Description
Effective August 23, 2004, the US Securities and Exchange Commission (SEC) requires all
firms to disclose restatements via an item 4.02 Form 8-K filing. However, a significant number of firms
continue to disclose restatements using means other than an 8-K. Commonly referred to as stealth
restatements, the purpose of this paper is to investigate the materiality of restatements disclosed in
either the 10-K or the 10-Q by comparing them to those disclosed via 8-K.
Accounting Research Journal
Do stealth restatements convey material information?
Afshad J . Irani Le (Emily) Xu
Article information:
To cite this document:
Afshad J . Irani Le (Emily) Xu, (2011),"Do stealth restatements convey material information?", Accounting
Research J ournal, Vol. 24 Iss 1 pp. 5 - 22
Permanent link to this document:
http://dx.doi.org/10.1108/10309611111148751
Downloaded on: 24 January 2016, At: 21:12 (PT)
References: this document contains references to 21 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 804 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.
23-49 http://dx.doi.org/10.1108/10309611111148760
Salma Ibrahim, Li Xu, Genese Rogers, (2011),"Real and accrual-based earnings management and its legal
consequences: Evidence from seasoned equity offerings", Accounting Research J ournal, Vol. 24 Iss 1 pp.
50-78 http://dx.doi.org/10.1108/10309611111148779
Magdy Farag, Rafik Elias, (2011),"Relative audit fees and client loyalty in the audit market", Accounting
Research J ournal, Vol. 24 Iss 1 pp. 79-93 http://dx.doi.org/10.1108/10309611111148788
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Do stealth restatements convey
material information?
Afshad J. Irani
Williams School of Commerce, Economics and Politics,
Washington and Lee University, Lexington, Virginia, USA, and
Le (Emily) Xu
Whittemore School of Business and Economics, University of New Hampshire,
Durham, New Hampshire, USA
Abstract
Purpose – Effective August 23, 2004, the US Securities and Exchange Commission (SEC) requires all
?rms to disclose restatements via an item 4.02 Form 8-K ?ling. However, a signi?cant number of ?rms
continue to disclose restatements using means other than an 8-K. Commonly referred to as stealth
restatements, the purpose of this paper is to investigate the materiality of restatements disclosed in
either the 10-K or the 10-Q by comparing them to those disclosed via 8-K.
Design/methodology/approach – Univariate and multivariate analyses compare the
characteristics of and the market reaction to 10-K/10-Q restatements to those of 8-K restatements.
Findings – The authors ?nd stealth restatements are more likely to be those not affecting net income,
with longer ?ling delays, not subject to SEC investigation and made by ?rms audited by non-big four
accounting ?rms. The authors document a negative market reaction to 8-K restatements around the
restatement disclosure date. However, for stealth restatements they ?nd no market reaction around the
10-K/10-Q ?ling date and for up to 22 trading days after the 10-K/10-Q ?lings.
Research limitations/implications – The study shows a signi?cant difference in materiality
between stealth and 8-K restatements.
Practical implications – The study is important to investors, regulators and academics because it
supports the notion that stealth restatements include less signi?cant information relative to that
disclosed in 8-K restatements. This result is in line with the SEC disclosure requirement.
Originality/value – The signi?cant number of stealth restatements since 2004 begs the question as to
what kind of information is being disclosed in these restatements. The paper responds to this question.
Keywords United States of America, Disclosure, Restatements, Stealth, Item 4.02, Form 8-K
Paper type Research paper
I. Introduction
We investigate the attributes of and the market reaction to ?nancial restatements ?rst
disclosed in quarterly or annual reports (10-K/10-Q) without being ?led via item 4.02
Form 8-K (hereafter referred to as 8-K restatements) expressly created by the US
Securities and Exchange Commission (SEC) for this purpose. Commonly referred to as
stealth restatements, this type of restatement has become increasingly common over
the past few years. About 10.7 percent of our restatement sample is stealth and this
percentage has annually increased over our 2005-2007 sample period. We examine
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
The authors thank Steve Balsam, Pieter Elgers, Steve Gill, Erin Moore and seminar participants
at the University of New Hampshire, Washington and Lee University and the Northeast Regional
Meeting of the American Accounting Association for their useful suggestions and insights. All
the data used in this study are available from public sources.
Stealth
restatements
5
Accounting Research Journal
Vol. 24 No. 1, 2011
pp. 5-22
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309611111148751
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the materiality of information released in stealth restatements on two fronts:
restatement attributes and the corresponding market reaction.
Compared to 8-K restatements, we ?nd that ?rms are more likely to use stealth
restatements if:
.
their restatements do not affect previously reported net income;
.
there is a longer lag between the end of the restatement period and the
restatement disclosure date; and
.
the restatement is not a result of an SEC investigation.
We also ?nd stealth restatements to be the likely disclosure venue for ?rms not
changing auditors between the end of the restatement period and the restatement ?ling
date, employing non-big four auditors and with higher book-to-market ratios.
In analyzing stock price reaction to restatement announcements, we ?nd stealth
restatements garner no signi?cant market reaction on the day of and the day following
the SEC ?ling date. On the other hand, 8-K restatements earn a 21.61 percent
(signi?cant) market-adjusted return over the day of and the day following their ?ling
date. This difference in market reaction between the two disclosure venues holds in a
multivariate setting after controlling for various auditor and ?rm attributes. In
addition, there is no long-term (up to 22 trading days or one calendar month) market
reaction to stealth restatements. Our market reaction ?ndings are consistent with our
aforementioned ?nding of stealth restatements possessing less severe restatement
attributes than 8-K restatements.
Pursuant to item 4.02 effective August 23, 2004, the SEC requires a company to ?le
Form 8-K under item 4.02(a) if and when the ?rm itself concludes that its previously
issued ?nancial statements can no longer be relied upon because of an error, or under
item4.02(b) if the company is advised by its independent accountant of the error[1]. The
?rm then has four business days to ?le an 8-K in which they need to disclose speci?c
information[2], [3]. However, not every revision to a company’s prior ?nancial
statements that results froman error requires an 8-Krestatement. This was the response
given by the SEC staff in their September 13, 2004 meeting with the AICPA’s SEC
Regulations Committee to a speci?c inquiry as to whether all restatements need to be
announced in an 8-K[4]. The SEC staff agreed with the Committee’s view that an 8-K
restatement need not be ?led if the ?rmand its independent accountant conclude that the
error is immaterial (based on a SAB 99 materiality assessment) and that the prior
?nancial statements may continue to be relied on. This position taken by the SEC
coupled with ?rm management’s possible desire to be less transparent when it comes to
disclosing restatements are what we believe to be plausible reasons for the continued
increase in the usage of stealth restatements. Results from our analysis of disclosure
venue attributes as well as those fromour stock market reaction tests are consistent with
the aforementioned materiality argument in that stealth restatements are not material.
We contribute to the restatement literature in a number of ways. To our knowledge,
this is one of the ?rst studies to explicitly examine stealth restatements and investigate
whether ?rms are following SEC guidelines for disclosing restatements. Questions
have repeatedly surfaced in the popular press regarding what stealth restatements
correct and whether that information is value relevant. We explicitly respond to those
concerns by examining stealth restatement characteristics and the corresponding
market reaction. Finally, our results documenting signi?cant differences between
ARJ
24,1
6
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the two disclosure venues should be of value to future restatement research in that
these two restatement venues should be explicitly controlled for in the empirical
analysis.
The rest of the paper is organized as follows. We provide our motivation and review
of related research in Section II followed by the research design in Section III. Section IV
presents descriptive statistics and our empirical ?ndings. Concluding remarks are
offered in Section V.
II. Motivation and related research
Restatements resulting from accounting errors are typically perceived as negative
signals regardless of their impact on the ?rm’s book value. In addition to suggesting
that past ?nancial statements can no longer be relied upon, they are indicative of a
weak accounting information system that can possibly lead to future problems.
Academic research focusing on ?nancial restatements has investigated their impact on
restating ?rm’s market value (Palmrose et al., 2004; Callen et al., 2006) as well as on
peer ?rm’s market value (Gleason et al., 2008) and found them to have a negative
impact on both. Wilson (2008) ?nds a reduction in the information content of restating
?rm’s earnings following a restatement. Restatements have also been found to lead to
turnover and signi?cant labor market penalties for both top management (Desai et al.,
2006) and ?rm directors (Srinivasan, 2005). Finally, restatements have been shown to
have a negative impact on analyst following (Grif?n, 2003) and transient institutional
investor holdings (Hribar et al., 2005)[5].
The number of ?nancial restatements has been on a rise since the passage of the
Sarbanes-Oxley Act (SOX) in 2002. Turner and Weirich (2006) report 1,323 restatements
?led by US public companies during 1997-2002. From2003 to 2005 that number grewto
2,319, out of which 320 were by companies making at least a second restatement. The
number of restatements is expected to stay at this level and perhaps even rise as small
companies (public ?oat of less than $75 million) begin fully complying with SOX
Section 404 provisions for ?scal years ending on or after December 15, 2009.
The increase in the number of restatements resulting fromaccounting errors has been
accompanied by ?rms’ increasing usage of alternative “less revealing” disclosure venues
possiblyinhopes of attractingminimal public attention. Eventhoughbelievers inef?cient
markets mayclaimthat disclosinginformationin“less revealing” venues still leads to that
information being incorporated in security prices, one can safely argue that such a
practice does not provide a level playing ?eld for all investors (especially small investors).
Item4.02 Form8-Kdisclosure, required starting August 23, 2004, was the SEC’s response
to combat such practices and level the playing ?eld by providing a uniform disclosure
venue for all restatements. According to the SEC (2004), the required disclosures:
[. . .] will bene?t markets by increasing the number of unquestionably or presumptively
material events that must be disclosed currently. They will also provide investors with better
and more timely disclosure of important corporate events.
The SEC went on to say that these amendments also further the goals of SOX Section
409 that requires companies to disclose on a “rapid and current” basis, information that
is necessary for the protection of investors and in the public interest (Prentice, 2005).
Contrary to the requirements outlined in the aforementioned SEC ruling but
consistent with the subsequent response by the SEC staff that immaterial restatements
Stealth
restatements
7
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
do not need to be ?led via item 4.02 Form 8-K disclosure, a summer 2006 report by the
Government Accountability Of?ce found that about 17 percent of public companies
restating results between August 2004 and September 2005 did not ?le it using an 8-K.
Similarly, according to a 2006 report by Glass, Lewis & Co., 14 percent of restatements
in 2005 are stealth compared to only 2 percent a year ago. In response, the SEC sent
letters to companies whose periodic ?lings (10-K/10-Q) were ?agged for containing
material restated numbers insisting that they disclose those restatements using Form
8-K. As anecdotal evidence, Reilly (2006) reports that Sun Microsystems, Inc. included
its restatement in the 10-K ?ling but later re-?led Form 8-K for the same restatement to
achieve “maximum transparency.” Similarly, Inter-Tel, Inc. had also used 10-K to
disclose a restatement. However, following comments from the SEC the company later
?led 8-K for the same restatement. These anecdotes suggest that stealth restatements
continue to exist despite the SEC’s effort to combat such practices.
The motivation behind this study is to seek a better understanding of the kind of
information that is disclosed in stealth restatements. If we ?nd that stealth restatements
only correct immaterial errors, then a ?rm’s decision to disclose themin 10-K’s/10-Q’s to
avoid negative publicity associated with restatements in general is understandable and
consistent with SEC guidance. Such a result would also negate the economic relevance
of the ?ndings reported in the 2006 GAO and Glass, Lewis & Co. reports.
However, if we ?nd correction of material errors in stealth restatements, then such
disclosures may be an indication of a ?rm’s desire to keep it hidden and avoid the
negative consequences associated with the value relevance of information disclosed in
those restatements[6]. In turn, this will create distrust in the market and an uneven
playing ?eld especially for small investors, who unlike institutional investors do not
have the time and/or the ability to discover and appropriately react to such disclosures.
In addition, stealth ?rms in the future may be subjected to SEC investigation and legal
action.
We conduct our investigation by examining both the attributes of stealth
restatements as well as the stock market reaction to their disclosure. Examining
attributes will provide an understanding of the kind of errors being disclosed in stealth
restatements, while the test of the market reaction to stealth restatements will re?ect on
the value relevance of the disclosed errors. We believe both these tests are
complimentary and necessary to re?ect on the materiality of information disclosed in
stealth restatements.
Several recent academic papers have investigated the prominence of the
restatement announcement within alternative disclosure venues. Using data from
1997-2002, Files et al. (2009) divide restatements disclosed in press releases into three
levels of prominence: high (disclosed in the headline), medium (described in the body,
headline is on a different subject) and low (change the comparative period amounts in
an earnings release with no direct mention of the restatement). After controlling for the
severity of the restatement, Files et al. ?nd the smallest negative market reaction
around the press release date (21, þ 1) to low prominence restatement disclosures. In
a related study using a similar sample period, Gordon et al. (2007) investigate the
impact of disclosure credibility on market reaction to restatements. They de?ne
disclosure credibility as “investors’ perceptions of the believability of a particular
disclosure” and measure it across two dimensions: ?rm management and restatement
attributes (disclosure prominence being one of the restatement attributes). They ?nd
ARJ
24,1
8
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
that more prominent restatement announcements (both in terms of inclusion in a
headline or being stand alone) garner a more negative stock price reaction. Note that
the sample periods used in both Files et al. (2009) and Gordon et al. (2007) predate the
item 4.02 Form 8-K disclosure requirement. In addition, both studies do not consider
stealth restatements, which have even lower prominence. Finally, unlike our study,
neither examines restatement attributes.
In a concurrent study, using data from August 24, 2004 to December 31, 2005, Sharp
(2007) investigates whether ?rms strategically manage the timing and transparency of
restatement disclosures. In addition to restatements disclosed in an 8-K (no obscurity)
and 10-K/10-Q (high obscurity/stealth restatements), the paper also includes
restatements disclosed in amended ?lings (10-K/A or 10-Q/A) which it identi?es as
those with medium obscurity. The results most relevant to our study include:
.
a positive association between usage of an 8-K and restatement attributes such
as negative and (small) positive earnings surprises, longer restated periods,
material weakness in internal control and more timely disclosure; and
.
for stealth restatements, less negative market reaction in the short run (22, þ 2)
and no market reaction in the long run (22, þ 20) after controlling for other
restatement attributes.
Our study differs from Sharp (2007) in two signi?cant ways. First, our study
incorporates a longer sample period and a signi?cantly larger ?rm sample (even
without including amended ?lings). This is relevant in terms of improving external
validity of the results, especially given the SEC’s push in 2006 to reduce/eliminate
stealth disclosures involving material restatements and an increase in the percentage
of stealth restatements as indicated by our sample. Second, the multivariate models
used in our study include important variables different than those used in Sharp (2007),
such as ?rms subject to an SEC investigation, auditor attributes and revenue
recognition errors.
III. Research design
1. Sample selection
We collect US ?nancial restatements resulting from accounting errors disclosed during
2005-2007 from Audit Analytics with restatement disclosure venues of 8-K, 10-K and
10-Q. Our sample period begins in 2005 since that is the ?rst full calendar year
following the effective date (August 23, 2004) of the item 4.02 Form 8-K disclosure
requirement. We adopt a conservative approach by not including restatements ?led
during 2004 in case those ?rms were either not familiar with the new SEC ruling or
chose to ignore it given the proximity of their ?ling date to the ruling’s effective date.
The initial sample comprises of 1,110 restatements. We lose 111 observations due to
missing ?rm data on COMPUSTAT and CRSP. In addition, we delete 35 observations
since their restatement period end dates are after their restatement ?ling dates. That
yields a preliminary sample of 964 restatements out of which 123 are stealth. To
con?rm the coding used by Audit Analytics, we then checked the 10-K/10-Q’s of each
of the 123 stealth restatements for any mention of a prior or simultaneous 8-K
disclosure. We also conducted a search on Lexis-Nexis Academic Universe on the days
following the 10-K/10-Q to make sure that there was no follow-up 8-K disclosure. We
found that 17 stealth restatements were accompanied by simultaneous 8-K disclosures
Stealth
restatements
9
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and three had 8-K disclosures within four business days following the 10-K/10-Q. We
reclassi?ed the 17 as 8-K restatements and dropped the three as we were unclear about
management intent. This yields our ?nal sample of 961 restatements disclosed by 789
?rms, out of which 103 are stealth restatements[7], [8].
2. Empirical model: determinants of disclosure venue
To investigate attributes of ?nancial restatement disclosures, we estimate the
following year ?xed effects logistic model:
Disc_Ven ¼ a
0
þa
1
Neg_Dir þa
2
Pos_Dir þa
3
Rev_Recog þa
4
Res_Yrs
þa
5
SEC_Inv þa
6
Filing_Delay þa
7
ACC þa
8
AC þa
9
B4
þa
10
Size_1 þa
11
B=M_1 þa
12
DumROE_1 þa
13
DumRet_1
þa
14
Dum_Prior8K þa
15
Dum_PriorSt þ1
ð1Þ
where:
Disc_Ven ¼ 1 for stealth restatements, 0 for 8-K restatements.
Neg_Dir ¼ 1 if the restatement negatively impacts net income, 0 otherwise.
Pos_Dir ¼ 1 if the restatement positively impacts net income, 0 otherwise.
Rev_Recog ¼ 1 if it is a revenue recognition restatement, 0 otherwise.
Res_Yrs ¼ Number of restated years.
SEC_Inv ¼ 1 if the restatement was under SEC investigation, 0 otherwise.
Filing_Delay ¼ Number of days between the end of the restatement period and
the restatement disclosure date divided by 365.
ACC ¼ 1 if the restatement is related to lease accounting, stock-based
compensation (SFAS 123 (R)) or hedge accounting, 0 otherwise.
AC ¼ 1 if the auditor at the end of the restatement period is different
from the auditor at the restatement disclosure date, 0 otherwise.
B4 ¼ 1 if the auditor on the restatement disclosure date is one of the big
four accounting ?rms, 0 otherwise.
Size_1 ¼ Log of market value of equity at the end of the ?scal year prior to
the restatement disclosure date, in millions of dollars.
B/M_1 ¼ Book value of equity divided by market value of equity at the end
of the ?scal year prior to the restatement disclosure date.
DumROE_1 ¼ 1 if the restating ?rm’s ROE in the year prior to the restatement
disclosure date is within the top or bottom three deciles,
0 otherwise. ROE is de?ned as income before extraordinary items
divided by market value of equity.
DumRet_1 ¼ 1 if the restating ?rm’s compounded annual return for the ?scal
year prior to the restatement disclosure date is within the top or
bottom three deciles, 0 otherwise.
ARJ
24,1
10
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Dum_Prior8K ¼ 1 if the restating ?rm had a previous 8-K restatement since year
2000, 0 otherwise.
Dum_PriorSt ¼ 1 if the restating ?rm had a previous stealth restatement since
year 2000, 0 otherwise.
We use Neg_Dir, Pos_Dir, Rev_Recog, Res_Yrs and SEC_Inv to proxy for the
materiality of the restatement. Following Palmrose et al. (2004) and Wilson (2008), we
consider restatements affecting net income (Neg_Dir and Pos_Dir), involving revenue
recognition errors (Rev_Recog), affecting longer time periods (Res_Yrs), and under SEC
investigation (SEC_Inv) to be more material. If ?rms do in fact follow the SEC
requirement and disclose all material restatements using Form 8-K, then we expect a
negative sign on all these variables. While ?rms can ?le an 8-K on any business day,
?rms with stealth restatements wait until their next annual or quarterly ?ling (10-K or
10-Q) to include the restatement information. Therefore, stealth restatements are
expected to have longer ?ling delays and consequently we expect a positive coef?cient
on Filing_Delay.
We include ACC to address three speci?c accounting issues related to leasing,
compensation and hedging addressed by recent restatements. In 2005, the SEC issued
an open letter clarifying the correct way to account for leases and that led to a
signi?cant number of restatements (Branson and Pagach, 2005). Following the highly
publicized hedge accounting misconduct at Fannie Mae, a lot of companies reassessed
their practice of recording hedges, thereby resulting in a signi?cant number of
restatements. Finally, following SFAS 123(R) taking effect in 2005, many companies
restated their ?nancial statements to correct errors related to share-based payment
computations. We include these three accounting issues in one (control) variable
because all three accounting issues affected a signi?cant number of ?rms and received
much public attention. Given the publicity related to these issues, we expect affected
?rms to have used 8-K restatements (i.e. a negative a
7
).
In addition to SEC_Inv, we include two monitoring variables related to auditor
attributes, AC and B4. AC captures the change in auditors from the end of the
restatement period to the restatement disclosure date. We expect its coef?cient to be
negative because a different auditor at the time of disclosure does not have any
incentives to hide restatements that occurred during the tenure of the previous auditor.
Therefore, an 8-K is more likely to be used as the disclosure venue. We also expect the
coef?cient on B4 to be negative. Big four accounting ?rms are generally considered to
be higher quality auditors (Beatty, 1989; Teoh and Wong, 1993) and therefore more
likely to require their clients to ?le an 8-K restatement.
Lastly, we included six ?rm-speci?c attributes. Since bigger ?rms are under more
public scrutiny, we expect these ?rms to disclose 8-K restatements and thus the
predictive sign for the coef?cient on Size_1 is negative. We do not predict signs for
coef?cients of B/M_1, DUMROE_1 and DUMRET_1. A ?rm with higher (lower) book
to market indicates that it is a value (growth) ?rm. DUMROE_1 and DUMRET_1
indicate whether a ?rm performed very well or very poorly relative to its peers, as
measured by their return on equity and stock market performance at the end of the
?scal year prior to their restatement disclosure date. A growth ?rm and/or a ?rm that
has performed better or worse than its peers may have incentives to disguise their
restatement due to the possible economic loss, but they may also want to be more
Stealth
restatements
11
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
transparent because they are more visible and do not want to lose their credibility in
the eyes of the market participants.
We include Dum_Prior8K and Dum_PriorSt to capture the existence of prior 8-K
and stealth restatements, respectively, since the beginning of 2000[9]. We expect ?rms
with prior 8-Krestatements to continue to use an 8-Kfor their current restatement ?ling
(a
14
, 0). Firms that have made prior stealth restatements may also follow their past
practice which would indicate a positive coef?cient on Dum_PriorSt. However, their
prior restatements may have been ?led prior to August 23, 2004 at which time there
was no explicit SEC requirement for using an 8-K. Consequently, for restatements ?led
in 2005 onwards these same ?rms may use the required item 4.02 Form 8-K disclosure
venue. In addition, these ?rms may have experienced signi?cant economic and
litigation costs related to their prior stealth restatement that would prevent them from
repeating their past practice. Consequently, we do not predict a sign on the coef?cient
of Dum_PriorSt.
3. Empirical model: multivariate analysis of market reaction to restatements
We use the following year ?xed effects regression model to investigate the market
reaction:
BHMAR ð0; 1Þ ¼ b
0
þb
1
Disc_Ven þb
2
AC þb
3
B4 þb
4
AQ_INC
þb
5
AQ_DEC þb
6
Size_1 þb
7
B=M_1 þb
8
Dum_Prior8k
þb
9
Dum_PriorSt þp
ð2Þ
where:
BHMAR (0, 1) ¼ Two-day buy and hold market-adjusted returns.
AQ_Inc ¼ 1 if the auditor at the end of restatement period is one of the
non-big four accounting ?rms and the auditor at the time of
restatement disclosure is one of the big four accounting ?rms,
0 otherwise.
AQ_Dec ¼ 1 if the auditor at the end of restatement period is one of the big
four accounting ?rms and the auditor at the time of restatement
disclosure is one of the non-big four accounting ?rms,
0 otherwise.
and other independent variables are as previously de?ned in Section III.2.
Disc_Ven is our variable of interest as it indicates whether the short-run market
reaction differs between the two restatement disclosure venues after controlling for
auditor and ?rm attributes. Based on prior literature (Palmrose et al., 2004; Callen et al.,
2006), the market reacts negatively when ?rms announce restatements. These studies
have primarily focused on restatements that were announced publicly. Therefore,
these results are related to 8-K restatements in our study. In comparison, stealth
restatements are disclosed quietly through regular SEC ?lings. If these stealth
restatements contain immaterial information, then we expect a less negative (or zero)
market reaction. Consequently, we expect b
1
to be positive.
The next four variables focus on auditor attributes. When the auditor at the
restatement disclosure date is different from the auditor at the end of the
ARJ
24,1
12
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
restatement period, the market may lend more credibility to the announced restatement
and thus react less negatively. On the other hand, the market may attribute the
accounting irregularities (being addressed by the restatement) as one of the reasons
behind the auditor change and react even more negatively due to increased
uncertainty. Consequently, we do not predict a sign on the coef?cient of AC. We expect
the coef?cient on B4 to be positive as big four auditors are perceived to be of higher
quality. Consequently, the market may consider ?rms with big four auditors to have a
better ?nancial reporting system and may feel less uncertain about the scope and
impact of the restatement. We expect coef?cients on AQ_INC (AQ_DEC) to also be
positive (negative) since an auditor change from non-big four (big four) to big four
(non-big four) is perceived as an increase (decrease) in the auditor quality.
We expect negative coef?cients on Dum_Prior8k and Dum_PriorSt. An existing
restatement coupled with at least one previous restatement (8-K or stealth) may signal
continuing problems with the ?rm’s ?nancial reporting system and thus cause the
market to react more negatively to the current restatement. Finally, we include Size_1
and B/M_1 as control variables[10].
IV. Empirical results
1. Descriptive statistics
Table I panel A presents the number of restatements by restatement and auditor
attributes for the two restatement disclosure venues. Our sample consists of 858 8-K
restatements and 103 stealth restatements. The percentage of stealth restatements has
steadily increased from 6.9 (25/361) in 2005 to 11.4 (42/367) in 2006 to 15.5 (36/233) in
2007[11]. Relative to 8-K restatements, a lower percentage of stealth restatements
impact net income (45 vs 77 percent for 8-K), are a result of revenue recognition errors
(12 vs 15 percent for 8-K), and are involved in a SEC investigation (5 vs 14 percent for
8-K). Table II panel B shows that 2,414 (245) corrections were included in the 858 (103)
8-K (stealth) restatements which yields an average of 2.8 (2.4) corrections per 8-K
(stealth) restatement. Roughly, one-fourth of errors disclosed in stealth restatements
are corrections of classi?cation issues. These ?ndings are consistent with the SEC’s
position requiring item 4.02 Form 8-K disclosures for material restatements.
In addition, consistent with our expectation, Table I panel A shows proportionately
more 8-K restatements related to lease, compensation and hedging activities (ACC).
Also as expected, proportionately less ?rms with stealth restatements use big four
accounting ?rms as their auditors (67 vs 79 percent for 8-K). Both disclosure venues
have a similar distribution across industries, with industry de?ned using the GICS
two-digit industry classi?cation codes (Bhojraj et al., 2003)[12].
Table III panel C provides descriptive statistics for restatement and ?rm attributes
across the two disclosure venues. Tests of signi?cance for the difference in the means
and medians of the corresponding variables (using t- and Wilcoxon statistics,
respectively) suggest that stealth restatements have signi?cantly less negative impact
on net income (median only) and less number of restated years[13]. These results are
consistent with the ?ndings in panels A and B that stealth restatements appear to be
less severe than 8-K restatements. In addition, panel C shows that stealth restatements
have signi?cantly longer ?ling delays and higher book-to-market ratios.
Table IV presents the Pearson and Spearman correlation coef?cients for the
exogenous variables used in our study. Most of the signi?cant correlations found
Stealth
restatements
13
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
in the table were expected. By construction, Neg_Dir and Pos_Dir are negatively
correlated (20.508). AQ_DEC and AC are also highly correlated because there is only
one observation in our sample that has increased auditor quality. So, most of the
sample ?rms that have changed auditors either have decreased their auditor quality or
remained at the same level. In addition, B4 is highly correlated with Size_1 (0.502).
This is not surprising since larger ?rms are more likely to be audited by one of the big
four accounting ?rms. Furthermore, we test for the potential effect of multicollinearity
on standard errors both by examining variance in?ation factors as well as running
step-wise regressions and ?nd our results to be unaffected.
2. Attributes of restatement disclosure venue
Table V presents the empirical results for the attributes of restatement disclosure
venue (model (1)). The logistic model generates a pseudo-R
2
of 14.21 percent. Signs on
most coef?cient estimates are consistent with our expectations. a
1
and a
2
are negative
and signi?cant at the 1 percent level suggesting that restatements impacting
Number of restatements
a
Total 8-K Stealth
Full sample 961 858 103
2005 361 336 (39%) 25 (24%)
2006 367 325 (38%) 42 (41%)
2007 233 197 (23%) 36 (35%)
Restatement attributes
Neg_Dir 562 524 (61%) 38 (37%)
Pos_Dir 149 141 (16%) 8 (8%)
Rev_Recog 141 129 (15%) 12 (12%)
Sec_Inv 121 116 (14%) 5 (5%)
Dum_Prior8K 154 136 (16%) 18 (17%)
Dum_PriorSt 331 292 (34%) 39 (38%)
ACC 225 212 (25%) 13 (13%)
Auditor attributes
AC 50 44 (5%) 6 (6%)
B4 744 675 (79%) 69 (67%)
AQ_Inc 1 1 (0.1%) 0 (0%)
AQ_Dec 19 15 (2%) 4 (4%)
GICS two-digit industry classi?cation
Energy 52 46 (5%) 6 (6%)
Materials 39 38 (4%) 1 (1%)
Industrials 110 98 (11%) 12 (12%)
Consumer discretionary 248 215 (25%) 33 (32%)
Consumer staples 24 20 (2%) 4 (4%)
Health care 99 88 (10%) 11 (11%)
Financials 134 125 (15%) 9 (9%)
Information technology 208 182 (21%) 26 (25%)
Telecommunication services 26 26 (3%) 0 (0%)
Utilities 21 20 (2%) 1 (1%)
Notes: n ¼ 961;
a
% in parenthesis indicates the proportion of regular (858) or stealth (103)
restatements for that speci?c restatement or auditor attribute; for example, 61 percent (524/858) of all
8-K restatements have negative impact on net income
Table I.
Descriptive statistics –
panel A: number of
restatements by
restatement attributes
and auditor attributes for
different disclosure
venues
ARJ
24,1
14
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
8-K restatements Stealth restatements
Number Percentage Number Percentage
PPE, intangible or ?xed asset issues and related
depreciation, depletion or amortization errors 217 9 7 3
Accounts/loans receivables, investment and cash issues 68 3 18 7
Deferred stock-based compensation 153 6 9 4
Liabilities, payables, reserves and accrual estimate
failures 110 5 18 7
Lease, SFAS 5, legal, contingency and commitment
issues 152 6 7 3
Revenue recognition issues 129 5 12 5
Expense (cost of goods sold, payroll, SGA, other)
recording issues and capitalization of expenditures 269 11 18 7
Tax expense/bene?t/deferral/other (FAS109) issue 156 6 14 6
Acquisitions, mergers, disposals and re-org acct issues 105 4 15 6
Classi?cation issues on ?nancial statements 179 7 58 24
Other 876 36 69 28
Total 2,414 100 245 100
Table II.
Descriptive
statistics – panel B:
number of restatements
by restatement
accounts for different
disclosure venues
Mean Median
8-K
(n ¼ 858)
Stealth
(n ¼ 103)
8-K
(n ¼ 858)
Stealth
(n ¼ 103)
Restatement attributes
Negative change in net income per restated year
divided by book value of equity at the beginning
of the restatement period 20.088 20.040 20.021
*
20.009
Positive change in net income per restated year
divided by book value of equity at the beginning
of the restatement period 0.041
*
0.011 0.013 0.006
Res_Yrs 2.871
* *
2.013 2.750
* *
1.500
Filing_Delay 0.460
* *
0.909 0.367
* *
1.093
Firm attributes
Market value of equity at the end of the ?scal year
prior to the restatement disclosure date, in millions
of dollars 3,235 1,833 537 345
B/M_1 0.474
*
0.557 0.448
* *
0.525
Income before extraordinary items divided by market
value of equity at the end of the ?scal year prior to the
restatement disclosure date 20.018 20.029 0.030 0.020
Compounded annual return for the ?scal year prior
to restatement disclosure date 0.168 0.089 0.079 0.015
Notes:
*
,
* *
indicates whether the restatement and ?rm attributes are signi?cantly different between
regular and stealth restatements at the 5 and 1 percent signi?cance level; statistical signi?cance is
based on the t- and Wilcoxon tests for the means and medians, respectively:
Disc_Ven ¼ a
0
þa
1
Neg_Dir þa
2
Pos_Dir þa
3
Rev_Recog þa
4
Res_Yrs þa
5
SEC_Inv
þa
6
Filing_Delay þa
7
ACC þa
8
AC þa
9
B4 þa
10
Size_1 þa
11
B=M_1
þa
12
DumROE_1 þa
13
DumRet_1 þa
14
Dum_Prior8K þa
15
Dum_PriorSt þ1
ð1Þ
Table III.
Descriptive statistics –
panel C: descriptive
statistics for restatement
and ?rm attributes by
disclosure venue
Stealth
restatements
15
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
N
e
g
_
D
i
r
P
o
s
_
D
i
r
R
e
v
_
R
e
c
o
g
R
e
s
_
Y
r
s
S
E
C
_
I
n
v
F
i
l
i
n
g
_
D
e
l
a
y
D
u
m
_
P
r
i
o
r
8
k
D
u
m
_
P
r
i
o
r
S
t
A
C
C
A
C
B
4
A
Q
_
I
n
c
A
Q
_
D
e
c
S
i
z
e
_
1
B
/
M
_
1
R
O
E
_
1
R
e
t
_
1
N
e
g
_
D
i
r
1
2
0
.
5
0
8
*
*
0
.
0
9
9
*
*
0
.
2
0
8
*
*
0
.
0
4
6
2
0
.
1
0
1
*
*
2
0
.
0
1
2
2
0
.
0
0
7
0
.
1
5
2
*
*
2
0
.
0
4
0
0
.
0
2
0
2
0
.
0
3
8
2
0
.
0
6
2
0
.
0
5
5
2
0
.
0
7
4
*
2
0
.
0
2
6
0
.
0
0
8
P
o
s
_
D
i
r
2
0
.
5
0
8
*
*
1
2
0
.
0
5
6
2
0
.
0
3
4
2
0
.
0
5
0
2
0
.
0
2
8
0
.
0
1
7
2
0
.
0
0
2
0
.
0
8
9
*
0
.
0
0
3
0
.
0
0
4
2
0
.
0
1
4
0
.
0
2
2
0
.
0
3
2
0
.
0
3
2
2
0
.
0
3
3
2
0
.
0
0
5
R
e
v
_
R
e
c
o
g
0
.
0
9
9
*
*
2
0
.
0
5
6
1
2
0
.
0
5
3
0
.
0
1
1
2
0
.
0
0
1
0
.
0
4
3
0
.
0
7
1
*
2
0
.
0
7
6
*
2
0
.
0
0
4
0
.
0
3
4
2
0
.
0
1
3
0
.
0
0
4
0
.
0
0
4
2
0
.
0
2
8
2
0
.
0
5
0
2
0
.
0
6
4
*
R
e
s
_
Y
r
s
0
.
2
1
0
*
*
2
0
.
0
5
7
2
0
.
0
5
4
1
0
.
2
3
9
*
*
2
0
.
1
4
4
*
*
2
0
.
0
4
0
2
0
.
0
3
6
0
.
1
7
9
*
*
2
0
.
0
9
3
*
*
0
.
2
3
6
*
*
2
0
.
0
4
5
2
0
.
0
2
8
0
.
2
8
3
*
*
2
0
.
0
2
3
0
.
1
2
4
*
*
0
.
0
8
0
*
S
E
C
_
I
n
v
0
.
0
4
6
2
0
.
0
5
0
0
.
0
1
1
0
.
2
4
8
*
*
1
2
0
.
1
3
1
*
*
2
0
.
0
2
0
2
0
.
0
1
8
2
0
.
0
6
9
*
2
0
.
0
0
4
0
.
0
1
0
2
0
.
0
1
2
2
0
.
0
3
1
0
.
1
2
4
*
*
0
.
0
0
9
2
0
.
0
1
4
2
0
.
0
3
1
F
i
l
i
n
g
_
D
e
l
a
y
2
0
.
1
2
5
*
*
2
0
.
0
5
5
2
0
.
0
2
1
2
0
.
0
9
1
*
*
2
0
.
0
4
9
1
2
0
.
0
2
1
0
.
0
3
3
2
0
.
0
8
6
*
*
0
.
1
3
8
*
*
2
0
.
0
6
1
2
0
.
0
1
4
0
.
1
1
5
*
*
2
0
.
0
8
2
*
0
.
0
1
0
2
0
.
0
5
2
2
0
.
0
2
5
D
u
m
_
P
r
i
o
r
8
K
2
0
.
0
1
2
0
.
0
1
7
0
.
0
4
3
2
0
.
0
1
3
2
0
.
0
2
0
2
0
.
0
0
1
1
0
.
6
0
3
*
*
2
0
.
0
2
7
0
.
1
1
5
*
*
0
.
0
1
2
2
0
.
0
1
4
0
.
0
4
0
2
0
.
0
0
7
0
.
0
4
5
2
0
.
0
2
2
0
.
0
4
2
D
u
m
_
P
r
i
o
r
S
t
2
0
.
0
0
7
2
0
.
0
0
2
0
.
0
7
1
*
2
0
.
0
0
8
2
0
.
0
1
8
0
.
0
2
9
0
.
6
0
3
*
*
1
2
0
.
0
6
5
*
0
.
1
3
6
*
*
0
.
0
2
5
0
.
0
4
5
0
.
0
2
3
0
.
0
3
8
2
0
.
0
0
3
2
0
.
0
9
6
*
*
0
.
0
1
6
A
C
C
0
.
1
5
2
*
*
0
.
0
8
9
*
*
2
0
.
0
7
6
*
0
.
0
7
1
*
2
0
.
0
6
9
*
2
0
.
1
1
0
*
*
2
0
.
0
2
7
2
0
.
0
6
5
*
1
2
0
.
0
3
0
0
.
1
2
8
*
*
2
0
.
0
1
8
0
.
0
1
0
0
.
1
6
9
*
*
0
.
0
1
0
0
.
1
4
7
*
*
0
.
0
9
4
*
*
A
C
2
0
.
0
4
0
0
.
0
0
3
2
0
.
0
0
4
2
0
.
0
9
3
*
*
2
0
.
0
0
4
0
.
1
3
7
*
*
0
.
1
1
5
*
*
0
.
1
3
6
*
*
2
0
.
0
3
0
1
2
0
.
2
6
6
*
*
0
.
1
3
8
*
*
0
.
6
0
6
*
*
2
0
.
0
9
9
*
*
2
0
.
0
1
2
2
0
.
0
2
5
2
0
.
0
3
1
B
4
0
.
0
2
0
0
.
0
0
4
0
.
0
3
4
0
.
1
7
4
*
*
0
.
0
1
0
2
0
.
0
3
2
0
.
0
1
2
0
.
0
2
5
0
.
1
2
8
*
*
2
0
.
2
6
6
*
*
1
0
.
0
1
7
2
0
.
2
6
3
*
*
0
.
5
0
6
*
*
0
.
0
2
4
0
.
1
5
9
*
*
0
.
0
8
2
*
*
A
Q
_
I
n
c
2
0
.
0
3
8
2
0
.
0
1
4
2
0
.
0
1
3
2
0
.
0
3
0
2
0
.
0
1
2
2
0
.
0
1
3
2
0
.
0
1
4
0
.
0
4
5
2
0
.
0
1
8
0
.
1
3
8
*
*
0
.
0
1
7
1
2
0
.
0
0
5
2
0
.
0
1
1
2
0
.
0
0
3
2
0
.
0
2
8
2
0
.
0
1
5
A
Q
_
D
e
c
2
0
.
0
6
2
0
.
0
2
2
0
.
0
0
4
2
0
.
0
4
4
2
0
.
0
3
1
0
.
1
1
5
*
*
0
.
0
4
0
0
.
0
2
3
0
.
0
1
0
0
.
6
0
6
*
*
2
0
.
2
6
3
*
*
2
0
.
0
0
5
1
2
0
.
0
6
1
2
0
.
0
1
7
2
0
.
0
2
0
2
0
.
0
2
1
S
i
z
e
_
1
0
.
0
5
7
0
.
0
3
4
0
.
0
0
4
0
.
2
4
1
*
*
0
.
1
1
7
*
*
2
0
.
0
6
6
*
2
0
.
0
1
1
0
.
0
5
0
0
.
1
7
2
*
*
2
0
.
0
9
6
*
*
0
.
5
0
2
*
*
2
0
.
0
0
8
2
0
.
0
5
8
1
2
0
.
1
4
8
*
*
0
.
2
6
4
*
*
0
.
2
1
0
*
*
B
/
M
_
1
2
0
.
0
4
9
0
.
0
1
8
2
0
.
0
1
2
2
0
.
0
3
7
0
.
0
4
4
0
.
0
1
8
0
.
0
2
8
2
0
.
0
0
8
0
.
0
1
1
2
0
.
0
1
4
0
.
0
3
3
2
0
.
0
0
4
2
0
.
0
1
4
2
0
.
1
2
3
*
*
1
0
.
1
0
2
*
*
2
0
.
2
4
3
*
*
R
O
E
_
1
2
0
.
0
0
8
0
.
0
3
5
0
.
0
1
6
0
.
1
0
1
*
*
0
.
0
0
4
2
0
.
0
2
8
2
0
.
0
0
6
2
0
.
0
7
6
*
0
.
1
0
4
*
*
2
0
.
0
1
7
0
.
1
5
8
*
*
0
.
0
0
0
2
0
.
0
2
9
0
.
2
4
8
*
*
0
.
1
7
8
*
*
1
0
.
3
6
8
*
*
R
e
t
_
1
0
.
0
1
9
2
0
.
0
1
7
2
0
.
0
6
0
*
0
.
0
2
0
2
0
.
0
4
2
2
0
.
0
4
7
0
.
0
1
9
0
.
0
0
3
0
.
0
5
7
0
.
0
0
8
0
.
0
1
2
2
0
.
0
1
3
0
.
0
3
3
0
.
0
8
7
*
*
2
0
.
2
0
1
*
*
0
.
1
7
0
*
*
1
N
o
t
e
s
:
n
¼
9
6
1
;
s
i
g
n
i
?
c
a
n
c
e
a
t
:
*
5
a
n
d
*
*
1
p
e
r
c
e
n
t
(
t
w
o
-
t
a
i
l
e
d
)
;
P
e
a
r
s
o
n
(
S
p
e
a
r
m
a
n
)
c
o
r
r
e
l
a
t
i
o
n
c
o
e
f
?
c
i
e
n
t
s
a
r
e
p
r
e
s
e
n
t
e
d
b
e
l
o
w
(
a
b
o
v
e
)
t
h
e
d
i
a
g
o
n
a
l
Table IV.
Pearson (Spearman)
correlation matrix
ARJ
24,1
16
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
net income, either positively or negatively, are more likely to be disclosed in an 8-K.
Restatements that do not impact net income are more likely to be stealth. In addition,
results show that restatements under SEC investigation are also more likely to be
disclosed in an 8-K.
As expected, we ?nd that restatements with longer ?ling delays are more likely to
be stealth. Coef?cients on both AC and B4 are signi?cantly negative indicating that
restating ?rms that currently have an auditor different from the one they had at the
end of the restatement period and those with one of the big four auditors are more
likely to ?le an 8-K restatement. The results also show that restating ?rms with higher
book-to-market ratios are more likely to use stealth restatements. In summary, results
given in Table V are consistent with the univariate ?ndings and the SEC’s
recommendation that 8-K restatements are required to correct material errors[14].
3. Market reaction to restatement disclosure venue
Figure 1 shows the market reaction around the restatement disclosure date for stealth
and 8-K restatements. We examine several sub-periods around the disclosure date
since we do not know whether ?rms use stealth restatements to ?le immaterial errors
or to hide material ones. Use of a 22-day trading window (approximately one calendar
month) following the restatement date provides market participants enough time to
discover and appropriately react to the two types of restatements. If ?rms use stealth
restatements to hide material errors, then upon discovery (which we expect to occur
within 22 trading days) we anticipate market participants to react negatively. The plot
shows a statistically (and economically) signi?cant market reaction of 21.61 percent
Coef?cient Expected sign Estimate
Intercept a
0
? 22.577
* *
Restatement attributes
Neg_Dir a
1
– 21.063
* *
Pos_Dir a
2
– 21.420
* *
Rev_Recog a
3
– 20.078
Res_Yrs a
4
– 20.102
SEC_Inv a
5
– 21.253
*
Filing_Delay a
6
þ 2.094
* *
ACC a
7
– 0.014
Auditor attributes
AC a
8
2 20.997
*
B4 a
9
2 20.919
* *
Firm attributes
Size_1 a
10
– 0.070
B/M_1 a
11
? 0.760
* *
DUMROE_1 a
12
? 20.410
DUMRET_1 a
13
? 0.104
Dum_Prior8K a
14
– 0.008
Dum_PriorSt a
15
? 0.111
Pseudo-R
2
14.21%
Notes: n ¼ 961; signi?cance at:
*
5 and
* *
1 percent; equation (1); Disc_Ven equals 1 for stealth
restatements and 0 for 8-K restatements; year ?xed effects are not reported; A one-tailed (two-tailed)
t-test applies to variables with expected (unknown) signs
Table V.
Attributes of restatement
disclosure venue
Stealth
restatements
17
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
on the day of and the day following the 8-K restatement. There is no signi?cant market
reaction for the long return window (þ2, þ 22) for 8-K restatements. These results are
consistent with the market ef?ciently re?ecting the information included in the 8-K
restatements.
For stealth restatements, there is no signi?cant market reaction in the short run
(0, þ 1) or the long run (þ2, þ 22). Since the largest buy and hold market-adjusted
return on the plot for stealth restatement (1.22 percent) is earned over (0, þ 5), we also
test that interval and ?nd it to be insigni?cantly different from zero (buy and hold
market-adjusted return of 21.53 percent for 8-K restatements over this same window
is signi?cantly less than zero). These ?ndings are consistent with the notion that
stealth restatements do not convey material information.
Table VI presents the multivariate results for model (2)[15]. Consistent with our
univariate results, we ?nd that Disc_Ven is positively associated with BHMAR (0, 1).
In other words, stealth restatements garner a less negative reaction than 8-K
restatements (b
1
. 0). In addition, also consistent with our univariate ?ndings, the
multivariate analyses show no market reaction to stealth restatements (b
0
þ b
1
is
insigni?cant) suggesting that they are not value relevant. We ?nd the presence of a big
four auditor to have a positive effect on the market reaction while ?rms with auditor
changes and prior stealth restatements garner a more negative reaction[16].
4. Robustness tests
We conduct several tests to ensure that our results are robust across different sample,
variable and model variations. Our main ?ndings are robust (1) across the two stealth
restatement venues (10-K and 10-Q), (2) to including only one restatement per ?rm,
Figure 1.
Market-adjusted
reaction to stealth
and 8-K restatements
0.015
0.01
0.005
0
–0.005
–5 –4 –3 –2 –1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 19 20 21 22 18
–0.01
–0.015
–0.02
: 8-K restatements : Stealth restatements
8-K restatements
(n = 858) (%)
Stealth restatements
(n = 103) (%)
BHMAR (–5, –1) –0.05 0.14
BHMAR (0, +1) –1.61** 0.65
BHMAR (0, +5) –1.53** 1.22
BHMAR (+2, +22) 0.04 –1.04
Notes: BHMAR is the buy and hold market-adjusted return over the specified event windows;
** (*) indicates significance at 1% (5%)
ARJ
24,1
18
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and (3) to including the dollar effect of the restatement on net income, instead of the
directional effect on net income (Neg_Dir and Pos_Dir).
Further, the market reaction results in Table VI hold when using a (0, þ 5) event
window. Finally, to control for the possibility that the (expected) market reaction
following the restatement disclosure may affect the choice of disclosure venue, we run
a simultaneous equation model of equation (2) where we include BHMAR(0, 1) as an
independent variable and Disc_Ven as a dependent variable while keeping all the other
independent variables. The coef?cient on Disc_Ven in equation (2) remains signi?cant
in explaining the market reaction around the restatement disclosure date.
V. Conclusion
We investigate the attributes of and the market reaction to restatements disclosed via
two restatement disclosure venues (8-Ks and 10-Ks/10-Qs). This inquiry is important
because pursuant to item 4.02 effective August 23, 2004 requiring ?rms to use Form
8-K for ?nancial restatements, the use of 10-Ks/10-Qs to disclose restatements
continues to be observed. We document that stealth restatements have less severe
restatement attributes than 8-K restatements. Speci?cally, restatements not affecting
net income, with longer ?ling delays, and not subject to SEC investigation are more
likely to be stealth. In addition, our results suggest that restating ?rms that changed
audit ?rms since the end of the restatement period and their current auditor is one of
the big four accounting ?rms are more likely to disclose 8-K restatements.
Consistent with prior literature, we also document a negative market reaction to 8-K
restatements around the restatement disclosure date. However, for the stealth
restatements we ?nd no signi?cant market reaction around the 10-K/10-Q ?ling date
and up to 22 trading days after the ?ling date. These market results are robust after
controlling for auditor and ?rm attributes. Collectively, both sets of results suggest
that stealth restatements correct immaterial errors, consistent with the SEC staff’s
Coef?cient Expected sign Estimate
Intercept b
0
– 20.025
* *
Disc_Ven b
1
þ 0.013
* *
Auditor attributes
AC b
2
? 20.018
*
B4 b
3
þ 0.007
*
AQ_INC b
4
þ 20.018
AQ_DEC b
5
– 0.023
Firm attributes
Size_1 b
6
? 0.001
B/M_1 b
7
? 0.007
Dum_Prior8K B
8
– 0.011
Dum_PriorSt b
9
– 20.014
* *
b
0
þ b
1
Not signi?cant 20.012
Adjusted-R
2
9.97%
Notes: n ¼ 913; signi?cance at:
*
5 and
* *
1 percent; equation (2); BHMAR (0, 1) is the two-day buy
and hold market-adjusted return; year ?xed effects are not reported; a one-tailed (two-tailed) t-test
applies to variables with expected (unknown) signs; observations from model (2) with absolute values
of studentized t greater than 2.0 or Cook’s D greater than 1.0 are omitted, yielding 913 observations
Table VI.
Multivariate analysis
of market reaction
to restatements
Stealth
restatements
19
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
position that all material restatements have to be ?led using item 4.02 Form 8-K. They
also support SEC’s attempt of leveling the playing ?eld for all investors.
Notes
1. The best practice would be for companies to ?le an 8-K as soon as it knows it will restate.
Companies should also ?le their restatements using amended ?lings (10-K/A, 10-Q/A, etc.)
that explain exactly what was corrected (Turner and Weirich, 2006).
2. It is in the eyes of the restating ?rm as to when the four-day period begins. For instance, it is
conceivable that a company’s board and its auditors argue for months as to whether a
restatement is necessary only to agree four days (or less) before they are scheduled to ?le
their ?nancial statements (Turner and Weirich, 2006). Such situations are even more likely
as a result of a safe harbor from public and private claims under Exchange Act Section 10(b)
and Rule 10b-5 granted to companies by the SEC for a failure to ?le a timely 4.02 disclosure.
3. Additional Form 8-K Disclosure Requirements and Acceleration of Filing Date (SEC Release
No. 33-8400).
4. See web site: http://thecaq.aicpa.org/NR/rdonlyres/20040913highlights2086nwzdmgcxb
sqwtu?hltbpkesrgskhgiq.pdf (p. 25).
5. Transient institutional investors are de?ned as those with shorter investor horizons and
higher portfolio turnover.
6. As mentioned earlier, prior research has shown signi?cant negative market reaction and
labor market penalties for top management and ?rm directors following restatement
announcements (Palmrose et al., 2004; Srinivasan, 2005; Callen et al., 2006; Desai et al., 2006).
Karpoff et al. (2008) show that litigation penalties average $23.5 million per ?rm for their
sample. In addition, they show that for every dollar of in?ated market value that results from
a ?nancial misstatement, ?rms loose the in?ated dollar plus an additional $3.08, of which
$0.36 is related to litigation cost and the remainder to reputation cost.
7. In total, 654 ?rms have one restatement, 108 ?rms have two restatements, 20 ?rms with
three restatements, four ?rms with four restatements and three ?rms with ?ve restatements.
8. SAB 108 was issued on September 13, 2006, prompting a signi?cant amount of restatements
regarding materiality. Our study does not include restatements related to SAB 108 given the
data constraint at the time of empirical analysis.
9. The year 2000 is the ?rst calendar year for which Audit Analytics provides restatement
data.
10. Since stealth restatements are disclosed through 10-K/10-Q ?lings, a question remains as to
whether a market reaction to stealth restatements is affected by a market reaction to
unexpected earnings and/or other information disclosed through the regular SEC ?lings.
For the 10-K ?lings included in this study, earnings are announced, on average, 21 days
earlier than when 10-Ks are ?led. Since earnings announcements usually occur earlier than
the 10-K/10-Q ?ling dates, we do not expect a market reaction to unexpected earnings around
the ?ling dates. Being unable to control for the market reaction caused by all other
information disclosed in the ?lings is a limitation of our study.
11. We require our sample observations to have returns for 22 days (approximately one calendar
month) following the restatement disclosure date. Consequently, we exclude 11 restatements
disclosed in December 2007, resulting in a total of 961 restatements.
12. About 46 percent of the restatements come from ?rms belonging to two industries: consumer
discretionary and information technology. To rule out the industry effect, we re-ran all the
analyses without these two industries and our inferences remained the same.
ARJ
24,1
20
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
13. Mean of Neg_Impact for 8-K restatements is less than the corresponding mean for stealth
restatements at the 10 percent signi?cance level.
14. Hennes et al. (2008) suggest the importance of distinguishing accounting errors from
irregularities, with irregularities (errors) de?ned as intentional (unintentional) mistakes.
They report that market reaction to accounting irregularities is more pronounced than its
reaction to accounting errors and suggest the use of fraud-related class action lawsuits as a
proxy for accounting irregularities. Since we do not have access to class action lawsuit data,
we cannot directly test the impact of differentiating accounting errors from irregularities on
the choice of disclosure venues. However, the results from model (1) suggest that stealth
restatements typically include immaterial information while 8-K restatements include
material information. Therefore, one may conclude that an 8-K ?ling is more likely to be used
for restatements with accounting irregularities while a 10-K/10-Q ?ling is more likely to be
used for restatements with (immaterial) accounting errors.
15. To mitigate the potential effects of outliers, we delete observations with absolute values of
studentized t greater than 2.0 or Cook’s D greater than 1.0, obtained from the OLS analysis of
model (2). This screening yields 913 observations.
16. To make our results comparable to those reported in Sharp (2007), we run two additional
regressions. In one analysis, we include the seven restatement attributes (used earlier in
model (1)) but not Disc_Ven, while in the other analysis we include the seven restatement
attributes and Disc_Ven. Inclusion of the restatement attributes makes the coef?cient on
Disc_Ven insigni?cant. This implies that after controlling for the restatement attributes,
disclosing restatements in the 10-K/10-Q as opposed to the 8-K does not garner any market
reaction. This result is consistent with market participants reacting to the information
contained in the restatement and not merely the disclosure venue.
References
Beatty, R. (1989), “Auditor reputation and the pricing of initial public offerings”, The Accounting
Review, Vol. 64 No. 4, pp. 693-709.
Bhojraj, S., Lee, C. and Oler, D. (2003), “What’s my line? A comparison of industry classi?cation
schemes for capital market research”, Journal of Accounting Research, Vol. 41 No. 5,
pp. 745-74.
Branson, B. and Pagach, D. (2005), “Recent controversies in accounting for operating leases and
leasehold improvements”, The CPA Journal, Vol. 75 No. 10, pp. 40-1.
Callen, J., Livnat, J. and Segal, D. (2006), “Accounting restatements: are they always bad news for
investors?”, The Journal of Investing, Fall, pp. 57-68.
Desai, H., Hogan, C. and Wilkins, M. (2006), “The reputational penalty for aggressive accounting:
earnings restatements and management turnover”, The Accounting Review, Vol. 81 No. 1,
pp. 83-112.
Files, R., Swanson, E. and Tse, S. (2009), “Stealth disclosure of accounting restatements”,
The Accounting Review, Vol. 84 No. 5, pp. 1495-520.
Gleason, C., Jenkins, N. and Johnson, W. (2008), “The contagion effects of accounting
restatements”, The Accounting Review, Vol. 83 No. 1, pp. 83-110.
Gordon, E., Henry, E., Peytcheva, M. and Sun, L. (2007), “Disclosure credibility and market
reaction to restatements”, working paper, Rutgers Business School, Newark, NJ.
Grif?n, P. (2003), “A league of their own? Financial analysts’ responses to restatements and
corrective disclosures”, Journal of Accounting, Auditing & Finance, Vol. 18 No. 4,
pp. 479-518.
Stealth
restatements
21
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Hennes, K., Leone, A. and Miller, B. (2008), “The importance of distinguishing errors from
irregularities in restatement research: the case of restatements and CEO/CFO turnover”,
Accounting Review, Vol. 83 No. 6, pp. 1487-519.
Hribar, P., Jenkins, N. and Wang, J. (2005), “Institutional investors and accounting restatements”,
working paper, University of Iowa, Iowa City, IA.
Karpoff, J., Lee, D. and Martin, G. (2008), “The cost to ?rms of cooking the books”, Journal of
Financial and Quantitative Analysis, Vol. 43 No. 3, pp. 581-611.
Palmrose, Z., Richardson, V. and Scholz, S. (2004), “Determinants of market reactions to
restatement announcements”, Journal of Accounting and Economics, Vol. 37 No. 1,
pp. 59-89.
Prentice, R. (2005), Student Guide to the Sarbanes-Oxley Act, Thomson Publishing, Mason, OH.
Reilly, D. (2006), “No more ‘stealth restating’”, The Wall Street Journal, September 21.
Securities and Exchange Commission (2004), Final Rule: Additional Form 8-K Disclosure
Requirements an Acceleration of Filing Date, Government Printing Of?ce, Washington, DC.
Sharp, N. (2007), “Do ?rms strategically manage the timing and transparency of accounting
restatement disclosures?”, working paper, Texas A&M University, College Station, TX.
Srinivasan, S. (2005), “Consequences of ?nancial reporting failure for outside directors: evidence
from accounting restatements and audit committee members”, Journal of Accounting
Research, Vol. 43 No. 2, pp. 291-334.
Teoh, S. and Wong, T. (1993), “Perceived auditor quality and the earnings response coef?cients”,
The Accounting Review, Vol. 68 No. 2, pp. 346-66.
Turner, L. and Weirich, T. (2006), “A closer look at ?nancial statement restatements: analyzing
the reasons behind the trend”, The CPA Journal, December, pp. 13-23.
Wilson, W. (2008), “An empirical analysis of the decline in the information content of earnings
following restatements”, The Accounting Review, Vol. 83 No. 2, pp. 519-48.
About the authors
Afshad J. Irani is an Associate Professor of Accounting at Washington and Lee University.
He received his PhD in Business Administration with an emphasis in Accounting from
The Pennsylvania State University in 1998. His research interests are in examining management
behavior as re?ected in activities such as voluntary disclosure, earnings management, insider
trading and ?nancial restatements.
Le (Emily) Xu is an Associate Professor of Accounting at the University of New Hampshire.
She obtained her PhD from the University of Massachusetts Amherst in 2003. Her primary
research interests are ?nancial restatements, ?nancial analysts’ forecasts of earnings and value
relevance of accounting information. Le (Emily) Xu is the corresponding author and can be
contacted at: [email protected]
ARJ
24,1
22
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Afshad J. Irani, Stefanie L. Tate, Le (Emily) Xu. 2015. Restatements: Do They Affect Auditor Reputation
for Quality?. Accounting Horizons 29, 829-851. [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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_678869063.pdf
Effective August 23, 2004, the US Securities and Exchange Commission (SEC) requires all
firms to disclose restatements via an item 4.02 Form 8-K filing. However, a significant number of firms
continue to disclose restatements using means other than an 8-K. Commonly referred to as stealth
restatements, the purpose of this paper is to investigate the materiality of restatements disclosed in
either the 10-K or the 10-Q by comparing them to those disclosed via 8-K.
Accounting Research Journal
Do stealth restatements convey material information?
Afshad J . Irani Le (Emily) Xu
Article information:
To cite this document:
Afshad J . Irani Le (Emily) Xu, (2011),"Do stealth restatements convey material information?", Accounting
Research J ournal, Vol. 24 Iss 1 pp. 5 - 22
Permanent link to this document:
http://dx.doi.org/10.1108/10309611111148751
Downloaded on: 24 January 2016, At: 21:12 (PT)
References: this document contains references to 21 other documents.
To copy this document: [email protected]
The fulltext of this document has been downloaded 804 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.
23-49 http://dx.doi.org/10.1108/10309611111148760
Salma Ibrahim, Li Xu, Genese Rogers, (2011),"Real and accrual-based earnings management and its legal
consequences: Evidence from seasoned equity offerings", Accounting Research J ournal, Vol. 24 Iss 1 pp.
50-78 http://dx.doi.org/10.1108/10309611111148779
Magdy Farag, Rafik Elias, (2011),"Relative audit fees and client loyalty in the audit market", Accounting
Research J ournal, Vol. 24 Iss 1 pp. 79-93 http://dx.doi.org/10.1108/10309611111148788
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Do stealth restatements convey
material information?
Afshad J. Irani
Williams School of Commerce, Economics and Politics,
Washington and Lee University, Lexington, Virginia, USA, and
Le (Emily) Xu
Whittemore School of Business and Economics, University of New Hampshire,
Durham, New Hampshire, USA
Abstract
Purpose – Effective August 23, 2004, the US Securities and Exchange Commission (SEC) requires all
?rms to disclose restatements via an item 4.02 Form 8-K ?ling. However, a signi?cant number of ?rms
continue to disclose restatements using means other than an 8-K. Commonly referred to as stealth
restatements, the purpose of this paper is to investigate the materiality of restatements disclosed in
either the 10-K or the 10-Q by comparing them to those disclosed via 8-K.
Design/methodology/approach – Univariate and multivariate analyses compare the
characteristics of and the market reaction to 10-K/10-Q restatements to those of 8-K restatements.
Findings – The authors ?nd stealth restatements are more likely to be those not affecting net income,
with longer ?ling delays, not subject to SEC investigation and made by ?rms audited by non-big four
accounting ?rms. The authors document a negative market reaction to 8-K restatements around the
restatement disclosure date. However, for stealth restatements they ?nd no market reaction around the
10-K/10-Q ?ling date and for up to 22 trading days after the 10-K/10-Q ?lings.
Research limitations/implications – The study shows a signi?cant difference in materiality
between stealth and 8-K restatements.
Practical implications – The study is important to investors, regulators and academics because it
supports the notion that stealth restatements include less signi?cant information relative to that
disclosed in 8-K restatements. This result is in line with the SEC disclosure requirement.
Originality/value – The signi?cant number of stealth restatements since 2004 begs the question as to
what kind of information is being disclosed in these restatements. The paper responds to this question.
Keywords United States of America, Disclosure, Restatements, Stealth, Item 4.02, Form 8-K
Paper type Research paper
I. Introduction
We investigate the attributes of and the market reaction to ?nancial restatements ?rst
disclosed in quarterly or annual reports (10-K/10-Q) without being ?led via item 4.02
Form 8-K (hereafter referred to as 8-K restatements) expressly created by the US
Securities and Exchange Commission (SEC) for this purpose. Commonly referred to as
stealth restatements, this type of restatement has become increasingly common over
the past few years. About 10.7 percent of our restatement sample is stealth and this
percentage has annually increased over our 2005-2007 sample period. We examine
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1030-9616.htm
The authors thank Steve Balsam, Pieter Elgers, Steve Gill, Erin Moore and seminar participants
at the University of New Hampshire, Washington and Lee University and the Northeast Regional
Meeting of the American Accounting Association for their useful suggestions and insights. All
the data used in this study are available from public sources.
Stealth
restatements
5
Accounting Research Journal
Vol. 24 No. 1, 2011
pp. 5-22
qEmerald Group Publishing Limited
1030-9616
DOI 10.1108/10309611111148751
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the materiality of information released in stealth restatements on two fronts:
restatement attributes and the corresponding market reaction.
Compared to 8-K restatements, we ?nd that ?rms are more likely to use stealth
restatements if:
.
their restatements do not affect previously reported net income;
.
there is a longer lag between the end of the restatement period and the
restatement disclosure date; and
.
the restatement is not a result of an SEC investigation.
We also ?nd stealth restatements to be the likely disclosure venue for ?rms not
changing auditors between the end of the restatement period and the restatement ?ling
date, employing non-big four auditors and with higher book-to-market ratios.
In analyzing stock price reaction to restatement announcements, we ?nd stealth
restatements garner no signi?cant market reaction on the day of and the day following
the SEC ?ling date. On the other hand, 8-K restatements earn a 21.61 percent
(signi?cant) market-adjusted return over the day of and the day following their ?ling
date. This difference in market reaction between the two disclosure venues holds in a
multivariate setting after controlling for various auditor and ?rm attributes. In
addition, there is no long-term (up to 22 trading days or one calendar month) market
reaction to stealth restatements. Our market reaction ?ndings are consistent with our
aforementioned ?nding of stealth restatements possessing less severe restatement
attributes than 8-K restatements.
Pursuant to item 4.02 effective August 23, 2004, the SEC requires a company to ?le
Form 8-K under item 4.02(a) if and when the ?rm itself concludes that its previously
issued ?nancial statements can no longer be relied upon because of an error, or under
item4.02(b) if the company is advised by its independent accountant of the error[1]. The
?rm then has four business days to ?le an 8-K in which they need to disclose speci?c
information[2], [3]. However, not every revision to a company’s prior ?nancial
statements that results froman error requires an 8-Krestatement. This was the response
given by the SEC staff in their September 13, 2004 meeting with the AICPA’s SEC
Regulations Committee to a speci?c inquiry as to whether all restatements need to be
announced in an 8-K[4]. The SEC staff agreed with the Committee’s view that an 8-K
restatement need not be ?led if the ?rmand its independent accountant conclude that the
error is immaterial (based on a SAB 99 materiality assessment) and that the prior
?nancial statements may continue to be relied on. This position taken by the SEC
coupled with ?rm management’s possible desire to be less transparent when it comes to
disclosing restatements are what we believe to be plausible reasons for the continued
increase in the usage of stealth restatements. Results from our analysis of disclosure
venue attributes as well as those fromour stock market reaction tests are consistent with
the aforementioned materiality argument in that stealth restatements are not material.
We contribute to the restatement literature in a number of ways. To our knowledge,
this is one of the ?rst studies to explicitly examine stealth restatements and investigate
whether ?rms are following SEC guidelines for disclosing restatements. Questions
have repeatedly surfaced in the popular press regarding what stealth restatements
correct and whether that information is value relevant. We explicitly respond to those
concerns by examining stealth restatement characteristics and the corresponding
market reaction. Finally, our results documenting signi?cant differences between
ARJ
24,1
6
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
the two disclosure venues should be of value to future restatement research in that
these two restatement venues should be explicitly controlled for in the empirical
analysis.
The rest of the paper is organized as follows. We provide our motivation and review
of related research in Section II followed by the research design in Section III. Section IV
presents descriptive statistics and our empirical ?ndings. Concluding remarks are
offered in Section V.
II. Motivation and related research
Restatements resulting from accounting errors are typically perceived as negative
signals regardless of their impact on the ?rm’s book value. In addition to suggesting
that past ?nancial statements can no longer be relied upon, they are indicative of a
weak accounting information system that can possibly lead to future problems.
Academic research focusing on ?nancial restatements has investigated their impact on
restating ?rm’s market value (Palmrose et al., 2004; Callen et al., 2006) as well as on
peer ?rm’s market value (Gleason et al., 2008) and found them to have a negative
impact on both. Wilson (2008) ?nds a reduction in the information content of restating
?rm’s earnings following a restatement. Restatements have also been found to lead to
turnover and signi?cant labor market penalties for both top management (Desai et al.,
2006) and ?rm directors (Srinivasan, 2005). Finally, restatements have been shown to
have a negative impact on analyst following (Grif?n, 2003) and transient institutional
investor holdings (Hribar et al., 2005)[5].
The number of ?nancial restatements has been on a rise since the passage of the
Sarbanes-Oxley Act (SOX) in 2002. Turner and Weirich (2006) report 1,323 restatements
?led by US public companies during 1997-2002. From2003 to 2005 that number grewto
2,319, out of which 320 were by companies making at least a second restatement. The
number of restatements is expected to stay at this level and perhaps even rise as small
companies (public ?oat of less than $75 million) begin fully complying with SOX
Section 404 provisions for ?scal years ending on or after December 15, 2009.
The increase in the number of restatements resulting fromaccounting errors has been
accompanied by ?rms’ increasing usage of alternative “less revealing” disclosure venues
possiblyinhopes of attractingminimal public attention. Eventhoughbelievers inef?cient
markets mayclaimthat disclosinginformationin“less revealing” venues still leads to that
information being incorporated in security prices, one can safely argue that such a
practice does not provide a level playing ?eld for all investors (especially small investors).
Item4.02 Form8-Kdisclosure, required starting August 23, 2004, was the SEC’s response
to combat such practices and level the playing ?eld by providing a uniform disclosure
venue for all restatements. According to the SEC (2004), the required disclosures:
[. . .] will bene?t markets by increasing the number of unquestionably or presumptively
material events that must be disclosed currently. They will also provide investors with better
and more timely disclosure of important corporate events.
The SEC went on to say that these amendments also further the goals of SOX Section
409 that requires companies to disclose on a “rapid and current” basis, information that
is necessary for the protection of investors and in the public interest (Prentice, 2005).
Contrary to the requirements outlined in the aforementioned SEC ruling but
consistent with the subsequent response by the SEC staff that immaterial restatements
Stealth
restatements
7
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
do not need to be ?led via item 4.02 Form 8-K disclosure, a summer 2006 report by the
Government Accountability Of?ce found that about 17 percent of public companies
restating results between August 2004 and September 2005 did not ?le it using an 8-K.
Similarly, according to a 2006 report by Glass, Lewis & Co., 14 percent of restatements
in 2005 are stealth compared to only 2 percent a year ago. In response, the SEC sent
letters to companies whose periodic ?lings (10-K/10-Q) were ?agged for containing
material restated numbers insisting that they disclose those restatements using Form
8-K. As anecdotal evidence, Reilly (2006) reports that Sun Microsystems, Inc. included
its restatement in the 10-K ?ling but later re-?led Form 8-K for the same restatement to
achieve “maximum transparency.” Similarly, Inter-Tel, Inc. had also used 10-K to
disclose a restatement. However, following comments from the SEC the company later
?led 8-K for the same restatement. These anecdotes suggest that stealth restatements
continue to exist despite the SEC’s effort to combat such practices.
The motivation behind this study is to seek a better understanding of the kind of
information that is disclosed in stealth restatements. If we ?nd that stealth restatements
only correct immaterial errors, then a ?rm’s decision to disclose themin 10-K’s/10-Q’s to
avoid negative publicity associated with restatements in general is understandable and
consistent with SEC guidance. Such a result would also negate the economic relevance
of the ?ndings reported in the 2006 GAO and Glass, Lewis & Co. reports.
However, if we ?nd correction of material errors in stealth restatements, then such
disclosures may be an indication of a ?rm’s desire to keep it hidden and avoid the
negative consequences associated with the value relevance of information disclosed in
those restatements[6]. In turn, this will create distrust in the market and an uneven
playing ?eld especially for small investors, who unlike institutional investors do not
have the time and/or the ability to discover and appropriately react to such disclosures.
In addition, stealth ?rms in the future may be subjected to SEC investigation and legal
action.
We conduct our investigation by examining both the attributes of stealth
restatements as well as the stock market reaction to their disclosure. Examining
attributes will provide an understanding of the kind of errors being disclosed in stealth
restatements, while the test of the market reaction to stealth restatements will re?ect on
the value relevance of the disclosed errors. We believe both these tests are
complimentary and necessary to re?ect on the materiality of information disclosed in
stealth restatements.
Several recent academic papers have investigated the prominence of the
restatement announcement within alternative disclosure venues. Using data from
1997-2002, Files et al. (2009) divide restatements disclosed in press releases into three
levels of prominence: high (disclosed in the headline), medium (described in the body,
headline is on a different subject) and low (change the comparative period amounts in
an earnings release with no direct mention of the restatement). After controlling for the
severity of the restatement, Files et al. ?nd the smallest negative market reaction
around the press release date (21, þ 1) to low prominence restatement disclosures. In
a related study using a similar sample period, Gordon et al. (2007) investigate the
impact of disclosure credibility on market reaction to restatements. They de?ne
disclosure credibility as “investors’ perceptions of the believability of a particular
disclosure” and measure it across two dimensions: ?rm management and restatement
attributes (disclosure prominence being one of the restatement attributes). They ?nd
ARJ
24,1
8
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
that more prominent restatement announcements (both in terms of inclusion in a
headline or being stand alone) garner a more negative stock price reaction. Note that
the sample periods used in both Files et al. (2009) and Gordon et al. (2007) predate the
item 4.02 Form 8-K disclosure requirement. In addition, both studies do not consider
stealth restatements, which have even lower prominence. Finally, unlike our study,
neither examines restatement attributes.
In a concurrent study, using data from August 24, 2004 to December 31, 2005, Sharp
(2007) investigates whether ?rms strategically manage the timing and transparency of
restatement disclosures. In addition to restatements disclosed in an 8-K (no obscurity)
and 10-K/10-Q (high obscurity/stealth restatements), the paper also includes
restatements disclosed in amended ?lings (10-K/A or 10-Q/A) which it identi?es as
those with medium obscurity. The results most relevant to our study include:
.
a positive association between usage of an 8-K and restatement attributes such
as negative and (small) positive earnings surprises, longer restated periods,
material weakness in internal control and more timely disclosure; and
.
for stealth restatements, less negative market reaction in the short run (22, þ 2)
and no market reaction in the long run (22, þ 20) after controlling for other
restatement attributes.
Our study differs from Sharp (2007) in two signi?cant ways. First, our study
incorporates a longer sample period and a signi?cantly larger ?rm sample (even
without including amended ?lings). This is relevant in terms of improving external
validity of the results, especially given the SEC’s push in 2006 to reduce/eliminate
stealth disclosures involving material restatements and an increase in the percentage
of stealth restatements as indicated by our sample. Second, the multivariate models
used in our study include important variables different than those used in Sharp (2007),
such as ?rms subject to an SEC investigation, auditor attributes and revenue
recognition errors.
III. Research design
1. Sample selection
We collect US ?nancial restatements resulting from accounting errors disclosed during
2005-2007 from Audit Analytics with restatement disclosure venues of 8-K, 10-K and
10-Q. Our sample period begins in 2005 since that is the ?rst full calendar year
following the effective date (August 23, 2004) of the item 4.02 Form 8-K disclosure
requirement. We adopt a conservative approach by not including restatements ?led
during 2004 in case those ?rms were either not familiar with the new SEC ruling or
chose to ignore it given the proximity of their ?ling date to the ruling’s effective date.
The initial sample comprises of 1,110 restatements. We lose 111 observations due to
missing ?rm data on COMPUSTAT and CRSP. In addition, we delete 35 observations
since their restatement period end dates are after their restatement ?ling dates. That
yields a preliminary sample of 964 restatements out of which 123 are stealth. To
con?rm the coding used by Audit Analytics, we then checked the 10-K/10-Q’s of each
of the 123 stealth restatements for any mention of a prior or simultaneous 8-K
disclosure. We also conducted a search on Lexis-Nexis Academic Universe on the days
following the 10-K/10-Q to make sure that there was no follow-up 8-K disclosure. We
found that 17 stealth restatements were accompanied by simultaneous 8-K disclosures
Stealth
restatements
9
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and three had 8-K disclosures within four business days following the 10-K/10-Q. We
reclassi?ed the 17 as 8-K restatements and dropped the three as we were unclear about
management intent. This yields our ?nal sample of 961 restatements disclosed by 789
?rms, out of which 103 are stealth restatements[7], [8].
2. Empirical model: determinants of disclosure venue
To investigate attributes of ?nancial restatement disclosures, we estimate the
following year ?xed effects logistic model:
Disc_Ven ¼ a
0
þa
1
Neg_Dir þa
2
Pos_Dir þa
3
Rev_Recog þa
4
Res_Yrs
þa
5
SEC_Inv þa
6
Filing_Delay þa
7
ACC þa
8
AC þa
9
B4
þa
10
Size_1 þa
11
B=M_1 þa
12
DumROE_1 þa
13
DumRet_1
þa
14
Dum_Prior8K þa
15
Dum_PriorSt þ1
ð1Þ
where:
Disc_Ven ¼ 1 for stealth restatements, 0 for 8-K restatements.
Neg_Dir ¼ 1 if the restatement negatively impacts net income, 0 otherwise.
Pos_Dir ¼ 1 if the restatement positively impacts net income, 0 otherwise.
Rev_Recog ¼ 1 if it is a revenue recognition restatement, 0 otherwise.
Res_Yrs ¼ Number of restated years.
SEC_Inv ¼ 1 if the restatement was under SEC investigation, 0 otherwise.
Filing_Delay ¼ Number of days between the end of the restatement period and
the restatement disclosure date divided by 365.
ACC ¼ 1 if the restatement is related to lease accounting, stock-based
compensation (SFAS 123 (R)) or hedge accounting, 0 otherwise.
AC ¼ 1 if the auditor at the end of the restatement period is different
from the auditor at the restatement disclosure date, 0 otherwise.
B4 ¼ 1 if the auditor on the restatement disclosure date is one of the big
four accounting ?rms, 0 otherwise.
Size_1 ¼ Log of market value of equity at the end of the ?scal year prior to
the restatement disclosure date, in millions of dollars.
B/M_1 ¼ Book value of equity divided by market value of equity at the end
of the ?scal year prior to the restatement disclosure date.
DumROE_1 ¼ 1 if the restating ?rm’s ROE in the year prior to the restatement
disclosure date is within the top or bottom three deciles,
0 otherwise. ROE is de?ned as income before extraordinary items
divided by market value of equity.
DumRet_1 ¼ 1 if the restating ?rm’s compounded annual return for the ?scal
year prior to the restatement disclosure date is within the top or
bottom three deciles, 0 otherwise.
ARJ
24,1
10
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Dum_Prior8K ¼ 1 if the restating ?rm had a previous 8-K restatement since year
2000, 0 otherwise.
Dum_PriorSt ¼ 1 if the restating ?rm had a previous stealth restatement since
year 2000, 0 otherwise.
We use Neg_Dir, Pos_Dir, Rev_Recog, Res_Yrs and SEC_Inv to proxy for the
materiality of the restatement. Following Palmrose et al. (2004) and Wilson (2008), we
consider restatements affecting net income (Neg_Dir and Pos_Dir), involving revenue
recognition errors (Rev_Recog), affecting longer time periods (Res_Yrs), and under SEC
investigation (SEC_Inv) to be more material. If ?rms do in fact follow the SEC
requirement and disclose all material restatements using Form 8-K, then we expect a
negative sign on all these variables. While ?rms can ?le an 8-K on any business day,
?rms with stealth restatements wait until their next annual or quarterly ?ling (10-K or
10-Q) to include the restatement information. Therefore, stealth restatements are
expected to have longer ?ling delays and consequently we expect a positive coef?cient
on Filing_Delay.
We include ACC to address three speci?c accounting issues related to leasing,
compensation and hedging addressed by recent restatements. In 2005, the SEC issued
an open letter clarifying the correct way to account for leases and that led to a
signi?cant number of restatements (Branson and Pagach, 2005). Following the highly
publicized hedge accounting misconduct at Fannie Mae, a lot of companies reassessed
their practice of recording hedges, thereby resulting in a signi?cant number of
restatements. Finally, following SFAS 123(R) taking effect in 2005, many companies
restated their ?nancial statements to correct errors related to share-based payment
computations. We include these three accounting issues in one (control) variable
because all three accounting issues affected a signi?cant number of ?rms and received
much public attention. Given the publicity related to these issues, we expect affected
?rms to have used 8-K restatements (i.e. a negative a
7
).
In addition to SEC_Inv, we include two monitoring variables related to auditor
attributes, AC and B4. AC captures the change in auditors from the end of the
restatement period to the restatement disclosure date. We expect its coef?cient to be
negative because a different auditor at the time of disclosure does not have any
incentives to hide restatements that occurred during the tenure of the previous auditor.
Therefore, an 8-K is more likely to be used as the disclosure venue. We also expect the
coef?cient on B4 to be negative. Big four accounting ?rms are generally considered to
be higher quality auditors (Beatty, 1989; Teoh and Wong, 1993) and therefore more
likely to require their clients to ?le an 8-K restatement.
Lastly, we included six ?rm-speci?c attributes. Since bigger ?rms are under more
public scrutiny, we expect these ?rms to disclose 8-K restatements and thus the
predictive sign for the coef?cient on Size_1 is negative. We do not predict signs for
coef?cients of B/M_1, DUMROE_1 and DUMRET_1. A ?rm with higher (lower) book
to market indicates that it is a value (growth) ?rm. DUMROE_1 and DUMRET_1
indicate whether a ?rm performed very well or very poorly relative to its peers, as
measured by their return on equity and stock market performance at the end of the
?scal year prior to their restatement disclosure date. A growth ?rm and/or a ?rm that
has performed better or worse than its peers may have incentives to disguise their
restatement due to the possible economic loss, but they may also want to be more
Stealth
restatements
11
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
transparent because they are more visible and do not want to lose their credibility in
the eyes of the market participants.
We include Dum_Prior8K and Dum_PriorSt to capture the existence of prior 8-K
and stealth restatements, respectively, since the beginning of 2000[9]. We expect ?rms
with prior 8-Krestatements to continue to use an 8-Kfor their current restatement ?ling
(a
14
, 0). Firms that have made prior stealth restatements may also follow their past
practice which would indicate a positive coef?cient on Dum_PriorSt. However, their
prior restatements may have been ?led prior to August 23, 2004 at which time there
was no explicit SEC requirement for using an 8-K. Consequently, for restatements ?led
in 2005 onwards these same ?rms may use the required item 4.02 Form 8-K disclosure
venue. In addition, these ?rms may have experienced signi?cant economic and
litigation costs related to their prior stealth restatement that would prevent them from
repeating their past practice. Consequently, we do not predict a sign on the coef?cient
of Dum_PriorSt.
3. Empirical model: multivariate analysis of market reaction to restatements
We use the following year ?xed effects regression model to investigate the market
reaction:
BHMAR ð0; 1Þ ¼ b
0
þb
1
Disc_Ven þb
2
AC þb
3
B4 þb
4
AQ_INC
þb
5
AQ_DEC þb
6
Size_1 þb
7
B=M_1 þb
8
Dum_Prior8k
þb
9
Dum_PriorSt þp
ð2Þ
where:
BHMAR (0, 1) ¼ Two-day buy and hold market-adjusted returns.
AQ_Inc ¼ 1 if the auditor at the end of restatement period is one of the
non-big four accounting ?rms and the auditor at the time of
restatement disclosure is one of the big four accounting ?rms,
0 otherwise.
AQ_Dec ¼ 1 if the auditor at the end of restatement period is one of the big
four accounting ?rms and the auditor at the time of restatement
disclosure is one of the non-big four accounting ?rms,
0 otherwise.
and other independent variables are as previously de?ned in Section III.2.
Disc_Ven is our variable of interest as it indicates whether the short-run market
reaction differs between the two restatement disclosure venues after controlling for
auditor and ?rm attributes. Based on prior literature (Palmrose et al., 2004; Callen et al.,
2006), the market reacts negatively when ?rms announce restatements. These studies
have primarily focused on restatements that were announced publicly. Therefore,
these results are related to 8-K restatements in our study. In comparison, stealth
restatements are disclosed quietly through regular SEC ?lings. If these stealth
restatements contain immaterial information, then we expect a less negative (or zero)
market reaction. Consequently, we expect b
1
to be positive.
The next four variables focus on auditor attributes. When the auditor at the
restatement disclosure date is different from the auditor at the end of the
ARJ
24,1
12
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
restatement period, the market may lend more credibility to the announced restatement
and thus react less negatively. On the other hand, the market may attribute the
accounting irregularities (being addressed by the restatement) as one of the reasons
behind the auditor change and react even more negatively due to increased
uncertainty. Consequently, we do not predict a sign on the coef?cient of AC. We expect
the coef?cient on B4 to be positive as big four auditors are perceived to be of higher
quality. Consequently, the market may consider ?rms with big four auditors to have a
better ?nancial reporting system and may feel less uncertain about the scope and
impact of the restatement. We expect coef?cients on AQ_INC (AQ_DEC) to also be
positive (negative) since an auditor change from non-big four (big four) to big four
(non-big four) is perceived as an increase (decrease) in the auditor quality.
We expect negative coef?cients on Dum_Prior8k and Dum_PriorSt. An existing
restatement coupled with at least one previous restatement (8-K or stealth) may signal
continuing problems with the ?rm’s ?nancial reporting system and thus cause the
market to react more negatively to the current restatement. Finally, we include Size_1
and B/M_1 as control variables[10].
IV. Empirical results
1. Descriptive statistics
Table I panel A presents the number of restatements by restatement and auditor
attributes for the two restatement disclosure venues. Our sample consists of 858 8-K
restatements and 103 stealth restatements. The percentage of stealth restatements has
steadily increased from 6.9 (25/361) in 2005 to 11.4 (42/367) in 2006 to 15.5 (36/233) in
2007[11]. Relative to 8-K restatements, a lower percentage of stealth restatements
impact net income (45 vs 77 percent for 8-K), are a result of revenue recognition errors
(12 vs 15 percent for 8-K), and are involved in a SEC investigation (5 vs 14 percent for
8-K). Table II panel B shows that 2,414 (245) corrections were included in the 858 (103)
8-K (stealth) restatements which yields an average of 2.8 (2.4) corrections per 8-K
(stealth) restatement. Roughly, one-fourth of errors disclosed in stealth restatements
are corrections of classi?cation issues. These ?ndings are consistent with the SEC’s
position requiring item 4.02 Form 8-K disclosures for material restatements.
In addition, consistent with our expectation, Table I panel A shows proportionately
more 8-K restatements related to lease, compensation and hedging activities (ACC).
Also as expected, proportionately less ?rms with stealth restatements use big four
accounting ?rms as their auditors (67 vs 79 percent for 8-K). Both disclosure venues
have a similar distribution across industries, with industry de?ned using the GICS
two-digit industry classi?cation codes (Bhojraj et al., 2003)[12].
Table III panel C provides descriptive statistics for restatement and ?rm attributes
across the two disclosure venues. Tests of signi?cance for the difference in the means
and medians of the corresponding variables (using t- and Wilcoxon statistics,
respectively) suggest that stealth restatements have signi?cantly less negative impact
on net income (median only) and less number of restated years[13]. These results are
consistent with the ?ndings in panels A and B that stealth restatements appear to be
less severe than 8-K restatements. In addition, panel C shows that stealth restatements
have signi?cantly longer ?ling delays and higher book-to-market ratios.
Table IV presents the Pearson and Spearman correlation coef?cients for the
exogenous variables used in our study. Most of the signi?cant correlations found
Stealth
restatements
13
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
in the table were expected. By construction, Neg_Dir and Pos_Dir are negatively
correlated (20.508). AQ_DEC and AC are also highly correlated because there is only
one observation in our sample that has increased auditor quality. So, most of the
sample ?rms that have changed auditors either have decreased their auditor quality or
remained at the same level. In addition, B4 is highly correlated with Size_1 (0.502).
This is not surprising since larger ?rms are more likely to be audited by one of the big
four accounting ?rms. Furthermore, we test for the potential effect of multicollinearity
on standard errors both by examining variance in?ation factors as well as running
step-wise regressions and ?nd our results to be unaffected.
2. Attributes of restatement disclosure venue
Table V presents the empirical results for the attributes of restatement disclosure
venue (model (1)). The logistic model generates a pseudo-R
2
of 14.21 percent. Signs on
most coef?cient estimates are consistent with our expectations. a
1
and a
2
are negative
and signi?cant at the 1 percent level suggesting that restatements impacting
Number of restatements
a
Total 8-K Stealth
Full sample 961 858 103
2005 361 336 (39%) 25 (24%)
2006 367 325 (38%) 42 (41%)
2007 233 197 (23%) 36 (35%)
Restatement attributes
Neg_Dir 562 524 (61%) 38 (37%)
Pos_Dir 149 141 (16%) 8 (8%)
Rev_Recog 141 129 (15%) 12 (12%)
Sec_Inv 121 116 (14%) 5 (5%)
Dum_Prior8K 154 136 (16%) 18 (17%)
Dum_PriorSt 331 292 (34%) 39 (38%)
ACC 225 212 (25%) 13 (13%)
Auditor attributes
AC 50 44 (5%) 6 (6%)
B4 744 675 (79%) 69 (67%)
AQ_Inc 1 1 (0.1%) 0 (0%)
AQ_Dec 19 15 (2%) 4 (4%)
GICS two-digit industry classi?cation
Energy 52 46 (5%) 6 (6%)
Materials 39 38 (4%) 1 (1%)
Industrials 110 98 (11%) 12 (12%)
Consumer discretionary 248 215 (25%) 33 (32%)
Consumer staples 24 20 (2%) 4 (4%)
Health care 99 88 (10%) 11 (11%)
Financials 134 125 (15%) 9 (9%)
Information technology 208 182 (21%) 26 (25%)
Telecommunication services 26 26 (3%) 0 (0%)
Utilities 21 20 (2%) 1 (1%)
Notes: n ¼ 961;
a
% in parenthesis indicates the proportion of regular (858) or stealth (103)
restatements for that speci?c restatement or auditor attribute; for example, 61 percent (524/858) of all
8-K restatements have negative impact on net income
Table I.
Descriptive statistics –
panel A: number of
restatements by
restatement attributes
and auditor attributes for
different disclosure
venues
ARJ
24,1
14
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
8-K restatements Stealth restatements
Number Percentage Number Percentage
PPE, intangible or ?xed asset issues and related
depreciation, depletion or amortization errors 217 9 7 3
Accounts/loans receivables, investment and cash issues 68 3 18 7
Deferred stock-based compensation 153 6 9 4
Liabilities, payables, reserves and accrual estimate
failures 110 5 18 7
Lease, SFAS 5, legal, contingency and commitment
issues 152 6 7 3
Revenue recognition issues 129 5 12 5
Expense (cost of goods sold, payroll, SGA, other)
recording issues and capitalization of expenditures 269 11 18 7
Tax expense/bene?t/deferral/other (FAS109) issue 156 6 14 6
Acquisitions, mergers, disposals and re-org acct issues 105 4 15 6
Classi?cation issues on ?nancial statements 179 7 58 24
Other 876 36 69 28
Total 2,414 100 245 100
Table II.
Descriptive
statistics – panel B:
number of restatements
by restatement
accounts for different
disclosure venues
Mean Median
8-K
(n ¼ 858)
Stealth
(n ¼ 103)
8-K
(n ¼ 858)
Stealth
(n ¼ 103)
Restatement attributes
Negative change in net income per restated year
divided by book value of equity at the beginning
of the restatement period 20.088 20.040 20.021
*
20.009
Positive change in net income per restated year
divided by book value of equity at the beginning
of the restatement period 0.041
*
0.011 0.013 0.006
Res_Yrs 2.871
* *
2.013 2.750
* *
1.500
Filing_Delay 0.460
* *
0.909 0.367
* *
1.093
Firm attributes
Market value of equity at the end of the ?scal year
prior to the restatement disclosure date, in millions
of dollars 3,235 1,833 537 345
B/M_1 0.474
*
0.557 0.448
* *
0.525
Income before extraordinary items divided by market
value of equity at the end of the ?scal year prior to the
restatement disclosure date 20.018 20.029 0.030 0.020
Compounded annual return for the ?scal year prior
to restatement disclosure date 0.168 0.089 0.079 0.015
Notes:
*
,
* *
indicates whether the restatement and ?rm attributes are signi?cantly different between
regular and stealth restatements at the 5 and 1 percent signi?cance level; statistical signi?cance is
based on the t- and Wilcoxon tests for the means and medians, respectively:
Disc_Ven ¼ a
0
þa
1
Neg_Dir þa
2
Pos_Dir þa
3
Rev_Recog þa
4
Res_Yrs þa
5
SEC_Inv
þa
6
Filing_Delay þa
7
ACC þa
8
AC þa
9
B4 þa
10
Size_1 þa
11
B=M_1
þa
12
DumROE_1 þa
13
DumRet_1 þa
14
Dum_Prior8K þa
15
Dum_PriorSt þ1
ð1Þ
Table III.
Descriptive statistics –
panel C: descriptive
statistics for restatement
and ?rm attributes by
disclosure venue
Stealth
restatements
15
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
N
e
g
_
D
i
r
P
o
s
_
D
i
r
R
e
v
_
R
e
c
o
g
R
e
s
_
Y
r
s
S
E
C
_
I
n
v
F
i
l
i
n
g
_
D
e
l
a
y
D
u
m
_
P
r
i
o
r
8
k
D
u
m
_
P
r
i
o
r
S
t
A
C
C
A
C
B
4
A
Q
_
I
n
c
A
Q
_
D
e
c
S
i
z
e
_
1
B
/
M
_
1
R
O
E
_
1
R
e
t
_
1
N
e
g
_
D
i
r
1
2
0
.
5
0
8
*
*
0
.
0
9
9
*
*
0
.
2
0
8
*
*
0
.
0
4
6
2
0
.
1
0
1
*
*
2
0
.
0
1
2
2
0
.
0
0
7
0
.
1
5
2
*
*
2
0
.
0
4
0
0
.
0
2
0
2
0
.
0
3
8
2
0
.
0
6
2
0
.
0
5
5
2
0
.
0
7
4
*
2
0
.
0
2
6
0
.
0
0
8
P
o
s
_
D
i
r
2
0
.
5
0
8
*
*
1
2
0
.
0
5
6
2
0
.
0
3
4
2
0
.
0
5
0
2
0
.
0
2
8
0
.
0
1
7
2
0
.
0
0
2
0
.
0
8
9
*
0
.
0
0
3
0
.
0
0
4
2
0
.
0
1
4
0
.
0
2
2
0
.
0
3
2
0
.
0
3
2
2
0
.
0
3
3
2
0
.
0
0
5
R
e
v
_
R
e
c
o
g
0
.
0
9
9
*
*
2
0
.
0
5
6
1
2
0
.
0
5
3
0
.
0
1
1
2
0
.
0
0
1
0
.
0
4
3
0
.
0
7
1
*
2
0
.
0
7
6
*
2
0
.
0
0
4
0
.
0
3
4
2
0
.
0
1
3
0
.
0
0
4
0
.
0
0
4
2
0
.
0
2
8
2
0
.
0
5
0
2
0
.
0
6
4
*
R
e
s
_
Y
r
s
0
.
2
1
0
*
*
2
0
.
0
5
7
2
0
.
0
5
4
1
0
.
2
3
9
*
*
2
0
.
1
4
4
*
*
2
0
.
0
4
0
2
0
.
0
3
6
0
.
1
7
9
*
*
2
0
.
0
9
3
*
*
0
.
2
3
6
*
*
2
0
.
0
4
5
2
0
.
0
2
8
0
.
2
8
3
*
*
2
0
.
0
2
3
0
.
1
2
4
*
*
0
.
0
8
0
*
S
E
C
_
I
n
v
0
.
0
4
6
2
0
.
0
5
0
0
.
0
1
1
0
.
2
4
8
*
*
1
2
0
.
1
3
1
*
*
2
0
.
0
2
0
2
0
.
0
1
8
2
0
.
0
6
9
*
2
0
.
0
0
4
0
.
0
1
0
2
0
.
0
1
2
2
0
.
0
3
1
0
.
1
2
4
*
*
0
.
0
0
9
2
0
.
0
1
4
2
0
.
0
3
1
F
i
l
i
n
g
_
D
e
l
a
y
2
0
.
1
2
5
*
*
2
0
.
0
5
5
2
0
.
0
2
1
2
0
.
0
9
1
*
*
2
0
.
0
4
9
1
2
0
.
0
2
1
0
.
0
3
3
2
0
.
0
8
6
*
*
0
.
1
3
8
*
*
2
0
.
0
6
1
2
0
.
0
1
4
0
.
1
1
5
*
*
2
0
.
0
8
2
*
0
.
0
1
0
2
0
.
0
5
2
2
0
.
0
2
5
D
u
m
_
P
r
i
o
r
8
K
2
0
.
0
1
2
0
.
0
1
7
0
.
0
4
3
2
0
.
0
1
3
2
0
.
0
2
0
2
0
.
0
0
1
1
0
.
6
0
3
*
*
2
0
.
0
2
7
0
.
1
1
5
*
*
0
.
0
1
2
2
0
.
0
1
4
0
.
0
4
0
2
0
.
0
0
7
0
.
0
4
5
2
0
.
0
2
2
0
.
0
4
2
D
u
m
_
P
r
i
o
r
S
t
2
0
.
0
0
7
2
0
.
0
0
2
0
.
0
7
1
*
2
0
.
0
0
8
2
0
.
0
1
8
0
.
0
2
9
0
.
6
0
3
*
*
1
2
0
.
0
6
5
*
0
.
1
3
6
*
*
0
.
0
2
5
0
.
0
4
5
0
.
0
2
3
0
.
0
3
8
2
0
.
0
0
3
2
0
.
0
9
6
*
*
0
.
0
1
6
A
C
C
0
.
1
5
2
*
*
0
.
0
8
9
*
*
2
0
.
0
7
6
*
0
.
0
7
1
*
2
0
.
0
6
9
*
2
0
.
1
1
0
*
*
2
0
.
0
2
7
2
0
.
0
6
5
*
1
2
0
.
0
3
0
0
.
1
2
8
*
*
2
0
.
0
1
8
0
.
0
1
0
0
.
1
6
9
*
*
0
.
0
1
0
0
.
1
4
7
*
*
0
.
0
9
4
*
*
A
C
2
0
.
0
4
0
0
.
0
0
3
2
0
.
0
0
4
2
0
.
0
9
3
*
*
2
0
.
0
0
4
0
.
1
3
7
*
*
0
.
1
1
5
*
*
0
.
1
3
6
*
*
2
0
.
0
3
0
1
2
0
.
2
6
6
*
*
0
.
1
3
8
*
*
0
.
6
0
6
*
*
2
0
.
0
9
9
*
*
2
0
.
0
1
2
2
0
.
0
2
5
2
0
.
0
3
1
B
4
0
.
0
2
0
0
.
0
0
4
0
.
0
3
4
0
.
1
7
4
*
*
0
.
0
1
0
2
0
.
0
3
2
0
.
0
1
2
0
.
0
2
5
0
.
1
2
8
*
*
2
0
.
2
6
6
*
*
1
0
.
0
1
7
2
0
.
2
6
3
*
*
0
.
5
0
6
*
*
0
.
0
2
4
0
.
1
5
9
*
*
0
.
0
8
2
*
*
A
Q
_
I
n
c
2
0
.
0
3
8
2
0
.
0
1
4
2
0
.
0
1
3
2
0
.
0
3
0
2
0
.
0
1
2
2
0
.
0
1
3
2
0
.
0
1
4
0
.
0
4
5
2
0
.
0
1
8
0
.
1
3
8
*
*
0
.
0
1
7
1
2
0
.
0
0
5
2
0
.
0
1
1
2
0
.
0
0
3
2
0
.
0
2
8
2
0
.
0
1
5
A
Q
_
D
e
c
2
0
.
0
6
2
0
.
0
2
2
0
.
0
0
4
2
0
.
0
4
4
2
0
.
0
3
1
0
.
1
1
5
*
*
0
.
0
4
0
0
.
0
2
3
0
.
0
1
0
0
.
6
0
6
*
*
2
0
.
2
6
3
*
*
2
0
.
0
0
5
1
2
0
.
0
6
1
2
0
.
0
1
7
2
0
.
0
2
0
2
0
.
0
2
1
S
i
z
e
_
1
0
.
0
5
7
0
.
0
3
4
0
.
0
0
4
0
.
2
4
1
*
*
0
.
1
1
7
*
*
2
0
.
0
6
6
*
2
0
.
0
1
1
0
.
0
5
0
0
.
1
7
2
*
*
2
0
.
0
9
6
*
*
0
.
5
0
2
*
*
2
0
.
0
0
8
2
0
.
0
5
8
1
2
0
.
1
4
8
*
*
0
.
2
6
4
*
*
0
.
2
1
0
*
*
B
/
M
_
1
2
0
.
0
4
9
0
.
0
1
8
2
0
.
0
1
2
2
0
.
0
3
7
0
.
0
4
4
0
.
0
1
8
0
.
0
2
8
2
0
.
0
0
8
0
.
0
1
1
2
0
.
0
1
4
0
.
0
3
3
2
0
.
0
0
4
2
0
.
0
1
4
2
0
.
1
2
3
*
*
1
0
.
1
0
2
*
*
2
0
.
2
4
3
*
*
R
O
E
_
1
2
0
.
0
0
8
0
.
0
3
5
0
.
0
1
6
0
.
1
0
1
*
*
0
.
0
0
4
2
0
.
0
2
8
2
0
.
0
0
6
2
0
.
0
7
6
*
0
.
1
0
4
*
*
2
0
.
0
1
7
0
.
1
5
8
*
*
0
.
0
0
0
2
0
.
0
2
9
0
.
2
4
8
*
*
0
.
1
7
8
*
*
1
0
.
3
6
8
*
*
R
e
t
_
1
0
.
0
1
9
2
0
.
0
1
7
2
0
.
0
6
0
*
0
.
0
2
0
2
0
.
0
4
2
2
0
.
0
4
7
0
.
0
1
9
0
.
0
0
3
0
.
0
5
7
0
.
0
0
8
0
.
0
1
2
2
0
.
0
1
3
0
.
0
3
3
0
.
0
8
7
*
*
2
0
.
2
0
1
*
*
0
.
1
7
0
*
*
1
N
o
t
e
s
:
n
¼
9
6
1
;
s
i
g
n
i
?
c
a
n
c
e
a
t
:
*
5
a
n
d
*
*
1
p
e
r
c
e
n
t
(
t
w
o
-
t
a
i
l
e
d
)
;
P
e
a
r
s
o
n
(
S
p
e
a
r
m
a
n
)
c
o
r
r
e
l
a
t
i
o
n
c
o
e
f
?
c
i
e
n
t
s
a
r
e
p
r
e
s
e
n
t
e
d
b
e
l
o
w
(
a
b
o
v
e
)
t
h
e
d
i
a
g
o
n
a
l
Table IV.
Pearson (Spearman)
correlation matrix
ARJ
24,1
16
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
net income, either positively or negatively, are more likely to be disclosed in an 8-K.
Restatements that do not impact net income are more likely to be stealth. In addition,
results show that restatements under SEC investigation are also more likely to be
disclosed in an 8-K.
As expected, we ?nd that restatements with longer ?ling delays are more likely to
be stealth. Coef?cients on both AC and B4 are signi?cantly negative indicating that
restating ?rms that currently have an auditor different from the one they had at the
end of the restatement period and those with one of the big four auditors are more
likely to ?le an 8-K restatement. The results also show that restating ?rms with higher
book-to-market ratios are more likely to use stealth restatements. In summary, results
given in Table V are consistent with the univariate ?ndings and the SEC’s
recommendation that 8-K restatements are required to correct material errors[14].
3. Market reaction to restatement disclosure venue
Figure 1 shows the market reaction around the restatement disclosure date for stealth
and 8-K restatements. We examine several sub-periods around the disclosure date
since we do not know whether ?rms use stealth restatements to ?le immaterial errors
or to hide material ones. Use of a 22-day trading window (approximately one calendar
month) following the restatement date provides market participants enough time to
discover and appropriately react to the two types of restatements. If ?rms use stealth
restatements to hide material errors, then upon discovery (which we expect to occur
within 22 trading days) we anticipate market participants to react negatively. The plot
shows a statistically (and economically) signi?cant market reaction of 21.61 percent
Coef?cient Expected sign Estimate
Intercept a
0
? 22.577
* *
Restatement attributes
Neg_Dir a
1
– 21.063
* *
Pos_Dir a
2
– 21.420
* *
Rev_Recog a
3
– 20.078
Res_Yrs a
4
– 20.102
SEC_Inv a
5
– 21.253
*
Filing_Delay a
6
þ 2.094
* *
ACC a
7
– 0.014
Auditor attributes
AC a
8
2 20.997
*
B4 a
9
2 20.919
* *
Firm attributes
Size_1 a
10
– 0.070
B/M_1 a
11
? 0.760
* *
DUMROE_1 a
12
? 20.410
DUMRET_1 a
13
? 0.104
Dum_Prior8K a
14
– 0.008
Dum_PriorSt a
15
? 0.111
Pseudo-R
2
14.21%
Notes: n ¼ 961; signi?cance at:
*
5 and
* *
1 percent; equation (1); Disc_Ven equals 1 for stealth
restatements and 0 for 8-K restatements; year ?xed effects are not reported; A one-tailed (two-tailed)
t-test applies to variables with expected (unknown) signs
Table V.
Attributes of restatement
disclosure venue
Stealth
restatements
17
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
on the day of and the day following the 8-K restatement. There is no signi?cant market
reaction for the long return window (þ2, þ 22) for 8-K restatements. These results are
consistent with the market ef?ciently re?ecting the information included in the 8-K
restatements.
For stealth restatements, there is no signi?cant market reaction in the short run
(0, þ 1) or the long run (þ2, þ 22). Since the largest buy and hold market-adjusted
return on the plot for stealth restatement (1.22 percent) is earned over (0, þ 5), we also
test that interval and ?nd it to be insigni?cantly different from zero (buy and hold
market-adjusted return of 21.53 percent for 8-K restatements over this same window
is signi?cantly less than zero). These ?ndings are consistent with the notion that
stealth restatements do not convey material information.
Table VI presents the multivariate results for model (2)[15]. Consistent with our
univariate results, we ?nd that Disc_Ven is positively associated with BHMAR (0, 1).
In other words, stealth restatements garner a less negative reaction than 8-K
restatements (b
1
. 0). In addition, also consistent with our univariate ?ndings, the
multivariate analyses show no market reaction to stealth restatements (b
0
þ b
1
is
insigni?cant) suggesting that they are not value relevant. We ?nd the presence of a big
four auditor to have a positive effect on the market reaction while ?rms with auditor
changes and prior stealth restatements garner a more negative reaction[16].
4. Robustness tests
We conduct several tests to ensure that our results are robust across different sample,
variable and model variations. Our main ?ndings are robust (1) across the two stealth
restatement venues (10-K and 10-Q), (2) to including only one restatement per ?rm,
Figure 1.
Market-adjusted
reaction to stealth
and 8-K restatements
0.015
0.01
0.005
0
–0.005
–5 –4 –3 –2 –1 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 19 20 21 22 18
–0.01
–0.015
–0.02
: 8-K restatements : Stealth restatements
8-K restatements
(n = 858) (%)
Stealth restatements
(n = 103) (%)
BHMAR (–5, –1) –0.05 0.14
BHMAR (0, +1) –1.61** 0.65
BHMAR (0, +5) –1.53** 1.22
BHMAR (+2, +22) 0.04 –1.04
Notes: BHMAR is the buy and hold market-adjusted return over the specified event windows;
** (*) indicates significance at 1% (5%)
ARJ
24,1
18
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
and (3) to including the dollar effect of the restatement on net income, instead of the
directional effect on net income (Neg_Dir and Pos_Dir).
Further, the market reaction results in Table VI hold when using a (0, þ 5) event
window. Finally, to control for the possibility that the (expected) market reaction
following the restatement disclosure may affect the choice of disclosure venue, we run
a simultaneous equation model of equation (2) where we include BHMAR(0, 1) as an
independent variable and Disc_Ven as a dependent variable while keeping all the other
independent variables. The coef?cient on Disc_Ven in equation (2) remains signi?cant
in explaining the market reaction around the restatement disclosure date.
V. Conclusion
We investigate the attributes of and the market reaction to restatements disclosed via
two restatement disclosure venues (8-Ks and 10-Ks/10-Qs). This inquiry is important
because pursuant to item 4.02 effective August 23, 2004 requiring ?rms to use Form
8-K for ?nancial restatements, the use of 10-Ks/10-Qs to disclose restatements
continues to be observed. We document that stealth restatements have less severe
restatement attributes than 8-K restatements. Speci?cally, restatements not affecting
net income, with longer ?ling delays, and not subject to SEC investigation are more
likely to be stealth. In addition, our results suggest that restating ?rms that changed
audit ?rms since the end of the restatement period and their current auditor is one of
the big four accounting ?rms are more likely to disclose 8-K restatements.
Consistent with prior literature, we also document a negative market reaction to 8-K
restatements around the restatement disclosure date. However, for the stealth
restatements we ?nd no signi?cant market reaction around the 10-K/10-Q ?ling date
and up to 22 trading days after the ?ling date. These market results are robust after
controlling for auditor and ?rm attributes. Collectively, both sets of results suggest
that stealth restatements correct immaterial errors, consistent with the SEC staff’s
Coef?cient Expected sign Estimate
Intercept b
0
– 20.025
* *
Disc_Ven b
1
þ 0.013
* *
Auditor attributes
AC b
2
? 20.018
*
B4 b
3
þ 0.007
*
AQ_INC b
4
þ 20.018
AQ_DEC b
5
– 0.023
Firm attributes
Size_1 b
6
? 0.001
B/M_1 b
7
? 0.007
Dum_Prior8K B
8
– 0.011
Dum_PriorSt b
9
– 20.014
* *
b
0
þ b
1
Not signi?cant 20.012
Adjusted-R
2
9.97%
Notes: n ¼ 913; signi?cance at:
*
5 and
* *
1 percent; equation (2); BHMAR (0, 1) is the two-day buy
and hold market-adjusted return; year ?xed effects are not reported; a one-tailed (two-tailed) t-test
applies to variables with expected (unknown) signs; observations from model (2) with absolute values
of studentized t greater than 2.0 or Cook’s D greater than 1.0 are omitted, yielding 913 observations
Table VI.
Multivariate analysis
of market reaction
to restatements
Stealth
restatements
19
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
position that all material restatements have to be ?led using item 4.02 Form 8-K. They
also support SEC’s attempt of leveling the playing ?eld for all investors.
Notes
1. The best practice would be for companies to ?le an 8-K as soon as it knows it will restate.
Companies should also ?le their restatements using amended ?lings (10-K/A, 10-Q/A, etc.)
that explain exactly what was corrected (Turner and Weirich, 2006).
2. It is in the eyes of the restating ?rm as to when the four-day period begins. For instance, it is
conceivable that a company’s board and its auditors argue for months as to whether a
restatement is necessary only to agree four days (or less) before they are scheduled to ?le
their ?nancial statements (Turner and Weirich, 2006). Such situations are even more likely
as a result of a safe harbor from public and private claims under Exchange Act Section 10(b)
and Rule 10b-5 granted to companies by the SEC for a failure to ?le a timely 4.02 disclosure.
3. Additional Form 8-K Disclosure Requirements and Acceleration of Filing Date (SEC Release
No. 33-8400).
4. See web site: http://thecaq.aicpa.org/NR/rdonlyres/20040913highlights2086nwzdmgcxb
sqwtu?hltbpkesrgskhgiq.pdf (p. 25).
5. Transient institutional investors are de?ned as those with shorter investor horizons and
higher portfolio turnover.
6. As mentioned earlier, prior research has shown signi?cant negative market reaction and
labor market penalties for top management and ?rm directors following restatement
announcements (Palmrose et al., 2004; Srinivasan, 2005; Callen et al., 2006; Desai et al., 2006).
Karpoff et al. (2008) show that litigation penalties average $23.5 million per ?rm for their
sample. In addition, they show that for every dollar of in?ated market value that results from
a ?nancial misstatement, ?rms loose the in?ated dollar plus an additional $3.08, of which
$0.36 is related to litigation cost and the remainder to reputation cost.
7. In total, 654 ?rms have one restatement, 108 ?rms have two restatements, 20 ?rms with
three restatements, four ?rms with four restatements and three ?rms with ?ve restatements.
8. SAB 108 was issued on September 13, 2006, prompting a signi?cant amount of restatements
regarding materiality. Our study does not include restatements related to SAB 108 given the
data constraint at the time of empirical analysis.
9. The year 2000 is the ?rst calendar year for which Audit Analytics provides restatement
data.
10. Since stealth restatements are disclosed through 10-K/10-Q ?lings, a question remains as to
whether a market reaction to stealth restatements is affected by a market reaction to
unexpected earnings and/or other information disclosed through the regular SEC ?lings.
For the 10-K ?lings included in this study, earnings are announced, on average, 21 days
earlier than when 10-Ks are ?led. Since earnings announcements usually occur earlier than
the 10-K/10-Q ?ling dates, we do not expect a market reaction to unexpected earnings around
the ?ling dates. Being unable to control for the market reaction caused by all other
information disclosed in the ?lings is a limitation of our study.
11. We require our sample observations to have returns for 22 days (approximately one calendar
month) following the restatement disclosure date. Consequently, we exclude 11 restatements
disclosed in December 2007, resulting in a total of 961 restatements.
12. About 46 percent of the restatements come from ?rms belonging to two industries: consumer
discretionary and information technology. To rule out the industry effect, we re-ran all the
analyses without these two industries and our inferences remained the same.
ARJ
24,1
20
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
13. Mean of Neg_Impact for 8-K restatements is less than the corresponding mean for stealth
restatements at the 10 percent signi?cance level.
14. Hennes et al. (2008) suggest the importance of distinguishing accounting errors from
irregularities, with irregularities (errors) de?ned as intentional (unintentional) mistakes.
They report that market reaction to accounting irregularities is more pronounced than its
reaction to accounting errors and suggest the use of fraud-related class action lawsuits as a
proxy for accounting irregularities. Since we do not have access to class action lawsuit data,
we cannot directly test the impact of differentiating accounting errors from irregularities on
the choice of disclosure venues. However, the results from model (1) suggest that stealth
restatements typically include immaterial information while 8-K restatements include
material information. Therefore, one may conclude that an 8-K ?ling is more likely to be used
for restatements with accounting irregularities while a 10-K/10-Q ?ling is more likely to be
used for restatements with (immaterial) accounting errors.
15. To mitigate the potential effects of outliers, we delete observations with absolute values of
studentized t greater than 2.0 or Cook’s D greater than 1.0, obtained from the OLS analysis of
model (2). This screening yields 913 observations.
16. To make our results comparable to those reported in Sharp (2007), we run two additional
regressions. In one analysis, we include the seven restatement attributes (used earlier in
model (1)) but not Disc_Ven, while in the other analysis we include the seven restatement
attributes and Disc_Ven. Inclusion of the restatement attributes makes the coef?cient on
Disc_Ven insigni?cant. This implies that after controlling for the restatement attributes,
disclosing restatements in the 10-K/10-Q as opposed to the 8-K does not garner any market
reaction. This result is consistent with market participants reacting to the information
contained in the restatement and not merely the disclosure venue.
References
Beatty, R. (1989), “Auditor reputation and the pricing of initial public offerings”, The Accounting
Review, Vol. 64 No. 4, pp. 693-709.
Bhojraj, S., Lee, C. and Oler, D. (2003), “What’s my line? A comparison of industry classi?cation
schemes for capital market research”, Journal of Accounting Research, Vol. 41 No. 5,
pp. 745-74.
Branson, B. and Pagach, D. (2005), “Recent controversies in accounting for operating leases and
leasehold improvements”, The CPA Journal, Vol. 75 No. 10, pp. 40-1.
Callen, J., Livnat, J. and Segal, D. (2006), “Accounting restatements: are they always bad news for
investors?”, The Journal of Investing, Fall, pp. 57-68.
Desai, H., Hogan, C. and Wilkins, M. (2006), “The reputational penalty for aggressive accounting:
earnings restatements and management turnover”, The Accounting Review, Vol. 81 No. 1,
pp. 83-112.
Files, R., Swanson, E. and Tse, S. (2009), “Stealth disclosure of accounting restatements”,
The Accounting Review, Vol. 84 No. 5, pp. 1495-520.
Gleason, C., Jenkins, N. and Johnson, W. (2008), “The contagion effects of accounting
restatements”, The Accounting Review, Vol. 83 No. 1, pp. 83-110.
Gordon, E., Henry, E., Peytcheva, M. and Sun, L. (2007), “Disclosure credibility and market
reaction to restatements”, working paper, Rutgers Business School, Newark, NJ.
Grif?n, P. (2003), “A league of their own? Financial analysts’ responses to restatements and
corrective disclosures”, Journal of Accounting, Auditing & Finance, Vol. 18 No. 4,
pp. 479-518.
Stealth
restatements
21
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
Hennes, K., Leone, A. and Miller, B. (2008), “The importance of distinguishing errors from
irregularities in restatement research: the case of restatements and CEO/CFO turnover”,
Accounting Review, Vol. 83 No. 6, pp. 1487-519.
Hribar, P., Jenkins, N. and Wang, J. (2005), “Institutional investors and accounting restatements”,
working paper, University of Iowa, Iowa City, IA.
Karpoff, J., Lee, D. and Martin, G. (2008), “The cost to ?rms of cooking the books”, Journal of
Financial and Quantitative Analysis, Vol. 43 No. 3, pp. 581-611.
Palmrose, Z., Richardson, V. and Scholz, S. (2004), “Determinants of market reactions to
restatement announcements”, Journal of Accounting and Economics, Vol. 37 No. 1,
pp. 59-89.
Prentice, R. (2005), Student Guide to the Sarbanes-Oxley Act, Thomson Publishing, Mason, OH.
Reilly, D. (2006), “No more ‘stealth restating’”, The Wall Street Journal, September 21.
Securities and Exchange Commission (2004), Final Rule: Additional Form 8-K Disclosure
Requirements an Acceleration of Filing Date, Government Printing Of?ce, Washington, DC.
Sharp, N. (2007), “Do ?rms strategically manage the timing and transparency of accounting
restatement disclosures?”, working paper, Texas A&M University, College Station, TX.
Srinivasan, S. (2005), “Consequences of ?nancial reporting failure for outside directors: evidence
from accounting restatements and audit committee members”, Journal of Accounting
Research, Vol. 43 No. 2, pp. 291-334.
Teoh, S. and Wong, T. (1993), “Perceived auditor quality and the earnings response coef?cients”,
The Accounting Review, Vol. 68 No. 2, pp. 346-66.
Turner, L. and Weirich, T. (2006), “A closer look at ?nancial statement restatements: analyzing
the reasons behind the trend”, The CPA Journal, December, pp. 13-23.
Wilson, W. (2008), “An empirical analysis of the decline in the information content of earnings
following restatements”, The Accounting Review, Vol. 83 No. 2, pp. 519-48.
About the authors
Afshad J. Irani is an Associate Professor of Accounting at Washington and Lee University.
He received his PhD in Business Administration with an emphasis in Accounting from
The Pennsylvania State University in 1998. His research interests are in examining management
behavior as re?ected in activities such as voluntary disclosure, earnings management, insider
trading and ?nancial restatements.
Le (Emily) Xu is an Associate Professor of Accounting at the University of New Hampshire.
She obtained her PhD from the University of Massachusetts Amherst in 2003. Her primary
research interests are ?nancial restatements, ?nancial analysts’ forecasts of earnings and value
relevance of accounting information. Le (Emily) Xu is the corresponding author and can be
contacted at: [email protected]
ARJ
24,1
22
To purchase reprints of this article please e-mail: [email protected]
Or visit our web site for further details: www.emeraldinsight.com/reprints
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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
This article has been cited by:
1. Afshad J. Irani, Stefanie L. Tate, Le (Emily) Xu. 2015. Restatements: Do They Affect Auditor Reputation
for Quality?. Accounting Horizons 29, 829-851. [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
2
2
4
J
a
n
u
a
r
y
2
0
1
6
(
P
T
)
doc_678869063.pdf