Description
Prior research suggests that firms manipulate earnings through accruals to achieve
certain reporting objectives. Recently, especially following the Sarbanes-Oxley (SarbOx) Act,
researchers have turned their attention to real account manipulation as an alternative. However, there
is no evidence on whether the likelihood of being detected by outsiders is different for firms using
these alternative manipulation methods. The purpose of this paper is to examine this research question
in the context of seasoned equity offerings (SEOs).
Accounting Research Journal
Real and accrual-based earnings management and its legal consequences: Evidence
from seasoned equity offerings
Salma Ibrahim Li Xu Genese Rogers
Article information:
To cite this document:
Salma Ibrahim Li Xu Genese Rogers, (2011),"Real and accrual-based earnings management and its legal
consequences", Accounting Research J ournal, Vol. 24 Iss 1 pp. 50 - 78
Permanent link to this document:http://dx.doi.org/10.1108/10309611111148779
Downloaded on: 24 January 2016, At: 21:13 (PT)
References: this document contains references to 31 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 2129 times since 2011*
Users who downloaded this article also downloaded:
Tony Mortensen, Richard Fisher, (2011),"The meaning of cash in the context of alternative accounting
standards: IFRS convergence and classification decisions", Accounting Research J ournal, Vol. 24 Iss 1 pp.
23-49http://dx.doi.org/10.1108/10309611111148760
Magdy Farag, Rafik Elias, (2011),"Relative audit fees and client loyalty in the audit market", Accounting
Research J ournal, Vol. 24 Iss 1 pp. 79-93http://dx.doi.org/10.1108/10309611111148788
Arifur Khan, Paul Mather, (2013),"The value of executive director share ownership and discretionary
accruals", Accounting Research J ournal, Vol. 26 Iss 1 pp. 35-55http://dx.doi.org/10.1108/
ARJ -02-2012-0011
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Real and accrual-based earnings
management and its legal
consequences
Evidence from seasoned equity offerings
Salma Ibrahim
Department of Accounting and Finance, Kingston University,
Kingston upon Thames, UK
Li Xu
Department of Accounting and Finance, Southern Illinois University,
Carbondale, Illinois, USA, and
Genese Rogers
Department of Accounting and Finance, Morgan State University,
Baltimore, Maryland, USA
Abstract
Purpose – Prior research suggests that ?rms manipulate earnings through accruals to achieve
certain reporting objectives. Recently, especially following the Sarbanes-Oxley (SarbOx) Act,
researchers have turned their attention to real account manipulation as an alternative. However, there
is no evidence on whether the likelihood of being detected by outsiders is different for ?rms using
these alternative manipulation methods. The purpose of this paper is to examine this research question
in the context of seasoned equity offerings (SEOs).
Design/methodology/approach – First, the authors compare SEOs to a matched sample of
non-SEOs to document income-increasing manipulation. Next, they identify SEOs that prompt
lawsuits and compare sued and non-sued ?rms to determine whether using a particular method of
manipulation is more likely to be detected and associated with litigation.
Findings – The authors ?nd evidence of income-increasing accrual and real manipulation for SEOs
in the year prior to the offering in the pre-SarbOx period, and ?nd some evidence of a shift to real
account manipulation post-SarbOx. The authors examine the subsequent litigation pattern of these
SEOs, and ?nd that ?rms that are subsequently sued have a higher prevalence of income-increasing
discretionary accruals when the lawsuit allegations involve accounting issues. Following SarbOx,
investors are paying less attention to accrual manipulation through accounts receivable and there is
more scrutiny of real account manipulation.
Originality/value – The implication in this paper is that ?rms that engage in income-increasing
earnings management are more likely to be sued when they engage in accrual manipulation while
other forms of manipulation may be less understood. This ?nding is important to investors and
regulators.
Keywords United States of America, Accrual manipulation, Real manipulation, Earnings management,
Seasoned equity offerings, Sarbanes-Oxley, Litigation
Paper type Research paper
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
Data are available from the authors upon request.
ARJ
24,1
50
Accounting Research Journal
Vol. 24 No. 1, 2011
pp. 50-78
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309611111148779
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
1. Introduction
Prior research suggests that managers of ?rms manipulate earnings through accruals
under various situations, including around public equity offerings. Recently, especially
following the Sarbanes-Oxley (SarbOx) Act of 2002, researchers have turned their
attention to real account manipulation as an alternative to accrual manipulation,
e.g. through eliminating discretionary expenses, such as advertising or research and
development expenditures, or through accelerating the timing of sales through
increased price discounts or more lenient credit terms, or through reporting lower costs
of goods sold through increased production (Cohen et al., 2008). However, there is no
evidence on whether the likelihood of the manipulation being detected by investors and
outsiders, and of the ?rm facing litigation is different when employing these
alternative manipulation methods. This study addresses this gap and speci?cally
answers the following question: whether the likelihood of litigation varies with the
method(s) used to reach earnings goals around these stock issues. This research
question is addressed by examining the differences in manipulation behavior between
?rms that are subject to litigation and those that are not, in the seasoned equity
offering (SEO) setting. We also examine the effect of the SarbOx Act; the passage of the
Act was in part due to the increase in earnings management through accrual
manipulation and was aimed at reducing its prevalence. Whereas the Act has reduced
accrual manipulation (Cohen et al., 2008), prior evidence suggests that ?rms have
shifted to real account manipulation. Whether the stakeholders are able to see through
this is an important question for investors and other stakeholders of public ?rms,
as well as academicians and public policy makers.
We argue that the dif?culty for outside investors to detect earnings management
varies with the alternative methods used to manipulate earnings. We expect that,
for all ?rms that engage in earnings management around equity offerings, earnings
manipulation through accruals is more likely to be detected than managing other
components of the income statement and balance sheet. We propose to explore the role
of alternative manipulation in litigation over equity offerings.
As in earlier work, we ?rst document the presence of real and accrual earnings
management in the equity offering setting using measures of discretionary/abnormal
accruals, expenditures and cash ?ows. We focus on SEOs as our test sample. This sample
has been frequently used in earnings management studies, since it is argued that there is
both the opportunity and the incentive by management to engage in income-increasing
manipulationaroundSEOs, e.g. Teohet al. (1998a), Rangan(1998) and Shivakumar (2000),
in order to increase the market price of the stock at the time of the offering.
The SarbOx Act of 2002 addresses internal control issues of ?rms andestablishes rules
for auditors and management to limit the probability of earnings management through
accruals. We examine the effect of SarbOx on earnings management behavior and extend
the research that shows that earnings management has shifted fromaccrual management
to real account management (Cohen et al., 2008) after the passage of SarbOx.
We next identify offers that prompt lawsuits from our equity offering population
and compare sued and non-sued ?rms to examine whether there are signi?cant
differences in the discretionary or abnormal component of accruals and real accounts
for these two groups. We argue that it is harder to prove that real account manipulation
has occurred as management and auditors can contend that these abnormalities are
normal changes required during the course of operations.
Real and accrual-
based earnings
51
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This paper contributes to the accounting and ?nance literature in several ways: ?rst,
to the best of our knowledge, our paper is the ?rst attempt to address the association
between both types of earnings management and litigation around equity offerings.
This study is important in light of the recent trend of earnings management in public
?rms, which ultimately hurts users of ?nancial statements, particularly investors.
Also, we provide further evidence on which methods are used to manage earnings
around equity offerings and how this has changed following the passage of SarbOx.
To test our hypotheses, we ?rst identify a sample of ?rms that underwent a SEO
and test which accounts show abnormal behavior (accrual or real accounts) in the year
prior to the offering. Our test sample includes 1,871 SEOs in the years 1990-2004[1].
Out of this sample, 218 ?rms subsequently faced litigation. Some of the allegations
(n ¼ 85) against SEOs involve the manipulation of the market price by underwriters at
the time of the offering and other non-accounting irregularities. These allegations are
unrelated to earnings manipulation; we therefore eliminate these litigation cases and
examine only 133 lawsuits that involve accounting allegations.
The results indicate that SEOs have a prevalence of income-increasing manipulation
using both accrual and some real account manipulation in the year prior to the offering.
We ?nd some evidence of a shift to real account manipulation in the post-SarbOx period.
Furthermore, we ?nd that SEO ?rms that are subject to class action litigation by
investors have a higher prevalence of discretionary accruals, which is consistent with
prior research on litigation that considered only accrual manipulation (Lu, 2004;
Ibrahim et al., 2008). On the other hand, there is no evidence of overall manipulation in
real accounts in sued SEOs, when compared to non-sued SEOs. However, sued SEOs
have lower abnormal cash?ows (one measure of real account manipulation). We also
?nd that the probability of litigation is directly related to the level of discretionary
accruals; the higher income-increasing discretionary accruals, the more likely the SEO
will result in accounting-related litigation. In the post-SarbOx period, there is less
emphasis on accrual manipulation through accounts receivable and more emphasis on
income-increasing abnormal accounts in litigation decisions. The implications of these
?ndings are that ?rms that engage in income-increasing earnings management through
accruals are more likely to be sued. Other forms of manipulation are less likely to be
considered by investors and regulators in initiating litigation. There is evidence that this
may have changed somewhat in the post-SarbOx period; investors are paying more
attention to non-accrual manipulation following SarbOx.
The rest of the paper is organized as follows. Section 2 presents prior research and
the development of the hypotheses. Data selection and methodology is presented in
Section 3, followed by the results in Section 4. Section 5 concludes.
2. Prior research and hypotheses development
This section discusses prior research related to accrual and real account manipulation
around equity offerings (Section 2.1) as well as the effect of SarbOx on earnings
management behavior (Section 2.2) and litigation risk in relation to earnings
management (Section 2.3). The hypotheses are developed in Section 2.4.
2.1 Earnings management around equity offerings
Prior research has documented that ?rms manage earnings upwards using
accruals around initial public offerings (IPOs) (Aharony et al., 1993; Friedlan, 1994;
ARJ
24,1
52
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Teoh et al., 1998b) and SEOs (Teoh et al., 1998a; Rangan, 1998; Shivakumar, 2000)
in order to increase the market price of the stock. The bene?ts from earnings
management in this context are especially high because the proceeds of an equity
offering are based on the stock price at one point in time. Prior studies also ?nd that in
order for earnings management to have the maximumprice impact, ?rms issuing equity,
either through initial or seasoned offerings, will prefer to manage earnings using
recurring rather than non-recurring income statement items. In particular, ?rms issuing
equity are likely to manage earnings through revenue accounts, and speci?cally through
accounts receivable (Marquardt and Wiedman, 2004). However, Ball and Shivakumar
(2008) and Armstrong et al. (2009) argue that results from prior research on earnings
management around large transactions and other large events, especially around an
IPO, may be biased for several reasons. Most notably, they argue that using the year of
the event as the test year may overestimate discretionary accruals, given that this year is
accompanied with high increases in working capital which will in?ate discretionary
accrual measures.
Research on real manipulation around equity offerings similarly provides evidence of
income-increasing manipulation. Cohen and Zarowin (2010) ?nd that ?rms use
income-increasing real, as well as accrual-based, earnings management tools in the
years surrounding SEOs and they show how the tendency to trade-off real versus
accrual-based earnings management activities varies by industry. Speci?cally, they ?nd
that SEOs have signi?cant positive abnormal production costs, which would lead to a
decrease in cost of goods sold, in the year of the offering. They also ?nd lower than
expected discretionary expenses, including advertising, research and development and
selling, general and administrative expenses in the years prior to as well as the year of
the offering. Finally, they ?nd lower than expected cash ?ows from operations,
indicating less cash receipts from offering more lenient terms to customers in the years
leading to as well as the year of the offering. All of these effects wouldtend to abnormally
increase earnings through real account changes. Mizik and Jacobson (2007) show that
earnings are in?ated around equity offerings using both real- and accrual-based
management. They also show that stock overvaluation is more prevalent for ?rms with
real manipulation, i.e. investors are fooled by this real account manipulation.
2.2 Earnings management and SarbOx
The SarbOx Act, passed in July 2002, was enacted to protect investors by improving
the accuracy and reliability of corporate disclosures made pursuant to the securities
laws[2]. Li et al. (2008) show that accrual earnings management has declined following
the passage of SarbOx. The results in Li et al. (2008) are consistent with investors
expecting that the more extensively ?rms managed their earnings, the more SarbOx
would constrain earnings management and enhance the quality of ?nancial reporting.
Accrual and real earnings management are seen as substitutes in managing
earnings (Graham et al., 2005; Zang, 2007). Cohen et al. (2008) test the shift between
accrual and real earnings management in the periods before and after the passage of
SarbOx. They ?nd that ?rms have shifted away from accrual management to real
account management following SarbOx. Their evidence implies that the increased
costs of detection of accrual management post-SarbOx has led to management turning
to other types of manipulation, even though they may be costlier to the ?rm in the
long run.
Real and accrual-
based earnings
53
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
2.3 Earnings management and litigation risk
Several studies have considered the role of accrual earnings management in class
action securities litigation. Jones (1998) studies the determinants of securities litigation
risk using a sample of 69 ?rms sued during the period 1989-1992 and a control sample
of non-sued ?rms with a 10 percent drop in stock price. He ?nds a negative but
insigni?cant association between litigation risk and abnormal current accruals
estimated from a term-adjusted version of Jones (1991) model. Lu (2004) examines the
relation between earnings management and securities class action litigation. In a
sample of 781 ?rms sued in class action securities litigation from 1988 to 2000, Lu ?nds
that accruals and the changes in revenue are abnormally high during alleged periods of
manipulation, and tend to reverse subsequently. Moreover, the magnitude of accruals
overstatement is greatest for defendant ?rms subject to SEC enforcement actions or
having made accounting restatements, and least for defendant ?rms not facing
accounting allegations.
Furthermore, data collected by PricewaterhouseCoopers in 2000 since the enactment
of the Private Securities Litigation Reform Act of 1995 show that the majority of suits
allege some sort of revenue recognition violation. This suggests that earnings
management through revenue misstatement increases the likelihood of litigation,
which is consistent with Palmrose et al. (2004) who demonstrate a positive relationship
between litigation and misstatement of revenue.
In sum, the evidence from prior literature shows that accrual earnings management
activities increase ?rms’ litigation risk in general.
In the context of equity offerings, DuCharme et al. (2004) examine the association
between accrual earnings management by ?rms issuing stocks and the incidence of
litigation, allegations of earnings management, and lawsuit settlement amounts. They
use a litigation sample consisting of 150 SEOs and 72 IPOs from 1988 to 1997, and a
control sample consisting of all SEOs and IPOs that are not subject to litigation during
the same periods. After controlling for characteristics of the stock offerings, they only
?nd a signi?cant and positive association between abnormal current accruals and the
incidence of lawsuit ?lings for the SEO ?rms, but not for the IPO ?rms. They also fail
to ?nd a signi?cant relation between abnormal current accruals and allegations of
earnings management. Ibrahim et al. (2008) also study the association between
litigation and discretionary components of accruals, namely accounts receivable,
inventory, accounts payable, other working capital, depreciation, and special items in
an equity offering setting. They ?nd a higher prevalence of accounts receivable
manipulation in litigated ?rms, consistent with revenue recognition problems.
There is no prior research studying litigation risk and real account manipulation.
However, prior research implies that real manipulation is less likely to result in
litigation, since it can be indistinguishable from optimal business decisions and thus
less likely to be detected (Graham et al., 2005; Cohen and Zarowin, 2010).
2.4 Hypotheses development
To the best of our knowledge, no prior research has studied the relationship between
the alternative methods of earnings management, accrual and real account
manipulation, and litigation risk. This study examines this question. We assume
that the likelihood of litigation varies with the method(s) used to reach earnings goals
for ?rms issuing equity. In particular, ?rms can manage earnings by managing either
ARJ
24,1
54
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
accruals or real accounts. Income can be manipulated upwards through real accounts
by cutting real expenditures, reducing cost of goods sold by increasing production,
or increasing sales through offering lenient terms and discounts to customers, which
would reduce operating cash ?ows. We recognize that lawsuits occur over equity
offerings on cases with positive expected values for plaintiffs or plaintiff attorneys
(Priest and Klein, 1984; Alexander, 1991). That is, lawsuits are ?led when the likelihood
of success times the expected recovery exceeds the expected litigation costs (Palmrose,
1991). We conjecture that the assessment of the likelihood of success will change with
the method of manipulation. Speci?cally, we argue that real account manipulation will
be harder to detect by outsiders, since it cannot be distinguished from normal changes
in operations, and it will lead to lower incidence of litigation than accrual manipulation.
We test the following hypothesis:
H1. Defendant SEO?rms engage in a higher magnitude of earnings overstatement
through accrual manipulation in the year prior to the offering, but not real
account manipulation, compared to non-defendant offering ?rms.
Whereas the SarbOx act has reduced accrual manipulation, the evidence suggests that
?rms have shifted to real manipulation (Cohen et al., 2008). Given this evidence, we
conjecture that investors would be more wary of real account manipulation following
SarbOx and may initiate litigation not only based on suspected accrual manipulation,
but also real account manipulation. We therefore test the following hypothesis:
H2. In the post-SarbOx period, defendant SEO ?rms engage in a higher
magnitude of earnings overstatement through real account manipulation in
the year prior to the offering, compared to non-defendant offering ?rms.
Whereas we believe that litigation decisions will be based more on accrual manipulation
in the years prior to SarbOx, we hypothesize that investors will place more scrutiny
on both types of earnings management in more recent years. Therefore, defendant
?rms will show evidence of higher accrual and real manipulation, as compared to
non-defendant ?rms.
3. Empirical methodology
In this section, we explain the sample selection procedures (Section 3.1) and introduce
the measures of discretionary accruals (Section 3.2) and abnormal real accounts
(Section 3.3) followed by the lawsuit data collection methodology (Section 3.4).
3.1 Sample selection
We identify ?rms that underwent an SEO from Thomson Financial’s SDC Platinume
database in the years 1991-2005. Financial information of these SEO?rms in addition to
all other ?rms in the same industries is collected from Standard and Poor’s Compustat
database. We use the year prior to the offering to test for earnings management behavior.
This is typical of earnings management studies, as the incentives to manage earnings are
strongest in the year prior to the offering. Therefore, our SEO sample covers the years
1990-2004. We eliminate ?rms in the utilities and ?nance industries (SIC between 4,000
and 4,999 and between 6,000 and 6,999) and industry-year combinations that have less
than tenobservations since the empirical analysis is based oncross-sectional analysis by
industryandbyyear.[3] We eliminate extreme observations, identi?ed as those inthe top
Real and accrual-
based earnings
55
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and bottom percentile of the total assets, total accruals, production costs, discretionary
expenses, and cash from operations distributions. The ?nal sample consists of all
?rm/year observations with necessary data in industries that include a ?rm that
underwent a SEO in the sample period. This results in a total of 31,207 ?rm-year
observations, which includes 1,871 SEO ?rm/year observations and 29,336 non-SEO
?rm/year observations. This sample is used to calculate discretionary/abnormal accrual
and real accounts, as explained in Sections 3.2 and 3.3. For the empirical analysis, we
adopt a performance-matching approach as suggested by Kothari et al. (2005). More
speci?cally, each SEOobservation is matched with an observation belonging to a ?rmin
the same industry (four-digit SICcode) and year, with the closest level of returnonassets.
If a match is not found, we select fromobservations in the three-digit SIC code. If a match
is still not found, we select from observations in the two-digit SIC code. The empirical
analysis is based on this sample of 1,871 SEO and 1,871 non-SEO ?rm/year
observations.[4] This matching alleviates the problems associated with misspeci?cation
of manipulation measures in observations with extreme performance (Kothari et al.,
2005; Armstrong et al., 2009). Table I presents the yearly and industrial distribution of
the 1,871 SEO ?rms[5].
Panel A of Table I presents the temporal distribution of the sample. The SEO
observations are evenly distributed among the sample years, 1990-2004. The number
of SEO observations each year ranges from 93 (in year 2004) to 165 (in year 2003).
Panel B presents the industrial distribution of the sample. SEOs are most prevalent
in the machinery and equipment industry followed by the wholesale and retail,
business services, and natural resources industry (number of SEOs ¼ 625, 274, 228,
and 198, in machinery and equipment, wholesale and retail, business services and
natural resources industries, respectively). The remaining SEOs (29 percent of the
distribution) are distributed among the remaining nine industries.
3.2 Accrual earnings management: discretionary accruals measures
We estimate accrual manipulation following the modi?ed Jones model ( Jones, 1991;
Dechow et al., 1995) with the proposed Kothari et al. (2005) modi?cations through the
following cross-sectional regression[6]:
TAC
t
A
t21
¼ b
0
þb
1
1
A
t21
þb
2
DREV
t
2DAR
t
A
t21
þb
3
PPE
t
A
t21
þb
4
ðROA
t
Þ
þ DAC
t
whereTAC
t
istotal accrualsfromthecash?owstatement (Compustat No. 123-(Nos308-124))
scaled by beginning total assets (Compustat No. 6) and the independent variables include
change in revenue (Compustat No. 12) less change in accounts receivable (Compustat No. 2)
(DREV
t
2DAR
t
), the level of gross property, plant, and equipment (Compustat No. 7)
(PPE
t
), and return on assets in year t (ROA
t
), which is calculated as net income in year t
(Compustat No. 172) divided by total assets in year t (Compustat No. 6). All variables, other
thanROA
t
, are scaledbybeginningtotal assets toadjust for heteroskedasticity. The residual
from this regression is the estimate of discretionary accruals, DAC
t
. The coef?cients are
estimated by running cross-sectional regressions by four-digit SIC codes and by year. This
methodology eliminates the constraints of ?xing the coef?cients in these regressions over
time or over all industries (Kothari et al., 2005). As previouslymentioned, the year prior to the
offering is used as the test year.
ARJ
24,1
56
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
We also estimate discretionary accounts receivable, following Ibrahim (2009), since
this is the accrual that is most signi?cantly associated with litigation, as follows:
DAR
t
A
t21
¼ b
01
þb
11
1
A
t21
þb
21
DREV
t
A
t21
þb
31
ðROA
t
Þ þ DAR
t
where DAR
t
is the change in accounts receivable (Compustat No. 2) and the residual from
the regression, DAR
t
, is the discretionary accounts receivable accrual. DAR
t
represents
the unexpected portion of accounts receivable, which may occur in conjunction with
revenue manipulation, e.g. through early booking of revenue or arti?cial sales
transactions. This methodology for calculating this component of accruals was shown
No. of ?rms %
Panel A: distribution of SEO observations by year
Year
1990 108 5.77
1991 99 5.29
1992 128 6.84
1993 100 5.34
1994 142 7.59
1995 157 8.39
1996 141 7.54
1997 120 6.41
1998 145 7.75
1999 130 6.95
2000 103 5.51
2001 110 5.88
2002 130 6.95
2003 165 8.82
2004 93 4.97
Total 1,871 100.00
Panel B: distribution of SEO observations by industry
Industry
(1) Natural resources 198 10.58
(2) Construction and metal 82 4.38
(3) Food and tobacco 13 0.69
(4) Consumer goods 47 2.51
(5) Paper and printing 28 1.50
(6) Chemical and petroleum 166 8.87
(7) Machinery and equipment 625 33.40
(8) Transportation related 46 2.46
(9) Wholesale and retail 274 14.64
(10) Entertainment 36 1.92
(11) Business services 228 12.19
(12) Health and other services 118 6.31
(13) Non-classi?able 10 0.53
Total 1,871 100.00
Notes: Panel A presents the distribution of SEOs across the sample years; panel B provides industry
compositions of the SEOsample; the industries (in the order presented in panel B) are classi?ed based on
two-digit SICcodes as follows: (1) 0-9,10-14; (2) 15-19, 30, 32-34; (3) 20-21; (4) 22-23, 25, 31, 39; (5) 24, 26-27;
(6) 28-29; (7) 35-36, 38; (8) 37, 40-47; (9) 50-59; (10) 78-79; (11) 73, 81; (12) 70, 72, 75-76, 80, 82-89; (13) 99
Table I.
Temporal and industry
distribution of SEO
observations
Real and accrual-
based earnings
57
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
to be more powerful than the Marquardt and Wiedman (2004) methodology in detecting
earnings management[7].
3.3 Real earnings manipulation measures
We rely on prior studies that developed proxies for real earnings management.
Cohen et al. (2008) base their measures of real earnings management on Roychowdhury
(2006) and others. Speci?cally, they focus on three manipulation methods that would
increase bottom-line earnings:
(1) acceleration of the timing of sales through increased price discounts or more
lenient credit terms, which would abnormally decrease cash from operations;
(2) reporting lower cost of goods sold through increased production; and
(3) decreases in discretionary expenses including advertising expense, research
and development expenses, and selling, general, and administrative (SG&A)
expenses.
The discretionary or abnormal levels of the above real accounts are measured
as follows:
CFO
t
A
t21
¼ a
11
1
A
t21
þa
12
REV
t
A
t21
þa
13
DREV
t
A
t21
þ ACFO
t
PROD
t
A
t21
¼ a
21
1
A
t
þa
22
REV
t
A
t21
þa
23
DREV
t
A
t21
Þ þa
24
DREV
t21
A
t21
þ APROD
t
DEXP
A
t21
¼ a
31
1
A
t21
þa
32
REV
t21
A
t21
þ ADEXP
t
where CFO
t
is cash from operations in period t (Compustat Nos 308-124), PROD
t
is
production costs in period t, de?ned as the sum of cost of goods sold (Compustat No. 41)
and the change in inventory in period t (Compustat No. 3), and DEXP
t
is discretionary
expenses, de?ned as the sum of advertising expenses (Compustat No. 45), R&D
expenses (Compustat No. 46) and SG&A (Compustat No. 189)[8]. The abnormal cash
from operations (ACFO
t
), abnormal production costs (APROD
t
), and abnormal
discretionary expenses (ADEXP
t
) are the residuals from the above regressions. The
regressions are run separately for each four-digit SIC code and for each year. Anegative
ACFO
t
and ADEXP
t
as well as a positive APROD
t
would indicate income-increasing
earnings management[9]. All variables are de?ated by the beginning total assets to
adjust for heteroskedasticity.
As in Cohen et al. (2008), we also provide a measure of total manipulation through
real accounts as follows:
AREAL
t
¼ ACFO
t
þ APROD
t
þ ADEXP
t
Since the three separate real account manipulation measures have different effects on
bottom-line earnings, we do not make predictions about the sign of the measure,
AREAL
t
, but only consider its magnitude.
ARJ
24,1
58
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
All above variables in addition to the ?nancial variables collected are de?ned in
Appendix 1. The Pearson correlation coef?cients between the ?nancial and
manipulation variables appear in Table II.
All coef?cients that are above 0.8 are shown in italic. Most notably, the performance
measures are highly correlated. INC
t
is highly correlated with CFO
t
( p ¼ 0.812), while
CFO
t
is highly correlated with abnormal cash ?ow from operations, ACFO
t
( p ¼ 0.876).
Net revenues, REV
t
, are highly correlated with production costs, PROD
t
( p ¼ 0.926). The
discretionary accruals measures are not highly correlated with the real account
manipulation measures. The highest correlation is between ACFO
t
and ADEXP
t
( p ¼ 20.499). We do not believe that these correlations will impact the empirical analysis.
3.4 Lawsuit data selection
Information regarding the lawsuits, such as class periods and the nature of the
allegations made therein was taken from the LEXIS/NEXIS Academic Universe
Business News, searching on company name and the keywords “Class Action” and
“Litigation” in the three years following the offering year. Data were also collected
from the Stanford Law Securities Class Action Clearinghouse which hosts a list of
?rms that faced class action litigation. The search identi?ed 218 SEOs that were sued
subsequent to their offering. Out of these observations, 85 involved litigation due to
underwriter problems of disclosure or other non-accounting irregularities and are thus
excluded from the analysis. The remaining 133 SEOs faced accounting allegations in
their lawsuits. There were 1,653 SEOs that were not subsequently subject to litigation.
The empirical analysis is based on the 133 sued SEOs and the 1,653 non-sued SEOs.
4. Tests and empirical results
We ?rst begin by testing the accrual and real account manipulation in the sample of
SEOs (Section 4.1) followed by the hypotheses testing (Section 4.2).
4.1 Accrual and real earnings management around SEOs
First, we replicate prior studies to show that our sample of SEOs engage in
income-increasing accrual and real account manipulation prior to the offering. Panel A
of Table III reports descriptive statistics and differences between the SEOs (n ¼ 1,871)
and the matched non-SEO observations.
The SEO observations are larger and less pro?table than the matched sample
(difference in median A
t
is $55m and difference in mean INC
t
is 20.015, signi?cant at
the 1 and 5 percent levels, respectively). SEO observations also have higher revenues
(difference in mean REV
t
is 0.072, and difference in median is 0.058, both signi?cant at
the 5 percent level). The results of differences in discretionary accruals and real
accounts appear in the second part of the panel. Discretionary accruals are higher in
the SEO observations (difference in mean DAC
t
is 0.003, which is not signi?cant, and
difference in median DAC
t
is 0.005, signi?cant at the 10 percent level). Discretionary
accounts receivable are also higher in the SEO sample (difference in mean DAR
t
is
0.014, and difference in median is 0.005, both signi?cant at the 1 percent level). Both are
consistent with income-increasing manipulation prior to the offering. Abnormal cash
from operations is lower in SEO observations (difference in mean ACFO
t
is 20.022 and
difference in median is 20.008, signi?cant at the 1 and 5 percent levels, respectively),
which is consistent with income-increasing real account manipulation. However,
Real and accrual-
based earnings
59
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
R
E
V
t
C
F
O
t
P
R
O
D
t
D
E
X
P
t
T
A
C
t
R
O
A
t
A
t
D
A
C
t
D
A
R
t
A
C
F
O
t
A
P
R
O
D
t
A
D
E
X
P
t
A
R
E
A
L
t
I
N
C
t
0
.
2
8
7
0
.
8
1
2
0
.
2
0
9
2
0
.
4
4
0
0
.
5
5
3
0
.
7
6
8
0
.
0
6
5
0
.
2
3
3
2
0
.
1
0
3
0
.
6
8
0
2
0
.
1
0
0
2
0
.
5
6
4
2
0
.
4
2
9
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
5
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
R
E
V
t
1
.
0
0
0
0
.
1
9
3
0
.
9
2
6
0
.
2
9
7
0
.
2
1
5
0
.
2
6
4
2
0
.
0
5
3
0
.
0
3
5
0
.
0
7
2
0
.
0
8
6
0
.
1
0
2
2
0
.
0
5
7
0
.
0
6
0
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
2
1
)
(
0
.
1
3
1
)
(
0
.
0
0
2
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
1
4
)
(
0
.
0
0
9
)
C
F
O
t
1
.
0
0
0
0
.
1
0
5
2
0
.
4
3
6
2
0
.
0
3
9
0
.
6
6
0
0
.
1
1
1
2
0
.
2
5
3
2
0
.
1
5
9
0
.
8
7
6
2
0
.
1
8
1
2
0
.
5
1
4
2
0
.
2
7
6
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
9
6
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
P
R
O
D
t
1
.
0
0
0
0
.
0
5
6
0
.
2
0
8
0
.
1
9
5
2
0
.
0
1
0
0
.
0
4
4
0
.
0
4
1
0
.
0
0
9
0
.
3
5
1
2
0
.
2
0
6
2
0
.
0
1
7
(
0
.
0
1
5
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
6
6
8
)
(
0
.
0
5
9
)
(
0
.
0
7
3
)
(
0
.
7
0
5
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
4
6
6
)
D
E
X
P
t
1
.
0
0
0
2
0
.
1
3
1
2
0
.
3
0
3
2
0
.
1
7
7
2
0
.
0
4
8
0
.
1
5
3
2
0
.
3
9
6
2
0
.
3
0
7
0
.
7
8
1
0
.
6
3
2
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
3
8
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
T
A
C
t
1
.
0
0
0
0
.
3
7
1
2
0
.
0
4
8
0
.
7
6
0
0
.
0
5
0
2
0
.
0
8
6
0
.
0
8
6
2
0
.
2
3
1
2
0
.
3
4
0
(
0
.
0
0
1
)
(
0
.
0
3
8
)
(
0
.
0
3
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
R
O
A
t
1
.
0
0
0
0
.
0
5
9
0
.
0
0
4
2
0
.
0
2
6
0
.
5
4
2
2
0
.
0
7
8
2
0
.
3
8
8
2
0
.
2
4
9
(
0
.
0
1
1
)
(
0
.
8
7
6
)
(
0
.
2
6
2
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
A
t
1
.
0
0
0
2
0
.
0
4
9
2
0
.
0
2
4
0
.
1
0
6
0
.
0
0
6
2
0
.
1
1
7
2
0
.
0
9
3
(
0
.
0
3
5
)
(
0
.
2
9
7
)
(
0
.
0
0
1
)
(
0
.
7
9
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
D
A
C
t
1
.
0
0
0
0
.
0
7
4
2
0
.
2
9
8
0
.
1
2
0
2
0
.
1
1
4
2
0
.
2
9
2
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
D
A
R
t
1
.
0
0
0
2
0
.
1
8
3
2
0
.
0
3
0
0
.
1
5
6
0
.
0
7
4
(
0
.
0
0
1
)
(
0
.
1
8
9
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
A
C
F
O
t
1
.
0
0
0
2
0
.
2
5
6
2
0
.
4
9
9
2
0
.
2
2
4
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
A
P
R
O
D
t
1
.
0
0
0
2
0
.
4
2
8
2
0
.
0
1
7
(
0
.
0
0
1
)
(
0
.
4
5
3
)
A
D
E
X
P
t
1
.
0
0
0
0
.
7
8
6
(
0
.
0
0
1
)
N
o
t
e
s
:
n
¼
1
,
8
7
1
;
a
l
l
c
o
r
r
e
l
a
t
i
o
n
c
o
e
f
?
c
i
e
n
t
s
a
b
o
v
e
0
.
8
a
r
e
s
h
o
w
n
i
n
i
t
a
l
i
c
;
a
l
l
v
a
r
i
a
b
l
e
s
a
r
e
d
e
?
n
e
d
i
n
A
p
p
e
n
d
i
x
1
Table II.
Pearson correlation
coef?cients ( p-values)
of ?nancial variables,
discretionary accruals,
and abnormal real
accounts in SEO sample
ARJ
24,1
60
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
SEO observations
(n ¼ 1,871)
Matched observations
(n ¼ 1,871) Difference
Variable Mean Median Mean Median Mean Median
Panel A: mean and median differences in ?nancial variables, discretionary accruals, and abnormal real
accounts between SEO observations and matched non-SEO observations in all years
Financial variables
INC
t
20.020 0.048 20.005 0.041 20.015
* *
0.007
* *
REV
t
1.444 1.248 1.372 1.190 0.072
* *
0.058
* *
DREV
t
0.303 0.204 0.167 0.099 0.136
* * *
0.104
* * *
CFO
t
0.045 0.082 0.057 0.081 20.012
* *
0.001
PROD
t
0.949 0.727 0.901 0.696 0.048
* *
0.031
DEXP
t
0.524 0.388 0.462 0.364 0.063
* * *
0.024
* *
TAC
t
20.065 20.058 20.062 20.056 20.004 20.002
ROA
t
20.030 0.036 20.022 0.037 20.008 20.001
A
t
931.821 180.490 956.504 125.410 224.682 55.080
* * *
Discretionary accruals and real accounts
DAC
t
0.003 0.005 0.000 0.000 0.003 0.005
*
DAR
t
0.013 0.002 20.001 20.003 0.014
* * *
0.005
* * *
ACFO
t
20.011 0.011 0.011 0.019 20.022
* * *
20.008
* *
APROD
t
20.025 20.022 20.014 20.014 20.011
* *
20.008
*
ADEXP
t
0.136 0.034 0.036 0.007 0.101
* * *
0.027
* * *
AREAL
t
0.100 0.041 0.033 0.008 0.067
* * *
0.034
* * *
Panel B: mean and median differences in discretionary accruals and abnormal real accounts between
SEO observations and matched non-SEO observations in the pre- and post-SarbOx periods
Pre-SarbOx
DAC
t
0.004 0.006 20.001 20.001 0.006
*
0.006
* *
DAR
t
0.014 0.002 20.001 20.004 0.016
* * *
0.005
* * *
ACFO
t
20.010 0.014 0.014 0.021 20.024
* * *
20.006
* *
APROD
t
20.024 20.022 20.011 20.012 20.013
*
20.009
ADEXP
t
0.140 0.036 0.035 0.010 0.105
* * *
0.026
* * *
AREAL
t
0.007 0.001 20.001 0.000 0.008
* * *
0.001
* * *
Post-SarbOx
DAC
t
20.001 20.001 0.005 0.004 20.006 20.005
DAR
t
0.006 0.002 0.000 20.002 0.006
*
0.004
* * *
ACFO
t
20.017 0.006 0.000 0.011 20.017
*
20.006
APROD
t
20.026 20.029 20.022 20.019 20.005 20.011
ADEXP
t
0.124 0.030 0.040 20.004 0.084
* * *
0.034
* * *
AREAL
t
20.013 0.001 20.003 0.000 20.010
* * *
0.001
* * *
Difference between pre- and post-SarbOx
DAC
t
20.005 20.006
* *
0.006 0.005
DAR
t
20.008
*
0.000 0.002 0.002
ACFO
t
20.007 20.009
* *
20.013
*
20.010
*
APROD
t
20.002 20.008 20.010 20.006
ADEXP
t
20.016 20.006 0.005 20.014
AREAL
t
20.020
*
0.000
* * *
20.002
*
0.000
Notes: Signi?cance at:
*
10,
* *
5, and
* * *
1 per cent levels (one tailed), respectively, using t-test and
Wilcoxon two-sample test; the matched observations are chosen from the same four-digit SIC code as
the SEO observation in the same year as the SEO, with the closest ROA; if no match is found, then the
matching is based on the three-digit SIC code; if still no match is found, then the matching is based on
the two-digit SIC code; all variables are de?ned in Appendix 1; n refers to the number of observations;
pre-SarbOx refers to years 1990-2001 and post-SarbOx refers to years 2002-2004
Table III.
Univariate tests of
differences in ?nancial
variables, discretionary
accruals, and real
accounts in SEO and
matched observations
Real and accrual-
based earnings
61
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
abnormal production costs are lower for SEO observations (difference in mean
APROD
t
is 20.011 and difference in median is 20.008, signi?cant at the 5 and
10 percent levels, respectively), which is not consistent with income-increasing
manipulation. Finally, abnormal discretionary expenses are higher in SEO
observations (difference in mean ADEXP
t
is 0.101 and difference in median is 0.027,
both signi?cant at the 1 percent level), which is not consistent with income-increasing
manipulation. The total real manipulation measure, AREAL
t
, is signi?cantly larger in
SEO observations (difference in mean is 0.067 and difference in median is 0.034, both
signi?cant at the 1 percent level). The magnitude of the abnormal real accounts in the
non-SEO sample is close to the numbers reported in Cohen et al. (2008, Table I, p. 767).
Panel B presents the univariate differences in the pre- and post-SarbOx periods.
There are 1,483 SEO observations in the pre-SarbOx period and 388 SEO observations
in the post-SarbOx period. In the pre-SarbOx period, the results are consistent with the
full sample. DAC
t
and DAR
t
are both higher in the SEO observations (difference in
mean DAC
t
is 0.006 and difference in median is 0.006, signi?cant at the 10 and 5 percent
levels, respectively; difference in mean DAR
t
is 0.016 and difference in median is 0.005,
both signi?cant at the 1 percent level). ACFO
t
is lower in the SEO observations
(difference in mean is 20.024 and difference in median is 20.006, signi?cant at the
1 and 5 percent levels, respectively), consistent with income-increasing manipulation.
However, APROD
t
and ADEXP
t
do not exhibit income-increasing behavior (difference
in mean APROD
t
is 20.013 and difference in mean ADEXP
t
is 0.105, signi?cant at the
10 and 1 percent levels, respectively). As in the full sample, the total real manipulation
measure, AREAL
t
, is signi?cantly larger in SEO observations (mean difference is 0.008
and median difference is 0.001, both signi?cant at the 1 percent level).
In the post-SarbOx period, DAC
t
is not signi?cantly different between the SEO and
the non-SEO observations. DAR
t
however is still larger in SEO observations (difference
in mean is 0.006 and difference in median is 0.004, signi?cant at the 10 and 1 percent
levels, respectively). There is some evidence of income-increasing manipulation
through real accounts. ACFO
t
is slightly lower in the SEO ?rms (difference in mean
is 20.017, signi?cant at the 10 percent level). However, the remaining real accounts do
not indicate income-increasing manipulation. ADEXP
t
is higher (difference in mean is
0.084 and difference in median is 0.034, both signi?cant at the 1 percent level), and
APROD
t
is lower in SEOs but the difference is not signi?cant. The total manipulation
measure, AREAL
t
, is signi?cantly different between SEO and non-SEO observations
(difference in mean is 20.010 and difference in median is 0.001, both signi?cant at the
1 percent level); however, it is not clear whether it is an income-increasing or
decreasing manipulation.
The third section of Panel B presents the mean and median differences in
discretionary/abnormal accrual and real accounts between the pre- and post-SarbOx
periods. We ?nd signi?cant differences between these two periods for the SEO ?rms.
Speci?cally, discretionary accruals are lower in the post-SarbOx period (difference in
median is 20.006, signi?cant at the 5 percent level), indicating that SEOs engage in
less accrual manipulation following the passage of SarbOx; discretionary accounts
receivables are also lower (difference in mean is 20.008, signi?cant at the 10 percent
level). There is some evidence of shifting to real manipulation as abnormal cash from
operations is lower (difference in median is 20.009, signi?cant at the 5 percent level).
The real manipulation measure, AREAL
t
, is smaller in the post-SarbOx period
ARJ
24,1
62
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(difference in median is 20.020, signi?cant at the 10 percent level). For the non-SEO
observations, the only signi?cant differences between both periods is in the abnormal
cash from operations (mean difference is 20.013 and median difference is 20.010, both
signi?cant at the 10 percent level) indicating income-increasing real account
manipulation. These results corroborate prior evidence that shows that SEO ?rms
engage in less earnings management through accruals in the post-SarbOx period but
still show some evidence of real account manipulation.
To further test the differences between SEO and non-SEO observations, we conduct
a multivariate analysis including other variables that might cause the differences in the
discretionary accrual and abnormal real account measures observed above.
Speci?cally, we test the following regression:
DEPENDENT
it
¼ a
1i
þb
1i
SEO
t
þb
2i
SIZE
t
þb
3i
LEV
t
þb
4i
ROA
t
þb
5i
MTB
t
þb
6i
LOSS
t
þb
7i
LITRISK
t
þ
X
j
b
ji
Industry Dummies
t
ð1Þ
where:
DEPENDENT
it
¼ the discretionary accrual or abnormal real account
measure and i ¼ 1 2 6 for DAC
t
, DAR
t
, ACFO
t
,
APROD
t
, ADEXP, and AREAL
t
as the dependent variable;
SEO
t
¼ indicator variable that takes on the value 1 for an SEO
observation in the year prior to the offering and 0 for
matched non-SEO observations;
SIZE
t
¼ Log (total assets at year end);
LEV
t
¼ leverage ¼ total of short- and long-term debt scaled by
average total assets;
ROA
t
¼ return on assets measured as net income divided by total
assets;
MTB
t
¼ market-to-book value measured as market value of assets
divided by book value of assets at year end;
LOSS
t
¼ indicator variable that takes on the value 1 if the ?rm has
a net loss during the year, and 0 otherwise;
LITRISK
t
¼ indicator variable that takes on the value 1 if the
?rm operates in a high-risk environment, as de?ned in
Francis et al. (1994), and 0 otherwise[10]; and
Industry Dummies
t
¼ indicator variable that takes on the value of 1 if observation
belongs to a speci?c industry, and 0 otherwise.
We control for several factors that are associated with incentives to manage earnings
as well as the discretionary accruals measure (Frankel et al., 2002). First, we control
for the size of the ?rm since larger ?rms tend to have larger accruals and larger
discretionary accruals. Larger ?rms will also tend to have higher expenses and cash
from operations. We control for leverage, since ?rms with higher leverage have more
Real and accrual-
based earnings
63
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
incentives to manage earnings to avoid covenant violations. We control for
performance (ROA) since ?rms with better performance tend to have higher
accruals and cash from operations. We control for MTB since ?rms with growth
prospects (with higher market-to-book values) have higher incentives to manage
earnings. We control for loss since loss ?rms have different incentives to manage
earnings. We control for litigation risk since ?rms that more likely to face litigation
have higher incentives to manage earnings to avoid it. We also control for the industry
membership of the observation. We use the same designations in Table I based on the
four-digit SIC code to construct the industry dummies.
The coef?cients b
1i
are used to test for earnings management behavior with
i ¼ 1 2 6 in the regressions with DAC
t
, DAR
t
, ACFO
t
, APROD
t
, ADEXP
t
, and
AREAL
t
as the dependent variables. The results of the above regressions appear in
Table IV. All t-statistics in the regressions are calculated from White’s
heteroskedasticity-corrected standard errors.
The results in panel Aare consistent with the univariate results. DAC
t
and DAR
t
are
signi?cantly higher in SEO observations, controlling for other factors typically related
to discretionary accruals (coef?cient ¼ 0.007 and 0.011 for SEO indicator variable in
the regression with DAC
t
and DAR
t
as the dependent variable, respectively, signi?cant
at the 5 and 1 percent levels, respectively). The magnitude of the differences between
SEO and non-SEO observations is quite signi?cant, which warrants further discussion.
Given that the mean (median) total assets for SEO observations is $932m ($181m)
(Table III), the above coef?cients show that DAC
t
is higher in SEO observations by an
average of $6.5m (median of $1.3m) and that DAR
t
is higher by an average of $10m
(median of $2m). Furthermore, these differences constitute an average of 35 per cent of
net income for DAC
t
and 55 per cent for DAR
t
. Economically speaking, these are quite
signi?cant amounts that could affect the decision of users of the ?nancial statements.
As for the abnormal real accounts, only ACFO
t
is consistent with income-increasing
manipulation (coef?cient of SEO ¼ 20.021 in the regression with ACFO
t
as the
dependent variable, signi?cant at the 1 percent level). This means that SEO
observations have lower ACFO
t
by an average of $20m (median of $4m). But there is
evidence of real account manipulation (coef?cient of SEO ¼ 0.065 with AREAL
t
as the
dependent variable, signi?cant at the 1 percent level). The control variables are
generally signi?cant in these regressions. The adjusted R
2
of these regressions range
from 2.51 percent (with APROD
t
as the dependent variable) to 26.81 percent (with
ACFO
t
as the dependent variable). In panel B, to test the differences between the pre-
and post-SarbOx periods in a multivariate setting, we run the following regression:
DEPENDENT
it
¼ a
1i
þb
1i
SEO
t
þb
2i
SarbOx
t
þb
3i
SEO
*
t
SarbOx
t
þb
4i
SIZE
t
þb
5i
LEV
t
þb
6i
ROA
t
þb
7i
MTB
t
þb
8i
LOSS
t
þb
9i
LITRISK
t
þ
X
j
b
ji
Industry Dummies
ð2Þ
where SarbOx
t
– an indicator variable that takes on the value 1 if the observation is in
the years 2002-2004, and 0 otherwise. All remaining variables are as de?ned above.
The main test variable is the interaction variable SEO
*
t
SarbOx
t
, The coef?cient b
3i
measures the amount by which the change in the value of the dependent variable for
SEOs is affected by SarbOx. Hence, we examine the magnitude and the sign of the
coef?cients b
3i
in the empirical tests.
ARJ
24,1
64
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
D
e
p
.
v
a
r
i
a
b
l
e
D
A
C
D
A
R
A
C
F
O
A
P
R
O
D
A
D
E
X
P
A
R
E
A
L
P
a
n
e
l
A
:
c
o
e
f
?
c
i
e
n
t
s
(
t
-
s
t
a
t
i
s
t
i
c
s
)
f
r
o
m
r
e
g
r
e
s
s
i
o
n
s
o
f
t
h
e
f
o
r
m
:
n
¼
3
,
7
4
2
a
S
E
O
t
0
.
0
0
7
(
1
.
9
6
)
*
*
0
.
0
1
1
(
4
.
7
7
)
*
*
*
2
0
.
0
2
1
(
2
4
.
3
0
)
*
*
*
2
0
.
0
1
4
(
2
2
.
1
8
)
*
*
0
.
1
0
0
(
9
.
0
5
)
*
*
*
0
.
0
6
5
(
8
.
1
8
)
*
*
*
S
I
Z
E
t
2
0
.
0
0
5
(
2
5
.
1
0
)
*
*
*
2
0
.
0
0
1
(
2
2
.
0
3
)
*
*
0
.
0
0
4
(
2
.
5
5
)
*
*
*
0
.
0
0
3
(
1
.
6
3
)
*
2
0
.
0
0
5
(
2
1
.
6
3
)
*
0
.
0
0
2
(
0
.
8
1
)
L
E
V
t
2
0
.
0
2
9
(
2
2
.
3
4
)
*
*
*
0
.
0
4
0
(
3
.
0
0
)
*
*
*
0
.
0
0
4
(
0
.
4
8
)
0
.
0
3
6
(
2
.
7
4
)
*
*
*
2
0
.
0
5
1
(
2
2
.
4
0
)
*
*
*
2
0
.
0
1
0
(
2
0
.
7
4
)
R
O
A
t
0
.
0
1
3
(
0
.
7
1
)
0
.
0
0
2
(
0
.
4
5
)
0
.
2
5
1
(
5
.
8
3
)
*
*
*
2
0
.
0
3
9
(
2
2
.
6
0
)
*
*
*
2
0
.
3
4
0
(
2
5
.
4
3
)
*
*
*
2
0
.
1
2
8
(
2
4
.
0
3
)
*
*
*
L
O
S
S
t
2
0
.
0
2
3
(
2
3
.
9
3
)
*
*
*
2
0
.
0
0
1
(
2
0
.
4
7
)
2
0
.
0
5
5
(
2
4
.
7
2
)
*
*
*
0
.
0
2
9
(
3
.
5
3
)
*
*
*
0
.
0
5
5
(
2
.
8
2
)
*
*
*
0
.
0
2
9
(
2
.
3
5
)
*
*
*
L
I
T
R
I
S
K
t
0
.
0
0
0
(
2
0
.
0
5
)
0
.
0
0
4
(
1
.
1
2
)
0
.
0
0
9
(
1
.
4
2
)
*
2
0
.
0
0
5
(
2
0
.
6
1
)
0
.
0
2
4
(
1
.
6
1
)
*
0
.
0
2
7
(
2
.
6
5
)
*
*
*
M
T
B
t
0
.
0
0
0
(
1
.
2
3
)
0
.
0
0
0
(
1
.
1
6
)
2
0
.
0
0
1
(
2
1
.
7
1
)
*
*
2
0
.
0
0
1
(
2
1
.
9
7
)
*
*
0
.
0
0
2
(
1
.
5
4
)
*
0
.
0
0
1
(
1
.
2
2
)
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
A
d
j
u
s
t
e
d
R
2
(
%
)
2
.
6
0
4
.
5
9
2
6
.
8
1
2
.
5
1
1
5
.
6
1
9
.
3
4
P
a
n
e
l
B
:
c
o
e
f
?
c
i
e
n
t
s
(
t
-
s
t
a
t
i
s
t
i
c
s
)
f
r
o
m
r
e
g
r
e
s
s
i
o
n
s
o
f
t
h
e
f
o
r
m
:
n
¼
3
,
7
4
2
b
S
E
O
t
0
.
0
0
9
(
2
.
2
0
)
*
*
0
.
0
1
3
(
4
.
8
7
)
*
*
*
2
0
.
0
2
2
(
2
3
.
8
3
)
*
*
*
2
0
.
0
1
6
(
2
2
.
0
9
)
*
*
*
0
.
1
0
3
(
8
.
1
2
)
*
*
*
0
.
0
6
5
(
7
.
2
6
)
*
*
*
S
a
r
b
O
x
t
0
.
0
1
1
(
1
.
9
4
)
*
0
.
0
0
3
(
0
.
8
5
)
2
0
.
0
0
3
(
2
0
.
4
5
)
2
0
.
0
1
4
(
2
1
.
3
8
)
2
0
.
0
1
3
(
2
0
.
8
3
)
2
0
.
0
3
0
(
2
2
.
7
1
)
*
*
*
S
E
O
*t
S
a
r
b
O
x
t
2
0
.
0
1
0
(
2
1
.
3
0
)
2
0
.
0
0
9
(
2
1
.
7
8
)
*
0
.
0
0
4
(
0
.
3
9
)
0
.
0
0
7
(
0
.
4
9
)
2
0
.
0
1
7
(
2
0
.
7
2
)
2
0
.
0
0
6
(
2
0
.
3
5
)
S
I
Z
E
t
2
0
.
0
0
5
(
2
5
.
1
1
)
*
*
*
2
0
.
0
0
1
(
2
1
.
8
5
)
*
0
.
0
0
4
(
2
.
5
3
)
*
*
*
0
.
0
0
4
(
1
.
7
6
)
*
2
0
.
0
0
4
(
2
1
.
3
3
)
0
.
0
0
3
(
1
.
3
4
)
L
E
V
t
2
0
.
0
2
9
(
2
2
.
3
1
)
*
*
0
.
0
4
0
(
3
.
0
0
)
*
*
*
0
.
0
0
4
(
0
.
4
7
)
0
.
0
3
6
(
2
.
7
1
)
*
*
*
2
0
.
0
5
2
(
2
2
.
4
4
)
*
*
*
2
0
.
0
1
2
(
2
0
.
8
6
)
R
O
A
t
0
.
0
1
3
(
0
.
7
3
)
0
.
0
0
2
(
0
.
4
2
)
0
.
2
5
1
(
5
.
8
1
)
*
*
*
2
0
.
0
3
9
(
2
2
.
6
3
)
*
*
*
2
0
.
3
4
1
(
2
5
.
4
3
)
*
*
*
2
0
.
1
3
0
(
2
4
.
0
6
)
*
*
*
L
O
S
S
t
2
0
.
0
2
4
(
2
3
.
9
9
)
*
*
*
2
0
.
0
0
1
(
2
0
.
4
3
)
2
0
.
0
5
5
(
2
4
.
7
4
)
*
*
*
0
.
0
3
0
(
3
.
6
2
)
*
*
*
0
.
0
5
6
(
2
.
9
2
)
*
*
*
0
.
0
3
2
(
2
.
5
5
)
*
*
*
L
I
T
R
I
S
K
t
0
.
0
0
0
(
0
.
0
8
)
0
.
0
0
4
(
1
.
1
3
)
0
.
0
0
9
(
1
.
4
2
)
2
0
.
0
0
5
(
2
0
.
5
8
)
0
.
0
2
4
(
1
.
6
5
)
0
.
0
2
8
(
2
.
7
3
)
*
*
*
M
T
B
t
0
.
0
0
0
(
1
.
2
4
)
0
.
0
0
0
(
1
.
1
3
)
2
0
.
0
0
1
(
2
1
.
7
1
)
*
2
0
.
0
0
1
(
2
1
.
9
8
)
*
*
0
.
0
0
2
(
1
.
5
3
)
0
.
0
0
1
(
1
.
1
9
)
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
A
d
j
u
s
t
e
d
R
2
(
%
)
2
.
6
3
4
.
6
1
2
6
.
7
7
2
.
5
0
1
5
.
6
3
9
.
5
8
N
o
t
e
s
:
S
i
g
n
i
?
c
a
n
c
e
a
t
:
*
1
0
,
*
*
5
,
a
n
d
*
*
*
1
p
e
r
c
e
n
t
l
e
v
e
l
s
(
o
n
e
t
a
i
l
e
d
)
,
r
e
s
p
e
c
t
i
v
e
l
y
:
a
D
E
P
E
N
D
E
N
T
i
t
¼
a
1
i
þ
b
1
i
S
E
O
t
þ
b
2
i
S
I
Z
E
t
þ
b
3
i
L
E
V
t
þ
b
4
i
R
O
A
t
þ
b
5
i
M
T
B
t
þ
b
6
i
L
O
S
S
t
þ
b
7
i
L
I
T
R
I
S
K
t
þ
X
j
b
j
i
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
t
b
D
E
P
E
N
D
E
N
T
i
t
¼
a
1
i
þ
b
1
i
S
E
O
t
þ
b
2
i
S
a
r
b
O
x
t
þ
b
3
i
S
E
O
*t
S
a
r
b
O
x
t
þ
b
4
i
S
I
Z
E
t
þ
b
5
i
L
E
V
t
þ
b
6
i
R
O
A
t
þ
b
7
i
M
T
B
t
þ
b
8
i
L
O
S
S
t
þ
b
9
i
L
I
T
R
I
S
K
t
þ
X
j
b
j
i
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
w
h
e
r
e
D
E
P
E
N
D
E
N
T
i
t
r
e
p
r
e
s
e
n
t
s
d
i
s
c
r
e
t
i
o
n
a
r
y
a
c
c
r
u
a
l
o
r
a
b
n
o
r
m
a
l
r
e
a
l
a
c
c
o
u
n
t
v
a
r
i
a
b
l
e
;
a
l
l
v
a
r
i
a
b
l
e
s
a
r
e
d
e
?
n
e
d
i
n
A
p
p
e
n
d
i
x
1
;
t
h
e
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
c
a
l
c
u
l
a
t
e
d
f
r
o
m
W
h
i
t
e
’
s
h
e
t
e
r
o
s
k
e
d
a
s
t
i
c
i
t
y
-
c
o
r
r
e
c
t
e
d
s
t
a
n
d
a
r
d
e
r
r
o
r
s
Table IV.
Multivariate tests
of differences in
discretionary accruals
and real accounts
in SEO and matched
observations
Real and accrual-
based earnings
65
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The results in panel B indicate that discretionary accounts receivable is signi?cantly
lower in the post-SarbOx period (coef?cient of SEO
*
t
SarbOx
t
¼ 20:009 with DAR
t
as the dependent variable, signi?cant at the 10 percent level). This means that DAR
t
is lower by 0.9 per cent of total assets in SEOs following SarbOx. This is equivalent to an
average of $8.4m (median of $1.6m). Abnormal real accounts are not signi?cantly
different following SarbOx (coef?cient of SEO
*
t
SarbOx
t
¼ 20:006 and insigni?cant
with AREAL
t
as the dependent variable). Therefore, there is evidence of a decline in one
type of accrual manipulation, namely accounts receivable, for SEOs following SarbOx.
However, there is evidence that the overall real account manipulation behavior has
changed in the post-SarbOx period (coef?cient on SarbOx
t
¼ 20.030, signi?cant at the
1 percent level with AREAL
t
as the dependent variable). Overall, the results indicate that
SEOs engage in accrual manipulation and some form of real account manipulation to
increase income in the year prior to the offering in the pre-SarbOx period. It is not evident
that SEOs reduce discretionary expenses (SG&A, R&D, and advertising expenses) or
increase production in order to positively in?uence earnings. There is some evidence of
a reduction in accrual manipulation in the post-SarbOx period in the SEO setting and
a shift to real accrual manipulation in all ?rms.
4.2 Accrual and real manipulation in sued and non-sued SEOs
To test whether litigation subsequent to SEOs is associated with the type of manipulation,
we ?rst examine univariate differences in the mean and median discretionary accrual and
real accounts for sued and non-sued SEOs. From the sample of 1,871 SEOs, a total of
218 issues resulted in subsequent litigation. However, only 133 ?rms were sued over
accounting issues. We only include these 133 SEO observations in the analysis. The
remaining SEO observations (1,653) did not face litigation subsequent to the SEO date.
Table V presents the univariate differences between sued and non-sued SEOs. Panel
A presents the differences in all periods, whereas panel B presents the results
separately in the pre- and post-SarbOx periods.
Sued SEOs appear to be more pro?table in the year prior to the offering but have
less cash from operations (difference in median INC
t
is 0.093 and difference in median
CFO
t
is 20.004, both signi?cant at the 10 percent level). DAC
t
is larger in magnitude in
sued SEO ?rms (difference in mean is 0.017 and difference in median is 0.004,
signi?cant at the 5 and 10 percent levels, respectively). DAR
t
is not signi?cantly
different in sued and non-sued SEOs. Abnormal discretionary expenses are higher in
sued SEOs (difference in median is 0.037, signi?cant at the 5 percent level) but this
represents income-decreasing manipulation. The total real account manipulation
measure, AREAL
t
, is signi?cantly higher in sued SEOs (difference in mean is 0.036 and
difference in median is 0.027, signi?cant at the 10 and 5 percent levels, respectively).
Panel B presents the results separately in the pre- and post-SarbOx periods. In the
pre-SarbOx period, there are 104 SEO observations that led to subsequent class action
litigation for accounting irregularities (7 percent of all SEOs in that period). In the post-
SarbOx period, there are 29 sued SEOs (8 percent of all SEOs in that period). The results
in the pre-SarbOx period are similar to those in the overall sample. DAC
t
is higher
in sued SEOs (difference in mean is 0.019, signi?cant at the 10 percent level).
Abnormal discretionary expenses are higher in sued SEOs (median difference is 0.038,
signi?cant at the 10 percent level) but this is inconsistent with income-increasing
manipulation. The total real account manipulation measure, AREAL
t
, is also
ARJ
24,1
66
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
signi?cantly higher in sued SEOs (mean difference is 0.051 and median difference is
0.037, both signi?cant at the 5 percent level).
In the post-SarbOx period, abnormal cash from operations is lower in sued SEOs
(mean difference is 20.041 and median difference is 20.045, both signi?cant at the
Sued SEOs Non-sued SEOs Difference
Variable Mean Median Mean Median Mean Median
Panel A: mean and median differences in ?nancial variables, discretionary accruals, and real accounts
between sued and non-sued SEO ?rms in all years
Financial variables n ¼ 133 n ¼ 1,653
INC
t
1.511 1.332 1.440 1.239 0.071 0.093
*
REV
t
20.051 20.047 20.062 20.058 0.011 0.011
CFO
t
20.044 0.032 20.027 0.036 20.017 20.004
*
TAC
t
20.018 0.049 20.012 0.049 20.006 0.000
ROA
t
0.033 0.062 0.050 0.084 20.017 20.022
Discretionary accruals and real accounts
DAC
t
0.022 0.009 0.004 0.005 0.017
* *
0.004
*
DAR
t
0.012 0.002 0.012 0.002 0.000 0.000
ACFO
t
20.014 20.003 20.009 0.013 20.005 20.016
APROD
t
20.020 20.010 20.025 20.020 0.005 0.010
ADEXP
t
0.160 0.065 0.125 0.028 0.036 0.037
* *
AREAL
t
0.126 0.062 0.090 0.035 0.036
*
0.027
* *
Panel B: mean and median differences in discretionary accruals and real accounts between sued and
non-sued SEO ?rms in pre- and post-SarbOx periods
Pre-SarbOx n ¼ 104 n ¼ 1,304
DAC
t
0.025 0.013 0.005 0.006 0.019
*
0.007
DAR
t
0.016 0.004 0.014 0.002 0.003 0.002
ACFO
t
20.004 0.001 20.009 0.016 0.005 20.015
APROD
t
20.024 20.020 20.023 20.020 20.001 0.000
ADEXP
t
0.172 0.065 0.125 0.027 0.047 0.038
*
AREAL
t
0.144 0.075 0.093 0.038 0.051
* *
0.037
* *
Post-SarbOx n ¼ 29 n ¼ 349
DAC
t
0.010 0.008 0.001 20.002 0.010 0.010
DAR
t
20.003 20.001 0.007 0.002 20.010 20.003
ACFO
t
20.051 20.036 20.010 0.009 20.041
*
20.045
*
APROD
t
20.004 0.010 20.033 20.040 0.029 0.050
ADEXP
t
0.119 0.129 0.122 0.029 20.003 0.100
AREAL
t
0.064 0.047 0.079 0.029 20.015 0.018
Difference between pre- and post-SarbOx
DAC
t
20.014 20.005 20.005 20.008
* *
DAR
t
20.020 20.005 20.007
*
0.001
ACFO
t
20.047 20.037
*
20.001 20.007
APROD
t
0.021 0.030 20.010 20.020
* *
ADEXP
t
20.053 0.065 20.003 0.002
AREAL
t
20.080
* *
20.028 20.014 20.009
* *
Notes: Signi?cance levels of
*
10,
* *
5 and
* * *
1 percent (one-tailed), respectively, using t-test and
Wilcoxon two-sample test; the table reports descriptive information for the SEO ?rms with subsequent
litigation (only accounting allegations) and the SEO ?rms without subsequent litigation for the entire
period; any litigation for non-accounting allegations are excluded; the far right columns report the
differences betweenmeanandmedianvalues; all variables arede?nedinAppendix1; nreferstothe number
of observations; Pre-SarbOx refers to years 1990-2001 and post-SarbOx refers to years 2002-2004
Table V.
Univariate tests of
differences between
sued and non-sued SEO
observations
Real and accrual-
based earnings
67
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
10 percent level). This is consistent with income-increasing manipulation. There are no
other signi?cant differences.
The third section of Panel B presents the differences in the manipulation measures
between the pre- and post-SarbOx periods. ACFO
t
is signi?cantly lower in the
post-SarbOx period in sued SEOs (difference in median is 20.037, signi?cant at
the 10 percent level). Furthermore, total real account manipulation, AREAL
t
, is
signi?cantly lower in the post-SarbOx period (mean difference is 20.080, signi?cant at
the 5 percent level). As for differences between non-sued SEO ?rms in both periods,
DAC
t
is lower post-SarbOx (difference in median is 20.008, signi?cant at the 5 percent
level), DAR
t
is lower (difference in mean is 20.007, signi?cant at the 10 percent level),
and APROD
t
is lower (difference in median is 20.020, signi?cant at the 5 percent
level). These are indicative of a reduction in income-increasing accrual manipulation in
the non-sued observations, consistent with prior research. Firms that do not face
litigation have signi?cantly less income-increasing discretionary accruals in the
post-SarbOx period. However, ?rms that face litigation have no signi?cant differences
in discretionary accruals post-SarbOx; this implies that ?rms facing litigation are still
being targeted for accrual manipulation.
As before, we also conduct multivariate analyses while controlling for variables
typically associated with discretionary/abnormal accrual and real accounts. There is a
concern that the manipulation measures and the litigation variable are endogenous,
stemming from the causality between them. This would lead to a correlation between
the litigation variable and the error term in any regression including the discretionary
accrual and abnormal real account variables as the dependent variable and litigation as
an independent variable, and thus will lead to biased results. We therefore employ a
two-stage least square analysis. In the ?rst step, we measure a predicted value of
litigation based on exogenous variables that affect the probability of litigation (detailed
in Appendix 2). In the second step, we use the predicted value of the probability of
litigation in the following two regressions:
DEPENDENT
it
¼a
1i
þb
1i
P_LIT
t
þb
2i
SIZE
t
þb
3i
LEV
t
þb
4i
ROA
t
þb
5i
MTB
t
þb
6i
LOSS
t
þb
7i
LITRISK
t
þ
X
j
b
ji
Industry Dummies
ð3Þ
DEPENDENT
it
¼a
1i
þb
1i
P_LIT
t
þb
2i
SarbOx
t
þb
3i
LIT
*
t
SarbOx
t
þb
4i
SIZE
t
þb
5i
LEV
t
þb
6i
ROA
t
þb
7i
MTB
t
þb
8i
LOSS
t
þb
9i
LITRISK
t
þ
X
j
b
ji
Industry Dummies
ð4Þ
where DEPENDENT
it
is the discretionary accrual or abnormal real account measure,
P_LIT is the predicted value of the probability of litigation, and the remaining
variables are as previously de?ned. As before, the coef?cients of interest are b
1i
in
regression (3) and b
3i
in regression (4). Given our H1, we expect a positive and
signi?cant coef?cient b
11
when the dependent variable is discretionary accruals.
This would indicate accrual manipulation in the year prior to the offering in defendant
?rms as compared to non-defendant ?rms. We do not expect a signi?cant coef?cient in
ARJ
24,1
68
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the ?nal regression (b
16
) as this represents real account manipulation. Given our H2,
we expect a signi?cant coef?cient b
36
using regression (4), with the dependent variable
being the real manipulation measure, to represent more scrutiny of real account
manipulation. The results appear in Table VI.
The results of regression (1) appear in Panel A. We ?nd that ?rms that are sued
have higher DAC
t
(coef?cient on P_LIT
t
¼ 0.177 in the regression with DAC
t
as the
dependent variable, signi?cant at the 5 percent level). This is in line with
income-increasing accrual manipulation. However, it does not appear that this accrual
manipulation is achieved through accounts receivable (consistent with revenue
manipulation) as the variable P_LIT
t
is insigni?cant in the regression with DAR
t
as the
dependent variable. We also ?nd that ?rms that are sued do not show evidence of real
account manipulation as the variable P_LIT
t
is insigni?cant in the regressions with
AREAL
t
as the dependent variable. However, there is evidence of income-increasing
manipulation using abnormal cash?ows (coef?cient on P_LIT
t
¼ 20.366 in the
regression with ACFO
t
as the dependent variable, signi?cant at the 1 percent level).
The magnitude of earnings management in sued SEOs is harder to interpret from
these coef?cients as the test variable is a predicted variable and not the LIT variable
directly. These results are in line with our H1. We ?nd that defendant ?rms engage in
accrual manipulation but not real manipulation in the year prior to the offering,
which means that investors focus on accrual manipulation in initiating class action
litigation.
In panel B, we ?nd that the only signi?cant change in litigation behavior post-SarbOx
is due to DAR
t
(coef?cient on P_LIT
*
SarbOx ¼ 20.205 in the regression with DAR
t
as
the dependent variable, signi?cant at the 1 percent level). This indicates that investors
have reduced their scrutiny of discretionary accounts receivable in the post-SarbOx
period. However, we do not ?nd more scrutiny of real account manipulation (coef?cient
on P_LIT
*
SarbOx ¼ 20.288 in the regression with DREAL
t
as the dependent variable,
and not signi?cant). Therefore, there is no support for our H2 that investors have more
scrutiny of real account manipulation in the post-SarbOx period.
Overall, the results indicate that sued SEOs have a higher prevalence of accrual
manipulation. However, there is also a higher prevalence of income-increasing real
account manipulation through cash ?ow accounts. Following the enactment of SarbOx,
investors seem to have reduced their reliance on accrual manipulation in the form of
accounts receivable. Overall, it appears that investors focus on abnormal accrual
patterns in instigating litigation.
4.3 Tests of probability of litigation
As a robustness check and in order to more easily interpret the magnitude of
earnings management in sued SEOs, in this section we directly test the probability of
litigation in relation to the various accrual and real account manipulation variables.
We follow the analysis in DuCharme et al. (2004) and run the following two
regressions[11]:
ProbðLIT
t
Þ ¼a
1
þb
1
DAC
t
þb
2
DAR
t
þb
3
ACFO
t
þb
4
APROD
t
þb
5
ADEXP
t
þb
6
AUD
t
þb
7
Principal
t
þb
8
Secondary
t
þb
9
High 2Tech
t
ð5Þ
Real and accrual-
based earnings
69
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
D
e
p
.
v
a
r
i
a
b
l
e
D
A
C
D
A
R
A
C
F
O
A
P
R
O
D
A
D
E
X
P
A
R
E
A
L
P
a
n
e
l
A
:
c
o
e
f
?
c
i
e
n
t
s
(
t
-
s
t
a
t
i
s
t
i
c
s
)
f
r
o
m
r
e
g
r
e
s
s
i
o
n
s
o
f
t
h
e
f
o
r
m
a
P
_
L
I
T
t
0
.
1
7
7
(
1
.
9
8
)
*
*
0
.
0
6
5
(
1
.
0
1
)
2
0
.
3
6
6
(
2
2
.
9
9
)
*
*
*
0
.
2
6
0
(
1
.
5
7
)
0
.
4
0
2
(
1
.
5
5
)
0
.
2
9
7
(
1
.
6
1
)
S
I
Z
E
t
2
0
.
0
0
9
(
2
4
.
7
2
)
*
*
*
2
0
.
0
0
5
(
2
3
.
3
8
)
*
*
*
0
.
0
1
6
(
5
.
9
0
)
*
*
*
0
.
0
0
2
(
0
.
5
8
)
2
0
.
0
3
5
(
2
6
.
0
7
)
*
*
*
2
0
.
0
1
7
(
2
4
.
1
4
)
*
*
*
L
E
V
t
2
0
.
0
3
3
(
2
4
.
4
1
)
*
*
*
0
.
0
7
0
(
1
3
.
0
0
)
*
*
*
2
0
.
0
0
9
(
2
0
.
8
7
)
0
.
0
4
4
(
3
.
2
1
)
*
*
*
2
0
.
0
3
5
(
2
1
.
6
0
)
0
.
0
0
1
(
0
.
0
7
)
R
O
A
t
2
0
.
0
2
0
(
2
1
.
8
1
)
*
0
.
0
0
1
(
0
.
1
1
)
0
.
2
7
0
(
1
7
.
5
9
)
*
*
*
2
0
.
0
4
6
(
2
2
.
2
2
)
*
*
2
0
.
3
5
9
(
2
1
0
.
9
8
)
*
*
*
2
0
.
1
3
5
(
2
5
.
8
2
)
*
*
*
L
O
S
S
t
2
0
.
0
3
5
(
2
5
.
0
4
)
*
*
*
2
0
.
0
0
2
(
2
0
.
4
3
)
2
0
.
0
6
0
(
2
6
.
1
9
)
*
*
*
0
.
0
1
6
(
1
.
2
4
)
0
.
0
7
5
(
3
.
6
4
)
*
*
*
0
.
0
3
1
(
2
.
1
2
)
*
*
L
I
T
R
I
S
K
t
2
0
.
0
1
5
(
2
1
.
7
4
)
*
0
.
0
0
5
(
0
.
8
2
)
0
.
0
3
0
(
2
.
5
6
)
*
*
*
2
0
.
0
0
7
(
2
0
.
4
4
)
2
0
.
0
1
9
(
2
0
.
7
9
)
0
.
0
0
4
(
0
.
2
1
)
M
T
B
t
0
.
0
0
0
(
1
.
3
1
)
0
.
0
0
0
(
0
.
9
5
)
2
0
.
0
0
1
(
2
2
.
9
5
)
*
*
*
2
0
.
0
0
1
(
2
1
.
2
9
)
0
.
0
0
3
(
3
.
8
0
)
*
*
*
0
.
0
0
1
(
2
.
2
5
)
*
*
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
A
d
j
u
s
t
e
d
R
2
(
%
)
3
.
8
1
9
.
2
7
3
3
.
9
1
4
.
1
0
2
4
.
3
2
1
3
.
4
5
P
a
n
e
l
B
:
c
o
e
f
?
c
i
e
n
t
s
(
t
-
s
t
a
t
i
s
t
i
c
s
)
f
r
o
m
r
e
g
r
e
s
s
i
o
n
s
o
f
t
h
e
f
o
r
m
b
P
_
L
I
T
t
0
.
1
6
6
(
1
.
7
8
)
*
0
.
1
1
9
(
1
.
7
8
)
*
2
0
.
4
0
8
(
2
3
.
1
9
)
*
*
*
0
.
2
4
9
(
1
.
4
4
)
0
.
5
2
3
(
1
.
9
2
)
*
0
.
3
6
6
(
1
.
9
0
)
*
S
a
r
b
O
x
t
0
.
0
0
0
(
0
.
0
0
)
0
.
0
1
2
(
1
.
6
5
)
*
2
0
.
0
1
7
(
2
1
.
2
1
)
2
0
.
0
1
2
(
2
0
.
6
3
)
0
.
0
3
2
(
1
.
0
7
)
0
.
0
0
3
(
0
.
1
4
)
P
_
L
I
T
*t
S
a
r
b
O
x
t
0
.
0
4
7
(
2
0
.
4
8
)
2
0
.
2
0
5
(
2
2
.
9
2
)
*
*
*
0
.
1
3
9
(
1
.
0
4
)
0
.
0
1
4
(
0
.
0
7
)
2
0
.
4
4
1
(
2
1
.
5
4
)
2
0
.
2
8
8
(
2
1
.
4
3
)
S
I
Z
E
t
2
0
.
0
0
9
(
2
4
.
6
9
)
*
*
*
2
0
.
0
0
5
(
2
3
.
3
1
)
*
*
*
0
.
0
1
6
(
5
.
9
6
)
*
*
*
0
.
0
0
3
(
0
.
7
5
)
2
0
.
0
3
5
(
2
5
.
9
9
)
*
*
*
2
0
.
0
1
6
(
2
3
.
8
4
)
*
*
*
L
E
V
t
2
0
.
0
3
3
(
2
4
.
3
5
)
*
*
*
0
.
0
6
9
(
1
2
.
9
2
)
*
*
*
2
0
.
0
0
9
(
2
0
.
9
2
)
0
.
0
4
4
(
3
.
1
4
)
*
*
*
2
0
.
0
3
5
(
2
1
.
6
1
)
2
0
.
0
0
1
(
2
0
.
0
4
)
R
O
A
t
2
0
.
0
2
0
(
2
1
.
7
9
)
*
0
.
0
0
1
(
0
.
0
8
)
0
.
2
7
0
(
1
7
.
5
4
)
*
*
*
2
0
.
0
4
7
(
2
2
.
2
6
)
*
*
2
0
.
3
5
9
(
2
1
0
.
9
7
)
*
*
*
2
0
.
1
3
6
(
2
5
.
8
8
)
*
*
*
L
O
S
S
t
2
0
.
0
3
6
(
2
5
.
0
8
)
*
*
*
2
0
.
0
0
1
(
2
0
.
2
5
)
2
0
.
0
6
0
(
2
6
.
1
5
)
*
*
*
0
.
0
1
7
(
1
.
3
1
)
0
.
0
7
6
(
3
.
6
9
)
*
*
*
0
.
0
3
3
(
2
.
2
9
)
*
*
L
I
T
R
I
S
K
t
2
0
.
0
1
5
(
2
1
.
7
5
)
*
0
.
0
0
5
(
0
.
7
7
)
0
.
0
3
0
(
2
.
6
3
)
*
*
*
2
0
.
0
0
6
(
2
0
.
3
8
)
2
0
.
0
2
0
(
2
0
.
8
3
)
0
.
0
0
4
(
0
.
2
5
)
M
T
B
t
0
.
0
0
0
(
1
.
3
4
)
0
.
0
0
0
(
0
.
8
3
)
2
0
.
0
0
1
(
2
2
.
9
3
)
*
*
*
2
0
.
0
0
1
(
2
1
.
3
0
)
0
.
0
0
3
(
3
.
7
5
)
*
*
*
0
.
0
0
1
(
2
.
1
7
)
*
*
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
A
d
j
u
s
t
e
d
R
2
(
%
)
3
.
7
3
9
.
6
4
3
3
.
8
9
4
.
0
3
2
4
.
3
4
1
3
.
5
5
N
o
t
e
s
:
S
i
g
n
i
?
c
a
n
c
e
a
t
:
*
1
0
,
*
*
5
a
n
d
*
*
*
1
p
e
r
c
e
n
t
l
e
v
e
l
s
(
o
n
e
-
t
a
i
l
e
d
)
,
r
e
s
p
e
c
t
i
v
e
l
y
;
n
¼
1
,
7
8
6
;
a
l
l
v
a
r
i
a
b
l
e
s
a
r
e
d
e
?
n
e
d
i
n
A
p
p
e
n
d
i
x
1
:
a
D
E
P
E
N
D
E
N
T
i
t
¼
a
1
i
þ
b
1
i
P
_
L
I
T
t
þ
b
2
i
S
I
Z
E
t
þ
b
3
i
L
E
V
t
þ
b
4
i
R
O
A
t
þ
b
5
i
M
T
B
t
þ
b
6
i
L
O
S
S
t
þ
b
7
i
L
I
T
R
I
S
K
t
þ
X
j
b
j
i
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
b
D
E
P
E
N
D
E
N
T
i
t
¼
a
1
i
þ
b
1
i
P
_
L
I
T
t
þ
b
2
i
S
a
r
b
O
x
t
þ
b
3
i
L
I
T
*t
S
a
r
b
O
x
t
þ
b
4
i
S
I
Z
E
t
þ
b
5
i
L
E
V
t
þ
b
6
i
R
O
A
t
þ
b
7
i
M
T
B
t
þ
b
8
i
L
O
S
S
t
þ
b
9
i
L
I
T
R
I
S
K
t
þ
X
j
b
j
i
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
w
h
e
r
e
D
E
P
E
N
D
E
N
T
i
t
r
e
p
r
e
s
e
n
t
s
d
i
s
c
r
e
t
i
o
n
a
r
y
a
c
c
r
u
a
l
o
r
a
b
n
o
r
m
a
l
r
e
a
l
a
c
c
o
u
n
t
v
a
r
i
a
b
l
e
Table VI.
Multivariate tests of
differences between sued
and non-sued SEOs
ARJ
24,1
70
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ProbðLIT
t
Þ ¼a
1
þb
1
DAC
t
þb
2
DAR
t
þb
3
ACFO
t
þb
4
APROD
t
þb
5
ADEXP
t
þb
6
SarbOx
t
þb
7
DAC
*
t
SarbOx
t
þb
8
DAR
*
t
SarbOx
t
þb
9
ACFO
*
t
SarbOx
t
þb
10
APROD
*
t
SarbOx
t
þb
11
ADEXP
*
t
SarbOx
t
þb
12
AUD
t
þb
13
Principal
t
þb
14
Secondary
t
þb
15
High 2Tech
t
ð6Þ
where LIT is an indicator variable that takes on the value of 1 if the SEO is
subsequently sued for accounting irregularities, and 0 if it is not subsequently sued,
DAC
t
, DAR
t
, ACFO
t
, APROD
t
, and ADEXP
t
are the discretionary/abnormal accrual
and real accounts as previously de?ned. AREAL
t
is not included in the regression since
it is the sum of the three abnormal real accounts. The second regression includes the
SarbOx variable as well as interaction variables for SarbOx and all
discretionary/abnormal accrual and real accounts. The control variables used are:
AUD
t
¼ indicator variable that takes on the value of 1 if the auditor is a
prestigious one (de?ned as Compustat item no. 149 (the auditor
component) between 1 and 8, and 0 otherwise.
Principal
t
¼ proceeds from offer (offer size)/beginning total assets.
Secondary
t
¼ secondary shares issued/total shares issued.
High-Tech
t
¼ indicator variable that takes on the value of 1 if the SEO belongs to
a high-technology industry and 0 otherwise.
The high-technology industries include computers and of?ce equipment, consumer
electronics, communications equipment, electronic components and accessories,
semiconductors, industrial electronics, photonics, defense electronics, electro-medical
equipment, communications services, and software and computer-related services[12].
The variable AUD is used as a control variable since offerings with prestigious
auditors are unlikely to attract lawsuits because the risk of dramatically poor post-offer
stock returns is low. On the other hand, auditors may be named as codefendants,
alongside offering ?rms. The deep pockets that tend to accompany prestige may attract
lawsuits. Therefore, it is unclear what the direction of the relationship between litigation
and AUD will be. The variable Principal is used to control for the offer size which may
also affect the incidence of lawsuits. If the offer is small, the potential for dollar damages
to participating investors is also small. It may not be worth suing ?rms over a small
offer, if there are ?xed costs of litigation. On the other hand, well-known established
?rms make most of the large offers. These offers may be among the least risky and
therefore least likely to precede very low rates of return to stockholders. The variable
Secondary is used as the fraction of the offer that is secondary also has an impact, albeit
ambiguous, on litigation risk. This fraction may be large when a ?rm’s principal
shareholders are substantially divesting their stakes and may seek to deceive investors
about factors affecting ?rm value. Deceptive behavior in connection with a stock offer
would tend to increase the risk of a subsequent lawsuit. However, investors and
regulatory authorities may closely scrutinize offers that contain large secondary
fractions. Other things equal, this would increase the chances that illegal deceptive
behavior is later exposed and punished. The increased risk of punishment would tend
Real and accrual-
based earnings
71
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
to discourage deceptive behavior and reduce the incidence of related lawsuits. Hence, it
is not clear whether, on balance, the incidence of lawsuits should be positively or
negatively related to the fraction of an offer that is secondary. Finally, speci?c industries
may also attract more litigation. For example, in the period after year 2000, there was an
abundance of litigation targeted towards ?rms in the high-technology industry.
Since the dependent variable, LIT, is a dichotomous variable, ordinary least square
is not appropriate for estimating the coef?cients. We instead use probit regression
using a logistic distribution. In essence, the regression measures the probability of
being sued given the independent variables in the model. The results appear in
Table VII. To test our H1, we examine the coef?cients b
1
2 b
5
. We expect that b
1
(and
perhaps b
2
) will be positive and signi?cant.
We ?nd that discretionary accruals (DAC
t
) are positively associated with the
probability of litigation (coef?cient ¼ 1.784, signi?cant at the 10 percent level). This
means that SEOs are almost twice as likely to be sued if their discretionary accruals are
higher by one unit (in this case an amount equivalent to 1 percent of total assets).
Abnormal cash from operations is also associated with the probability of litigation
Indep. variable Without SarbOx interaction terms With SarbOx interaction terms
DAC
t
1.784 (3.77)
*
2.163 (4.69)
* *
DAR
t
20.298 (0.07) 20.038 (0.00)
ACFO
t
1.338 (3.20)
*
2.061 (6.39)
* *
APROD
t
0.672 (1.43) 0.942 (2.30)
ADEXP
t
0.536 (2.07) 0.842 (4.07)
* *
SarbOx
t
0.303 (1.56)
DAC
*
t
SarbOx
t
22.000 (0.64)
DAR
*
t
SarbOx
t
26.884 (2.81)
*
ACFO
*
t
SarbOx
t
25.283 (6.64)
* * *
APROD
*
t
SarbOx
t
21.220 (0.56)
ADEXP
*
t
SarbOx
t
22.139 (4.71)
* *
AUD
t
0.826 (3.40)
*
0.890 (3.84)
*
Principal
t
0.057 (2.96)
*
0.067 (3.62)
*
Secondary
t
21.001 (5.67)
* *
20.952 (5.02)
* *
High-Tech
t
0.783 (18.09)
* * *
0.800 (18.55)
* * *
Log Likelihood 2454.74 2450.28
Notes: Signi?cance levels of
*
10,
* *
5 and
* * *
1 percent (one-tailed), respectively; n ¼ 1,786; all
variables are de?ned in Appendix 1; the table presents coef?cients (x
2
-values) from the following
probit regressions:
ProbðLIT
t
Þ ¼ a
1
þb
1
DAC
t
þb
2
DAR
t
þb
3
ACFO
t
þb
4
APROD
t
þb
5
ADEXP
t
þb
6
AUD
t
þb
7
Principal
t
þb
8
Secondary
t
þb
9
High 2Tech
t
ProbðLIT
t
Þ ¼a
1
þb
1
DAC
t
þb
2
DAR
t
þb
3
ACFO
t
þb
4
APROD
t
þb
5
ADEXP
t
þb
6
SarbOx
t
þb
7
DAC
*
t
SarbOx
t
þb
8
DAR
*
t
SarbOx
t
þb
9
ACFO
*
t
SarbOx
t
þb
10
APROD
*
t
SarbOx
t
þb
11
ADEXP
*
t
SarbOx
t
þb
12
AUD
t
þb
13
Principal
t
þb
14
Secondary
t
þb
15
High 2Tech
t
Table VII.
Tests of probability
of litigation
ARJ
24,1
72
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(coef?cient for ACFO
t
¼ 1.338 and signi?cant at the 10 percent level). However,
this indicates that the probability of litigation increases as abnormal cash ?ows increase
which is inconsistent with income-increasing manipulation. All other abnormal
accounts are not signi?cantly related to the probability of litigation. We ?nd that the
probability of litigation is signi?cantly related to the type of auditor (coef?cient of
AUD
t
¼ 0.826, signi?cant at the 10 percent level), indicating that ?rms are more likely to
be sued if the auditor is a prestigious one. The likelihood of litigation is also positively
related to the principal amount of the offering (coef?cient of Principal
t
¼ 0.057,
signi?cant at the 10 percent level), and being ina high-technology industry(coef?cient of
High-Tech
t
¼ 0.783, signi?cant at the 1 percent level). On the other hand, the likelihood
of litigation is negatively associated with the share of the offer that is secondary
(coef?cient of Secondary
t
¼ 21.001, signi?cant at the 5 percent level).
The second regression shows the effect of SarbOx on the probability of litigation. The
coef?cients on the interaction terms between SarbOx and the discretionary/abnormal
accruals and real accounts show how the probability of litigation has changed
following the enactment of SarbOx. We expect a signi?cant coef?cient for one or more
of the real manipulation measures (b
9
2 b
11
). We ?nd that investors are paying less
attention to income-increasing discretionary accounts receivable following
SarbOx (coef?cient ¼ 26.884, signi?cant at the 10 percent level). However, ?rms
with lower abnormal cash from operations are scrutinized more in the SarbOx period
(coef?cient ¼ 25.283, signi?cant at the 1 percent level). Also, ?rms with lower
abnormal discretionary expenses are more likely to be sued (coef?cient ¼ 22.139,
signi?cant at the 5 percent level). This provides some support for our H2.
Overall, we ?nd that the probability of litigation is associated with
income-increasing accrual earnings management as well as income-decreasing real
account manipulation. The enactment of SarbOx has shifted the focus somewhat from
accrual to real account manipulation.
5. Conclusion
In this paper, we study the alternative methods of manipulating bottom-line earnings
in a SEO setting and address whether litigation for accounting irregularities varies
with the method(s) used to reach earnings goals around these stock issues.
We identify a sample of 1,871 SEO observations and compare them to a matched
sample of 1,871 ?rm/year observations that did not undergo an equity offering. We ?nd
that the SEO ?rms engage in income-increasing manipulation using more than one
method. Speci?cally, they engage in both accrual and real account manipulation in the
year prior to the offering, through discretionary accruals, and by increasing sales
through more lenient terms (shown as a lower level of abnormal cash from operations).
SEOs in general have higher abnormal discretionary expenses but this is most likely
related to the types of ?rms that initiate SEOs. We then examine which of the ?rms
were sued from the above sample. We ?rst employ a two-stage least square
methodology to account for endogeneity between the probability of litigation and the
accrual and real account manipulation measures and ?nd that ?rms that are
subsequently sued by investors following the SEO have a higher prevalence of
discretionary accruals, when the lawsuit allegations are related to accounting issues.
This is indicative of earnings manipulation using accrual accounts and is
consistent with prior research on litigation subsequent to SEOs (Ibrahim et al., 2008).
Real and accrual-
based earnings
73
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
There is no evidence of overall real account manipulation in the sued SEO ?rms but
there is some evidence of one particular type of real manipulation. These results are in
line with our expectations. Our H1 states that SEO ?rms that are subsequently sued for
accounting irregularities will have a higher prevalence of accrual manipulation but not
real account manipulation when compared to non-defendant SEO ?rms. In robustness
tests, we test how the probability of litigation is affected by these manipulation
measures. We ?nd that SEOs are more likely to be sued when they have higher
income-increasing discretionary accruals. There is no evidence that the probability of
litigation is related to income-increasing real account manipulation. These ?ndings are
important to academicians as well as investors. They indicate that even though there is
evidence of real account manipulation, it appears that investors still focus on accrual
manipulation in initiating litigation against ?rms and may not able to detect real
account manipulation. This paper provides the ?rst evidence of the relationship
between real account manipulation and litigation risk.
The SarbOx Act of 2002 was designed to reduce earnings management, speci?cally
through accrual accounts. There is evidence that this was successful in the SEO
setting. There is some evidence that SEO ?rms have shifted to real account
manipulation post-SarbOx. This is in line with prior research and indicates that accrual
and real account manipulation are substitutes used to manage earnings. Following the
enactment of SarbOx, we ?nd that investors are paying less attention to one type of
accrual manipulation, namely account receivable manipulation. We also hypothesize
and ?nd some support that investors may be paying more attention to real
manipulation measures in their litigation decisions in the post-SarbOx period.
Some limitations of the study include generalisability and methodology issues.
Speci?cally, this study focuses on a very speci?c setting, SEOs. The conclusions of the
study may not be generalisable to other settings. This is an avenue for future research.
Furthermore, the post-SarbOx period is small due to data limitations which may
impact the results. Finally, as in all earnings management studies, the manipulation
measures may not fully capture the magnitude and nature of manipulation.
Notes
1. The offerings were made in the years 1991-2005; however our test sample is in the years
1990-2004 since we test manipulation in the year prior to the offering.
2. The full act can be accessed at: www.gpo.gov/fdsys/pkg/PLAW-107publ204/pdf/PLAW-
107publ204.pdf
3. We use the four-digit SIC code to designate an industry.
4. We eliminate ?rms that underwent an IPO from the non-SEO observations.
5. The distribution of the matched ?rms is identical.
6. The ?rm and industry subscripts have been suppressed from this and following regressions.
7. We omit DAR
t
from the right-hand side of the regression, unlike Ibrahim (2009).
8. If SG&A is not missing but one or both of the other discretionary expenses are missing, they
are set to 0.
9. Abnormally increasing production, keeping the level of sales constant, would shift some of
the costs to the balance sheet, in the form of unsold inventory, abnormally decreasing cost of
goods sold.
ARJ
24,1
74
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
10. Francis et al. designate ?rms in the biotechnology, computers, electronics, and retailing
industries as subject to high litigation risk.
11. See Table VIII in DuCharme et al. (2004). Their test variable is abnormal working capital
rather than the breakdown of discretionary/abnormal accruals and real accounts in this
paper. They also include a variable for cumulative abnormal return and the type of
underwriter. We cannot obtain information regarding the type of underwriter and thus do
not include it in the regressions. In untabulated results, we include a measure of cumulative
abnormal returns over the 30 days following the offering date, with similar results. The
number of observations was lower due to missing data.
12. The speci?c SIC codes designated as high tech are available upon request.
References
Aharony, J., Lin, C.J. and Loeb, M.P. (1993), “Initial public offerings, accounting choices, and
earnings management”, Contemporary Accounting Research, Vol. 10, pp. 61-81.
Alexander, J.C. (1991), “Do the merits matter? A study of settlements in securities class actions”,
Stanford Law Review, Vol. 43, pp. 497-598.
Armstrong, C., Foster, G. and Taylor, D. (2009), “Earnings management around initial public
offerings: a re-examination”, Rock Center for Corporate Governance Working Paper No. 23,
University of Pennsylvania, Philadelphia, PA, September.
Ball, R. and Shivakumar, L. (2008), “Earnings quality at initial public offerings”, Journal of
Accounting and Economics, Vol. 45, pp. 324-49.
Cohen, D.A. andZarowin, P. (2010), “Accrual-basedandreal earnings management activities around
seasoned equity offerings”, Journal of Accounting and Economics, Vol. 50 No. 1, pp. 2-19.
Cohen, D.A., Dey, A. and Lys, T. (2008), “Real and accrual-based earnings management in the
pre- and post-Sarbanes Oxley periods”, The Accounting Review, Vol. 83, May, pp. 757-88.
Dechow, P., Sloan, R. and Sweeney, A. (1995), “Detecting earnings management”, The Accounting
Review, Vol. 70, pp. 193-225.
DuCharme, L.L., Malatesta, P.H. and Sefcik, S.E. (2004), “Earnings management, stock issues,
and shareholder lawsuits”, Journal of Financial Economics, Vol. 71 No. 1, pp. 27-49.
Francis, J.D., Philbrick, D. and Schipper, K. (1994), “Shareholder litigation and corporate
disclosures”, Journal of Accounting Research, Vol. 32, pp. 137-164.
Frankel, R.M., Johnson, M.F. and Nelson, K.K. (2002), “The relation between auditors’ fees for
nonaudit services and earnings management”, The Accounting Review, Vol. 77, pp. 71-105
(supplement).
Friedlan, J. (1994), “Accounting choices by issuers of initial public offerings”, Contemporary
Accounting Research, Vol. 11, pp. 1-32.
Graham, J.R., Harvey, C.R. and Rajgopal, S. (2005), “The economic implications of corporate
?nancial reporting”, Journal of Accounting and Economics, Vol. 40, pp. 3-73.
Ibrahim, S. (2009), “The usefulness of measures of consistency of discretionary components of
accruals in the detection of earnings management”, Journal of Business Finance &
Accounting, Vol. 9/10, pp. 1087-116.
Ibrahim, S., Xu, L. and Deal, L. (2008), “The legal consequences of management of earnings
components: evidence from equity offerings”, working paper, Morgan State University,
Baltimore, MD, June.
Real and accrual-
based earnings
75
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Jones, C.L. (1998), “The association of earnings management with current returns, current market
values, future returns, executive compensation and the likelihood of being a target of 10B-5
litigation”, unpublished dissertation, Stanford University, Palo Alto, CA.
Jones, J. (1991), “Earnings management during import relief investigation”, Journal of Accounting
Research, Vol. 29 No. 22, pp. 193-228.
Kothari, S.P., Leone, A. and Wasley, C. (2005), “Performance matched discretionary accrual
measures”, Journal of Accounting and Economics, Vol. 39, February, pp. 163-97.
Li, H., Pincus, M. and Rego, S.O. (2008), “Market reaction to events surrounding the
Sarbanes-Oxley Act of 2002 and earnings management”, The Journal of Law &Economics,
Vol. 51, pp. 111-34.
Lu, Y. (2004), “Earnings management and securities litigation”, working paper, University of
Southern California, Los Angeles, CA.
Marquardt, C. and Wiedman, C. (2004), “How are earnings managed? An examination of speci?c
accruals”, Contemporary Accounting Research, Vol. 21, Summer, pp. 461-89.
Mizik, N. and Jacobson, R. (2007), “Earnings in?ation through accruals and real activity
manipulation: its prevalence at the time of an SEOand the ?nancial market consequences”,
working paper, Columbia University, SSRN Abstract No. 1031006, June, New York, NY.
Palmrose, Z. (1991), “Trials of legal disputes involving independent auditors: some empirical
evidence”, Journal of Accounting Research, Vol. 29, pp. 149-85 (supplement).
Palmrose, Z., Richardson, V. and Scholz, S. (2004), “Determinants of market reactions to
restatement announcements”, Journal of Accounting and Economics, Vol. 37, pp. 59-89.
Priest, G.L. and Klein, B. (1984), “The selection of disputes for litigation”, Journal of Legal Studies,
Vol. 13, pp. 1-55.
Rangan, S. (1998), “Earnings before seasoned equity offerings: are they overstated?”, Journal of
Financial Economics, Vol. 50, pp. 101-22.
Roychowdhury, S. (2006), “Earnings management through real activities manipulation”, Journal
of Accounting and Economics, Vol. 42, pp. 335-70.
Shivakumar, L. (2000), “Do ?rms mislead investors by overstating earnings before seasoned
equity offerings?”, Journal of Accounting and Economics, Vol. 29 No. 3, pp. 339-71.
Teoh, S.H., Welch, I. and Wong, T.J. (1998a), “Earnings management and the post-issue
performance of seasoned equity offerings”, Journal of Financial Economics, Vol. 50,
pp. 63-99.
Teoh, S.H., Wong, T.J. and Rao, G.R. (1998b), “Are accruals during initial public offerings
opportunistic?”, Review of Accounting Studies, Vol. 3, pp. 175-208.
Zang, A. (2007), “Evidence on the tradeoff between real manipulation and accrual manipulation”,
working paper, Hong Kong University of Science and Technology, Kowloon, March.
Further reading
PricewaterhouseCoopers (2000), Securities Litigation Study, available at:http://10b5.pwc.com/
PDF/2000STUDY.PDF (accessed June 2007).
ARJ
24,1
76
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Appendix 1
Variables De?nition and data source
Financial and accrual variables
A
t
Total assets in year t [6]
INC
t
Income before extraordinary items in year t [18]/A
t21
(total assets in year t 2 1) [6]
REV
t
Net revenues (sales) in year t [12]/A
t21
DREV
t
Change in net sales from year t 2 1 to year t [D12]/A
t21
DREV
t21
Change in net sales from year t 2 2 to year t 2 1 [D12]/A
t21
DAR
t
Change in accounts receivable from year t 2 1 to t [D2]/A
t21
PPE
t
Gross property, plant, and equipment in year t [7]/A
t21
CFO
t
Cash from operations in year t [308 2 124]/A
t21
COGS
t
Cost of goods sold in year t [41]/A
t21
DINV
t
Change in inventory from year t 2 1 to year t [D3]/A
t21
PROD
t
Production costs in year t [41 þ D3]/A
t21
ADV
t
Advertising expenses in year t [45]/A
t21
RD
t
Research and development expenses in year t [46]/A
t21
SGA
t
Selling, general and administrative expenses in year t [189]/A
t21
DEXP
t
Discretionary expenses in year t [45 þ 46 þ 189]/A
t21
TAC
t
Total accruals ¼ INC
t
2 CFO
t
ROA
t
Return on assets in year t ¼ NI
t
[172]/A
t
Discretionary accruals and abnormal real accounts
DAC
t
Discretionary accruals in year t from performance-modi?ed Jones
model ¼ TAC
t
2b
0
þ b
1
ð1=A
t21
Þ þ b
2
ðDREV
t
2DAR
t
Þ þ b
3
PPE
t
þ b
4
ðROA
t
Þ
DAR
t
Discretionary accounts receivable in year t from performance-modi?ed Jones
model ¼ DAR
t
2b
01
þ b
11
ð1=A
t21
Þ þ b
21
ðDREV
t
Þ þ b
31
PPE
t
þ b
41
ðROA
t
Þ
ACFO
t
Abnormal or discretionary cash ?ow from operations in year
t ¼ CFO
t
2a
11
ð1=A
t21
Þ þ a
12
ðREV
t
Þ þ a
13
ðDREV
t
Þ
APROD
t
Abnormal or discretionary production
costs ¼ PROD
t
2a
21
ð1=A
t21
Þ þ a
22
ðREV
t
Þ þ a
23
ðDREV
t
Þ þ a
24
ðDREV
t21
Þ
ADEXP
t
Abnormal or discretionary expenditures ¼ DEXP
t
2a
31
ð1=A
t21
Þ þ a
32
ðREV
t21
Þ
AREAL
t
Total abnormal real accounts ¼ ACFO
t
þ APROD
t
þ ADEXP
t
Control variables in multivariate tests
SIZE
t
Log (A
t
)
LEV
t
Leverage ¼ (short-term debt [34] þ long-term debt [9])/A
t
LOSS
t
An indicator variable equal to one if the observation has negative earnings before extraordinary
items in year t, and zero otherwise
MTB
t
Market-to-book value at year end (market value of equity divided by book value of equity)
LITRISK
t
An indicator variable equal to one if the observation is from an industry with high litigation risk,
and zero otherwise
AUD
t
An indicator variable equal to one if the auditor is a prestigious one (de?ned as a Big 4 auditor or
predecessor), and zero otherwise
Principal
t
Proceeds from issuance/A
t21
Secondary
t
Secondary shares issued/total share issued
High-Tech
t
An indicator variable equal to one if the observation belongs to a high-technology industry, and
zero otherwise
Test variables in multivariate tests
SEO An indicator variable equal to one for SEO observation in the year prior to the offering, and zero
for non-SEO observation
LIT An indicator variable equal to one if the SEO ?rm was subsequently sued for accounting
irregularities, and zero if the SEO ?rm was not subsequently sued
P_LIT Predicted value of probability of litigation measured from probit regression of LIT on exogenous
variables
SarbOx An indicator variable equal to one if the observation is in year 2002-2004, and zero otherwise
Note: All Compustat data item numbers are provided in square brackets
Table AI.
Variable de?nitions
Real and accrual-
based earnings
77
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Appendix 2. Regression used in two-stage least square analysis
The ?rst stage of the two-stage least square measures the predicted litigation based on various
exogenous variables related to litigation as follows:
ProbðLIT
t
Þ ¼a
1
þb
1
AUD
t
þb
2
Principal
t
þb
3
Secondary
t
þb
4
SIZE
t
þb
5
LEV
t
þb
6
ROA
t
þb
7
LOSS
t
þb
8
LITRISK
t
þb
9
MTB
t
þ
X
j
b
j
IndustryDummies
t
þ1
where:
LIT ¼ an indicator variable that takes on the value of 1 if the SEO is
subsequently sued for accounting irregularities, and 0 if it is not
subsequently sued.
AUD ¼ indicator variable that takes on the value of 1 if the auditor is a
prestigious one (de?ned as Compustat item no. 149 (the auditor
component) between 1 and 8) and 0 otherwise.
Principal ¼ proceeds from offer (offer size)/beginning total assets.
Secondary ¼ secondary shares issued/total shares issued.
SIZE
t
¼ Log (total assets at year end).
LEV
t
¼ leverage ¼ total of short- and long-term debt scaled by average total
assets.
ROA
t
¼ return on assets measured as net income divided by total assets.
MTB
t
¼ market-to-book value measured as market value of assets divided
by book value of assets at year end.
LOSS
t
¼ indicator variable that takes on the value 1 if the ?rm has a net loss,
and 0 otherwise.
LITRISK
t
¼ indicator variable that takes on the value 1 if the ?rm operates in
a high-risk environment, as de?ned in Francis et al. (1994), and
0 otherwise.
Industry Dummies
t
¼ indicator variable that takes on the value of 1 if observation belongs
to a speci?c industry, and 0 otherwise.
The predicted value of litigation probability from the above regression is used as an independent
variable in regressions (3) and (4).
Corresponding author
Salma Ibrahim can be contacted at: [email protected]
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
ARJ
24,1
78
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Richard Kent, James Routledge. 2015. Use of benchmarks in predicting earnings management?. Accounting
& Finance n/a-n/a. [CrossRef]
2. Walter Aerts, Shuyu Zhang. 2014. Management’s causal reasoning on performance and earnings
management. European Management Journal 32, 770-783. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_401618023.pdf
Prior research suggests that firms manipulate earnings through accruals to achieve
certain reporting objectives. Recently, especially following the Sarbanes-Oxley (SarbOx) Act,
researchers have turned their attention to real account manipulation as an alternative. However, there
is no evidence on whether the likelihood of being detected by outsiders is different for firms using
these alternative manipulation methods. The purpose of this paper is to examine this research question
in the context of seasoned equity offerings (SEOs).
Accounting Research Journal
Real and accrual-based earnings management and its legal consequences: Evidence
from seasoned equity offerings
Salma Ibrahim Li Xu Genese Rogers
Article information:
To cite this document:
Salma Ibrahim Li Xu Genese Rogers, (2011),"Real and accrual-based earnings management and its legal
consequences", Accounting Research J ournal, Vol. 24 Iss 1 pp. 50 - 78
Permanent link to this document:http://dx.doi.org/10.1108/10309611111148779
Downloaded on: 24 January 2016, At: 21:13 (PT)
References: this document contains references to 31 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 2129 times since 2011*
Users who downloaded this article also downloaded:
Tony Mortensen, Richard Fisher, (2011),"The meaning of cash in the context of alternative accounting
standards: IFRS convergence and classification decisions", Accounting Research J ournal, Vol. 24 Iss 1 pp.
23-49http://dx.doi.org/10.1108/10309611111148760
Magdy Farag, Rafik Elias, (2011),"Relative audit fees and client loyalty in the audit market", Accounting
Research J ournal, Vol. 24 Iss 1 pp. 79-93http://dx.doi.org/10.1108/10309611111148788
Arifur Khan, Paul Mather, (2013),"The value of executive director share ownership and discretionary
accruals", Accounting Research J ournal, Vol. 26 Iss 1 pp. 35-55http://dx.doi.org/10.1108/
ARJ -02-2012-0011
Access to this document was granted through an Emerald subscription provided by emerald-srm:115632 []
For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for
Authors service information about how to choose which publication to write for and submission guidelines
are available for all. Please visit www.emeraldinsight.com/authors for more information.
About Emerald www.emeraldinsight.com
Emerald is a global publisher linking research and practice to the benefit of society. The company
manages a portfolio of more than 290 journals and over 2,350 books and book series volumes, as well as
providing an extensive range of online products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee
on Publication Ethics (COPE) and also works with Portico and the LOCKSS initiative for digital archive
preservation.
*Related content and download information correct at time of download.
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Real and accrual-based earnings
management and its legal
consequences
Evidence from seasoned equity offerings
Salma Ibrahim
Department of Accounting and Finance, Kingston University,
Kingston upon Thames, UK
Li Xu
Department of Accounting and Finance, Southern Illinois University,
Carbondale, Illinois, USA, and
Genese Rogers
Department of Accounting and Finance, Morgan State University,
Baltimore, Maryland, USA
Abstract
Purpose – Prior research suggests that ?rms manipulate earnings through accruals to achieve
certain reporting objectives. Recently, especially following the Sarbanes-Oxley (SarbOx) Act,
researchers have turned their attention to real account manipulation as an alternative. However, there
is no evidence on whether the likelihood of being detected by outsiders is different for ?rms using
these alternative manipulation methods. The purpose of this paper is to examine this research question
in the context of seasoned equity offerings (SEOs).
Design/methodology/approach – First, the authors compare SEOs to a matched sample of
non-SEOs to document income-increasing manipulation. Next, they identify SEOs that prompt
lawsuits and compare sued and non-sued ?rms to determine whether using a particular method of
manipulation is more likely to be detected and associated with litigation.
Findings – The authors ?nd evidence of income-increasing accrual and real manipulation for SEOs
in the year prior to the offering in the pre-SarbOx period, and ?nd some evidence of a shift to real
account manipulation post-SarbOx. The authors examine the subsequent litigation pattern of these
SEOs, and ?nd that ?rms that are subsequently sued have a higher prevalence of income-increasing
discretionary accruals when the lawsuit allegations involve accounting issues. Following SarbOx,
investors are paying less attention to accrual manipulation through accounts receivable and there is
more scrutiny of real account manipulation.
Originality/value – The implication in this paper is that ?rms that engage in income-increasing
earnings management are more likely to be sued when they engage in accrual manipulation while
other forms of manipulation may be less understood. This ?nding is important to investors and
regulators.
Keywords United States of America, Accrual manipulation, Real manipulation, Earnings management,
Seasoned equity offerings, Sarbanes-Oxley, Litigation
Paper type Research paper
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
Data are available from the authors upon request.
ARJ
24,1
50
Accounting Research Journal
Vol. 24 No. 1, 2011
pp. 50-78
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309611111148779
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
1. Introduction
Prior research suggests that managers of ?rms manipulate earnings through accruals
under various situations, including around public equity offerings. Recently, especially
following the Sarbanes-Oxley (SarbOx) Act of 2002, researchers have turned their
attention to real account manipulation as an alternative to accrual manipulation,
e.g. through eliminating discretionary expenses, such as advertising or research and
development expenditures, or through accelerating the timing of sales through
increased price discounts or more lenient credit terms, or through reporting lower costs
of goods sold through increased production (Cohen et al., 2008). However, there is no
evidence on whether the likelihood of the manipulation being detected by investors and
outsiders, and of the ?rm facing litigation is different when employing these
alternative manipulation methods. This study addresses this gap and speci?cally
answers the following question: whether the likelihood of litigation varies with the
method(s) used to reach earnings goals around these stock issues. This research
question is addressed by examining the differences in manipulation behavior between
?rms that are subject to litigation and those that are not, in the seasoned equity
offering (SEO) setting. We also examine the effect of the SarbOx Act; the passage of the
Act was in part due to the increase in earnings management through accrual
manipulation and was aimed at reducing its prevalence. Whereas the Act has reduced
accrual manipulation (Cohen et al., 2008), prior evidence suggests that ?rms have
shifted to real account manipulation. Whether the stakeholders are able to see through
this is an important question for investors and other stakeholders of public ?rms,
as well as academicians and public policy makers.
We argue that the dif?culty for outside investors to detect earnings management
varies with the alternative methods used to manipulate earnings. We expect that,
for all ?rms that engage in earnings management around equity offerings, earnings
manipulation through accruals is more likely to be detected than managing other
components of the income statement and balance sheet. We propose to explore the role
of alternative manipulation in litigation over equity offerings.
As in earlier work, we ?rst document the presence of real and accrual earnings
management in the equity offering setting using measures of discretionary/abnormal
accruals, expenditures and cash ?ows. We focus on SEOs as our test sample. This sample
has been frequently used in earnings management studies, since it is argued that there is
both the opportunity and the incentive by management to engage in income-increasing
manipulationaroundSEOs, e.g. Teohet al. (1998a), Rangan(1998) and Shivakumar (2000),
in order to increase the market price of the stock at the time of the offering.
The SarbOx Act of 2002 addresses internal control issues of ?rms andestablishes rules
for auditors and management to limit the probability of earnings management through
accruals. We examine the effect of SarbOx on earnings management behavior and extend
the research that shows that earnings management has shifted fromaccrual management
to real account management (Cohen et al., 2008) after the passage of SarbOx.
We next identify offers that prompt lawsuits from our equity offering population
and compare sued and non-sued ?rms to examine whether there are signi?cant
differences in the discretionary or abnormal component of accruals and real accounts
for these two groups. We argue that it is harder to prove that real account manipulation
has occurred as management and auditors can contend that these abnormalities are
normal changes required during the course of operations.
Real and accrual-
based earnings
51
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This paper contributes to the accounting and ?nance literature in several ways: ?rst,
to the best of our knowledge, our paper is the ?rst attempt to address the association
between both types of earnings management and litigation around equity offerings.
This study is important in light of the recent trend of earnings management in public
?rms, which ultimately hurts users of ?nancial statements, particularly investors.
Also, we provide further evidence on which methods are used to manage earnings
around equity offerings and how this has changed following the passage of SarbOx.
To test our hypotheses, we ?rst identify a sample of ?rms that underwent a SEO
and test which accounts show abnormal behavior (accrual or real accounts) in the year
prior to the offering. Our test sample includes 1,871 SEOs in the years 1990-2004[1].
Out of this sample, 218 ?rms subsequently faced litigation. Some of the allegations
(n ¼ 85) against SEOs involve the manipulation of the market price by underwriters at
the time of the offering and other non-accounting irregularities. These allegations are
unrelated to earnings manipulation; we therefore eliminate these litigation cases and
examine only 133 lawsuits that involve accounting allegations.
The results indicate that SEOs have a prevalence of income-increasing manipulation
using both accrual and some real account manipulation in the year prior to the offering.
We ?nd some evidence of a shift to real account manipulation in the post-SarbOx period.
Furthermore, we ?nd that SEO ?rms that are subject to class action litigation by
investors have a higher prevalence of discretionary accruals, which is consistent with
prior research on litigation that considered only accrual manipulation (Lu, 2004;
Ibrahim et al., 2008). On the other hand, there is no evidence of overall manipulation in
real accounts in sued SEOs, when compared to non-sued SEOs. However, sued SEOs
have lower abnormal cash?ows (one measure of real account manipulation). We also
?nd that the probability of litigation is directly related to the level of discretionary
accruals; the higher income-increasing discretionary accruals, the more likely the SEO
will result in accounting-related litigation. In the post-SarbOx period, there is less
emphasis on accrual manipulation through accounts receivable and more emphasis on
income-increasing abnormal accounts in litigation decisions. The implications of these
?ndings are that ?rms that engage in income-increasing earnings management through
accruals are more likely to be sued. Other forms of manipulation are less likely to be
considered by investors and regulators in initiating litigation. There is evidence that this
may have changed somewhat in the post-SarbOx period; investors are paying more
attention to non-accrual manipulation following SarbOx.
The rest of the paper is organized as follows. Section 2 presents prior research and
the development of the hypotheses. Data selection and methodology is presented in
Section 3, followed by the results in Section 4. Section 5 concludes.
2. Prior research and hypotheses development
This section discusses prior research related to accrual and real account manipulation
around equity offerings (Section 2.1) as well as the effect of SarbOx on earnings
management behavior (Section 2.2) and litigation risk in relation to earnings
management (Section 2.3). The hypotheses are developed in Section 2.4.
2.1 Earnings management around equity offerings
Prior research has documented that ?rms manage earnings upwards using
accruals around initial public offerings (IPOs) (Aharony et al., 1993; Friedlan, 1994;
ARJ
24,1
52
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Teoh et al., 1998b) and SEOs (Teoh et al., 1998a; Rangan, 1998; Shivakumar, 2000)
in order to increase the market price of the stock. The bene?ts from earnings
management in this context are especially high because the proceeds of an equity
offering are based on the stock price at one point in time. Prior studies also ?nd that in
order for earnings management to have the maximumprice impact, ?rms issuing equity,
either through initial or seasoned offerings, will prefer to manage earnings using
recurring rather than non-recurring income statement items. In particular, ?rms issuing
equity are likely to manage earnings through revenue accounts, and speci?cally through
accounts receivable (Marquardt and Wiedman, 2004). However, Ball and Shivakumar
(2008) and Armstrong et al. (2009) argue that results from prior research on earnings
management around large transactions and other large events, especially around an
IPO, may be biased for several reasons. Most notably, they argue that using the year of
the event as the test year may overestimate discretionary accruals, given that this year is
accompanied with high increases in working capital which will in?ate discretionary
accrual measures.
Research on real manipulation around equity offerings similarly provides evidence of
income-increasing manipulation. Cohen and Zarowin (2010) ?nd that ?rms use
income-increasing real, as well as accrual-based, earnings management tools in the
years surrounding SEOs and they show how the tendency to trade-off real versus
accrual-based earnings management activities varies by industry. Speci?cally, they ?nd
that SEOs have signi?cant positive abnormal production costs, which would lead to a
decrease in cost of goods sold, in the year of the offering. They also ?nd lower than
expected discretionary expenses, including advertising, research and development and
selling, general and administrative expenses in the years prior to as well as the year of
the offering. Finally, they ?nd lower than expected cash ?ows from operations,
indicating less cash receipts from offering more lenient terms to customers in the years
leading to as well as the year of the offering. All of these effects wouldtend to abnormally
increase earnings through real account changes. Mizik and Jacobson (2007) show that
earnings are in?ated around equity offerings using both real- and accrual-based
management. They also show that stock overvaluation is more prevalent for ?rms with
real manipulation, i.e. investors are fooled by this real account manipulation.
2.2 Earnings management and SarbOx
The SarbOx Act, passed in July 2002, was enacted to protect investors by improving
the accuracy and reliability of corporate disclosures made pursuant to the securities
laws[2]. Li et al. (2008) show that accrual earnings management has declined following
the passage of SarbOx. The results in Li et al. (2008) are consistent with investors
expecting that the more extensively ?rms managed their earnings, the more SarbOx
would constrain earnings management and enhance the quality of ?nancial reporting.
Accrual and real earnings management are seen as substitutes in managing
earnings (Graham et al., 2005; Zang, 2007). Cohen et al. (2008) test the shift between
accrual and real earnings management in the periods before and after the passage of
SarbOx. They ?nd that ?rms have shifted away from accrual management to real
account management following SarbOx. Their evidence implies that the increased
costs of detection of accrual management post-SarbOx has led to management turning
to other types of manipulation, even though they may be costlier to the ?rm in the
long run.
Real and accrual-
based earnings
53
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
2.3 Earnings management and litigation risk
Several studies have considered the role of accrual earnings management in class
action securities litigation. Jones (1998) studies the determinants of securities litigation
risk using a sample of 69 ?rms sued during the period 1989-1992 and a control sample
of non-sued ?rms with a 10 percent drop in stock price. He ?nds a negative but
insigni?cant association between litigation risk and abnormal current accruals
estimated from a term-adjusted version of Jones (1991) model. Lu (2004) examines the
relation between earnings management and securities class action litigation. In a
sample of 781 ?rms sued in class action securities litigation from 1988 to 2000, Lu ?nds
that accruals and the changes in revenue are abnormally high during alleged periods of
manipulation, and tend to reverse subsequently. Moreover, the magnitude of accruals
overstatement is greatest for defendant ?rms subject to SEC enforcement actions or
having made accounting restatements, and least for defendant ?rms not facing
accounting allegations.
Furthermore, data collected by PricewaterhouseCoopers in 2000 since the enactment
of the Private Securities Litigation Reform Act of 1995 show that the majority of suits
allege some sort of revenue recognition violation. This suggests that earnings
management through revenue misstatement increases the likelihood of litigation,
which is consistent with Palmrose et al. (2004) who demonstrate a positive relationship
between litigation and misstatement of revenue.
In sum, the evidence from prior literature shows that accrual earnings management
activities increase ?rms’ litigation risk in general.
In the context of equity offerings, DuCharme et al. (2004) examine the association
between accrual earnings management by ?rms issuing stocks and the incidence of
litigation, allegations of earnings management, and lawsuit settlement amounts. They
use a litigation sample consisting of 150 SEOs and 72 IPOs from 1988 to 1997, and a
control sample consisting of all SEOs and IPOs that are not subject to litigation during
the same periods. After controlling for characteristics of the stock offerings, they only
?nd a signi?cant and positive association between abnormal current accruals and the
incidence of lawsuit ?lings for the SEO ?rms, but not for the IPO ?rms. They also fail
to ?nd a signi?cant relation between abnormal current accruals and allegations of
earnings management. Ibrahim et al. (2008) also study the association between
litigation and discretionary components of accruals, namely accounts receivable,
inventory, accounts payable, other working capital, depreciation, and special items in
an equity offering setting. They ?nd a higher prevalence of accounts receivable
manipulation in litigated ?rms, consistent with revenue recognition problems.
There is no prior research studying litigation risk and real account manipulation.
However, prior research implies that real manipulation is less likely to result in
litigation, since it can be indistinguishable from optimal business decisions and thus
less likely to be detected (Graham et al., 2005; Cohen and Zarowin, 2010).
2.4 Hypotheses development
To the best of our knowledge, no prior research has studied the relationship between
the alternative methods of earnings management, accrual and real account
manipulation, and litigation risk. This study examines this question. We assume
that the likelihood of litigation varies with the method(s) used to reach earnings goals
for ?rms issuing equity. In particular, ?rms can manage earnings by managing either
ARJ
24,1
54
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
accruals or real accounts. Income can be manipulated upwards through real accounts
by cutting real expenditures, reducing cost of goods sold by increasing production,
or increasing sales through offering lenient terms and discounts to customers, which
would reduce operating cash ?ows. We recognize that lawsuits occur over equity
offerings on cases with positive expected values for plaintiffs or plaintiff attorneys
(Priest and Klein, 1984; Alexander, 1991). That is, lawsuits are ?led when the likelihood
of success times the expected recovery exceeds the expected litigation costs (Palmrose,
1991). We conjecture that the assessment of the likelihood of success will change with
the method of manipulation. Speci?cally, we argue that real account manipulation will
be harder to detect by outsiders, since it cannot be distinguished from normal changes
in operations, and it will lead to lower incidence of litigation than accrual manipulation.
We test the following hypothesis:
H1. Defendant SEO?rms engage in a higher magnitude of earnings overstatement
through accrual manipulation in the year prior to the offering, but not real
account manipulation, compared to non-defendant offering ?rms.
Whereas the SarbOx act has reduced accrual manipulation, the evidence suggests that
?rms have shifted to real manipulation (Cohen et al., 2008). Given this evidence, we
conjecture that investors would be more wary of real account manipulation following
SarbOx and may initiate litigation not only based on suspected accrual manipulation,
but also real account manipulation. We therefore test the following hypothesis:
H2. In the post-SarbOx period, defendant SEO ?rms engage in a higher
magnitude of earnings overstatement through real account manipulation in
the year prior to the offering, compared to non-defendant offering ?rms.
Whereas we believe that litigation decisions will be based more on accrual manipulation
in the years prior to SarbOx, we hypothesize that investors will place more scrutiny
on both types of earnings management in more recent years. Therefore, defendant
?rms will show evidence of higher accrual and real manipulation, as compared to
non-defendant ?rms.
3. Empirical methodology
In this section, we explain the sample selection procedures (Section 3.1) and introduce
the measures of discretionary accruals (Section 3.2) and abnormal real accounts
(Section 3.3) followed by the lawsuit data collection methodology (Section 3.4).
3.1 Sample selection
We identify ?rms that underwent an SEO from Thomson Financial’s SDC Platinume
database in the years 1991-2005. Financial information of these SEO?rms in addition to
all other ?rms in the same industries is collected from Standard and Poor’s Compustat
database. We use the year prior to the offering to test for earnings management behavior.
This is typical of earnings management studies, as the incentives to manage earnings are
strongest in the year prior to the offering. Therefore, our SEO sample covers the years
1990-2004. We eliminate ?rms in the utilities and ?nance industries (SIC between 4,000
and 4,999 and between 6,000 and 6,999) and industry-year combinations that have less
than tenobservations since the empirical analysis is based oncross-sectional analysis by
industryandbyyear.[3] We eliminate extreme observations, identi?ed as those inthe top
Real and accrual-
based earnings
55
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and bottom percentile of the total assets, total accruals, production costs, discretionary
expenses, and cash from operations distributions. The ?nal sample consists of all
?rm/year observations with necessary data in industries that include a ?rm that
underwent a SEO in the sample period. This results in a total of 31,207 ?rm-year
observations, which includes 1,871 SEO ?rm/year observations and 29,336 non-SEO
?rm/year observations. This sample is used to calculate discretionary/abnormal accrual
and real accounts, as explained in Sections 3.2 and 3.3. For the empirical analysis, we
adopt a performance-matching approach as suggested by Kothari et al. (2005). More
speci?cally, each SEOobservation is matched with an observation belonging to a ?rmin
the same industry (four-digit SICcode) and year, with the closest level of returnonassets.
If a match is not found, we select fromobservations in the three-digit SIC code. If a match
is still not found, we select from observations in the two-digit SIC code. The empirical
analysis is based on this sample of 1,871 SEO and 1,871 non-SEO ?rm/year
observations.[4] This matching alleviates the problems associated with misspeci?cation
of manipulation measures in observations with extreme performance (Kothari et al.,
2005; Armstrong et al., 2009). Table I presents the yearly and industrial distribution of
the 1,871 SEO ?rms[5].
Panel A of Table I presents the temporal distribution of the sample. The SEO
observations are evenly distributed among the sample years, 1990-2004. The number
of SEO observations each year ranges from 93 (in year 2004) to 165 (in year 2003).
Panel B presents the industrial distribution of the sample. SEOs are most prevalent
in the machinery and equipment industry followed by the wholesale and retail,
business services, and natural resources industry (number of SEOs ¼ 625, 274, 228,
and 198, in machinery and equipment, wholesale and retail, business services and
natural resources industries, respectively). The remaining SEOs (29 percent of the
distribution) are distributed among the remaining nine industries.
3.2 Accrual earnings management: discretionary accruals measures
We estimate accrual manipulation following the modi?ed Jones model ( Jones, 1991;
Dechow et al., 1995) with the proposed Kothari et al. (2005) modi?cations through the
following cross-sectional regression[6]:
TAC
t
A
t21
¼ b
0
þb
1
1
A
t21
þb
2
DREV
t
2DAR
t
A
t21
þb
3
PPE
t
A
t21
þb
4
ðROA
t
Þ
þ DAC
t
whereTAC
t
istotal accrualsfromthecash?owstatement (Compustat No. 123-(Nos308-124))
scaled by beginning total assets (Compustat No. 6) and the independent variables include
change in revenue (Compustat No. 12) less change in accounts receivable (Compustat No. 2)
(DREV
t
2DAR
t
), the level of gross property, plant, and equipment (Compustat No. 7)
(PPE
t
), and return on assets in year t (ROA
t
), which is calculated as net income in year t
(Compustat No. 172) divided by total assets in year t (Compustat No. 6). All variables, other
thanROA
t
, are scaledbybeginningtotal assets toadjust for heteroskedasticity. The residual
from this regression is the estimate of discretionary accruals, DAC
t
. The coef?cients are
estimated by running cross-sectional regressions by four-digit SIC codes and by year. This
methodology eliminates the constraints of ?xing the coef?cients in these regressions over
time or over all industries (Kothari et al., 2005). As previouslymentioned, the year prior to the
offering is used as the test year.
ARJ
24,1
56
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
We also estimate discretionary accounts receivable, following Ibrahim (2009), since
this is the accrual that is most signi?cantly associated with litigation, as follows:
DAR
t
A
t21
¼ b
01
þb
11
1
A
t21
þb
21
DREV
t
A
t21
þb
31
ðROA
t
Þ þ DAR
t
where DAR
t
is the change in accounts receivable (Compustat No. 2) and the residual from
the regression, DAR
t
, is the discretionary accounts receivable accrual. DAR
t
represents
the unexpected portion of accounts receivable, which may occur in conjunction with
revenue manipulation, e.g. through early booking of revenue or arti?cial sales
transactions. This methodology for calculating this component of accruals was shown
No. of ?rms %
Panel A: distribution of SEO observations by year
Year
1990 108 5.77
1991 99 5.29
1992 128 6.84
1993 100 5.34
1994 142 7.59
1995 157 8.39
1996 141 7.54
1997 120 6.41
1998 145 7.75
1999 130 6.95
2000 103 5.51
2001 110 5.88
2002 130 6.95
2003 165 8.82
2004 93 4.97
Total 1,871 100.00
Panel B: distribution of SEO observations by industry
Industry
(1) Natural resources 198 10.58
(2) Construction and metal 82 4.38
(3) Food and tobacco 13 0.69
(4) Consumer goods 47 2.51
(5) Paper and printing 28 1.50
(6) Chemical and petroleum 166 8.87
(7) Machinery and equipment 625 33.40
(8) Transportation related 46 2.46
(9) Wholesale and retail 274 14.64
(10) Entertainment 36 1.92
(11) Business services 228 12.19
(12) Health and other services 118 6.31
(13) Non-classi?able 10 0.53
Total 1,871 100.00
Notes: Panel A presents the distribution of SEOs across the sample years; panel B provides industry
compositions of the SEOsample; the industries (in the order presented in panel B) are classi?ed based on
two-digit SICcodes as follows: (1) 0-9,10-14; (2) 15-19, 30, 32-34; (3) 20-21; (4) 22-23, 25, 31, 39; (5) 24, 26-27;
(6) 28-29; (7) 35-36, 38; (8) 37, 40-47; (9) 50-59; (10) 78-79; (11) 73, 81; (12) 70, 72, 75-76, 80, 82-89; (13) 99
Table I.
Temporal and industry
distribution of SEO
observations
Real and accrual-
based earnings
57
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
to be more powerful than the Marquardt and Wiedman (2004) methodology in detecting
earnings management[7].
3.3 Real earnings manipulation measures
We rely on prior studies that developed proxies for real earnings management.
Cohen et al. (2008) base their measures of real earnings management on Roychowdhury
(2006) and others. Speci?cally, they focus on three manipulation methods that would
increase bottom-line earnings:
(1) acceleration of the timing of sales through increased price discounts or more
lenient credit terms, which would abnormally decrease cash from operations;
(2) reporting lower cost of goods sold through increased production; and
(3) decreases in discretionary expenses including advertising expense, research
and development expenses, and selling, general, and administrative (SG&A)
expenses.
The discretionary or abnormal levels of the above real accounts are measured
as follows:
CFO
t
A
t21
¼ a
11
1
A
t21
þa
12
REV
t
A
t21
þa
13
DREV
t
A
t21
þ ACFO
t
PROD
t
A
t21
¼ a
21
1
A
t
þa
22
REV
t
A
t21
þa
23
DREV
t
A
t21
Þ þa
24
DREV
t21
A
t21
þ APROD
t
DEXP
A
t21
¼ a
31
1
A
t21
þa
32
REV
t21
A
t21
þ ADEXP
t
where CFO
t
is cash from operations in period t (Compustat Nos 308-124), PROD
t
is
production costs in period t, de?ned as the sum of cost of goods sold (Compustat No. 41)
and the change in inventory in period t (Compustat No. 3), and DEXP
t
is discretionary
expenses, de?ned as the sum of advertising expenses (Compustat No. 45), R&D
expenses (Compustat No. 46) and SG&A (Compustat No. 189)[8]. The abnormal cash
from operations (ACFO
t
), abnormal production costs (APROD
t
), and abnormal
discretionary expenses (ADEXP
t
) are the residuals from the above regressions. The
regressions are run separately for each four-digit SIC code and for each year. Anegative
ACFO
t
and ADEXP
t
as well as a positive APROD
t
would indicate income-increasing
earnings management[9]. All variables are de?ated by the beginning total assets to
adjust for heteroskedasticity.
As in Cohen et al. (2008), we also provide a measure of total manipulation through
real accounts as follows:
AREAL
t
¼ ACFO
t
þ APROD
t
þ ADEXP
t
Since the three separate real account manipulation measures have different effects on
bottom-line earnings, we do not make predictions about the sign of the measure,
AREAL
t
, but only consider its magnitude.
ARJ
24,1
58
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
All above variables in addition to the ?nancial variables collected are de?ned in
Appendix 1. The Pearson correlation coef?cients between the ?nancial and
manipulation variables appear in Table II.
All coef?cients that are above 0.8 are shown in italic. Most notably, the performance
measures are highly correlated. INC
t
is highly correlated with CFO
t
( p ¼ 0.812), while
CFO
t
is highly correlated with abnormal cash ?ow from operations, ACFO
t
( p ¼ 0.876).
Net revenues, REV
t
, are highly correlated with production costs, PROD
t
( p ¼ 0.926). The
discretionary accruals measures are not highly correlated with the real account
manipulation measures. The highest correlation is between ACFO
t
and ADEXP
t
( p ¼ 20.499). We do not believe that these correlations will impact the empirical analysis.
3.4 Lawsuit data selection
Information regarding the lawsuits, such as class periods and the nature of the
allegations made therein was taken from the LEXIS/NEXIS Academic Universe
Business News, searching on company name and the keywords “Class Action” and
“Litigation” in the three years following the offering year. Data were also collected
from the Stanford Law Securities Class Action Clearinghouse which hosts a list of
?rms that faced class action litigation. The search identi?ed 218 SEOs that were sued
subsequent to their offering. Out of these observations, 85 involved litigation due to
underwriter problems of disclosure or other non-accounting irregularities and are thus
excluded from the analysis. The remaining 133 SEOs faced accounting allegations in
their lawsuits. There were 1,653 SEOs that were not subsequently subject to litigation.
The empirical analysis is based on the 133 sued SEOs and the 1,653 non-sued SEOs.
4. Tests and empirical results
We ?rst begin by testing the accrual and real account manipulation in the sample of
SEOs (Section 4.1) followed by the hypotheses testing (Section 4.2).
4.1 Accrual and real earnings management around SEOs
First, we replicate prior studies to show that our sample of SEOs engage in
income-increasing accrual and real account manipulation prior to the offering. Panel A
of Table III reports descriptive statistics and differences between the SEOs (n ¼ 1,871)
and the matched non-SEO observations.
The SEO observations are larger and less pro?table than the matched sample
(difference in median A
t
is $55m and difference in mean INC
t
is 20.015, signi?cant at
the 1 and 5 percent levels, respectively). SEO observations also have higher revenues
(difference in mean REV
t
is 0.072, and difference in median is 0.058, both signi?cant at
the 5 percent level). The results of differences in discretionary accruals and real
accounts appear in the second part of the panel. Discretionary accruals are higher in
the SEO observations (difference in mean DAC
t
is 0.003, which is not signi?cant, and
difference in median DAC
t
is 0.005, signi?cant at the 10 percent level). Discretionary
accounts receivable are also higher in the SEO sample (difference in mean DAR
t
is
0.014, and difference in median is 0.005, both signi?cant at the 1 percent level). Both are
consistent with income-increasing manipulation prior to the offering. Abnormal cash
from operations is lower in SEO observations (difference in mean ACFO
t
is 20.022 and
difference in median is 20.008, signi?cant at the 1 and 5 percent levels, respectively),
which is consistent with income-increasing real account manipulation. However,
Real and accrual-
based earnings
59
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
R
E
V
t
C
F
O
t
P
R
O
D
t
D
E
X
P
t
T
A
C
t
R
O
A
t
A
t
D
A
C
t
D
A
R
t
A
C
F
O
t
A
P
R
O
D
t
A
D
E
X
P
t
A
R
E
A
L
t
I
N
C
t
0
.
2
8
7
0
.
8
1
2
0
.
2
0
9
2
0
.
4
4
0
0
.
5
5
3
0
.
7
6
8
0
.
0
6
5
0
.
2
3
3
2
0
.
1
0
3
0
.
6
8
0
2
0
.
1
0
0
2
0
.
5
6
4
2
0
.
4
2
9
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
5
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
R
E
V
t
1
.
0
0
0
0
.
1
9
3
0
.
9
2
6
0
.
2
9
7
0
.
2
1
5
0
.
2
6
4
2
0
.
0
5
3
0
.
0
3
5
0
.
0
7
2
0
.
0
8
6
0
.
1
0
2
2
0
.
0
5
7
0
.
0
6
0
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
2
1
)
(
0
.
1
3
1
)
(
0
.
0
0
2
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
1
4
)
(
0
.
0
0
9
)
C
F
O
t
1
.
0
0
0
0
.
1
0
5
2
0
.
4
3
6
2
0
.
0
3
9
0
.
6
6
0
0
.
1
1
1
2
0
.
2
5
3
2
0
.
1
5
9
0
.
8
7
6
2
0
.
1
8
1
2
0
.
5
1
4
2
0
.
2
7
6
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
9
6
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
P
R
O
D
t
1
.
0
0
0
0
.
0
5
6
0
.
2
0
8
0
.
1
9
5
2
0
.
0
1
0
0
.
0
4
4
0
.
0
4
1
0
.
0
0
9
0
.
3
5
1
2
0
.
2
0
6
2
0
.
0
1
7
(
0
.
0
1
5
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
6
6
8
)
(
0
.
0
5
9
)
(
0
.
0
7
3
)
(
0
.
7
0
5
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
4
6
6
)
D
E
X
P
t
1
.
0
0
0
2
0
.
1
3
1
2
0
.
3
0
3
2
0
.
1
7
7
2
0
.
0
4
8
0
.
1
5
3
2
0
.
3
9
6
2
0
.
3
0
7
0
.
7
8
1
0
.
6
3
2
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
3
8
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
T
A
C
t
1
.
0
0
0
0
.
3
7
1
2
0
.
0
4
8
0
.
7
6
0
0
.
0
5
0
2
0
.
0
8
6
0
.
0
8
6
2
0
.
2
3
1
2
0
.
3
4
0
(
0
.
0
0
1
)
(
0
.
0
3
8
)
(
0
.
0
3
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
R
O
A
t
1
.
0
0
0
0
.
0
5
9
0
.
0
0
4
2
0
.
0
2
6
0
.
5
4
2
2
0
.
0
7
8
2
0
.
3
8
8
2
0
.
2
4
9
(
0
.
0
1
1
)
(
0
.
8
7
6
)
(
0
.
2
6
2
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
A
t
1
.
0
0
0
2
0
.
0
4
9
2
0
.
0
2
4
0
.
1
0
6
0
.
0
0
6
2
0
.
1
1
7
2
0
.
0
9
3
(
0
.
0
3
5
)
(
0
.
2
9
7
)
(
0
.
0
0
1
)
(
0
.
7
9
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
D
A
C
t
1
.
0
0
0
0
.
0
7
4
2
0
.
2
9
8
0
.
1
2
0
2
0
.
1
1
4
2
0
.
2
9
2
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
D
A
R
t
1
.
0
0
0
2
0
.
1
8
3
2
0
.
0
3
0
0
.
1
5
6
0
.
0
7
4
(
0
.
0
0
1
)
(
0
.
1
8
9
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
A
C
F
O
t
1
.
0
0
0
2
0
.
2
5
6
2
0
.
4
9
9
2
0
.
2
2
4
(
0
.
0
0
1
)
(
0
.
0
0
1
)
(
0
.
0
0
1
)
A
P
R
O
D
t
1
.
0
0
0
2
0
.
4
2
8
2
0
.
0
1
7
(
0
.
0
0
1
)
(
0
.
4
5
3
)
A
D
E
X
P
t
1
.
0
0
0
0
.
7
8
6
(
0
.
0
0
1
)
N
o
t
e
s
:
n
¼
1
,
8
7
1
;
a
l
l
c
o
r
r
e
l
a
t
i
o
n
c
o
e
f
?
c
i
e
n
t
s
a
b
o
v
e
0
.
8
a
r
e
s
h
o
w
n
i
n
i
t
a
l
i
c
;
a
l
l
v
a
r
i
a
b
l
e
s
a
r
e
d
e
?
n
e
d
i
n
A
p
p
e
n
d
i
x
1
Table II.
Pearson correlation
coef?cients ( p-values)
of ?nancial variables,
discretionary accruals,
and abnormal real
accounts in SEO sample
ARJ
24,1
60
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
SEO observations
(n ¼ 1,871)
Matched observations
(n ¼ 1,871) Difference
Variable Mean Median Mean Median Mean Median
Panel A: mean and median differences in ?nancial variables, discretionary accruals, and abnormal real
accounts between SEO observations and matched non-SEO observations in all years
Financial variables
INC
t
20.020 0.048 20.005 0.041 20.015
* *
0.007
* *
REV
t
1.444 1.248 1.372 1.190 0.072
* *
0.058
* *
DREV
t
0.303 0.204 0.167 0.099 0.136
* * *
0.104
* * *
CFO
t
0.045 0.082 0.057 0.081 20.012
* *
0.001
PROD
t
0.949 0.727 0.901 0.696 0.048
* *
0.031
DEXP
t
0.524 0.388 0.462 0.364 0.063
* * *
0.024
* *
TAC
t
20.065 20.058 20.062 20.056 20.004 20.002
ROA
t
20.030 0.036 20.022 0.037 20.008 20.001
A
t
931.821 180.490 956.504 125.410 224.682 55.080
* * *
Discretionary accruals and real accounts
DAC
t
0.003 0.005 0.000 0.000 0.003 0.005
*
DAR
t
0.013 0.002 20.001 20.003 0.014
* * *
0.005
* * *
ACFO
t
20.011 0.011 0.011 0.019 20.022
* * *
20.008
* *
APROD
t
20.025 20.022 20.014 20.014 20.011
* *
20.008
*
ADEXP
t
0.136 0.034 0.036 0.007 0.101
* * *
0.027
* * *
AREAL
t
0.100 0.041 0.033 0.008 0.067
* * *
0.034
* * *
Panel B: mean and median differences in discretionary accruals and abnormal real accounts between
SEO observations and matched non-SEO observations in the pre- and post-SarbOx periods
Pre-SarbOx
DAC
t
0.004 0.006 20.001 20.001 0.006
*
0.006
* *
DAR
t
0.014 0.002 20.001 20.004 0.016
* * *
0.005
* * *
ACFO
t
20.010 0.014 0.014 0.021 20.024
* * *
20.006
* *
APROD
t
20.024 20.022 20.011 20.012 20.013
*
20.009
ADEXP
t
0.140 0.036 0.035 0.010 0.105
* * *
0.026
* * *
AREAL
t
0.007 0.001 20.001 0.000 0.008
* * *
0.001
* * *
Post-SarbOx
DAC
t
20.001 20.001 0.005 0.004 20.006 20.005
DAR
t
0.006 0.002 0.000 20.002 0.006
*
0.004
* * *
ACFO
t
20.017 0.006 0.000 0.011 20.017
*
20.006
APROD
t
20.026 20.029 20.022 20.019 20.005 20.011
ADEXP
t
0.124 0.030 0.040 20.004 0.084
* * *
0.034
* * *
AREAL
t
20.013 0.001 20.003 0.000 20.010
* * *
0.001
* * *
Difference between pre- and post-SarbOx
DAC
t
20.005 20.006
* *
0.006 0.005
DAR
t
20.008
*
0.000 0.002 0.002
ACFO
t
20.007 20.009
* *
20.013
*
20.010
*
APROD
t
20.002 20.008 20.010 20.006
ADEXP
t
20.016 20.006 0.005 20.014
AREAL
t
20.020
*
0.000
* * *
20.002
*
0.000
Notes: Signi?cance at:
*
10,
* *
5, and
* * *
1 per cent levels (one tailed), respectively, using t-test and
Wilcoxon two-sample test; the matched observations are chosen from the same four-digit SIC code as
the SEO observation in the same year as the SEO, with the closest ROA; if no match is found, then the
matching is based on the three-digit SIC code; if still no match is found, then the matching is based on
the two-digit SIC code; all variables are de?ned in Appendix 1; n refers to the number of observations;
pre-SarbOx refers to years 1990-2001 and post-SarbOx refers to years 2002-2004
Table III.
Univariate tests of
differences in ?nancial
variables, discretionary
accruals, and real
accounts in SEO and
matched observations
Real and accrual-
based earnings
61
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
abnormal production costs are lower for SEO observations (difference in mean
APROD
t
is 20.011 and difference in median is 20.008, signi?cant at the 5 and
10 percent levels, respectively), which is not consistent with income-increasing
manipulation. Finally, abnormal discretionary expenses are higher in SEO
observations (difference in mean ADEXP
t
is 0.101 and difference in median is 0.027,
both signi?cant at the 1 percent level), which is not consistent with income-increasing
manipulation. The total real manipulation measure, AREAL
t
, is signi?cantly larger in
SEO observations (difference in mean is 0.067 and difference in median is 0.034, both
signi?cant at the 1 percent level). The magnitude of the abnormal real accounts in the
non-SEO sample is close to the numbers reported in Cohen et al. (2008, Table I, p. 767).
Panel B presents the univariate differences in the pre- and post-SarbOx periods.
There are 1,483 SEO observations in the pre-SarbOx period and 388 SEO observations
in the post-SarbOx period. In the pre-SarbOx period, the results are consistent with the
full sample. DAC
t
and DAR
t
are both higher in the SEO observations (difference in
mean DAC
t
is 0.006 and difference in median is 0.006, signi?cant at the 10 and 5 percent
levels, respectively; difference in mean DAR
t
is 0.016 and difference in median is 0.005,
both signi?cant at the 1 percent level). ACFO
t
is lower in the SEO observations
(difference in mean is 20.024 and difference in median is 20.006, signi?cant at the
1 and 5 percent levels, respectively), consistent with income-increasing manipulation.
However, APROD
t
and ADEXP
t
do not exhibit income-increasing behavior (difference
in mean APROD
t
is 20.013 and difference in mean ADEXP
t
is 0.105, signi?cant at the
10 and 1 percent levels, respectively). As in the full sample, the total real manipulation
measure, AREAL
t
, is signi?cantly larger in SEO observations (mean difference is 0.008
and median difference is 0.001, both signi?cant at the 1 percent level).
In the post-SarbOx period, DAC
t
is not signi?cantly different between the SEO and
the non-SEO observations. DAR
t
however is still larger in SEO observations (difference
in mean is 0.006 and difference in median is 0.004, signi?cant at the 10 and 1 percent
levels, respectively). There is some evidence of income-increasing manipulation
through real accounts. ACFO
t
is slightly lower in the SEO ?rms (difference in mean
is 20.017, signi?cant at the 10 percent level). However, the remaining real accounts do
not indicate income-increasing manipulation. ADEXP
t
is higher (difference in mean is
0.084 and difference in median is 0.034, both signi?cant at the 1 percent level), and
APROD
t
is lower in SEOs but the difference is not signi?cant. The total manipulation
measure, AREAL
t
, is signi?cantly different between SEO and non-SEO observations
(difference in mean is 20.010 and difference in median is 0.001, both signi?cant at the
1 percent level); however, it is not clear whether it is an income-increasing or
decreasing manipulation.
The third section of Panel B presents the mean and median differences in
discretionary/abnormal accrual and real accounts between the pre- and post-SarbOx
periods. We ?nd signi?cant differences between these two periods for the SEO ?rms.
Speci?cally, discretionary accruals are lower in the post-SarbOx period (difference in
median is 20.006, signi?cant at the 5 percent level), indicating that SEOs engage in
less accrual manipulation following the passage of SarbOx; discretionary accounts
receivables are also lower (difference in mean is 20.008, signi?cant at the 10 percent
level). There is some evidence of shifting to real manipulation as abnormal cash from
operations is lower (difference in median is 20.009, signi?cant at the 5 percent level).
The real manipulation measure, AREAL
t
, is smaller in the post-SarbOx period
ARJ
24,1
62
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(difference in median is 20.020, signi?cant at the 10 percent level). For the non-SEO
observations, the only signi?cant differences between both periods is in the abnormal
cash from operations (mean difference is 20.013 and median difference is 20.010, both
signi?cant at the 10 percent level) indicating income-increasing real account
manipulation. These results corroborate prior evidence that shows that SEO ?rms
engage in less earnings management through accruals in the post-SarbOx period but
still show some evidence of real account manipulation.
To further test the differences between SEO and non-SEO observations, we conduct
a multivariate analysis including other variables that might cause the differences in the
discretionary accrual and abnormal real account measures observed above.
Speci?cally, we test the following regression:
DEPENDENT
it
¼ a
1i
þb
1i
SEO
t
þb
2i
SIZE
t
þb
3i
LEV
t
þb
4i
ROA
t
þb
5i
MTB
t
þb
6i
LOSS
t
þb
7i
LITRISK
t
þ
X
j
b
ji
Industry Dummies
t
ð1Þ
where:
DEPENDENT
it
¼ the discretionary accrual or abnormal real account
measure and i ¼ 1 2 6 for DAC
t
, DAR
t
, ACFO
t
,
APROD
t
, ADEXP, and AREAL
t
as the dependent variable;
SEO
t
¼ indicator variable that takes on the value 1 for an SEO
observation in the year prior to the offering and 0 for
matched non-SEO observations;
SIZE
t
¼ Log (total assets at year end);
LEV
t
¼ leverage ¼ total of short- and long-term debt scaled by
average total assets;
ROA
t
¼ return on assets measured as net income divided by total
assets;
MTB
t
¼ market-to-book value measured as market value of assets
divided by book value of assets at year end;
LOSS
t
¼ indicator variable that takes on the value 1 if the ?rm has
a net loss during the year, and 0 otherwise;
LITRISK
t
¼ indicator variable that takes on the value 1 if the
?rm operates in a high-risk environment, as de?ned in
Francis et al. (1994), and 0 otherwise[10]; and
Industry Dummies
t
¼ indicator variable that takes on the value of 1 if observation
belongs to a speci?c industry, and 0 otherwise.
We control for several factors that are associated with incentives to manage earnings
as well as the discretionary accruals measure (Frankel et al., 2002). First, we control
for the size of the ?rm since larger ?rms tend to have larger accruals and larger
discretionary accruals. Larger ?rms will also tend to have higher expenses and cash
from operations. We control for leverage, since ?rms with higher leverage have more
Real and accrual-
based earnings
63
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
incentives to manage earnings to avoid covenant violations. We control for
performance (ROA) since ?rms with better performance tend to have higher
accruals and cash from operations. We control for MTB since ?rms with growth
prospects (with higher market-to-book values) have higher incentives to manage
earnings. We control for loss since loss ?rms have different incentives to manage
earnings. We control for litigation risk since ?rms that more likely to face litigation
have higher incentives to manage earnings to avoid it. We also control for the industry
membership of the observation. We use the same designations in Table I based on the
four-digit SIC code to construct the industry dummies.
The coef?cients b
1i
are used to test for earnings management behavior with
i ¼ 1 2 6 in the regressions with DAC
t
, DAR
t
, ACFO
t
, APROD
t
, ADEXP
t
, and
AREAL
t
as the dependent variables. The results of the above regressions appear in
Table IV. All t-statistics in the regressions are calculated from White’s
heteroskedasticity-corrected standard errors.
The results in panel Aare consistent with the univariate results. DAC
t
and DAR
t
are
signi?cantly higher in SEO observations, controlling for other factors typically related
to discretionary accruals (coef?cient ¼ 0.007 and 0.011 for SEO indicator variable in
the regression with DAC
t
and DAR
t
as the dependent variable, respectively, signi?cant
at the 5 and 1 percent levels, respectively). The magnitude of the differences between
SEO and non-SEO observations is quite signi?cant, which warrants further discussion.
Given that the mean (median) total assets for SEO observations is $932m ($181m)
(Table III), the above coef?cients show that DAC
t
is higher in SEO observations by an
average of $6.5m (median of $1.3m) and that DAR
t
is higher by an average of $10m
(median of $2m). Furthermore, these differences constitute an average of 35 per cent of
net income for DAC
t
and 55 per cent for DAR
t
. Economically speaking, these are quite
signi?cant amounts that could affect the decision of users of the ?nancial statements.
As for the abnormal real accounts, only ACFO
t
is consistent with income-increasing
manipulation (coef?cient of SEO ¼ 20.021 in the regression with ACFO
t
as the
dependent variable, signi?cant at the 1 percent level). This means that SEO
observations have lower ACFO
t
by an average of $20m (median of $4m). But there is
evidence of real account manipulation (coef?cient of SEO ¼ 0.065 with AREAL
t
as the
dependent variable, signi?cant at the 1 percent level). The control variables are
generally signi?cant in these regressions. The adjusted R
2
of these regressions range
from 2.51 percent (with APROD
t
as the dependent variable) to 26.81 percent (with
ACFO
t
as the dependent variable). In panel B, to test the differences between the pre-
and post-SarbOx periods in a multivariate setting, we run the following regression:
DEPENDENT
it
¼ a
1i
þb
1i
SEO
t
þb
2i
SarbOx
t
þb
3i
SEO
*
t
SarbOx
t
þb
4i
SIZE
t
þb
5i
LEV
t
þb
6i
ROA
t
þb
7i
MTB
t
þb
8i
LOSS
t
þb
9i
LITRISK
t
þ
X
j
b
ji
Industry Dummies
ð2Þ
where SarbOx
t
– an indicator variable that takes on the value 1 if the observation is in
the years 2002-2004, and 0 otherwise. All remaining variables are as de?ned above.
The main test variable is the interaction variable SEO
*
t
SarbOx
t
, The coef?cient b
3i
measures the amount by which the change in the value of the dependent variable for
SEOs is affected by SarbOx. Hence, we examine the magnitude and the sign of the
coef?cients b
3i
in the empirical tests.
ARJ
24,1
64
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
D
e
p
.
v
a
r
i
a
b
l
e
D
A
C
D
A
R
A
C
F
O
A
P
R
O
D
A
D
E
X
P
A
R
E
A
L
P
a
n
e
l
A
:
c
o
e
f
?
c
i
e
n
t
s
(
t
-
s
t
a
t
i
s
t
i
c
s
)
f
r
o
m
r
e
g
r
e
s
s
i
o
n
s
o
f
t
h
e
f
o
r
m
:
n
¼
3
,
7
4
2
a
S
E
O
t
0
.
0
0
7
(
1
.
9
6
)
*
*
0
.
0
1
1
(
4
.
7
7
)
*
*
*
2
0
.
0
2
1
(
2
4
.
3
0
)
*
*
*
2
0
.
0
1
4
(
2
2
.
1
8
)
*
*
0
.
1
0
0
(
9
.
0
5
)
*
*
*
0
.
0
6
5
(
8
.
1
8
)
*
*
*
S
I
Z
E
t
2
0
.
0
0
5
(
2
5
.
1
0
)
*
*
*
2
0
.
0
0
1
(
2
2
.
0
3
)
*
*
0
.
0
0
4
(
2
.
5
5
)
*
*
*
0
.
0
0
3
(
1
.
6
3
)
*
2
0
.
0
0
5
(
2
1
.
6
3
)
*
0
.
0
0
2
(
0
.
8
1
)
L
E
V
t
2
0
.
0
2
9
(
2
2
.
3
4
)
*
*
*
0
.
0
4
0
(
3
.
0
0
)
*
*
*
0
.
0
0
4
(
0
.
4
8
)
0
.
0
3
6
(
2
.
7
4
)
*
*
*
2
0
.
0
5
1
(
2
2
.
4
0
)
*
*
*
2
0
.
0
1
0
(
2
0
.
7
4
)
R
O
A
t
0
.
0
1
3
(
0
.
7
1
)
0
.
0
0
2
(
0
.
4
5
)
0
.
2
5
1
(
5
.
8
3
)
*
*
*
2
0
.
0
3
9
(
2
2
.
6
0
)
*
*
*
2
0
.
3
4
0
(
2
5
.
4
3
)
*
*
*
2
0
.
1
2
8
(
2
4
.
0
3
)
*
*
*
L
O
S
S
t
2
0
.
0
2
3
(
2
3
.
9
3
)
*
*
*
2
0
.
0
0
1
(
2
0
.
4
7
)
2
0
.
0
5
5
(
2
4
.
7
2
)
*
*
*
0
.
0
2
9
(
3
.
5
3
)
*
*
*
0
.
0
5
5
(
2
.
8
2
)
*
*
*
0
.
0
2
9
(
2
.
3
5
)
*
*
*
L
I
T
R
I
S
K
t
0
.
0
0
0
(
2
0
.
0
5
)
0
.
0
0
4
(
1
.
1
2
)
0
.
0
0
9
(
1
.
4
2
)
*
2
0
.
0
0
5
(
2
0
.
6
1
)
0
.
0
2
4
(
1
.
6
1
)
*
0
.
0
2
7
(
2
.
6
5
)
*
*
*
M
T
B
t
0
.
0
0
0
(
1
.
2
3
)
0
.
0
0
0
(
1
.
1
6
)
2
0
.
0
0
1
(
2
1
.
7
1
)
*
*
2
0
.
0
0
1
(
2
1
.
9
7
)
*
*
0
.
0
0
2
(
1
.
5
4
)
*
0
.
0
0
1
(
1
.
2
2
)
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
A
d
j
u
s
t
e
d
R
2
(
%
)
2
.
6
0
4
.
5
9
2
6
.
8
1
2
.
5
1
1
5
.
6
1
9
.
3
4
P
a
n
e
l
B
:
c
o
e
f
?
c
i
e
n
t
s
(
t
-
s
t
a
t
i
s
t
i
c
s
)
f
r
o
m
r
e
g
r
e
s
s
i
o
n
s
o
f
t
h
e
f
o
r
m
:
n
¼
3
,
7
4
2
b
S
E
O
t
0
.
0
0
9
(
2
.
2
0
)
*
*
0
.
0
1
3
(
4
.
8
7
)
*
*
*
2
0
.
0
2
2
(
2
3
.
8
3
)
*
*
*
2
0
.
0
1
6
(
2
2
.
0
9
)
*
*
*
0
.
1
0
3
(
8
.
1
2
)
*
*
*
0
.
0
6
5
(
7
.
2
6
)
*
*
*
S
a
r
b
O
x
t
0
.
0
1
1
(
1
.
9
4
)
*
0
.
0
0
3
(
0
.
8
5
)
2
0
.
0
0
3
(
2
0
.
4
5
)
2
0
.
0
1
4
(
2
1
.
3
8
)
2
0
.
0
1
3
(
2
0
.
8
3
)
2
0
.
0
3
0
(
2
2
.
7
1
)
*
*
*
S
E
O
*t
S
a
r
b
O
x
t
2
0
.
0
1
0
(
2
1
.
3
0
)
2
0
.
0
0
9
(
2
1
.
7
8
)
*
0
.
0
0
4
(
0
.
3
9
)
0
.
0
0
7
(
0
.
4
9
)
2
0
.
0
1
7
(
2
0
.
7
2
)
2
0
.
0
0
6
(
2
0
.
3
5
)
S
I
Z
E
t
2
0
.
0
0
5
(
2
5
.
1
1
)
*
*
*
2
0
.
0
0
1
(
2
1
.
8
5
)
*
0
.
0
0
4
(
2
.
5
3
)
*
*
*
0
.
0
0
4
(
1
.
7
6
)
*
2
0
.
0
0
4
(
2
1
.
3
3
)
0
.
0
0
3
(
1
.
3
4
)
L
E
V
t
2
0
.
0
2
9
(
2
2
.
3
1
)
*
*
0
.
0
4
0
(
3
.
0
0
)
*
*
*
0
.
0
0
4
(
0
.
4
7
)
0
.
0
3
6
(
2
.
7
1
)
*
*
*
2
0
.
0
5
2
(
2
2
.
4
4
)
*
*
*
2
0
.
0
1
2
(
2
0
.
8
6
)
R
O
A
t
0
.
0
1
3
(
0
.
7
3
)
0
.
0
0
2
(
0
.
4
2
)
0
.
2
5
1
(
5
.
8
1
)
*
*
*
2
0
.
0
3
9
(
2
2
.
6
3
)
*
*
*
2
0
.
3
4
1
(
2
5
.
4
3
)
*
*
*
2
0
.
1
3
0
(
2
4
.
0
6
)
*
*
*
L
O
S
S
t
2
0
.
0
2
4
(
2
3
.
9
9
)
*
*
*
2
0
.
0
0
1
(
2
0
.
4
3
)
2
0
.
0
5
5
(
2
4
.
7
4
)
*
*
*
0
.
0
3
0
(
3
.
6
2
)
*
*
*
0
.
0
5
6
(
2
.
9
2
)
*
*
*
0
.
0
3
2
(
2
.
5
5
)
*
*
*
L
I
T
R
I
S
K
t
0
.
0
0
0
(
0
.
0
8
)
0
.
0
0
4
(
1
.
1
3
)
0
.
0
0
9
(
1
.
4
2
)
2
0
.
0
0
5
(
2
0
.
5
8
)
0
.
0
2
4
(
1
.
6
5
)
0
.
0
2
8
(
2
.
7
3
)
*
*
*
M
T
B
t
0
.
0
0
0
(
1
.
2
4
)
0
.
0
0
0
(
1
.
1
3
)
2
0
.
0
0
1
(
2
1
.
7
1
)
*
2
0
.
0
0
1
(
2
1
.
9
8
)
*
*
0
.
0
0
2
(
1
.
5
3
)
0
.
0
0
1
(
1
.
1
9
)
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
A
d
j
u
s
t
e
d
R
2
(
%
)
2
.
6
3
4
.
6
1
2
6
.
7
7
2
.
5
0
1
5
.
6
3
9
.
5
8
N
o
t
e
s
:
S
i
g
n
i
?
c
a
n
c
e
a
t
:
*
1
0
,
*
*
5
,
a
n
d
*
*
*
1
p
e
r
c
e
n
t
l
e
v
e
l
s
(
o
n
e
t
a
i
l
e
d
)
,
r
e
s
p
e
c
t
i
v
e
l
y
:
a
D
E
P
E
N
D
E
N
T
i
t
¼
a
1
i
þ
b
1
i
S
E
O
t
þ
b
2
i
S
I
Z
E
t
þ
b
3
i
L
E
V
t
þ
b
4
i
R
O
A
t
þ
b
5
i
M
T
B
t
þ
b
6
i
L
O
S
S
t
þ
b
7
i
L
I
T
R
I
S
K
t
þ
X
j
b
j
i
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
t
b
D
E
P
E
N
D
E
N
T
i
t
¼
a
1
i
þ
b
1
i
S
E
O
t
þ
b
2
i
S
a
r
b
O
x
t
þ
b
3
i
S
E
O
*t
S
a
r
b
O
x
t
þ
b
4
i
S
I
Z
E
t
þ
b
5
i
L
E
V
t
þ
b
6
i
R
O
A
t
þ
b
7
i
M
T
B
t
þ
b
8
i
L
O
S
S
t
þ
b
9
i
L
I
T
R
I
S
K
t
þ
X
j
b
j
i
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
w
h
e
r
e
D
E
P
E
N
D
E
N
T
i
t
r
e
p
r
e
s
e
n
t
s
d
i
s
c
r
e
t
i
o
n
a
r
y
a
c
c
r
u
a
l
o
r
a
b
n
o
r
m
a
l
r
e
a
l
a
c
c
o
u
n
t
v
a
r
i
a
b
l
e
;
a
l
l
v
a
r
i
a
b
l
e
s
a
r
e
d
e
?
n
e
d
i
n
A
p
p
e
n
d
i
x
1
;
t
h
e
t
-
s
t
a
t
i
s
t
i
c
s
a
r
e
c
a
l
c
u
l
a
t
e
d
f
r
o
m
W
h
i
t
e
’
s
h
e
t
e
r
o
s
k
e
d
a
s
t
i
c
i
t
y
-
c
o
r
r
e
c
t
e
d
s
t
a
n
d
a
r
d
e
r
r
o
r
s
Table IV.
Multivariate tests
of differences in
discretionary accruals
and real accounts
in SEO and matched
observations
Real and accrual-
based earnings
65
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
The results in panel B indicate that discretionary accounts receivable is signi?cantly
lower in the post-SarbOx period (coef?cient of SEO
*
t
SarbOx
t
¼ 20:009 with DAR
t
as the dependent variable, signi?cant at the 10 percent level). This means that DAR
t
is lower by 0.9 per cent of total assets in SEOs following SarbOx. This is equivalent to an
average of $8.4m (median of $1.6m). Abnormal real accounts are not signi?cantly
different following SarbOx (coef?cient of SEO
*
t
SarbOx
t
¼ 20:006 and insigni?cant
with AREAL
t
as the dependent variable). Therefore, there is evidence of a decline in one
type of accrual manipulation, namely accounts receivable, for SEOs following SarbOx.
However, there is evidence that the overall real account manipulation behavior has
changed in the post-SarbOx period (coef?cient on SarbOx
t
¼ 20.030, signi?cant at the
1 percent level with AREAL
t
as the dependent variable). Overall, the results indicate that
SEOs engage in accrual manipulation and some form of real account manipulation to
increase income in the year prior to the offering in the pre-SarbOx period. It is not evident
that SEOs reduce discretionary expenses (SG&A, R&D, and advertising expenses) or
increase production in order to positively in?uence earnings. There is some evidence of
a reduction in accrual manipulation in the post-SarbOx period in the SEO setting and
a shift to real accrual manipulation in all ?rms.
4.2 Accrual and real manipulation in sued and non-sued SEOs
To test whether litigation subsequent to SEOs is associated with the type of manipulation,
we ?rst examine univariate differences in the mean and median discretionary accrual and
real accounts for sued and non-sued SEOs. From the sample of 1,871 SEOs, a total of
218 issues resulted in subsequent litigation. However, only 133 ?rms were sued over
accounting issues. We only include these 133 SEO observations in the analysis. The
remaining SEO observations (1,653) did not face litigation subsequent to the SEO date.
Table V presents the univariate differences between sued and non-sued SEOs. Panel
A presents the differences in all periods, whereas panel B presents the results
separately in the pre- and post-SarbOx periods.
Sued SEOs appear to be more pro?table in the year prior to the offering but have
less cash from operations (difference in median INC
t
is 0.093 and difference in median
CFO
t
is 20.004, both signi?cant at the 10 percent level). DAC
t
is larger in magnitude in
sued SEO ?rms (difference in mean is 0.017 and difference in median is 0.004,
signi?cant at the 5 and 10 percent levels, respectively). DAR
t
is not signi?cantly
different in sued and non-sued SEOs. Abnormal discretionary expenses are higher in
sued SEOs (difference in median is 0.037, signi?cant at the 5 percent level) but this
represents income-decreasing manipulation. The total real account manipulation
measure, AREAL
t
, is signi?cantly higher in sued SEOs (difference in mean is 0.036 and
difference in median is 0.027, signi?cant at the 10 and 5 percent levels, respectively).
Panel B presents the results separately in the pre- and post-SarbOx periods. In the
pre-SarbOx period, there are 104 SEO observations that led to subsequent class action
litigation for accounting irregularities (7 percent of all SEOs in that period). In the post-
SarbOx period, there are 29 sued SEOs (8 percent of all SEOs in that period). The results
in the pre-SarbOx period are similar to those in the overall sample. DAC
t
is higher
in sued SEOs (difference in mean is 0.019, signi?cant at the 10 percent level).
Abnormal discretionary expenses are higher in sued SEOs (median difference is 0.038,
signi?cant at the 10 percent level) but this is inconsistent with income-increasing
manipulation. The total real account manipulation measure, AREAL
t
, is also
ARJ
24,1
66
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
signi?cantly higher in sued SEOs (mean difference is 0.051 and median difference is
0.037, both signi?cant at the 5 percent level).
In the post-SarbOx period, abnormal cash from operations is lower in sued SEOs
(mean difference is 20.041 and median difference is 20.045, both signi?cant at the
Sued SEOs Non-sued SEOs Difference
Variable Mean Median Mean Median Mean Median
Panel A: mean and median differences in ?nancial variables, discretionary accruals, and real accounts
between sued and non-sued SEO ?rms in all years
Financial variables n ¼ 133 n ¼ 1,653
INC
t
1.511 1.332 1.440 1.239 0.071 0.093
*
REV
t
20.051 20.047 20.062 20.058 0.011 0.011
CFO
t
20.044 0.032 20.027 0.036 20.017 20.004
*
TAC
t
20.018 0.049 20.012 0.049 20.006 0.000
ROA
t
0.033 0.062 0.050 0.084 20.017 20.022
Discretionary accruals and real accounts
DAC
t
0.022 0.009 0.004 0.005 0.017
* *
0.004
*
DAR
t
0.012 0.002 0.012 0.002 0.000 0.000
ACFO
t
20.014 20.003 20.009 0.013 20.005 20.016
APROD
t
20.020 20.010 20.025 20.020 0.005 0.010
ADEXP
t
0.160 0.065 0.125 0.028 0.036 0.037
* *
AREAL
t
0.126 0.062 0.090 0.035 0.036
*
0.027
* *
Panel B: mean and median differences in discretionary accruals and real accounts between sued and
non-sued SEO ?rms in pre- and post-SarbOx periods
Pre-SarbOx n ¼ 104 n ¼ 1,304
DAC
t
0.025 0.013 0.005 0.006 0.019
*
0.007
DAR
t
0.016 0.004 0.014 0.002 0.003 0.002
ACFO
t
20.004 0.001 20.009 0.016 0.005 20.015
APROD
t
20.024 20.020 20.023 20.020 20.001 0.000
ADEXP
t
0.172 0.065 0.125 0.027 0.047 0.038
*
AREAL
t
0.144 0.075 0.093 0.038 0.051
* *
0.037
* *
Post-SarbOx n ¼ 29 n ¼ 349
DAC
t
0.010 0.008 0.001 20.002 0.010 0.010
DAR
t
20.003 20.001 0.007 0.002 20.010 20.003
ACFO
t
20.051 20.036 20.010 0.009 20.041
*
20.045
*
APROD
t
20.004 0.010 20.033 20.040 0.029 0.050
ADEXP
t
0.119 0.129 0.122 0.029 20.003 0.100
AREAL
t
0.064 0.047 0.079 0.029 20.015 0.018
Difference between pre- and post-SarbOx
DAC
t
20.014 20.005 20.005 20.008
* *
DAR
t
20.020 20.005 20.007
*
0.001
ACFO
t
20.047 20.037
*
20.001 20.007
APROD
t
0.021 0.030 20.010 20.020
* *
ADEXP
t
20.053 0.065 20.003 0.002
AREAL
t
20.080
* *
20.028 20.014 20.009
* *
Notes: Signi?cance levels of
*
10,
* *
5 and
* * *
1 percent (one-tailed), respectively, using t-test and
Wilcoxon two-sample test; the table reports descriptive information for the SEO ?rms with subsequent
litigation (only accounting allegations) and the SEO ?rms without subsequent litigation for the entire
period; any litigation for non-accounting allegations are excluded; the far right columns report the
differences betweenmeanandmedianvalues; all variables arede?nedinAppendix1; nreferstothe number
of observations; Pre-SarbOx refers to years 1990-2001 and post-SarbOx refers to years 2002-2004
Table V.
Univariate tests of
differences between
sued and non-sued SEO
observations
Real and accrual-
based earnings
67
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
10 percent level). This is consistent with income-increasing manipulation. There are no
other signi?cant differences.
The third section of Panel B presents the differences in the manipulation measures
between the pre- and post-SarbOx periods. ACFO
t
is signi?cantly lower in the
post-SarbOx period in sued SEOs (difference in median is 20.037, signi?cant at
the 10 percent level). Furthermore, total real account manipulation, AREAL
t
, is
signi?cantly lower in the post-SarbOx period (mean difference is 20.080, signi?cant at
the 5 percent level). As for differences between non-sued SEO ?rms in both periods,
DAC
t
is lower post-SarbOx (difference in median is 20.008, signi?cant at the 5 percent
level), DAR
t
is lower (difference in mean is 20.007, signi?cant at the 10 percent level),
and APROD
t
is lower (difference in median is 20.020, signi?cant at the 5 percent
level). These are indicative of a reduction in income-increasing accrual manipulation in
the non-sued observations, consistent with prior research. Firms that do not face
litigation have signi?cantly less income-increasing discretionary accruals in the
post-SarbOx period. However, ?rms that face litigation have no signi?cant differences
in discretionary accruals post-SarbOx; this implies that ?rms facing litigation are still
being targeted for accrual manipulation.
As before, we also conduct multivariate analyses while controlling for variables
typically associated with discretionary/abnormal accrual and real accounts. There is a
concern that the manipulation measures and the litigation variable are endogenous,
stemming from the causality between them. This would lead to a correlation between
the litigation variable and the error term in any regression including the discretionary
accrual and abnormal real account variables as the dependent variable and litigation as
an independent variable, and thus will lead to biased results. We therefore employ a
two-stage least square analysis. In the ?rst step, we measure a predicted value of
litigation based on exogenous variables that affect the probability of litigation (detailed
in Appendix 2). In the second step, we use the predicted value of the probability of
litigation in the following two regressions:
DEPENDENT
it
¼a
1i
þb
1i
P_LIT
t
þb
2i
SIZE
t
þb
3i
LEV
t
þb
4i
ROA
t
þb
5i
MTB
t
þb
6i
LOSS
t
þb
7i
LITRISK
t
þ
X
j
b
ji
Industry Dummies
ð3Þ
DEPENDENT
it
¼a
1i
þb
1i
P_LIT
t
þb
2i
SarbOx
t
þb
3i
LIT
*
t
SarbOx
t
þb
4i
SIZE
t
þb
5i
LEV
t
þb
6i
ROA
t
þb
7i
MTB
t
þb
8i
LOSS
t
þb
9i
LITRISK
t
þ
X
j
b
ji
Industry Dummies
ð4Þ
where DEPENDENT
it
is the discretionary accrual or abnormal real account measure,
P_LIT is the predicted value of the probability of litigation, and the remaining
variables are as previously de?ned. As before, the coef?cients of interest are b
1i
in
regression (3) and b
3i
in regression (4). Given our H1, we expect a positive and
signi?cant coef?cient b
11
when the dependent variable is discretionary accruals.
This would indicate accrual manipulation in the year prior to the offering in defendant
?rms as compared to non-defendant ?rms. We do not expect a signi?cant coef?cient in
ARJ
24,1
68
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the ?nal regression (b
16
) as this represents real account manipulation. Given our H2,
we expect a signi?cant coef?cient b
36
using regression (4), with the dependent variable
being the real manipulation measure, to represent more scrutiny of real account
manipulation. The results appear in Table VI.
The results of regression (1) appear in Panel A. We ?nd that ?rms that are sued
have higher DAC
t
(coef?cient on P_LIT
t
¼ 0.177 in the regression with DAC
t
as the
dependent variable, signi?cant at the 5 percent level). This is in line with
income-increasing accrual manipulation. However, it does not appear that this accrual
manipulation is achieved through accounts receivable (consistent with revenue
manipulation) as the variable P_LIT
t
is insigni?cant in the regression with DAR
t
as the
dependent variable. We also ?nd that ?rms that are sued do not show evidence of real
account manipulation as the variable P_LIT
t
is insigni?cant in the regressions with
AREAL
t
as the dependent variable. However, there is evidence of income-increasing
manipulation using abnormal cash?ows (coef?cient on P_LIT
t
¼ 20.366 in the
regression with ACFO
t
as the dependent variable, signi?cant at the 1 percent level).
The magnitude of earnings management in sued SEOs is harder to interpret from
these coef?cients as the test variable is a predicted variable and not the LIT variable
directly. These results are in line with our H1. We ?nd that defendant ?rms engage in
accrual manipulation but not real manipulation in the year prior to the offering,
which means that investors focus on accrual manipulation in initiating class action
litigation.
In panel B, we ?nd that the only signi?cant change in litigation behavior post-SarbOx
is due to DAR
t
(coef?cient on P_LIT
*
SarbOx ¼ 20.205 in the regression with DAR
t
as
the dependent variable, signi?cant at the 1 percent level). This indicates that investors
have reduced their scrutiny of discretionary accounts receivable in the post-SarbOx
period. However, we do not ?nd more scrutiny of real account manipulation (coef?cient
on P_LIT
*
SarbOx ¼ 20.288 in the regression with DREAL
t
as the dependent variable,
and not signi?cant). Therefore, there is no support for our H2 that investors have more
scrutiny of real account manipulation in the post-SarbOx period.
Overall, the results indicate that sued SEOs have a higher prevalence of accrual
manipulation. However, there is also a higher prevalence of income-increasing real
account manipulation through cash ?ow accounts. Following the enactment of SarbOx,
investors seem to have reduced their reliance on accrual manipulation in the form of
accounts receivable. Overall, it appears that investors focus on abnormal accrual
patterns in instigating litigation.
4.3 Tests of probability of litigation
As a robustness check and in order to more easily interpret the magnitude of
earnings management in sued SEOs, in this section we directly test the probability of
litigation in relation to the various accrual and real account manipulation variables.
We follow the analysis in DuCharme et al. (2004) and run the following two
regressions[11]:
ProbðLIT
t
Þ ¼a
1
þb
1
DAC
t
þb
2
DAR
t
þb
3
ACFO
t
þb
4
APROD
t
þb
5
ADEXP
t
þb
6
AUD
t
þb
7
Principal
t
þb
8
Secondary
t
þb
9
High 2Tech
t
ð5Þ
Real and accrual-
based earnings
69
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
D
e
p
.
v
a
r
i
a
b
l
e
D
A
C
D
A
R
A
C
F
O
A
P
R
O
D
A
D
E
X
P
A
R
E
A
L
P
a
n
e
l
A
:
c
o
e
f
?
c
i
e
n
t
s
(
t
-
s
t
a
t
i
s
t
i
c
s
)
f
r
o
m
r
e
g
r
e
s
s
i
o
n
s
o
f
t
h
e
f
o
r
m
a
P
_
L
I
T
t
0
.
1
7
7
(
1
.
9
8
)
*
*
0
.
0
6
5
(
1
.
0
1
)
2
0
.
3
6
6
(
2
2
.
9
9
)
*
*
*
0
.
2
6
0
(
1
.
5
7
)
0
.
4
0
2
(
1
.
5
5
)
0
.
2
9
7
(
1
.
6
1
)
S
I
Z
E
t
2
0
.
0
0
9
(
2
4
.
7
2
)
*
*
*
2
0
.
0
0
5
(
2
3
.
3
8
)
*
*
*
0
.
0
1
6
(
5
.
9
0
)
*
*
*
0
.
0
0
2
(
0
.
5
8
)
2
0
.
0
3
5
(
2
6
.
0
7
)
*
*
*
2
0
.
0
1
7
(
2
4
.
1
4
)
*
*
*
L
E
V
t
2
0
.
0
3
3
(
2
4
.
4
1
)
*
*
*
0
.
0
7
0
(
1
3
.
0
0
)
*
*
*
2
0
.
0
0
9
(
2
0
.
8
7
)
0
.
0
4
4
(
3
.
2
1
)
*
*
*
2
0
.
0
3
5
(
2
1
.
6
0
)
0
.
0
0
1
(
0
.
0
7
)
R
O
A
t
2
0
.
0
2
0
(
2
1
.
8
1
)
*
0
.
0
0
1
(
0
.
1
1
)
0
.
2
7
0
(
1
7
.
5
9
)
*
*
*
2
0
.
0
4
6
(
2
2
.
2
2
)
*
*
2
0
.
3
5
9
(
2
1
0
.
9
8
)
*
*
*
2
0
.
1
3
5
(
2
5
.
8
2
)
*
*
*
L
O
S
S
t
2
0
.
0
3
5
(
2
5
.
0
4
)
*
*
*
2
0
.
0
0
2
(
2
0
.
4
3
)
2
0
.
0
6
0
(
2
6
.
1
9
)
*
*
*
0
.
0
1
6
(
1
.
2
4
)
0
.
0
7
5
(
3
.
6
4
)
*
*
*
0
.
0
3
1
(
2
.
1
2
)
*
*
L
I
T
R
I
S
K
t
2
0
.
0
1
5
(
2
1
.
7
4
)
*
0
.
0
0
5
(
0
.
8
2
)
0
.
0
3
0
(
2
.
5
6
)
*
*
*
2
0
.
0
0
7
(
2
0
.
4
4
)
2
0
.
0
1
9
(
2
0
.
7
9
)
0
.
0
0
4
(
0
.
2
1
)
M
T
B
t
0
.
0
0
0
(
1
.
3
1
)
0
.
0
0
0
(
0
.
9
5
)
2
0
.
0
0
1
(
2
2
.
9
5
)
*
*
*
2
0
.
0
0
1
(
2
1
.
2
9
)
0
.
0
0
3
(
3
.
8
0
)
*
*
*
0
.
0
0
1
(
2
.
2
5
)
*
*
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
A
d
j
u
s
t
e
d
R
2
(
%
)
3
.
8
1
9
.
2
7
3
3
.
9
1
4
.
1
0
2
4
.
3
2
1
3
.
4
5
P
a
n
e
l
B
:
c
o
e
f
?
c
i
e
n
t
s
(
t
-
s
t
a
t
i
s
t
i
c
s
)
f
r
o
m
r
e
g
r
e
s
s
i
o
n
s
o
f
t
h
e
f
o
r
m
b
P
_
L
I
T
t
0
.
1
6
6
(
1
.
7
8
)
*
0
.
1
1
9
(
1
.
7
8
)
*
2
0
.
4
0
8
(
2
3
.
1
9
)
*
*
*
0
.
2
4
9
(
1
.
4
4
)
0
.
5
2
3
(
1
.
9
2
)
*
0
.
3
6
6
(
1
.
9
0
)
*
S
a
r
b
O
x
t
0
.
0
0
0
(
0
.
0
0
)
0
.
0
1
2
(
1
.
6
5
)
*
2
0
.
0
1
7
(
2
1
.
2
1
)
2
0
.
0
1
2
(
2
0
.
6
3
)
0
.
0
3
2
(
1
.
0
7
)
0
.
0
0
3
(
0
.
1
4
)
P
_
L
I
T
*t
S
a
r
b
O
x
t
0
.
0
4
7
(
2
0
.
4
8
)
2
0
.
2
0
5
(
2
2
.
9
2
)
*
*
*
0
.
1
3
9
(
1
.
0
4
)
0
.
0
1
4
(
0
.
0
7
)
2
0
.
4
4
1
(
2
1
.
5
4
)
2
0
.
2
8
8
(
2
1
.
4
3
)
S
I
Z
E
t
2
0
.
0
0
9
(
2
4
.
6
9
)
*
*
*
2
0
.
0
0
5
(
2
3
.
3
1
)
*
*
*
0
.
0
1
6
(
5
.
9
6
)
*
*
*
0
.
0
0
3
(
0
.
7
5
)
2
0
.
0
3
5
(
2
5
.
9
9
)
*
*
*
2
0
.
0
1
6
(
2
3
.
8
4
)
*
*
*
L
E
V
t
2
0
.
0
3
3
(
2
4
.
3
5
)
*
*
*
0
.
0
6
9
(
1
2
.
9
2
)
*
*
*
2
0
.
0
0
9
(
2
0
.
9
2
)
0
.
0
4
4
(
3
.
1
4
)
*
*
*
2
0
.
0
3
5
(
2
1
.
6
1
)
2
0
.
0
0
1
(
2
0
.
0
4
)
R
O
A
t
2
0
.
0
2
0
(
2
1
.
7
9
)
*
0
.
0
0
1
(
0
.
0
8
)
0
.
2
7
0
(
1
7
.
5
4
)
*
*
*
2
0
.
0
4
7
(
2
2
.
2
6
)
*
*
2
0
.
3
5
9
(
2
1
0
.
9
7
)
*
*
*
2
0
.
1
3
6
(
2
5
.
8
8
)
*
*
*
L
O
S
S
t
2
0
.
0
3
6
(
2
5
.
0
8
)
*
*
*
2
0
.
0
0
1
(
2
0
.
2
5
)
2
0
.
0
6
0
(
2
6
.
1
5
)
*
*
*
0
.
0
1
7
(
1
.
3
1
)
0
.
0
7
6
(
3
.
6
9
)
*
*
*
0
.
0
3
3
(
2
.
2
9
)
*
*
L
I
T
R
I
S
K
t
2
0
.
0
1
5
(
2
1
.
7
5
)
*
0
.
0
0
5
(
0
.
7
7
)
0
.
0
3
0
(
2
.
6
3
)
*
*
*
2
0
.
0
0
6
(
2
0
.
3
8
)
2
0
.
0
2
0
(
2
0
.
8
3
)
0
.
0
0
4
(
0
.
2
5
)
M
T
B
t
0
.
0
0
0
(
1
.
3
4
)
0
.
0
0
0
(
0
.
8
3
)
2
0
.
0
0
1
(
2
2
.
9
3
)
*
*
*
2
0
.
0
0
1
(
2
1
.
3
0
)
0
.
0
0
3
(
3
.
7
5
)
*
*
*
0
.
0
0
1
(
2
.
1
7
)
*
*
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
I
n
c
l
u
d
e
d
A
d
j
u
s
t
e
d
R
2
(
%
)
3
.
7
3
9
.
6
4
3
3
.
8
9
4
.
0
3
2
4
.
3
4
1
3
.
5
5
N
o
t
e
s
:
S
i
g
n
i
?
c
a
n
c
e
a
t
:
*
1
0
,
*
*
5
a
n
d
*
*
*
1
p
e
r
c
e
n
t
l
e
v
e
l
s
(
o
n
e
-
t
a
i
l
e
d
)
,
r
e
s
p
e
c
t
i
v
e
l
y
;
n
¼
1
,
7
8
6
;
a
l
l
v
a
r
i
a
b
l
e
s
a
r
e
d
e
?
n
e
d
i
n
A
p
p
e
n
d
i
x
1
:
a
D
E
P
E
N
D
E
N
T
i
t
¼
a
1
i
þ
b
1
i
P
_
L
I
T
t
þ
b
2
i
S
I
Z
E
t
þ
b
3
i
L
E
V
t
þ
b
4
i
R
O
A
t
þ
b
5
i
M
T
B
t
þ
b
6
i
L
O
S
S
t
þ
b
7
i
L
I
T
R
I
S
K
t
þ
X
j
b
j
i
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
b
D
E
P
E
N
D
E
N
T
i
t
¼
a
1
i
þ
b
1
i
P
_
L
I
T
t
þ
b
2
i
S
a
r
b
O
x
t
þ
b
3
i
L
I
T
*t
S
a
r
b
O
x
t
þ
b
4
i
S
I
Z
E
t
þ
b
5
i
L
E
V
t
þ
b
6
i
R
O
A
t
þ
b
7
i
M
T
B
t
þ
b
8
i
L
O
S
S
t
þ
b
9
i
L
I
T
R
I
S
K
t
þ
X
j
b
j
i
I
n
d
u
s
t
r
y
D
u
m
m
i
e
s
w
h
e
r
e
D
E
P
E
N
D
E
N
T
i
t
r
e
p
r
e
s
e
n
t
s
d
i
s
c
r
e
t
i
o
n
a
r
y
a
c
c
r
u
a
l
o
r
a
b
n
o
r
m
a
l
r
e
a
l
a
c
c
o
u
n
t
v
a
r
i
a
b
l
e
Table VI.
Multivariate tests of
differences between sued
and non-sued SEOs
ARJ
24,1
70
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
ProbðLIT
t
Þ ¼a
1
þb
1
DAC
t
þb
2
DAR
t
þb
3
ACFO
t
þb
4
APROD
t
þb
5
ADEXP
t
þb
6
SarbOx
t
þb
7
DAC
*
t
SarbOx
t
þb
8
DAR
*
t
SarbOx
t
þb
9
ACFO
*
t
SarbOx
t
þb
10
APROD
*
t
SarbOx
t
þb
11
ADEXP
*
t
SarbOx
t
þb
12
AUD
t
þb
13
Principal
t
þb
14
Secondary
t
þb
15
High 2Tech
t
ð6Þ
where LIT is an indicator variable that takes on the value of 1 if the SEO is
subsequently sued for accounting irregularities, and 0 if it is not subsequently sued,
DAC
t
, DAR
t
, ACFO
t
, APROD
t
, and ADEXP
t
are the discretionary/abnormal accrual
and real accounts as previously de?ned. AREAL
t
is not included in the regression since
it is the sum of the three abnormal real accounts. The second regression includes the
SarbOx variable as well as interaction variables for SarbOx and all
discretionary/abnormal accrual and real accounts. The control variables used are:
AUD
t
¼ indicator variable that takes on the value of 1 if the auditor is a
prestigious one (de?ned as Compustat item no. 149 (the auditor
component) between 1 and 8, and 0 otherwise.
Principal
t
¼ proceeds from offer (offer size)/beginning total assets.
Secondary
t
¼ secondary shares issued/total shares issued.
High-Tech
t
¼ indicator variable that takes on the value of 1 if the SEO belongs to
a high-technology industry and 0 otherwise.
The high-technology industries include computers and of?ce equipment, consumer
electronics, communications equipment, electronic components and accessories,
semiconductors, industrial electronics, photonics, defense electronics, electro-medical
equipment, communications services, and software and computer-related services[12].
The variable AUD is used as a control variable since offerings with prestigious
auditors are unlikely to attract lawsuits because the risk of dramatically poor post-offer
stock returns is low. On the other hand, auditors may be named as codefendants,
alongside offering ?rms. The deep pockets that tend to accompany prestige may attract
lawsuits. Therefore, it is unclear what the direction of the relationship between litigation
and AUD will be. The variable Principal is used to control for the offer size which may
also affect the incidence of lawsuits. If the offer is small, the potential for dollar damages
to participating investors is also small. It may not be worth suing ?rms over a small
offer, if there are ?xed costs of litigation. On the other hand, well-known established
?rms make most of the large offers. These offers may be among the least risky and
therefore least likely to precede very low rates of return to stockholders. The variable
Secondary is used as the fraction of the offer that is secondary also has an impact, albeit
ambiguous, on litigation risk. This fraction may be large when a ?rm’s principal
shareholders are substantially divesting their stakes and may seek to deceive investors
about factors affecting ?rm value. Deceptive behavior in connection with a stock offer
would tend to increase the risk of a subsequent lawsuit. However, investors and
regulatory authorities may closely scrutinize offers that contain large secondary
fractions. Other things equal, this would increase the chances that illegal deceptive
behavior is later exposed and punished. The increased risk of punishment would tend
Real and accrual-
based earnings
71
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
to discourage deceptive behavior and reduce the incidence of related lawsuits. Hence, it
is not clear whether, on balance, the incidence of lawsuits should be positively or
negatively related to the fraction of an offer that is secondary. Finally, speci?c industries
may also attract more litigation. For example, in the period after year 2000, there was an
abundance of litigation targeted towards ?rms in the high-technology industry.
Since the dependent variable, LIT, is a dichotomous variable, ordinary least square
is not appropriate for estimating the coef?cients. We instead use probit regression
using a logistic distribution. In essence, the regression measures the probability of
being sued given the independent variables in the model. The results appear in
Table VII. To test our H1, we examine the coef?cients b
1
2 b
5
. We expect that b
1
(and
perhaps b
2
) will be positive and signi?cant.
We ?nd that discretionary accruals (DAC
t
) are positively associated with the
probability of litigation (coef?cient ¼ 1.784, signi?cant at the 10 percent level). This
means that SEOs are almost twice as likely to be sued if their discretionary accruals are
higher by one unit (in this case an amount equivalent to 1 percent of total assets).
Abnormal cash from operations is also associated with the probability of litigation
Indep. variable Without SarbOx interaction terms With SarbOx interaction terms
DAC
t
1.784 (3.77)
*
2.163 (4.69)
* *
DAR
t
20.298 (0.07) 20.038 (0.00)
ACFO
t
1.338 (3.20)
*
2.061 (6.39)
* *
APROD
t
0.672 (1.43) 0.942 (2.30)
ADEXP
t
0.536 (2.07) 0.842 (4.07)
* *
SarbOx
t
0.303 (1.56)
DAC
*
t
SarbOx
t
22.000 (0.64)
DAR
*
t
SarbOx
t
26.884 (2.81)
*
ACFO
*
t
SarbOx
t
25.283 (6.64)
* * *
APROD
*
t
SarbOx
t
21.220 (0.56)
ADEXP
*
t
SarbOx
t
22.139 (4.71)
* *
AUD
t
0.826 (3.40)
*
0.890 (3.84)
*
Principal
t
0.057 (2.96)
*
0.067 (3.62)
*
Secondary
t
21.001 (5.67)
* *
20.952 (5.02)
* *
High-Tech
t
0.783 (18.09)
* * *
0.800 (18.55)
* * *
Log Likelihood 2454.74 2450.28
Notes: Signi?cance levels of
*
10,
* *
5 and
* * *
1 percent (one-tailed), respectively; n ¼ 1,786; all
variables are de?ned in Appendix 1; the table presents coef?cients (x
2
-values) from the following
probit regressions:
ProbðLIT
t
Þ ¼ a
1
þb
1
DAC
t
þb
2
DAR
t
þb
3
ACFO
t
þb
4
APROD
t
þb
5
ADEXP
t
þb
6
AUD
t
þb
7
Principal
t
þb
8
Secondary
t
þb
9
High 2Tech
t
ProbðLIT
t
Þ ¼a
1
þb
1
DAC
t
þb
2
DAR
t
þb
3
ACFO
t
þb
4
APROD
t
þb
5
ADEXP
t
þb
6
SarbOx
t
þb
7
DAC
*
t
SarbOx
t
þb
8
DAR
*
t
SarbOx
t
þb
9
ACFO
*
t
SarbOx
t
þb
10
APROD
*
t
SarbOx
t
þb
11
ADEXP
*
t
SarbOx
t
þb
12
AUD
t
þb
13
Principal
t
þb
14
Secondary
t
þb
15
High 2Tech
t
Table VII.
Tests of probability
of litigation
ARJ
24,1
72
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
(coef?cient for ACFO
t
¼ 1.338 and signi?cant at the 10 percent level). However,
this indicates that the probability of litigation increases as abnormal cash ?ows increase
which is inconsistent with income-increasing manipulation. All other abnormal
accounts are not signi?cantly related to the probability of litigation. We ?nd that the
probability of litigation is signi?cantly related to the type of auditor (coef?cient of
AUD
t
¼ 0.826, signi?cant at the 10 percent level), indicating that ?rms are more likely to
be sued if the auditor is a prestigious one. The likelihood of litigation is also positively
related to the principal amount of the offering (coef?cient of Principal
t
¼ 0.057,
signi?cant at the 10 percent level), and being ina high-technology industry(coef?cient of
High-Tech
t
¼ 0.783, signi?cant at the 1 percent level). On the other hand, the likelihood
of litigation is negatively associated with the share of the offer that is secondary
(coef?cient of Secondary
t
¼ 21.001, signi?cant at the 5 percent level).
The second regression shows the effect of SarbOx on the probability of litigation. The
coef?cients on the interaction terms between SarbOx and the discretionary/abnormal
accruals and real accounts show how the probability of litigation has changed
following the enactment of SarbOx. We expect a signi?cant coef?cient for one or more
of the real manipulation measures (b
9
2 b
11
). We ?nd that investors are paying less
attention to income-increasing discretionary accounts receivable following
SarbOx (coef?cient ¼ 26.884, signi?cant at the 10 percent level). However, ?rms
with lower abnormal cash from operations are scrutinized more in the SarbOx period
(coef?cient ¼ 25.283, signi?cant at the 1 percent level). Also, ?rms with lower
abnormal discretionary expenses are more likely to be sued (coef?cient ¼ 22.139,
signi?cant at the 5 percent level). This provides some support for our H2.
Overall, we ?nd that the probability of litigation is associated with
income-increasing accrual earnings management as well as income-decreasing real
account manipulation. The enactment of SarbOx has shifted the focus somewhat from
accrual to real account manipulation.
5. Conclusion
In this paper, we study the alternative methods of manipulating bottom-line earnings
in a SEO setting and address whether litigation for accounting irregularities varies
with the method(s) used to reach earnings goals around these stock issues.
We identify a sample of 1,871 SEO observations and compare them to a matched
sample of 1,871 ?rm/year observations that did not undergo an equity offering. We ?nd
that the SEO ?rms engage in income-increasing manipulation using more than one
method. Speci?cally, they engage in both accrual and real account manipulation in the
year prior to the offering, through discretionary accruals, and by increasing sales
through more lenient terms (shown as a lower level of abnormal cash from operations).
SEOs in general have higher abnormal discretionary expenses but this is most likely
related to the types of ?rms that initiate SEOs. We then examine which of the ?rms
were sued from the above sample. We ?rst employ a two-stage least square
methodology to account for endogeneity between the probability of litigation and the
accrual and real account manipulation measures and ?nd that ?rms that are
subsequently sued by investors following the SEO have a higher prevalence of
discretionary accruals, when the lawsuit allegations are related to accounting issues.
This is indicative of earnings manipulation using accrual accounts and is
consistent with prior research on litigation subsequent to SEOs (Ibrahim et al., 2008).
Real and accrual-
based earnings
73
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
There is no evidence of overall real account manipulation in the sued SEO ?rms but
there is some evidence of one particular type of real manipulation. These results are in
line with our expectations. Our H1 states that SEO ?rms that are subsequently sued for
accounting irregularities will have a higher prevalence of accrual manipulation but not
real account manipulation when compared to non-defendant SEO ?rms. In robustness
tests, we test how the probability of litigation is affected by these manipulation
measures. We ?nd that SEOs are more likely to be sued when they have higher
income-increasing discretionary accruals. There is no evidence that the probability of
litigation is related to income-increasing real account manipulation. These ?ndings are
important to academicians as well as investors. They indicate that even though there is
evidence of real account manipulation, it appears that investors still focus on accrual
manipulation in initiating litigation against ?rms and may not able to detect real
account manipulation. This paper provides the ?rst evidence of the relationship
between real account manipulation and litigation risk.
The SarbOx Act of 2002 was designed to reduce earnings management, speci?cally
through accrual accounts. There is evidence that this was successful in the SEO
setting. There is some evidence that SEO ?rms have shifted to real account
manipulation post-SarbOx. This is in line with prior research and indicates that accrual
and real account manipulation are substitutes used to manage earnings. Following the
enactment of SarbOx, we ?nd that investors are paying less attention to one type of
accrual manipulation, namely account receivable manipulation. We also hypothesize
and ?nd some support that investors may be paying more attention to real
manipulation measures in their litigation decisions in the post-SarbOx period.
Some limitations of the study include generalisability and methodology issues.
Speci?cally, this study focuses on a very speci?c setting, SEOs. The conclusions of the
study may not be generalisable to other settings. This is an avenue for future research.
Furthermore, the post-SarbOx period is small due to data limitations which may
impact the results. Finally, as in all earnings management studies, the manipulation
measures may not fully capture the magnitude and nature of manipulation.
Notes
1. The offerings were made in the years 1991-2005; however our test sample is in the years
1990-2004 since we test manipulation in the year prior to the offering.
2. The full act can be accessed at: www.gpo.gov/fdsys/pkg/PLAW-107publ204/pdf/PLAW-
107publ204.pdf
3. We use the four-digit SIC code to designate an industry.
4. We eliminate ?rms that underwent an IPO from the non-SEO observations.
5. The distribution of the matched ?rms is identical.
6. The ?rm and industry subscripts have been suppressed from this and following regressions.
7. We omit DAR
t
from the right-hand side of the regression, unlike Ibrahim (2009).
8. If SG&A is not missing but one or both of the other discretionary expenses are missing, they
are set to 0.
9. Abnormally increasing production, keeping the level of sales constant, would shift some of
the costs to the balance sheet, in the form of unsold inventory, abnormally decreasing cost of
goods sold.
ARJ
24,1
74
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
10. Francis et al. designate ?rms in the biotechnology, computers, electronics, and retailing
industries as subject to high litigation risk.
11. See Table VIII in DuCharme et al. (2004). Their test variable is abnormal working capital
rather than the breakdown of discretionary/abnormal accruals and real accounts in this
paper. They also include a variable for cumulative abnormal return and the type of
underwriter. We cannot obtain information regarding the type of underwriter and thus do
not include it in the regressions. In untabulated results, we include a measure of cumulative
abnormal returns over the 30 days following the offering date, with similar results. The
number of observations was lower due to missing data.
12. The speci?c SIC codes designated as high tech are available upon request.
References
Aharony, J., Lin, C.J. and Loeb, M.P. (1993), “Initial public offerings, accounting choices, and
earnings management”, Contemporary Accounting Research, Vol. 10, pp. 61-81.
Alexander, J.C. (1991), “Do the merits matter? A study of settlements in securities class actions”,
Stanford Law Review, Vol. 43, pp. 497-598.
Armstrong, C., Foster, G. and Taylor, D. (2009), “Earnings management around initial public
offerings: a re-examination”, Rock Center for Corporate Governance Working Paper No. 23,
University of Pennsylvania, Philadelphia, PA, September.
Ball, R. and Shivakumar, L. (2008), “Earnings quality at initial public offerings”, Journal of
Accounting and Economics, Vol. 45, pp. 324-49.
Cohen, D.A. andZarowin, P. (2010), “Accrual-basedandreal earnings management activities around
seasoned equity offerings”, Journal of Accounting and Economics, Vol. 50 No. 1, pp. 2-19.
Cohen, D.A., Dey, A. and Lys, T. (2008), “Real and accrual-based earnings management in the
pre- and post-Sarbanes Oxley periods”, The Accounting Review, Vol. 83, May, pp. 757-88.
Dechow, P., Sloan, R. and Sweeney, A. (1995), “Detecting earnings management”, The Accounting
Review, Vol. 70, pp. 193-225.
DuCharme, L.L., Malatesta, P.H. and Sefcik, S.E. (2004), “Earnings management, stock issues,
and shareholder lawsuits”, Journal of Financial Economics, Vol. 71 No. 1, pp. 27-49.
Francis, J.D., Philbrick, D. and Schipper, K. (1994), “Shareholder litigation and corporate
disclosures”, Journal of Accounting Research, Vol. 32, pp. 137-164.
Frankel, R.M., Johnson, M.F. and Nelson, K.K. (2002), “The relation between auditors’ fees for
nonaudit services and earnings management”, The Accounting Review, Vol. 77, pp. 71-105
(supplement).
Friedlan, J. (1994), “Accounting choices by issuers of initial public offerings”, Contemporary
Accounting Research, Vol. 11, pp. 1-32.
Graham, J.R., Harvey, C.R. and Rajgopal, S. (2005), “The economic implications of corporate
?nancial reporting”, Journal of Accounting and Economics, Vol. 40, pp. 3-73.
Ibrahim, S. (2009), “The usefulness of measures of consistency of discretionary components of
accruals in the detection of earnings management”, Journal of Business Finance &
Accounting, Vol. 9/10, pp. 1087-116.
Ibrahim, S., Xu, L. and Deal, L. (2008), “The legal consequences of management of earnings
components: evidence from equity offerings”, working paper, Morgan State University,
Baltimore, MD, June.
Real and accrual-
based earnings
75
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Jones, C.L. (1998), “The association of earnings management with current returns, current market
values, future returns, executive compensation and the likelihood of being a target of 10B-5
litigation”, unpublished dissertation, Stanford University, Palo Alto, CA.
Jones, J. (1991), “Earnings management during import relief investigation”, Journal of Accounting
Research, Vol. 29 No. 22, pp. 193-228.
Kothari, S.P., Leone, A. and Wasley, C. (2005), “Performance matched discretionary accrual
measures”, Journal of Accounting and Economics, Vol. 39, February, pp. 163-97.
Li, H., Pincus, M. and Rego, S.O. (2008), “Market reaction to events surrounding the
Sarbanes-Oxley Act of 2002 and earnings management”, The Journal of Law &Economics,
Vol. 51, pp. 111-34.
Lu, Y. (2004), “Earnings management and securities litigation”, working paper, University of
Southern California, Los Angeles, CA.
Marquardt, C. and Wiedman, C. (2004), “How are earnings managed? An examination of speci?c
accruals”, Contemporary Accounting Research, Vol. 21, Summer, pp. 461-89.
Mizik, N. and Jacobson, R. (2007), “Earnings in?ation through accruals and real activity
manipulation: its prevalence at the time of an SEOand the ?nancial market consequences”,
working paper, Columbia University, SSRN Abstract No. 1031006, June, New York, NY.
Palmrose, Z. (1991), “Trials of legal disputes involving independent auditors: some empirical
evidence”, Journal of Accounting Research, Vol. 29, pp. 149-85 (supplement).
Palmrose, Z., Richardson, V. and Scholz, S. (2004), “Determinants of market reactions to
restatement announcements”, Journal of Accounting and Economics, Vol. 37, pp. 59-89.
Priest, G.L. and Klein, B. (1984), “The selection of disputes for litigation”, Journal of Legal Studies,
Vol. 13, pp. 1-55.
Rangan, S. (1998), “Earnings before seasoned equity offerings: are they overstated?”, Journal of
Financial Economics, Vol. 50, pp. 101-22.
Roychowdhury, S. (2006), “Earnings management through real activities manipulation”, Journal
of Accounting and Economics, Vol. 42, pp. 335-70.
Shivakumar, L. (2000), “Do ?rms mislead investors by overstating earnings before seasoned
equity offerings?”, Journal of Accounting and Economics, Vol. 29 No. 3, pp. 339-71.
Teoh, S.H., Welch, I. and Wong, T.J. (1998a), “Earnings management and the post-issue
performance of seasoned equity offerings”, Journal of Financial Economics, Vol. 50,
pp. 63-99.
Teoh, S.H., Wong, T.J. and Rao, G.R. (1998b), “Are accruals during initial public offerings
opportunistic?”, Review of Accounting Studies, Vol. 3, pp. 175-208.
Zang, A. (2007), “Evidence on the tradeoff between real manipulation and accrual manipulation”,
working paper, Hong Kong University of Science and Technology, Kowloon, March.
Further reading
PricewaterhouseCoopers (2000), Securities Litigation Study, available at:http://10b5.pwc.com/
PDF/2000STUDY.PDF (accessed June 2007).
ARJ
24,1
76
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Appendix 1
Variables De?nition and data source
Financial and accrual variables
A
t
Total assets in year t [6]
INC
t
Income before extraordinary items in year t [18]/A
t21
(total assets in year t 2 1) [6]
REV
t
Net revenues (sales) in year t [12]/A
t21
DREV
t
Change in net sales from year t 2 1 to year t [D12]/A
t21
DREV
t21
Change in net sales from year t 2 2 to year t 2 1 [D12]/A
t21
DAR
t
Change in accounts receivable from year t 2 1 to t [D2]/A
t21
PPE
t
Gross property, plant, and equipment in year t [7]/A
t21
CFO
t
Cash from operations in year t [308 2 124]/A
t21
COGS
t
Cost of goods sold in year t [41]/A
t21
DINV
t
Change in inventory from year t 2 1 to year t [D3]/A
t21
PROD
t
Production costs in year t [41 þ D3]/A
t21
ADV
t
Advertising expenses in year t [45]/A
t21
RD
t
Research and development expenses in year t [46]/A
t21
SGA
t
Selling, general and administrative expenses in year t [189]/A
t21
DEXP
t
Discretionary expenses in year t [45 þ 46 þ 189]/A
t21
TAC
t
Total accruals ¼ INC
t
2 CFO
t
ROA
t
Return on assets in year t ¼ NI
t
[172]/A
t
Discretionary accruals and abnormal real accounts
DAC
t
Discretionary accruals in year t from performance-modi?ed Jones
model ¼ TAC
t
2b
0
þ b
1
ð1=A
t21
Þ þ b
2
ðDREV
t
2DAR
t
Þ þ b
3
PPE
t
þ b
4
ðROA
t
Þ
DAR
t
Discretionary accounts receivable in year t from performance-modi?ed Jones
model ¼ DAR
t
2b
01
þ b
11
ð1=A
t21
Þ þ b
21
ðDREV
t
Þ þ b
31
PPE
t
þ b
41
ðROA
t
Þ
ACFO
t
Abnormal or discretionary cash ?ow from operations in year
t ¼ CFO
t
2a
11
ð1=A
t21
Þ þ a
12
ðREV
t
Þ þ a
13
ðDREV
t
Þ
APROD
t
Abnormal or discretionary production
costs ¼ PROD
t
2a
21
ð1=A
t21
Þ þ a
22
ðREV
t
Þ þ a
23
ðDREV
t
Þ þ a
24
ðDREV
t21
Þ
ADEXP
t
Abnormal or discretionary expenditures ¼ DEXP
t
2a
31
ð1=A
t21
Þ þ a
32
ðREV
t21
Þ
AREAL
t
Total abnormal real accounts ¼ ACFO
t
þ APROD
t
þ ADEXP
t
Control variables in multivariate tests
SIZE
t
Log (A
t
)
LEV
t
Leverage ¼ (short-term debt [34] þ long-term debt [9])/A
t
LOSS
t
An indicator variable equal to one if the observation has negative earnings before extraordinary
items in year t, and zero otherwise
MTB
t
Market-to-book value at year end (market value of equity divided by book value of equity)
LITRISK
t
An indicator variable equal to one if the observation is from an industry with high litigation risk,
and zero otherwise
AUD
t
An indicator variable equal to one if the auditor is a prestigious one (de?ned as a Big 4 auditor or
predecessor), and zero otherwise
Principal
t
Proceeds from issuance/A
t21
Secondary
t
Secondary shares issued/total share issued
High-Tech
t
An indicator variable equal to one if the observation belongs to a high-technology industry, and
zero otherwise
Test variables in multivariate tests
SEO An indicator variable equal to one for SEO observation in the year prior to the offering, and zero
for non-SEO observation
LIT An indicator variable equal to one if the SEO ?rm was subsequently sued for accounting
irregularities, and zero if the SEO ?rm was not subsequently sued
P_LIT Predicted value of probability of litigation measured from probit regression of LIT on exogenous
variables
SarbOx An indicator variable equal to one if the observation is in year 2002-2004, and zero otherwise
Note: All Compustat data item numbers are provided in square brackets
Table AI.
Variable de?nitions
Real and accrual-
based earnings
77
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Appendix 2. Regression used in two-stage least square analysis
The ?rst stage of the two-stage least square measures the predicted litigation based on various
exogenous variables related to litigation as follows:
ProbðLIT
t
Þ ¼a
1
þb
1
AUD
t
þb
2
Principal
t
þb
3
Secondary
t
þb
4
SIZE
t
þb
5
LEV
t
þb
6
ROA
t
þb
7
LOSS
t
þb
8
LITRISK
t
þb
9
MTB
t
þ
X
j
b
j
IndustryDummies
t
þ1
where:
LIT ¼ an indicator variable that takes on the value of 1 if the SEO is
subsequently sued for accounting irregularities, and 0 if it is not
subsequently sued.
AUD ¼ indicator variable that takes on the value of 1 if the auditor is a
prestigious one (de?ned as Compustat item no. 149 (the auditor
component) between 1 and 8) and 0 otherwise.
Principal ¼ proceeds from offer (offer size)/beginning total assets.
Secondary ¼ secondary shares issued/total shares issued.
SIZE
t
¼ Log (total assets at year end).
LEV
t
¼ leverage ¼ total of short- and long-term debt scaled by average total
assets.
ROA
t
¼ return on assets measured as net income divided by total assets.
MTB
t
¼ market-to-book value measured as market value of assets divided
by book value of assets at year end.
LOSS
t
¼ indicator variable that takes on the value 1 if the ?rm has a net loss,
and 0 otherwise.
LITRISK
t
¼ indicator variable that takes on the value 1 if the ?rm operates in
a high-risk environment, as de?ned in Francis et al. (1994), and
0 otherwise.
Industry Dummies
t
¼ indicator variable that takes on the value of 1 if observation belongs
to a speci?c industry, and 0 otherwise.
The predicted value of litigation probability from the above regression is used as an independent
variable in regressions (3) and (4).
Corresponding author
Salma Ibrahim can be contacted at: [email protected]
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
ARJ
24,1
78
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Richard Kent, James Routledge. 2015. Use of benchmarks in predicting earnings management?. Accounting
& Finance n/a-n/a. [CrossRef]
2. Walter Aerts, Shuyu Zhang. 2014. Management’s causal reasoning on performance and earnings
management. European Management Journal 32, 770-783. [CrossRef]
D
o
w
n
l
o
a
d
e
d
b
y
P
O
N
D
I
C
H
E
R
R
Y
U
N
I
V
E
R
S
I
T
Y
A
t
2
1
:
1
3
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_401618023.pdf