualifications. For the 223 fraud companies where we were
the last year of the fraud period. The next tier of four national able to obtain audit opinion data from COMPUSTAT®, we
audit firms beyond the Big Six/Four24 audited 6 percent (n = determined that 97 of those 223 audit reports (43 percent)
13) of the fraud firm financial statements, while the remaining contained unqualified auditor opinions with no explanatory
15 percent (n = 33) of fraud firm financial statements were language. An additional 125 of the 223 fraud companies’
audited by non-national firms. These percentages were financial statements (56 percent) contained an auditor’s
similar to the mix of auditor type for the 247 no-fraud firms report that included an unqualified opinion along with
where we could locate auditor information. explanatory language. Only one of the 223 auditor opinions
was qualified, and no audit opinion was issued for another of
We also reviewed information about the nature of the the 223 fraud companies examined (collectively 1 percent).
auditor’s opinion on the last set of financial statements

23
Our intent was not to compare the nature of audit opinion for a fraud firm and its similar no-fraud firm.
Instead, we were interested in comparing auditor characteristics as a whole for each group (fraud firms and
no-fraud firms). Thus, we did not need equal numbers of fraud and no-fraud firms for our auditor analysis.
24
The next tier of four national audit firms is Grant Thornton LLP, BDO Seidman LLP, Crowe Chizek and
www.co s o.o rg Company LLC, and McGladrey & Pullen LLP.


Table 21. Size of Audit Firms Issuing Reports

Fraud Companies No-Fraud Companies
Big Six/Four
auditor
15% 12%
Next tier
of 4 national 5%
6% firms
Non-national
firms

79% 83%

These results differ from the no-fraud firms. The majority No audit opinion was issued for one of the 247 no-fraud
of no-fraud firms (64 percent (n = 158) of the 247 no-fraud companies examined. See the pie charts in Table 22. More
firms where we had auditor report information) received research is needed to examine the nature of the audit
unqualified audit opinions without any explanatory report modification and to determine if there is any relation
language, while the remaining 36 percent (n = 88) received between the report modification and the nature of the fraud
unqualified opinions accompanied by explanatory language. technique employed.

Table 22. Types of Auditor Reports

Unqualified

Other

Unqualified 36%
43% with
explanatory
paragraph
56%

64%
0%

1%

www.co s o.o rg


Analysis of Auditor Reports on internal However, in nine of these instances, the auditor’s report
Control Over Financial Reporting cited a financial restatement that had already occurred, and
in the remaining case, the auditor cited amendments to the
Because of the significant time lag between the occurrence original 10-K that were filed immediately after the original
of fraud and the subsequent issuance by the SEC of an 10-K filing (due to auditor-detected issues). Thus, in all of the
AAER, only a small percentage of the fraud companies instances where the auditor concluded that controls were
had fraud periods extending into 2004 or later, the period ineffective, there had already been a financial restatement
when SOX Section 404 internal control audits became or other amendment of the 10-K. Finally, in two cases (10
mandatory for accelerated filers (effective for fiscal percent), the original Section 404 opinion indicated effective
years ended November 15, 2004 or later). We identified 24 controls, but the opinion was subsequently restated to
fraud companies (40 company years, as some companies indicate ineffective controls. In both cases, a company
had fraud in 2004, 2005, and/or 2006) with fraud periods financial restatement apparently triggered the restatement
including 2004 or later that might be subject to the Section of the auditor’s Section 404 opinion.
404 requirements, if the company was large enough to be
an accelerated filer. Of these 40 company years, 18 did not Overall, the analysis of Section 404 opinions for the 22
appear to involve accelerated filers or the company failed company years with available data indicates that the opinions
to issue a 10-K. This leaves 22 company years for analysis. indicated effective controls unless there had already been
While we do provide this analysis, we caution readers about a financial restatement or amended 10-K. Thus, it does not
drawing conclusions about the impact of Section 404 based appear that adverse Section 404 opinions were diagnostic
on this very small sample size. of future misstatements, but rather simply reflected
already detected misstatements that resulted in financial
In 10 of the 22 cases (45 percent), the Section 404 opinion restatements or amended 10-Ks. However, the small sample
indicated that the company had effective internal control size provides a very limited perspective about Section 404
over financial reporting (see the pie chart in Table 23). providing fraud detection capability. Further research is
Thus, the auditor concluded that controls were effective, warranted to determine whether there are ways to improve
even though the company was later determined to have auditors’ ability to detect internal control weaknesses that
fraudulently misstated its financial statements for this may lead to fraud in the future. Additionally, we are unable
period. In another 10 cases (45 percent), the Section to assess whether Section 404 serves as a deterrent for
404 opinion was adverse, indicating ineffective controls. management to engage in fraudulent financial reporting.

Table 23. Analysis of Section 404 Internal Control Opinions
(n = 22 company years)

Section 404
opinion indicated
e ective controls 10%

Section 404
opinion indicated
ine ective controls
Original Section 404 45%
opinion indicating
e ective controls was 45%
later restated to indicate
ine ective controls

Note: In all cases where the Section 404 opinion indicated ineffective controls, the opinion cited a financial restatement or
other amendment of the 10-K.

www.co s o.o rg


Alleged Auditor involvement in the Fraud In the remaining 51 cases where the auditor was named,
the auditor was accused of violating Rule 102(e) of the 1934
In 23 percent of the cases (78 of the 342 fraud cases where Securities Exchange Act mostly for performing an alleged
the AAERs named individuals), the external auditor was substandard audit. Out of these 51 cases, 22 involved a
named in an AAER. In five of these 78 fraud cases, two national audit firm, while 29 involved a non-national audit firm.
different audit firms were named. Thus, the data in Table 24
describe the accusations against a total of 83 audit firms.25 The relative infrequency of enforcement actions against
national firms relative to non-national firms is particularly
As indicated in Table 24, out of the 83 cases where the auditor striking, given that most of the fraud firms were audited by
was named, 32 audit firms were charged with violating a national audit firm. Table 21 indicates that 85 percent of
antifraud statutes (either Rule 10(b)-5 of the 1934 Securities the fraud companies were audited by a national audit firm,
Exchange Act or charged with aiding and abetting others in yet only 40 percent of the enforcement actions (33 of 83
a violation of Rule 10(b)-5). Of those 32 cases, 11 involved a enforcement actions) were against a national audit firm (see
national audit firm (Big Six/Four or the next tier of four national Table 24).
audit firms) and 21 involved a non-national audit firm.

Table 24. Frequency of Audit Firms Named in enforcement Actions
SeC Alleged Audit Auditors Named Number of National Number of Non-National
Firm Violation in AAeR Firms Named Firms Named

Anti-fraud statutes 32 11 21
Non-fraud statutes 51 22 29
including Rule 102(e)
Total 83 33 50
Note: There were 78 fraud cases in which the SEC named an individual at an audit firm or the audit firm itself in the AAER.
For five of the 78 cases, the SEC named individuals at two different audit firms or two different audit firms.

Auditor Changes During Fraud Period Most (60 percent) of the fraud firm auditor switches
occurred during the fraud period, while the remaining 40
To determine whether fraud companies changed auditors percent of fraud firm auditor switches occurred at the end of
just prior to or during the fraud period, we gathered the last clean financial statement period (i.e., just before the
COMPUSTAT® data to compare the name of the auditor fraud period began). Of the 47 instances of fraud companies
associated with the last clean financial statements to the switching auditors, 35 companies (74 percent) switched
name of the auditor who issued an audit report on the last from one national audit firm to another national audit firm,
fraudulent financial statements. We were able to make that five (11 percent) switched from a national audit firm to a
comparison for 184 similarly paired fraud and no-fraud firms. non-national audit firm, five (11 percent) switched from a
As reflected in the pie charts in Table 25, we found that 47 non-national audit firm to a national audit firm, and two (4
of the 184 fraud firms (26 percent) percent) switched among non-national
changed auditors between the period Twenty-six percent of fraud firms. In contrast, 16 of the 22 no-fraud
that the company issued the last clean companies switched auditors firms (73 percent) switching auditors
financial statements and the period between the issuance of the changed from one national audit firm
the company issued the last set of last clean financial statements to another national audit firm, five (23
fraudulent financial statements. In and the last set of percent) switched from a national audit
contrast, only 22 of the 184 no-fraud fraudulently misstated firm to a non-national audit firm, and one
firms (12 percent) switched auditors financial statements, while (4 percent) switched from a non-national
over that same time frame. This 12% of no-fraud firms firm to a national firm.
difference was statistically significant switched auditors during the
(p-value < 0.01). same time period.

25
The SEC commonly names an individual auditor in the AAER instead of naming the entire audit firm. For
ease of discussion, we refer to the “audit firm” to mean the employer of the named auditor or the firm itself. www.co s o.o rg


Table 25. Auditor Changes

Changed
auditors
12%
No change
26% in auditors

74%
88%

Consequences for the Company and individuals involved
We attempted to identify consequences for companies engaging in fraudulent financial reporting once the fraud was
revealed. First, we noted consequences described in the AAERs for each of the 347 fraud companies.

Table 26 presents information in the AAERs on the sanctions outside professionals (e.g., accountants, attorneys, etc.) can
imposed by the SEC against both companies and individuals be barred from practicing before the SEC, either temporarily
as a result of the fraud.26 The most common sanctions were or permanently. In 46 percent of the fraud cases, one or
cease and desist orders, officer and director and SEC bars, more outside professionals were subject to an SEC bar.
and monetary penalties. A cease and desist order compels a
party to stop engaging in certain behavior, and the recipient of SEC sanctions can involve monetary penalties, either fines
such an order can be a company or an individual. A cease and or disgorgements. Fines can
desist order is the mildest sanction that the SEC can impose be levied against companies
in a fraud case, and it was the most commonly employed and individuals, and were Fines were imposed
sanction (89 percent of the fraud companies received a imposed in 65 percent of the in 65% of the
cease and desist order). Generally, the SEC issues a cease fraud cases. A disgorgement cases, while
and desist order in addition to other sanctions. However, in involves returning monies disgorgements
29 cases, the SEC issued a cease and desist order without inappropriately received were imposed in

43%
issuing any other sanctions. as a result of the fraud. For of the cases.
example, an individual might
The SEC can bar an individual from serving as an officer or be required to disgorge a
a director of a public company, either for a period of time or bonus received based on
permanently. This is a severe sanction, as it seriously affects fraudulently reported income or the proceeds from a stock
the economic situation of an individual receiving such a bar. sale when the stock price was inflated as a result of the
In almost half of the fraud cases (47 percent), one or more fraud. Disgorgements were ordered in 43 percent of the
individuals received an officer and director bar. In addition, fraud cases.

26
Frequencies of consequences reported in this section are inherently understated given that we were
only able to identify consequences explicitly noted in an AAER or in business press articles. Given that the
business press often does not cover smaller or otherwise less visible companies, there were likely to be many
www.co s o.o rg consequences that occurred that we were unable to identify for some of our sample firms.


Table 26. Consequences Based on AAER Information
(n = 347)
100%
89

80%

65
60%
47 46 43
40%

20%

0%

Cease & Desist SEC Bar Disgorgement
O cer/Director Bar Fines-Civil

Table 27 presents information on the monetary penalties all 221 fraud cases imposed by the SEC totaled $2.74 billion.
imposed by the SEC – for cases in which the dollar amounts The mean and median disgorgement amounts were $18.1
were disclosed in the AAERs. The average fine imposed million and $195,000, respectively (maximum amount of a
by the SEC was $12.4 million, and the median was $100,000 disgorgement was $700 million). Cumulative disgorgements
(maximum fine was $750 million). The cumulative fines for across all of the 146 fraud cases totaled $2.65 billion.

Table 27. Fines and Disgorgements Based on AAeR information
(n = 347)

Number of Cumulative Mean Amount Median Amount
Description Companies Amount Paid by Paid by a Paid by a Maximum
of Penalty identified All Companies Single Company Single Company Paid

Fines and settlements 221 $2.74 billion $12.4 million $100,000 $750 million
Disgorgements 146 $2.65 billion $18.1 million $195,000 $700 million

Other Consequences for Companies was different for fraud companies relative to a similar set
of companies not engaging in fraud during the same time
To identify other consequences to the fraud companies for periods.
engaging in fraudulent financial reporting, we performed
extensive searches of electronic databases of business Recall in our earlier analysis of board governance
press articles appearing during the period beginning with characteristics that we were able to generate a sample of
the calendar year that coincides with the last year of the 203 fraud and 203 similar no-fraud firms. As discussed in
fraud and ending with the calendar year two years after the that section, we were unable to examine board governance
SEC issued the last AAER related to the fraud. variables for the full sample of 347 firms due to the lack of
available proxy statements for some firms. For our business
We also performed the search of business press articles press searches, we were able to expand our sample to 311
for the sample of no-fraud companies. This allows us to fraud and 311 no-fraud firms (we were not able to identify a
determine whether the rate of subsequent consequences similar no-fraud company in 36 cases).

www.co s o.o rg


In addition to SEC sanctions (e.g., fines, disgorgements,
cease and desist orders), companies can suffer other Fraud firms filed for bankruptcy or were
consequences either directly, or indirectly, as a result of delisted from a stock exchange significantly more
the fraud. We examined the incidence of financial failure often in the time period following the fraud than
(bankruptcy, liquidation, etc.), stock exchange delisting, their counterpart no-fraud firms in
and material asset sales for the fraud companies and the that same time period.
comparable percentages for the no-fraud sample.

As shown in Table 28, 28 percent of the fraud companies
failed (filed for bankruptcy, were liquidated, etc.) within two companies compared to 31 percent of no-fraud companies
years of the latest AAER issued by the SEC. The comparable engaged in a material asset sale (p-value < 0.001).
percentage for the no-fraud companies was 13 percent.
The probability of bankruptcy or other failure for a fraud firm All of the above metrics clearly indicate that fraud firms
was statistically larger than for a no-fraud firm (p-value < were more likely to suffer adverse financial outcomes than
0.001). Similarly, fraud firms were significantly more likely no-fraud firms. These differences were likely due to the
than no-fraud firms to be involuntarily delisted from a stock fact that companies that experienced operating difficulties
exchange (p-value < 0.001). Almost half (47 percent) of chose to engage in fraud to mask these difficulties, and to
fraud firms suffered by being delisted by a stock exchange, the direct and indirect costs associated with fraud (e.g.,
whereas only 20 percent of no-fraud firms were delisted legal fees, fines, investigations, reputation damage, loss of
during a similar time period. Finally, 62 percent of fraud personnel, loss of customers, etc.).

Table 28. Other Consequences to Companya
(n = 311)b

Fraud Percentage of No-Fraud Percentage of
Subsequent Companies Fraud Companies Companies No-Fraud Companies
Consequences Affected Affected Affected Affected p-valuec

Bankrupt, 86 28% 39 13% .001
liquidated, etc.
involuntary stock 147 47% 61 20% .001
exchange delistingd
Material asset sales 193 62% 96 31% .001
a
The consequences of the fraud were examined from the beginning of the last fraud year until two years after the end of the
year of the last AAER related to the fraud. The occurrence of these events at the no-fraud companies was examined during the
identical time period.
b
There were 311 fraud companies where a similar no-fraud company could be identified.
c
Tests of statistical differences were performed using Wilcoxon’s signed rank test.
d
Includes revocation of a firm’s registration with the SEC.
(two-tailed).

Stock Price Reaction
We identified the date of the initial disclosure of an alleged
To further examine the effect of the fraud on the company, financial statement fraud by searching for the initial press
we examined the stock price reaction for two different disclosure of a potential accounting impropriety. We took
dates related to the disclosure of the alleged financial this approach since the initial press disclosure of an alleged
statement fraud. First, we examined the stock price reaction accounting impropriety may not specifically indicate that
to the initial disclosure of the fraud. Second, we examined a fraud has occurred, given that in many instances an
the stock price reaction to the initial disclosure of an investigation has yet to be commenced. Because that
investigation by the SEC or the Department of Justice. initial disclosure may or may not suggest to the markets the

www.co s o.o rg


existence of a possible fraud, we identified a second date of disclosure, which captures the stock price reaction on the
to measure stock market reactions. That date represents next trading day following the date of disclosure (referred to
the date of the SEC’s or Department of Justice’s first public as Day +1).
disclosure of an investigation.
Table 29 provides information about the abnormal returns
We measured the stock reaction on these two different measured on each of these three days surrounding the first
disclosure dates by calculating the abnormal stock returns public disclosure of an alleged financial statement fraud.
using methodologies widely used in research to capture That table also shows the cumulative abnormal return for
unique stock reactions to disclosures of new information Days 0 and +1 on a combined basis, which is consistent with
to the capital markets. An abnormal stock return basically typical abnormal stock return research. The abnormal returns
captures the portion of the change in stock price attributable reported in Table 29 are
to the company-specific news disclosed on that date and shown in percentage form
Stock prices declined
does not include normal changes for that firm’s stock given to provide an indication of
changes in overall market conditions.27 17% on average the percentage change
(beyond normal
in stock price to the initial
market movement)
initial Disclosure of Potential disclosure of alleged
across two days
Accounting improprieties surrounding the financial statement fraud.
initial disclosure of
For each of the two different disclosure events described alleged fraud. As expected, the average
above, we measured the abnormal returns over three abnormal returns for each
different days. First, we measured the abnormal return on of the three days and
the day prior to the initial disclosure of the fraud (referred the cumulative two days
to as Day -1). Measuring stock market reactions on the day (Days 0 and +1) surrounding the first public disclosure of
prior to the date of disclosure captures any stock market an alleged fraud were negative. The p-values for each day
reaction to potential leakage of information in the day prior indicate that all of the negative abnormal stock returns
to disclosure. Next, we measured the abnormal return on were highly statistically significant. The mean abnormal
the day of disclosure (referred to as Day 0). Finally, we return on Day -1 was -1.4 percent, suggesting some market
measured the abnormal return on the day following the date reaction to potential leakage of news of an alleged fraud.

Table 29. Abnormal Stock Returns Surrounding First Public Disclosure
of Potential Accounting irregularities
Percentage Abnormal Stock Return
Day -1 Day 0 Day +1 Days 0 and +1
(n=221)a (n=213)a (n=198)a (n=215)a
Mean -1.4% -10.0% -7.3% -16.7%
Standard deviation .07 .19 .17 .23
1 quartile
st
-2.5% -17.2% -12.6% -28.6%
Median -.5% -3.3% -2.2% -11.1%
3 quartile
rd
1.4% .4% 1.5% -1.7%
t-statistic -3.21 -23.19 -16.89 -27.25
p-value .001 .0001 .0001 .0001
a
Of the 347 fraud firms, stock price information was not provided for 73 firms in the CRSP database, and we were unable
to identify a unique date of the public disclosure of the potential accounting irregularity for 15 additional firms. Finally, stock
price information for some of the days (-1, 0, or +1) was missing for between 38 and 61 firms, depending on the date of interest.
Thus, the number of firms for each of the measurement dates differed slightly.
(two-tailed).

27
We calculated abnormal returns using the Center for Research in Security Prices (CRSP) database and
the Eventus program, using the market model with an equally weighted index consistent with prior research
methodologies (see DeFond et al. (2007) and MacKinlay (1997)). We estimated the market model parameters
using a 120-day estimation window consistent with prior research methodologies (see Palmrose et al. 2004).
Given the small size of some of the fraud companies, firms were retained in the sample if 30 days or more of
stock returns were available during the 120-day estimation window. www.co s o.o rg


The mean stock price reaction on the day of disclosure (Day However, the mean abnormal returns on day 0 and +1 were
0) jumped to -10.0 percent, followed by an additional -7.3 -4.9 percent and -2.5 percent, respectively, which were both
percent return on Day +1. The cumulative average negative statistically significant. Thus, the disclosure of a government
abnormal return of -16.7 percent on Days 0 and +1 indicates investigation of alleged financial statement fraud resulted
an abnormal stock price decline of 16.7 percent over the in an average abnormal stock price decline over a two-day
two-day period surrounding the initial news of fraudulent period of 7.3 percent.
financial reporting.
It is interesting to note that these stock price declines were
initial Disclosure of SeC/Department smaller in magnitude than those surrounding the initial
of Justice investigation press disclosure of the potential accounting improprieties.
Nevertheless, the announcement of a governmental
Table 30 provides information about the investigation, while not typically
abnormal stock returns for the three days providing the initial disclosure of the
(-1, 0, and +1) surrounding the first public Stock prices declined 7% potential accounting improprieties, did
disclosure of a governmental investigation on average (beyond normal provide incremental information to the
of the potential accounting improprieties, market movement) over the market. The market may have reacted
whether that investigation was commenced two-day period surrounding to the realities of costs associated
by the SEC or by the U.S. Department of announcement of an SeC with responding to a governmental
or Department of Justice
Justice. The average abnormal stock return investigation and to the adverse
investigation about
was -0.5 percent on Day -1, but this was reputational consequences for the firm.
alleged financial
not statistically significant, suggesting statement fraud.
that the announcement of a governmental
investigation did not leak into the market
before the investigation was announced.

Table 30. Abnormal Stock Returns Surrounding First Public Disclosure
of SeC or Department of Justice investigation
Percentage Abnormal Stock Return
Day -1 Day 0 Day +1 Days 0 and +1
(n=142)a (n=142)a (n=140)a (n=143)a
Mean -.5% -4.9% -2.5% -7.3%
Standard deviation .09 .13 .11 .16
1st quartile -1.5% -8.0% -6.3% -13.6%
Median -.2% -2.2% -1.2% -4.0%
3rd quartile 1.6% .5% 1.9% -.3%
t-statistic -1.09 -10.29 -5.39 -10.96
p-value .28 .0001 .0001 .0001
a
Of the 347 fraud firms, stock price information was not provided for 73 firms in the CRSP database, and we were unable
to identify a unique date of the public disclosure of the SEC’s or Department of Justice’s investigation for 78 firms. Finally,
stock price information for some of the days (-1, 0, or +1) was missing for between 53 and 56 firms, depending on the date of
interest. Thus, the number of firms for each of the measurement dates differed slightly.
(two-tailed).

www.co s o.o rg


Other Consequences for individuals Twenty-one percent (17 percent) of the fraud CEOs (CFOs)
were criminally indicted, whereas virtually none of the no-
In addition to SEC sanctions (e.g., fines, disgorgements, SEC fraud CEOs or CFOs was criminally indicated over the same
officer and director bars, etc.) described earlier, individuals time periods (statistically significant (p-value = 0.001)). For
involved with a fraud can suffer other consequences in the fraud firms, 64 percent of the CEOs criminally indicted were
labor market. We examined turnover (including the specific convicted (41/64), whereas 75 percent of the CFOs criminally
reason for the turnover) for the CEO, CFO, chairman of the indicted were convicted (39/52). This difference likely
board, and other board members. In addition, we considered reflects the greater difficulty that CFOs have in denying that
criminal indictments and convictions of the they had any knowledge of the fraud given the
CEO and CFO. We examined consequences CFO’s responsibility for the firm’s finances.
of the fraud for individuals that occurred
between the beginning of the last fraud
80% or more of For fraud firms, approximately two-thirds of
CeOs and CFOs
year through two years after the year of turned over following
board chairs left the board, whereas only
the last AAER related to the fraud. the disclosure 25 percent of board chairs left the boards of
of the alleged fraud. no-fraud firms, which is significantly lower
Table 31 presents this information. (p-value = 0.001). In addition, 68 percent of
Because we captured information about fraud firms experienced turnover of at least
management changes and other events, one other board member as compared to 40
it was important that we contrasted the percent of no-fraud firms (p-value = 0.001). As
experience of fraud firms with a similar set of no-fraud firms. was the case with CEO and CFO turnover, resignation was
Thus, Table 31 shows results for both fraud and no-fraud the most common reason given for the departure. Fifteen
firms. We tracked similar consequences for no-fraud firms percent (32 of 211 instances) of the board chair turnover at
over the same time frame used for their related fraud firms. fraud firms was due to the board chair being terminated,
We were able to find similar no-fraud firms for 311 of the 347 whereas only 11 percent (9 of 79 instances) of the board
fraud firms. chair turnover at no-fraud firms was due to a termination.
Also, if there was turnover of the CEO, CFO, or board chair,
In most cases, turnover occurred for the CEO and CFO the board chair was most likely to be fired. Turnover of
positions of companies committing fraud. Eighty-two percent other board members at both fraud and no-fraud firms was
of the CEOs and 80 percent of the CFOs of fraud firms overwhelmingly due to resignations, but terminations of
experienced turnover. The comparable percentages for the other board members occurred in six percent of the fraud
no-fraud firms were 47 percent for CEOs and 49 percent for firms. We identified only one instance where a non-chair
CFOs, significantly lower (p-value = 0.001). A large majority of board member at a no-fraud firm was terminated.
the CEO and CFO turnover was due to resignations, although
we cannot observe how many of these resignations were
forced. Seven percent of the fraud companies experienced Approximately 20%
CEO terminations, while 59 percent experienced CEO of CeOs and CFOs of fraud
companies were
resignations. In contrast, during the same time period no-
criminally indicted, and
fraud firms terminated two percent of their CEOs, while
about two-thirds of those
21 percent of the no-fraud firm CEOs resigned. A similar indictments ultimately
pattern existed for CFO turnover at the fraud and no-fraud led to criminal
companies. That suggests that fraud revelations often result convictions.
in significantly greater management changes.

www.co s o.o rg


Table 31. Consequences to individuals
(n = 311)

Number of Percentage of Number of Percentage of
Subsequent Fraud Firms Fraud Firms No-Fraud Firms No-Fraud Firms
Consequencesa Affected Affected Affected Affected p-valueb

CeO turnover: 255 82% 147 47% .001
Firing/dismissal 23 7% 5 2%
Resignation 185 59% 66 21%
Retirement 17 6% 24 8%
Another position 14 5% 17 5%
Other 16 5% 35 11%
CFO turnover: 250 80% 153 49% .001
Other 15 5% 13 4%
CeO criminal indictment 64 21% 1 < 1% .001
CeO criminal conviction 41 13% 1 < 1% .010
CFO criminal indictment 52 17% 1 < 1% .001
CFO criminal conviction 39 13% 0 0% .010
Chairman of board turnover: 211 68% 79 25% .001
Other 12 4% 10 3%
Other board turnover: 212 68% 123 40% .001
Firing/dismissal 19 6% 1 < 1%
Other 10 3% 7 2%
a
The consequences of the fraud for individuals were examined from the beginning of the last fraud year until two years
after the year of the last AAER related to the fraud. The occurrence of these events for individuals at the no-fraud firms was
examined during the identical time period.
b
Tests of statistical significance were performed using the Wilcoxon’s signed rank test.

(two-tailed).

www.co s o.o rg


iV. Conclusion
We believe that our analysis of fraudulent financial reporting from 1998-2007 reveals several key
messages. First, the financial statement fraud problem still exists and warrants continued attention.
The SEC alleged that 347 public companies committed fraud over the ten-year period 1998-2007. The
magnitude of individual fraud cases and the size of fraud companies both increased markedly from
COSO’s 1999 report. The major accounting scandals of the early 2000s involved larger frauds and larger
companies, which contributed to the nearly $120 billion in cumulative misstatement or misappropriation
across all frauds in the ten-year period. Because the number of frauds examined in this study involving
financial reporting periods after the passage of SOX is very limited, further research is needed to
assess the effects of SOX in addressing fraud.

Second, the SEC continues to name individuals in the C-suite Fifth, fraud companies are twice as likely to change
for some alleged involvement in the fraud, even more so auditors as no-fraud firms between the last clean financial
than in the past. During 1998-2007, the CEO and/or CFO were statements and the last fraudulent financial statements.
named in an AAER in nearly 90 percent of the cases. Boards, More research is needed to fully understand the relation
auditors, and regulators need to seek additional tools to between auditor change and fraudulent financial reporting.
assess management integrity and susceptibility to fraud
pressures. Research about leadership and organizational Finally, the consequences of fraud are severe for individuals
behavior may help to provide insights about potential drivers and companies. Individuals may face civil fines, SEC bars,
of financial statement fraud. disgorgement, and criminal prosecution. Fraud companies
experience significant negative abnormal stock price
Third, revenue fraud continues to emerge as the leading type declines, and they face bankruptcy, delisting, and material
of fraud, now accounting for over 60 percent of SEC fraud asset sales at much higher rates than do no-fraud firms.
cases. Additional research into revenue fraud methods,
especially industry-specific studies, may reveal new ways to
address this risk area.

Fourth, board governance characteristics often do not differ
meaningfully between fraud and no-fraud firms. These
characteristics have been the focus of recent regulation,
thus reducing or even eliminating previous fraud/no-fraud
differences. Future research on governance processes and
the interaction of various governance mechanisms may be
needed to identify less-observable governance differences
associated with fraudulent financial reporting.

www.co s o.o rg


Call for Further Research and Analysis
COSO sponsored this study, Fraudulent Financial Reporting: 1998-2007, to provide a comprehensive analysis of fraudulent
financial reporting occurrences investigated by the U.S. Securities and Exchange Commission between January 1998 and
December 2007. This study updates our understanding of fraudulent financial reporting since COSO’s 1999 issuance of
Fraudulent Financial Reporting: 1987-1997.

COSO’s mission is to provide thought leadership through the development of comprehensive frameworks, guidance, and
research on enterprise risk management, internal control, and fraud deterrence. COSO’s efforts are designed to improve
organizational performance and governance and to reduce the extent of fraud in organizations.

COSO hopes that those involved in financial reporting will carefully consider the results reported in this study and recommit
their efforts to improve the prevention, deterrence, and detection of fraudulent financial reporting. While several insights
from this study are discussed within this document, more research is needed to better understand fraudulent financial
reporting. COSO encourages other thought leaders to creatively explore new and different ways to reduce occurrences of
fraudulent financial reporting.

COSO, 2010

www.co s o.o rg

V.

Authors
Mark S. Beasley is the Deloitte Professor of Enterprise Risk Management and
Professor of Accounting at North Carolina State University. He is the Director of NC
State’s Enterprise Risk Management (ERM) Initiative, which provides thought leadership
about ERM practices and their integration with strategy and corporate governance.
Mark currently is serving on the board for the Committee of Sponsoring Organizations
of the Treadway Commission (COSO). Mark has previously served as President of the
American Accounting Association’s Auditing Section and on several national task
forces and working groups, including the Auditing Standards Board SAS No. 99 Fraud
Task Force. His research has appeared in such journals as The Accounting Review,
Journal of Accounting Research, Contemporary Accounting Research, Auditing: A
Journal of Practice & Theory, and Accounting Horizons. He is a frequent speaker
at national and international conferences on ERM, internal controls, and corporate
governance, including audit committee practices.

Mark can be reached at: mark_beasley@ncsu.edu.

Joseph V. Carcello is the Ernst & Young Professor and co-founder and Director
of Research at the University of Tennessee’s Corporate Governance Center. He has
published in such journals as The Accounting Review, Journal of Accounting Research,
Contemporary Accounting Research, Auditing: A Journal of Practice & Theory, and
Accounting Horizons. Joe is a member of the Public Company Accounting Oversight
Board’s (PCAOB’s) Investor Advisory Group, he served on the PCAOB’s Standing
Advisory Group, he testified before the U.S. Treasury Department Advisory Committee
on the Auditing Profession, and he served on COSO’s Small Business Control Guidance
Advisory Group Task Force. Joe has served as Vice President – Finance of the
American Accounting Association, and as President of the Auditing Section of the AAA.
He has served as an expert witness in cases involving fraudulent financial reporting for
the U.S. Securities and Exchange Commission, and as an expert in evaluating corporate
governance reforms instituted as part of legal settlements of shareholder claims in
federal and state courts.

Joe can be reached at: jcarcell@utk.edu.

www.co s o.o rg


Authors - continued
Dana R. Hermanson is Dinos Eminent Scholar
Chair of Private Enterprise, Professor of Accounting,
and Director of Research in the Corporate Governance
Center at Kennesaw State University. Among the most
prolific researchers in accounting, he has published in
such journals as Contemporary Accounting Research,
Auditing: A Journal of Practice & Theory, Journal of
Accounting and Public Policy, Journal of Accounting
Literature, Accounting Horizons, Behavioral Research
in Accounting, Journal of Information Systems, and
Issues in Accounting Education. Dana is co-editor
of Accounting Horizons and was founding co-editor
of Current Issues in Auditing. He and his co-authors
received the 2008 Deloitte/AAA Wildman Medal, which
recognizes research judged “to have made or be
likely to make the most significant contribution to the
advancement of the public practice of accountancy.” Dana was a member of the NACD
Blue Ribbon Commission on Audit Committees, and his work has appeared in such outlets
as The Wall Street Journal and BusinessWeek.

Dana can be reached at: dhermans@kennesaw.edu.

Terry L. Neal is the Dennis Hendrix Professor of Accounting in the Department of
Accounting and Information Management at the University of Tennessee and a Research
Fellow at the University of Tennessee’s Corporate Governance Center. He also serves
as the director of the Ph.D. program in Accounting. Terry’s research has been published
in The Accounting Review, Contemporary Accounting Research, Auditing: A Journal of
Practice & Theory, Journal of Accounting and Public Policy, Accounting Horizons, The
International Journal of Accounting, and Corporate Governance: an international review.
He currently serves or has served on the editorial boards of The Accounting Review,
Auditing: A Journal of Practice & Theory, Accounting Horizons, and Current Issues in
Auditing. Terry also has served as an ad-hoc reviewer for several other journals including
Contemporary Accounting Research, Journal of Accounting and Public Policy, Journal of
Accounting Literature, and Issues in Accounting Education.

Terry can be reached at: tneal3@utk.edu.

Note: Beasley, Carcello, and Hermanson are co-authors of the preceding
COSO-sponsored study, Fraudulent Financial Reporting: 1987-1997,
An Analysis of U.S. Public Companies.

www.co s o.o rg


References
Advisory Committee on Smaller Public Companies (Advisory Committee). 2006. Final Report of the
Advisory Committee on Smaller Public Companies to the United States Securities and Exchange Commission.
Washington, DC: SEC.

Beasley, M. S., J. V. Carcello, and D. R. Hermanson. 1999. Fraudulent Financial Reporting: 1987-1997, An Analysis
of U.S. Public Companies. New York: COSO.

Beasley, M. S., J. V. Carcello, D. R. Hermanson, and T. L. Neal. 2009. The audit committee oversight process.
Contemporary Accounting Research 26 (Spring): 65-122.

Blue Ribbon Committee on improving the effectiveness of Corporate Audit Committees (BRC). 1999. Report
and Recommendations of the Blue Ribbon Committee on Improving the Effectiveness of Corporate Audit
Committees. New York: NYSE and NASD.

Carcello, J. V., T. L. Neal, Z. V. Palmrose, and S. Scholz. 2010. CEO involvement in selecting board members,
audit committee effectiveness, and restatements. Working paper.

Cohen, J., G. Krishnamoorthy, and A. M. Wright. 2004. The corporate governance mosaic and financial
reporting quality. Journal of Accounting Literature 23: 87–152.

DeFond, M., M. Hung, and R. Trezevant. 2007. Investor protection and the information content of annual
earnings announcements: International evidence. Journal of Accounting and Economics 43 (March): 37-67.

Deloitte LLP. 2008a. Ten Things About Financial Statement Fraud – Second Edition. New York: Deloitte Forensic.

Deloitte LLP. 2008b. Ten Things About the Consequences of Financial Statement Fraud. New York: Deloitte
Forensic.

Deloitte LLP. 2009. Ten Things About Financial Statement Fraud – Third Edition. New York: Deloitte Forensic.

DeZoort, F. T., D. R. Hermanson, D. Archambeault, and S. Reed. 2002. Audit committee effectiveness: A
synthesis of the empirical audit committee literature. Journal of Accounting Literature 21: 38–75.

MacKinlay, A. C. 1997. Event studies in economics and finance. Journal of Economic Literature 35 (March): 13-39.

National Commission on Fraudulent Financial Reporting (NCFFR). 1987. Report of the National Commission on
Fraudulent Financial Reporting. New York: AICPA.

Palmrose, Z-V., V. J. Richardson, and S. Scholz. 2004. Determinants of market reactions to restatement
announcements. Journal of Accounting and Economics 37 (September): 59-89.

Ramamoorti, S. 2008. The psychology and sociology of fraud: Integrating the behavioral sciences component
into fraud and forensic accounting curricula. Issues in Accounting Education 23 (November): 521-533.

Securities and exchange Commission (SEC). 2003. Report Pursuant to Section 704 of the Sarbanes-Oxley Act
of 2002. Washington, DC: SEC.

www.co s o.o rg


Notes

www.co s o.o rg

Committee of Sponsoring Organizations
of the Treadway Commission

www.co s o.o rg

F R A U D U L e N T
F i N A N C i A L
R e P O R T i N g
1 9 9 8 - 2 0 0 7
A n A n a l ys i s of U.S. P ublic Companies

Committee of Sponsoring Organizations of the Treadway Commission

www.co s o.o rg

990023HI

Cosofraudstudy2010

More Related Content

Similar to Cosofraudstudy2010 (20)

More from ACTUS Foundation for Financial Research (20)

Recently uploaded (20)

Cosofraudstudy2010