The Effect of Auditors’ Incentives on the Disclosure Prominence of Identified

Misstatements

Brant Christensen
Price College of Business
University of Oklahoma

Roy Schmardebeck
Haslam College of Business
The University of Tennessee, Knoxville

Timothy Seidel
BYU Marriott School of Business
Brigham Young University

June 2019

Acknowledgements: We are grateful for helpful comments and suggestions from Eric Condie,
Lauren Cunningham, Keith Jones, Sarah Rice, Michael Wilkins, Michael Willenborg, workshop
participants at the University of Tennessee, Knoxville, and conference participants at the
Brigham Young University Accounting Research Symposium. We thank Lyndon Orton, Taylor
Hauser, Changwei Song, Andrew Glover, and Logan Buchheit for their assistance hand
collecting out-of-period adjustment information. Timothy Seidel gratefully acknowledges
financial support from the Glen D. Ardis Fellowship at Brigham Young University.

0
 The Effect of Auditors’ Incentives on the Disclosure Prominence of Identified
Misstatements

Abstract: This paper investigates how auditor incentives affect the disclosure of misstatements
from previously audited financial statements. We identify a sample of misstatement disclosures
that vary in prominence between non-reliance restatements, revision restatements, and out-of-
period adjustments. Requiring that the auditor during the misstated period be the same as at the
misstatement disclosure date, we find that greater auditor engagement risk (i.e., the risk of
litigation and reputational damage) and greater client importance to an audit office are associated
with less prominent disclosure of prior period misstatements. We find consistent results when the
magnitude of the misstatement is held constant across disclosure types. Further tests show that
this association does not exist when misstatements are clearly material but does occur when the
materiality of the misstatement is close to auditors’ quantitative materiality thresholds. Our
results are supported by several analyses used to rule out alternative explanations. These findings
suggest that while the potential for litigation and reputation risk encourage auditors to provide
higher quality during the current period, those same incentives align with the desire to appease
the client when auditors are faced with acknowledging failed audits from prior periods. These
results provide important insights regarding auditors’ response to incentives.
Keywords: Litigation, reputation, restatements, audit quality, incentives, engagement risk
1. Introduction
Auditors have a responsibility to provide reasonable—but not absolute—assurance that

financial statements are free of material misstatement. Accordingly, auditors necessarily bear

some risk that the opinion they provide could subsequently prove inaccurate, potentially

exposing them to financial penalties, future litigation, and the loss of reputation. Exposure to

litigation and reputation risk are the main components of what is referred to as engagement risk

(Messier, Glover, and Prawitt 2014). Prior research shows that higher engagement risk generally

incentivizes auditors to improve audit quality (Geiger and Raghunandan 2001; Lee and Mande

2003; Venkataraman, Weber, and Willenborg 2008). Importantly, this incentive to avoid

subsequent litigation and reputational damages tends to overcome the opposing incentive to cater

to important clients’ preferences for more aggressive accounting choices (Craswell, Stokes, and

Laughton 2002; Reynolds and Francis 2000; Chung and Kallapur 2003). However, when

auditors are faced with evaluating and disclosing misstatements related to their previously issued

audit opinions, auditors may prefer the disclosure method that minimizes their immediate

litigation and reputational damage, thus bringing their incentives into alignment with client

preferences. Thus, we examine whether auditor incentives to 1) avoid litigation and reputational

damage and 2) acquiesce to important clients are associated with less prominent disclosure of

misstatements from their clients’ previously audited financial statements.

Although the decision to correct previously issued financial statements ultimately lies

with management, auditors play an important role in this decision for at least two reasons. First,

current and prior year errors are often identified during the audit (e.g., Bedard and Graham

2011).1 Second, when the auditor becomes aware that the financial statements on which s/he

1
We note that management’s own incentives to avoid litigation and subsequent employment termination are aligned
with those of the auditor, in that increased risk of subsequent negative outcomes also incentivizes management to
prefer less prominent disclosure (Desai, Hogan, and Wilkins 2006). We control for management’s exposure to

1
opined are materially misstated, auditing standards require the auditor to discuss the matter with

the client and take action to prevent future reliance on the audit report, which includes advising

the client to make appropriate disclosure if the misstatement is material.2 However, materiality

thresholds are based on professional judgment and differ across client- and audit firm-specific

situations (Eilifsen and Messier 2014). Research finds that audit materiality plays a key role in

determining whether prior period accounting errors result in restatements or adjustments (Acito,

Burks, and Johnson 2009).3

This subjectivity in materiality judgments enables manager and auditor incentives to

influence how misstatements are disclosed. All else equal, managers likely prefer less prominent

disclosure of prior period errors (Files, Swanson, and Tse 2009). On the other hand, auditors

exercising due professional care would be less likely to encourage or support client decisions to

disclose misstatements less prominently, unless clearly immaterial. However, the presence of

strong auditor incentives to avoid litigation, reputational damage, and the loss of an important

client could lead auditors to align their disclosure preferences with those of management,

particularly when the disclosure of the misstatement retracts reliance on the previously issued

audit opinion.4 Consistent with this notion, Lynn Turner, former Chief Accountant at the

Securities and Exchange Commission stated that “…once an [audit] firm has issued a report on

dissemination of negative news through analyst following. In untabulated analysis, our results also hold after
controlling for executives’ stock-based compensation incentives (Core and Guay 1999). Additionally, we find that
our results regarding auditor incentives continue to hold even when management incentives to avoid prominent
disclosures are lower. Finally, while we recognize that we cannot completely disentangle auditors’ and
management’s incentives, subsequent tests involving auditors unconnected to the misstated period provide further
support that our results relate to auditor incentives rather than management incentives.
2
See PCAOB AS 2905.
3
While Acito et al. (2009) examine the impact of materiality on error disclosures, our examination focuses on
whether auditor incentives influence the way that companies correct and disclose prior period errors while
controlling for, or matching on, the materiality of the error.
4
Our arguments suggest that more prominently disclosed misstatements are associated with an increased frequency
of litigation against the auditor and subsequent client losses. We provide evidence supporting this premise in Section
4.2.

2
the financial statements of a company, there is an inherent conflict in later concluding that the

financial statements were wrong” (PCAOB 2011). Thus, we predict that as auditors’ ex-ante

incentives to avoid prominent misstatement disclosure increases, the disclosure prominence of

misstatements related to previously audited financial statements decreases.

We examine our prediction by testing the association between auditors’ incentives to

avoid litigation and reputation damage, auditors’ incentives to appease important clients, and the

subsequent disclosure prominence of misstatements from previously audited financial

statements. We examine three channels through which misstatements are revealed to the public.

The most prominent form of misstatement disclosure comes through filing an 8-K with the

Securities and Exchange Commission (SEC) and filing restated financial statements. This 8-K

requirement is triggered when it is determined that the previously issued financial statements and

related audit report should no longer be relied upon by the investing public (i.e., non-reliance

restatements). In this instance, amended financial statements are filed with the SEC along with a

revised audit opinion. The next most prominent form is a ‘revision restatement’, in which prior

period numbers are restated within the current period’s comparative financial statements instead

of issuing a standalone amendment to a prior filing.5 These misstatements are considered

immaterial to each individual financial reporting period but are cumulatively material in the

current set of comparative financial statements. Thus, the auditor does not need to issue a revised

audit opinion regarding prior financial reporting periods.

Finally, the least-prominent form of disclosure is relatively new to the academic

accounting literature: an out-of-period adjustment, in which misstatements from prior periods are

charged to current period accounts, resulting in no restatement or revision of prior period

5
Some prior literature refers to these revision restatements as “little r” restatements (e.g., Tan and Young 2015).

3
financial statement account balances. These are misstatements that are considered immaterial to

each individual financial reporting period and cumulatively immaterial to the current reporting

period. Because the prior individual financial reporting periods were not materially misstated,

there is no need for the auditor to issue a revised audit opinion. While out-of-period adjustments

are inherently smaller in magnitude than either form of restatement, Choudhary, Merkley, and

Schipper (2017) find that out-of-period adjustments are predictive of future material and

immaterial errors and thus still constitute meaningful adjustments.

To perform our tests, we identify a sample of 2,868 restatement and out-of-period

adjustment disclosures from 2004 through 2014. We require that the auditor during the misstated

period be the same as the auditor during the disclosure period, thus ensuring that auditors in our

sample have the theorized incentives to seek less prominent disclosure methods regarding errors

missed during prior period audits. Consistent with our hypotheses, we find less prominent

disclosures of prior period errors, on average, when auditors face higher engagement risk or

when the client in question is more important to the audit office. Further, we obtain these results

controlling for various characteristics of the misstatement including its magnitude, income effect,

duration, and timeliness of the disclosure, and management’s exposure to the quick

dissemination of negative news. We also find that these results hold after matching observations

across the three disclosure types based on the magnitude of the misstatement’s net income

impact. Thus, our results are not due to fundamental differences in error magnitude between the

disclosure types.

In additional analysis, we find that the effects are present among misstatements whose

magnitudes are most uncertain in terms of materiality (i.e., near auditors’ overall quantitative

materiality thresholds). However, we do not find evidence of an association between disclosure

4
prominence and auditor incentives when the magnitude of the misstatement is clearly material.

Thus, when the quantitative materiality of the misstatement is less obvious, but still nontrivial,

auditor incentives to avoid litigation and reputation damage and appease important clients are

associated with less-prominent disclosure channels.

We perform several analyses to alleviate concerns about alternative explanations. First,

we re-perform our analyses on a sample of misstatement disclosures where the audit firm

changed after the misstated period (i.e., the auditor at the time of the disclosure differs from the

auditor during the misstated period). Because auditors uninvolved with the misstated period

should not be concerned about the potential for damaged reputation or increased litigation risk

when considering the disclosure of errors missed by the previous auditor, we do not expect an

association between auditor incentives to avoid prominent misstatement disclosure and the

disclosure prominence of misstatements in this subsample. Consistent with our expectations, we

do not find a significant association between auditor incentives and disclosure prominence in this

subsample. Taken together with our main results, these results are important because they

provide evidence that our documented results pertain to how the auditor’s incentives affect the

subsequent disclosure prominence of identified misstatements, not management’s.

Another potential explanation for our results is that increased engagement risk leads

auditors to increase effort in the current year, thereby decreasing the likelihood of a large error

going undetected and reducing the severity of any subsequent adjustments. To address this

possibility, we limit our sample to company-year observations with lower than expected audit

fees in the latest misstated year. To the extent that abnormal (or unexpected) fees capture audit

effort (Blankley, Hurtt, and MacGregor 2012), the observations in this subsample demonstrate

abnormally lower levels of auditor effort and thus reduces the possibility that the association

5
between engagement risk and subsequent disclosure prominence is due to greater auditor effort

in the misstated year. Within this subsample, we continue to find results related to engagement

risk, suggesting that our main results are not an artifact of higher audit effort during misstatement

years that would reduce the size of any errors detected in subsequent years.

Our study makes several important contributions. First, if auditors are economic agents

and gatekeepers to the public interest, then it is important to understand the incentives that guide

their actions (Antle 1982). We find evidence that auditors behave in accordance with their

incentives to avoid prominently disclosing misstatements. However, it is important to note that

auditor incentives do not affect the disclosure prominence of clearly material misstatements.

Second, we contribute to the stream of research examining whether auditors cater to important

clients. Although experimental research suggests this is a concern (Hackenbrack and Nelson

1996; Kadous, Kennedy, and Peecher 2003), prior archival research generally does not find on-

average evidence of auditors catering to important clients (Craswell et al. 2002; Reynolds and

Francis 2000; Chung and Kallapur 2003). We contribute to this research by identifying a setting

in which auditors appear to concede to preferences of important clients – supporting or

encouraging less prominent disclosure of misstatements of previously audited work. Finally, it

has been noted that empirical evidence regarding the effects of reputation on audit quality is

limited (DeFond and Zhang 2014). Our results show that reputational concerns are relevant to

auditor decision-making.

2. Background and Hypothesis

In this section, we first discuss the influence of auditor engagement risk and economic

bonding on audit quality in the current year. We then discuss the various methods used to

disclose misstatements related to prior periods. Finally, we develop our hypothesis examining

6
whether auditor incentives to avoid litigation and reputation damage and to appease important

clients influence the disclosure prominence of misstatements from previously audited financial

statements.

2.1 Engagement Risk and Contemporaneous Audit Quality

Because auditors do not provide absolute assurance that the company’s financial

statements are fairly stated, auditors incur some level of risk they will issue an unqualified

opinion in the presence of an undetected material misstatement. When this occurs, auditors are

then exposed to litigation and reputation damage, collectively referred to as engagement risk

(Messier et al. 2014). These risk factors influence client acceptance/continuance decisions, but

once a client is accepted, auditors respond to higher engagement risk by lowering the acceptable

level of audit risk (i.e., increasing audit effort) (Bedard and Johnstone 2004; Johnstone and

Bedard 2003, 2004). For example, Hackenbrack and Nelson (1996) find experimental evidence

that auditors respond to high engagement risk by requiring more conservative application of

accounting standards.

Consistent with this notion, most theoretical research investigating one component of

engagement risk—litigation risk—suggests that higher (lower) litigation risk is associated with

higher (lower) contemporaneous audit quality (Chan and Pae 1998; Liu and Wang 2006;

Schwartz 1997; Zhang and Thoman 1999). Further, a wide range of empirical studies show that

auditors charge higher audit fees for firms with higher litigation risk, suggesting greater auditor

effort (Choi, Kim, Liu, and Simunic 2008; Choi, Kim, Liu, and Simunic 2009; Magnan 2008;

Seetharaman, Gul, and Lynn 2002; Simunic 1980; Simunic and Stein 1996).6 Similarly, a

number of studies find that auditors respond to higher (lower) litigation risk by becoming more

6
Auditors plan additional audit hours in the presence of increased litigation risk, which implies greater auditor
effort and not just a litigation risk premium (see Simunic and Stein 1996).

7
(less) conservative in financial reporting (Geiger and Raghunandan 2001; Lee and Mande 2003;

Venkataraman et al. 2008). Taken together, research supports the notion that auditors increase

contemporaneous audit effort, thereby lowering the level of audit risk they are willing to accept,

in the presence of higher litigation risk.

DeAngelo (1981) argues that the other component of engagement risk—the risk of

reputational damage—should have a similar effect on acceptable audit risk.7 Empirically, Keune

and Johnstone (2012) use audit fees to proxy for the risk of auditor reputation loss, arguing that

large, high profile clients pay higher audit fees, thus exposing the audit firm to greater potential

reputation loss should the audit of those clients fail. The authors find that as the risk of reputation

loss increases, auditors are less likely to waive current-period misstatements (i.e., are more likely

to request management to correct the misstatement). Thus, the potential exposure to adverse

publicity and loss of reputation leads auditors to lower acceptable audit risk (i.e., achieve higher

audit quality) when considering identified errors related to the current year.

2.2 Client Importance and Contemporaneous Audit Quality

A large stream of research examines the proposition that auditors report more favorably

for larger clients or acquiesce to these clients’ accounting preferences due to economic bonding.

Although Kadous et al. (2003) find experimental evidence consistent with this idea, archival

evidence either fails to find conclusive evidence that economic bonding leads to more favorable

going-concern reporting or increased managerial discretion in accruals (Craswell et al. 2002;

Chung and Kallapur 2003) or finds more conservative reporting (Reynolds and Francis 2000; Li

7
Concern about negative outcomes following reputation damage are well-founded. Weber, Willenborg, and Zhang
(2008) examine a high-profile accounting scandal in Germany and find that clients of the audit firm involved in the
scandal suffered negative abnormal stock returns and the audit firm itself lost a number of clients, presumably due to
the scandal’s detrimental effect on the firm’s reputation. Similarly, Skinner and Srinivasan (2012) show that one
audit firm lost approximately 25 percent of its client base following an accounting fraud in Japan.

8
2009). Reynolds and Francis (2000) suggest that greater litigation risk and exposure to reputation

loss are “sufficient to motivate auditors to be independent, despite the presence of economic

dependence” (Reynolds and Francis 2000, 377). Findings in Chen, Sun, and Wu (2010) shed

additional insight. They find evidence in China that individual auditors were less likely to issue

modified audit opinions to important clients when the institutional environment for investor

protection was weaker, suggesting these auditors likely compromised their objectivity for these

clients. However, when investor protection was strengthened through litigation and regulatory

reform in 2001, the propensity to issue modified audit opinions to important clients increased.

Although this finding is pronounced at the individual auditor level, it aligns with U.S. based

evidence that exposure to litigation and reputation loss (i.e., engagement risk) can offset (or

overcome) auditor incentives to report more favorably for important clients.

2.3 Methods of Disclosing Identified Misstatements

The way identified misstatements are disclosed varies in practice in part because of a lack

of “bright line” rules on what constitutes a material misstatement. When an identified

misstatement is deemed material and the previously issued financial statements and audit report

should no longer be relied upon, SEC rules require companies to file an 8-K with Item 4.02

within four business days of the non-reliance judgment and an amended filing with restated

financial statements and an amended audit opinion.

However, some companies choose to restate prior period financial statements in a

‘revision restatement’ rather than issuing a press release and related 8-K filing. A revision

restatement refers to instances in which prior periods are materially correct on a standalone basis,

but an accumulation of errors has become material in the current year. Thus, this cumulative

error is corrected through prior year accounts and displayed in the current periodic filing without

9
issuing a separate amended filing or revised audit opinion for the misstated financial statements.

Finally, other companies simply correct prior period errors by making and disclosing out-of-

period adjustments to the current year’s financial statements. In this case, prior period statements

and figures are neither restated nor revised. Instead, the misstatement is charged to current-year

accounts. Even though out-of-period adjustments are likely to be smaller in magnitude than some

restatements, recent research finds them relevant to investors and related to future financial

reporting quality (Choudhary et al. 2017). Additionally, the fact that these errors eventually

required correction and disclosure of any kind signals their importance in the eyes of the issuer

and its auditor.

2.4 Hypothesis Development

Prior research findings suggest that auditor’s response to engagement risk leads to higher

quality audits, thereby overcoming incentives arising from economic dependence on important

clients (Reynolds and Francis 2000; Chung and Kallapur 2003). However, our study examines

how these incentives influence decisions of how to disclose previously unreported misstatements

related to the auditor’s prior audit opinion. This examination is important because unlike

previous research settings where auditor incentives related to engagement risk and economic

bonding are in opposition, in our setting these incentives align with each other and with

management preferences to avoid prominent disclosure of prior period errors. Although

materiality should be the primary consideration for how to disclose identified misstatements,

discretion in materiality judgments enables manager and auditor incentives to influence these

disclosure decisions. For instance, managers are likely motivated to disclose misstatements less

prominently to avoid a negative stock market reaction (Palmrose, Richardson, and Scholz 2004;

Files et al. 2009) and reduce the likelihood of management turnover (Desai et al. 2006).

10
Although the financial statements are the responsibility of management, Bedard and Graham

(2011) find that most internal control deficiencies are identified by audit testing, and the same

fact pattern likely extends to the identification of misstatements. Because auditing standards

require the auditor to discuss the prior period misstatement with the client and take action to

prevent future reliance on the audit report if the misstatement is deemed material, auditor

incentives likely influence auditors’ materiality assessment.

Prior research in organizational behavior and human performance investigates whether

individuals will revise past decisions or behavior when faced with negative consequences (Staw

1976). This research finds that individuals generally increase their commitment to prior decisions

when contradictory information is subsequently identified. This escalation of commitment results

in individuals biasing their future actions to support their initial decision—in this case, that the

original financial statements were fairly stated.8 Increased engagement risk and its associated

costs should escalate auditors’ commitment to previously audited work.

Regarding the reputation component of engagement risk more specifically, auditors

develop their reputation over time by consistently producing high quality audits. However, the

disclosure of a misstatement generally indicates the error was undetected in a previous audit.

Kanodia, Bushman, and Dickhaut (1989) argue that changing prior decisions informs the public

that the decision maker’s judgments are fallible, thereby damaging the individual’s reputation.

Prior audit literature supports this notion in that auditors are more likely dismissed and lose

market share following restatement announcements (Hennes, Leone, and Miller 2014; Swanquist

8
Although the argument could be made that these adjustments are clearly immaterial, Choudhary et al. (2017) find
that out-of-period adjustments are predictive of future material and immaterial errors. This finding could suggest
that auditors may be more willing to agree to an immaterial adjustment related to prior periods (i.e., less than the
auditor’s materiality threshold) that is perhaps smaller than the true prior period error to avoid more prominent
disclosure, which ultimately leads to needed adjustments in the future.

11
and Whited 2015). When the client is faced with the decision to announce a misstatement of

previously audited financial statements, auditors may prefer the disclosure method that

minimizes reputational damage. Given that investor attention is limited (Hirshleifer and Teoh

2003) and that prominence or placement of disclosure can influence the weight placed on

extracted information (Hirst and Hopkins 1998; Maines and McDaniel 2000), more prominent

disclosures of misstatements are likely to be more damaging to an auditor’s reputation. As

such—and holding the magnitude of the misstatement constant—we expect that auditors will

prefer less prominent disclosure of identified misstatements when ex-ante potential reputational

costs are high.

A prominently disclosed misstatement can also increase the likelihood of costly legal

action. The stock price drop that often accompanies the disclosure of a misstatement can lead

investors to seek legal recourse in the form of class action lawsuits, including against the audit

firm (e.g., Hogan, Lambert, and Schmidt 2014; Palmrose and Scholz 2004; Rice, Weber, and Wu

2015). As such, a prominently disclosed correction of a misstatement is expected to increase the

likelihood of costly legal action, even if settled out of court.

These arguments suggest that when auditor’s ex-ante litigation and reputational risk are

high, auditors are likely to encourage or to support clients’ decisions to disclose identified

misstatements less prominently. Additionally, auditor and management incentives align in this

instance to avoid prominent disclosure of prior period misstatements, thus enabling auditors to

cater more to important clients. As such, our hypothesis is as follows:

Hypothesis: Auditor incentives to avoid litigation and reputational damage and appease
important clients are negatively associated with the disclosure prominence of identified
misstatements.

12
3. Research Design and Measurement

3.1 Research Design

We use the Audit Analytics Advanced Restatement database to identify misstatements

that are subsequently disclosed through 1) Item 4.02 non-reliance restatements, 2) other

restatements occurring through a periodic filing (i.e., revision restatements), or 3) out-of-period

adjustments. We then limit our sample to disclosures in which the auditor during the misstated

period(s) is the same at the date of the misstatement disclosure.9 This requirement is key,

because an auditor unaffiliated with the misstated financial statements would not have the same

incentive to avoid admitting that prior-period financial statements were misstated.

3.2 Auditor Engagement Risk

We measure the construct of auditor engagement risk by using previously established

proxies for its main components: auditor litigation risk and the risk of reputation loss. We first

measure an auditor’s risk of litigation using three proxies. We first follow Shu (2000) by

estimating a logistic model of the likelihood of auditor-related litigation. This measure of

auditor-specific litigation captures litigation risk at the level of the client, thereby capturing the

litigation risk faced by the auditors that could potentially influence the misstatement disclosure

decision. The litigation risk score computation is as follows, consistent with Table 3 of Shu

(2000):

LitRisk_Shu = -10.049 + 0.276*Size + 1.153*INV + 2.075*REC + 1.251*ROA –

0.088*Current + 1.501*Leverage + 0.301*SalesGrowth – 0.371*Returns –
2.309*RETVOL + 0.235*Beta + 1.464*Turnover + 1.060*Delist + 0.928*TECH +
0.463*QUALOPIN (1)

We define the variable definitions in the Appendix. To test the effect of auditors’

9
Misstated periods for restatements are obtained from Audit Analytics using the restatement begin/end dates.
Misstated periods for out-of-period adjustments were obtained by manually searching and reading the context and
details of the adjustment in the SEC filing that includes the disclosure.

13
consideration of litigation risk in deciding how to disclose identified misstatements, we must

measure the auditor’s litigation risk before the misstatement disclosure. Therefore, we measure

LitRisk_Shu at the most recent fiscal year-end before the misstatement disclosure.10 Finally, we

rank the LitRisk_Shu score into deciles consistent with prior research (e.g., Iliev, Miller, and

Roth 2014), in which a higher score represents a higher likelihood of litigation against the

auditor.

Our second measure of litigation risk follows model 2 of Table 7 in Kim and Skinner

(2012) estimated as follows:

LitRisk_KS = -7.718 + 0.180*FPS+0.463*Sizet-1 + 0.553*SalesGrowtht-1 – 0.498*Returns

– 0.359*Returns Skewness + 14.437*RETVOL + 0.0004*Turnover (2)

We define the variable definitions in the Appendix. Similar to the Shu (2000) measure, we

measure LitRisk_KS at the most recent fiscal year-end before the misstatement disclosure and we

rank the LitRisk_KS score into deciles. Our third measure of litigation risk (InstOwn%) is the

percentage of company stock held by institutional investors. Cassell, Drake, and Dyer (2018)

document a positive association between audit fees and institutional ownership suggesting that

auditors are sensitive to litigation risk related to their clients having a greater number/proportion

of sophisticated investors who rely on the audited financial reports.

We also measure auditor reputation risk using three accepted proxies. First, following

Keune and Johnstone (2012), we use a client’s audit fees in the year prior to the misstatement

disclosure as a proxy for auditor reputation risk (Fees). Second, following the theoretical

10
For example, if a company announces during its 2014 fiscal year that prior periods were misstated, we measure
components of engagement risk at the 2013 fiscal year-end. This approach captures auditors’ concerns about
litigation and reputation risk before the disclosure decision is made. It is important to note that we cannot measure
components of engagement risk in the same year as the misstatement is disclosed because doing so would artificially
increase LitRisk_Shu (which is partially measured by stock returns, which are directly affected by error disclosures)
and Fees (which increase during error disclosure years).

14
argument of DeAngelo (1981), we use an indicator for whether the auditor is one of the Big 4

(BigN). Finally, Francis and Michas (2013) find evidence that some audit offices have consistent

audit quality problems, but larger offices perform higher quality audits to respond to the risk of

losing reputational capital, consistent with higher audit quality among larger Big 4 auditor

offices (Francis and Yu 2009). Following this argument, we also measure reputation risk using

OfficeSize measured in the year preceding the misstatement disclosure.

We acknowledge that auditor litigation risk and reputational concerns are related and

very challenging to separate empirically (DeFond and Zhang 2014). For example, increased

visibility resulting from litigation can lead to reputational damage. Additionally, as the risk of

public exposure increases based on the size and familiarity of the client, the potential for lawsuits

can increase. Thus, it is not surprising that correlations between our proxies for litigation and

reputation risk (presented in Table 3) are large and significant (e.g., ranging from 0.225 to

0.745). Due to these high correlations and the inability to separate the constructs of litigation and

reputation risk empirically in our setting, we use the principal components method of factor

analysis to reduce these six variables to a single construct of auditor engagement risk.

This procedure results in only one factor with an eigenvalue greater than one, further

supporting the notion that the six variables proxy for a single construct (see Hair, Anderson,

Tathum, and Black 1995).11 We label this factor AudEngageRisk. The final communality

estimates, which capture the correlation between the common factor and the six individual

variables, are 0.552 for LitRisk_Shu, 0.639 for LitRisk_KS, 0.251 for InstOwn%, 0.673 for Fees,

11
We retain one factor with a score greater than one (3.217); the remaining factors are all below one. Our one
retained factor explains 53.62 percent of variation.

15
and 0.548 for BigN, and 0.553 for OfficeSize.12 We use this factor (AudEngageRisk) in our main

analyses, but also report the results using each of the individual variables. Because this factor is

composed of proxies for litigation and reputation risk previously used in the literature, we do not

hold AudEngageRisk out as a “new” measure; rather, using this combined factor is necessary due

to the empirical difficulty of separating litigation and reputation risk.

3.3 Client Importance

Reynolds and Francis (2000) argue that the local practice office is the primary level for

decision-making related to client engagement, client retention, administering audits, and issuing

opinions. Following prior research, we measure client importance by scaling client total fees by

total fees for public clients of the respective audit office (Reynolds and Francis 2000; Chung and

Kallapur 2003).

3.4 Sample Selection

Our sample includes company-year observations between 2004 and 2014. We begin the

sample period in 2004 because of the addition of Item 4.02 as an 8-K triggering event to identify

non-reliance restatements. We use the Audit Analytics Advanced Restatement database to collect

out-of-period adjustments.13 Our sample consists of companies in non-regulated industries (i.e.,

we exclude companies with SIC codes 4400-4999 and 6000-6999) to limit the effect of

regulatory influence on the method of misstatement disclosure. Because auditor incentives to

prefer less prominent disclosure should only be manifest when the auditor provided the opinion

on the misstated financial statements, we limit our sample to misstatement disclosures in which

the auditor during the misstated period(s) is the same at the misstatement disclosure date. Table

12
We note that two elements of AudEngageRisk are ranked (LitRisk_Shu and LitRisk_KS), while the others are not.
Factor analysis results are similar when we also decile rank InstOwn%, Fees, and OfficeSize before performing the
factor analysis.
13
As of May 2016, Audit Analytics Advanced Restatement database contains out-of-period adjustments from filings
starting in 1998. We only include errors related to out-of-period adjustments from 2004 onward in our sample.

16
1, Panel A, describes how we arrived at our main sample of 2,868 company-year observations,

each of which is a unique misstatement disclosure.

Table 1, Panel B, presents the sample breakout, by year, for the various disclosure

methods. We find that within our sample, 1,082 misstatements (37.7 percent) are revealed

through a non-reliance restatement, 1,124 (39.2 percent) are revealed through a restatement other

than a non-reliance restatement (i.e., revision restatements), and 662 (23.1 percent) are revealed

through an out-of-period adjustment. We find a decreasing trend in non-reliance restatements

from 2004 through 2014, consistent with Scholz (2008). Other restatements experience an

upward trend after 2009. We find an increasing trend in out-of-period adjustments through 2008,

remaining stable in the years following.14

3.5 Model Specification

To test our hypothesis, we use the following ordered logistic regression model with year

and industry fixed effects:

RestateCategoryt+1 = α0 + α1AudEngageRiskit + α2ClientImportanceit + α3Sizeit + α4MTBit

+ α5Leverageit + α6Lossit + α7ICMWit + α8NTFilerit + α9Foreignit + α10M&Ait +
α11Restructureit + α12Segmentsit + α13ShareTurnoverit + α14NegIncomeImpact +
α15PosIncomeImpact + α16YearsMisstated + α17DaysToDisclose +
α18NumAnalystsit + α19NR_Restate_Pr2it-2, t-1 + α20Revision_Restate_Pr2it-2, t-1 +
α21OOPA_Pr2it-2, t-1 + αjIndustryFE + αkYearFE (3)
where:

RestateCategory = an ordered variable based on the prominence of misstatement disclosure;

non-reliance restatements = 2, other restatements = 1, and out-of-period
adjustments = 0;

14
Table 1 shows that non-reliance restatements decrease in frequency over our sample period and that out-of-period
adjustments increase in frequency through 2008. Therefore, it is possible that the hypothesized relationship between
engagement risk and less-prominent disclosures could merely capture these time trends. To alleviate this concern,
we include year-fixed effects in all regressions. Results are also robust to only including observations post-2008 in
which these trends are less pronounced.

17
AudEngageRisk = the common factor for LitRisk_Shu, LitRisk_KS, InstOwn%, Fees, BigN,
and OfficeSize derived from the principal components method of factor
analysis;

ClientImportance = total client fees divided by total fees of publicly listed clients of the
respective auditor office; and

all other variables are defined in the Appendix. To test our hypothesis, the coefficients of interest

are α1, the coefficient on AudEngageRisk, and α2, the coefficient on ClientImportance. We

expect both coefficients to be negative, indicating less prominent disclosures.

We control for company characteristics and events/transactions that could potentially

influence the severity and magnitude of the misstatement, thereby influencing the method of

disclosure. All control variables are measured as of the most recent fiscal year-end before the

misstatement disclosure.15 Specifically, we control for company size (Size), expected growth

(MTB), leverage (Leverage), financial distress (Loss), the strength of the financial reporting

environment (ICMW), delays in financial reporting (NTFiler), operating complexity (Foreign

and Segments), involvement in merger and acquisition activity (M&A), and involvement in

restructuring activities (Restructure). Additionally, because investors react more negatively to

more prominently disclosed restatements (Files et al. 2009), greater investor attention could

influence how a misstatement is disclosed. As such, we include share turnover (ShareTurnover)

as a measure of investor attention.

Because materiality should be one of the most important determinants for how to disclose

misstatements, we control for the magnitude of the misstatement as it relates to error adjustments

that decrease net income (NegIncomeImpact) and, separately, adjustments that increase net

income (PosIncomeImpact). We separate these variables because the perception of the

15
Measuring control variables in the year of the misstatement disclosure is problematic given that restatements can
be announced at various times throughout the year, thus making it difficult to ensure that control variables precede
the announcement date.

18
severity/magnitude of the error may not be symmetric. To further control for the severity of the

misstatement, we control for the number of years that were misstated (YearsMisstated) and the

time elapsed between the misstatement period and the disclosure date (DaysToDisclose)

(Schmidt and Wilkins 2013).

To better isolate the auditor’s incentives, we explicitly control for management’s

reputation risk as measured by analyst attention (NumAnalysts). To control for clients’ past

preference for error disclosure method, we control for whether companies have issued non-

reliance restatements, other restatements, or out-of-period adjustments in the prior two years

(NR_Restate_Pr2, Revision_Restate_Pr2, OOPA_Pr2). Finally, we include industry and year

fixed effects to control for variation in disclosure methods for misstatements across industries

and over time, and we cluster standard errors by company.16 All continuous variables are

winsorized at 1 and 99 percent.

4. Results

4.1 Descriptive Statistics and Correlations

Table 2 presents descriptive statistics for our sample. 80.4 percent of companies in our

sample are audited by a Big N firm (BigN). Average (unlogged) audit fees in the fiscal year

before the disclosure are $2.5 million (Fees). In the fiscal year preceding the misstatement

disclosure, 36.2 percent report a loss (Loss), 17.7 percent file the annual report after the initial

deadline (NTFiler), and 23.6 percent report a material weakness in internal control. This rate of

material weakness disclosure prior to misstatement disclosure highlights the subjectivity in

16
We use aggregated industry indicators following Ashbaugh, LaFond, and Mayhew (2003) to avoid the incidental
parameters problem of including a large number of fixed effects in a binary dependent variable model (see Greene
2004). In untabulated analyses, we re-estimate our models as linear probability models (rather than logistic
regression models) and include industry indicators based on 2-digit SIC codes. All inferences remain robust to this
alternative model specification.

19
materiality decisions and is consistent with evidence in Rice and Weber (2012) that companies

do not always report material weaknesses in a timely manner. In terms of misstatement

magnitude, the average cumulative net income impact for misstatements that decrease income is

1.1 percent of total assets (NegIncomeImpact), while the average cumulative net income impact

for misstatements that increase income is 0.166 percent of total assets (PosIncomeImpact). On

average 2.3 years are misstated (YearsMisstated) and the misstatement disclosure occurs on

average 236.7 days after the latest misstated period (DaysToDisclose).17

Table 3 presents the Spearman and Pearson correlations among the misstatement

disclosure categories, the individual proxies for auditor litigation and reputation risk, and the

combined factor. Consistent with our hypothesis, we find a positive and significant correlation

between engagement risk (AudEngageRisk) and revealing a misstatement through an out-of-

period adjustment (OOPA, p < 0.001) but a negative relationship with non-reliance restatements

(NR_Restate, p < 0.001). These results provide initial evidence supporting our hypothesis.

Table 4 presents descriptive statistics for select variables between the three different

disclosure methods. As expected, we find that misstatement magnitudes are generally larger

when disclosure is through more prominent channels. For example, mean IncomeImpact (the

signed cumulative net income impact) and ABS_IncomeImpact (the absolute value of the

cumulative net income impact) are significantly larger for misstatements disclosed through a

non-reliance restatement than through a revision restatement or an out-of-period adjustment (p <

0.01) and for revision restatements compared to out-of-period adjustments.18 Non-reliance

17
Descriptive statistics are consistent with a sample of misstated financial statements (i.e., non-timely filings, higher
incidence of material weaknesses, etc.) because we measure control variables in the year before the misstatement
disclosure and because most disclosures happen within a year of the end of the misstated period (DaysToDisclose).
18
We present descriptive statistics of cumulative income impact instead of signed income impact
(NegIncomeImpact, PosIncomeImpact) because the frequency of 0 counts in signed versions make descriptive
information less useful.

20
restatement disclosures are also more likely to reduce net income (IncomeImpact) and cover

more years (YearsMisstated) than other forms of restatement and out-of-period adjustments (p <

0.01 in all cases). However, relative to revision restatements and out-of-period adjustments, non-

reliance restatements are timelier, in that the time between the last misstated period and the

restatement announcement is shorter (DaysToDisclose, p < 0.01 in all cases). This is likely

attributable to SEC rules requiring timely disclosure of this important corporate event.19

Regarding the misstated accounts, we find some differences in the reporting of Revenue,

Inventory/COGS, Liabilities/Reserves, and Other restatements across the three categories. Out-

of-period adjustments are less likely to be reported for Expenses, Debt/Equity, and

Liability/Reserves misstatements, but more likely to be reported for Tax and Other error

corrections.

4.2 Validation of Disclosure Prominence Categories

Prior research suggests that restatement announcements influence clients’ auditor

retention decisions (Hennes et al. 2014), which can negatively impact auditors’ market share

(Swanquist and Whited 2015). Therefore, our hypothesis presumes that auditors face a greater

risk of litigation and reputation damage when the disclosure of misstatements from previously

audited financial statements is more prominent. To empirically support this presumption, we

investigate whether misstatement disclosure prominence impacts the number of lawsuits brought

against an auditor and the number of public clients audited in the year after a misstatement

disclosure.

We perform our tests at the audit firm-year level by aggregating client information by

19
According to SEC rules, once a prior period misstatement is determined material and reportable under 8-K filing
rules, companies are required to file the related 8-K within four business days (see e.g.,
https://www.sec.gov/rules/final/33-8400.htm).

21
audit firm-year during our sample period. We capture the prominence of the misstatement

disclosure with variables representing the number of non-reliance restatements

(Sum_NR_Restate), revision restatements (Sum_Revision_Restate), and out-of-period

adjustments (Sum_OOPA) disclosed by clients of an audit firm during the current year. These

variables are defined in the Appendix. Our models also include audit firm and year fixed effects

to control for changes in the dependent variable to due to time and time-invariant characteristics

of the audit firm and its client portfolio.

First, we examine whether disclosure prominence affects the number of lawsuits brought

against an audit firm. In Column (1) of Table 5, we regress the sum of open litigation cases

brought against an audit firm (Sum_Auditor_Litigation) per the Audit Analytics Legal Case and

Legal Parties database on the number of misstatements disclosed during the year based on

disclosure prominence. We find that the number of public audit client non-reliance restatements

(Sum_NR_Restate) announced in year t is significantly associated with the number of lawsuits

brought against the audit firm in that year. We also find that the audit firms are named in fewer

lawsuits as the number of misstatements revealed through out-of-period adjustments

(Sum_OOPA) increases.

In Columns (2) and (3) of Table 5, we regress the net change in public audit clients

audited by an auditor from the current to subsequent year (NetChangeClients) and the percentage

change in public audit clients (%ChangeClients) on our disclosure prominence measures. We

find that as the number of non-reliance restatement disclosures increases, there is a future loss of

market share, suggesting that more prominently disclosed misstatements can negatively impact

an auditor’s reputation. Taken together, results in Table 5 provide direct evidence that

misstatement disclosure prominence impacts litigation against the auditor and changes in

22
auditors’ market share, consistent with our argument that auditors should be aware of the link

between disclosure prominence and engagement risk, and thus act according to their incentives.

4.3 Hypothesis Test

Table 6 presents the results of our hypothesis test. We test our hypothesis with an ordered

logistic regression in which the dependent variable RestateCategory is equal to 2 for non-

reliance restatements, 1 for revision restatements, and 0 for out-of-period adjustments. Thus,

higher values of RestateCategory capture more prominent misstatement disclosures. Regarding

our variables of interest, and consistent with our hypothesis, we find that both higher levels of

auditor engagement risk (AudEngageRisk) and greater economic bonding to the client

(ClientImportance) are negatively associated with more prominent disclosure channels (p < 0.01

and p < 0.10, respectively). Table 6 also reports the results of regression estimates when

AudEngageRisk is replaced with its individual components. Specifically, results suggest that

elements of litigation risk (LitRisk_KS, p <0.05) and reputation risk (BigN, OfficeSize, p < 0.01)

are associated with disclosure prominence. Thus, our results regarding AudEngageRisk is not

merely an artifact of the factor analysis process and we conclude that results from Table 6

support our hypothesis.

Results of control variables in Table 6 suggest that companies with disclosed material

weaknesses (ICMW) and that file after the initial filing deadline (NTFiler) are more likely to

disclose misstatements through more prominent channels. Companies with more foreign sales

(Foreign) are less likely to disclose misstatements prominently. Further, companies are more

likely to disclose the misstatement prominently when the magnitude of the cumulative impact to

23
net income is larger in either direction (NegIncomeImpact and PosIncomeImpact).20 More

prominent disclosure is used when the misstatement covers more years (YearsMisstated) but

occurs more quickly (DaysToDisclose).

In Table 7, we further examine the impact of engagement risk on preferred disclosure

channels by creating an alternative sample in which we match misstatements corrected through

out-of-period adjustments with non-reliance and revision restatements based on the signed,

cumulative impact of the misstatement on income.21 Specifically, we match OOPA, NR Restate,

and Revision Restate observations in our sample within the same industry and year and closest

signed, cumulative net income impact of the misstatement (i.e., IncomeImpact, with a maximum

difference of +/- 0.03).22 Out of 662 potential matches, this results in 202 successful matches for

a sample of 606 observations.

Panel A of Table 7 provides differences in mean and median signed, cumulative net

income impact (IncomeImpact) between these three groups. We do not find significant

differences in mean or median values. Panel B of Table 7 presents the results of estimating

Equation (3) with RestateCategory as the dependent variable. Although our three groups are well

balanced on the cumulative income impact of the misstatement, we continue to control for the

magnitude of the impact (i.e., NegIncomeImpact and PosIncomeImpact) in our model as there is

still variation between matched sets. We continue to find evidence that among a group of

companies with similar sized misstatements to income, auditor engagement risk

20
Because NegIncomeImpact is signed, the negative coefficient in Table 6 indicates less prominent disclosure
methods as the negative income impact approaches zero.
21
Results from this matched-magnitude test are consistent when we limit the sample to restatements related to one
annual report, thus removing any noise from considering multiple periods.
22
In untabulated analysis we instead match on Size and find consistent results. We match on the signed income
impact (IncomeImpact) because matching on the absolute value could match income-increasing misstatements with
income-decreasing misstatements, which are fundamentally different types of management errors. By matching on
the signed income impact, we are more likely to match misstatements of similar magnitude and management intent.

24
(AudEngageRisk, p < 0.001) and client importance (ClientImportance, p < 0.10) are negatively

associated with prominent disclosure of the misstatement. Further, results persist when

AudEngageRisk is separated into its components (LitRisk_KS, BigN, OfficeSize,

ClientImportance, p < 0.10). These results provide corroborating evidence that auditor incentives

influence the prominence of disclosure of misstatements even when the income impact of the

misstatement is similar.

5. Additional Analyses

In this section, we 1) perform additional analyses to further examine the role of

misstatement magnitude in our results, 2) provide support for our inferences regarding auditor

incentives, and 3) perform other sensitivity analyses.

5.1 Magnitude of Misstatement and Disclosure Method

We recognize that although materiality decisions require professional judgment,

materiality should be the primary consideration for how to disclose misstatements (Acito et al.

2009). As such, auditors’ incentive to avoid prominent disclosure of misstatements should play a

smaller role in the face of increasing misstatement magnitude. To test this assumption, we limit

the sample to observations where the materiality of the signed error is less certain (i.e.,

IncomeImpact is between 0.25 and 1.0 percent of total assets) and where the cumulative signed

net income impact is clearly material (i.e., IncomeImpact is greater than 1.0 percent of total

assets).23

Table 8 reports multivariate tests to this effect. Specifically, in the first column we find

that the auditor incentive effects captured by AudEngageRisk remain negative and significant.

We also find that the coefficient on ClientImportance is negative, but insignificant at

23
The more uncertain materiality range is based on the ranges auditors typically use when establishing overall
materiality (Eilifsen and Messier 2014).

25
conventional levels (p = 0.128). In the second column, we examine our results when the

misstatement is clearly material. In this setting, we find that both AudEngageRisk and

ClientImportance are insignificant at conventional levels.24 Thus, our main results from Tables 6

and 7 do not suggest that auditors ignore materiality and intentionally conceal large

misstatements through out-of-period adjustments. Instead, the results reflect an average incentive

for auditors to, when possible, use the least-prominent disclosure method as engagement risk

increases. Thus, one key takeaway from our study is that auditors’ incentives play a role in error

disclosure prominence when materiality is uncertain.

5.2 Auditor during Misstated Period(s) Differs from Auditor at Disclosure Date

To provide further evidence that empirical results from Tables 6 through 8 are influenced

by auditor and not management incentives, we examine whether our results differ in a sample of

misstatement disclosures in which the auditor during the misstated period(s) is different than the

auditor at the disclosure date. In these cases, because the disclosing auditor’s prior work is not

called into question and the risk of reputational damage and litigation risk is not a central issue,

we would not expect auditor engagement risk to affect disclosure prominence. To perform this

analysis, we incorporate all misstatement disclosures where the auditor at the disclosure date is

different from the misstated periods.

Using this sample of observations, we re-estimate Equation (3) with RestateCategory as

the dependent variable and AudEngageRisk and ClientImportance as the variables of interest.25

The results of this test are presented in Table 9. In this test, we find insignificant coefficients on

24
To test the difference between the coefficients on AudEngageRisk in the two different subsample estimations, we
calculate a Z-statistic following Clogg, Petkova, and Haritou (1995) as (βpre - βpost) / √(SEβpre2 + SEβpost2) and
find a significant difference (Z-statistic = 1.302, p-value = 0.093).
25
In this test, we are unable to control for NR_Restate_Pr2, Revision_Restate_Pr2, or OOPA_ Pr2 because of
insufficient variation in the control variables in the reduced different-auditor sample.

26
both AudEngageRisk and ClientImportance (p = 0.330 and 0.224, respectively). In fact, when

AudEngageRisk is broken into its individual components, we find some evidence that

misstatement disclosure becomes more prominent when the new auditor is from a large office or

auditing an important client (OfficeSize, ClientImportance, p < 0.10). These results suggest that

new auditors are not averse to prominently disclosing clients’ misstatements related to the prior

auditor’s work. Together, these results suggest that engagement risk and client importance have a

fundamentally different effect on misstatement disclosure prominence when the auditor at the

disclosure date differs from the auditor during the misstated period. This test provides support

that our main inferences are influenced by auditor incentives and not those of management.

5.4 Additional Sensitivity Analyses

An alternative explanation for our observed results is that because engagement risk is not

likely to change drastically over time, the inferences we attribute to the post-audit consideration

of engagement risk are instead due to current period consideration of engagement risk. Prior

literature shows that higher current period engagement risk leads auditors to lower acceptable

audit risk and perform more work, thereby decreasing the likelihood of a material error not being

detected—and thus reducing the subsequent need for a prominent disclosure. However, if our

main results were simply due to greater auditor effort in the current year, the identified

association between engagement risk and subsequent misstatement disclosure prominence would

not exist in lower effort audits. To examine this, we estimate abnormal audit fees within the

broader population of available company-years between 2004 and 2014.26 After merging this

26
To estimate abnormal audit fees, we use the residual from a regression of the natural log of audit fees on company
characteristics (i.e., size, leverage, performance, expected growth, filer status, internal control strength, debt/equity
issuances, restructuring activities, merger and acquisition activities, financial distress, auditor size, industry and year
fixed effects). The residual captures the variance in logged audit fees not attributable to these client, auditor,
industry, and year effects and should represent higher or lower than expected audit fees after controlling for these
characteristics.

27
variable into our sample dataset, we limit our sample to company-year observations where the

abnormal (or unexpected) audit fees in the latest misstated year are not positive (i.e., lower-than-

expected fees/effort), resulting in a sample of 1,329 observations. With this sample

(untabulated), we continue to find a negative and significant coefficient on AudEngageRisk (p <

0.05) and a negative coefficient on ClientImportance that is insignificant at conventional levels

(p = 0.188). To the extent abnormal fees capture audit effort, these results suggest that auditors

act according to their post-audit incentives and that our results are not an artifact of higher effort

in the current period.

We recognize that the misstatements in our sample vary in terms of the length of time

corrected, which may impact the cumulative net income impact of the misstatement. In

sensitivity analysis, we limit the sample to corrections related to one annual report, thereby

removing any noise from considering multiple periods. This results in a limited sample of 1,295

observations. We find consistent results with our main tests using this more limited sample (p

<0.01 on AudEngageRisk, untabulated), but ClientImportance is insignificant (p = 0.627).

6. Conclusion

Due to the inherent nature of a risk-based audit, auditors are exposed to engagement risk,

which is the risk that costs will be incurred for providing an inaccurate opinion on a company’s

financial statements. The risk of losses due to litigation or diminished reputation can incentivize

auditors to perform high quality audits to minimize this exposure. Research finds that these

incentives help achieve higher contemporaneous audit quality, effectively overcoming auditors’

competing incentives to acquiesce to the demands of economically important clients. However, it

is unclear how these competing incentives of reducing engagement risk and appeasing the client

combine to affect disclosure decisions when auditors are faced with admitting missed errors from

28
a previous year’s audit. In this instance, auditors and management incentives to prefer less

prominent disclosure of prior period misstatements are aligned. We identify a sample of 2,868

misstatement disclosures from 2004 through 2014 and investigate the effect of engagement risk

and client importance on the disclosure channel through which misstatements are corrected and

communicated to the public.

We find that both higher auditor engagement risk and client importance are associated

with a lower likelihood of disclosing misstatements through prominent channels. While in all

cases the company admits to a misstatement, less prominent disclosures do not retract reliance on

the previously issued audit opinion, thus potentially sparing the auditor (and client) reputational

damage and negative publicity. Importantly, we find that this association is manifest when

materiality is uncertain (i.e., around an auditor’s overall quantitative materiality threshold) but

not when the misstatement is clearly material.

Our results contribute to the literature in several ways. First, while extant literature shows

how engagement risk serves to increase contemporaneous audit quality, our results demonstrate

how this risk affects auditors and clients when considering previously issued financial

statements. Second, we provide evidence of the effect that reputational concerns can have on the

audit, a topic currently receiving little attention in the literature (DeFond and Zhang 2014).

Third, we provide archival evidence of a setting in which auditors are more likely to favor

important clients, providing support to past experimental research (Kadous et al. 2003). Fourth,

we document that incentives to use less prominent disclosure methods are somewhat offset by

the materiality of the misstatement, suggesting that clients and their auditors do not disclose

clearly material misstatements through less prominent methods. Rather, results indicate that

clients and their auditors operate within general materiality guidelines and act in accordance with

29
their incentives by seeking to minimize the prominence of misstatement disclosures when the

materiality is less certain. These results shed light on the effect of incentives on auditor

judgment.

Our results are subject to the limitation that the final decision to disclose misstatements,

and how to do so, is the responsibility of management. Although our analyses control for

management’s exposure to dissemination of negative news and we perform additional analyses

to strengthen inferences that auditor incentives are driving our observed results, we recognize

that we cannot fully disentangle management and auditor incentives. However, the collective

evidence suggests that auditor incentives, and not just those of management, play a role in the

prominence of misstatement disclosures.

30
 Appendix
Variable Definitions

Variable Definition
%ChangeClients The percentage change in the number of public clients audited by an audit
firm from the current to the subsequent year

ABS_IncomeImpact The absolute value of the cumulative net income impact of the misstatement,
scaled by total assets and multiplied by 100 for expositional convenience

AudEngageRisk The common factor for LitRisk_Shu, LitRisk_KS, InstOwn%, Fees, BigN, and
OfficeSize derived from the principal components method of factor analysis

Beta The slope coefficient of a regression of daily stock returns on equal-weighted

market returns

BigN An indicator variable set equal to 1 if the auditor is from the Big 4, and 0
otherwise

ClientImportance Total client fees divided by total fees of publicly listed clients of the
respective auditor office

Current The current ratio (current assets divided by current liabilities)

DaysToDisclose The number of days between the misstatement period and the disclosure date

Delist One if the company delists in the year following the two year window
following the year under question, and zero otherwise

Fees Logged total audit fees during the year

Foreign An indicator variable equal to 1 if the company reports income from foreign
operations in year t, and 0 otherwise

ICMW An indicator variable set equal to 1 if a material weakness in internal controls

over financial reporting is disclosed in the year, and 0 otherwise

IncomeImpact The signed value of the cumulative net income impact of the misstatement,
scaled by total assets and multiplied by 100 for expositional convenience

IndustryFE Industry fixed effects using SIC codes to define industries as follows
(Ashbaugh et al. 2003): agriculture (0100-0999), mining and construction
(1000-1999, excluding 1300-1399), food (2000-2111), textiles and
printing/publishing (2200-2799), chemicals (2800-2824; 2840-2899),
pharmaceuticals (2830-2836), extractive (1300-1399; 2900-2999), durable
manufacturers (3000-3999, excluding 3570-3579 and 3670-3679),
transportation (4000-4899), retail (5000-5999), services (7000-8999,
excluding 7370-7379), computers (3570-3579; 3670-3679; 7370-7379), and
utilities (4900-4999)

InstOwn% The percentage of company stock held by institutions

INV Inventory scaled by prior year total assets

31
Leverage Long-term debt plus the current portion of long-term debt divided by total
assets

LitRisk_KS The sample decile ranking of the probability of litigation following model 2 of
Table 7 of Kim and Skinner (2012)

LitRisk_Shu The sample decile ranking of the probability of litigation following Shu
(2000)

Loss An indicator variable set equal to 1 if net income is less than zero, and 0
otherwise

M&A An indicator variable set equal to 1 if there was a merger or acquisition in the
year, and 0 otherwise

MTB The market-to-book ratio, calculated as the market value of equity divided by
the book value of equity

NegIncomeImpact the signed value of the cumulative net income impact of the misstatement,
scaled by total assets and multiplied by 100 for expositional convenience,
when net income is reduced, and 0 when the misstatement does not reduce net
income

NetChangeClients The net change in the public audit clients audited by an auditor from the
current to the subsequent year

NR_Restate An indicator variable equal to 1 if the misstatement was subsequently revealed

through a non-reliance restatement, and zero otherwise

NR_Restate_Pr2 An indicator variable equal to 1 if a non-reliance restatement was disclosed in

the previous two years, and zero otherwise

NTFiler An indicator variable set equal to 1 if the company files an ‘NT 10-K’, and 0
otherwise

NumAnalysts The number of analysts making earnings estimates during the year

OfficeSize Size of the audit office that performs the audit, measured as the natural log of
total fees received from publicly listed clients of the office

OOPA An indicator variable equal to 1 if the misstatement was subsequently revealed

through an out-of-period adjustment, and zero otherwise

OOPA_Pr2 An indicator variable equal to 1 if an out-of-period adjustment was disclosed

in the previous two years, and zero otherwise

PosIncomeImpact the signed value of the cumulative net income impact of the misstatement,
scaled by total assets and multiplied by 100 for expositional convenience,
when net income is increased, and 0 when the misstatement does not increase
net income

QUALOPIN An indicator variable set equal to 1 if the company received a qualified audit
opinion, and 0 otherwise

REC Accounts receivable scaled by prior year total assets

32
RestateCategory An ordered variable based on the prominence of disclosure of misstatements;
misstatements disclosed through non-reliance restatements = 2, misstatements
disclosed through other restatements = 1, and misstatements disclosed as out-
of-period adjustments = 0

Restructure An indicator variable equal to 1 if the company reports restructuring charges

in year t, and 0 otherwise

Returns Market-adjusted 12 month stock returns

Returns Skewness Skewness of the company’s 12 monthstock return

RETVOL Equals the standard deviation of the company’s 12 month stock return

Revision_Restate An indicator variable equal to 1 if the misstatement was subsequently revealed

through a revision restatement, and zero otherwise

Revision_Restate_Pr2 An indicator variable equal to 1 if a revision restatement was disclosed in the

previous two years, and zero otherwise

ROA Return on assets measured as net income divided by total assets

SalesGrowth Sales growth over the previous year

Segments The natural log of the number of a company’s business segments

ShareTurnover Share turnover during the year

Size The natural log of total assets

Sum_Auditor_Litigation The number of law suits against an audit firm that are open during the fiscal
year

Sum_OOPA The number of public audit clients for an audit firm that report OOPAs during
the fiscal year

Sum_NR_Restate The number of public audit clients for an audit firm that report non-reliance
restatements during the fiscal year

Sum_Revision_Restate The number of public audit clients for an audit firm that report revision
restatements during the fiscal year

TECH One if the company is in a technology industry (i.e., SIC codes 2830s, 3570s,
7370s, 8730s, and between 3825 and 3839), and zero otherwise

Turnover Trading volume accumulated over the 12 month period scaled by

beginning of the year shares outstanding

YearFE Indicator variables for each year in the sample period

YearsMisstated The number of misstated years

33
 References

Acito, A. A., J. J. Burks, and W. B. Johnson. 2009. Materiality Decisions and the Correction of
Accounting Errors. The Accounting Review 84 (3): 659-688.

Antle, R. 1982. The Auditor as an Economic Agent.’ Journal of Accounting Research 20 (2):
503-527.

Ashbaugh, H., R. LaFond, and B. W. Mayhew. 2003. Do Nonaudit Services Compromise

Auditor Independence? Further Evidence. The Accounting Review 78 (3): 611-639.

Bedard, J.C., and K.M. Johnstone. 2004. Earnings Manipulation Risk, Corporate Governance
Risk, and Auditors' Planning and Pricing Decisions. The Accounting Review 79 (2): 277-304.

Bedard, J. C. and L. Graham. 2011. Detection and severity classifications of Sarbanes-Oxley

Section 404 internal control deficiencies. The Accounting Review 86 (3): 825-855.

Blankley, A. I., D. N. Hurtt, and J. E. Macgregor. 2012. Abnormal audit fees and restatements.
Auditing: A Journal of Practice & Theory 31 (1): 79-96.

Cassell, C. A., M. S. Drake, and T. A. Dyer. 2017. Auditor Litigation Risk and the Number of
Institutional Investors. Auditing: A Journal of Practice & Theory 37 (3): 71-90.

Chan, D. K., and S. Pae. 1998. An Analysis of the Economic Consequences of the Proportionate
Liability Rule. Contemporary Accounting Research 15 (4): 457-480.

Chen, S., S. Y. Sun, and D. Wu. 2010. Client importance, institutional improvements, and audit
quality in China: An office and individual auditor level analysis. The Accounting Review 85 (1):
127-158.

Choi, J. H., J.B. Kim, X. Liu, and D. A. Simunic. ‘Audit Pricing, Legal Liability Regimes, And
Big 4 Premiums: Theory And Cross‐Country Evidence.’ Contemporary Accounting Research 25
(2008): 55-99.

Choi, J. H., J. B. Kim, X. Liu, and D. A. Simunic. 2009. Cross-Listing Audit Fee Premiums:
Theory and Evidence. The Accounting Review 84 (5): 1429-1463.

Choudhary, P., K.J. Merkley, and K. Schipper. 2017. Qualitative Characteristics of Financial
Reporting Errors Deemed Immaterial by Managers. Working paper, University of Arizona,
Cornell University, and Duke University.

Chung, H., and S. Kallapur. 2003. Client importance, nonaudit services, and abnormal accruals.
The Accounting Review 78 (4): 931-955.

Clogg, C. C., E. Petkova, and A. Haritou. 1995. Statistical methods for comparing regression
coefficients between models. American Journal of Sociology 100 (5): 1261-1293.

34
Core, J. and W. Guay. 1999. The use of equity grants to manage optimal equity incentive levels.
Journal of Accounting and Economics 28 (2): 151-184.

Craswell, A., D. J. Stokes, and J. Laughton. 2002. Auditor independence and fee dependence.
Journal of Accounting and Economics 33 (2): 253-275.

Deangelo, L. E. 1981. Auditor Size and Audit Quality. Journal of Accounting and Economics 3
(3): 183-199.

DeFond, M. L., And J. Zhang. 2014. A Review of Archival Auditing Research. Journal of
Accounting and Economics 58 (2-3): 275-326.

Desai, H., C. E. Hogan, and M. S. Wilkins. 2006. The Reputational Penalty for Aggressive
Accounting: Earnings Restatements and Management Turnover. The Accounting Review 81 (1):
83-112.

Eilifsen, A., and W. F. Messier Jr. 2014. Materiality guidance of the major public accounting
firms." Auditing: A Journal of Practice & Theory 34 (2): 3-26.

Files, R., E. P. Swanson, and S. Tse. 2009. Stealth Disclosure of Accounting Restatements. The
Accounting Review 84 (5): 1495-1520.

Francis, J. R., and P. N. Michas. 2013. The Contagion Effect of Low-Quality Audits. The
Accounting Review 88 (2): 521-552.

Francis, J., R., and M. D. Yu. 2009. Big 4 Office Size and Audit Quality. The Accounting Review
84 (5): 1521-1552.

Geiger, M. A., and K. Raghunandan. 2001. Bankruptcies, Audit Reports, and the Reform Act.
Auditing: A Journal of Practice & Theory 20 (1): 187-195.

Greene, W. 2004. The Behaviour of the Maximum Likelihood Estimator of Limited Dependent
Variable Models in the Presence of Fixed Effects. Econometrics Journal 7 (1): 98-119.

Hackenbrack, K., and M. W. Nelson. 1996. Auditors' incentives and their application of financial
accounting standards. Accounting Review 71 (1): 43-59.

Hair, J. F., R. E. Anderson, R. L. Tatham, and W. C. Black. 1995. Multivariate Data Analysis. 4th
edition. Englewood Cliffs, NJ: Prentice Hall.

Hennes, K. M., A. J. Leone, and B. P. Miller. 2014. Determinants and Market Consequences of
Auditor Dismissals after Accounting Restatements. The Accounting Review 89 (3): 1051-1082.

Hirshleifer, D., and S. H. Teoh. 2003. Limited Attention, Information Disclosure, and Financial
Reporting. Journal of Accounting and Economics 36 (1-3): 337-386.

35
Hirst, D. E., and P. E. Hopkins. 1998. Comprehensive Income Reporting and Analysts’
Valuation Judgments. Journal of Accounting Research 36 (Supplement): 47-75.

Hogan, C. E., T. A. Lambert, and J. J. Schmidt. 2014. Do Management Internal Control

Certifications Increase the Likelihood Of Restatement-Related Litigation? Working paper,
Michigan State University and University of Texas at Austin.

Iliev, P., D. P. Miller, and L. Roth. 2014. Uninvited U.S. Investors? Economic Consequences of
Involuntary Cross-Listings. Journal of Accounting Research 52 (2): 473-519.

Johnstone, K.M. and J.C. Bedard. 2003. Risk Management in Client Acceptance Decisions. The
Accounting Review 78 (4): 1003-1025.

Johnstone, K. M., and J. C. Bedard. 2004. Audit Firm Portfolio Management Decisions. Journal
of Accounting Research 42 (4): 659-690.

Kadous, K., S. J. Kennedy, and M. E. Peecher. 2003. The effect of quality assessment and
directional goal commitment on auditors' acceptance of client-preferred accounting methods. The
Accounting Review 78 (3): 759-778.

Kanodia, C., R. Bushman, and J. Dickhaut. 1989. Escalation Errors and The Sunk Cost Effect:
An Explanation Based On Reputation And Information Asymmetries. Journal of Accounting
Research 27 (1): 59-77.

Keune, M. B., and K. M. Johnstone. 2012. Materiality Judgments and the Resolution of Detected
Misstatements: The Role of Managers, Auditors, and Audit Committees. The Accounting Review
87 (5): 1641-1677.

Kim, I., and D. J. Skinner. 2012. Measuring securities litigation risk. Journal of Accounting and
Economics 53 (1-2): 290-310.

Lee, H. Y., and V. Mande. 2003. The Effect of the Private Securities Litigation Reform Act Of
1995 on Accounting Discretion of Client Managers of Big 6 and Non-Big 6 Auditors. Auditing:
A Journal of Practice & Theory 22 (1): 93-108.

Li, C. 2009. Does client importance affect auditor independence at the office level? Empirical
evidence from going‐concern opinions. Contemporary Accounting Research 26 (1): 201-230.

Liu, C., and T. Wang. 2006. Auditor Liability and Business Investment. Contemporary
Accounting Research 23 (4): 1051-1071.

Magnan, M. L. 2008. Discussion Of “Audit Pricing, Legal Liability Regimes, And Big 4
Premiums: Theory And Cross‐Country Evidence”. Contemporary Accounting Research 25 (1):
101-108.

36
Maines, L. A., and L. S. McDaniel. 2000. Effects of Comprehensive-Income Characteristics on
Nonprofessional Investors’ Judgments: The Role of Financial-Statement Presentation Format.
The Accounting Review 75 (2): 179-207.

Messier, W. F., S. M. Glover, and D. F. Prawitt. 2014. Auditing & Assurance Services: A
Systematic Approach. New York, NY: McGraw-Hill Irwin.

Palmrose, Z. V., and S. Scholz. 2004. The Circumstances and Legal Consequences of Non‐
GAAP Reporting: Evidence from Restatements. Contemporary Accounting Research 21 (1):
139-180.

Palmrose, Z. V., V.J. Richardson, and S. Scholz. 2004. Determinants of Market Reactions to
Restatement Announcements. Journal of Accounting and Economics 37 (1): 59-89.

Public Company Accounting Oversight Board (PCAOB). 2011. Concept Release on Auditor
Independence and Audit Firm Rotation. PCAOB Release No. 2011-006.

Reynolds, J. K., and J. R. Francis. 2000. Does size matter? The influence of large clients on
office-level auditor reporting decisions. Journal of accounting and economics 30 (3): 375-400.

Rice, S. C., and D. P. Weber. 2012. How Effective Is Internal Control Reporting Under Sox 404?
Determinants of the (Non‐) Disclosure of Existing Material Weaknesses. Journal of Accounting
Research 50 (3): 811-843.

Rice, S., D. Weber, and B. Wu. 2015. Does Sox 404 Have Teeth? Consequences of the Failure to
Report Existing Internal Control Weaknesses. The Accounting Review 90 (3): 1169-1200.

Schmidt, J., and M.S. Wilkins. 2013. Bringing Darkness to Light: The Influence of Auditor
Quality and Audit Committee Expertise on the Timeliness of Financial Statement Restatement
Disclosures. Auditing: A Journal of Practice & Theory 32 (1): 221-244.

Scholz, S. 2008. The Changing Nature and Consequences of Public Company Financial
Restatements. The US Department of the Treasury.
Schwartz, R. 1997. Legal Regimes, Audit Quality and Investment. The Accounting Review 72
(3): 385-406.

Seetharaman, A., F. A. Gul., and S. G. Lynn. 2002. Litigation Risk and Audit Fees: Evidence
from UK Firms Cross-Listed on Us Markets. Journal of Accounting and Economics 33 (1): 91-
115.

Shu, S. Z. 2000. Auditor Resignations: Clientele Effects and Legal Liability. Journal of
Accounting and Economics 29 (2): 173-205.

Simunic, D. A. 1980. The Pricing of Audit Services: Theory and Evidence. Journal of
Accounting Research (1): 161-190.

37
Simunic, D., and M. Stein. 1996. The Impact of Litigation Risk on Audit Pricing: A Review of
the Economics and the Evidence. Auditing: A Journal of Practice & Theory 15 (Supplement):
119-134.

Skinner, D. J., and S. Srinivasan. 2012. Audit Quality and Auditor Reputation: Evidence from
Japan. The Accounting Review 87 (5): 1737-1765.

Staw, B. M. 1976. Knee-Deep In The Big Muddy: A Study Of Escalating Commitment To A

Chosen Course Of Action. Organizational Behavior and Human Performance 16 (1): 27-44.

Swanquist, Q. T., and R. L. Whited. 2015. Do Clients Avoid “Contaminated” Offices? The
Economic Consequences of Low-Quality Audits. The Accounting Review 90 (6): 2537-2570.

Tan, C. E., and S.M. Young. 2015. An Analysis of “Little r” Restatements. Accounting Horizons
29 (3): 667-693.

Venkataraman, R., J. P. Weber, and M. Willenborg. 2008. Litigation Risk, Audit Quality, and
Audit Fees: Evidence from Initial Public Offerings. The Accounting Review 83 (5): 1315-1345.

Weber, J., M. Willenborg, and J. Zhang. 2008. Does Auditor Reputation Matter? The Case of
KPMG Germany and Comroad Ag. Journal of Accounting Research 46 (4): 941-972.

Zhang, P., and L. Thoman. 1999. Pre-Trial Settlement and the Value of Audits. The Accounting
Review 74 (4): 473-491.

38
 Table 1
Sample Selection
Panel A: Sample Selection
N
Restatement and out-of-period announcements from Audit Analytics Advanced
Restatement Database covering misstatement disclosures between 2004 and 2014 13,642
Less: Restatement and out-of-period announcements in regulated industries in
non-regulated industries (where regulated industries capture SIC codes 4400-
4999 and 6000- 6999) (2,948)
Less: Restatement and out-of-period announcements with missing data for model
variables (7,674)
Less: Restatement and out-of-period announcements where the auditor at the
announcement date is different from the auditor during the misstated period(s) (152)
2,868

Where auditor is the same during misstatement period and public disclosure
Non-reliance restatement announcements 1,082
Other restatements 1,124
Out-of-period adjustments 662
2,868

Panel B: Misstatement Disclosures by Year

Misstatement Disclosure: 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Total
Non-reliance restatement 276 268 143 87 55 59 51 36 52 29 26 1,082
Other restatement 93 103 69 61 77 74 80 112 153 165 137 1,124
Out-of-period adjustment 9 15 30 62 71 91 75 83 72 73 81 662
Total 378 386 242 210 203 224 206 231 277 267 244 2,868

39
 Table 2
Descriptive Statistics
25th 75th
Variable N Mean STD Percentile Median Percentile
AudEngageRisk 2,868 0.029 0.907 -0.486 0.185 0.691
BigN 2,868 0.804 0.397 1.000 1.000 1.000
ClientImportance 2,868 0.123 0.208 0.015 0.041 0.123
DaysToDisclose 2,868 236.760 202.643 121.000 162.000 365.000
Fees 2,868 14.016 1.180 13.303 14.044 14.754
Foreign 2,868 0.364 0.481 0.000 0.000 1.000
ICMW 2,868 0.236 0.425 0.000 0.000 0.000
Inst% 2,868 0.421 0.354 0.029 0.389 0.749
Leverage 2,868 0.207 0.210 0.009 0.157 0.326
LitRisk_Shu 2,868 5.368 2.806 3.000 6.000 8.000
LitRisk_KS 2,868 4.189 2.743 2.000 4.000 6.000
Loss 2,868 0.362 0.481 0.000 0.000 1.000
M&A 2,868 0.200 0.400 0.000 0.000 0.000
MTB 2,868 3.014 4.878 1.220 1.960 3.488
NegIncomeImpact 2,868 -1.128 3.635 -0.546 0.000 0.000
NR_Restate 2,868 0.377 0.485 0.000 0.000 1.000
NR_Restate_Pr2 2,868 0.174 0.379 0.000 0.000 0.000
NTFiler 2,868 0.177 0.382 0.000 0.000 0.000
NumAnalysts 2,868 5.929 6.707 1.000 3.833 8.545
OfficeSize 2,868 17.262 1.737 16.176 17.607 18.604
OOPA 2,868 0.231 0.421 0.000 0.000 0.000
OOPA_Pr2 2,868 0.097 0.296 0.000 0.000 0.000
PosIncomeImpact 2,868 0.166 0.655 0.000 0.000 0.000
Restructure 2,868 0.369 0.482 0.000 0.000 1.000
Revision_Restate 2,868 0.392 0.488 0.000 0.000 1.000
Revision_Restate_Pr2 2,868 0.129 0.335 0.000 0.000 0.000
Segments 2,868 0.159 0.603 0.000 0.000 0.000
ShareTurnover 2,868 2.266 2.067 0.813 1.720 2.999
Size 2,868 6.283 1.830 5.025 6.255 7.553
YearsMisstated 2,868 2.328 1.805 1.000 2.000 3.000
Notes: Descriptive statistics for variables included in our multivariate hypothesis test in Equation (3). All variables
are defined in the Appendix.

40
 Table 3
Univariate Correlations
Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
(1) NR_Restate -0.625 -0.426 -0.076 -0.143 -0.078 -0.192 -0.214 -0.139 -0.179 -0.020
(2) Revision_Restate -0.625 -0.440 -0.014 0.047 -0.010 0.047 0.065 -0.009 0.018 0.053
(3) OOPA -0.426 -0.440 0.103 0.111 0.101 0.166 0.170 0.171 0.185 -0.038
(4) LitRisk_Shu -0.071 -0.016 0.101 0.497 0.302 0.472 0.315 0.284 0.579 -0.026
(5) LitRisk_KS -0.146 0.052 0.108 0.496 0.273 0.491 0.330 0.291 0.718 0.005
(6) InstOwn% -0.065 -0.012 0.089 0.253 0.225 0.331 0.259 0.288 0.543 -0.129
(7) Fees -0.192 0.049 0.164 0.450 0.491 0.286 0.535 0.553 0.760 0.052
(8) BigN -0.214 0.065 0.170 0.289 0.332 0.214 0.518 0.678 0.658 -0.257
(9) OfficeSize -0.132 -0.015 0.169 0.230 0.268 0.228 0.485 0.616 0.655 -0.607
(10) AudEngageRisk -0.162 0.006 0.179 0.573 0.745 0.507 0.731 0.575 0.574 -0.173
(11) ClientImportance 0.014 0.051 -0.075 0.049 0.056 -0.069 0.151 -0.380 -0.741 -0.120
Notes: The table summarizes correlations for key variables used in our analyses. Pearson (Spearman) correlations are above (below) the diagonal. Correlations
that are significant at p-value < 0.05 are bolded. See the Appendix for variable definitions.

41
 Table 4
Select Descriptive Statistics by Disclosure Method

(1) (2) (3)

NR_Restate Revision_Restate OOPA
(N=1,082) (N=1,124) (N=662)
Variable Mean Median Mean Median Mean Median
IncomeImpact -0.020 -0.004 -0.003 0.000 -0.001 0.000
ABS_IncomeImpact 0.025 0.008 0.006 0.000 0.003 0.001
YearsMisstated 2.864 2.000 2.075 2.000 1.882 1.000
DaysToDisclose 160.567 135.000 263.153 218.000 316.480 365.000

Restated Accounts:
Revenue 0.166 0.000 0.107 0.000 0.086 0.000
Inventory/COGS 0.110 0.000 0.083 0.000 0.076 0.000
Expenses 0.193 0.000 0.094 0.000 0.017 0.000
Debt/Equity 0.140 0.000 0.114 0.000 0.020 0.000
Liability/Reserves 0.107 0.000 0.084 0.000 0.060 0.000
Tax 0.166 0.000 0.198 0.000 0.288 0.000
Other 0.348 0.000 0.462 0.000 0.484 0.000

(1) vs (2) (1) vs (3) (2) vs (3)

Diff in Diff in Diff in Diff in Diff in Diff in
Variable Mean Median Mean Median Mean Median
IncomeImpact -0.017*** -0.004*** -0.019*** -0.004*** -0.003*** 0.000*
ABS_IncomeImpact 0.020*** 0.008*** 0.023*** 0.006*** 0.003*** -0.001
YearsMisstated 0.789*** 0.000*** 0.982*** 1.000*** 0.193*** 1.000***
DaysToDisclose -102.586*** -83.000*** -155.913*** -230.000*** -53.327*** -147.000

Revenue 0.060*** 0.000*** 0.080*** 0.000*** 0.021 0.000

Inventory/COGS 0.027** 0.000** 0.034** 0.000** 0.007 0.000
Expenses 0.099*** 0.000*** 0.177*** 0.000*** 0.078*** 0.000***
Debt/Equity 0.027* 0.000* 0.121*** 0.000*** 0.094*** 0.000***
Liability/Reserves 0.024* 0.000* 0.047*** 0.000*** 0.023* 0.000*
Tax -0.032* 0.000* -0.122*** 0.000*** -0.090*** 0.000***
Other -0.113*** 0.000*** -0.135*** 0.000*** -0.022* 0.000*
Notes: This table presents the differences in mean and median values of variables capturing the severity and
magnitude of misstated-years (IncomeImpact, ABS_IncomeImpact, YearsMisstated, and DaysToDisclose) and
affected accounts between disclosure method (non-reliance restatement, revision restatement, and out-of-period
adjustment). Differences in sample means for severity and magnitude are based on t-tests, and the tests for
differences in sample medians are based on Wilcoxon-Mann-Whitney tests. P-values are two-tailed. *, **, and ***
denote statistical significance at 0.10, 0.05, and 0.01 levels, respectively.

42
 Table 5
The Association between the Number of Misstatement Disclosures (by Prominence) and Subsequent Litigation against the
Audit Firm and Changes in Public Audit Clients at the Audit-Firm Level
(1) (2) (3)
DV = Sum_Auditor_Litigation DV = NetChangeClients DV = %ChangeClients
Variable Coefficient t-stat Coefficient t-stat Coefficient t-stat
Sum_NR_Restate 0.043 *** 3.350 -1.570 * -1.350 -0.007 *** -2.560
Sum_Revision_Restate 0.009 0.320 1.030 0.750 0.003 0.800
Sum_OOPA -0.042 ** -2.260 3.211 * 1.920 0.004 1.170
Year FE Included Included Included
Audit Firm FE Included Included Included
N 234 234 234
Adjusted R2 0.607 0.439 0.846
Notes: This table presents the results of tests examining the association between the number of misstatement disclosures (by prominence) by public audit clients
of an audit firm and the number of lawsuits brought against an audit firm, the change in the number public audit clients in the following year and percentage
change in the number of public audit clients in the following year. P-values are two-tailed. *, **, and *** denote statistical significance at 0.10, 0.05, and 0.01
levels, respectively. Standard errors are clustered by company. All variables are defined in the Appendix.

43
 Table 6
The Association between Misstatement Disclosure Prominence and Auditor Incentives
Ordered Logistic Regression
DV = RestateCategory
Variable Coefficient z-statistic Coefficient z-statistic
AudEngageRisk (-) -0.273 *** -3.490
LitRisk_Shu (-) 0.006 0.200
LitRisk_KS (-) -0.039 ** -2.020
InstOwn% (-) 0.083 0.640
Fees (-) -0.044 -0.530
BigN (-) -0.448 *** -2.710
OfficeSize (-) -0.100 *** -2.000
ClientImportance (-) -0.270 * -1.400 -0.769 *** -2.630
Size -0.037 -0.870 0.017 0.290
MTB -0.004 -0.430 -0.002 -0.200
Leverage 0.514 ** 2.400 0.421 * 1.920
Loss -0.110 -1.140 -0.071 -0.730
ICMW 1.560 *** 12.880 1.577 *** 12.960
NTFiler 0.899 *** 6.160 0.876 *** 5.880
Foreign -0.263 *** -2.990 -0.224 ** -2.490
M&A 0.026 0.250 0.024 0.230
Restructure 0.027 0.280 0.011 0.110
Segments -0.062 -1.010 -0.062 -1.020
ShareTurnover 0.000 0.020 -0.011 -0.320
NegIncomeImpact -0.202 *** -3.550 -0.199 *** -3.480
PosIncomeImpact 0.385 *** 4.770 0.383 *** 4.720
YearsMisstated 0.115 *** 3.830 0.113 *** 3.760
DaysToDisclose -0.003 *** -11.700 -0.003 *** -11.620
NumAnalysts 0.008 1.040 0.006 0.770
NR_Restate_Pr2 0.028 0.260 0.033 0.310
Revision_Restate_Pr2 0.146 1.280 0.149 1.300
OOPA_ Pr2 -1.147 *** -6.950 -1.128 *** -6.840
Industry FE Included Included
Year FE Included Included
N 2,868 2,868
N NR_Restate 1,082 1,082
N Revision_Restate 1,124 1,124
N OOPA 662 662
Pseudo R2 0.290 0.294
Notes: P-values are two-tailed. Predicted direction for our variable of interest is in parentheses. *, **, and ***
denote statistical significance at 0.10, 0.05, and 0.01 levels, respectively. Standard errors are clustered by company.
All variables are defined in the Appendix.

44
 Table 7
Matched Sample on Adjustment Magnitude

Panel A: Differences in Mean and Median IncomeImpact between Matched Companies

(1) OOPA (2) Revision Restate (3) NR_Restate
(N = 202) (N = 202) (N = 202)
Variable Mean Median Mean Median Mean Median
IncomeImpact -0.075 0.000 -0.086 0.000 -0.100 0.000

(1) vs (2) (1) vs (3) (2) vs (3)

Mean Median Mean Median Mean Median
Difference 0.010 0.000 0.025 0.000 0.015 0.000
p-value 0.792 0.930 0.543 0.712 0.715 0.567

Notes: This panel presents the differences in mean and median IncomeImpact between observations with an out-of-
period adjustment, revision restatement, or non-reliance restatement matched in the same year and industry and
magnitude of the adjustment (with a maximum difference of +/- 0.03). This procedure resulted in 202 successful
matches (out of 662 potential matches).

45
Panel B: Multiple Regression Analysis Using Matched Sample
Ordered Logistic Regression
DV = RestateCategory
Variable Coefficient z-statistic Coefficient z-statistic
AudEngageRisk (-) -0.582 *** -3.66
LitRisk_Shu (-) 0.022 0.340
LitRisk_KS (-) -0.090 *** -2.130
InstOwn% (-) -0.015 -0.050
Fees (-) -0.085 -0.460
BigN (-) -0.461 * -1.300
OfficeSize (-) -0.271 ** -2.250
ClientImportance (-) -0.587 * -1.47 -1.677 *** -2.410
Size -0.025 -0.31 0.085 0.680
MTB -0.002 -0.07 0.003 0.130
Leverage 1.599 *** 3.33 1.382 *** 2.700
Loss 0.167 0.81 0.282 1.330
ICMW 1.888 *** 7.59 1.934 *** 7.430
NTFiler 0.818 *** 2.86 0.712 ** 2.340
Foreign -0.023 -0.13 -0.006 -0.030
M&A 0.229 0.99 0.217 0.920
Restructure -0.074 -0.32 -0.103 -0.450
Segments -0.244 -1.57 -0.242 -1.660
ShareTurnover 0.039 0.83 0.020 0.270
NegIncomeImpact 0.260 0.84 0.220 0.690
PosIncomeImpact -0.702 -1.57 -0.729 -1.630
YearsMisstated 0.013 0.21 0.015 0.250
DaysToDisclose -0.002 *** -5.02 -0.002 *** -5.070
NumAnalysts -0.002 -0.1 -0.012 -0.600
NR_Restate_Pr2 0.075 0.3 0.128 0.500
Revision_Restate_Pr2 0.067 0.27 0.022 0.090
OOPA_Pr2 -1.999 *** -4.12 -2.018 *** -4.190
Industry FE Included Included
Year FE Included Included
N 606 606
N NR_Restate 202 202
N Revision_Restate 202 202
N OOPA 202 202
2
Pseudo R 0.206 0.217
Notes: This panel presents the results of our matched sample (where OOPA observations are matched with a non-
OOPA observations in the same year, industry, and closest adjustment magnitude). Predicted direction for our
variable of interest is in parentheses. P-values are two-tailed. *, **, and *** denote statistical significance at 0.10,
0.05, and 0.01 levels, respectively. Standard errors are clustered by company. All variables are defined in the
Appendix.

46
 Table 8
Misstatement Magnitude
NI impact <0.25% and <1% of assets NI impact > 1% of assets
DV=RestateCategory DV=RestateCategory
Variable Coefficient z-statistics Coefficient z-statistics
AudEngageRisk (- / ?) -0.551 *** -3.290 -0.205 -0.990
ClientImportance (- / ?) -0.466 -1.140 -0.338 -0.660
Controls Included Included
Industry and Year FE Included Included
N 661 638
N NR_Restate 272 476
N Revision_Restate 225 132
N OOPA 164 30
Pseudo R2 0.336 0.287
Notes: P-values are two-tailed unless a prediction is made. Predicted directions for our variable of interest are in parentheses. *, **, and *** denote statistical
significance at 0.10, 0.05, and 0.01 levels, respectively. Standard errors are clustered by company. All variables are defined in the Appendix.

47
 Table 9
Different Auditor during Misstatement Period(s) from Disclosure Date
DV = RestateCategory
Variable Coefficient z-statistic Coefficient z-statistic
AudEngageRisk (?) -0.316 -0.970
LitRisk_Shu (?) -0.038 -0.260
LitRisk_KS (?) -0.071 -0.650
InstOwn% (?) -0.479 -0.470
Fees (?) -0.466 -0.980
BigN (?) -1.110 -1.340
OfficeSize (?) 0.450 * 1.760
ClientImportance (?) 1.285 1.220 3.046 ** 1.980
Controls Included Included
Industry FE Included Included
Year FE Included Included
N 152 152
N NR_Restate 63 63
N Revision_Restate 46 46
N OOPA 43 43
Pseudo R2 0.338 0.353
Notes: P-values are two-tailed unless a prediction is made. Predicted directions for our variable of interest are in
parentheses. *, **, and *** denote statistical significance at 0.10, 0.05, and 0.01 levels, respectively. Standard errors
are clustered by company. All variables are defined in the Appendix.

48