Does the Organization and Culture of the
Largest Audit Firms Influence their Audit
Quality and Efficiency?
October 14, 2016
Daniel Aobdia
Kellogg School of Management, Northwestern University
DISCLAIMERS
I completed this study while a Senior Economic Research
Fellow in the Center for Economic Analysis at the PCAOB
The PCAOB, as a matter of policy disclaims responsibility for
any private publication or statement by any of its Economic
Research Fellows and employees
The views expressed are my own and do not necessarily
reflect the views of the Board, its members, or staff of the
PCAOB
RESEARCH QUESTIONS
How much influence do audit firms’ quality control (QC)
systems, have on audit quality, audit efficiency, and audit
pricing (in the U.S.)?
Do all QC systems matter? Are some, such as culture and
audit tools, more important than others?
To answer these questions, use of a unique dataset: The
deficiencies identified by the PCAOB in the firms’ QC
systems (Part II Findings)
Is remediation of QC issues (identified by the PCAOB)
associated with better audit quality? Efficiency?
THEORETICAL BACKGROUND
ON THE ROLE OF QC SYSTEMS AND CULTURE
Only the audit opinion is observed, whereas effort is
unobservable, giving incentives to individual auditors to
shirk. This moral hazard can be reduced if mutual monitoring
takes place (Balachandran and Ramakrishnan 1987)
While audit firms have reputation and litigation incentives
to provide high quality audits, these incentives do not
necessarily apply to their individual employees
Company culture is relevant because employees will face
choices that cannot be properly regulated ex-ante, leading
culture to act as a safeguard (O’Reilly 1989, Kreps 1989,
Guiso et al. 2015)
ANECDOTAL EVIDENCE ON THE INFLUENCE
OF CULTURE AND QC SYSTEMS IN AN AUDIT FIRM:
ARTHUR ANDERSEN IN THE 90s / EARLY 2000s
The rise of consulting at Arthur Andersen led to a culture
change focusing on revenue to the detriment of audit quality,
contrary to the founder’s principles
This lead to several failed audits, including Boston
Chicken, Waste Management and Enron (Brown and
Dugan 2002, Eichenwald and Norris 2002, Richard and
Thurm 2002, Toffler and Reingold 2003, Wyatt 2004,
Gendron and Spira 2009)
Enron: The engagement team was able to overrule the
authority of their national specialists on the engagement
(Eichenwald and Norris 2002)
MOTIVATION
Despite theoretical predictions and anecdotal evidence, very limited
prior empirical literature about firm-wide QC systems
Francis 2011, p138: “Research on the relation between
accounting firms and audit quality is severely limited by the
availability of data on characteristics of accounting firms. To
date, research on this topic has relied on variables that can be
constructed from public disclosures such as client-based measures
of industry expertise or office size. However, these measures do
not go inside the “black-box” of the accounting firm’s
organizational structure and operations” Francis, 2011 p138
DeFond and Zhang 2014, p304: “We currently know little about
basic characteristics of audit firms such as their choice of
ownership structure, governance systems, audit quality control
systems, compensation schemes, or audit technology.”
MOTIVATION
The question is also informative about the role of the PCAOB
inspections of the firms’ QC systems, and the Part II Findings
From an audit quality standpoint: Given outside concerns,
do PCAOB inspections measure true deficiencies in audit
quality? Or only documentation issues?
Are the results in inspections of individual engagements (risk
based) generalizable to the remainder of the firm?
Does remediation of QC issues identified by the PCAOB
improve audit quality?
Are there some other undocumented potential benefits from
the PCAOB inspections process, besides potential
improvement in audit quality?
BACKGROUND ON PCAOB INSPECTIONS
The PCAOB (established by SOX) regularly inspects audit firms
that audit public companies (issuers)
Annual inspections for firms with more than 100 issuers and
triennial inspections for firms with less than 100 issuers
A typical PCAOB inspection incorporates two elements:
A review of selected portions of individual audit
engagements (chosen using a risk-based approach)
Deficiencies are called Part I Findings
A Part I Finding: The audit work is not sufficient to
support the audit opinion, based on applicable standards
A review of the audit firm’s quality control (QC) systems
Deficiencies are called Part II Findings
OVERVIEW OF THE PCAOB REVIEW OF
LARGE AUDIT FIRMS’ QC SYSTEMS
Eight Major Top-Down areas inspected by the PCAOB Based on specific analyses that focus on the firm’s QC policies, and
from inferences from review of individual engagements: Tone at the top (culture), partner management, Independence policies, Client acceptance / retention, Internal inspections, Audit methodology,
Policies related to foreign affiliates, Other practice monitoring
Audit 1 Audit 2 Audit 3 Audit 4 Audit 5 Audit 6
Audit Performance Deficiencies Identified from inspections of individual engagements
Inspection Of QC
Systems:
Inspection Of QC
Systems:
Inspection Of
individual audits:
AN EXAMPLE OF A PART I FINDING
FROM PUBLIC 2011 INSPECTION REPORT OF DELOITTE
1. Issuer A
In this audit, the Firm failed to identify a departure from generally accepted
accounting principles ("GAAP") that it should have identified and addressed
before issuing its audit opinion. Specifically, the issuer inappropriately
allocated to goodwill, rather than to a definite-lived intangible asset, a
portion of the purchase price of a group of assets.
In addition, the Firm failed to perform sufficient procedures to test the
valuation of goodwill for one of the issuer's segments. The issuer used
revenue and earnings projections in its evaluation of the possible impairment of
goodwill and, for this segment, the projected growth rates were significantly
higher than the issuer's recent historical results and projections for the issuer's
industry, which were included in the Firm's work papers. The Firm failed to
sufficiently test the projected growth rates for this segment. Specifically, the
Firm relied on controls related to the issuer's budget without testing the
effectiveness of controls over the development of the assumptions used in the
budget process, and the Firm failed to evaluate, beyond inquiry of
management, the reasonableness of the issuer's revenue and earnings
projections.
PART II FINDING SECTION
FROM 2011 INSPECTION REPORT OF DELOITTE
Section 104(g)(2) of SOX: All Part II Findings will remain
nonpublic if “remediated” within one year
PART II FINDING SECTION
FROM EXPANDED 2007 INSPECTION REPORT OF DELOITTE
A. Audit Performance
A firm's system of quality control should provide reasonable assurance that the
firm's audit work will meet professional standards and regulatory requirements. Not
every deficiency in an audit indicates that a firm's quality control system is
insufficient to provide that assurance, and this report does not discuss every
auditing deficiency observed by the inspection team. On the other hand, some
deficiencies, or repeated instances of a similar deficiency, may indicate a
potentially significant defect in a firm's quality control system even if the
deficiency has not resulted in an insufficiently supported audit opinion. As
described below, some deficiencies reported by the inspection team do suggest that
the Firm's system of quality control may in some respects fail to provide sufficient
assurance that the Firm's audit work will meet applicable standards and requirements.
PART II FINDING SECTION
FROM EXPANDED 2007 INSPECTION REPORT OF DELOITTE
A. Audit Performance
…
b. Management Estimates
The engagement reviews provide cause for concern that the Firm's system of
quality control may not do enough to assure that the Firm performs appropriate
procedures to audit significant estimates, including evaluating management's
assumptions and testing the data supporting the estimates. In addition to
seven engagements described in Part I.A, the inspection team identified eight
engagements (two of which are also discussed in Part I.A with respect to different
estimates) with deficiencies in the Firm's testing of management estimates.
Specifically -
…
PART II FINDING SECTION
FROM EXPANDED 2012 INSPECTION REPORT OF BDO
Deficiencies in Quality Controls Related to the Firm's Internal Inspection
Program
…
In 2012, the PCAOB inspection team inspected three audits that the internal
inspection team had reviewed, including two audits that the internal inspection team
had rated as "unsatisfactory." In each of these three audits, the PCAOB inspection
team identified one or more deficiencies that were not identified by the internal
inspectors but are of such significance that they are included in Part I.A of this
report. The inspection team identified a total of six such deficiencies in these audits
that were not identified by the internal inspectors, even though they had reviewed the
relevant area.
The above discrepancies may suggest that the Firm's internal inspection teams
do not always appropriately apply professional skepticism to the performance of
internal inspections or to the evaluation of the results of those inspections.
…
DATA AND SAMPLE
PCAOB data:
Part II Findings identified from nonpublic parts of the PCAOB
reports
Restricted to the U.S. operations of the largest eight audit
firms (Big 4, GT, BDO, McGladrey and Crowe Horwath) for
comparability purposes
Use of individual engagement inspection deficiencies (Part I
Findings) as one (but not the only one) measure of audit
quality
Audit hours for 2008 till 2012, for the U.S. engagements of
the largest auditors
Merged with S&P Compustat and Audit Analytics data (2004
till 2013) for measures of audit quality (restatements…), audit
fees, client switches and other control variables
MEASURE OF QC DEFICIENCIES
Word count in the Part II Section of the report
Split between audit performance deficiencies and others
(top-down) to determine whether they matter differently
Index of top-down deficiencies
Indices are commonly used in the academic literature (e.g.,
La Porta et al., 1998; Gompers et al., 2003; Bebchuk et al.,
2009; Garmaise, 2011)
Main concern: Remove any element of subjectivity in the measures
and use natural sections in each report
INDEX OF TOP DOWN QC DEFICIENCIES
One point is added to the index for each category if Part II Findings
are identified in the following areas
1. Tone at the top
2. Partner management
3. Independence policies
4. Client acceptance and retention policies
5. Internal inspection program
6. Audit policies, procedures and methodologies, and training
7. Policies related to foreign affiliates
8. (Other) practice monitoring
Index = 0: No major deficiencies in QC system (“good”)
Index = 8: Deficiencies identified in all categories of the index (“bad”)
Later on, separate Tone at the top and Audit methodology issues
and create a sub-index
PROPORTION OF FIRM-YEARS WITH DEFICIENCIES IN
Proportion of firm-years (%)
0 20 40 60 80 100
Tone at the top
Partner management
Independence policies
Client acceptance and retention policies
Internal inspection program
Audit policies, methodology and training
Policies related to foreign affiliates
(Other) practice monitoring
SUMMARY OF RESULTS
H1: An audit firm’s QC deficiencies are negatively associated with audit
quality
Negative association between QC issues and audit quality
For both “audit performance” and “top-down” deficiencies, including
tone at the top
H2: There is no association between audit firm’s QC deficiencies and audit
pricing
No relationship between QC issues and audit pricing
SUMMARY OF RESULTS
H3: There is no association between audit firms’ QC deficiencies and audit
efficiency
Positive association between top-down QC issues and audit hours
Consistent with “inefficiency” hypothesis
Mainly driven by tone and methodology issues
Additional results:
Some evidence that remediation improves audit quality and efficiency
Positive association between QC deficiencies and client switches
TYPICAL INSPECTION PROCESS TIMELINE
Year t-1 Year t Year t+1
End of Fiscal Year t-1
End of Audit Year t-1
Fieldwork (March to November) • Year t-1 audits inspected • QC systems at t? t-1? inspected
End of Fiscal Year t
End of Audit Year t
The analysis of the firm’s QC systems is related to the prior year’s
audits, but some analyses appear to use current information (in
contrast with the inspection of individual audits)
Use the deficiencies in the firm’s QC systems identified in
Year t for audit quality and efficiency measured in Year t
(results also robust if look at Year t-1 output variables)
Also use the inspection findings identified in Year t+1 for the
Year t audits as a measure of audit quality (different year)
Fieldwork (March to November) • Year t audits inspected • QC systems at t? t+1? inspected
RESEARCH DESIGN:
CROSS-SECTIONAL REGRESSION
For several measures (audit quality, hours, fees), run the
following regression
Measure = α + β.QC Issues + γ.Controls + ε
QC issues: Number of words (audit performance and others) in
report, QC Index, Reduced Index, Tone, Methodology
Measures of audit quality (see Aobdia 2015): Restatements,
propensity to meet/beat the zero earnings threshold, accruals,
Part I Findings (for inspected engagements the following year)
Extensive battery of controls, including year fixed effects (results
are not particularly sensitive to inclusion)
AUDIT QUALITY RESULTS (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables: Restatement Meet/Beat PartIFinding |Accruals/CFO|
Number_Words 0.202*** 0.077*** 0.220*** -0.032
[5.905] [2.784] [3.288] [-0.901]
Year Fixed Effects Yes Yes Yes Yes
Control Variables Yes Yes Yes Yes
Observations 27,837 27,837 2,452 27,837
Pseudo R-squared 0.038 0.087 0.101
Adjusted R-squared 0.134
Clustering Issuer Issuer Issuer Issuer
Model Logistic Logistic Logistic OLS
AUDIT QUALITY RESULTS (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables: Restatement Meet/Beat PartIFinding |Accruals/CFO|
Number_Words 0.202*** 0.077*** 0.220*** -0.032
[5.905] [2.784] [3.288] [-0.901]
Year Fixed Effects Yes Yes Yes Yes
Control Variables Yes Yes Yes Yes
Observations 27,837 27,837 2,452 27,837
Pseudo R-squared 0.038 0.087 0.101
Adjusted R-squared 0.134
Clustering Issuer Issuer Issuer Issuer
Model Logistic Logistic Logistic OLS
AUDIT QUALITY RESULTS (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables: Restatement Meet/Beat PartIFinding |Accruals/CFO|
Number_Words_Perf 0.208*** -0.000 0.135* -0.050
[5.459] [0.003] [1.860] [-1.333]
Reduced_QC_Index -0.003 0.030* 0.095** 0.044**
[-0.134] [1.824] [2.260] [2.128]
Tone 0.026 0.098** -0.014 -0.026
[0.430] [1.838] [-0.093] [-0.408]
Methodology 0.139** 0.126** 0.233* -0.009
[2.328] [2.426] [1.668] [-0.137]
Full Set of Controls Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 27,837 27,837 2,452 27,837
Pseudo R-squared 0.040 0.087 0.102
Adjusted R-squared 0.134
AUDIT QUALITY RESULTS (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables: Restatement Meet/Beat PartIFinding |Accruals/CFO|
Number_Words_Perf 0.208*** -0.000 0.135* -0.050
[5.459] [0.003] [1.860] [-1.333]
Reduced_QC_Index -0.003 0.030* 0.095** 0.044**
[-0.134] [1.824] [2.260] [2.128]
Tone 0.026 0.098** -0.014 -0.026
[0.430] [1.838] [-0.093] [-0.408]
Methodology 0.139** 0.126** 0.233* -0.009
[2.328] [2.426] [1.668] [-0.137]
Full Set of Controls Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 27,837 27,837 2,452 27,837
Pseudo R-squared 0.040 0.087 0.102
Adjusted R-squared 0.134
AUDIT QUALITY RESULTS (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables: Restatement Meet/Beat PartIFinding |Accruals/CFO|
Number_Words_Perf 0.208*** -0.000 0.135* -0.050
[5.459] [0.003] [1.860] [-1.333]
Reduced_QC_Index -0.003 0.030* 0.095** 0.044**
[-0.134] [1.824] [2.260] [2.128]
Tone 0.026 0.098** -0.014 -0.026
[0.430] [1.838] [-0.093] [-0.408]
Methodology 0.139** 0.126** 0.233* -0.009
[2.328] [2.426] [1.668] [-0.137]
Full Set of Controls Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 27,837 27,837 2,452 27,837
Pseudo R-squared 0.040 0.087 0.102
Adjusted R-squared 0.134
AUDIT HOURS RESULTS (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables:
Audit
Hours
Partner
Hours
EQR
Hours
Fees per
Hour
Number_Words_Perf -0.071*** -0.133*** -0.229*** 18.522***
[-8.829] [-14.468] [-27.969] [10.979]
Reduced_QC_Index -0.000 0.038*** 0.039*** 1.386
[-0.085] [7.232] [8.899] [1.134]
Tone 0.125*** 0.136*** 0.279*** -33.041***
[10.402] [8.003] [19.761] [-10.841]
Methodology 0.092*** 0.075*** 0.048*** -18.401***
[6.553] [4.928] [3.458] [-5.574]
Full Set of Controls Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 13,802 13,793 13,698 13,802
Adjusted R-squared 0.735 0.578 0.384 0.095
AUDIT HOURS RESULTS (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables:
Audit
Hours
Partner
Hours
EQR
Hours
Fees per
Hour
Number_Words_Perf -0.071*** -0.133*** -0.229*** 18.522***
[-8.829] [-14.468] [-27.969] [10.979]
Reduced_QC_Index -0.000 0.038*** 0.039*** 1.386
[-0.085] [7.232] [8.899] [1.134]
Tone 0.125*** 0.136*** 0.279*** -33.041***
[10.402] [8.003] [19.761] [-10.841]
Methodology 0.092*** 0.075*** 0.048*** -18.401***
[6.553] [4.928] [3.458] [-5.574]
Full Set of Controls Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 13,802 13,793 13,698 13,802
Adjusted R-squared 0.735 0.578 0.384 0.095
AUDIT HOURS RESULTS (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables:
Audit
Hours
Partner
Hours
EQR
Hours
Fees per
Hour
Number_Words_Perf -0.071*** -0.133*** -0.229*** 18.522***
[-8.829] [-14.468] [-27.969] [10.979]
Reduced_QC_Index -0.000 0.038*** 0.039*** 1.386
[-0.085] [7.232] [8.899] [1.134]
Tone 0.125*** 0.136*** 0.279*** -33.041***
[10.402] [8.003] [19.761] [-10.841]
Methodology 0.092*** 0.075*** 0.048*** -18.401***
[6.553] [4.928] [3.458] [-5.574]
Full Set of Controls Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 13,802 13,793 13,698 13,802
Adjusted R-squared 0.735 0.578 0.384 0.095
EVIDENCE ON REMEDIATION: AUDIT QUALITY (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables: Restatement Meet/Beat PartIFinding ScaledAccrualsCFO
Number_Words 0.189*** 0.074*** 0.157** -0.032
[5.433] [2.630] [2.219] [-0.853]
Proportion_Remediated -0.401** 0.011 -2.454*** 0.135
[-2.284] [0.072] [-5.369] [0.623]
Full Set of Controls Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 23,572 23,572 2,089 23,572
Adjusted/Pseudo R-squared 0.032 0.085 0.126 0.136
EVIDENCE ON REMEDIATION: AUDIT QUALITY (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables: Restatement Meet/Beat PartIFinding ScaledAccrualsCFO
Number_Words 0.189*** 0.074*** 0.157** -0.032
[5.433] [2.630] [2.219] [-0.853]
Proportion_Remediated -0.401** 0.011 -2.454*** 0.135
[-2.284] [0.072] [-5.369] [0.623]
Full Set of Controls Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 23,572 23,572 2,089 23,572
Adjusted/Pseudo R-squared 0.032 0.085 0.126 0.136
EVIDENCE ON REMEDIATION: AUDIT HOURS (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables: Logaudithours Logpartnerhours Logeqrhours Fees_per_hour
Number_Words -0.003 -0.056*** -0.061*** 1.192
[-0.580] [-7.487] [-9.684] [0.934]
Proportion_Remediated -0.075*** -0.135*** 0.129*** 16.750***
[-3.235] [-4.555] [4.237] [3.399]
Full Set of Controls Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 13,802 13,793 13,698 13,802
Adjusted/Pseudo R-squared 0.731 0.565 0.322 0.082
EVIDENCE ON REMEDIATION: AUDIT HOURS (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables: Logaudithours Logpartnerhours Logeqrhours Fees_per_hour
Number_Words -0.003 -0.056*** -0.061*** 1.192
[-0.580] [-7.487] [-9.684] [0.934]
Proportion_Remediated -0.075*** -0.135*** 0.129*** 16.750***
[-3.235] [-4.555] [4.237] [3.399]
Full Set of Controls Yes Yes Yes Yes
Year Fixed Effects Yes Yes Yes Yes
Observations 13,802 13,793 13,698 13,802
Adjusted/Pseudo R-squared 0.731 0.565 0.322 0.082
CLIENT SWITCHES RESULTS (*: Statistically significant at 10%, **: 5%, ***, 1%)
Dependent Variables: Switch Switch Switch Switch
Number_Words_Perf 0.142*** 0.136*** 0.111** 0.105*
[2.721] [2.597] [2.050] [1.915]
QCIndex 0.090*** 0.080***
[3.349] [3.000]
Reduced_QC_Index 0.071** 0.062**
[2.503] [2.190]
Methodology 0.265*** 0.247***
[3.036] [2.803]
Tone 0.113 0.111
[1.450] [1.419]
Full Set of Controls No Yes No Yes
Reduced Set of Controls Yes No Yes No
Year Fixed Effects Yes Yes Yes Yes
Observations 25,675 25,675 25,675 25,675
Pseudo R-squared 0.101 0.123 0.102 0.123
CONCLUSION
QC systems matter from an audit quality, efficiency, and client
standpoint
Remediation of QC deficiencies has the potential to improve both
audit quality and efficiency
Some of the QC deficiencies identified by the PCAOB appear to
measure nontrivial issues
To a certain extent, the results of individual PCAOB inspections, risk
based, can be generalized to the remainder of the firm
THANK YOU