Stealth Disclosure of Accounting Irregularities: Is Silence Golden? Edward P. Swanson* Professor Texas A&M University [email protected]Senyo Tse Professor Texas A&M University [email protected]Rebecca Wynalda Ph.D. Student Texas A&M University [email protected]*Corresponding Author. Edward P. Swanson Mays Business School Texas A&M University 4353 TAMU College Station, TX 77843-4353 Current Draft: August 21, 2007 Keywords: accounting restatements, press release disclosure, voluntary disclosure Data Availability: The data used in this study are publicly available from the sources indicated in the text. Acknowledgments: We would like to thank Devin Shanthikumar, Nate Sharp, and workshop participants at Texas A&M University and at the 2007 AAA Annual Meeting for their comments. We are indebted to Jason Call, Vikas Hegde, and Natasha Thomas for their research assistance. The authors gratefully acknowledge financial support from the Mays Business School of Texas A&M. Swanson would also like to acknowledge funding from the Durst Chair in Accounting.
55
Embed
Stealth Disclosure of Accounting Irregularities: Is ...web-docs.stern.nyu.edu/salomon/docs/conferences/swanson et al.pdf · Stealth Disclosure of Accounting Irregularities: Is Silence
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Stealth Disclosure of Accounting Irregularities: Is Silence Golden?
body of the press release (medium prominence). And the remaining firms simply restate
prior-period comparative balances in an earnings release, with no disclosure of the
restatement other than a footnote stating that the financial figures for the prior year have been
changed (low prominence). With the last two practices, an investor who does not read each
press release might not be aware of the accounting irregularity. A report by Glass Lewis
(2006) uses the expression ―stealth restatements‖ to refer to the practice of disclosing an
accounting irregularity within an earnings release. The poor quality of many restatement
disclosures has recently come to the attention of the SEC and received press coverage (Reilly
2006a, b).2
We use these differences in disclosure prominence to investigate the broader question
of whether firms are rewarded or penalized for openness in disclosing bad news. This
question has been studied previously in the context of earnings warnings. Kasznik and Lev
(1995) find that returns for warning firms are lower than returns for firms with similar risks
and earnings news that do not warn. Their paper led to an article in The Economist, ―Silence
is golden‖ (1994, 91), which concludes: ―But so long as openness is penalized, it is unlikely
to become the voluntary norm.‖ 3 Tucker (2007) replicates Kasznik and Lev‘s (1995) result
using a similar window. She then extends the window to five (or more) months and finds
that warning firms have more unfavorable non-earnings news (e.g., product news, alliances,
and management changes), suggesting that their long-window returns would have been lower
2 Glass Lewis suggests that companies hope investors ―may dismiss a restatement as relatively minor‖ if it is
tucked ―away quietly in the current period‘s results.‖ Subsequent to our test period, the SEC clarified that
restatements ―constitute information that is material to investors and thus needs to be formally disclosed in a
restatement filing clearly labeled as such.‖ (Reilly 2006a). However, a recent working paper by Sharp (2007)
shows that the SEC clarification has not ended the practice. We focus on the press release because we find little
market reaction to the SEC filings. 3 Consistent with The Economist, we use expressions such as penalty for openness or reward for silence
interchangeably.
3
than non-warning firms‘ returns even without the warning. She concludes that firms are not
penalized in the long-run for providing voluntary management forecasts.
The question of whether firms are rewarded or penalized for openness in disclosing
bad news has also been studied in the context of shareholder litigation. Skinner (1994, 1997)
conjectures that litigation concerns could explain why some firms warn investors in advance
of a large negative earnings surprise. Francis, Philbrick and Shipper (1994), however, find
that firms that issue an earnings warning are more likely to be sued than a control sample of
firms that do not warn, which suggests that silence is rewarded. However, Field, Lowry, and
Shu (2005) use a research design that attempts to control for the endogenous relation between
earnings warnings and litigation and conclude that warnings may deter litigation. Most
recently, Rogers and Van Buskirk (2007) find that firms that are sued reduce the frequency of
management forecasts (and conference calls) in the year following the lawsuit filing. This
finding suggests that managers believe reduced disclosure lowers the likelihood of another
lawsuit.
Existing evidence is therefore mixed about whether firms are rewarded or penalized
for openness in disclosing bad news. Our paper provides new evidence, but in the context of
disclosures of accounting irregularities rather than earnings shortfalls. Accounting
irregularities provide an important setting in which to extend this line of research. Since the
passage of the Private Securities Litigation Reform Act of 1995, accounting litigation has
shifted to firms disclosing the need for an accounting restatement and away from class
actions alleging a failure to provide timely warning of an earnings shortfall (Grundfest and
Perino 1997).
4
Our first finding is that the magnitude of the market response to a restatement is
related to press release format. The differences are substantial with average returns for the
three categories of disclosure prominence of -8.3%, -4.0%, and -1.5% for high, medium, and
low prominence, respectively. (The average return for low-prominence disclosures is not
statistically different from zero.) We consider two scenarios in which this return pattern
could occur. In the first scenario, press release prominence conveys the severity of the firm‘s
accounting irregularity, and the return responses simply reflect misstatement severity. This
scenario is consistent with market efficiency. In the second scenario, press release
prominence is not closely related to misstatement severity. This could occur if some
managers provide less prominent press release disclosure in an effort to reduce the stock
price decline and the likelihood of litigation. Specifically, some managers may use medium
(or even low) prominence disclosure for a severe accounting irregularity to avoid attracting
market attention. In this case, the less negative returns associated with less prominent
disclosure would constitute a ―reward for silence.‖ This scenario is consistent with limited
attention theory (Hirschleifer and Teoh 2003).
We evaluate the scenarios by first examining whether the return differences
associated with disclosure prominence decline over time, as investors presumably learn more
about the accounting irregularity. We observe a negative price drift in the 20 days after the
restatement announcement for firms that do not disclose the restatement in a headline (i.e.,
medium and low prominence), and this drift significantly reduces return differences across
the three categories. This response is consistent with the delayed market reaction to less
transparent disclosure predicted by limited attention theory (Hirshleifer and Teoh 2003).
Importantly, over the longer window (-1, +20), returns for each category differ significantly
5
from zero (-7.9%, -6.4%, and -3.2%). This evidence indicates that market participants are
initially misled by the differences in press release format used to announce restatements.
We next examine whether press release format is associated with announcement date
returns (-1, +1) when we control for misstatement severity. Our prominence variable should
no longer be significant if press release format simply reflects misstatement severity, but we
find that prominence is still highly significant. We then use the same model to explain
returns over the longer window (-1, +20) that allows market participants more time to assess
the accounting irregularity, and find that press release prominence is no longer statistically
significant. In sum, our returns tests indicate that firms providing less prominent press
release disclosure of an accounting irregularity are rewarded with less negative returns at the
announcement date; however, much of the reward is short-term.
Next, we examine the association between disclosure prominence and the incidence
of a class action lawsuit. About twenty percent of the companies in our sample are sued as a
result of their accounting restatement, with the frequency declining monotonically across the
three categories of disclosure prominence (27%, 16%, and 0% for the high, medium, and low
prominence firms, respectively). To determine whether the lower rate of litigation for
restatements disclosed without a headline (i.e., medium and low prominence) is due to a less
severe accounting irregularity or a less transparent press release, we estimate a logistic
regression model of the likelihood of litigation that includes controls for misstatement
severity, determinants of lawsuits identified in prior research, and announcement-date
returns. We find that the coefficient of the prominence index is positive and significant,
despite the inclusion of controls for misstatement severity. This finding suggests that press
release format is a separate determinant of lawsuit likelihood, and that managers not
6
providing headline disclosure are rewarded with a lower likelihood of litigation. We find this
result is robust to the treatment of dismissed lawsuits, so in contrast to the return differences,
the effect is permanent.
The results of this study should be of interest to regulators, corporate executives, and
academics with an interest in corporate disclosure. This paper adds to the (limited) evidence
indicating that press release emphasis affects investor responses to the accounting
information reported (e.g. Bowen, Davis and Matsumoto 2005, Gordon et al. 2007). The
paper also provides evidence about whether firms are rewarded or penalized for openness in
disclosing bad news. Prior studies on earnings warnings provide conflicting evidence about
whether the market and legal system exact a penalty for openness. Our paper provides new
evidence of a penalty. Regulators, especially the SEC, may want to standardize press release
disclosure and thereby reduce the large differences in returns and lawsuit outcomes for
substantively similar economic events. Lastly, our findings should be of interest to the
commission formed by Secretary Paulson, in their effort to understand how restatements
impact investors and the capital markets (see introductory paragraph).
The rest of the paper is organized as follows. In Section II, we present the
hypotheses. In section III, we describe the sample. We summarize the multivariate models
used to explain returns and litigation in section IV. In section V, we present the results based
on market returns, including robustness checks and extensions. In section VI, we report
litigation results. We conclude in section VII.
7
II. HYPOTHESES
Guidelines about how to write an effective press release stress the importance of the
headline. Mahoney (1991, 284) comments, ―The headlines drive an idea, instead of simply
identifying a subject.‖ And Feid (2005) states ―Don‘t Bury the News—A headline should
convey the entire key message at the top of the press release. Then the lead paragraph should
reiterate it with the basic facts.‖
To-date, the effect of press release format on market reactions has received little
attention in the accounting literature. A relevant exception is a study by Bowen, Davis and
Matsumoto (2005) which investigates whether investors‘ relative response to GAAP earnings
versus pro forma earnings depends on placement in the firm‘s press release. They find that
the relative strength of the price response to each earnings metric is greater if that metric is
emphasized by its placement in the press release. Using an approach similar to that in
Bowen et al. (2005), we define a three-level index of disclosure prominence. We code as 3
(high prominence) any press release in which the restatement is mentioned in the headline;
we use 2 (medium prominence) for any press release in which the restatement is not
mentioned in the headline but is discussed in some detail within the body of the report;4 and
we assign 1 (low prominence) to those press releases that mention the restatement only in the
footnotes to comparative period results.5
4 Firms in this category typically present other company information in the press release. Fifty-eight percent
present current earnings information, and the remaining firms issue a variety of other announcements, such as a
new credit facility (Sybron Dental 2002), the settlement of past litigation charges (Cognos 2002), and the
adoption of new tax rules (First American Financial Corporation 2000). Kothari, Shu, and Wysocki (2005, 3)
posit that managers often delay disclosing bad news, apparently gambling that subsequent corporate events (i.e.,
future corporate turnaround or restructuring) will allow them to ―bury‖ the bad news.
5 Bowen et al. (2005) use a 4-point scale for each earnings metric: 1 = discussed in the headline; 2 = discussed
in the 1st or 2
nd paragraph; 3 = discussed further down in the body of the release; and 4 = only discussed in the
financial statements provided at the end of the release (i.e., the footnotes). Our index combines scores 2 and 3
into one measure.
8
Concurrent research by Gordon et al. (2007) also examines the effect of press release
characteristics on returns in a three-day window centered on the restatement announcement.
Two of their characteristics are similar to those we consider, and their results are consistent
with our findings. First, they find that firms reporting the restatement in the headline (using a
dummy variable) experience a larger price decline. Second, they find that firms providing
more details about the restatement, including the restatement amount, experience a less
negative market reaction. In other respects, the papers are complementary but quite
different.6
Theory differs about whether the format of a press release would affect investor
response. In an efficient market, security prices respond promptly and fully to all publicly
available information, so it should not matter whether the restatement is in the headline.
However, limited attention theory (Hirshleifer and Teoh 2003) predicts that the speed and
completeness of price reactions are both reduced when information is disclosed in a form that
investors may overlook. Thus, price reactions to restatements that are omitted from the press
release headline (for example, restatement disclosures in an earnings press release) may be
smaller and/or extend over a longer period than headline announcements. We therefore test
the following null hypothesis:
H0 1: The extent of prominence given to an accounting restatement in the press
release does not affect the magnitude of the market‘s response.
Disclosure prominence may also affect the likelihood of shareholder litigation. This
could occur if potential litigants are themselves subject to limited attention. For example,
6 Gordon et al. (2007) focus on determining whether disclosure credibility is associated with market
reactions to restatement announcements. Disclosure credibility consists of situational incentives at the time
of the disclosure (which they argue is homogenous among restating firms), management credibility
(determined by the amount and tone of prior discretionary disclosures), and the characteristics of the
restatement announcement. Their research therefore complements and extends the capital markets analysis
in our paper. Note that Gordon et al. (2007) do not consider whether restatement disclosure practices
influence the likelihood of litigation.
9
class action law firms typically have staff assigned to clip newspaper articles about corporate
events with lawsuit potential, and a headlined event is less likely to be overlooked. We test
the following null hypothesis:
H0 2: The extent of prominence given to an accounting restatement in the press
release does not affect the likelihood of class action litigation.
Note that lawsuit likelihood may increase with the stock price decline that triggers the
lawsuit, and those returns may in turn be driven by disclosure prominence. We control for
this possibility by including announcement-period returns in testing H0 2.
III. SAMPLE SELECTION AND DATA DESCRIPTION
Our sample is taken from the GAO (2003) database, which lists 919 companies that
announced an accounting restatement between January 1, 1997 and June 30, 2002. The GAO
list is restricted to ―accounting irregularities‖ and excludes restatements that arise from
normal corporate activity or simple presentation issues.7 The GAO tracks restatements based
on the press release announcement date, not the date on which adjusted financial statements
are filed. In addition to the company name and announcement date, the database reports the
ticker, entity that prompted the restatement (e.g. auditor, company, SEC, unknown), and the
accounting issue(s) underlying the misstatement (selected from nine categories). The GAO
list of restating companies is important because it influenced passage of the Sarbanes-Oxley
Act of 2002 by showing Congress that accounting irregularities were widespread. About
7 The GAO report (2002, 2) states: ―For the purposes of this report, an accounting irregularity is defined as an
instance in which a company restates its financial statements because they were not fairly presented in
accordance with generally accepted accounting principles (GAAP). This would include material errors and
fraud.‖ The GAO report further states that accounting irregularities include so-called ―aggressive‖ accounting
practices, intentional and unintentional misuse of facts applied to financial statements, oversight or
misinterpretation of accounting rules, and fraud.‖ An important advantage of the GAO list is that the researcher
does not make the judgment about which restatements constitute aggressive accounting practices. A second
advantage is that the list is publicly available, so other researchers can extend (or replicate) research results.
10
10% of all companies listed on the NYSE, NASDAQ and Amex have a restatement cited in
the database (GAO 2002, 16).
Table 1 reconciles our sample to the GAO list. We require that our sample firms
have short-interest data, and drop 452 firms (from the 919) that do not meet this requirement.
Most of the eliminated firms are small or trade over-the-counter. We use the level of short
interest to control for information available to a sophisticated investor about accounting
problems. Desai et al. (2006) and Efendi et al. (2005) show that the level of short interest
increases before a restatement announcement. Of direct importance to our use of short
interest, Efendi et al. (2005) show that the entire increase occurs for companies with severe
accounting irregularities. 8
Our sample is further reduced by 61 firms for which we could not
locate the press release; 9 17 firms that do not have return information on CRSP; and eight
firms whose returns are in the top or bottom one percent of the cumulative market-adjusted
returns distribution. The final sample consists of 381 restatement firms.
<Insert Table 1 Here>
We obtain information about litigation from the Securities Class Action
Clearinghouse (http://securities.stanford.edu/). This website covers more than 2500 issuers
that have been named in federal class action securities fraud lawsuits since passage of the
Private Securities Litigation Reform Act of 1995. It includes information about complaints,
briefs, filings, and other litigation-related materials. We review this information to determine
8 We would like to acknowledge these authors for allowing us to use their proprietary database, which contains
short interest for firms listed on the NYSE, AMEX, and NASDAQ. 9 To ensure that we are capturing management‘s disclosure choices, we use press releases sent out over either
PR Newswire or Business Wire, when available. These services distribute the exact press release issued by a
firm, while other media outlets may alter the press release to conform to their format guidelines, space
requirements, etc. Almost all of the press releases in our sample are from one of these two sources.
11
whether any litigation filed against a sample firm is a direct consequence of the accounting
restatement.
Table 2 provides descriptive information about the sample. Panel A shows that
restating firms are widely distributed across 49 two-digit industries. To provide insight into
whether some industries have a higher than normal rate of restatements, we compare the
industry distribution of the sample to the Compustat population. The differences are modest:
Business services provides 15.49% of the restatement sample (59 of the 381 companies),
which is higher than the industry‘s 11.01% representation in Compustat. Only five other
industries provide more than 12 sample firms, and only two of them have at least a three
percent difference from Compustat: 1) Industrial and commercial machinery and computer
equipment (8.66% vs. 4.89%); and 2) electronic and other electrical components, except
computer equipment (8.66% vs. 5.58%). Table 2, Panel B reports the distribution of the
sample announcements over time. Restatement frequency increases over the sample period,
peaking at 117 restatements in 2001.10
Table 2, Panel C provides a cross-tabulation of
accounting issue with the three categories of disclosure prominence. Because some firms
restate for more than one accounting issue, the 442 accounting issues exceed the 381
companies in the restatement sample. Revenue recognition is the most common type of
accounting misstatement for each of the three prominence categories. Of particular note, a
large portion of the companies providing footnote-only disclosure restate revenue (25 of the
37). This is the primary reason that the 2.17 prominence score for revenue restatements is
low compared to the full sample average of 2.33. (Later in the paper, we report robustness
tests to determine the effect of revenue restatements on our findings.)
<Insert Table 2 Here>
10
Recall that the GAO dataset ends June 30, 2002 so our sample does not include all restatements for 2002.
12
IV. MULTIVARIATE MODELS
For H0 1, we use an OLS regression model to investigate whether disclosure
prominence is associated with the magnitude of returns around the restatement
announcement. For H0 2, we use a logistic regression to examine whether disclosure
prominence influences the likelihood of class action litigation. For each hypothesis, we
control for factors other than press release format that could influence the response to the
Restatement firms in General Accounting Office studya 919
Less: firms without short selling informationb (452)
Less: firms without press release information available (61)
Less: firms without CRSP data (17)
Less: possible outliersc (8)
Final Sample 381
a The General Accounting Office (GAO 2002) prepared a report for the U.S. Senate
Committee on Banking, Housing and Urban Affairs, chaired by Senator Sarbanes.
This report identifies 919 firm restatements from January 1, 1997 to June 30, 2002 that
correct ―accounting irregularities.‖ (See text for further description.) b Short selling information is taken from a proprietary database which contains short
interest for firms listed on the NYSE, AMEX, and Nasdaq. Most of the companies
deleted are small and trade in the over-the-counter (OTC) market. c Observations whose abnormal returns around the restatement date fall in the top or
bottom 1% of the distribution are deleted as possible outliers.
35
TABLE 2
Sample Description
Panel A: Distribution of Restatement Firms by Industry
SIC
Industry Description
Sample
Firms
Sample
Percentage
Compustat
Percentagea
10 Metal Mining 1 0.26 1.32%
13 Oil and gas extraction 5 1.31 3.76%
15 General building contractors 2 0.52 0.54%
16 Heavy construction other than building
construction contractors
3 0.79 0.23%
20 Food and kindred products 10 2.62 1.74%
23 Apparel and other textile products 4 1.05 0.74%
24 Lumber and wood products 2 0.52 0.39%
25 Furniture and fixtures 2 0.52 0.42%
26 Paper and allied products 4 1.05 0.74%
27 Printing, publishing and allied industries 4 1.05 1.09%
28 Chemicals and allied products 22 5.77 5.69%
29 Petroleum refining and related industries 2 0.52 0.47%
30 Rubber and misc. plastics products 4 1.05 0.98%
32 Stone, clay, glass and concrete products 2 0.52 0.56%
33 Primary metal industries 2 0.52 1.02%
34 Fabricated metal products; except machinery
and transport equip.
3 0.79 1.18%
35 Industrial and commercial machinery and
computer equipment
33 8.66 4.89%
36 Electronic and other electrical components;
except computer equip.
33 8.66 5.58%
37 Transportation equipment 6 1.57 1.54%
38 Measuring, analyzing and controlling
instruments
26 6.82 4.77%
39 Misc. manufacturing industries 2 0.52 0.99%
42 Motor freight transportation and warehousing 2 0.52 0.58%
44 Water transportation 1 0.26 0.40%
47 Transportation services 1 0.26 0.34%
48 Communications 6 1.57 3.98%
49 Electric, gas and sanitary services 12 3.15 2.99%
50 Wholesale trade-durable goods 8 2.10 2.22%
51 Wholesale trade-non-durable goods 4 1.05 1.37%
52 Building materials, garden supply and mobile 1 0.26 0.20%
53 General merchandise stores 5 1.31 0.48%
54 Food stores 3 0.79 0.60%
55 Automotive dealers and service stations 1 0.26 0.29%
56 Apparel and accessory stores 10 2.62 0.54%
57 Home furniture, furnishings and equipment
stores
7 1.84 0.47%
58 Eating and drinking places 2 0.52 1.35%
59 Misc. retail 5 1.31 1.71%
60 Depository institutions 29 7.61 8.53%
36
61 Non depository credit institutions 1 0.26 1.45%
62 Security, commodity brokers and services 3 0.79 1.03%
63 Insurance carriers 11 2.89 2.12%
65 Real estate 2 0.52 1.19%
67 Holding and other investment offices 10 2.62 6.97%
73 Business services 59 15.49 11.01%
78 Motion pictures 1 0.26 0.93%
79 Amusement and recreation services 3 0.79 1.06%
80 Health services 6 1.57 1.67%
82 Educational services 4 1.05 0.29%
87 Engineering and management services 4 1.05 1.80%
99 Unclassified establishments 8 2.10 1.35%
381 100% 100%
Panel B: Distribution of Announcement Dates
Time Period Number of Firms Percentage of Sample
Calendar 1997 24 6.3%
Calendar 1998 20 5.3%
Calendar 1999 75 19.7%
Calendar 2000 87 22.8%
Calendar 2001 117 30.7%
To June 30, 2002 58 15.2%
Panel C: Cross Tabulation of Accounting Issue by Press Release Prominence
Coded as
High
Prominence
(3)
Medium
Prominence
(2)
Low
Prominence
(1)
Frequencyb
Average
Prominence
Scorec
Revenue 52 81 25 158 2.17
Cost or expense 28 42 3 73 2.34
Restructuring, assets or
inventory 26 24 3
53
2.43
Other 17 13 2 32 2.47
Acquisitions and
mergers 14 13 1 28 2.46
Securities related 12 12 0 24 2.50
IPR&D 4 15 2 21 2.10
Reclassification 8 10 1 19 2.37
Related-party
transactions 6 7 0 13 2.46
Unspecified 6 3 0 9 2.67
Loan-loss 6 1 0 7 2.86
Tax related 2 3 0 5 2.40
Total Obs. 181 224 37 442 2.33
a
Based on Compustat companies in 2000.
b Each restatement can be caused by one or more accounting issues. Therefore, the total number of accounting issues
(442) is greater than the number of firms in our sample (381). c Disclosure emphasis is coded as 3 (high prominence) for any press release that mentions the earnings restatement
in the headline. Disclosure quality is coded 2 (medium prominence) for press releases that do not mention the
37
restatement in the headline, but that discuss the restatement in some detail within the body of the report. A code of 1
(low prominence) is assigned to press releases that only mention the restatement in the footnotes of the report.
Average Prominence Score = [(# High Prominence)*(3) + (# Medium Prominence)*(2) +
(# Low Prominence)*(1)] / # Total
38
TABLE 3
Mean Abnormal Returns Around the Restatement Announcement for the Full
Variable Definitions: CAR = cumulative abnormal return, calculated as the raw stock return minus the CRSP equally-weighted market portfolio return measured over the
three-day window (-1,+1) with day 0 indicating the restatement announcement ; LITIGATION = 1 if a class action lawsuit was filed against the firm as a response to the
restatement of their financial statements and 0 otherwise; PROMINENCE INDEX takes on the value of either 1, 2, or 3. Disclosure emphasis is coded as 3 (high
prominence) for any press release that mentions the earnings restatement in the headline. Disclosure quality is coded 2 (medium prominence) for press releases that do not
mention the restatement in the headline, but that discuss the restatement in some detail within the body of the report. A code of 1 (low prominence) is assigned to press
releases that only mention the restatement in the footnotes of the report. RESTATEMENT AMT = 1 if the press release includes information on the amount of the
restatement and 0 otherwise; SEC = 1 if the press release indicates an SEC inquiry and 0 otherwise; FRAUD = 1 if the press release mentions an allegation of fraud and 0
otherwise; REVENUE = 1 if any part of the restatement is due to revenue recognition problems and 0 otherwise; RULE CHANGE = 1 if a change in accounting rules was
mentioned in the press release as the reason for the restatement and 0 otherwise; COUNT = the number of different accounting issues, as reported in the GAO database;
SHORT INTEREST = the percent of short interest [100 * (number of shares sold short)/(total shares outstanding)] as of the month prior to the restatement announcement;
SHARE TURNOVER = [1-Π(1-volume tradedt/total sharest)], accumulated over the 1-year period ending on the second day prior to the restatement announcement date;
MKTCAP = market capitalization (closing stock price * common shares outstanding) measured as of the end of the fiscal year prior to t.
41
TABLE 5
Multivariate Analysis of Abnormal Returns around Restatement Announcement
Panel A: Model with Short CAR Window (-1,+1)
CAR = α + β1PROMINENCE INDEX+ β2RESTATEMENT AMT + β3SEC + β4FRAUD + β5REVENUE +