Accepted Manuscript

Do Practitioner Assessments Agree with Academic Proxies for Audit

Quality? Evidence from PCAOB and Internal Inspections

Daniel Aobdia

PII: S0165-4101(18)30106-X
DOI: https://doi.org/10.1016/j.jacceco.2018.09.001
Reference: JAE 1213

To appear in: Journal of Accounting and Economics

Received date: 19 March 2016

Revised date: 5 September 2018
Accepted date: 6 September 2018

Please cite this article as: Daniel Aobdia , Do Practitioner Assessments Agree with Academic Prox-
ies for Audit Quality? Evidence from PCAOB and Internal Inspections, Journal of Accounting and
Economics (2018), doi: https://doi.org/10.1016/j.jacceco.2018.09.001

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service
to our customers we are providing this early version of the manuscript. The manuscript will undergo
copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please
note that during the production process errors may be discovered which could affect the content, and
all legal disclaimers that apply to the journal pertain.
 ACCEPTED MANUSCRIPT

Do Practitioner Assessments Agree with Academic Proxies for Audit Quality?

Evidence from PCAOB and Internal Inspections

Daniel Aobdia

Kellogg School of Management, Northwestern University

T
d-aobdia@kellogg.northwestern.edu

IP
CR
This version: September 2018

US
Abstract
This study investigates the degree of concordance between fifteen measures of audit quality used
AN
in academia and two measures of audit process quality determined either by audit firms’ internal
inspections or by Public Company Accounting Oversight Board inspections of individual
engagements. Using two confidential datasets of these assessments of audit process quality, I
M

find that three of the measures of audit quality used by academics have significant associations
with both measures of audit process deficiencies used by auditors and regulators: (i) the
propensity to restate financial statements, (ii) the propensity to meet or beat the zero earnings
ED

threshold, and (iii) audit fees. Seven academic proxies are significantly associated with only one
audit process quality measure, and five have insignificant associations with both practitioner
PT

assessments. Overall, the significant associations indicate that practitioners and academics share
common ground in identifying low-quality audits. These findings can provide guidance for
future studies in selecting audit-quality proxies suitable for different research questions.
CE
AC

Keywords: Audit Process Quality, PCAOB Inspections, Internal Inspections, Measures of Audit
Quality, Audit fees, Restatements.

JEL Classification: M42, C80.

1
 ACCEPTED MANUSCRIPT

1. Introduction
This study examines the degree of concordance between widely used academic measures of

audit quality and the views of audit firms and regulators (practitioners). Prior accounting

research introduces multiple input and output measures of audit quality based on externally

observable data. Output measures are mostly derived from the financial reporting quality of

T
client companies, and input measures are primarily derived from observable auditor attributes. In

IP
their reviews of the literature, Francis (2011), Knechel et al. (2013), DeFond and Zhang (2014),

CR
and Gaynor et al. (2016) argue that many of these proxies represent valid measurements of audit

quality. However, each construct indirectly measures audit quality, and has weaknesses. 1

US
An alternative approach to measure audit quality is to consider practitioners’ assessments of
AN
what constitutes a high-quality audit. Such assessments typically involve a detailed examination

of the audit process and focus on two elements: (i) whether an audit is conducted according to
M

relevant audit standards or to an audit firm’s internal methodology, and (ii) whether the evidence

gathered by the engagement team is sufficient to support the audit opinion (PCAOB, 2014; Bell
ED

et al., 2015). A major advantage of such assessments is that they provide an understanding of

how well an audit is executed from auditor effort, competence, and independence standpoints,
PT

which is what most measures used by academics indirectly attempt to capture (Bell et al., 2015).
CE

Thus, understanding the degree of concordance between practitioners’ assessments of audit

quality and academic measures can help determine which observable measures map the most
AC

with such assessments. However, ex ante, the level of agreement between practitioners’ and

academic views of audit quality is unclear. On the one hand, auditors aim to obtain reasonable

1
This perhaps explains why prior audit literature sometimes finds conflicting results depending on the audit quality
measure used. For example, the literature considers whether the provision of non-audit services impairs audit quality,
and finds conflicting results depending on the proxy of audit quality used (Lim and Tan, 2008; Bell et al., 2015).

2
 ACCEPTED MANUSCRIPT

assurance about whether the financial statements are free of material misstatements (AU 110).2

Thus, auditors are more likely to focus on more material financial statement items and areas

where the risk of material misstatement is the greatest (AS 11 and AS 15). This suggests a direct

relationship between practitioners’ views of audit quality and academic measures, such as the

probability of a restatement. On the other hand, based on direct evidence from the audit process,

T
practitioners could judge an audit acceptable (not acceptable), even if external proxies indicate

IP
low (high) audit quality (Bell et al., 2015).3 Further, despite their advantages, practitioner

CR
measures of audit quality could suffer from several issues. Especially, (i) inherent limitations in

audit standards or audit firms’ methodologies could lead to inaccurate practitioners’

US
understanding of how much evidence needs to be gathered to support the audit opinion; and (ii)
AN
practitioners’ assessments are often subjective and could be affected by particular incentives.

I use in this study two proprietary measures of practitioners’ views of audit quality, both
M

obtained from the Public Company Accounting Oversight Board (PCAOB). The first one reflects

deficiencies identified during PCAOB inspections of individual audits. The second one reflects
ED

audit firms’ internal assessments of their own audits. The former represents the regulator’s views

and the latter the audit firms’ views of what constitutes improperly conducted audits.
PT

The PCAOB is a nonprofit corporation established by the Sarbanes-Oxley Act of 2002 (SOX)
CE

to oversee the audit of public companies (issuers hereafter). Per Section 104 of SOX, the
AC

2
The PCAOB adopted a reorganization of the numbering of its standards on March 31 2015, effective December 31,
2016. Because of the timing of this paper, I use the standard numbering prior to the reorganization.
3
Illustratively, PCAOB inspection reports note that when PCAOB inspectors judge an audit to be deficient, this
does not necessarily indicate that the financial statements are misstated (e.g., PCAOB, 2014). To take an extreme
example, suppose that an auditor does not conduct any work and signs an unqualified audit opinion. The auditor
could be lucky if the pre-audit financial statements of the issuer are free of misstatements and exhibit high financial
reporting quality. Academic measures of audit quality would exhibit very high quality, but practitioners would
consider this audit to be inadequate due to the lack of work. Conversely, an audit intends to provide reasonable, not
absolute, assurance that the financial statements are free of material misstatement (AU 110). Thus, an audit could be
adequately executed from a practitioner’s standpoint, but several academic measures of audit quality, such as the
probability of a restatement, could still exhibit low audit quality.

3
 ACCEPTED MANUSCRIPT

PCAOB inspects public accounting firms that audit issuers and, as part of each inspection,

reviews audit engagements after they are completed. The PCAOB issues a Part I Finding when

its inspectors determine that, based on applicable audit standards, the work completed by the

engagement team was not sufficient to support its audit opinion.

Because PCAOB inspections are designed to ensure that the engagement team fulfilled its

T
IP
role and because PCAOB inspectors are all experienced former auditors who devote a significant

amount of effort to the inspection process, Part I Findings have the potential, from a

CR
practitioner’s standpoint, to be a precise assessment of audit process quality. Consistent with this

US
view, several studies find evidence of audit quality as well as real economic consequences of

PCAOB inspection findings (e.g., DeFond and Lennox, 2017; Aobdia and Shroff, 2017; Shroff,
AN
2017). Nevertheless, auditors and the academic community have often expressed concerns about

this notion. Their concerns go beyond the PCAOB and are generally applicable to any
M

regulator’s competence, incentives, and potential capture (e.g., Stigler, 1971; Mahoney, 2011).

To mitigate this issue, I also use a proprietary dataset of internal inspections conducted by the
ED

largest audit firms themselves after their audits are completed. In comparison with PCAOB
PT

inspections, internal inspections are carried out by practicing auditors and generally touch on

more parts of an engagement (Houston and Stefaniak, 2013). However, internal inspectors might
CE

be reluctant to identify deficiencies. While, taken separately, each PCAOB and internal

assessment of audit process quality perhaps has some limitations, finding similar results with
AC

both assessments enhances confidence that the results of my analyses are not driven by specific

issues in either.

I use two reasonably large confidential datasets of 5,309 engagements inspected by the

PCAOB and 2,286 internal inspections conducted by the seven largest audit firms, which overlap

4
 ACCEPTED MANUSCRIPT

for 204 observations. For these, the correlation between PCAOB Part I Findings and internal

audit deficiencies is 24%, which indicates that the PCAOB and audit firms somewhat agree on

what constitutes a low quality audit.4 However, internal inspectors, perhaps because of different

incentives, identify comparatively fewer deficiencies. I merge these datasets with Compustat and

Audit Analytics to compute several measures of audit quality and other variables. I rely on prior

T
literature to compute the following fifteen measures of audit quality. First, five output proxies

IP
based on accruals: signed and unsigned discretionary accruals, residuals from the Dechow and

CR
Dichev model, unsigned total accruals, and unsigned total accruals deflated by cash flow from

operations. I particularly focus on accruals because they are widely used in the audit literature.

US
Next, I consider five non-accruals output proxies: the propensities for an issuer to restate its
AN
financial statements, meet or beat the zero earnings threshold, meet or beat prior year’s earnings,

the issuance of going concern opinions and type I errors in going concern opinions. I also
M

evaluate five input proxies for audit quality: Auditor industry specialization, audit fees, audit

hours, whether the client is new, and office size.5 I generally find, with a few exceptions, that the
ED

correlations among publicly available measures of audit quality are low. This is consistent with

DeFond and Zhang (2014, p 276) who mention that publicly available audit quality proxies
PT

reflect different dimensions of audit quality.

CE

Next, I determine whether each measure is associated with practitioners’ assessments of audit

quality. I find that unsigned discretionary accruals, unsigned total accruals, and unsigned
AC

4
Because PCAOB inspectors examine only certain areas of an engagement, whereas internal inspections are more
comprehensive in their scope, internal inspectors sometimes identify deficiencies in areas that are not inspected by
the PCAOB. This mechanically reduces the correlation between PCAOB and internal inspection deficiencies. The
correlation would increase to above 0.5 if I assume that the PCAOB would have identified the deficiencies found by
the internal inspectors, had the PCAOB inspected the areas of the audit where such deficiencies occurred.
5
I also control for whether the audit is conducted by one of the Big 4 audit firms in all specifications that focus on
PCAOB inspections. Therefore my analyses also indicate whether Big 4 audit firms provide higher quality audits. I
cannot conduct this analysis for the internal inspections, which differ across audit firms. Instead I include audit firm
fixed effects in these analyses.

5
 ACCEPTED MANUSCRIPT

accruals deflated by cash flow from operations are associated with Part I Findings, but not with

internal inspection outcomes. I also find that restatements and the propensity to meet or beat the

zero earnings threshold are associated with both Part I Findings and internal inspection

deficiencies. Illustratively, they increase the probability of a Part I Finding by 10.1% and 5.0%,

respectively (to be compared with an average probability of a Part I Finding of 27.5%).

T
IP
For the input measures, audit fees are negatively associated with both Part I Findings and

internal inspection deficiencies. An interquartile range increase (75th percentile less 25th

CR
percentile value) in audit fees decreases the probability of a Part I Finding by 10.7%. First-year

US
clients (Big 4 audit firms) are positively (negatively) associated with Part I Findings, and audit

hours and office size are negatively associated with internal inspection deficiencies. Table 1
AN
provides a summary of which academic measures of audit quality are associated with practitioner

assessments.
M

[Insert Table 1 About Here]

ED

I include in the regressions all measures of audit quality that are consistently associated with

the practitioners’ views. First, I find that restatements, small profits, accruals (absolute
PT

discretionary accruals, total accruals, and accruals scaled by cash flow from operations), and

audit fees are jointly predictive of Part I Findings. Second, I find that restatements, small profits,
CE

and audit fees or hours are jointly predictive of internal inspection ratings. This suggests that
AC

each measure captures a different dimension of audit quality and confirms the need, commonly

adopted in the audit literature, to triangulate audit quality using several proxies (e.g., Lim and

Tan, 2008; Francis and Yu, 2009). However, collectively, these measures explain a small portion,

between 3% and 19%, of the practitioners’ views. This is consistent with two ideas. Either

observable measures of audit quality represent weak proxies for (unobservable) audit processes

6
 ACCEPTED MANUSCRIPT

that practitioners care about, or disagreements exist between practitioners and academics about

what constitutes audit quality. 6

I conduct several additional tests, which focus on PCAOB inspections because the data are

more extensive. I focus on Part I Findings in the auditing of complex accounting estimates, a

recurring focus of the PCAOB (PCAOB, 2016).7 I find that accrual measures and the propensity

T
IP
to meet or beat the zero earnings threshold are associated with these deficiencies, but not with

other types of Part I Findings. I also conduct several tests to mitigate a concern that the

CR
relationship between restatements and Part I Findings is mechanical. In particular, I find that

US
PCAOB inspectors are unlikely to suffer from a hindsight bias whereby they would issue Part I

Findings to engagements for which they are aware of imminent restatements.

AN
Overall, this study contributes to the literature in several ways. First, the significant

associations in this study show that practitioners and academic researchers to a certain extent
M

agree about what constitutes proper audit quality, even though they use widely different
ED

approaches to define and measure it. My study offers guidance about which publicly available

measures of audit quality, for U.S. listed issuers, are predictive of the practitioners’ views. It
PT

extends several review studies that provide a qualitative assessment of observable measures of

audit quality but, due to lack of data, cannot evaluate the degree of concordance with
CE

practitioners’ measures of audit process quality (Francis, 2011; Knechel et al., 2013; DeFond and

Zhang, 2014; Gaynor et al., 2016). The most promising output measures are the issuance of a
AC

restatement and the propensity to meet or beat the zero earnings threshold. The most promising

6
Because PCAOB inspections focus on specific areas of selected engagements, this could result in a lower
predictive power of academic measures of audit quality on PCAOB inspection outcomes. This problem is less severe
for internal inspection outcomes because internal inspections are more extensive in scope.
7
The internal inspection data does not report the nature of the deficiency. Thus, I cannot conduct similar analyses
using this dataset.

7
 ACCEPTED MANUSCRIPT

input measure is audit fees. All three measures are significantly associated with both PCAOB

and internal inspection outcomes. A note of caution is that the predictive power of the combined

measures on PCAOB inspection and internal inspection outcomes appears limited. 8 This suggests

that many unobservable factors influence practitioners’ views of audit quality, and calls for

additional research directed to find more theoretically appealing and empirically precise proxies.

T
In the meantime, the results in this study support current academic practice that uses more than

IP
one proxy of audit quality to reduce the risk of type 1 error.

CR
My results, from a different perspective, also indicate that a poorly conducted audit as per

US
applicable standards, and identified as deficient by practitioners, is associated with worse

reporting outcomes, higher accruals and a higher probability of restatements. Thus, these results
AN
also speak of the value of audits conducted according to audit standards and audit firms’

methodologies, and of the value of practitioners’ assessments of audit process quality. My study
M

also provides unique descriptive evidence about the PCAOB and internal inspection processes,

based on direct access to these data. In particular, this study is among the first to provide
ED

evidence on the differences between high and low quality audits as per PCAOB Part I Findings.
PT

One caveat is that, because PCAOB inspections are risk-based, the results perhaps cannot be

generalized outside of inspected engagements.9 Input-based measures of audit quality could be

CE
AC

8
Another note of caution is the lack of association between going concern opinions and practitioners’ assessments,
possibly because practitioners focus on going concern opinions less than academics do when assessing the overall
quality of an audit. In practice, the work spent on going concern opinions represents only a small portion of an audit
and, thus, they are unlikely to represent a general measure of audit quality. From an academic standpoint, going
concern opinions serve a primary role: a signal of auditor independence (Francis, 2011; DeFond and Zhang, 2014).
Recent research also suggests a disclaimer role: perceived or actual litigation risk is lower following the issuance of
a going concern opinion (Mutchler, 1984; Carcello and Palmrose, 1994; Kaplan and Williams, 2013). The auditor
thus could have an incentive to lower effort for the remainder of the audit (Dye, 1993). I find in several univariate
tests a negative association between going concern opinions and Part I Findings, consistent with the disclaimer role.
9
Aobdia et al. (2017) find that selection bias using the engagements inspected by the PCAOB appears not to be
severe. However, whether the evidence can be generalized outside of their study remains unclear.

8
 ACCEPTED MANUSCRIPT

more sensitive to this risk-based approach.10 Similar concerns could exist for internal inspections,

given that internally inspected engagements are on average larger. However, the associations in

this study still provide evidence about the behavior of audit-quality proxies amount the sub-

sample of inspected audits. Further, the estimates do not vary much when conducting sensitivity

analyses similar to Altonji et al. (2005), which assume different correlations between unobserved

T
factors that could determine the selection for a PCAOB inspection and an inspection outcome.

IP
The remainder of this paper is structured as follows. Section 2 provides background on

CR
academic and practitioners’ views of audit quality, Section 3 discusses the data, Section 4

US
considers the main empirical tests, and Section 5 reports on additional tests. Section 6 concludes.

2. Background on Academic and Practitioners’ Views of Audit Quality

AN
2.1 Academic views of audit quality
The academic literature generally uses an outcome-based definition of audit quality.
M

DeAngelo (1981) defines audit quality as the probability that the auditor will both discover and

report material misstatements. DeFond and Zhang (2014, p276) refine this definition to “greater
ED

assurance that the financial statements faithfully reflect the firm’s underlying economics,

conditioned on its financial reporting system and innate characteristics.” Overall, these
PT

definitions relate to an auditor’s competence, effort level, and independence (Bell et al., 2015).
CE

Lack of effort or competence prevents an auditor from detecting issues to be resolved, and lack

of independence prevents an auditor from correcting issues identified in the client’s pre-audit
AC

financial statements. However, these constructs are not directly or publicly observable and

archival audit researchers instead use a wide set of indirect proxies based on measurable inputs

10
Illustratively, the PCAOB can fully implement its risk-based approach for Big 4 auditors that audit many issuers.
In contrast, the PCAOB can only inspect one issuer for a small auditor with only one client. Thus, the negative
coefficient on Big 4 auditors in the analyses of Part I Findings could be biased toward zero, because the riskiest
engagements of the Big 4 are compared with more typical engagements of smaller audit firms.

9
 ACCEPTED MANUSCRIPT

and outputs of the audit process (e.g., Knechel et al., 2013). The Online Appendix provides a

description of commonly used output and input measures of audit quality.

Output measures generally focus on the characteristics of audited financial statements; input

measures, on characteristics of the auditor. Importantly, most output proxies of audit quality,

except going concern opinions, are a joint function of the issuer’s innate characteristics, its pre-

T
IP
audit financial reporting system, and the work of the auditor (DeFond and Zhang, 2014, p284).

Thus, they cannot differentiate a low quality from a high quality audit when the issuer’s pre-audit

CR
financial statements are of high quality. A major limitation of the input measures is that they are

US
restricted to observable characteristics of the audit firms, which are limited under current

disclosures.
AN
In contrast, practitioners’ assessments of audit quality concentrate directly on the level of

competence, effort, and independence of an auditor. Practitioners have access to detailed

M

information that allows them to determine whether sufficient audit work was completed in light

of the issuer’s circumstances (which depend on the issuer’s pre-audit financial reporting systems,
ED

internal control systems, and economics). Thus, they proxy more directly for what academic
PT

researchers define as audit quality but can only measure using indirect proxies. However,

practitioners’ assessments also strive to pursue a definition of audit quality that can be reliably
CE

verified, in order to assess compliance with regulation. They are based on applicable audit

standards and, therefore, could be noisy or even inaccurate if the standards are inadequate.
AC

Further, due to the complexity of audits and the extensive use of judgment involved, such

assessments are inherently subjective.

2.2 PCAOB Inspections

2.2.1 Background

10
 ACCEPTED MANUSCRIPT

As part of SOX, Congress established independent oversight of the accounting profession by

the PCAOB for audits of issuers. Since its creation, the PCAOB has, each year, conducted

hundreds of inspections of registered public accounting firms that audit issuers. These

inspections are annual for audit firms that regularly provide audit reports for more than 100

issuers and at least triennial otherwise (Section 104 of SOX).

T
IP
Per Section 104 of SOX, a PCAOB inspection focuses on two elements: a review of

individual audit engagements (the focus of this paper), and a review of the overall quality control

CR
systems of an audit firm. 11 Before an inspection begins, the PCAOB notifies the audit firm and

US
requests information about the audit firm’s public engagements and the personnel conducting

these audits (CAQ, 2012). The PCAOB cannot inspect all the engagements of a particular audit
AN
firm every year and uses a risk-based approach to select the engagements that will be inspected.

According to CAQ (2012, p3),

M

“The PCAOB has developed a variety of tools to identify audits that may pose difficult or
complex issues. Risk factors include the nature of the company, including its industry and market
capitalization; audit issues likely to be encountered; and whether the company has significant
ED

operations in emerging markets. Other factors that influence engagement selection are specific
to the inspected firm, such as the type and range of its public company engagements, the results
of prior PCAOB inspections, and findings from the firms’ internal risk management and
PT

inspections processes. The inspection staff also considers the assignments and inspection history
of the partners who audit public companies.”
An illustrative timeline of the reviews of an auditor’s engagements is presented in Figure 1.
CE

PCAOB inspection fieldwork of an engagement typically begins a few months after the audit has
AC

been completed. This fieldwork usually is conducted at the audit office in charge of the

engagement. Sometimes, for smaller audit firms, the audit work papers are reviewed in PCAOB

offices and interviews are conducted by phone (CAQ, 2012). The fieldwork lasts approximately

one week, although some inspections can finish in four business days or last two weeks or more

11
See Aobdia (2018) for a description of the PCAOB’s inspection of audit firms’ quality control systems.

11
 ACCEPTED MANUSCRIPT

(Riley et al., 2008; Johnson et al., 2014).12 During the fieldwork, PCAOB inspectors typically

focus on three areas of the audit, with an area defined by the type of transaction. The most

frequently selected financial reporting areas in 2015 include revenue and accounts receivable,

nonfinancial assets, and inventory and financial instruments (PCAOB, 2016). Audit firms are not

allowed to influence the selection of engagements or parts of engagements inspected by the

T
PCAOB (PCAOB, 2014).

IP
[Insert Figure 1 About Here]

CR
When on-site, PCAOB inspectors dissect the audit work papers, extensively interact with the

US
engagement team to understand the work completed during the audit, and determine whether the

work performed by the engagement team is sufficient to support the audit opinion. 13 When the
AN
inspection team identifies a potential issue, it first discusses the issue with the engagement team.

If the inspection team is unable to resolve the issue through discussion and any review of
M

additional work papers, it issues a comment form shortly after the inspection fieldwork (e.g.,
ED

PCAOB, 2014). The audit firm then has the opportunity to provide a written response to the

comment form.14 If the response does not address the inspection team’s concerns, and the issue is
PT

material, the PCAOB issues a Part I Finding for that specific engagement. Three examples of

Part I Findings, provided in Appendix B, show that PCAOB inspectors can often identify
CE

deficiencies even if the issuer’s financial statements eventually exhibit high financial reporting
AC

12
In many instances, additional work is also conducted within the PCAOB offices outside of the one-week
fieldwork window (Riley et al., 2008).
13
One potential issue regarding the inspection program is that PCAOB inspectors have direct access only to audit
work papers, not client data. In some instances, the auditors might not include in the audit work papers some issues
they are aware of to avoid regulatory scrutiny or the risk of litigation.
14
Some audit firms may already conduct additional audit work at this point, in accordance with standards AU 390
(consideration of omitted procedures after the report date) and AU 561 (subsequent discovery of facts existing at the
date of the auditor’s report), because they agree with the PCAOB findings. However, this additional work will not
be taken into account by the PCAOB in the determination of a Part I Finding because it is conducted after the audit
opinion’s issuance date (in other words, the audit firm cannot avoid a Part I Finding by conducting additional work).

12
 ACCEPTED MANUSCRIPT

quality. Part I Findings are made public in the inspection reports of individual audit firms

disclosed by the PCAOB. However, the name of the issuer is masked in the public reports.

Further, the specific engagements selected for inspection are not publicly disclosed. Thus, an

important part of the inspection process is not disclosed to the public and only aggregate

inference can be made with publicly available data.

T
IP
2.2.2 Advantages of using PCAOB inspections as a practitioner’s assessment of audit quality
Using PCAOB inspection outcomes has three advantages. First, they derive from a standards-

CR
based view of audit quality and focus on the work of the auditor conditional on the issuer’s

circumstances. 15 Appendix B presents several examples of PCAOB deficiencies. They show that

US
PCAOB inspections focus on whether each audit appropriately responds to particular issuer
AN
circumstances. Consistent with the arguments in DeFond et al. (2018), PCAOB inspectors focus

on professional judgment from auditors to ensure that financial statements are fairly presented. In
M

Appendix B.1, the auditor did not assess why the issuer recorded a different amount on a

contingent liability than estimated by the specialist hired for this purpose. In Appendix B.2, the
ED

auditor did not sufficiently test the issuer’s revenue recognition process. In Appendix B.3, the

auditor did not test the assumptions made by the issuer to compute its allowance for doubtful
PT

accounts. These issues could potentially be serious and impact the issuer’s financial statements.
CE

Second, the PCAOB spends considerable effort conducting inspections, suggesting that Part I

Findings could be a precise signal of poor audit process quality. Inspection teams are composed
AC

of experienced and knowledgeable former auditors (PCAOB, 2011). Lennox and Pittman (2010)

report that inspectors average 12 years of public practice experience, an observation corroborated

15
Audit standards instruct auditors to focus on such circumstances. For example, AU 342 focuses on the auditing of
management estimates and instructs the auditor to evaluate the reasonableness of the estimate.

13
 ACCEPTED MANUSCRIPT

by the current job requirements for inspectors (see the Online Appendix).16 Many inspectors

previously worked for the Big 4 audit firms at the manager level and above, audited publicly

traded companies, and have extensive knowledge of auditing practices. Further, as full-time

PCAOB employees, they are independent from the public accounting profession and build

extensive inspection expertise (Lennox and Pittman, 2010; PCAOB, 2011; Carcello et al., 2011).

T
The inspectors spend a significant amount of time analyzing the work papers of the inspected

IP
engagement, and interacting with the audit team, to refine their understanding of the particular

CR
engagement and the mindset of and the procedures used by the engagement team. Further, audit

firms are given several opportunities during the inspection process to provide additional

US
perspectives about the audit (PCAOB, 2012; CAQ, 2012), thereby reducing potential noise in the
AN
inspection process and increasing the accuracy of Part I Findings. Survey evidence also suggests

that in general PCAOB inspectors have the proper qualifications to conduct their role and that
M

their fieldwork is appropriate. Houston and Stefaniak (2013) survey 107 partners of large

auditing firms and find that partners agree with the following statements: The PCAOB
ED

inspections are very detailed within the areas inspected, and the inspectors concentrate on

assessing sufficiency of evidence gathered, judgments made during substantive testing, audit
PT

documentation, and internal control evaluations. Based on survey evidence, triennially inspected
CE

audit firms also agree that PCAOB inspectors possess adequate technical knowledge, exercise

appropriate professional conduct, and that their focus is appropriate (Daugherty and Tervo, 2010).
AC

Third, survey data finds that PCAOB inspections are reasonably unpredictable (Houston and

Stefaniak, 2013), suggesting that engagement teams are unable to strategically improve quality

16
PCAOB (2007) also mentions that the inspection team leaders for large inspections have an average of 25 years of
relevant work experience and that all other inspection team members average 14 years of relevant work experience.

14
 ACCEPTED MANUSCRIPT

on their audits in anticipation of an inspection. Such strategic actions have the potential to distort

the relationship between publicly available measures of audit quality and the inspection results.

2.2.3 Concerns about using PCAOB inspection data as a practitioner’s assessment

Several concerns arise about using PCAOB inspections data. Specifically, whether a Part I

Finding is an appropriate measure of audit process quality crucially depends on the PCAOB’s

T
competence, incentives, and regulatory capture (e.g., Stigler, 1971; Mahoney, 2001).

IP
PCAOB inspectors are experienced former auditors. Nevertheless, prior literature questions

CR
whether their expertise is sufficient (Palmrose, 2006). The PCAOB thus could call attention to

US
the lack of some procedures, even if they are not necessary in the first place. In Johnson et al.

(2014), five interviewees out of 19 mention that PCAOB inspectors question too much the
AN
auditors’ judgments and conclusions, despite insufficient knowledge of the history of particular

clients.17
M

PCAOB incentives might also lead inspectors to make Type I errors in identifying
ED

deficiencies. Velikonja (2016) finds that the SEC, as well as other government agencies, reports

flawed statistics to protect its ability to continue enforcing the law. Multiple reporting statutes
PT

authorize Congress to cut agencies’ budgets for failing to meet performance targets. While it is

an empirical question whether, given its design, the PCAOB suffers from similar issues, Houston
CE

and Stefaniak (2013) report that their surveyed audit partners agree that the primary focus of

PCAOB inspectors is to find deficiencies. Further, 17 partners mention in open-ended comments

AC

that PCAOB inspectors make big issues out of immaterial ones.

17
Big 4 audit firms’ replies to earlier inspection reports echo this perception. In its 2007 inspection reply, Deloitte &
Touche (U.S.) states: “We believe that the observations included in the draft report reflect the fact that professional
judgments are involved both in auditing an issuer’s financial statements as well as in subsequently inspecting any
such audits. Professional judgments of reasonable and highly competent people may differ as to the nature and
extent of necessary auditing procedures, conclusions reached and required documentation. We believe that
reasonable judgments should not be second guessed…”

15
 ACCEPTED MANUSCRIPT

To reduce potential regulatory capture, SOX designed the PCAOB as a distinct, nonprofit

organization. Two of the PCAOB’s five Board members must be Certified Public Accountants,

and three must be independent of the accounting profession. The aim is for the PCAOB to have

sufficient expertise but not to be captured (Ferguson, 2012). PCAOB funding sources also ensure

that the PCAOB is not financially dependent on the largest audit firms. However, the SEC, a

T
government agency, appoints the PCAOB Board members and must approve the PCAOB’s

IP
budget, litigation, and rules. While the PCAOB may not be regulatory captured by the auditing

CR
profession, the SEC, especially its office of the chief accountant, might be (Levinson, 2015).

Furthermore, while PCAOB inspectors are full-time employees, they are usually hired from the

US
Big 4 audit firms and the risk of revolving door is always possible. 18
AN
Additional concerns stem from the inspection dataset. The most salient is sample loss

resulting from the limited number of inspections the PCAOB is able to conduct every year. This
M

sample loss is not random, due to the risk-based nature of PCAOB inspections. Further, because

the number of engagements selected for inspection is limited, the probability that a given
ED

engagement is selected several times over the sample period is low, which results in a lack of
PT

time-series data. Thus, the results of this paper are relevant mostly for studies that focus on

cross-sectional analyses.19 An additional issue is that a Part I Finding is a zero-one outcome. No

CE

information is available on the audits that passed the inspection, especially how close they were

to the failure threshold. Last, PCAOB inspections focus on specific areas of the engagement,
AC

often those that appear to the inspectors to be the most critical for the audit (Hanson, 2012;

18
As discussed in Levinson (2015) and per my experience working as a Fellow, the PCAOB has strong legal
controls to prevent its employees from having any real or perceived conflict of interests. However, the revolving
door could still be a risk at the PCAOB, as evidenced by the recent KPMG leak case (e.g., Rapoport, 2018).
19
Many auditing papers use panel data but examine cross-sectional predictors of audit quality. Consequently, the
results of this study are applicable to them. However, a handful of papers, including Gul et al. (2013) and Aobdia et
al. (2015), use time-series or difference-in-difference research designs.

16
 ACCEPTED MANUSCRIPT

PCAOB, 2012; CAQ, 2012). Even though audit partners of large audit firms neither agree nor

disagree with the statement that a PCAOB inspection touches on all audit areas in an engagement

(Houston and Stefaniak, 2013), suggesting that in practice the inspectors do concentrate on the

most critical areas of the audit, the PCAOB focus on sub-areas of the audit indicates that, for a

given engagement, not all audit deficiencies may be identified by the inspectors. Noise thus

T
could be introduced in the measure, reducing the power of the tests and accounting at least

IP
partially for the low explanatory power of some regressions. 20

CR
2.3 Mitigating several of the concerns of the PCAOB inspection data with internal inspections
Large audit firms, as part of their quality control systems, conduct internal inspections on a

US
sample of their own audits, shortly after their completion. The aim is to evaluate the performance
AN
of the engagement team and assure that the audit meets the methodologies and policies of the

audit firm as well as audit standards. Experienced internal reviewers conduct internal inspections.
M

Separate lines of authority, professional training, and specific incentives generally ensure that

reviewers are competent and independent (e.g., Deloitte, 2010; Bell et al., 2015).
ED

Internal inspections are important because they represent the views of the audit firms on the

quality of their own audits. Thus, they are devoid of regulatory incentives. However, audit firms
PT

may have other incentives when conducting internal reviews. Survey evidence suggests that
CE

internal inspections can complement PCAOB inspections. In comparison with PCAOB

inspections, internal inspections consider more aspects of an audit, understand the audit firm’s
AC

methodology better, are less inclined to seek out deficiencies, and provide more timely feedback

(Houston and Stefaniak, 2013). Another advantage of using internal inspection data is that the

20
In other words, higher audit quality cannot be unanimously inferred from the absence of a Part I Finding.

17
 ACCEPTED MANUSCRIPT

level of agreement between assessments of audit quality by the PCAOB and auditors can be

measured for engagements that are inspected by both.

3. PCAOB Sample Data construction and measures of audit quality

3.1 PCAOB Sample Construction
I obtain PCAOB inspection and Part I Finding data from the PCAOB. These data cover the

T
inspected engagements for fiscal years between 2003 and 2013, and they include the name of the

IP
issuer, its Central Index Key (CIK), its auditor and fiscal year, and whether a Part I Finding is

CR
issued. In case of a Part I Finding, the data also include the text of the finding, the audit standards

that are not met, and the accounting area with the deficiency. While the data do not report

US
whether PCAOB inspectors identify a departure from GAAP, this information can be inferred

from the text of the Part I Finding. The dataset contains 9,696 unique inspection engagement
AN
reviews. I merge it with Compustat and Audit Analytics to obtain appropriate measures of audit

quality and control variables, reducing the sample size to 7,724 observations. I restrict the
M

sample to two-digit SIC industries with more than ten observations each year to permit the
ED

computation of discretionary accruals. After inclusion of all control variables, the final sample is

restricted to 5,309 inspected engagements.21

PT

3.2 Internal Inspections Sample Construction

As part of its inspection of the quality control systems of an audit firm, the PCAOB has
CE

access to the internal inspection ratings of the U.S. operations of the seven largest audit firms
AC

(e.g., Aobdia, 2018). I obtain several internal inspection documents, which indicate when the

internal inspection took place, the identity of the issuer inspected, and the internal inspection

21
Some studies exclude from the analyses financial sector (SIC codes 60-69) and regulated industries (SIC codes
44-49). The results are generally unchanged when excluding these industries.

18
 ACCEPTED MANUSCRIPT

rating.22 While the rating scales can differ across audit firms and even across time within the

same audit firm, they can be grouped into three categories that are used in the majority of the

sample: satisfactory, satisfactory with comments, and unsatisfactory. The difference between the

last two categories is based on whether the identified issues are immaterial or material.

Because the central index code (CIK) of the issuer is not available in the spreadsheets, I

T
IP
match the identity of the issuer with Compustat based on the name. After inclusion of all control

variables to conduct meaningful analyses, I obtain a final sample composed of 2,286

CR
observations. The dataset spans 2005 to 2013, but the audit firm coverage is limited prior to 2008.

3.3 Descriptive Statistics

US
Descriptive statistics on PCAOB and internal inspections are presented in Table 2. Panel A
AN
shows the number of inspections. The average probability of inspection in the sample is 10% for

PCAOB inspections, and 15% for internal inspections.23

M

[Insert Table 2 About Here]

ED

Panel B reports that 27.5% of the PCAOB inspections result in Part I Findings. In contrast,

only 7.7% of internal inspections result in unsatisfactory outcomes. Audit firms and the PCAOB
PT

may have different thresholds about what constitutes an unsupported audit opinion. And,

PCAOB inspectors (internal inspectors) could face incentives to identify more (fewer) audit
CE

deficiencies. Compared with PCAOB inspections, internal inspections outcomes also provide

more variation with an intermediate category, “satisfactory with comments”, where issues are
AC

22
In contrast with the PCAOB inspection data the internal inspection data incorporate only the final rating, not the
nature of the deficiencies. Thus I am unable to conduct analyses focusing on the nature of the deficiencies.
23
For comparability purposes, non-inspected engagements are restricted to issuer-year observations for which all
control variables are available, similar to the inspected engagements. This restriction tends to overstate the
probability of an inspection by the PCAOB, which would be lower, at approximately 6%, if no restriction on the
control variables were imposed for both inspected and non-inspected samples. Furthermore, the observations are
restricted to audit firms and years for which internal inspection data are available for the internal inspection sample.

19
 ACCEPTED MANUSCRIPT

identified but are considered immaterial by the audit firm. 11.1% of internal inspections are

deemed satisfactory but with comments.

There is a subsample of 204 engagements that are inspected by both PCAOB and internal

inspections (this overlap could suffer from selection bias given that the PCAOB does not

necessarily randomly inspect engagements that were already internally inspected). For this

T
IP
subsample, the correlation between PCAOB and internal inspection outcomes equals 0.24

(significant at 1%), indicating that the PCAOB and auditors do statistically agree, albeit not

CR
always, on the quality of the engagements they inspect. Further, untabulated analyses indicate

US
that this correlation has been increasing over time (significant at 10%). While a correlation of

0.24 may appear low, apparently indicating some disagreement between the PCAOB and audit
AN
firms about the quality of a given audit, PCAOB inspections tend to be narrower in scope than

internal inspections (Houston and Stefaniak, 2013). Thus, in some instances, internal inspectors
M

may identify deficiencies in areas that PCAOB inspectors did not inspect.
ED

Panel C splits the sample between inspections of Big 4 and non-Big 4 auditors, as well as by

year. For PCAOB inspections, 56% of the engagements inspected are for the Big 4. This
PT

proportion is smaller than the entire population at the intersection of Audit Analytics and

Compustat, with Big 4 audit firms representing 68% of the audits. The total number of
CE

engagements inspected varies from a low of 315 in 2009 to a high of 647 in 2007. Regarding

internal inspections, 9% of the sample is composed of internal inspections from non-Big 4 audit
AC

firms. The number of internal inspections in the sample greatly varies over time due to limited

audit firm coverage before 2008.

Panel D compares the engagements based on whether they were selected for internal

inspection. Differences in means are presented but the differences in distributions are similar

20
 ACCEPTED MANUSCRIPT

(untabulated). Internally inspected engagements are larger than non-inspected engagements.

Average assets are $1.3 billion compared with $800 million. Reputation and litigation risk

incentivize large audit firms to conduct higher quality audits. These risks, which are higher for

larger clients (e.g., Reynolds and Francis, 2000), encourage audit firms to monitor their audits

more closely.24 Untabulated analyses show that the difference in size often explains why the

T
measures of audit quality and several other control variables are different across both subsamples.

IP
I am unable to report similar statistics for PCAOB inspections, due to the need to preserve

CR
the confidentiality of the inspection process. However, it is plausible that the PCAOB also

US
concentrates on larger engagements, for similar reasons. For example, CAQ (2012, p 3) states

that, for smaller audit firms, “the PCAOB generally focuses on audits of companies with the
AN
largest market capitalizations.” Further, the PCAOB, for its 2014 inspections of the Big 4 in the

U.S., began reporting the revenue ranges of the inspected engagements. Based on the 2014
M

reports, 39% of the inspected engagements had revenues below $1 billion. This contrasts with

approximately 50% of such engagements in the entire population audited by the Big 4.
ED

3.4 Measures of Audit Quality

3.4.1 Accruals and Discretionary Accruals
PT

Following prior literature, I estimate discretionary accruals using the cross-sectional

CE

modified Jones model (Dechow et al., 1995; Kothari et al., 2005; Reichelt and Wang, 2010):

TAi,t/ASSETi,t-1=β1(1/ASSETi,t-1)+β2(ΔSALESi,t)/ASSETi,t-1+β3PPEi,t/ASSETi,t-1+ β4ROAi,t-1+ εi,t, (1)

AC

where TA is total accruals measured as earnings before extraordinary items (IB) minus net cash

flow from operations excluding extraordinary items and discontinued operations (OANCF-

XIDOC), ΔSALES is change in sales, PPE is gross property, plant, and equipment, and ROA is

24
Increased internal monitoring of larger clients may serve as an explanation for why auditors report more
conservatively for larger clients, despite higher economic dependence on such clients (Reynolds and Francis, 2000).

21
 ACCEPTED MANUSCRIPT

return on assets. I estimate model (1) by industry (two-digit SIC code) and year. Discretionary

accrual (Disc. Accruals) is the residual from the model, εi,t. I use both Disc. Accruals and its

absolute value, Abs(Disc. Accruals), as potential measures.

Following Leuz et al. (2003), I also consider two accrual-based measures of audit quality:

Accruals equal to TA/ASSET, and Accruals/CFO, equal to total accruals (TA) deflated by the

T
IP
absolute value of the issuer’s cash flow from operations (OANCF-XIDOC). I take their absolute

values, denominated Abs(Accruals) and Abs(Accruals/CFO).

CR
3.4.2 Dechow and Dichev residuals
I also estimate the Dechow and Dichev (2002) residuals.25 I augment the Dechow and Dichev
US
model following McNichols (2002) and Francis, Lafond et al. (2005).
AN
TCAi,t = β0+β1CFOi,t-1 + β2CFOi,t + β3CFOi,t+1+ β4 ΔSALESi,t+ β5PPEi,t+ εi,t, (2)

where TCA is total current accruals (= Δ inventories + Δ accounts receivable - Δ accounts

M

payable - Δ income tax payable + net change in other assets and liabilities, all from the cash flow

statement), and CFO is cash flow from operations (OANCF-XIDOC). All variables are scaled by
ED

average assets during the year. I estimate equation (2) by industry (two-digit SIC code) and year.

The absolute value of the residual, DD Residual, is a potential measure of audit quality.26
PT

3.4.3 Other measures of audit quality

CE

Restatements are commonly used as a measure of audit quality. Restatement is an indicator

variable for whether the financial statements for the year are restated. I consider only fiscal year-
AC

end restatements, consistent with Lobo and Zhao (2013). The data are from Audit Analytics.

25
Although the most frequently used accrual measure of audit quality is based on the Jones (1991) model, several
studies measure audit quality with the Dechow and Dichev (2002) model (DeFond and Zhang, 2014).
26
Dechow et al. (2010) and Dechow and Dichev (2002) suggest that the absolute value of the residual is an
appropriate measure of accrual quality. They also propose using the standard deviation of the residuals. While this
measure is appropriate from an accruals quality standpoint, the main disadvantage, from an audit quality standpoint,
is that it is not directly linked with a particular audit anymore, but with a series of audits over the estimation period.
Untabulated analyses indicate that the results are unchanged if I instead use the standard deviation of the residuals.

22
 ACCEPTED MANUSCRIPT

Going concern opinions are also commonly used. Going Concern is an indicator variable for

whether the auditor issued a going concern opinion. The data are from Audit Analytics. Because

most issuers with going concern opinions do not go bankrupt, I also define Type I Going

Concern as an indicator if the auditor issues a going concern opinion and the issuer does not

declare bankruptcy the following year. I use both UCLA Lopucki and SDC databases to

T
determine whether an issuer declares bankruptcy. Because going concern opinions are usually

IP
issued to distressed issuers, when using Going Concern and Type I Going Concern as dependent

CR
variables, I restrict the sample to distressed issuers, defined as reporting negative income before

extraordinary items or negative cash flow from operations, consistent with prior literature

(Reynolds and Francis, 2000; DeFond et al., 2002). US

AN
Several studies use the propensity to meet or beat earnings thresholds as measures of audit

quality. Following prior studies (e.g., Francis and Yu, 2009), Small Profit is an indicator variable
M

for whether the ROA (income before extraordinary items deflated by beginning assets) is

between 0% and 3%.27 Prior ROA Meet is an indicator variable for whether the year-on-year
ED

change in ROA is less than 1%.

PT

Following prior studies, Industry Specialization is the within-industry auditor market share,

based on audit fees charged by the auditor in the client’s two-digit SIC industry (e.g., Minutti-
CE

Meza, 2013). Big 4 is an indicator variable for whether the auditor is a Big 4 audit firm. I include

this variable in all regressions with PCAOB outcomes as the dependent variable. New Client is
AC

an indicator variable for whether the auditor-client relationship is in its first year. Office Size is

the natural logarithm of one plus the total audit office’s fees.

27
Results are qualitatively unchanged when using other numbers, such as 1%, 4%, and 5%.

23
 ACCEPTED MANUSCRIPT

I also use Audit Fees, the natural logarithm of audit fees reported in Audit Analytics. Because

audit fees are the product of unit price and the quantity of audit services required, and thus

potentially proxy for different factors (Simunic, 1980), I split audit fees between Audit Hours,

the logarithm of audit hours reported by the audit firms to the PCAOB, and Fees Per Hour, audit

fees divided by audit hours. I consider Fees Per Hour in untabulated analyses and do not find

T
any association with Part I Findings or internal inspection deficiencies. Audit hour data are

IP
available from 2008 to 2013 for the U.S. operations of the largest audit firms. The advantage of

CR
using audit hours is that they represent a direct measure of auditor effort. They can also be

correlated with audit fees, to determine whether audit fees represent an accurate proxy for

auditor effort. US
AN
4. Main Empirical Tests
4.1 Research Design
I estimate whether a given measure of audit quality is associated with PCAOB or internal
M

inspection outcomes using the following regressions:

ED

Part I Findingi,t=β0+β1.Audit Quality Measurei,t+β.Controlsi,t+ Industry-Year Effects + εi,t (3)

and

Internal Ratingi,t =β0+β1.Audit Quality Measurei,t+β.Controlsi,t+Industry-Year Effects+Auditor

PT

Fixed Effects + εi,t, (4)

corresponding to the PCAOB and internal inspection samples. Part I Finding is an indicator
CE

variable for whether the PCAOB identifies a Part I Finding. 28 Because of the binary nature of the
AC

dependent variable, I estimate Model (3) using a logistic specification. Internal Rating shows the

outcome of the internal inspection, which takes the value one for satisfactory engagements, two

for satisfactory engagements with comments, and three for unsatisfactory engagements. The

28
In untabulated analyses I consider a slightly different research design, replacing Part I Finding with the logarithm
of one plus the count of Part I Findings identified by the PCAOB for each inspected engagement. The tenor of the
results remains unchanged.

24
 ACCEPTED MANUSCRIPT

higher the rating, the worse the assessment of audit quality. Audit Quality Measure is one of the

fifteen measures of audit quality defined in section 3.4 and is either one of the five measures of

accruals, one of the five non-accruals output measures of audit quality, or one of the five input

measures of audit quality. Controls is a vector of control variables that prior research has

identified as influencing audit fees and quality (Francis et al., 2005; Francis and Yu, 2009; and

T
Reichelt and Wang, 2010). This vector of variables, which could influence the issuance of a Part

IP
I Finding, is composed of Size, the natural logarithm of the issuer’s assets, Foreign Income, the

CR
absolute value of pretax income from foreign operations divided by the absolute value of pretax

income, Geographic Segments, the number of geographic segments of the issuer, and Business

US
Segments, the number of business segments of the issuer, to control for issuer and therefore audit
AN
complexity. Larger and multinational corporations are likely to require multi-location audits that

may necessitate more auditor coordination. December Year End is an indicator variable equal to
M

one if the issuer’s fiscal year ends in December, to control for the auditor’s busy season and

potential time constraints during the audit period. I also include variables that are related to
ED

issuer risk, because such issuers may be more difficult to audit.29 The variables are Leverage, the

issuer’s leverage ratio (total debt divided by total debt plus book equity), BTM, its book-to-
PT

market ratio, CFO, the issuer’s cash flow from operations deflated by beginning assets,
CE

Std(CFO), the standard deviation of CFO computed from years t-3 to year t, Sale Growth, the

year-on-year sales growth of the issuer, and Litigation, an indicator variable for whether the
AC

issuer is in a high-litigation industry.30,31 Because the PCAOB scrutiny is more intense for

29
For example, Matsumoto (2002) finds that issuers with high litigation risk and growth prospects are more likely to
be concerned about missing earnings benchmarks.
30
Because earnings equal accruals plus cash flow from operations, including CFO could create mechanical
relationships in the regressions for accruals based measures. Results are qualitatively unchanged if I exclude CFO
from the regressions or measure CFO for the prior year.
31
Prior audit literature that focuses on accruals also often controls for return on assets and the propensity to realize
losses, instead of cash flows (Reichelt and Wang, 2010; Minutti-Meza, 2013). I do so in untabulated analyses when

25
 ACCEPTED MANUSCRIPT

annually inspected larger audit firms, I also include Big 4, an indicator variable equal to one

when the audit firm is one of the Big 4. I also add in the analyses of going concern opinions

Altman Z Score, the issuer’s Altman Z-score, to control for the probability of financial distress.

Given that the PCAOB often issues Part I Findings in the domain of internal control over

financial reporting (ICFR), I also include Integrated Audit, an indicator variable for whether the

T
engagement is an integrated audit of ICFR and financial statements, and Material Weakness, an

IP
indicator variable for whether the auditor identifies a material weakness. The relationship

CR
between Material Weakness and Part I Finding is unclear ex-ante. On the one hand, the PCAOB

is often concerned about the non-issuance of a material weakness when it is warranted, thereby

US
suggesting a negative relationship. On the other hand, issuers with material weaknesses are likely
AN
to be more difficult to audit, thereby suggesting a positive relationship. Appendix A provides

detailed variable definitions. I winsorize continuous variables at the 1st and 99th percentiles to
M

reduce the impact of outliers, and cluster standard errors at the issuer level.32 I also include

industry times year fixed effects based on Fama French ten industry groups to control for
ED

specific industry focus over time of PCAOB and internal inspections. I also add auditor fixed

effects in Model (4) because different audit firms may conduct their internal inspections
PT

differently.33 Thus, Big 4 is not identified in Model (4).

CE

4.2 Descriptive Statistics and Correlations

Descriptive statistics on the PCAOB inspection sample are presented in Panel A of Table 3.
AC

Out of the 5,309 inspections in the sample, 1,933, or 36% of the total, are inspections of

using accrual measures as dependent variables and generally find qualitatively unchanged results. However, the
significance of the coefficient on absolute discretionary accruals in the regression of PCAOB Part I Findings falls
below conventional statistical levels.
32
Results are qualitatively unchanged if I cluster the standard errors at the year level.
33
One issue of adding auditor fixed effects in Model (3) is that doing so removes important cross-sectional variation
in the data, especially for smaller auditors that are inspected infrequently and do not have much variation in their
inspection outcomes. Most results in Model (3) are robust to including auditor fixed effects, but the statistical
significance of the coefficients on accruals falls below conventional levels.

26
 ACCEPTED MANUSCRIPT

engagements of distressed issuers. Further, 12% of the issuers’ financial statements are

eventually restated, and 14% of issuers report small profits. Also, 21% of inspected engagements

of distressed issuers receive going concern opinions, and 71% of the issuers have December

year-ends and thus correspond to busy-season audits.

Panel A also reports the means of the subsamples partitioned on whether the PCAOB

T
IP
identifies a Part I Finding. Inspections with Part I Findings are more likely to receive going

concern opinions, at 26% compared with 19% (for distressed issuers), and the issuer’s financial

CR
statements are more likely to be restated, at 16% compared with 11%. Further, they are more

US
likely to report small profits, at 16% compared with 13%, and their absolute discretionary

accruals, absolute accruals, absolute accruals deflated by cash flow from operations, and Dechow
AN
and Dichev residuals are higher (all results are significant at 5%).34 An audit deemed deficient by

the PCAOB is thus associated with worse financial reporting outcomes, including higher
M

accruals, an increased probability of restatements, and a higher propensity to meet or beat the

zero earnings thresholds. This confirms the concordance between a low-quality audit, based on
ED

the PCAOB assessment, and output measures of audit quality. However, the direction of the
PT

going concern opinion goes opposite to prior claims in the literature, which could be consistent

with a disclaimer effect on the audit of the going concern opinion. There is no difference
CE

between groups in signed accruals and in the propensity to beat last year’s earnings. Further,

engagements receiving Part I Findings are less likely to be audited by Big 4 audit firms and
AC

industry specialist auditors, and include more new clients. Reaching any conclusion for Audit

34
Some of these results are related to Gunny and Zhang (2013), who run auditor-level regressions using the public
portions of the PCAOB inspection reports and find, for triennially inspected audit firms, a weak association between
audit deficiencies and abnormal current accruals and restatements. However, they find conflicting results for
annually inspected auditors. Furthermore, due to the nature of the PCAOB public reports, their explanatory variable
can distinguish only inspected audit firms without any Part I Findings from those with at least one Part I Finding.
This does not provide any identification for larger audit firms which receive Part I Findings every year.

27
 ACCEPTED MANUSCRIPT

Fees, which is lower for the Part I Finding group, is difficult without controlling for the size of

the issuer, which is also lower. Some of the other control variables are different across groups,

confirming the need to control for them in multivariate tests.

[Insert Table 3 About Here]

T
Descriptive statistics on the internal inspection sample are presented in Panel B. Out of the

IP
2,286 internal inspections, 594, or 26% of the total, are inspections of engagements of distressed

CR
issuers. 6% of distressed issuers receive a going concern opinion. These proportions are smaller

than for the PCAOB inspection sample, likely because the internal inspections sample is based

US
on larger auditors that have fewer clients at risk of bankruptcy. 204 engagements (9% of the

sample) are jointly inspected by the PCAOB and internal inspectors. The proportion of Part I
AN
Findings for this subsample, at 30%, is close to the average PCAOB inspections sample of 28%.

Panel B also reports the means of the subsamples partitioned on whether the internal
M

inspectors identify deficiencies (satisfactory with comments and unsatisfactory outcomes are
ED

grouped together). Some of the results are similar to those in Panel A. For engagements that do

not receive satisfactory assessments, the probability of restatements is higher, at 17% compared
PT

with 10%, audit fees and hours are lower, and the auditor is less likely to be a Big 4 audit firm.

However, accruals are not significantly different across the two samples. In terms of control
CE

variables, engagements with internal inspection deficiencies are generally not significantly
AC

different from engagements without deficiencies, except for size, which is slightly smaller

(marginally significant).

Correlations among the main variables of interest are presented in Table 4. The results in

Panel A, on PCAOB inspections, generally confirm those of Table 3. PCAOB Part I Findings are

positively correlated with unsigned accruals and the probabilities to restate, report small profits,

28
 ACCEPTED MANUSCRIPT

and receive a going concern opinion, with new clients, and negatively correlated with Big 4

auditors, auditor industry specialization, and audit fees. The correlations are modest, between 3%

and 13% in absolute value.

In terms of correlations among other proxies for audit quality, the correlations are somewhat

low. Exceptions are audit fees, which are correlated at 90% with audit hours. This confirms the

T
IP
applicability of using audit fees as a proxy for audit hours (e.g., Lobo and Zhao, 2013). Audit

fees are also negatively correlated with accruals, positively correlated with Big 4 and industry

CR
specialist auditors, larger offices, and negatively correlated with first-year clients. The positive

US
correlations between audit fees, and Big 4 or industry specialist auditors are consistent with

larger clients choosing Big 4 and industry specialist auditors (Lawrence et al., 2011; Minutti-
AN
Meza, 2013). Restatements are also generally positively correlated with unsigned accruals, but

the correlations are higher when using Part I Finding. Further, accruals measures are
M

mechanically correlated with each other. In particular, the correlation between unsigned

discretionary accruals and total accruals is high, at 0.66, consistent with Dechow et al. (2003).
ED

Going concern opinions are also positively correlated with unsigned discretionary accruals and
PT

total accruals, with correlations above 0.3. This result is consistent with Butler et al. (2004) who

find a positive association between going concern opinions and abnormal accruals.
CE

[Insert Table 4 About Here]

AC

Panel B presents comparable results when focusing on internal inspections. While

generally weaker, internal inspection ratings are still positively correlated with restatements and

negatively correlated with audit fees and audit hours, but they are not correlated with accruals.

4.3 Empirical Results

4.3.1 Accruals measures of audit quality

29
 ACCEPTED MANUSCRIPT

Table 5 presents the results of Models (3) and (4) with accruals as explanatory variables.

Panel A presents the results for PCAOB inspections. Unsigned discretionary accruals [Abs(Disc.

Accruals)], unsigned scaled accruals [Abs(Accruals)], and accruals deflated by cash flow from

operations [Abs(Accruals/CFO)] load positively in the specifications. Neither Dechow and

Dichev residuals (DD Residual) nor signed accruals (Disc. Accruals) are associated with Part I

T
Findings.

IP
[Insert Table 5 About Here]

CR
Some of the coefficients on the control variables suggest that an issuer’s circumstances can

US
predict PCAOB inspection deficiencies. Leverage and Std(CFO) load positively, suggesting that

riskier clients are more difficult to audit. Further, Big 4 loads negatively, consistent with
AN
arguments in prior literature that larger auditors provide higher quality audits (DeAngelo, 1981).

December Year End does not load, suggesting that busy season audits are not of significantly
M

lower quality due to time constraints.

ED

Panel B presents the results for internal inspections. There is no association between accruals

measures of audit quality and internal inspection outcomes, similar to the univariate statistics.
PT

Several of the coefficients on the control variables also suggest that an issuer’s circumstances

can predict internal inspection deficiencies. For example, similar to PCAOB inspections,
CE

Leverage is positively associated with internal ratings. December Year End is also negatively
AC

associated with internal ratings, despite the auditor’s time constraints during the busy season.

4.3.2 Non-accruals output measures of audit quality

Table 6 considers non-accruals output measures of audit quality. Panel A presents the results

for PCAOB inspections. I find positive associations between Part I Finding and both Small

Profit and Restatement, and no association with Prior ROA Meet. The previously identified

30
 ACCEPTED MANUSCRIPT

positive correlation between Going Concern and Part I Finding becomes insignificant in these

analyses, consistent with conflicting predictions existing for going concern opinions. This

suggests that researchers should ensure that tests using going concern opinions are clear tests of

auditor independence, not of general audit quality or auditor effort.

[Insert Table 6 About Here]

T
IP
Panel B reports the results when using the internal inspection ratings. Consistent with the

CR
results in Panel A, Restatement and Small Profit load positively, and Going Concern does not

load. A positive association also exists between Internal Rating and Prior ROA Meet (marginally

significant).

4.3.3 Input measures of audit quality

US
AN
Table 7 presents the results when using the input measures. Panel A focuses on PCAOB

inspections. I do not find, in contrast with the univariate tests, any association between Industry
M

Specialization and Part I Finding, a result consistent with Minutti-Meza (2013).35 I find a
ED

negative association between Audit Fees and Part I Finding, consistent with higher auditor effort

being associated with fewer audit deficiencies, and the arguments in Ball et al. (2012) and Lobo
PT

and Zhao (2013). The association between Audit Hours and Part I Finding, while negative, is

insignificant at conventional levels. This result is puzzling given that the correlation between
CE

Audit Fees and Audit Hours is approximately 90%, but is perhaps driven by the sample reduction

of almost 80%. New Client is positively associated with Part I Finding, consistent with prior
AC

35
In untabulated analyses I replace Industry Specialization with an indicator variable for whether the auditor has the
largest market share in the industry. The results remain unchanged. I also add another indicator variable for whether
the auditor has the largest market share in the industry in the area covered by a specific office, similar to Reichelt
and Wang (2010). I still find insignificant associations.

31
 ACCEPTED MANUSCRIPT

literature on auditor tenure (e.g., Bell et al., 2015). There is no association between Office Size

and Part I Finding.36

[Insert Table 7 About Here]

Panel B focuses on internal inspections. I find, consistent with the results in Panel A, a

T
negative association between Internal Rating and Audit Fees. Further, I find a negative

IP
association with Audit Hours. I do not find any association with New Client, but find a negative

CR
association with Office Size, consistent with Francis and Yu (2009).

4.4 Combined Regressions

US
I include in the regressions all the explanatory variables that worked consistently above, to

determine whether these variables are independently associated with PCAOB or internal
AN
inspection deficiencies. Results are presented in Table 8.

[Insert Table 8 About Here]

M

Panel A reports the results for PCAOB inspections. Each measure of audit quality still loads
ED

when controlling for the other ones.37 This suggests that each measure represents an independent

assessment of audit quality. Because I control for restatements in these regressions, which likely
PT

incorporate many departures from GAAP, the results on accruals and small profits suggest that
CE

these proxies measure within-GAAP manipulations, consistent with DeFond and Zhang (2014). I

find, in untabulated analyses based on Column (7), that a one interquartile range increase (75th
AC

percentile value less 25th percentile value) in Abs(Disc. Accruals) raises the probability of Part I

Finding by 0.6%, and that a restatement and small profit increases this probability by 10.1% and

36
The sample size is restricted to the Big 4 auditors for the office size analysis, consistent with prior literature
(Francis and Yu, 2009). Untabulated analyses indicate a negative association between Office Size and Part I Finding
when removing this restriction. However, the association appears to be driven by auditor size, not by office size.
37
I do not run the regressions including both unsigned discretionary accruals and unsigned total accruals, because
these two measures are substitutes for each other by construction.

32
 ACCEPTED MANUSCRIPT

5.0%, respectively. I also find that a one interquartile range increase in Abs(Accruals/CFO)

raises this probability by 0.4%, and an interquartile range increase in Audit Fees lowers this

probability by -10.7%. These numbers need to be compared with the 27.5% average probability

of a Part I Finding. They suggest that the publicly available measures of audit quality that proxy

the most for practitioners’ views are restatements, small profits, and audit fees. Accruals-based

T
measures of audit quality do not reach the same predictive power.

IP
I use the receiver operating characteristic (ROC) curves to estimate the predictive power of

CR
the regressions. The ROC curve is a parametric plot of the probability of detection versus the

US
false positive rate (e.g., Schmidt, 2012). I compute the area under the curve (AUC), a measure of

fit of the model. An AUC of 0.5 corresponds to a random model, and a value of 1.0 means
AN
perfect predictive power. Column (1) of Table 8 reports that the regression’s AUC, without the

test variables, is 0.688. Industry and issuer-specific circumstances thus explain up to 38% of the
M

issuance of a Part I Finding.38 Inclusion of all measures of audit quality in Column (8) increases

the model AUC only to 0.702, a change of 0.014 (significant at 1% in an untabulated chi-square
ED

test). Therefore, all measures of audit quality together have an incremental explanatory power of

only 3% on the probability of a Part I Finding. 39 One caveat is that PCAOB inspections delve
PT

into specific areas. Thus, the lack of power may be driven by noise introduced in the inspection
CE

outcome as a measure of practitioners’ views of the audit.

Panel B reports on internal inspections. Restatement, Small Profit, Audit Fees and Audit
AC

Hours are still associated with Internal Rating when included with each other. These findings are

38
This number is computed as (0.688 - 0.5)/(1 - 0.5)
39
This number is computed as (0.702 - 0.688)/(1 - 0.5). If I exclude all control variables from the specification but
the measures of audit quality, the ROC of this specification equals 0.596. Consequently, without inclusion of the
control variables an argument can be made that, collectively, publicly available measures of audit quality explain up
to 19% [computed as (0.596 - 0.5/(1 - 0.5)] of the Part I Findings.

33
 ACCEPTED MANUSCRIPT

consistent with the results in Panel A. Further, similar to Panel A, inclusion of the measures of

audit quality modestly increases the adjusted R-squared of the model from 7.7% to 9.1%.

5. Additional tests
Because the PCAOB inspection data are more extensive than internal inspection data,

additional tests focus primarily on PCAOB inspections.

T
IP
5.1 PCAOB sample partition between large and small audit firms
In light of prior research that often concentrates on the largest audit firms or the Big 4 (e.g.,

CR
Francis and Yu, 2009), a natural question is whether the relationships identified in Section 4

differ depending on whether the sample is restricted to the largest audit firms. I replicate the

US
prior tables, restricting the sample to the Big 4 auditors and the eight largest audit firms (the Big
AN
4 auditors plus BDO, Grant Thornton, McGladrey, and Crowe Horwath). In untabulated analyses,

I generally find qualitatively unchanged results, except for Abs(Accruals/CFO), and

M

Abs(Accruals), which become insignificant when restricting the sample to the eight largest audit

firms. New Client is also insignificant. I also run the analyses for the samples of non-Big 4 and
ED

non-eight largest audit firms. The untabulated results are generally consistent with those in

Tables 5 to 7. However, Abs(Disc. Accruals) and Small Profit tend to load below conventional
PT

significance levels in several regressions. This might be driven by the sample reduction.
CE

5.2 Analyses of the nature of the Part I Finding

I consider Part I Findings in the auditing of complex accounting estimates, a recurring focus
AC

of the PCAOB (PCAOB, 2016). Because areas of complex accounting estimates have more

accounting discretion, more room for earnings management is possible (Watts and Zimmerman,

1986; Dechow et al., 2010). Thus, issues in the auditing of complex accounting estimates could

lead the auditor not to detect earnings management (Healy and Wahlen, 1999). I expect positive

associations between such issues and accruals and earnings threshold measures. I define

34
 ACCEPTED MANUSCRIPT

Complex Estimates Part I as an indicator variable for a Part I Finding with any of the following

PCAOB standards not met: AU 342 (auditing accounting estimates), AU 328 (auditing fair value

measurements and disclosures), AU 332 (auditing derivative instruments, hedging activities, and

investments in securities), or AU 336 (using the work of a specialist).40 Conversely, I define Non

Complex Estimates Part I as an indicator variable for whether the Part I Finding is unrelated to

T
the issues raised in those standards. Approximately 50% of the Part I Findings identified by the

IP
PCAOB in the sample correspond to issues in the auditing of complex accounting estimates.

CR
Many of these Part I Findings confirm the observation in DeFond et al. (2018) that “PCAOB

inspection reports provide many examples of how auditors are able (and indeed expected) to

US
adopt a fair presentation mindset rather than a compliance mindset.”
AN
In untabulated analyses, consistent with the predictions, I find positive associations between

the measures of accruals [Abs(Disc. Accruals), Abs(Accruals), and Abs(Accruals/CFO)] and the
M

propensity to meet the zero earnings threshold (Small Profit) with Complex Estimates Part I.

Importantly, the associations are not present when using Non Complex Estimates Part I as the
ED

dependent variable. These results are consistent with the nature of the audit deficiency being
PT

directly linked to the type of measure of audit quality considered.

5.3. Analyses of restatements

CE

5.3.1 Restatements Severity, Nature, and Part I Findings

I conduct supplemental analyses that focus on the severity of a restatement. Following
AC

Choudhary et al. (2016), I consider material and immaterial restatements. Material restatements,

which are more severe, correspond to errors that are judged to be material in the financial reports

40
According to AU 336, specialists possess special skills in fields other than accounting or auditing. Specialists
include appraisers, actuaries, environmental consultants and geologists, and are used by issuers to generate specific
estimates in financial statements, such as pension assumptions. Because such estimates require a large degree of
judgment, they are reasonably easy to manipulate. See Appendix B1 for an example of a Part I Finding.

35
 ACCEPTED MANUSCRIPT

and require a statement of non-reliance, disclosed on a separate 8-K filing under Item 4.02.

Immaterial restatements correspond to revisions that are immaterial for each individual reporting

period in which the error occurred, but collectively the errors can be material.

Columns (1) and (2) of Table 9 present analyses using material and immaterial restatements

as the explanatory variables (Material Restatement and Immaterial Restatement). Both are

T
IP
positively associated with the propensity to issue a Part I Finding (both load if included in the

same regression). The coefficient on Material Restatement is larger than the one on Immaterial

CR
Restatement (significant at 10%). Untabulated analyses also confirm that both Material

US
Restatement and Immaterial Restatement are associated with internal inspection deficiencies.

[Insert Table 9 About Here]

AN
The next analysis aims to increase confidence that the accounts impacted by an audit deemed

problematic by the PCAOB are directly related to the probability of a restatement in the same
M

area. I consider restatements and Part I Findings related to accruals and revenue, because the
ED

prevalence of restatements in these areas is high, and these areas are often targeted by the

PCAOB (PCAOB, 2016).41 I define Revenue Accruals Restatement (Revenue Accruals Part I
PT

Finding) as an indicator variable equal to one when a restatement (Part I Finding) is related to a

revenue or accrual account. Untabulated analyses indicate that restatements (Part I Findings) in
CE

revenue or accruals represent 42% (60%) of all restatements (Part I Findings). I also define Non
AC

Revenue Accruals Restatement (Non Revenue Accruals Part I Finding) equal to one for the

restatements (Part I Findings) related to other accounts. I then run Model (3) with Part I Findings

in accruals or revenues and other categories as the dependent variables, and the different types of

restatements as the explanatory variables. Columns (3) and (4) of Table 9 present the results.

41
See Appendix B.2 and B.3 for Part I Findings related to revenues and accounts receivable.

36
 ACCEPTED MANUSCRIPT

Only restatements in revenues and accruals are associated with Part I Findings in such areas. The

results in Column (4) confirm that only restatements related to other areas are positively

associated with other types of Part I Findings.42

5.3.2 Is the association between Restatements and Part I Findings Mechanical?

I conduct three analyses to mitigate concerns that the association between restatements and

T
Part I Findings could be mechanical, even though this possibility cannot be completely ruled out.

IP
First, I find that PCAOB inspectors are unlikely to suffer from a hindsight bias whereby they

CR
would issue Part I Findings to engagements for which they are aware of imminent restatements. I

US
obtain from the PCAOB detailed inspection fieldwork and comment form information timing for

a subsample of 55 material restatements corresponding to the 2010 to 2014 inspected

AN
engagements of the six largest audit firms. 43 Only in one instance the PCAOB inspectors became

aware during the fieldwork of a restatement in the area of the audit they inspected. In 24 cases,
M

the inspectors became aware, between the end of the fieldwork and the issuance of an inspection

report, of restatements relating to areas of the audits they reviewed. The PCAOB did not change
ED

its opinion based on the knowledge of a restatement.

PT

Second, a PCAOB inspection may directly cause the discovery of a misstatement. PCAOB

inspectors from time to time identify departures from GAAP that are noted in the Part I Finding.
CE

44
Issuers must restate if the SEC, auditors, and issuers determine the departure to be material. I

test a worst-case scenario in which I assume that all restatements that were connected with a
AC

42
Untabulated analyses also confirm that the coefficients on the two types of restatements are significantly different
from each other at 10% or better in both columns.
43
This sample is the same used in Aobdia, Choudhary and Sadka (2017). I am unable to collect additional
information because the collection effort is intensive and requires extensive PCAOB inspection personnel time.
While the sample is not random, the PCAOB was not allowed an opportunity to limit or influence the sample. Thus
there is no reason to believe that this sample would not be representative.
44
The PCAOB’s practice is to report this information to the SEC, which has jurisdiction to determine proper
accounting in issuer’s financial statements [see, for example, the 2013 inspection report for Deloitte (PCAOB, 2014,
p20, footnote 3)]. The PCAOB does not have jurisdiction to ask an issuer to restate its financial statements.

37
 ACCEPTED MANUSCRIPT

departure from GAAP Part I Finding would not have been detected had the PCAOB not

inspected the engagement. I split Part I Finding between Departure GAAP Part I Finding and

Non Departure GAAP Part I Finding. I infer a departure from GAAP based on the text of the

inspection reports. Approximately 7% of the Part I Findings are departures from GAAP

(untabulated). Column (5) of Table 9 confirms that Restatement is strongly associated with

T
Departure GAAP Part I Finding. Restatement remains positively associated with Non Departure

IP
GAAP Part I Finding in Column (6).

CR
Third, I also find no evidence that stock market reactions to the announcement of a

US
restatement are different based on whether an engagement received a Part I Finding, even when

PCAOB inspectors identified a departure from GAAP, which likely requires restatement. This
AN
result suggests that the threshold for restatement is not lower when the PCAOB reviews an

engagement.
M

5.4 Selection bias concerns

Because PCAOB inspections are risk-based, one legitimate concern is whether the analyses
ED

can be extended outside of inspected engagements. I regress the probability that an engagement

is selected for inspection by the PCAOB on the same variables as in Model (3) (untabulated).
PT

The area under the curve is equal to 0.65, indicating that a modest 30% of the probability of
CE

inspection is explained by the model. This suggests that selection on observable characteristics is

not too severe. However, selection on unobservable characteristics could influence both the
AC

probability of inspection and the inspection outcome. Because I do not have a proper exclusion

variable for a PCAOB inspection, I instead follow the sensitivity analysis approach of Altonji et

al. (2005). The basic idea is to use a bivariate probit model and, instead of identifying a specific

exclusion variable, assume a particular correlation between the unobserved factors that could

38
 ACCEPTED MANUSCRIPT

determine a PCAOB inspection and the inspection outcome. A high correlation would indicate

that the PCAOB has access to high-quality nonpublic information indicating that the audit was

poorly conducted, such as whistleblowers’ tips. In untabulated analyses, I replicate Columns (7)

and (8) of Panel A of Table 8 by assuming correlations for the error components in the PCAOB

inspection and outcome equations between 0 and 0.5, similar to Altonji et al. (2005). I find

T
robust results in all regressions.

IP
6. Conclusion

CR
This paper assesses the concordance between observable measures of audit quality used by

academic researchers, and two practitioners’ views of audit quality measured using PCAOB and

US
internal inspections. In the empirical analyses, I use fifteen proxies, five based on accruals, five
AN
based on additional output measures, and five based on input measures of audit quality.

I find that several of the observable measures are associated with practitioners’ views,
M

particularly restatements, whether the issuer meets or beats the zero earnings threshold, and audit

fees. I also find some evidence that unsigned accruals and audit hours are associated with
ED

practitioners’ views. Importantly, these measures predict practitioners’ assessments

independently from each other. However, collectively, observable measures of audit quality only
PT

explain a limited portion of practitioners’ views of audit quality. Overall, this confirms the need
CE

for audit studies to use more than one proxy of audit quality to limit type I errors. This study

demonstrates that practitioners and academics share common ground in identifying low-quality
AC

audits, even though they use widely different approaches to define and assess audit quality. My

findings can be useful to future audit-quality researchers in selecting not only theoretically

suitable, but also empirically powerful proxies, and may motivate additional research to enhance

the set of available proxies.

39
 ACCEPTED MANUSCRIPT

One caveat of this study is that, because PCAOB and internal inspections are not randomly

chosen, it is not certain that my results can be generalized outside of inspected engagements.

Future research may have the opportunity to revisit these analyses if the PCAOB gathers a

sizeable sample of randomly inspected audits, as currently under consideration (Doty, 2015).

Future research also may have the opportunity to further exploit the PCAOB data. Several

T
IP
avenues include understanding better the impact of PCAOB and internal inspections on audit

effort, quality, and fees; how PCAOB and internal inspectors’ competences and incentives

CR
influence the determination of practitioner deficiencies; the role of audit inputs on audit quality;

US
and, importantly, using PCAOB and internal inspection assessments as alternative measures of

audit quality to revisit prior literature based on observable measures of audit quality, or test
AN
entirely new economic predictions. Future research may further examine the relative costs and

benefits that practitioners and academics face in defining and measuring audit quality. In
M

particular, future behavioral and archival studies could determine whether observable measures

of audit quality represent weak proxies for (unobservable) audit processes that practitioners care
ED

about, or disagreements exist between practitioners and academics about what constitutes audit
PT

quality.
CE
AC

40
 ACCEPTED MANUSCRIPT

References
Altonji, J.G., T.E. Elder, and C.R. Taber. 2005. Selection on observed and unobserved variables:
Assessing the effectiveness of Catholic schools. Journal of Political Economy 113(1): 151-184.
Aobdia, D., C.-J. Lin, and R. Petacchi. 2015. Capital market consequences of audit partner
quality. The Accounting Review 90 (6): 2143-2176.
Aobdia, D., P. Choudhary, and G. Sadka. 2017. Do auditors correctly identify and assess internal
control deficiencies? Evidence from the PCAOB data. PCAOB working paper.
Aobdia, D., and N. Shroff. 2017. Regulatory oversight and auditor market share. Journal of
Accounting and Economics 63 (2-3): 262-287.
Aobdia, D. 2018. The economic consequences of audit firms’ quality control system deficiencies.

T
PCAOB working paper.

IP
Ball, R., S. Jayaraman, and L. Shivakumar. 2012. Audited financial reporting and voluntary
disclosure as complements: A test of the confirmation hypothesis. Journal of Accounting and
Economics 53 (1-2): 136-166.

CR
Bell, T.B., M. Causholli, and W.R. Knechel. 2015. Audit firm tenure, non-audit services, and
internal assessments of audit quality. Journal of Accounting Research 53 (3): 461-509.
Butler, M., A.J. Leone, and M. Willenborg. 2004. An empirical analysis of auditor reporting and

US
its association with abnormal accruals. Journal of Accounting and Economics 37 (2): 139-165.
CAQ (Center for Audit Quality). 2012. Guide to PCAOB Inspections.
Carcello, J.V., C. Hollingsworth, and S.A. Mastrolia. 2011. The effect of PCAOB inspections on
AN
Big 4 audit quality. Research in Accounting Regulation 23: 85-96.
Carcello, J.V., and Z.-V. Palmrose. 1994 Auditor litigation and modified reporting on bankrupt
clients. Journal of Accounting Research 32 (Supplement): 1-30.
Choudhary, P., K. Merkley, and K. Schipper. 2016. Qualitative characteristics of financial
M

reporting errors deemed immaterial by managers. University of Arizona working paper.

Daugherty B., and W. Tervo. 2010. PCAOB inspections of smaller CPA firms: The perspective
of inspected firms. Accounting Horizons 24 (2): 189-219.
ED

DeAngelo, L. 1981. Auditor size and audit quality. Journal of Accounting and Economics 3 (3):
183-199.
Dechow, P.M., R.G. Sloan, and A.P. Sweeney. 1995. Detecting earnings management. The
PT

Accounting Review 70 (2): 193-225.

Dechow, P.M., and I. Dichev. 2002. The quality of accruals and earnings: the role of accrual
estimation errors. The Accounting Review 77: 35-59.
CE

Dechow, P.M., S. Richardson, and I. Tuna. 2003. Why are earnings kinky? An examination of
the earnings management explanation. Review of Accounting Studies 8, 355-384.
Dechow, P.M., W. Ge, and C. Schrand. 2010. Understanding earnings quality: A review of the
AC

proxies, their determinants and their consequences. Journal of Accounting and Economics 50
(2-3): 344-401.
DeFond, M.L., K. Raghunandan, and K.R. Subramanyam. 2002. Do non-audit service fees
impair auditor independence? Evidence from going concern audit opinions. Journal of
Accounting Research 40 (4): 1247-1274.
DeFond, M.L., J. Zhang. 2014. A review of archival auditing research. Journal of Accounting
and Economics 58 (2-3): 275-326.
DeFond, M.L., and C.S. Lennox. 2017. Do PCAOB inspections improve the quality of internal
control audits? Journal of Accounting Research 55 (3): 591-627.

41
 ACCEPTED MANUSCRIPT

DeFond, M.L., C.S. Lennox, and J. Zhang. 2018. The primacy of fair presentation: Evidence
from PCAOB standards, federal legislation, and the courts. Accounting Horizons, Forthcoming.
Deloitte. 2010. Advancing quality through transparency Deloitte LLP inaugural report. Deloitte,
London, U.K.
Doty, J. 2015. Protecting the investing public’s interest in informative, accurate, and independent
audit reports. Keynote speech delivered at the AICPA Conference on Current SEC and
PCAOB developments, December 9.
Dye, R.A. 1993. Auditing standards, legal liability, and auditor wealth. Journal of Political
Economy 101 (5): 887-914.
Ferguson, L.H. 2012. Investor protection through audit oversight. Speech delivered at the

T
California State University 11th Annual SEC Financial Reporting Conference, September 21.
https://pcaobus.org/News/Speech/Pages/09212012_FergusonCalState.aspx (accessed February

IP
26, 2018).
Francis, J.R. 2011. A framework for understanding and researching audit quality. Auditing: A

CR
Journal of Practice & Theory 30 (2): 125-152.
Francis, J., R. Lafond, P. Olsson, and K. Schipper. 2005. The market pricing of accruals quality.
Journal of Accounting and Economics 39: 295-327.

US
Francis, J.R., K. Reichelt, and D. Wang. 2005. The pricing of national and city-specific
reputations for industry expertise in the U.S. audit market. The Accounting Review 80: 113–
136.
Francis, J.R., and M.D. Yu. 2009. Big 4 office size and audit quality. The Accounting Review 84
AN
(5): 1521-1552.
Gaynor, L.M., A.S. Kelton, M. Mercer, and T.L. Yohn. 2016. Understanding the relation
between financial reporting quality and audit quality. Auditing: A Journal of Practice &
M

Theory 35 (4): 1-22.

Gul, F.A., D. Wu, and Z. Yang. 2013. Do individual auditors affect audit quality? Evidence from
archival data. The Accounting Review 88 (6): 1993-2023.
ED

Gunny, K.A., and T.C. Zhang. 2013. PCAOB inspection reports and audit quality. Journal of
Accounting and Public Policy 32: 136-160.
Hanson, J.D. 2012. Reflections on the state of the audit profession. Speech delivered at the
American Accounting Association, Auditing Section, Midyear Meeting, January 13.
PT

http://pcaobus.org/News/Speech/Pages/01132012_HansonAAA.aspx (accessed on February 26,

2018).
Healy, P.M., and J.M. Wahlen. 1999. A review of the earnings management literature and its
CE

implications for standard setting. Accounting Horizons 13 (4): 365-383.

Houston, R.W., and C.M. Stefaniak. 2013. Audit partner perceptions of post-audit review
mechanisms: An examination of internal quality reviews and PCAOB inspections. Accounting
AC

Horizons 27 (1): 23-49.

Johnson, E., I.K. Khurana, and J.K. Reynolds. 2002. Audit-firm tenure and the quality of
financial reports. Contemporary Accounting Research 19 (4): 637-660.
Johnson, L.M., M.B. Keune, and J. Winchel. 2014. Auditor perceptions of the PCAOB oversight
process. University of Tennessee working Paper.
Jones, J. 1991. Earnings management during import relief investigations. Journal of Accounting
Research 29: 193-228.

42
 ACCEPTED MANUSCRIPT

Kaplan, S.E., and D.D. Williams. 2013. Do going concern audit reports protect auditors from
litigation? A simultaneous equations approach. The Accounting Review 88 (1): 199-232.
Knechel, W.R., G.V. Krishnan, M. Pevzner, L.B. Shefchik, and U.K. Velury. 2013. Audit quality:
Insights from the academic literature. Auditing: A Journal of Practice & Theory 32 (1): 385-
421.
Kothari, S.P., A.J. Leone, and C.E. Wasley. 2005. Performance matched discretionary accruals
measures. Journal of Accounting and Economics 39: 163-197.
Lawrence, A., M. Minutti-Meza, and P. Zhang. 2011. Can Big 4 versus non-Big 4 differences in
audit-quality proxies be attributed to client characteristics? The Accounting Review 86 (1): 259-
286.

T
Lennox, C., and J. Pittman. 2010. Auditing the auditors: Evidence on the recent reforms to the
external monitoring of audit firms. Journal of Accounting and Economics 49: 84-103.

IP
Leuz, C., Nanda, D., and P.D. Wysocki. 2003. Earnings management and investor protection: an
international comparison. Journal of Financial Economics 69 (3): 505-527.

CR
Levinson, C. 2015. Accounting industry and SEC hobble America’s audit watchdog. Reuters,
Dec 16.
Lim, C.-Y., and H.-T. Tan. 2008. Non-audit service fees and audit quality: The impact of auditor

US
specialization. Journal of Accounting Research 46 (1): 199-246.
Lobo, G.J., and Y. Zhao. 2013. Relation between audit effort and financial report misstatements:
Evidence from quarterly and annual restatements. The Accounting Review 88 (4): 1385-1412.
AN
Mahoney, P.G. 2001. The political economy of the Securities Act of 1933. Journal of Legal
Studies 30: 1-31.
Matsumoto, D.A. 2002. Management’s incentives to avoid negative earnings. The Accounting
Review 77 (3): 483-514.
M

McNichols, M. 2002. Discussion of “The quality of accruals and earnings: The role of accrual
estimation errors.” The Accounting Review 85: 315-341.
Minutti-Meza, M. 2013. Does auditor industry specialization improve audit quality? Journal of
ED

Accounting Research 51 (4): 779-817.

Mutchler, J. 1984. Auditor perceptions of the going concern opinion. Auditing: A Journal of
Practice & Theory 3 (1): 17-30.
PT

Palmrose, Z.-V. 2006. Maintaining the value and viability of independent auditors as gatekeepers
under SOX: An auditing master proposal. University of Sourthern California working paper.
PCAOB. 2007. PCAOB 2006 Annual Report. Public Company Accounting Oversight Board,
CE

Washington D.C.
PCAOB. 2011. Strategic Plan. 2011-2015. Public Company Accounting Oversight Board,
Washington D.C.
PCAOB. 2012. Information for audit committees about the PCAOB inspection process. Release
AC

No. 2012-003, August 1. Public Company Accounting Oversight Board, Washington D.C.
PCAOB. 2014. Report on 2013 Inspection of Deloitte & Touche LLP. Release No. 104-2014-
099, May 6. Public Company Accounting Oversight Board, Washington D.C.
PCAOB. 2016. Staff inspection brief. Information about 2016 inspections. Vol. 2016/3, July.
Public Company Accounting Oversight Board, Washington D.C.
Rapoport, M. 2018. What the KPMG conspiracy case revealed about its audits. Wall Street
Journal, Jan 24. Downloaded on February 22, 2018 from https://www.wsj.com/articles/what-
the-kpmg-conspiracy-case-revealed-about-its-audits-1516810866.

43
 ACCEPTED MANUSCRIPT

Reichelt, K.J., and D. Wang. 2010. National and office-specific measures of audit industry
expertise and effects on audit quality. Journal of Accounting Research 48 (3): 647-686.
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.
Riley, R.R., Jenkins, J.G., Roush, P.Y., and J.C. Thibodeau. 2008. Audit quality in the post-SOX
audit environment: What your accounting students must know about the PCAOB inspection
process. Currrent Issues in Auditing 2 (2): A17-A25.
Schmidt, J.J. 2012. Perceived auditor independence and audit litigation: The role of nonaudit
service fees. The Accounting Review 87 (3): 1033-1065.
Shroff, N. 2017. Does auditor regulatory oversight affect corporate financing and investment

T
decisions? M.I.T. working paper.

IP
Simunic, D.A. 1980. The pricing of audit services: Theory and evidence. Journal of Accounting
Research 18 (1): 161-190.
Stigler, G. 1971. The theory of economic regulation. Bell Journal of Economics and

CR
Management Science 2: 3-21.
Velikonja, U. 2016. Reporting agency performance: Behind the SEC’s enforcement statistics.
Cornell Law Review 101 (4): 901-980.

US
Watts, R., and J. Zimmerman. 1986. Positive theory of accounting. Englewood Cliffs, NY:
Prentice-Hall.
AN
M
ED
PT
CE
AC

44
 ACCEPTED MANUSCRIPT

Appendix A: Variables Definitions

Variable Definition
Dependent Variables:
Part I Finding Indicator variable equal to one if the inspection resulted in a Part I Finding.
Departure GAAP Indicator variable equal to one when the PCAOB identifies a departure from Generally
Part I Finding Accepted Accounting Principles (GAAP) in its Part I Finding.
Non Departure Indicator variable equal to one when the PCAOB identifies a Part I Finding without a
GAAP Part I departure from GAAP.
Finding
Revenue Accruals Indicator variable equal to one when the Part I Finding is related to revenue and accruals

T
Part I Finding financial statement areas.
Non Revenue Indicator variable equal to one when the Part I Finding is not related to revenue and

IP
Accruals Part I accruals financial statement areas.
Finding
Internal Rating Results of the internal inspection, equal to 1 when the inspection is satisfactory, 2 when it

CR
is satisfactory with comments, and 3 when it is unsatisfactory.
Test Variables:
Disc. Accruals Residual of a regression of accruals (deflated by beginning assets) on gross property, plant
and equipment (PP&E, deflated by beginning assets), the year-on-year change in revenues

US
(deflated by beginning assets), one over beginning assets, and prior year return on asset
(ROA, defined as income before extraordinary items deflated by average assets). Accruals
are defined as income before extraordinary items (Compustat IB), less cash flow from
operations (Compustat OANCF) excluding extraordinary items and discontinued
AN
operations (Compustat XIDOC).
Abs(Disc. Absolute value of Disc. Accruals.
Accruals)
DD Residual Absolute value of the residual of a regression of total current accruals deflated by average
assets on cash flow from operations deflated by average assets for the current year, the
M

following year, and the prior year, gross PP&E deflated by average assets, and change in
revenues deflated by average assets. Total current accruals are defined as change in
inventories plus change in accounts receivable minus change in accounts payable minus
change in income tax payable plus net change in other assets and liabilities. The variables
ED

are taken from the cash flow statement.

Abs(Accruals) Absolute value of accruals deflated by beginning assets.
Abs(Accruals/CFO) Absolute value of accruals deflated by cash flow from operations.
PT

Restatement Indicator variable equal to one if the fiscal year-end financial statements are restated.
Material Indicator variable equal to one if the fiscal year-end financial statements are restated and
Restatement the restatement is disclosed in a form 8-K item 4.02 indicating non reliance on prior period
results.
CE

Immaterial Indicator variable equal to one if the fiscal-year end financial statements are restated and
Restatement the restatement is not disclosed in a form 8-K item 4.02.
Revenue Accruals Indicator variable equal to one if the fiscal-year end financial statements are restated in the
Restatement area of revenue or accruals.
AC

Non Revenue Accruals Indicator variable equal to one if the fiscal-year end financial statements are restated in an
Restatement area other than revenue or accruals.
Small Profit Indicator variable equal to one if ROA is between 0% and 3%.
Prior ROA Meet Indicator variable equal to one if the year-on-year change in ROA is between 0% and 1%.
Going Concern Indicator variable equal to one if the auditor issued a going concern opinion.
Type I Going Concern Indicator variable equal to one if the auditor issued a going concern opinion and the issuer
did not declare bankruptcy the following year.
Big 4 Indicator variable equal to one if the audit firm is a Big 4, and zero otherwise.
Industry Specialization Auditor market share in the client’s industry, based on audit fees. Industries are defined at

45
 ACCEPTED MANUSCRIPT

Variable Definition
the two-digit SIC code.
Audit Fees Logarithm of the audit fees reported by the issuer, from Audit Analytics.
Audit Hours Logarithm of the total audit hours spent on the engagement, obtained from the PCAOB.
Fees Per Hour Audit fees divided by audit hours.
New Client Indicator variable equal to one if the auditor-client relationship is in its first year.
Office Size Logarithm of one plus total audit fees charged by one audit office.
Control Variables:
Integrated Audit Indicator variable equal to one when the audit is an integrated audit of financial statements
and internal controls.

T
Foreign Income Absolute value of pretax income from foreign operations (PIFO) divided by the absolute
value of pretax income (PI).

IP
Size Logarithm of the issuer’s assets.
Geographic Number of geographic segments, from GEOSEG in Compustat SEGMENTS.

CR
Segments
Business Segments Number of business segments, from BUSSEG in Compustat SEGMENTS.
December Year Indicator variable equal to one if the issuer’s fiscal year ends in December.
End
Std(CFO)

CFO
Leverage
US
Standard deviation of the issuer’s cash flow from operations deflated by beginning assets,
computed from year t − 3 to year t.
Issuer’s cash flow from operations deflated by beginning assets.
Total debt (short-term plus long-term) divided by the sum of total debt and equity.
AN
BTM Book-to-market ratio.
Litigation Indicator variable equal to one if the issuer is in a higher litigation industry (SIC code
between 2833 and 2836, 8731 and 8734, 3570 and 3577, 7370 and 7374, 3600 and 3674,
or 5200 and 5961).
M

Sales Growth Year-on-year sales growth of the issuer.

Material Weakness Indicator variable equal to one if a material weakness is reported for the year, per Audit
Analytics.
ED

Altman Z Score Altman Z score.

PT
CE
AC

46
 ACCEPTED MANUSCRIPT

Appendix B: Examples of Part I Findings from Public PCAOB Inspection Reports

B.1 2009 Deloitte (U.S.) Inspection

Issuer H
The issuer engaged a specialist to calculate the estimated amount of a significant contingent
liability. The amount calculated by the specialist exceeded the amount recorded by the issuer by
an amount that was approximately 13 times the Firm's planning materiality. The Firm failed to
perform sufficient procedures to test the contingent liability, as follows:

T
• The Firm failed to assess the reasons for the differences between the issuer's and the specialist's

IP
assumptions.
• The Firm's testing did not address the completeness of new claims filed, which is a key input

CR
used by the issuer in estimating the amount of this contingent liability, as it neither tested the
completeness of the new claims data nor obtained evidence about the operating effectiveness of
controls over the completeness of those data.

B.2 2007 PwC (U.S.) Inspection

US
AN
Issuer B

In this audit, the Firm failed in the following respects to obtain sufficient competent evidential
M

matter to support its audit opinion – …

Certain of the issuer's contracts involve multiple phases. For purposes of revenue recognition, the
ED

issuer allocates revenue to various phases. In certain cases, this practice results in different profit
margins for each phase. The Firm failed to test the reasonableness of the issuer's assertion that
the revenue that was allocated to each phase was representative of the fair value of the delivered
PT

elements of that phase.

CE

B.3 2008 Deloitte (U.S.) Inspection

Issuer A
AC

In this audit, the Firm failed in the following respects to obtain sufficient competent evidential
matter to support its audit opinion –

 The issuer calculated its allowance for doubtful accounts by applying estimated loss
factors to categories of accounts receivable grouped by days past due. The Firm failed to
test the assumptions that management had used to develop the estimated loss factors. …

47
 ACCEPTED MANUSCRIPT

Table 1: Summary of Results

This table summarizes the results of the multivariate analyses presented in Tables 5 to 7. These analyses evaluate the
association between PCAOB inspection deficiencies (Part I Findings) or internal inspection deficiencies as
dependent variables, and each individual measure of audit quality as the explanatory variable, controlling for other
potential determinants of inspection deficiencies. A blank cell indicates that no association is found at conventional
significance levels. + and – correspond to positive and negative associations, respectively. p < 0.10, p < 0.05, and p
< 0.01 correspond to statistical significance at the 10%, 5%, and 1% levels, respectively (two-tailed tests). In
addition to the results in the table, I find a negative association between Big 4 auditors and Part I Findings, and no
association between fees per hour and either Part I Findings or internal inspection deficiencies. See Appendix A for
detailed variable definitions.

Empirical

T
Association Empirical
Predicted with PCAOB Association with

IP
Proxy Name Description of Proxy
Association Deficiencies Internal Inspection
(Part I Deficiencies
Findings)

CR
Five accruals proxies
1) Disc. Accruals Signed discretionary accruals +
2) Abs(Disc. Unsigned discretionary
+ + (p <0.05)
Accruals) accruals
3) DD Residual

4) Abs(Accruals)
Unsigned Dechow and
Dichev model residuals
Unsigned total accruals
5) Abs(Accruals/CFO) Unsigned (total accruals /
US +
+ + (p <0.01)
AN
Cash flow from operations) + + (p <0.01)
Five non-accruals output proxies
1) Restatement Propensity to restate financial
+ + (p <0.01) + (p <0.01)
statements
M

2) Small Profit Propensity to meet or beat

+ + (p <0.01) + (p <0.05)
zero earnings threshold
3) Prior ROA Meet Propensity to meet or beat
+ + (p <0.10)
prior year's earnings
ED

4) Going Concern Propensity to issue a going

-
concern opinion
5) Type I Going Going concern opinions that
Concern do not result in bankruptcy -
PT

Five input proxies

1) Industry Auditor industry
-
Specialization specialization
2) Audit Fees Logarithm of audit fees
CE

- - (p <0.01) - (p <0.01)
3) Audit Hours Logarithm of audit hours - - (p <0.05)
4) New Client First year clients + + (p <0.05)
5) Office Size Total Big 4 office audit fees - - (p <0.10)
AC

48
 ACCEPTED MANUSCRIPT

Table 2: Sample Description

This table presents descriptive information about the two samples of PCAOB and internal inspections. Panel A
shows the number of inspected engagements in each sample. For comparability, non-inspected engagements are
restricted to the observations for which all control variables are available. For the internal inspection sample, the
number of non-inspected engagements is restricted to audit firm-years for which internal inspection data are
available. Panel B presents the inspection outcomes. PCAOB inspection outcomes are binary and deficiencies are
referred to as Part I Findings. Internal inspection deficiencies are classified into three types, from satisfactory (no
deficiency identified), to unsatisfactory (material deficiencies identified). Panel C partitions each sample by year and
between Big 4 and non-Big 4 auditors. Panel D partitions the total internal inspection sample based on whether the
audit is inspected. There are 13,401 non inspected engagements and 2,286 inspected engagements. A t-test compares
the differences in sample means. See Appendix A for variable definitions. Significance levels are * 10%, ** 5% and
*** 1%.

T
IP
Panel A: Inspections by year in the sample

CR
PCAOB Inspections Sample Internal Inspections Sample
Observations Percent Observations Percent
Not Inspected 47,719 90.0% Not Inspected 13,401 85.4%
Inspected 5,309 10.0% Inspected 2,286 14.6%
Total 53,028

Panel B: Inspection outcomes

100.0%
US Total 15,687 100.0%
AN
PCAOB Inspections Sample Internal Inspections Sample
Observations Percent Observations Percent
No Part I
Finding 3,849 72.5% 1: Satisfactory 1,857 81.2%
M

2: Satisfactory with
Part I Finding 1,460 27.5% comments 254 11.1%
Total 5,309 100.0% 3: Unsatisfactory 175 7.7%
ED

Total 2,286 100.0%

Panel C: Inspected samples partitioned between Big 4 and non-Big 4 for each year
PT

PCAOB Inspections Sample Internal Inspections Sample

Year Non-Big 4 Big 4 Total Year Non-Big 4 Big 4 Total
2003 187 322 509 2005 0 94 94
CE

2004 254 281 535 2006 0 102 102

2005 309 329 638 2007 0 76 76
2006 246 307 553 2008 18 172 190
2007 205 442 647 2009 15 212 227
AC

2008 246 180 426 2010 52 394 446

2009 179 136 315 2011 41 340 381
2010 185 229 414 2012 36 389 425
2011 204 231 435 2013 34 311 345
2012 190 247 437 Total 196 2,090 2,286
2013 159 241 400 Percentage 8.6% 91.4% 100.0%
Total 2,364 2,945 5,309
Percentage 44.5% 55.5% 100.0%

49
 ACCEPTED MANUSCRIPT

Panel D: Internal Inspection Sample Descriptive Statistics Partitioned by Whether the Audit is Inspected

Mean Mean Difference in Means

Variable Not inspected Inspected T-statistic
Disc. Accruals -0.02 -0.02 -0.22
Abs(Disc. Accruals) 0.10 0.08 -4.53 ***
DD Residual 0.04 0.04 -2.39 **

T
Abs(Accruals) 0.10 0.09 -3.75 ***
Abs(Accruals/CFO) 1.58 1.41 -1.89 *

IP
Restatement 0.10 0.11 2.02 **
Material Restatement 0.02 0.03 0.40

CR
Immaterial Restatement 0.07 0.09 2.00 **
Small Profit 0.16 0.18 2.09 **
Prior ROA Meet 0.17 0.19 2.91 ***
Going Concern
Type I Going Concern
Industry Specialization
0.08
0.08
0.21
US 0.06
0.05
0.23
-1.63
-1.94
7.27
*
***
AN
Audit Fees 13.97 14.30 13.55 ***
Audit Hours 8.71 9.00 11.66 ***
Fees Per Hour 226.29 231.28 1.42
New Client 0.03 0.06 7.35 ***
M

Office Size 17.51 17.65 4.66 ***

Integrated Audit 0.85 0.88 3.92 ***
Foreign Income 0.24 0.27 3.03 ***
ED

Size 6.67 7.20 11.68 ***

Geographic Segments 2.33 2.55 4.17 ***
Business Segments 2.06 2.18 3.20 ***
PT

December Year End 0.72 0.74 2.41 **

Std(CFO) 0.09 0.07 -2.69 ***
CFO 0.06 0.07 2.72 ***
CE

Leverage 0.34 0.36 2.46 **

BTM 0.51 0.50 -0.31
Litigation 0.32 0.31 -0.67
AC

Big 4 0.84 0.91 9.38 ***

Sales Growth 0.15 0.13 -1.46
Material Weakness 0.03 0.04 1.09
Altman Z Score 0.10 0.95 1.12

50
 ACCEPTED MANUSCRIPT

Table 3: Descriptive Statistics

Panels A and B present overall sample means and standard deviations for the PCAOB and internal inspection
samples, respectively. Each sample is also partitioned between inspections that received deficiencies and those that
did not (for internal inspections, satisfactory with comments and unsatisfactory outcomes are grouped together). A t-
test compares the difference in subsample means (Wilcoxon tests in differences in distributions are similar for most
variables). The number of observations is reduced for DD Residual, due to the requirement to obtain prior and future
year cash flow from operations data; for Going Concern, where the sample is restricted to distressed issuers; for
Audit Hours, where data are available only for the largest auditors between 2008 and 2013; and for Office Size,
which is restricted to the U.S. operations of the Big 4 audit firms. See Appendix A for variable definitions.
Significance levels are * 10%, ** 5% and *** 1%.

T
Panel A: Descriptive statistics for the PCAOB sample

IP
Mean No Mean Difference
Part I Finding Part I Finding in Means

CR
Variable N Mean Std Subsample Subsample T-statistic
Part I Finding 5,309 0.28 0.45
Disc. Accruals 5,309 -0.01 0.25 -0.01 -0.02 -0.94
Abs(Disc. Accruals) 5,309 0.14 0.25 0.13 0.17 5.23***
DD Residual
Abs(Accruals)
Abs(Accruals/CFO)
5,003
5,309
5,309
0.05
0.18
1.78
US
0.08
0.51
4.26
0.05
0.15
1.61
0.06
0.25
2.22
2.68***
6.58***
4.69***
AN
Restatement 5,309 0.12 0.33 0.11 0.16 5.14***
Material Restatement 5,309 0.06 0.24 0.05 0.09 4.50***
Immaterial Restatement 5,309 0.06 0.23 0.05 0.07 2.50**
Small Profit 5,309 0.14 0.35 0.13 0.16 2.41**
Prior ROA Meet 5,309 0.13 0.34 0.13 0.13 -0.67
M

Going Concern 1,933 0.21 0.41 0.19 0.26 3.33***

Type I Going Concern 1,933 0.20 0.40 0.18 0.25 3.29***
Industry Specialization 5,309 0.15 0.15 0.16 0.12 -7.92***
ED

Audit Fees 5,309 13.37 1.58 13.49 13.05 -9.07***

Audit Hours 1,150 8.99 0.95 8.99 8.97 -0.39
Fees Per Hour 1,150 229.80 136.12 228.55 232.32 0.44
New Client 5,309 0.11 0.32 0.10 0.14 3.18***
PT

Office Size 2,395 17.51 1.23 17.53 17.43 1.52

Integrated Audit 5,309 0.62 0.48 0.64 0.57 -5.11***
Foreign Income 5,309 0.20 0.41 0.20 0.18 -1.40
CE

Size 5,309 5.88 2.65 6.03 5.49 -6.62***

Geographic Segments 5,309 2.24 2.26 2.29 2.11 -2.61***
Business Segments 5,309 1.99 1.63 1.99 1.99 -0.03
December Year End 5,309 0.71 0.45 0.71 0.70 -0.57
AC

Std(CFO) 5,309 0.24 1.06 0.19 0.38 6.10***

CFO 5,309 0.00 0.48 0.02 -0.06 -5.00***
Leverage 5,309 0.31 0.59 0.30 0.33 1.26
BTM 5,309 0.45 1.30 0.44 0.48 1.11
Litigation 5,309 0.32 0.46 0.33 0.28 -3.72***
Big 4 5,309 0.55 0.50 0.59 0.45 -9.09***
Sales Growth 5,309 0.24 0.88 0.24 0.26 0.75
Material Weakness 5,309 0.05 0.22 0.05 0.04 -1.25
Altman Z Score 1,933 -7.64 55.11 -6.32 -10.41 -1.66*

51
 ACCEPTED MANUSCRIPT

Panel B: Descriptive statistics for the internal inspection sample

Mean Mean Difference

Satisfactory (1) Not Satisfactory In Means
Variable N Mean Std Subsample (2 or 3) Subsample T-statistic
Internal Rating 2,286 1.26 0.59 1.00 2.41
Part I Finding 204 0.30 0.46 0.25 0.48 0.23***
Disc. Accruals 2,286 -0.02 0.13 -0.02 -0.02 -0.43
Abs(Disc. Accruals) 2,286 0.08 0.11 0.08 0.08 -0.36
DD Residual 2,187 0.04 0.05 0.04 0.03 -0.73
Abs(Accruals) 2,286 0.09 0.12 0.09 0.08 -0.83

T
Abs(Accruals/CFO) 2,286 1.41 3.73 1.40 1.43 0.13

IP
Restatement 2,286 0.11 0.32 0.10 0.17 4.56***
Material Restatement 2,286 0.03 0.16 0.02 0.04 2.60***
Immaterial Restatement 2,286 0.09 0.28 0.08 0.13 3.49***

CR
Small Profit 2,286 0.18 0.39 0.18 0.21 1.38
Prior ROA Meet 2,286 0.19 0.39 0.19 0.21 1.31
Going Concern 594 0.06 0.24 0.06 0.07 0.39
Type I Going Concern
Industry Specialization
Audit Fees
594
2,286
2,286
0.05
0.23
14.30
US
0.23
0.12
1.15
0.05
0.23
14.34
0.06
0.23
14.13
0.40
0.19
-3.37***
AN
Audit Hours 1,870 9.00 1.02 9.04 8.84 -3.35***
Fees Per Hour 1,870 231.28 140.32 230.51 234.42 0.48
New Client 2,286 0.06 0.23 0.05 0.07 1.07
Office Size 2,090 17.65 1.25 17.65 17.63 -0.28
M

Integrated Audit 2,286 0.88 0.33 0.88 0.88 0.10

Foreign Income 2,286 0.27 0.45 0.28 0.25 -0.94
Size 2,286 7.20 2.07 7.23 7.04 -1.70*
ED

Geographic Segments 2,286 2.55 2.35 2.54 2.60 0.49

Business Segments 2,286 2.18 1.79 2.18 2.17 -0.14
December Year End 2,286 0.74 0.44 0.75 0.73 -0.72
Std(CFO) 2,286 0.07 0.18 0.07 0.08 1.06
PT

CFO 2,286 0.07 0.18 0.07 0.07 -0.83

Leverage 2,286 0.36 0.42 0.36 0.39 1.46
BTM 2,286 0.50 1.15 0.50 0.51 0.18
CE

Litigation 2,286 0.31 0.46 0.32 0.31 -0.45

Big 4 2,286 0.91 0.28 0.93 0.86 -4.46***
Sales Growth 2,286 0.13 0.57 0.13 0.14 0.18
AC

Material Weakness 2,286 0.04 0.19 0.04 0.04 0.12

Altman Z Score 594 0.95 10.41 0.72 1.95 1.13

52
 ACCEPTED MANUSCRIPT

T
IP
Table 4: Correlations

This table presents correlations among the main variables of interest. Spearman correlations are above the diagonal and Pearson correlations are below. See Appendix
A for variable definitions. Correlations that are significant at 5% are in bold. Column variables correspond to the rows with the same number [e.g., Column (3) in

CR
Panel A is Abs(Disc. Accruals)]. Panel A shows the correlations for the PCAOB inspection sample; Panel B, for the internal inspection sample. The samples are
restricted to distressed issuers for the correlations with Going Concern, and to the Big 4 for correlations with Office Size. Thus, the correlation between Office Size and
Big 4 cannot be computed. For the sake of space, the correlations with Type I Going Concern are not reported, because they are very similar to the ones with Going
Concern (most going concern opinions do not result in bankruptcy, therefore the correlation between both variables is very high).

Panel A: Correlations for the PCAOB sample

Variable (1) (2) (3) (4) (5) (6)

US (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17)
AN
(1) Part I Finding 0.00 0.05 0.03 0.07 0.09 0.07 0.03 -0.01 0.06 -0.13 -0.11 -0.01 0.02 0.04 -0.04 -0.12
(2) Disc. Accruals -0.01 -0.13 0.04 -0.22 -0.10 -0.01 0.03 0.00 0.01 -0.14 -0.16 -0.09 -0.03 0.05 -0.05 -0.13
(3) Abs(Disc. Accruals) 0.07 -0.13 0.39 0.48 0.21 0.04 -0.18 -0.20 0.30 -0.27 -0.28 -0.14 -0.02 0.06 0.04 -0.23
(4) DD Residual 0.04 0.01 0.49 0.30 0.12 0.05 -0.13 -0.15 0.30 -0.27 -0.30 -0.14 -0.01 0.08 0.00 -0.26
(5) Abs(Accruals) 0.09 -0.48 0.66 0.45 0.62 0.05 -0.16 -0.18 0.35 -0.21 -0.26 -0.19 -0.10 0.04 -0.03 -0.19
M

(6) Abs(Accruals/CFO) 0.06 -0.18 0.20 0.11 0.21 0.05 0.09 -0.10 -0.01 -0.13 -0.13 -0.06 -0.11 0.06 -0.01 -0.13
(7) Restatement 0.07 -0.02 0.06 0.05 0.05 0.03 0.01 -0.02 0.00 0.01 -0.03 0.05 -0.06 -0.02 -0.03 0.01
(8) Small Profit 0.03 0.01 -0.12 -0.11 -0.09 -0.08 0.01 0.12 -0.05 0.10 0.12 0.10 0.00 -0.04 0.03 0.09
(9) Prior ROA Meet -0.01 -0.01 -0.13 -0.11 -0.09 -0.08 -0.02 0.12 -0.09 0.12 0.16 0.15 0.02 -0.05 0.00 0.11
ED

(10) Going Concern 0.06 -0.04 0.34 0.37 0.39 0.06 0.00 -0.05 -0.09 -0.43 -0.43 -0.09 0.00 0.07 0.00 -0.31
(11) Industry Specialization -0.11 -0.05 -0.22 -0.23 -0.18 -0.10 0.02 0.09 0.10 -0.30 0.72 0.42 0.08 -0.21 0.10 0.86
(12) Audit Fees -0.12 -0.09 -0.32 -0.32 -0.28 -0.13 -0.03 0.13 0.17 -0.43 0.62 0.91 0.44 -0.17 0.29 0.69
(13) Audit Hours -0.02 -0.10 -0.13 -0.14 -0.11 -0.04 0.04 0.10 0.15 -0.10 0.40 0.90 0.09 -0.04 0.22 0.50
PT

(14) Fees Per Hour 0.03 0.01 -0.03 -0.01 -0.04 -0.04 -0.02 -0.04 0.03 0.10 0.04 0.29 -0.12 -0.25 0.16 0.02
(15) New Client 0.04 0.04 0.07 0.07 0.04 0.02 -0.02 -0.04 -0.05 0.07 -0.21 -0.17 -0.04 -0.14 -0.01 -0.23
(16) Office Size -0.03 -0.02 0.02 0.02 -0.02 -0.01 -0.03 0.03 0.01 -0.01 0.09 0.32 0.26 0.16 -0.02
(17) Big 4 -0.12 -0.06 -0.24 -0.26 -0.20 -0.11 0.01 0.09 0.11 -0.31 0.84 0.67 0.49 0.02 -0.23
CE
AC

53
 ACCEPTED MANUSCRIPT

T
IP
Panel B: Correlations for the internal inspections sample

Variable (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13) (14) (15) (16) (17)

CR
(1) Internal Rating -0.04 0.00 -0.01 0.00 0.01 0.10 0.03 0.03 0.01 0.01 -0.08 -0.07 -0.03 0.02 0.00 -0.10
(2) Disc. Accruals -0.01 -0.30 0.00 -0.30 -0.11 0.01 0.06 0.03 0.08 0.01 -0.08 -0.06 -0.03 0.02 -0.02 -0.02
(3) Abs(Disc. Accruals) -0.01 0.04 0.34 0.41 0.16 -0.01 -0.17 -0.21 0.09 -0.11 -0.19 -0.18 -0.05 0.03 0.06 -0.09
(4) DD Residual -0.01 0.21 0.41 0.22 0.07 0.01 -0.15 -0.16 0.12 -0.09 -0.17 -0.21 -0.01 0.03 0.04 -0.10
(5) Abs(Accruals) -0.02 -0.26 0.67 0.35 0.62 0.01 -0.10 -0.21 0.14 -0.05 -0.24 -0.26 -0.14 0.02 -0.03 -0.05
(6) Abs(Accruals/CFO)
(7) Restatement
(8) Small Profit
(9) Prior ROA Meet
0.00
0.09
0.03
0.03
-0.05
0.01
0.02
0.00
0.10
-0.02
-0.14
-0.17
0.06
0.02
-0.13
-0.13
0.20
-0.01
-0.10
-0.15
0.01
US
-0.07
-0.09
0.05

0.04
0.00
0.17
0.04

0.12
-0.13
0.00
0.12
-0.06
-0.02
-0.03
-0.02
-0.04
0.06
0.04
0.10
-0.11
0.05
0.09
0.16
-0.10
0.05
0.09
0.14
-0.14
-0.01
0.00
0.05
0.04
-0.03
0.02
-0.04
-0.01
0.02
0.00
0.02
-0.08
0.03
0.02
0.08
AN
(10) Going Concern 0.01 0.09 0.11 -0.02 0.11 -0.01 -0.02 -0.03 -0.02 -0.10 -0.15 -0.14 -0.01 -0.04 -0.05 -0.02
(11) Industry Specialization -0.01 0.02 -0.12 -0.07 -0.09 -0.05 0.08 0.05 0.11 -0.09 0.33 0.31 0.09 -0.10 0.09 0.48
(12) Audit Fees -0.09 -0.08 -0.19 -0.19 -0.20 -0.09 0.05 0.10 0.16 -0.13 0.33 0.93 0.50 -0.17 0.24 0.36
(13) Audit Hours -0.10 -0.05 -0.20 -0.23 -0.24 -0.09 0.05 0.10 0.14 -0.13 0.32 0.92 0.21 -0.11 0.20 0.33
M

(14) Fees Per Hour 0.00 -0.05 -0.02 -0.02 -0.05 -0.06 -0.02 0.00 0.05 -0.01 0.06 0.34 -0.02 -0.24 0.15 0.08
(15) New Client 0.02 0.01 0.06 0.01 0.08 0.04 -0.03 0.02 -0.04 -0.04 -0.10 -0.17 -0.11 -0.14 -0.04 -0.15
(16) Office Size -0.02 0.00 0.03 0.03 -0.02 -0.02 0.01 0.00 0.02 -0.05 0.06 0.28 0.23 0.11 -0.04
(17) Big 4 -0.10 0.01 -0.12 -0.07 -0.10 -0.05 0.03 0.02 0.08 -0.02 0.52 0.36 0.34 0.03 -0.15
ED
PT
CE
AC

54
 ACCEPTED MANUSCRIPT

Table 5: Results of the Inspection Deficiencies Association Models using Accruals

This table presents the results of Models (3) and (4) that evaluate the association between academic and practitioner
measures of audit quality, using accrual measures as the academic measures of audit quality. The dependent variable
in Panel A, Part I Finding, equals one if the PCAOB inspection results in the issuance of a Part I Finding. The
dependent variable in Panel B, Internal Rating, is equal to the rating assigned by the audit firms to their internally
inspected engagements. Each column shows the regression results for a different measure of audit quality, Audit
Quality Measure, shown at the top of each column. For example, Audit Quality Measure is Abs(Disc. Accruals) in
Column (2). Variable definitions are provided in Appendix A. AUC is the area under the curve and represents a
measure of fit of the model. The z- or t-statistic (in brackets) is below the coefficient. Standard-errors are clustered
at the issuer-level. Significance levels are * 10%, ** 5% and *** 1%.
Panel A: PCAOB Inspection Findings

T
(1) (2) (3) (4) (5)
Audit Quality Measure

IP
Dependent Variable: Disc. Abs(Disc. DD Abs Abs
Part I Finding Accruals Accruals) Residual (Accruals) (Accruals/CFO)

CR
Audit Quality Measure -0.164 0.333** 0.037 0.299*** 0.022***
[-1.207] [2.096] [0.073] [3.584] [3.009]
Integrated Audit -0.041 -0.040 -0.022 -0.033 -0.035
[-0.391] [-0.377] [-0.201] [-0.315] [-0.335]
Foreign Income

Size
-0.027
[-0.298]
-0.040*
[-1.747]
US -0.023
[-0.250]
-0.032
[-1.362]
-0.041
[-0.445]
-0.027
[-1.103]
-0.027
[-0.297]
-0.032
[-1.387]
-0.018
[-0.195]
-0.032
[-1.376]
AN
Geographic Segments -0.001 -0.001 0.001 -0.001 -0.003
[-0.075] [-0.047] [0.077] [-0.032] [-0.186]
Business Segments 0.029 0.030 0.026 0.031 0.029
[1.318] [1.343] [1.123] [1.395] [1.295]
M

December Year End -0.043 -0.045 -0.068 -0.052 -0.046

[-0.565] [-0.585] [-0.861] [-0.678] [-0.602]
Std(CFO) 0.111*** 0.114*** 0.114*** 0.100*** 0.118***
[3.228] [3.289] [3.199] [2.894] [3.466]
ED

CFO -0.052 0.016 -0.015 0.128 -0.059

[-0.670] [0.191] [-0.176] [1.415] [-0.763]
Leverage 0.133** 0.133** 0.107 0.146** 0.118*
[2.046] [2.088] [1.517] [2.276] [1.839]
PT

BTM 0.034 0.041 0.062* 0.043 0.039

[1.288] [1.505] [1.939] [1.575] [1.466]
Litigation -0.265** -0.267** -0.204* -0.275** -0.273**
CE

[-2.448] [-2.470] [-1.843] [-2.545] [-2.523]

Big 4 -0.375*** -0.376*** -0.402*** -0.387*** -0.373***
[-3.943] [-3.956] [-4.085] [-4.060] [-3.916]
Sales Growth 0.008 -0.001 0.019 -0.004 0.008
AC

[0.196] [-0.032] [0.462] [-0.092] [0.194]

Material Weakness 0.042 0.049 0.061 0.049 0.021
[0.260] [0.304] [0.370] [0.303] [0.129]

Observations 5,309 5,309 5,003 5,309 5,309

Pseudo R-squared 0.077 0.078 0.078 0.079 0.079
Sample All All All All All
AUC 0.689 0.690 0.690 0.691 0.691
Industry Year Fixed Effects Yes Yes Yes Yes Yes
Clustering Issuer Issuer Issuer Issuer Issuer

55
 ACCEPTED MANUSCRIPT

Panel B: Internal Inspection Ratings

(1) (2) (3) (4) (5)

Audit Quality Measure
Dependent Variable: Disc. Abs(Disc. DD Abs Abs
Internal Rating Accruals Accruals) Residual (Accruals) (AccrualsCFO)
Audit Quality Measure -0.054 -0.021 -0.249 -0.082 -0.001
[-0.593] [-0.176] [-0.867] [-1.069] [-0.158]
Integrated Audit 0.070* 0.069* 0.082* 0.066 0.069*
[1.681] [1.651] [1.918] [1.565] [1.664]

T
Foreign Income -0.046* -0.047* -0.054** -0.048* -0.047*
[-1.782] [-1.809] [-2.104] [-1.835] [-1.813]

IP
Size -0.017** -0.017** -0.019** -0.018** -0.017**
[-2.199] [-2.224] [-2.358] [-2.298] [-2.198]

CR
Geographic Segments 0.004 0.004 0.003 0.004 0.004
[0.639] [0.632] [0.497] [0.605] [0.628]
Business Segments 0.002 0.002 0.002 0.002 0.002
[0.281] [0.269] [0.238] [0.238] [0.266]
December Year End

Std(CFO)
-0.055*
[-1.741]
-0.010
[-0.104]
US
-0.054*
[-1.734]
-0.008
[-0.077]
-0.054*
[-1.662]
0.015
[0.165]
-0.054*
[-1.716]
-0.004
[-0.040]
-0.054*
[-1.735]
-0.009
[-0.098]
AN
CFO 0.020 0.034 0.054 0.031 0.036
[0.254] [0.442] [0.710] [0.408] [0.471]
Leverage 0.068** 0.069** 0.072** 0.070** 0.069**
[2.215] [2.235] [2.087] [2.289] [2.248]
M

BTM 0.015 0.015 0.020 0.015 0.015

[1.263] [1.245] [1.230] [1.213] [1.254]
Litigation 0.053 0.054 0.056 0.056 0.054
[1.345] [1.357] [1.383] [1.404] [1.365]
ED

Sales Growth 0.004 0.003 0.007 0.004 0.003

[0.165] [0.116] [0.268] [0.138] [0.100]
Material Weakness 0.014 0.014 0.026 0.015 0.015
[0.194] [0.200] [0.349] [0.210] [0.208]
PT

Observations 2,286 2,286 2,187 2,286 2,286

Adjusted R-squared 0.077 0.077 0.080 0.077 0.077
CE

Sample All All All All All

Industry Year Fixed Effects Yes Yes Yes Yes Yes
Auditor Fixed Effects Yes Yes Yes Yes Yes
Clustering Issuer Issuer Issuer Issuer Issuer
AC

56
 ACCEPTED MANUSCRIPT

Table 6: Results of the Inspection Deficiencies Association Models using Non-Accruals Output Measures
This table presents the results of Models (3) and (4) that evaluate the association between academic and practitioner
measures of audit quality, using non-accruals output measures as the academic measures of audit quality. The
dependent variable in Panel A, Part I Finding, equals one if the PCAOB inspection results in the issuance of a Part I
Finding. The dependent variable in Panel B, Internal Rating, is equal to the rating assigned by the audit firms to
their internally inspected engagements. Each column shows the regression results for a different measure of audit
quality, Audit Quality Measure, shown at the top of each column. For example, Audit Quality Measure is Small
Profit in Column (2). Variable definitions are provided in Appendix A. AUC is the area under the curve and
represents a measure of fit of the model. The z- or t-statistic (in brackets) is below the coefficient. Standard-errors
are clustered at the issuer-level. Significance levels are * 10%, ** 5% and *** 1%.
Panel A: PCAOB Inspection Findings
(1) (2) (3) (4) (5)

T
Audit Quality Measure
Dependent Variable: Prior ROA Going Type I Going

IP
Part I Finding Restatement Small Profit Meet Concern Concern
Audit Quality Measure 0.500*** 0.252*** 0.019 0.033 0.034

CR
[5.281] [2.687] [0.187] [0.205] [0.209]
Integrated Audit -0.034 -0.032 -0.041 0.241 0.241
[-0.328] [-0.304] [-0.393] [1.408] [1.408]
Foreign Income -0.028 -0.049 -0.026 0.051 0.051

Size

Geographic Segments
[-0.303]
-0.036
[-1.536]
-0.002
US [-0.542]
-0.045*
[-1.942]
0.002
[-0.292]
-0.039*
[-1.659]
-0.001
[0.377]
-0.157***
[-3.602]
-0.004
[0.377]
-0.157***
[-3.605]
-0.004
AN
[-0.086] [0.094] [-0.073] [-0.132] [-0.134]
Business Segments 0.027 0.028 0.029 0.079* 0.079*
[1.201] [1.268] [1.296] [1.902] [1.901]
December Year End -0.039 -0.039 -0.041 -0.123 -0.123
M

[-0.509] [-0.507] [-0.536] [-1.011] [-1.012]

Std(CFO) 0.114*** 0.117*** 0.116*** 0.124*** 0.124***
[3.265] [3.434] [3.374] [2.885] [2.884]
CFO -0.044 -0.049 -0.056 0.151 0.151
ED

[-0.574] [-0.629] [-0.715] [1.430] [1.431]

Leverage 0.121* 0.126** 0.128** 0.069 0.069
[1.880] [1.960] [1.986] [0.980] [0.983]
BTM 0.033 0.029 0.033 0.034 0.034
PT

[1.205] [1.085] [1.250] [1.129] [1.130]

Litigation -0.271** -0.263** -0.265** -0.336* -0.336*
[-2.510] [-2.426] [-2.452] [-1.920] [-1.920]
CE

Big 4 -0.395*** -0.374*** -0.374*** -0.339** -0.339**

[-4.150] [-3.930] [-3.937] [-2.160] [-2.160]
Sales Growth 0.001 0.012 0.008 0.011 0.011
[0.035] [0.303] [0.193] [0.247] [0.246]
AC

Material Weakness -0.015 0.041 0.046 -0.265 -0.265

[-0.093] [0.255] [0.290] [-1.038] [-1.036]
Altman Z Score 0.001 0.001
[0.742] [0.742]
Observations 5,309 5,309 5,309 1,902 1,902
Pseudo R-squared 0.081 0.078 0.077 0.103 0.103
Sample All All All Distressed Distressed
AUC 0.693 0.690 0.688 0.711 0.711
Industry Year Fixed Effects Yes Yes Yes Yes Yes
Clustering Issuer Issuer Issuer Issuer Issuer

57
 ACCEPTED MANUSCRIPT

Panel B: Internal Inspection Ratings

(1) (2) (3) (4) (5)

Audit Quality Measure
Dependent Variable: Prior ROA Going Type I Going
Internal Rating Restatement Small Profit Meet Concern Concern
Audit Quality Measure 0.149*** 0.081** 0.062* 0.116 0.073
[3.374] [2.293] [1.732] [0.849] [0.530]
Integrated Audit 0.064 0.070* 0.068 0.056 0.057
[1.559] [1.678] [1.634] [0.734] [0.738]
Foreign Income -0.047* -0.056** -0.044* -0.054 -0.054
[-1.813] [-2.126] [-1.690] [-1.162] [-1.159]

T
Size -0.017** -0.019** -0.020** -0.017 -0.017

IP
[-2.171] [-2.426] [-2.555] [-0.760] [-0.790]
Geographic Segments 0.003 0.005 0.004 0.019 0.019
[0.529] [0.735] [0.653] [1.379] [1.367]

CR
Business Segments 0.003 0.002 0.002 0.001 0.001
[0.360] [0.229] [0.204] [0.034] [0.041]
December Year End -0.057* -0.057* -0.056* -0.077 -0.077
[-1.832] [-1.815] [-1.770] [-1.089] [-1.091]
Std(CFO)

CFO
-0.013
[-0.127]
0.037
[0.480]
US-0.003
[-0.031]
0.048
[0.636]
-0.012
[-0.126]
0.039
[0.515]
0.044
[0.644]
0.113
[0.789]
0.046
[0.665]
0.106
[0.743]
AN
Leverage 0.060** 0.062** 0.070** 0.043 0.047
[1.967] [2.028] [2.293] [0.973] [1.086]
BTM 0.011 0.012 0.015 0.023 0.020
[0.935] [0.997] [1.315] [1.624] [1.550]
M

Litigation 0.055 0.056 0.057 0.102 0.099

[1.403] [1.405] [1.420] [1.167] [1.135]
Sales Growth 0.001 0.004 0.003 0.016 0.016
ED

[0.047] [0.144] [0.126] [0.469] [0.464]

Material Weakness -0.006 0.013 0.012 -0.134 -0.137
[-0.091] [0.183] [0.169] [-1.412] [-1.446]
Altman Z Score 0.001 0.001
PT

[0.574] [0.522]

Observations 2,286 2,286 2,286 594 594

Adjusted R-squared 0.083 0.079 0.078 0.057 0.056
CE

Sample All All All Distressed Distressed

Industry Year Fixed Effects Yes Yes Yes Yes Yes
Auditor Fixed Effects Yes Yes Yes Yes Yes
Clustering Issuer Issuer Issuer Issuer Issuer
AC

58
 ACCEPTED MANUSCRIPT

Table 7: Results of the Inspection Findings Association Models using Input Measures
This table presents the results of Models (3) and (4) that evaluate the association between academic and practitioner
measures of audit quality, using input measures as the academic measures of audit quality. The dependent variable
in Panel A, Part I Finding, equals one if the PCAOB inspection results in the issuance of a Part I Finding. The
dependent variable in Panel B, Internal Rating, is equal to the rating assigned by the audit firms to their internally
inspected engagements. Each column shows the regression results for a different measure of audit quality, Audit
Quality Measure, shown at the top of each column. For example, Audit Quality Measure is Audit Fees in Column
(2). Variable definitions are provided in Appendix A. AUC is the area under the curve and represents a measure of
fit of the model. The z- or t-statistic (in brackets) is below the coefficient. Standard-errors are clustered at the issuer-
level. Significance levels are * 10%, ** 5% and *** 1%.
Panel A: PCAOB Inspection Findings

T
(1) (2) (3) (4) (5)
Audit Quality Measure

IP
Dependent Variable: Industry Audit New Office
Part I Finding Specialization Fees Audit Hours Client Size
Audit Quality Measure -0.259 -0.245*** -0.107 0.219** -0.069

CR
[-0.599] [-4.242] [-0.870] [2.181] [-1.474]
Integrated Audit -0.042 0.059 -0.302 -0.039 -0.511*
[-0.402] [0.546] [-0.924] [-0.376] [-1.788]
Foreign Income

Size
-0.025
[-0.281]
-0.037
[-1.581]
US 0.043
[0.478]
0.068**
[2.084]
-0.014
[-0.087]
0.032
[0.480]
-0.023
[-0.255]
-0.039*
[-1.676]
-0.093
[-0.652]
0.039
[0.991]
AN
Geographic Segments -0.001 0.008 -0.039 -0.003 0.002
[-0.060] [0.447] [-1.190] [-0.153] [0.064]
Business Segments 0.029 0.037 0.032 0.029 -0.008
[1.308] [1.635] [0.772] [1.291] [-0.237]
December Year End -0.041 -0.045 -0.059 -0.043 0.017
M

[-0.543] [-0.584] [-0.356] [-0.567] [0.130]

Std(CFO) 0.115*** 0.110*** -1.935 0.116*** -0.753
[3.372] [3.231] [-1.110] [3.400] [-0.888]
ED

CFO -0.058 -0.113 -1.264** -0.054 -0.436

[-0.738] [-1.450] [-2.522] [-0.693] [-1.061]
Leverage 0.127** 0.136** 0.205 0.125* 0.106
[1.974] [2.126] [1.241] [1.926] [0.657]
PT

BTM 0.033 0.019 0.187** 0.034 0.084

[1.239] [0.718] [2.479] [1.262] [0.804]
Litigation -0.265** -0.260** -0.282 -0.271** -0.323*
CE

[-2.451] [-2.405] [-1.209] [-2.491] [-1.691]

Big 4 -0.315** -0.278*** -0.187 -0.341***
[-2.286] [-2.833] [-0.970] [-3.554]
Sales Growth 0.007 -0.005 -0.009 0.007 0.076
AC

[0.184] [-0.126] [-0.073] [0.182] [0.752]

Material Weakness 0.045 0.117 0.556* 0.017 0.072
[0.281] [0.721] [1.710] [0.107] [0.273]
Observations 5,309 5,309 1,134 5,309 2,395
Pseudo R-squared 0.077 0.080 0.076 0.078 0.104
Sample All All Hours Sample All Big 4
AUC 0.689 0.692 0.682 0.689 0.723
Industry Year Fixed Effects Yes Yes Yes Yes Yes
Clustering Issuer Issuer Issuer Issuer Issuer

59
 ACCEPTED MANUSCRIPT

Panel B: Internal Inspection Ratings

(1) (2) (3) (4) (5)

Audit Quality Measure
Dependent Variable: Industry New Office
Internal Rating Specialization Audit Fees Audit Hours Client Size
Audit Quality Measure -0.065 -0.097*** -0.066** 0.027 -0.021*
[-0.485] [-3.597] [-2.392] [0.474] [-1.820]
Integrated Audit 0.069* 0.061 0.075 0.069* 0.043
[1.673] [1.467] [1.610] [1.658] [0.977]
Foreign Income -0.046* -0.024 -0.054* -0.046* -0.054**
[-1.796] [-0.920] [-1.928] [-1.802] [-2.063]

T
Size -0.016** 0.028* 0.004 -0.017** -0.018**

IP
[-2.152] [1.877] [0.337] [-2.157] [-2.214]
Geographic Segments 0.004 0.009 0.014** 0.004 0.008
[0.658] [1.374] [1.974] [0.641] [1.175]

CR
Business Segments 0.002 0.005 0.004 0.002 0.004
[0.266] [0.673] [0.522] [0.265] [0.480]
December Year End -0.055* -0.055* -0.025 -0.055* -0.042
[-1.739] [-1.751] [-0.705] [-1.743] [-1.291]
Std(CFO)

CFO
-0.009
[-0.089]
0.037
[0.491]
US -0.000
[-0.002]
0.001
[0.012]
0.018
[0.197]
0.017
[0.205]
-0.011
[-0.109]
0.037
[0.488]
0.019
[0.200]
0.076
[0.947]
AN
Leverage 0.069** 0.070** 0.063* 0.068** 0.075**
[2.249] [2.313] [1.943] [2.217] [2.378]
BTM 0.015 0.011 0.011 0.015 0.017
[1.267] [0.922] [0.868] [1.227] [1.394]
M

Litigation 0.055 0.049 0.042 0.054 0.058

[1.370] [1.240] [0.935] [1.356] [1.457]
Sales Growth 0.003 0.001 -0.014 0.002 0.014
ED

[0.103] [0.026] [-0.518] [0.082] [0.458]

Material Weakness 0.014 0.038 0.090 0.011 0.008
[0.197] [0.545] [0.982] [0.161] [0.117]
Observations 2,286 2,286 1,870 2,286 2,090
PT

Adjusted R-squared 0.077 0.084 0.080 0.077 0.066

Sample All All Hours Sample All Big 4
Industry Year Fixed Effects Yes Yes Yes Yes Yes
Auditor Fixed Effects Yes Yes Yes Yes Yes
CE

Clustering Issuer Issuer Issuer Issuer Issuer

AC

60
 ACCEPTED MANUSCRIPT

T
IP
Table 8: Results of the Combined Regressions
This table presents the results of Models (3) and (4), using combined regressions that include all the academic proxies of audit quality that were predictive of Part I
Findings and internal inspection ratings in Tables 5 to 7. For brevity, the coefficients on the control variables included in the model are not reported. The dependent

CR
variable in Panel A, Part I Finding, equals one if the PCAOB inspection results in the issuance of a Part I Finding. The dependent variable in Panel B, Internal Rating,
is equal to the rating assigned by the audit firms to their internally inspected engagements. Column (1) corresponds to a baseline regression on control variables only.
Variable definitions are provided in Appendix A. AUC is the area under the curve and represents a measure of fit of the model. Standard-errors are clustered at the
issuer-level. The z- or t-statistic (in brackets) is below the coefficient. Significance levels are * 10%, ** 5% and *** 1%.
Panel A: PCAOB Inspection Findings

Dependent Variable: Part I Finding

Restatement

Small Profit
(1) (2)
0.498***
[5.256]
0.248***
US(3)
0.505***
[5.339]
0.232**
(4)
0.499***
[5.270]
0.244**
(5)
0.497***
[5.256]
0.242**
(6)
0.499***
[5.272]
0.250***
(7)
0.495***
[5.221]
0.257***
(8)
0.493***
[5.215]
0.254***
AN
[2.634] [2.445] [2.571] [2.552] [2.633] [2.713] [2.681]
Abs(Disc. Accruals) 0.366** 0.285*
[2.286] [1.771]
Abs(Accruals) 0.312*** 0.261***
[3.732] [3.054]
M

Abs(Accruals/CFO) 0.024*** 0.021*** 0.019**

[3.199] [2.841] [2.436]
Audit Fees -0.244*** -0.250*** -0.252*** -0.251*** -0.254*** -0.256***
[-4.208] [-4.303] [-4.324] [-4.325] [-4.387] [-4.397]
ED

Observations 5,309 5,309 5,309 5,309 5,309 5,309 5,309 5,309

Pseudo R-squared 0.077 0.083 0.086 0.087 0.088 0.088 0.088 0.089
AUC 0.688 0.695 0.698 0.700 0.701 0.701 0.702 0.702
Control Variables Yes Yes Yes Yes Yes Yes Yes Yes
Industry Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes
PT

Clustering Issuer Issuer Issuer Issuer Issuer Issuer Issuer Issuer

CE
AC

61
 ACCEPTED MANUSCRIPT

T
IP
Panel B: Internal Inspection Ratings

CR
Dependent Variable: Internal Rating (1) (2) (3) (4) (5) (6) (7) (8)
Restatement 0.147*** 0.150*** 0.150*** 0.150*** 0.158*** 0.159*** 0.159***
[3.312] [3.407] [3.408] [3.400] [3.296] [3.319] [3.310]
Small Profit 0.078** 0.065* 0.066* 0.064* 0.063 0.064* 0.061
[2.189] [1.874] [1.899] [1.860] [1.639] [1.652] [1.590]

US
Abs(Disc. Accruals) 0.049 0.097
[0.420] [0.654]
Abs(Accruals) -0.043 -0.051
[-0.548] [-0.368]
Abs(Accruals/CFO) 0.000 0.000 -0.002 -0.002
AN
[0.061] [0.124] [-0.778] [-0.710]
Audit Fees -0.094*** -0.095*** -0.094***
[-3.556] [-3.561] [-3.527]
Audit Hours -0.067** -0.066** -0.065**
[-2.416] [-2.359] [-2.348]
Observations 2,286 2,286 2,286 2,286 2,286 1,870 1,870 1,870
M

Adjusted R-squared 0.077 0.085 0.091 0.090 0.090 0.087 0.087 0.087
Control Variables Yes Yes Yes Yes Yes Yes Yes Yes
Industry Year Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes
Auditor Fixed Effects Yes Yes Yes Yes Yes Yes Yes Yes
ED

Clustering Issuer Issuer Issuer Issuer Issuer Issuer Issuer Issuer

PT
CE
AC

62
 ACCEPTED MANUSCRIPT

T
IP
Table 9: Additional Analyses on Restatements for PCAOB Inspections
This table presents additional analyses that focus specifically on restatements for the PCAOB inspections sample. Columns (1) and (2) focus on material and
immaterial restatements. The dependent variable in Column (3) [(4)] are Part I Findings [not] in revenue and accruals areas. The dependent variable in Column (5) [(6)]
are Part I Findings where the PCAOB does [does not] identify a departure from GAAP. Variable definitions are provided in Appendix A. AUC is the area under the

CR
curve and represents a measure of fit of the model. Standard-errors are clustered at the issuer-level. The z-statistic (in brackets) is below the coefficient. Significance
levels are * 10%, ** 5% and *** 1%.

Dependent Variable:
Restatement
(1)
Part I
Finding
(2)
Part I
Finding
(3)
US
Revenue Accruals
Part I Finding
(4)
Non Revenue Accruals
Part I Finding
(5)
Departure GAAP
Part I Finding
2.081***
(6)
Non Departure GAAP
Part I Finding
0.208**
AN
[8.941] [2.099]
Revenue Accruals Restatement 0.953*** 0.080
[6.387] [0.420]
Non Revenue Accruals Restatement 0.059 0.511***
[0.387] [3.310]
Material Restatement 0.633***
M

[5.035]
Immaterial Restatement 0.289**
[2.188]
Observations 5,309 5,309 5,206 5,184 3,475 5,309
ED

Pseudo R-squared 0.081 0.078 0.100 0.056 0.231 0.072

Sample All All All All All All
AUC 0.692 0.690 0.724 0.670 0.856 0.684
Control Variables Yes Yes Yes Yes Yes Yes
Industry Year Fixed Effects Yes Yes Yes Yes Yes Yes
PT

Clustering Issuer Issuer Issuer Issuer Issuer Issuer

CE
AC

63
 ACCEPTED MANUSCRIPT

Figure 1: PCAOB Inspection Process Timeline

This figure presents the typical timeline of a PCAOB inspection for one fictitious auditor. This auditor has four
engagements, three of which are inspected by the PCAOB during this inspection cycle, out of which two receive
deficiencies such that the work of the auditor was not sufficient to support the audit opinion (Part I Findings). All
issuers have calendar-year fiscal year ends for simplicity. Audits of Year t’s financial statements are typically
completed at the start of Year t+1, in this example between February and March. The PCAOB, as part of the
inspection of the audit firm, inspects these audits in Year t+1, shortly after their completion. PCAOB inspectors
spend approximately one week of fieldwork inspecting each individual engagement. Comment forms, when PCAOB
inspectors identify potential deficiencies in the audit, are issued shortly after the fieldwork is completed. The auditor
may need to conduct additional work on audits that received comment forms per AU 390 (consideration of omitted
procedures after the report date) and AU 561 (subsequent discovery of facts existing at the date of the auditor’s
report). These additional procedures do not influence the PCAOB’s decision to issue a Part I Finding. The inspection

T
report is typically released the year following the inspection, in Year t+2, at least for the U.S. operations of the Big 4.
Some inspection reports have been released after Year t+2 for inspections of smaller or international auditors.

IP
CR
US
AN
M
ED
PT
CE
AC

64
 ACCEPTED MANUSCRIPT

T
IP
CR
US
AN
M
ED
PT
CE
AC

65