which audit input matters? an analysis of the determinants ... audit input matters - aobdia...1...

66
1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability, and Audit Fees Using PCAOB Data Daniel Aobdia Kellogg School of Management, Northwestern University [email protected] Preeti Choudhary Eller College of Management, University of Arizona [email protected] Noah Newberger Public Company Accounting Oversight Board [email protected] This version: September 2018 Daniel Aobdia and Preeti Choudhary were Senior Economic Research Fellows in the Center of Economic Analysis (now called Office of Economic Research & Analysis ERA) at the Public Company Accounting Oversight Board (PCAOB), and conducted this research as part of this affiliation. Noah Newberger is a Senior Research Analyst in ERA 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 in this paper are ours and do not necessarily reflect the views of the Board, individual Board members, or staff of the PCAOB. We are thankful to PCAOB staff and seminar participants at the PCAOB for helpful comments and suggestions. Daniel Aobdia gratefully acknowledges generous financial support from the Kellogg School of Management and in particular the Lawrence Revsine Fellowship.

Upload: others

Post on 26-Dec-2019

3 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

1

Which Audit Input Matters? An Analysis of the Determinants of Audit

Quality, Profitability, and Audit Fees Using PCAOB Data

Daniel Aobdia

Kellogg School of Management, Northwestern University

[email protected]

Preeti Choudhary

Eller College of Management, University of Arizona

[email protected]

Noah Newberger

Public Company Accounting Oversight Board

[email protected]

This version: September 2018

Daniel Aobdia and Preeti Choudhary were Senior Economic Research Fellows in the Center of Economic Analysis

(now called Office of Economic Research & Analysis – ERA) at the Public Company Accounting Oversight Board

(PCAOB), and conducted this research as part of this affiliation. Noah Newberger is a Senior Research Analyst in

ERA 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 in this paper are ours and

do not necessarily reflect the views of the Board, individual Board members, or staff of the PCAOB. We are

thankful to PCAOB staff and seminar participants at the PCAOB for helpful comments and suggestions. Daniel

Aobdia gratefully acknowledges generous financial support from the Kellogg School of Management and in

particular the Lawrence Revsine Fellowship.

Page 2: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

2

Abstract

We exploit proprietary data from the Public Company Accounting Oversight Board (PCAOB) to

analyze how a variety of characteristics regarding the composition of the audit and audit team are

associated with audit quality measured via restatements and regulator-identified audit

deficiencies. We find that more time spent prior to the final phase of the audit is associated with

better audit quality, specifically when more time is spent by the core audit engagement team. We

also decompose the time spent and experience characteristics of the core audit engagement team

into the following roles: lead partner, engagement quality reviewer, and other experienced team

members (comprised of other audit partners, directors, senior managers, and managers). We

generally find that more time spent and more experience of experienced team members other

than the lead partner are associated with better audit quality. Some of the characteristics that

improve audit quality are costly to the client, but not necessarily to the audit firm. Overall, our

analysis highlights several cost-benefit tradeoffs to improving audit quality.

Page 3: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

1

1. Introduction

In this study, we exploit proprietary data from the Public Company Accounting Oversight

Board (PCAOB) to analyze how a variety of characteristics regarding the composition of the

audit and audit team are associated with audit quality measured via restatements and regulator-

identified (PCAOB) audit deficiencies. Specifically, we focus on the amount of time spent

during different phases of the audit as well as time spent and experience characteristics of the

lead audit engagement team. We partition the lead engagement team into the three following

mutually exclusive roles: lead partner, engagement quality reviewer (EQR), and other

experienced team members, where the latter are comprised of other audit partners, directors,

senior managers, and managers. Lastly, we evaluate whether some of the characteristics

associated with improved audit quality are costly for the audit firm or client, thereby highlighting

cost-quality trade-offs.

This analysis is important and timely given that many audit regulators and firms have

recently introduced audit quality indicator projects to better define and measure audit quality,

and, importantly, understand the determinants of audit quality (e.g., Christensen et al. 2016).1

Despite this increased interest from practitioners about understanding the determinants of audit

quality, archival evidence on the audit process remains scarce, due to lack of comprehensive data

that measures both engagement characteristics and audit quality (DeFond and Zhang 2014). Our

analysis helps open the black box of the role of audit process on audit quality.

1 For example, projects seeking to evaluate audit quality are on the agendas of the U.K. Financial Reporting Council

(FRC 2008), the International Auditing and Assurance Standards Board (IAASB 2013), the Center for Audit Quality

(CAQ 2014), the PCAOB (PCAOB 2015), the Canadian Public Accountability Board (CPAB 2018), Singapore’s

Accounting and Corporate Regulatory Authority (ACRA 2015), Netherlandse Beroepsorganisatie van Accountants

(NBA 2015), Switzerland’s Federal Audit Oversight Authority (FAOA 2015), Chartered Accountants of Australia

and New Zealand (CAANZ, 2015), International Organisation of Securities Commissions (IOSCO 2009), among

others.

Page 4: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

2

We exploit proprietary data obtained from the PCAOB for the U.S. operations of the

eight largest auditors. One of the roles of the PCAOB’s regulatory activities is to inspect

individual audit engagements of SEC-registered companies to determine whether these audits

were conducted according to relevant audit standards (Section 104 of the Sarbanes Oxley Act of

2002, SOX). If the PCAOB determines that the work performed was not sufficient to support the

audit opinion, the PCAOB issues a Part I Finding (an audit deficiency). As part of the inspection

process, the PCAOB collects specific information about the audit engagement characteristics.

We exploit this information, along with additional PCAOB data. Because our analysis is limited

to inspected engagements that are not randomly chosen, our analysis may not generalize to the

full population. In Section 4.9, we discuss several analyses that mitigate these concerns.

Our dataset contains 2,339 engagements inspected by the PCAOB between 2006 and

2015. It includes a breakdown of audit hours among the different phases of the audit and the

different personnel that form the core audit team, as well as the identity of the senior team

members, from the audit manager level, and their client specific, industry, and overall auditing

experiences. We combine this dataset with Compustat, Audit Analytics, and additional PCAOB

information to build several dependent and control variables. We consider two measures of audit

quality, one based on audit process quality, the other on the client’s financial reporting quality.

Audit process quality is measured by whether the PCAOB identifies any Part I Findings; under

the assumption that inspectors are objective audit experts, such findings yield a measure of

whether the audit was conducted in accordance with PCAOB standards (see Aobdia 2018a for

more details). Financial reporting quality is measured by whether the client restates its fiscal-year

end financial statements, which captures circumstances in which the auditor missed

misstatements at the time it issued the audit opinion. Restatements represent a joint outcome of

Page 5: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

3

both the client and management (e.g., Aobdia 2018a, Choudhary, Schipper, and Merkley 2018a).

We consider three measures of audit cost, including: audit fees, which capture the cost to the

client, the audit realization rate, which captures the engagement’s profitability, as well as the

total audit hours worked, which capture audit effort.2

Our first set of analyses evaluates the time spent prior to the final phase of the audit,

which typically corresponds to audit work conducted prior to the client’s fiscal year end. Ex ante,

it is unclear whether more time spent during the pre-final phase of the audit positively influences

audit quality. On the one hand, if auditors adjust personnel levels and expertise, as well as the

extent of audit procedures, based on the risk assessment conducted during the initial phases of

the audit (e.g., Johnstone and Bedard 2001) then time spent during the pre-final phase may

matter. Alternatively, if the link between risk assessment and the extent of testing is weak, then

time spent during the pre-final phase may not matter (e.g., Allen et al. 2006).

We find that engagements that spend more time during the pre-final phase exhibit better

audit process and financial reporting quality. These results are mainly driven by the time spent

by the core audit team, not by the information systems auditors, tax auditors, or audit specialists.

Our coefficients imply that an increase of one standard deviation in the proportion of the audit

time spent during the pre-final phase (10%) is associated with reductions in the probabilities of

Part I Findings and restatements of 1.8% and 2.5% points, respectively. The economic effects are

reasonably large given that the average Part I Finding and restatement rates in our sample are

28% and 12%, respectively. However, we find that this improvement in audit quality comes with

2 Realization rate is equal to total audit fees charged divided by the maximum audit fee the auditor would have

charged had all hours been billed at their undiscounted rate, which varies depending on whether the hours are

conducted by partners, managers, seniors or associates. Thus, a 100% realization rate corresponds to an audit where

the auditor works the planned number of hours and bills these hours at their maximum possible rate. The higher the

realization rate, the higher the audit profitability.

Page 6: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

4

a cost to the client in the form of higher audit fees, but not necessarily to the auditor, as we fail to

find any relation between time during the pre-final phase and audit firm profitability.

In our second set of analyses, we further drill down into the composition of the core audit

engagement team. We evaluate whether the time spent and expertise characteristics of

experienced team members that work under the guidance of the lead partner relate to audit

quality. On the one hand, PCAOB standard AS 1201 clearly states that the lead partner is

ultimately responsible for the engagement and its performance. Consistent with this idea, prior

research outside the U.S. finds significant individual lead partner effects on audit quality (e.g.,

Lennox and Wu 2018). Theoretical economics research about hierarchies also implies a more

important role for the person in charge (e.g., Garicano 2000, Garicano and Hubbard 2007). Yet

prior research within the U.S. surprisingly finds limited evidence that lead partner characteristics

influence audit quality (e.g., Laurion et al. 2016, Aobdia et al. 2018a, Gipper et al. 2018),

perhaps because lead partners delegate their audit quality responsibilities to their engagement

teams. It is also an empirical question whether trade-offs exist between audit quality and audit

cost for the engagement characteristics that positively influence audit quality. Some of these

characteristics might make audits both more efficient and effective; others may be costly to

either the client, the auditor or both.

To decompose the core audit engagement team, we consider the proportion of time spent,

split between lead partner, EQR, and other experienced team members. While we find evidence

consistent with a positive influence of the time spent by the EQR and other experienced team

members on audit quality, we do not find such results when focusing on the lead partner. We

also find that increased quality from other experienced team members’ time spent comes at a

reasonably high cost to the client in the form of higher audit fees. An increase of one standard

Page 7: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

5

deviation of other experienced team members time (6%) is associated with a 9.4% increase in

audit fees, and a 2.1% reduction in the probability of Part I Finding.

We also assess whether heterogeneity in the expertise of the core audit team experienced

members influences audit quality. Evaluating client-specific expertise, we find that increased

client expertise is associated with better audit quality. These results are driven by the other

experienced team members. A one-standard deviation increase in client specific expertise of

other experienced team members, 1.75 years, is associated with a 2% point decrease in the

probability of restatements. Again, this quality improvement comes with a cost to the client, in

the form of increased audit fees. However, we also find a positive association between client

expertise and audit realization rate, implying no trade-offs between audit profitability and audit

quality from an audit firm’s standpoint. Confirming the results in Gipper et al. (2018), we find no

association (a positive association) between lead partner’s client-specific expertise and audit

quality (audit realization rate).

We also find a positive association between average core audit team members’ industry

specialization and audit quality. The results are driven by the EQR and other experienced team

members. We also find a positive association between EQR overall expertise, proxied by the

EQR total years of audit experience, and audit quality. Consistent with Aobdia et al. (2018a), we

find no association (a positive association) between lead partner industry specialization and audit

quality (audit fees). We fail to find any association between lead partner overall experience and

audit quality. In additional analyses, we fail to find an association between the number of

experienced auditors in the team, or the team gender, and audit quality.

Overall, our results are consistent with several ideas. First, the time spent during the pre-

final phases of the audit positively influences audit quality. Second, other experienced audit team

Page 8: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

6

members beside the lead partner play an important role in terms of audit quality. Both their time

and client-specific and industry-expertise are positively associated with audit quality. We also

find evidence that the EQR plays an important role in terms of audit quality. In particular, an

EQR’s industry and general expertise positively influence audit quality, whereas client-specific

expertise appears to matter less. Third, while we fail to observe any trade-offs between audit

quality and auditor profitability, we do find that in many instances the audit characteristics that

are associated with better audit quality come with a cost to the client. Thus, depending on a

client’s willingness to pay, trade-offs exist between audit quality and the cost of the audit. We

also find evidence consistent with Laurion et al. (2016), Aobdia et al. (2018a), and Gipper et al.

(2018) that lead partner characteristics are largely unassociated with audit quality in the U.S.

Our analysis helps answer several major gaps in prior audit literature. First, there is an

older line of research on audit production that has access to audit process data but often assumes

that audit quality is constant within the same audit firm (e.g., O’Keefe et al. 1994, Davidson and

Gist 1996, Knechel et al. 2009). However, a different and more recent line of research

documents heterogeneity in audit quality within the same auditor, due to differences in industry

specialization, office characteristics such as size, and lead partner characteristics (e.g., Francis

and Yu 2009, Reichelt and Wang 2010, Gul et al. 2013). The second line of research is generally

constrained by lack of audit process data.3 We expand upon these two lines of research by

evaluating whether some previously unexplored characteristics of the audit team and time spent

are associated with heterogeneous audit quality effects within the same auditor.

3 A recent exception, Cameran et al. (2017), focuses on how audit teams are structured in the Italian setting and find

evidence consistent with audit team diversity affecting audit quality. However, Cameran et al. (2017) does not have

access to a strong measure of audit quality and relies on abnormal working capital accruals instead.

Page 9: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

7

Second, in exploring heterogeneity within the same audit firm, the audit literature

changed the unit of analysis from audit firms to audit offices, and more recently to engagement

partners (e.g., Lennox and Wu 2018). While some prior research on engagement partners

documents heterogeneities in audit quality due to partner differences (e.g., Gul et al. 2013,

Aobdia et al. 2015), most of the evidence is from outside of the United States. Recent studies

using the U.S. setting document little influence of lead partner characteristics on audit quality

(e.g., Laurion et al. 2016, Aobdia et al. 2018a, Gipper et al. 2018). Our study builds upon these

U.S. based studies to determine what characteristics of an audit engagement ultimately influence

audit quality in the U.S. One possible explanation for the lack of partner influence on audit

quality in the U.S. is that the role of the U.S. lead partner may relate to managing client relations,

whereas other audit team members may focus more on managing the audit. Accordingly, our

analysis of audit team members beyond the lead partner potentially provides additional

explanations for results that question the importance of the lead partner.

Third, prior research studying heterogeneity of audit firms often does not consider the

various trade-offs required to improve audit quality, primarily due to lack of data.4 While several

audit process or engagement team attributes might affect audit quality, these may also come with

a cost, either to the audit firm, or to the client. We exploit proprietary data to evaluate costs

beyond audit fees, including audit hours and engagement realization rates that reflect the

percentage discount in fees received over time spent by a variety of professionals on the audit.

Our analysis is able to pinpoint these trade-offs, thereby highlighting potential conflicts of

interest between auditor profitability or client costs, and audit quality. Generally, our results

4 Two exceptions are Aobdia et al. (2018a) and Gipper et al. (2018).

Page 10: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

8

regarding the cost/quality tradeoff show that the cost is typically borne by the client, not the

auditor.

Collectively, our results imply that audit practitioners and academics should focus more

on experienced audit team members other than the lead partner to understand their role on audit

quality, and performance in general. While an extensive literature focuses on the role of lead

partners, little is known about the roles below, due to lack of data (Lennox and Wu 2018). Our

results raise the question whether some of the results in prior literature that find that lead partners

influence audit quality or performance might actually be driven by their choices of engagement

team members. Nevertheless, two caveats are in order. First, our analysis is conducted on a

sample where the name of the lead partner is not public. In a regime where the lead partner

names are public, reputation effects might incentivize lead partners to achieve high quality

audits, rather than just striving to reach an acceptable level, and manage the audit process more

proactively. Second, our analyses are based on a sample of engagements inspected by the

PCAOB. Because the PCAOB does not randomly inspect engagements, it is not certain that our

results can generalize to non-inspected engagements.5 Therefore, we encourage academic

researchers to continue exploring this line of research to determine whether our results can

generalize to different samples, regulatory regimes, and countries.

The remainder of this paper is structured as follows. Section 2 includes a review of prior

literature and develops the main hypotheses; Section 3, describes the data and the sample

construction; Section 4, the empirical tests and results. Section 5 concludes.

2. Prior literature and main hypotheses

2.1 Prior literature

5 See the analyses in Section 4.9 that suggest that selection bias concerns are not too severe for our sample.

Page 11: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

9

At least two streams of literature are relevant to our study. The first is the literature that

studies how the audit is executed (i.e. audit production), which largely holds the output (audit

quality) constant. The second stream of literature on hierarchy and teams is largely theoretical,

and studies how different characteristics of team structure affect the output.

2.1.1 Literature on audit production

An older literature on audit production models an audit as a production function and

focuses on specific units of effort in the audit process necessary to achieve a certain level of

output (e.g., Knechel et al. 2009). In this line of research, the audit output represents the level of

assurance obtained by the audit process, or, in other words, audit quality (O’Keefe et al. 1994).

However, due to the lack of an appropriate measure of assurance obtained, the literature often

assumes that the audit market is competitive, and that audit quality is constant within the same

auditor (e.g., O’Keefe et al. 1994). As a result, prior studies often treat audit production as a cost

minimization problem, holding the output (audit quality) constant. For example, Simunic (1980)

uses audit fees to consider the joint role of audit costs and audit pricing, and models audit fee

determinants based on a variety of client characteristics. Subsequent research exploits direct data

on audit hours and focuses on audit costs only, using aggregated or disaggregated audit hours as

dependent variables, and a variety of client characteristics as explanatory variables (e.g.,

Palmrose 1986, 1989, Davis et al. 1993, O’Keefe et al. 1994, Davidson and Gist 1996,

Hackenbrack and Knechel 1997).

More recent studies, such as Dopuch et al. (2003), and Knechel et al. (2009), introduce

efficiency frontier techniques such as data envelopment analysis to examine relative efficiency in

audit production. In particular, Knechel et al. (2009) extend prior literature by considering audit

effort, i.e. audit hours, as the output, and the cost of staff resources as the input. The underlying

Page 12: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

10

assumption in Knechel et al. (2009) is that more audit effort in the form of more hours generates

more audit evidence and therefore a higher level of assurance; however, whether this assumption

is valid remains uncertain. In the Greek setting, Caramanis and Lennox (2008) find a positive

association between audit hours and audit quality; however, Aobdia (2018a) does not find any

association between audit hours and PCAOB Part I Findings. Additional studies such as Gong et

al. (2016) and Aobdia (2018b) also find that audit hours do not always move in tandem with

audit quality.

Overall, at least one main concern is that prior studies of audit production typically

assume that audit quality is constant within the same auditor, in part, because they do not have

access to a direct measure of audit quality. However, a more recent line of research documents

significant heterogeneity in audit quality, even within the same auditor. The literature initially

documents variation at the industry or office levels. For example, Balsam et al. (2003) and

Krishnan (2003) find evidence consistent with industry specialist auditors (measured at the

national level) conducting higher quality audits. Francis and Yu (2009) find evidence consistent

with larger audit offices providing higher quality audits, and Reichelt and Wang (2010) find

evidence consistent with industry specialist offices providing higher audit quality when the

auditor is also a national industry specialist. The literature subsequently documents heterogeneity

in audit quality at the partner level outside of the United States. For example, Gul et al. (2013),

Aobdia et al. (2015), Knechel et al. (2015), and Cameran et al. (2018) find evidence that audit

quality depends on the lead partner in China, Taiwan, Sweden, and the U.K., respectively. Chi et

al. (2017) find in Taiwan evidence consistent with lead partner client-specific and industry-

expertise influencing audit quality. Several other studies also show an influence of the lead

partner on audit quality outside of the United States (see Lennox and Wu 2018 for a recent

Page 13: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

11

review of the literature). Overall, these studies imply that research on the audit process cannot

assume that the output of an audit is constant within the same auditor, and that more precise

measures of the degree of assurance achieved by the audit need to be considered in audit

production models.

2.1.2 Literature on teams and hierarchy

An individual audit engagement requires the use of a team of auditors, with a hierarchy

within the team, to achieve a certain degree of assurance level. The analytical literature in

economics recognizes the importance of team dynamics on output. The literature generally either

focuses on moral hazard issues, or on frictions in the communication of knowledge among team

members. On the one hand, Holmstrom (1982) identifies incentives for team members to shirk,

especially in larger teams when effort is costly, implying that team efficiency and effectiveness

might decrease as the team becomes larger. Georgiadis (2015) studies a dynamic model with a

group of agents that collaborate to complete a project with costly effort. He finds that agents

work harder when the project is closer to completion, implying that team efficiency and

effectiveness might be greater during the final phase of a project. On the other hand, Garicano

(2000) and Garicano and Hubbard (2008) focus on hierarchies in organizations and find that a

hierarchy is a natural way to organize knowledge when matching problems with those who know

how to solve them is costly. Thus, individuals are more likely to influence the overall output of

the organization, as they are closer to the top of the hierarchy.

Despite this extensive theoretical literature on teams and hierarchy, the empirical

literature is more limited, due to lack of data. The results suggest that in our setting the

experienced professionals other than the top (lead) partner influence the output, partially in

Page 14: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

12

support of these theories. Further, the results suggest that frictions in the transmission of

knowledge are more prevalent in practice relative to moral hazard issues in our setting.

2.2 Hypotheses

2.2.1 Focus on the pre-final phases of the audit

Our first hypothesis evaluates the importance of the pre-final phases of the audit, which

represents audit work done prior to the client’s fiscal year end. The pre-final phase is comprised

of the planning, quarterly review, and interim field phases. During the planning phase of an

audit, PCAOB standards require an auditor to establish the overall audit strategy for the

engagement and to develop an audit plan, which includes planned risk assessment procedures

and planned responses to the risks of material misstatement (AS 2101). In particular, standards

require the engagement team to establish materiality levels and identify and assess the main risks

of material misstatements during the planning phase (AS 2105, AS 2110). Thus, PCAOB

standards grant an important role to planning activities. This is intuitively appealing given that a

proper assessment of risks can allow an auditor to develop an effective audit plan to reduce the

risk of misstatements. Consistent with this idea, O’Keefe et al. (1994) find a positive association

between inherent risk and senior and staff hours used in the audit. Johnstone and Bedard (2001)

and Bedard and Johnstone (2004) further find that audit firms respond to increased fraud and

error risk factors by assigning more high-risk specialist personnel, assigning more industry

expert personnel, and applying more intensive testing and reviews. Some behavioral studies also

find an influence of risk on audit effort decisions (e.g., Libby et al. 1985, Maletta and Kida 1993,

Zimbelman 1997). However, several studies also find that the link between risk assessment and

substantive testing is weaker than one may expect (Allen et al. 2006). For example Fukukawa et

al. (2005), focusing on Japanese audits, find that fraud risk factors have little influence on audit

Page 15: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

13

planning. Accordingly, it is unclear whether time spent in planning the audit matters for audit

quality.

Other pre-final phases, such as quarterly reviews and interim field, may also have a role

to play in improving audit quality. Quarterly review pertains to audit procedures performed

under AS 4105, Reviews of Interim Financial Statements, such as performing analytical

procedures and making inquiries based upon a review of quarterly financial statements. The

interim phase captures tests of accounting records through inspection, confirmation or

observation, tests of controls, and any substantive tests conducted prior to fiscal year end.

Because audit planning is a continuous and iterative process, quarterly review and interim

procedures may lead to changes in the audit plan (AS 2101). When the audit reaches its final

stage, the audit team may suffer from time compression, which might prevent the team from

properly responding to red flags discovered during the final phase. Consistent with this view,

several studies document a negative influence of time compression on audit quality (McDaniel

1990, Lopez and Peters 2012). Thus, we expect that if auditors properly respond to new evidence

gathered, more time spent prior to the final phase could be associated with better audit quality.

However, prior analytical literature shows that team members work harder as a project nears

completion, possibly implying a more effective output during the final phase of an audit

(Georgiadis 2015).

Eventually, given that prior evidence is mixed on the role of audit planning, quarterly

review, and interim phases on auditor response to increased risk, it remains an empirical question

whether more focus on planning tasks is associated with better audit quality. We state our first

hypothesis in its alternative form:

Page 16: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

14

H1: More audit effort during the earlier phases of the audit is associated with higher audit

quality.

2.2.2 Importance of other experienced audit team members besides the lead partner

PCAOB standards grant an important role to the lead audit partner, suggesting that his or

her expertise and the time spent on the audit can significantly influence audit quality.

Specifically, AS 1201 states that “[t]he engagement partner is responsible for the engagement

and its performance. Accordingly, the engagement partner is responsible for proper supervision

of the work of engagement team members and for compliance with PCAOB standards, including

standards regarding using the work of specialists, other auditors, internal auditors, and others

who are involved in testing controls.” Consistent with the view that lead partners play an

important role in the audit, prior international (non-U.S.) research finds evidence that lead

partners and their expertise influence audit quality (e.g., Gul et al. 2013, Chi et al. 2017). Further,

Garicano (2000)’s model of hierarchy implies that lead partners knowledge and expertise should

matter more than the knowledge and expertise of audit team members that immediately reports to

him or her. Overall, prior research suggests that lead partners should matter, and relatively more

than other audit team members, as a determinant of audit quality.

However, several forces also imply that the role of a lead audit partner may be more

limited than our priors suggest. First, AS 1201 authorizes lead partners to delegate their duties to

other team members. Second, the time spent by lead partners in U.S. audits is limited (Aobdia

2018b), which perhaps explains why recent studies using the U.S. setting surprisingly find little

evidence that lead partner characteristics influence audit quality (Laurion et al. 2016, Aobdia et

al. 2018a, Gipper et al. 2018). These findings suggest that other experienced team members may

have an important role to play in managing and conducting an audit. For example, AS 1220

Page 17: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

15

highlights the role of the EQR to evaluate the significant judgments made and conclusions

reached by the engagement team. The standard indicates that the EQR should have competence,

independence, integrity, and objectivity; in other words, the qualifications of an EQR are similar

to that of a lead partner. Audit managers, senior managers, directors and non-lead partners are

also significantly involved in the day-to-day activities of an audit, and depending on the amount

of delegation, in managing the audit as well. Accordingly, the time other experienced personnel

or the EQR spend on these activities, as well as their expertise, could affect audit quality.

Alternatively, characteristics and time spent by different audit team members may not

influence audit quality. Large audit firms in the U.S. have extensive quality control systems to

ensure high quality audits (e.g., Aobdia 2018c), which likely reduces differences in audit quality

across engagements. Recent evidence also suggests that audit firms in the U.S. may strive to

reach the minimum bar imposed by audit standards, and not necessarily go beyond this level

(Donovan et al. 2015, Aobdia 2018b). Either reason would suggest no association between other

experienced team members time or expertise and audit quality, consistent with prior results

found at the lead partner level in the U.S. Eventually, it remains an empirical question whether

the time and expertise of experienced audit team members, excluding the lead partner, influence

audit quality. We state our hypothesis in its alternative form:

H2: Non-lead partner experienced audit-team members’ time, client expertise, and overall

expertise are positively associated with audit quality.

2.2.3 Cost quality trade-offs

The hypotheses above, thus far, focus on the potential benefits of audit team

characteristics on audit quality. However, cost-quality trade-offs likely exist, such that the cost of

higher audit quality may be borne either by the client or the audit firm. For example, more time

Page 18: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

16

spent during the pre-final phase of the audit could result in the discovery of new risk areas,

requiring additional substantive testing that could be costly to either the audit firm or the client.

The staffing of more experienced team members could lead to higher client costs or lower audit

firm profits due to higher wages paid. Whether it is ultimately the client or the audit firm that

would bear the additional costs depends on clients’ willingness to pay for increased audit quality,

and their bargaining power vis-à-vis the audit firm.

Alternatively, specific engagement characteristics that positively influence audit quality

could also improve audit efficiency, such that we may not observe any trade-offs between audit

quality and the cost of the audit. For example, industry specialist auditors might be able to reduce

audit hours because of learning effects. Consistent with this idea, Bills et al. (2014) find evidence

of cost efficiencies for auditors that specialize in industries with homogenous operations and

complex accounting practices. Gong et al. (2016) find a significant reduction in audit hours

following audit firm mergers in China, without any reduction in audit quality. Eventually, it is an

empirical question whether the engagement characteristics that positively influence audit quality

are associated with increased audit costs, either in the form of higher audit fees or in the form of

reduced audit realization rate. We state our hypothesis in its alternative form.

H3: Some cost-benefit trade-offs may exist in the characteristics associated with better audit

quality.

3. Sample Data Construction and Measures of Engagement Characteristics

3.1 Data construction

Section 104 of SOX instructs the PCAOB to inspect individual engagements of public

accounting firms that audit SEC registered corporations. As part of these inspections, the

Page 19: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

17

PCAOB collects information from audit firms, which includes a breakdown of the engagement

hours across different audit phases and across different personnel that comprise the audit team.

Specifically, hours are split in a matrix of phases and personnel. The phases include the planning

hours, quarterly review hours, interim field hours, final field hours, as well as hours incurred

after the issuance of the audit report. The personnel generally includes the core audit team, such

as the lead partner, EQR, and other experienced team members, and personnel from other

functions such as IS and tax auditors, and specialists. The PCAOB also collects expertise

information about the senior core audit team members. Specifically, the following information is

available for the lead partner, the EQR, and other experienced members of the core audit team (at

the manager level or above): The number of years the individual has worked with the client, in

the client’s industry (industries are as defined by the audit firm), and in a particular role (such as

senior manager or lead partner).

We obtain engagement characteristics data from the inspection documents that result

from this process. To obtain permission to access these data we submitted a research proposal to

the PCAOB describing the nature of our study, the data necessary to conduct the study, and a

summary of related research and proposed research questions. As a condition of data access, the

PCAOB reviewed our research to approve the release of nonpublic information. After receiving

PCAOB approval, a joint effort was undertaken between PCAOB staff and us to arrange the

collected data into a usable database format for the purpose of several analyses. This process

took a significant amount of time, particularly due to the need to identify the different roles in an

audit team (these roles were not standardized in the original documents). The initial dataset

encompasses 3,051 individual inspections for audit engagements inspected between 2006 and

Page 20: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

18

2015 for the eight largest U.S. audit firms (Deloitte & Touche, Ernst & Young, Pricewaterhouse

Coopers, KPMG, BDO, Grant Thornton, Crowe, and McGladrey, renamed RSM in late 2015).

Out of the 3,051 inspections, we eliminate 270 inspections where hours or experience

data are missing. Importantly, we remove all the inspections where hour subtotals are not

reported, because their absence prevents us from checking whether data input or database

compilation mistakes occur. Out of the remaining observations, 175 inspections have errors in

subtotaling the hours across different phases and roles. We eliminate these from the sample. We

also eliminate 28 inspections corresponding to referred work, where the inspected auditor is not

the principal auditor and therefore the core audit team information is missing.

We combine the 2,578 remaining observations that have PCAOB engagement

characteristics with additional PCAOB data about the outcome of the PCAOB inspection (i.e.,

whether a Part I Finding is identified or not), the engagement realization rate (a measure of

profitability of the engagement), as well as Compustat and Audit Analytics data to compute

several dependent and control variables, such as audit fees and whether the fiscal year end

statements are eventually restated. We eliminate 239 observations where additional PCAOB,

Compustat or Audit Analytics data are missing, resulting in a primary sample of 2,339

inspections. Table 1 summarizes the sample selection process.

(Insert Table 1 About Here)

3.2 Sample statistics

Panel A of Table 2 reports the number of observations in the sample, partitioned by

inspection year and whether the auditor is a Big 4 or not. Each of the ten inspection years

Page 21: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

19

roughly comprises 10% of the sample, and the Big 4 observations comprise between 69% and

77% of the yearly number of inspections in the sample.

(Insert Table 2 About Here)

Panel B of Table 2 reports the number of observations in the sample split by whether the

inspection results in a Part I Finding, or the fiscal-year end statements are eventually restated.

Out of the 2,339 observations, 652, or 28% receive a Part I Finding, indicating that the PCAOB

believes the audit opinion was not supported according to relevant audit standards. There are 284

engagements, or 12% of the total, that are eventually restated. Consistent with Aobdia (2018a),

the proportion of restated engagements is higher, at 18%, when the PCAOB identifies a Part I

Finding versus 10% when the PCAOB does not identify a Part I Finding.

3.3 Measures of engagement characteristics

We consider 22 measures of engagement characteristics in our empirical analyses,

described below. Table 3, Panel A reports the descriptive statistics for these variables.

(Insert Table 3 About Here)

3.3.1 Pre-final phase

First, we focus on whether more audit effort during the earlier phases of the audit is

associated with higher audit quality (H1). Our variable of interest is %_Pre-final, equal to the

total lead U.S. office hours spent in planning, quarterly review, and interim phases of the audit

divided by the total lead U.S. office hours. Both numerators and denominators include hours for

all auditors in the U.S. lead office, including the core audit team (staff, senior, managers,

directors and partners), the information system (IS) and tax auditors, and the specialists involved

on the audit; they exclude the time spent by other U.S. locations, or the time spent by non-U.S.

Page 22: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

20

affiliate audit firms.6 Because our focus is on the quality of the work that leads to the audit

opinion, we also exclude hours incurred after the audit opinion is issued from the denominator.7

The mean %_Pre-final is 56%, and the interquartile ranges from 50-63%, indicating that a

significant portion of the audit is conducted during the pre-final phases. Untabulated analyses

indicate that, on average, quarterly review, planning, and interim field hours represent 18%,

10%, and 28% of the total lead U.S. office hours, respectively.

We also consider whether it is the core audit team that matters, or other team members,

such as the IS and tax auditors that are regularly present to assist with audit work. Accordingly,

we partition %_Pre-final between %_Audit_Pre-final, comprised of the total core audit team

hours during the pre-final phase of the audit divided by the total lead U.S. office hours, and

%_NonAudit_Pre-final, comprised of the total lead U.S. office hours spent by the IS auditors, tax

auditors, and specialists during the pre-final phase of the audit, divided by the total lead U.S.

office hours. The mean %_Pre-final (%_NonAudit_Pre-final) is 46% (10%) and the interquartile

ranges from 41-52% (6-14%), indicating that the bulk of the pre-final work is conducted by the

core audit team.

3.3.2 Time spent at each seniority level for the core audit team

We also evaluate the time spent by the lead audit partner, EQR, and other experienced

core audit team members, to understand their roles on audit quality (H2). %_LeadPartner,

%_EQR, and %OtherExperienced are the proportion of total lead U.S. office hours spent by the

lead partner, the EQR, and other experienced team members, with averages of 5%, 1%, and 18%,

6 Untabulated analyses indicate that these hours are small in comparison to the lead U.S. office hours, perhaps

because they are not well populated in our dataset. On average in our sample, other U.S. office and foreign affiliate

hours only represent 2.5% and 7.8% of total hours, respectively. 7 Hours incurred after the issuance of the audit opinion are generally small in comparison with total lead U.S. office

hours, on average 2.3%. They often relate to audit work-papers archiving tasks and, given that the audit opinion is

already issued at this stage, are unlikely to influence audit quality.

Page 23: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

21

respectively. Other experienced team members are comprised of other audit partners (excluding

the lead partner and EQR), directors, senior managers, and managers involved in the core audit

team. Overall, average lead partner and EQR hours represent a small part of the total hours. In

contrast, other experienced team members hours comprise three times the lead partner and EQR

hours combined.

Because we focus on the time spent during the pre-final phases of the audit, we also

consider %_LeadPartner_Pre-final, %_EQR_Pre-final, and %_OtherExperienced_Pre-final,

equal to the time spent by the lead partner, the EQR, and other experienced team members

during the pre-final phases of the audit, divided by the total lead U.S. office hours. The means

are 3%, 1%, and 10%, respectively.

3.3.3 Measures of expertise

We also consider 12 different measures of the core audit team expertise, first based on the

average of the team members, and then partitioned by seniority levels (i.e., lead partner, EQR,

and other experienced) to understand each of their roles on audit quality (H2).

First, we focus on client-specific expertise. Avg_ClientExp is equal to the number of

years of client experience reported by the engagement team, averaged over the lead partner,

EQR, and other experienced core audit team members; the mean is 3.1 years. We subsequently

partition this variable into LeadPartner_ClientExp, EQR_ClientExp, and

OtherExperienced_ClientExp, equal to the number of years of client experience for the lead

partner, the EQR, and the average of the other experienced team members, with reasonably low

means of 3.4, 2.5, and 3.2 years, respectively.8

8 Low client-specific expertise is partly explained by SOX requirements to rotate the lead partner and EQR off a

client every 5 years.

Page 24: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

22

Second, we focus on industry-specific expertise. Avg_IndustryExp is equal to the number

of years of industry experience reported by the engagement team, averaged over the lead partner,

EQR, and other experienced team members; the mean is 12.7 years. Similar to Avg_ClientExp,

we partition Avg_IndustryExp into LeadPartner_IndExp, EQR_IndExp, and

OtherExperienced_IndExp to focus on the industry expertise of the lead partner, the EQR, and

the other experienced core audit team members, with reasonably high means of 18.1, 19.0 and

8.0 years, respectively.

Third, we focus on the overall experience of the experienced core audit team, proxied by

the total years they spent in auditing. Avg_Seniority is equal to the number of years of total audit

experience, averaged over the lead partner, EQR, and other experienced team members; the

mean is 17.3 years. Because our dataset only provides the experience at a given seniority level,

we add 12 years of experience to the partner and director levels, 9 years for senior managers, and

6 for managers, consistent with typical promotion paths within audit firms. Similar to the other

expertise variables, we partition Avg_Seniority into LeadPartner_Seniority, EQR_Seniority, and

OtherExperienced_Seniority, to account separately for the overall experience of the lead partner,

the EQR, and the other experienced team members, with high means of 23.8, 25.5, and 11.4

years, respectively.

Finally, we also consider as an explanatory variable Count_OtherExperienced, equal to

the total number of other experienced team members; the mean is 3.1. We include this variable in

all specifications that focus on experience because average experience is mechanically related to

this variable. In other words, more managers or senior managers on an engagement have the

potential to mechanically reduce the average number of years of overall industry or general

expertise. Further, larger teams could suffer from increased free rider effect (e.g., Holmstrom

Page 25: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

23

1982), and thus not be as effective as smaller audit teams. We confirm in untabulated analyses

that all our results on audit quality are not sensitive to inclusion of this variable.

4. Empirical Analyses

4.1 Research design

To test our hypotheses, we assess whether a particular engagement characteristic is

associated with audit outcomes using the following regression:

Audit Quality or Audit Effort or Audit Profitability or Audit Feesi,t = α + β1.Engagement

Characteristici,t + γ.Controlsi,t +Auditor Fixed Efforts + Year Fixed Effects + Industry

Fixed Effects+ εi,t,

(1)

where the subscripts i and t correspond to clients and years, respectively.

Model (1) is estimated using ordinary least squares (OLS), but the results are

qualitatively unchanged if we use logistic specifications for binary dependent variables. We use

several dependent variables to assess the relationship between a particular engagement

characteristic and audit quality, effort, profitability and cost. We consider two main measures of

audit quality (for tests of H1 and H2) based on the audit process and the audit outcome.

PartIFinding is an indicator variable equal to one if the PCAOB identifies a Part I Finding on the

engagement. A Part I Finding indicates that the work conducted by the engagement team is not

sufficient to support the audit opinion. Thus, the measure is reflective of audit process issues.

One of its major advantages is that the PCAOB has access to extensive nonpublic information

about the engagement, such as the audit workpapers and direct interactions with the audit team,

to determine whether a Part I Finding exists on the engagement (see Aobdia 2018a for more

details). We also use two financial reporting quality measures based on the propensity for a client

to restate its financial statements. Restatement is an indicator variable equal to one if the fiscal-

year end statements are eventually restated, and Material_Restatement is an indicator variable

Page 26: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

24

equal to one when the restatement is considered material and is disclosed in a separate 8-K form

under item 4.02, non-reliance on previously issued financial statements. Material restatements

are more severe than revisions, which are immaterial (e.g. Choudhary, Merkley and Schipper

2017; both are included in the Audit Analytics restatement dataset).

We also consider the potential trade-offs between audit quality and audit effort,

profitability and cost (H3). To measure audit effort, we use Logaudithours, the logarithm of total

hours spent on an engagement (including at the main U.S. office, other U.S. offices, and

international affiliates). To measure audit profitability, we use Realization_Rate, the engagement

realization rate, equal to the total audit fees divided by the hours worked times their

undiscounted billing rates. The higher the realization rate, the higher the profitability of the

engagement. The main advantage of using the audit realization rate is that it represents a direct

measure of auditor profitability. However, the measure comes from a different PCAOB dataset,

is only available for inspections starting in 2009, unavailable for Crowe and RSM audit firms,

and some observations are missing, resulting in significant sample loss.9 To measure the audit

cost to the client, we use Logauditfees, equal to the logarithm of total audit fees charged to the

client from Audit Analytics.

Our main variable of interest, Engagement Characteristic is proxied by the 22

explanatory variables described in section 3.3. Each variable focuses on a different audit

characteristic, such as time spent during the pre-final phases of the audit, or client and industry

knowledge.

9 We consider in untabulated analyses whether this sample restriction could create a power issue for our tests.

Specifically, we restrict the analyses of audit fees and hours to the subsample where realization rate is available. We

generally find that our results on audit fees and hours are similar in the main and restricted samples. However, while

directionally similar, some of the associations with audit hours fall below conventional significance levels in the

restricted sample.

Page 27: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

25

Following prior literature (e.g., Francis et al. 2005, Francis and Yu 2009, Reichelt and

Wang 2010, Aobdia 2018a), we include a set of control variables that have been shown to

influence audit quality and audit fees. These control variables are composed of Logat (the natural

logarithm of the client’s total assets, to control for client size), and Leverage (total debt divided

by total debt plus equity, to control for capital structure). To control for client risk, we also

include Loss (an indicator variable equal to one if earnings before extraordinary items is

negative), BTM (the book to market ratio, equal to the client’s book equity divided by fiscal year

end market value), CFOat (cash flows from operations deflated by beginning assets), StdCFOat

(standard deviation of CFOat, measured from year t − 3 to year t), Salegrowth (year-on-year

percentage increase in the client’s revenues), Weakness (indicator variable if material

weaknesses are reported), and Litigation (indicator variable if the client is in a high litigation

industry). To control for client complexity, we also include ForeignPifo (absolute value of pretax

income from foreign operations divided by the absolute value of pretax income), Geoseg

(number of geographic segments, as per Compustat Segments), and Busseg (number of business

segments, as per Compustat Segments). We also control for two specific characteristics of the

audit: DecYE (a dummy that equals one for fiscal year ending in December), to account for

increased time constraints during the busy season, and FirstYear (indicator variable equal to one

for first year audits of new clients), to account for the increased difficulty of first year client

audits (e.g., Bell et al. 2015). Given that the PCAOB is likely to identify additional internal

control over financial reporting (ICFR) audit deficiencies for integrated audits of financial

statements and ICFR, we also include Integrated_Audit (an indicator variable equal to one when

the audit is an integrated audit of financial statements and ICFR). We also incorporate year, audit

firm, and Fama French 12 industry group fixed effects, cluster standard errors at the client level,

Page 28: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

26

and winsorize all continuous variables at the 1st and 99th percentiles to limit the influence of

outliers in the specifications. Detailed variable definitions are provided in the appendix.

4.2 Descriptive statistics

We present several descriptive statistics in Panel A of Table 3. In terms of dependent

variables, an average engagement has a 28% likelihood to receive a Part I Finding, a 12% (4%)

likelihood to be restated (have a material restatement), and commands audit fees of $1.5M,

corresponding to 7,300 hours of audit work. The average engagement realization rate is 49%,

indicative that auditors give significant discounts from the official billing rates.

In terms of control variables, our sample contains 77% of integrated audits of financial

statements and ICFR; 73% of the client-years with a December fiscal year end, and 25% that

report losses. Average client assets are $1.6B.

4.3 Sample Partition based on restatements and Part I Findings

Panel B of Table 3 reports the means of the subsamples partitioned on whether the

financial statements are eventually restated as well at t-tests of differences across the two

samples. Several interesting patterns emerge. Engagements of non-restating clients spend more

time during the pre-final phase of the audit as compared with non-restating clients (56.6% versus

53.7%; p<1%). Non-restating engagements also have more time spent by the other experienced

team members relative to restating engagements (18.3% versus 17.6%; p<10%), whereas the

time spent by the lead partner is similar across the two (4.8%; p>10%). The average client

expertise is also higher for non-restating engagements relative to restating ones (3.15 versus 2.84

years; p<1%), partly driven by higher expertise of the other experienced team members (3.3

versus 2.8 years; p<1%). While lead partner client expertise is higher for non-restating versus

Page 29: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

27

restating engagements (3.5 versus 3.2 years; p>10%), the difference is not statistically significant

at conventional levels. The overall team’s industry expertise is also higher for non-restating

versus restating engagements (12.8 versus 12.1 years; p<5%); these differences are partly driven

by higher EQR industry experience (19.2 versus 17.9 years; p<5%), and other team members

industry experience (8.0 versus 7.5 years; p<5%). Experienced team members’ seniority, other

dependent variables such as audit fees, hours, and realization rate, and control variables do not

show statistical differences between restating and non-restating groups (p>10%).

Overall, these preliminary univariate results are consistent with several ideas. First, more

time spent during the pre-final phase of the audit is beneficial to audit quality (H1). Second, both

time spent by other experienced team members, and their expertise, appear to be beneficial to

audit quality (H2). Our evidence is also consistent with a positive influence on audit quality of

the other experienced team members’ client-specific and industry-specific expertise. Finally,

consistent with Laurion et al. 2016, Aobdia et al. 2018a, and Gipper et al. 2018, we fail to find

any evidence that specific lead partner characteristics, such as time spent on the audit and

expertise, are associated with audit quality.

Panel C of Table 3 reports an analysis similar to Panel B, where the sample partition is

whether the engagement receives a Part I Finding. Some results in Panel C echo those in Panel

B. For example, engagements without Part I Findings spend more time during the pre-final phase

of the audit than engagements with Part I Findings. However, several results run opposite to

those of Panel B. For example, average industry experience is higher for engagements that

receive Part I Findings. We note that it might be more difficult than in Panel B to interpret these

univariate results without controlling for specific engagement characteristics, because the

engagements that receive Part I Findings appear different from those that do not receive them. In

Page 30: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

28

particular, engagements with Part I Findings in our sample are larger, with average assets of

$1.9B, compared with $1.5B for engagements without Part I Findings. Likely reflecting their

larger sizes, they are also comprised of integrated audits to a greater extent (85.9%), as compared

with 73.4% for engagements without Part I Findings. These differences are likely driven by the

PCAOB identifying additional ICFR audit deficiencies for engagements with integrated financial

statements and ICFR audits. Overall, the sample partition in Panel C of Table 3 confirms the

need to control in subsequent multivariate analyses for company-specific characteristics that may

influence both specific engagement characteristics and the probability to receive a Part I Finding.

4.4 Pre-final phase of the audit

Panel A of Table 4 reports multivariate tests of H1, whether more audit effort during the

earlier phases of the audit is associated with better audit quality. Our variable of interest is

%_Pre-final, the proportion of time spent during the pre-final phases of the audit. We estimate

model (1) with %_Pre-final and find negative associations with PartIFinding, Restatement, and

Material_Restatement (p<10%), suggesting that more time spent during the pre-final phases of

the audit positively influences audit quality (H1). A one standard-deviation decline in %_Pre-

final, (10%), is associated with a reduction in the probability of a Part I Finding of 1.8% and a

restatement of 2.5%. These economic effects are reasonably high in comparison with the average

Part I Finding and restatement rates in the sample of 28% and 12%, respectively.

We also find that the increased audit quality comes with a cost to the client, evidenced by

a positive association between %_Pre-final and Logauditfees, driven by a positive association

between %_Pre-final and Logaudithours. In terms of economic magnitude, an increase of one

standard-deviation of %_Pre-final is associated with a non-negligible 6.5% increase in audit

Page 31: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

29

fees.10 We do not find an association between %_Pre-final and audit profitability proxied by

Realization_Rate.

(Insert Table 4 About Here)

Next, we separately focus on the pre-final time spent by the core audit team, in

comparison with others involved in the process (IS and tax auditors and specialists). We partition

%_Pre-final into %_Audit_Pre-final and %_NonAudit_Pre-final in Panel B of Table 4. We find

that our initial results in Panel A are generally driven by the time spent by the core audit team in

the pre-final phase, with the exception of material restatements, which are influenced by both the

core audit team and the other auditors involved in the process. We also find positive associations

between both %_Audit_Pre-final and %_NonAudit_Pre-final with Logauditfees and Logaudit

hours, but not realization rate, suggesting the costs borne by the client are driven by both the

audit and non-audit personnel. In subsequent analyses, we focus on the core audit team given

that our data on the non-core audit team are more limited, and the quality effects appear

generally stronger for the core audit team.

In untabulated analyses, we also consider whether any particular pre-final phase of the

audit (quarterly review, planning, and interim field) matters more from an audit and financial

reporting quality standpoint. We partition %_Audit_Pre-final into these three components and

repeat our analyses. We find negative associations between each of the components of pre-final

hours and Restatement and Material_Restatement. However, these associations are insignificant

at conventional levels (p>10%). The main exception is a negative association between the

proportion of planning hours and PartIFinding, which suggests a larger role of the planning

10 This number is computed as e(0.633×0.10)-1.

Page 32: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

30

phase compared with the other ones, consistent with some of the arguments in Davidson and Gist

(1996).

Given that we find a positive association between %_Pre-final and Logaudithours in

Table 4, we also consider whether it is the increased audit hours, or the time spent by the auditor

during the pre-final phase that ultimately influence audit quality.11 We replicate our analyses on

audit quality adding audit hours as a control variable and continue to find robust results in the

relationship between %_Pre-final and audit quality (untabulated).

4.5 Time spent by seniority level

We focus in Panel A of Table 5 on the proportion of hours spent by the experienced core

audit team members, including the lead engagement partner (%_LeadPartner), the EQR

(%_EQR), and the other experienced team members (%_OtherExperienced). We find negative

associations between %_EQR and Restatement (p<10%), and between %_OtherExperienced and

PartIFinding (p<5%). The other associations with audit quality are insignificant (p>10%). These

associations provide some evidence that more time spent by more experienced team members

other than the lead partner improve audit quality (H2). However, we do not find any positive

association between the time spent by the lead partner and audit quality; in fact we note a weak

positive association between %_LeadPartner and Material_Restatement, suggesting that too

much time spent by the lead partner on the audit could be detrimental in some instances, perhaps

because it represents an inability to properly delegate to others.

The relationships between time spent by audit team members and cost measures are

mixed. The increased time spent by other experienced team members (%_OtherExperienced )

appears costly to the client with higher audit fees (Logauditfees; p<1%) and more audit hours

11 For example, Caramanis and Lennox (2008) find a positive association between audit hours and audit quality.

Page 33: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

31

(Logaudithours; p<1%), and no effect to auditor profitability (Realization Rate; p>10%). An

increase of one standard-deviation of %_OtherExperienced is associated with a reasonably high

9.4% increase in audit fees. In contrast, the additional time spent by the lead partner and EQR

seems to result in more efficient audits or lower audit hours (Logaudithours; p<1%), with

benefits passed to the client in terms of lower fees (Logauditfees; p<1%).

(Insert Table 5 About Here)

Panel B of Table 5 focuses on the time spent by experienced audit team members during

the pre-final phase of the audit. Our results are generally similar to those in Panel A, highlighting

the importance of the pre-final phase of the audit in terms of audit and financial reporting

quality.

4.6 Audit team members’ client expertise

We focus on the role of client-specific expertise in Table 6. Panel A reports the results

with the average client expertise for all experienced team members (Avg_ClientExp). We find

negative associations between Avg_ClientExp and Restatement as well as Material_Restatement

(p<5%). This is consistent with more client-specific expertise positively influencing audit

quality measured as financial reporting quality. A one standard-deviation increase in

Avg_ClientExp (1.33 years) is associated with a 2% reduced probability of restatements. This is

reasonably large given that 12% of the financial statements are eventually restated in our sample.

Focusing on the number of other experienced team members, we do not find any association

between Count_OtherExperienced and audit quality (p>10%); in other words, we fail to find

free-riding issues within larger audit teams.

Page 34: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

32

Next, we evaluate whether increased average client expertise comes with trade-offs or

efficiency benefits to the auditor or the client. We find a negative association between

Avg_ClientExp and Logaudithours (p<1%), as well as positive associations with

Realization_Rate and Logauditfees (p<10%). Overall, these results suggest that increased client

expertise has a positive influence on financial reporting quality, but also results in improved

audit efficiency that improves audit profitability. A one standard-deviation increase in

Avg_ClientExp is associated with a 2.73% increase in the audit realization rate. Nevertheless,

increased client expertise comes with a cost to the client in the form of higher audit fees

(p<10%).

(Insert Table 6 About Here)

We focus in Panel B of Table 6 on the separate influence of client expertise for the lead

partner (LeadPartner_ClientExp), the EQR (EQR_ClientExp), and the other experienced team

members (OtherExperienced_ClientExp). Consistent with Gipper et al. (2018), we do not find

any association between LeadPartner_ClientExp and audit quality. We also fail to find any

association between EQR_ClientExp and audit quality (p>10%), but we do find evidence of a

negative association between OtherExperienced_ClientExp and Restatement (p<1%). These

results are consistent with the results of Panel B of Table 5 that suggest that other experienced

team members characteristics are associated with audit quality measured as financial reporting

quality.

In terms of audit profitability, we find positive associations for all the measures of

expertise with Realization_Rate. Extending Gipper et al. (2018) who focus only on lead partner

characteristics, we not only find a negative (positive) association between lead partner client

experience and audit hours (audit profitability), we also find similar directional effects from

Page 35: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

33

other experienced team members’ and EQR client experience, suggesting that more client

experience improves audit efficiency and profitability. A one standard-deviation in

OtherExperienced_ClientExp (1.75 years) is associated with a 2.1% decrease in the probability

of restatements, and a 2% increase in realization rate. We also find a positive association with

Logauditfees, suggesting that client-specific experience of other experienced team members

comes with a cost to the client.

We replace the continuous variables of interest of Panel B on client expertise with

indicator variables that equal one when these variables are above their medians in Panel C; the

new variable names have >median appended to reflect the change in definition. We generally

find similar results as in Panel B. However, the association between

OtherExperienced_ClientExp> median and Restatement falls slightly below conventional

significance levels. Instead, we find a negative association between

OtherExperienced_ClientExp>median and PartIFinding. We also find a surprising, but weak,

positive association between LeadPartnerClientExp>median and PartIFinding.

Collectively, our results are consistent with the client specific expertise of other

experienced audit team members playing an important role in explaining financial reporting

quality and audit profitability. Importantly, because these team members are not subject to SOX

mandatory rotation requirements, our results suggest that an audit firm might be able to increase

both audit quality and profitability at the same time, by increasing the level of client expertise

among senior professionals on the audit team.

In additional analyses, we also consider whether non-linearities exist in the relationships

identified above. In particular, it is possible that some of the effects of client specific expertise

taper off as it increases. Following Davidson and Gist (1996), we add in untabulated analyses

Page 36: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

34

squared expertise terms in each of the regressions conducted in Table 6. We generally find

negative coefficients on the squared expertise terms in the regression of Realization_Rate,

consistent with the influence of client expertise on audit profitability tapering off as it increases.

We also find negative and positive associations between OtherExperienced_ClientExp and

OtherExperienced_ClientExp2 with PartIFinding, suggesting an initial positive influence of

other experienced team member’s expertise on audit process quality that tapers off. Further, the

association between other experienced team member’s client expertise and audit fees is

moderated when the number of years increase, evidenced by a negative association between

OtherExperienced_ClientExp2 and Logauditfees. Overall, these results are consistent with one

additional year of client specific expertise having a larger influence when client-specific

expertise is lower.

4.7 Audit team members’ industry expertise

In Table 7 we evaluate the role of industry expertise, as defined and reported by the audit

firms. Panel A reports the results when we consider the average industry expertise of the

experienced team members (Avg_IndustryExp). We find a negative association with

Material_Restatement (p<10%), suggesting industry expertise is positively associated with audit

quality. In terms of economic significance, one standard-deviation increase in Avg_IndustryExp,

4.67 years, is associated with a 0.9% decrease in the probability of material restatements, to be

compared with an average probability of 4% in our sample.

(Insert Table 7 About Here)

Panel B decomposes the average industry expertise among the lead partner

(LeadPartner_IndExp), the EQR (EQR_IndExp), and other experienced team members

(OtherExperienced_IndExp). Consistent with Aobdia et al. (2018a), we do not find an

Page 37: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

35

association between LeadPartner_IndExp and financial reporting quality, but we do find a

positive association with Logauditfees (p<10%).12 Consistent with our prior results, we find

negative associations between EQR_IndExp, OtherExperienced_IndExp, and

Material_Restatement (p<10%).

We replace the continuous variables of interest in Panel B on industry expertise with

indicator variables in Panel C when these variables are above their respective medians. We

append the names with >median to reflect the change in definition. We generally find similar

results as in Panel B. However, we also find a negative association between

OtherExperienced_IndExp>median and Restatement (p<5%), consistent with a positive effect of

other experienced team members’ industry expertise on audit quality. We also find a surprising

positive relationship between LeadPartnerClientExp>median and PartIFinding, similar to the

one of Panel C of Table 6 (p<1%), perhaps because lead partners become more overconfident or

document their audit procedures less as they gain significant expertise.

Similar to section 4.6, we also consider in untabulated analyses whether non-linearities

exist in the industry expertise relationship with outcome variables, by adding squared industry

expertise terms in the regressions. We find negative associations between

OtherExperienced_IndExp and positive associations between OtherExperienced_IndExp2 with

PartIFinding and Restatement, confirming the importance of the first few years of industry

expertise for other experienced team members.

4.8 Audit team members’ seniority

12 While we do not find any association between LeadPartner_IndExp and Logaudithours in the main sample, we

find a positive association when restricting the sample to where realization rate is available. The coefficient and

statistical significance of the association between LeadPartner_IndExp and Logauditfees also increases when

restricting the sample. Both results are consistent with Aobdia et al. (2018a). We find similar associations with audit

fees and hours and industry experience when focusing on the EQR in the subsample where realization rate is

available.

Page 38: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

36

We focus in Table 8 on experienced team member’s overall expertise proxied by the

number of years they worked in auditing. Panel A reports the results for the average years of

seniority in the team (Avg_Seniority). We do not find any association between Avg_Seniority and

audit quality (p>10%).

(Insert Table 8 About Here)

Next, we focus on the seniority of the lead partner (LeadPartner_Seniority), the EQR

(EQR_Seniority), and other experienced team members (OtherExperienced_Seniority) separately

in Panel B of Table 8. We find negative associations between EQR seniority and PartIFinding

(p<10%) as well as Material_Restatements (p<5%). Overall, these results are consistent with the

spirit of PCAOB standard AS 1220 which instructs the EQR to “perform an evaluation of the

significant judgments made by the engagement team and the related conclusions reached in

forming the overall conclusion on the engagement,” implying that significant general expertise is

required for an individual to properly fulfill the EQR role. We replace the continuous variables

of interest of Panel B on general expertise with indicator variables when these variables are

above their medians in Panel C. We append the names of the variables with >median to reflect

the changes in definition. We generally find similar results as in Panel B.

Collectively, our results are consistent with our hypotheses. First, time spent during the

pre-final phase of the audit is positively associated with audit quality (H1). Second, time spent

and expertise of experienced core audit team members other than the lead partner is also

positively associated with audit quality (H2). Third, while some particular engagement

characteristics come with a cost to the client, such as more time spent during the pre-final hours,

there is usually no negative impact on audit profitability proxied by the audit realization rate

(H3). Client-specific experience, on the other hand, tends to improve audit efficiency and audit

Page 39: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

37

profitability. We also confirm the results in prior literature that find little influence of specific

lead partner characteristics on audit quality in the U.S. (Laurion et al. 2016, Aobdia et al. 2018a,

Gipper et al. 2018).

4.9 Selection bias concerns

Our sample is based on audits selected for inspection by the PCAOB and, given that the

PCAOB does not randomly select engagements for inspection (Hanson 2012), our results may

not generalize to the full population. Prior research evaluates whether PCAOB-inspected samples

exhibit selection bias and generally finds evidence suggesting that selection bias concerns are not

severe (e.g., Aobdia 2018a, Aobdia et al. 2018b, Choudhary et al. 2018a and b). Further, many

of our explanatory variables of interest during the sample period are not available to the PCAOB

at the time of selection of engagements for inspection. Thus, selection based on these variables is

unlikely. However, the PCAOB has access to lead-partner level information prior to selecting

engagements (e.g., Aobdia et al. 2018a). Therefore, it is important to compare our results of lead

partner effects with those of prior literature, particularly Laurion et al. (2016), Aobdia et al.

(2018a), and Gipper et al. (2018), who analyze much broader samples, not restricted to PCAOB

inspected engagements. The lead partner results are generally consistent with these studies,

suggesting that our results may generalize to the full population.

Given that prior literature typically discusses selection bias issues as an omitted variable

concern (Heckman 1979), we apply a procedure, following Frank (2000) and Larcker and

Rusticus (2010), to estimate the amount of correlation an omitted variable must have with both

independent and dependent variables to overturn our reported results. For each variable of

interest, we compute the impact threshold for a confounding variable (ITCV), which is the

lowest product of the partial correlation between the dependent variable and the omitted variable,

Page 40: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

38

and the partial correlation between the explanatory variable of interest and the omitted variable

that makes the coefficient statistically insignificant. Theoretically, a high ITCV indicates that the

OLS results are robust to omitted variable concerns. Following Larcker and Rusticus (2010), we

also compute a benchmark impact for the size of likely correlations involving the hypothetical

(and unobserved) omitted variables, based on the control variables in our sample.

(Insert Table 9 About Here)

Table 9 reports the ITCV for each explanatory variable that has shown statistical

significance in the prior analyses. We also consider the control variables that have higher impact

(in absolute value) than the explanatory variable. In other words, an omitted variable would need

to have an impact comparable to the one of these control variables to overturn our results. For the

audit quality variables, our results indicate that the ITCV of the explanatory variables is high. In

most specifications, none of the control variables, including company size (Logat), has an impact

sufficiently large to overturn the results. For the audit hours and fees variables, only company

size (Logat) generally has a higher impact than our variables. Given that company size predicts a

significant proportion of audit fees and audit hours (over 60%), this is not surprising, and

suggests that the ITCV for the explanatory variables is high. Collectively, out of the 54

coefficients considered, 28 (17), and [9] have no control variable, (Logat only), or [Logat and at

least one other control variable, exactly one for 7 out of 9 cases and more than one for the 2

remaining], respectively, with impact sufficiently large to overturn the results. Overall, these

analyses suggest that our results are reasonably robust to omitted variable concerns.

4.10 Additional analyses

We consider in supplemental analyses the nature of the Part I Finding, in particular

whether the audit deficiency identified by the PCAOB is a financial statements or ICFR

Page 41: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

39

deficiency. We also use the count of Part I Findings identified by the PCAOB in each

engagement as an alternative measure of audit process quality. Untabulated analyses indicate that

our results are generally qualitatively unchanged, in comparison with PartIFinding, when we use

these alternative dependent variables. However, the proportion of time spent by other

experienced team members (%_OtherExperienced) is negatively associated with financial

statements Part I Findings and not with ICFR Part I Findings. We also find a negative association

between EQR time (%_EQR) and ICFR Part I Findings, but not with financial statements Part I

Findings. These results are consistent with the ICFR assessment being challenging and requiring

significant expertise (e.g., Hoitash et al. 2008, Bedard and Graham 2011, Kinney et al 2013), and

highlight an important role for the EQR in terms of audit process quality.

We also consider whether specific roles within the other experienced team members

matter more. In untabulated analyses, we decompose the other experienced team members into

two groups: other partners or directors, and senior managers or managers. We generally find that

the effect of client-specific and industry expertise are more pronounced for senior managers or

managers than for other partners or directors. However, we caution about over-interpreting this

evidence for the following reason: Less than half of the engagements in the sample have an audit

director or other audit partner, whereas most engagements have a manager or senior manager.

Thus, the weak results on director could be driven by sample loss (these results are conditional

on having a director or other partner present on the engagement). Consistent with the idea that

other partners and directors matter from an audit quality standpoint, we find in untabulated

analyses that engagements that have a higher proportion of director or other partner time have

better audit quality.

Page 42: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

40

We also evaluate the experienced core audit team’s gender. Prior research documents in

experimental and archival settings differences in team performance depending on the team

gender (e.g., Apesteguia et al. 2012, Huang and Kisgen 2013). In untabulated analyses, we count

the number of experienced team members, and consider the gender of each of them, based on

their first name.13 We do not find any association between team gender and any of our measures

of audit and financial reporting quality.

We also consider the introduction of new PCAOB standards. Specifically, we focus on

AS7 (renamed AS 1220 in 2015), which clarified the role of the EQR. We interact EQR related

variables (in Panel B of Tables 5 to 8) with an indicator variable for fiscal years beginning after

12/15/2009, when the standard became effective. In untabulated analyses, we find that several of

our audit quality results on EQR are driven by the post AS 1220-audits, evidenced by significant

F-tests on the sum of the EQR variable and its interaction with post 12/15/2009 fiscal years, but

insignificant coefficients on the main EQR variable, corresponding to pre-AS 1220 audits.

However, the interaction term is generally insignificant. Overall, these results provide weak

evidence that AS 1220 strengthened, or at least did not weaken, the role of the EQR on audit

quality.

We also focus on PCAOB release 2010-004, issued on August 5, 2010, that adopted eight

auditing standards related to the auditor's assessment of and response to risk and superseded six

of the Board's interim auditing standards and related amendments to PCAOB standards. This

standard may have changed the role of pre-final hours in the audit. We interact pre-final hours in

13 We use the database of names from the Social Security Administration of all babies born in the US between 1942-

1990. This database identifies the gender associated with each name for all names used at least 5 times. Because

some names are used for both males and females, we assign a given gender to a particular name if more than 90% of

occurrences of the name are used for a particular gender. For more ambiguous names, we exclude these names from

our analysis.

Page 43: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

41

Table 4 with an indicator variable for fiscal years beginning after 12/15/2009, when the standard

became effective. In untabulated analyses, we find a positive (negative) coefficient on the

interaction of this indicator variable and %_Pre-final with Restatement (PartIFinding) as the

dependent variable. This appears to indicate that the changes helped auditors better comply with

audit standards, but financial reporting quality did not eventually benefit.

5. Conclusion

In this study, we evaluate a variety of characteristics regarding the composition of the

audit and audit team to understand whether they influence audit quality. We consider both audit

process quality, measured by regulatory inspection findings, and financial reporting quality,

measured by the propensity of a client to restate the financial statements, as measures of audit

quality. We also explore the trade-offs between increased audit quality and the cost to the client

or the auditor in the form of reduced profitability versus improvements in audit efficiency.

We find that more time spent during the pre-final phase of the audit is associated with

better audit quality. The core audit team drives most of these results. We also find that greater

time spent by other experienced team members is associated with better audit quality. However,

these engagement characteristics come with increased costs to the client, reflecting trade-offs

between audit quality and the cost of the audit to the client. We also find evidence consistent

with EQR and other experienced team members’ expertise, including client specific expertise,

industry expertise, and overall expertise, positively influencing audit quality. Confirming prior

literature on audit partners in the U.S. (Laurion et al. 2016, Aobdia et al. 2018a, Gipper et al.

2018), we fail to find an influence of lead partner time and experience characteristics on audit

quality.

Page 44: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

42

Overall, our study is a first attempt at opening the black box of the role of audit processes

and personnel below the lead partner, and responds to recent regulatory initiatives that strive to

better understand the determinants of audit quality. Our results suggest that auditors, audit

regulators, and academics should focus more on non-lead partner level members in audit teams,

to better understand their roles from an audit performance standpoint. While prior audit research

has made significant progress at understanding the role of lead partners on audit performance

(e.g., Lennox and Wu 2018), and finds in several instances that individual partner characteristics

influence audit quality, our study raises the prospect that some of these results might be driven

by the quality of other team members comprising the engagement team, rather than the lead

partners themselves, in case lead partners tend to select specific team members. Future research

might have the opportunity to explore specific engagement characteristics in more details if audit

regulators, such as the PCAOB, or audit firms themselves, directly or through industry

associations, collect more engagement specific data and share them with academic researchers.

Page 45: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

43

References

Allen, R.D., D.R. Hermanson, T.M. Kozloski, and R.J. Ramsay. 2006. Auditor risk assessment:

Insights from the academic literature. Accounting Horizons 20 (2): 157-177.

Aobdia, D., Lin, C-J., and R. Petacchi. 2015. Capital market consequences of audit partner

quality. The Accounting Review 90: 2143–2176.

Aobdia, D., S. Siddiqui, and A. Vinelli. 2018a. Does engagement partner expertise matter?

Evidence from the U.S. operations of the Big 4 audit firms. Northwestern University

working paper.

Aobdia, D., P. Choudhary, and G. Sadka. 2018b. Do auditors correctly identify and assess

internal control deficiencies? Evidence from the PCAOB data. Northwestern University

working paper.

Aobdia, D. 2018a. Do practitioner assessments agree with academic proxies for audit quality?

Evidence from PCAOB and internal inspections. PCAOB and Northwestern University

working paper.

Aobdia, D. 2018b. The impact of the PCAOB individual engagement inspection process –

preliminary evidence. The Accounting Review 93 (4): 53-80.

Aobdia, D. 2018c. The economic consequences of audit firms’ quality control system

deficiencies. PCAOB and Northwestern University working paper.

Apesteguia, J., G. Azmat, and N. Iriberri. 2012. The impact of gender composition on team

performance and decision making: Evidence from the field. Management Science 58 (1):

78-93.

Balsam, S., J. Krishnan, and J. Yang. 2003. Auditor industry specialization and earnings quality.

Auditing: A Journal of Practice & Theory 22: 71-97.

Bedard, J.C. and K. Johnstone. 2004. Auditors’ assessments of and responses to earnings

management risk and corporate governance risk. The Accounting Review 79 (2): 277–304

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

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

Bell, T., 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.

Bills, K., D.C. Jeter, and S.E. Stein. 2014. Auditor industry specialization and evidence of cost

efficiencies in homogeneous industries. The Accounting Review 90 (5): 1721-1754.

Cameran, M., A. Ditillo, and A. Pettinicchio. 2017. Audit team attributes matter: How diversity

affects audit quality. European Accounting Review, forthcoming.

Cameran, M., D. Campa, and J. Francis. 2018. Audit effects of accounting firm organization

levels. Working paper available at:

https://papers.ssrn.com/sol3/papers.cfm?abstract_id=3157562

Canadian Public Accountability Board. 2018. Audit quality indicators. Final report.

Center for Audit Quality (CAQ). 2014. CAQ approach to audit quality indicators.

Page 46: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

44

Caramanis, C., and C. Lennox. 2008. Audit effort and earnings management. Journal of

Accounting and Economics 45 (1): 116-138.

Chi, W., L.A. Myers, T.C. Omer, and H. Xie. 2017. The effects of audit partner pre-client and

client-specific experience on audit quality and on perceptions of audit quality. Review of

Accounting Studies 22 (1): 361-391.

Choudhary, P., K. Merkley, and K. Schipper. 2017. Qualitative characteristics of financial

reporting errors deemed immaterial by managers. University of Arizona, Indiana

University and Duke University working paper.

Choudhary, P., K. Merkley, and K. Schipper. 2018a. The Last Chance to Improve Financial

Reporting Quality: Evidence from Recorded and Waived Audit Adjustments. University

of Arizona, Indiana University and Duke University working paper.

Choudhary, P., K. Merkley, and K. Schipper. 2018b. Direct Measures of Auditors’ Quantitative

Materiality Judgments: Properties, Determinants and Consequences for Audit

Characteristics and Financial Reporting Reliability. University of Arizona, Indiana

University and Duke University working paper.

Christensen, B.E., S.M. Glover, T.C. Omer, and M.K. Shelley. 2016. Understanding audit

quality: insights from audit professionals and investors. Contemporary Accounting

Research 33 (4): 1648-1684.

Davis, L., D. Ricchiute, and G. Trompeter. 1993. Audit effort, audit fees, and the provision of

nonaudit services to audit clients. The Accounting Review 68 (1): 135–150.

Davidson, R.A., and W.E. Gist. 1996. Empirical evidence on the functional relation between

audit planning and total audit effort. Journal of Accounting Research 34 (1): 111-124.

DeFond, M.L., and J. Zhang. 2014. A review of archival auditing research. Journal of

Accounting and Economics 58 (2-3): 275-326.

Financial Reporting Council. 2008. The audit quality framework. February.

Francis, J.R., and M.D. Yu. 2009. Big 4 office size and audit quality. The Accounting Review 84

(5): 1521-1552.

Frank, K.A. 2000. Impact of a confounding variable on a regression coefficient. Sociological

Methods & Research 29 (2): 147-174.

Fukukawa, H., T.J. Mock, and A. Wright. 2006. Audit programs and audit risks: A study of

Japanese practice. International Journal of Auditing 10 (1): 41-65.

Garicano, L. 2000. Hierarchies and the organization of knowledge in production. Journal of

Political Economy 108 (5): 874-904.

Garicano, L., and T.N. Hubbard. 2007. Managerial leverage is limited by the extent of the

market: Hierarchies, specialization, and the utilization of lawyers’ human capital. Journal

of Law and Economics 50 (1): 1-43.

Georgiadis, G. 2015. Projects and team dynamics. Review of Economic Studies 82: 187-218.

Gipper, B., L. Hail, and C. Leuz. 2018. On the economics of audit partner tenure and rotation:

Evidence from PCAOB data. NBER working paper No. 24018.

Page 47: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

45

Gong, Q., O.Z. Li, Y. Lin, and L. Wu. 2016. On the benefits of market consolidation: Evidence

from merged audit firms. The Accounting Review 91 (2): 463-488.

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.

Hackenbrack, K., and W. R. Knechel. 1997. Resource allocation decisions in audit engagements.

Contemporary Accounting Research 14 (3): 481–499.

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.

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

February 26, 2018).

Heckman, J.J. 1979. Sample selection bias as a specification error. Econometrica 47 (1): 153-

161.

Huang, J., and D.J. Kisgen. 2013. Gender and corporate finance: Are male executives

overconfident relative to female executives? Journal of Financial Economics 108 (3):

822-839.

Hoitash, R., U. Hoitash, and J.C. Bedard. 2008. Internal control quality and audit pricing under

the Sarbanes-Oxley Act. Auditing: A Journal of Practice & Theory 27: 105-126.

Holmstrom, B. 1982. Moral hazard in teams. Bell Journal of Economics 13: 324-340.

International Auditing and Assurance Standards Board. 2013. A framework for audit quality.

New York: The International Federation of Accountants.

Johnstone, K. and J. Bedard. 2001. Engagement planning, bid pricing, and client response in the

market for initial attest engagements. The Accounting Review 76 (2): 199–221.

Kinney, W. R., R.D. Martin, and M.L. Shepardson. 2013. Reflections on a decade of SOX

404(b) audit production and alternatives. Accounting Horizons 27 (4): 799-813.

Knechel, W.R., P. Rouse, and C. Schelleman. 2009. A modified audit production framework:

Evaluating the relative efficiency of audit engagements. The Accounting Review 84 (5):

1607-1638.

Knechel, W.R., A. Vanstraelen, and M. Zerni. 2015. Does the identity of engagement partners

matter? An analysis of audit partner reporting decisions. Contemporary Accounting

Research 32 (4): 1443-1478.

Krishnan, G. 2003. Does Big 6 auditor industry expertise constrain earnings management?

Accounting Horizons 17: 1-16.

Larcker, D.F., and T.O. Rusticus. 2010. On the use of instrumental variables in accounting

research. Journal of Accounting and Economics 49 (3): 186-205.

Laurion, H., A. Lawrence, and J.P. Ryans. 2017. U.S. audit partner rotations. The Accounting

Review 92 (3): 209-237.

Libby, R., J. Artman, and J. Willingham. 1985. Process susceptibility, control risk, and audit

planning. The Accounting Review 60 (2): 212–230.

Page 48: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

46

Lennox, C., and X. Wu. 2017. A review of the archival literature on audit partners. Accounting

Horizons Forthcoming.

Lopez, D.M., and G.F. Peters. 2012. The effect of workload compression on audit quality.

Auditing: A Journal of Practice & Theory 31 (4): 139-165.

Maletta, M., and T. Kida. 1993. The effect of risk factors on auditors’ configural information

processing. The Accounting Review 68 (3): 681–691.

McDaniel, L. 1990. The effects of time pressure and audit program structure on audit

performance. Journal of Accounting Research 28 (2): 267-285.

O’Keefe, T.B., D.A. Simunic, and M.T. Stein. 1994. The production of audit services: Evidence

from a major public accounting firm. Journal of Accounting Research 32 (2): 241-261.

Palmrose, Z.-V. 1986. Audit fees and auditor size: further evidence. Journal of Accounting

Research 24 (1): 97–110.

Palmrose, Z.-V. 1989. The relation of audit contract type to audit fees and audit hours. The

Accounting Review 64 (3): 488–499.

PCAOB. 2015. Concept release on audit quality indicators. PCAOB Release No. 2015-005. July

1.

Reichelt, K., and D. Wang. 2010. National and office-specific measures of auditor industry

expertise and effects on audit quality. Journal of Accounting Research 48: 647–686.

Simunic, D. 1980. The pricing of audit services: theory and evidence. Journal of Accounting

Research 18 (1): 161-190.

Zimbelman, M. F. 1997. The effects of SAS No. 82 on auditors’ attention to fraud risk factors

and audit planning decisions. Journal of Accounting Research 35 (Supplement): 75–97.

Page 49: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

47

Appendix: Variable Definitions

Variable Definition

Dependent Variables

PartIFinding An indicator variable equal to one if the inspection of the engagement resulted

in at least one Part I finding, from PCAOB proprietary data.

Restatement An indicator variable equal to one if the fiscal year-end financial statements

are subsequently restated as noted by Audit Analytics.

Material_Restatement An indicator variable equal to one if the fiscal year-end financial statements

are subsequently restated and the disclosure is made in a separate form 8-K

under item 4.02 (non-reliance on previously issued financial statements) as

noted by Audit Analytics.

Logaudithours The logarithm of the total audit hours spent on the engagement (including

other U.S. locations and non-affiliates hours), from PCAOB proprietary data.

Realization_Rate Total audit fees divided by total audit hours times their individual

undiscounted billing rates (which depend on an engagement team member's

seniority). A measure of audit profitability. Obtained from PCAOB proprietary

data but available only for most engagements inspected in 2009 and thereafter.

Logauditfees The logarithm of the engagement audit fees, from Audit Analytics.

Explanatory Variables All explanatory variables are obtained from PCAOB proprietary data.

%_Pre-final Total lead U.S. office hours spent in planning, quarterly review, and interim

phases divided by total lead U.S. office hours (total lead U.S. office hours

exclude hours spent by the non-lead U.S. offices and foreign affiliates). Both

the numerator and denominator include hours spent by IS auditors, tax

auditors, and specialists, but exclude any hours incurred after the audit opinion

is issued.

%_Audit_Pre-final Total lead U.S. office hours spent by the core audit team during the planning,

quarterly review, and interim audit phases divided by total lead U.S. office

hours. The core audit team excludes hours spent by IS auditors, tax auditors,

and specialists (but the denominator includes them). Both the numerator and

denominator exclude any hours incurred after the audit opinion is issued.

%_NonAudit_Pre-final Total lead U.S. office hours spent by the IS auditors, tax auditors, and

specialists in planning, quarterly review, and interim phases divided by total

lead U.S. office hours. The numerator excludes core audit team hours (but the

denominator includes them). Both the numerator and denominator exclude any

hours incurred after the audit opinion is issued.

%_LeadPartner Total lead partner hours divided by total lead U.S. office hours.

%_EQR Total engagement quality reviewer (EQR) hours divided by total lead U.S.

office hours.

%_OtherExperienced Total other experienced team members (other partners excluding the lead

partner and EQR, directors, senior managers and managers) divided by total

lead U.S. office hours

%_LeadPartner_Pre-final Total lead partner hours spent in planning, quarterly review, and interim

phases divided by total lead U.S. office hours.

%_EQR_Pre-final Total EQR hours spent in planning, quarterly review, and interim phases

divided by total lead U.S. office hours.

%_OtherExperienced_Pre-final Total other experienced team members (other partners excluding the lead

partner and EQR, directors, senior managers and managers) spent in planning,

quarterly review, and interim phases divided by total lead U.S. office hours.

Avg_ClientExp Number of years of client experience, averaged over lead partner, EQR, and

other experienced team members

LeadPartner_ClientExp Number of years of client experience for the lead partner.

EQR_ClientExp Number of years of client experience for the EQR.

OtherExperienced_ClientExp Number of years of client experience, averaged over the other experienced

Page 50: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

48

Variable Definition

team members.

Avg_IndustryExp Number of years of industry experience, averaged over lead partner, EQR, and

other experienced team members.

LeadPartner_IndExp Number of years of industry experience for the lead partner.

EQR_IndExp Number of years of industry experience for the EQR.

OtherExperienced_IndExp Number of years of industry experience, averaged over other experienced team

members.

Avg_Seniority Number of years of total audit experience, averaged over lead partner, EQR,

and other experienced team members. Because the dataset only provides

experience at a given level, 12 years of experience are added to the partner and

director levels, 9 for senior managers, and 6 for managers.

LeadPartner_Seniority Number of years of total audit experience for the lead partner.

EQR_Seniority Number of years of total audit experience for the EQR.

OtherExperienced_Seniority Number of years of total audit experience, averaged over other experienced

team members. Because the dataset only provides experience at a given role

(i.e. manager, senior manager, etc), 12 years of experience are added to the

partner and director levels, 9 for senior managers, and 6 for managers.

Count_OtherExperienced Total number of other experienced team members on the team.

Control Variables All data from Compustat unless otherwise noted

IntegratedAudit Indicator variable when the audit is an integrated audit of internal controls over

financial reporting and financial statements, from PCAOB proprietary data.

FirstYear Indicator variable when the client is a first-year audit client, from Audit

Analytics.

ForeignPifo Absolute value of pretax income from foreign operations (PIFO) divided by

the absolute value of pretax income (PI).

Logat Logarithm of assets.

Geoseg Number of geographic segments.

Busseg Number of business segments.

DecYE An indicator variable equal to one when the fiscal year ends in December.

StdCFOat Standard deviation of the client's cash flows from operations deflated by

beginning assets, computed over t-3 and t.

CFOat Client's cash flows from operations deflated by beginning assets.

Leverage Total debt divided by debt plus stockholder's equity.

Loss Indicator variable equal to one when income before extraordinary items (IB) is

negative.

BTM Shareholder's equity (book value) deflated by fiscal year end market

capitalization

Litigation Indicator variable if the client is in a high litigation industry (SIC code

between 2833 and 2836, 8731 and 8734, 3570 and 3577, 7370 and 7374, 3600

and 3674, or 5200 and 5961).

SaleGrowth Year-on-year sales growth of the client firm.

Weakness An indicator variable equal to one if the client reports a material weakness.

Page 51: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

49

Table 1: Sample Selection Process

This table presents the sample construction from the raw dataset obtained from the PCAOB.

Observations

Engagement Profiles. Inspection Years 2006 - 2015

3,051

Less:

Missing hours data or experience data 270

Errors in sub-totaling hours across phases and roles 175

Inspection of referred work 28

Missing Dependent Variables and Controls (Compustat, Audit Analytics, or PCAOB) 239

Final Sample (client-years)

2,339

Page 52: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

50

Table 2: Sample Statistics

This table presents several sample-level statistics. Panel A breaks down the sample by inspection year and auditor

affiliation (Big 4 and non-Big 4), and Panel B breaks down the sample by outcome (restatements and Part I

Findings).

Panel A: Observations by Inspection Year and Big 4

Inspection Year Non Big 4 Big 4 Total

2006 58 169 227

2007 72 184 256

2008 68 151 219

2009 67 196 263

2010 65 178 243

2011 65 178 243

2012 62 165 227

2013 67 182 249

2014 64 186 250

2015 38 124 162

Total 626 1,713 2,339

Panel B: Observations by Outcome (Restatements and Part I Findings)

No Part I Finding Part I Finding Total Part I Finding Rate

No Restatement 1,521 534 2,055 26%

Restatement 166 118 284 42%

Total 1,687 652 2,339 28%

Restatement Rate 10% 18% 12%

Page 53: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

51

Table 3: Descriptive Statistics

This table presents overall descriptive statistics in Panel A, and the sample means partitioned between observations

with and without restatements in Panel B, and observations with and without Part I Findings in Panel C. A t-test

assesses the differences in means across the subsamples in Panels B and C. Variable definitions are provided in the

Appendix. Significance levels are * 10%, ** 5% and *** 1%.

Panel A: Descriptive Statistics

Variable Observations Mean StDev Perc25th Median Perc75th

PartIFinding 2,339 0.28 0.45 0.00 0.00 1.00

Restatement 2,339 0.12 0.33 0.00 0.00 0.00

Material_Restatement 2,339 0.04 0.20 0.00 0.00 0.00

Logaudithours 2,339 8.90 0.93 8.25 8.83 9.47

Realization_Rate 1,399 48.52 22.54 30.33 43.00 65.00

Logauditfees 2,339 14.22 1.07 13.47 14.11 14.88

%_Pre-final 2,339 0.56 0.10 0.50 0.57 0.63

%_Audit_Pre-final 2,339 0.46 0.09 0.41 0.46 0.52

%_NonAudit_Pre-final 2,339 0.10 0.06 0.06 0.09 0.14

%_LeadPartner 2,339 0.05 0.02 0.03 0.04 0.06

%_EQR 2,339 0.01 0.01 0.01 0.01 0.01

%_OtherExperienced 2,339 0.18 0.06 0.14 0.17 0.22

%_LeadPartner_Pre-final 2,339 0.03 0.01 0.02 0.02 0.03

%_EQR_Pre-final 2,339 0.01 0.00 0.00 0.00 0.01

%_OtherExperienced_Pre-final 2,339 0.10 0.04 0.07 0.10 0.13

Avg_ClientExp 2,339 3.11 1.33 2.20 3.00 3.83

LeadPartner_ClientExp 2,339 3.43 2.57 2.00 3.00 4.00

EQR_ClientExp 2,339 2.51 1.36 1.00 2.00 4.00

OtherExperienced_ClientExp 2,334 3.21 1.75 2.00 3.00 4.20

Avg_IndustryExp 2,339 12.73 4.67 9.50 13.00 16.00

LeadPartner_IndExp 2,339 18.13 8.84 11.00 19.00 25.00

EQR_IndExp 2,339 19.02 9.22 12.00 20.00 25.00

OtherExperienced_IndExp 2,333 7.98 3.62 5.50 8.00 10.00

Avg_Seniority 2,339 17.31 3.02 15.00 16.83 19.00

LeadPartner_Seniority 2,339 23.82 6.52 19.00 22.75 28.00

EQR_Seniority 2,339 25.46 6.44 21.00 25.00 30.00

OtherExperienced_Seniority 2,333 11.37 2.97 9.50 11.00 13.00

Count_OtherExperienced 2,339 3.09 2.24 2.00 2.00 4.00

IntegratedAudit 2,339 0.77 0.42 1.00 1.00 1.00

FirstYear 2,339 0.08 0.27 0.00 0.00 0.00

ForeignPifo 2,339 0.28 0.60 0.00 0.00 0.35

Logat 2,339 7.40 1.79 6.10 7.30 8.50

Geoseg 2,339 2.45 2.40 1.00 2.00 4.00

Busseg 2,339 2.09 1.73 1.00 1.00 3.00

DecYE 2,339 0.73 0.44 0.00 1.00 1.00

StdCFOat 2,339 0.06 0.07 0.02 0.04 0.07

CFOat 2,339 0.09 0.11 0.02 0.08 0.14

Leverage 2,339 0.37 0.31 0.10 0.34 0.56

Loss 2,339 0.25 0.44 0.00 0.00 1.00

BTM 2,339 0.61 0.63 0.28 0.50 0.82

Litigation 2,339 0.28 0.45 0.00 0.00 1.00

SaleGrowth 2,339 0.15 0.39 -0.03 0.07 0.21

Weakness 2,339 0.05 0.22 0.00 0.00 0.00

Page 54: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

52

Panel B: Partition between No Restatements (2,055 observations) and Restatements (284 observations)

Difference in means

(No Restatement - Restatement)

Variable

Mean

No Restatement

Mean

Restatement t-statistic p-value

Logaudithours 8.894 8.926 -0.544 0.586

Realization_Rate 48.322 49.734 -0.808 0.419

Logauditfees 14.218 14.196 0.332 0.740

%_Pre-final 0.566 0.537 4.649*** 0.000

%_Audit_Pre-final 0.464 0.439 4.678*** 0.000

%_NonAudit_Pre-final 0.101 0.097 1.026 0.305

%_LeadPartner 0.048 0.048 -0.054 0.957

%_EQR 0.010 0.010 1.735* 0.083

%_OtherExperienced 0.183 0.176 1.817* 0.069

%_LeadPartner_Pre-final 0.027 0.026 0.980 0.327

%_EQR_Pre-final 0.005 0.005 2.039** 0.042

%_OtherExperienced_Pre-final 0.106 0.098 2.637*** 0.008

Avg_ClientExp 3.149 2.840 3.674*** 0.000

LeadPartner_ClientExp 3.455 3.246 1.280 0.201

EQR_ClientExp 2.501 2.543 -0.494 0.621

OtherExperienced_ClientExp 3.265 2.836 3.883*** 0.000

Avg_IndustryExp 12.811 12.117 2.348** 0.019

LeadPartner_IndExp 18.230 17.401 1.480 0.139

EQR_IndExp 19.171 17.921 2.143** 0.032

OtherExperienced_IndExp 8.046 7.516 2.319** 0.020

Avg_Seniority 17.317 17.271 0.242 0.809

LeadPartner_Seniority 23.868 23.469 0.967 0.334

EQR_Seniority 25.511 25.127 0.943 0.346

OtherExperienced_Seniority 11.375 11.334 0.216 0.829

Count_OtherExperienced 3.098 3.039 0.420 0.675

IntegratedAudit 0.770 0.757 0.497 0.619

FirstYear 0.079 0.077 0.080 0.936

ForeignPifo 0.283 0.288 -0.138 0.890

Logat 7.420 7.245 1.544 0.123

Geoseg 2.451 2.426 0.162 0.872

Busseg 2.089 2.074 0.134 0.894

DecYE 0.740 0.690 1.770* 0.077

StdCFOat 0.058 0.059 -0.170 0.865

CFOat 0.087 0.075 1.716* 0.086

Leverage 0.366 0.413 -2.360** 0.018

Loss 0.250 0.282 -1.145 0.252

BTM 0.610 0.622 -0.320 0.749

Litigation 0.276 0.292 -0.576 0.565

SaleGrowth 0.150 0.152 -0.059 0.953

Weakness 0.043 0.109 -4.790*** 0.000

Page 55: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

53

Panel C: Partition between No Part I Finding (1,687 observations) and Part I Finding (652 observations)

Difference in means (No Part I - Part I)

Variable

Mean

No Part I Finding

Mean

Part I Finding t-statistic p-value

Logaudithours 8.908 8.873 0.815 0.415

Realization_Rate 48.477 48.594 -0.091 0.927

Logauditfees 14.242 14.148 1.908* 0.056

%_Pre-final 0.568 0.547 4.542*** 0.000

%_Audit_Pre-final 0.465 0.450 3.743*** 0.000

%_NonAudit_Pre-final 0.102 0.096 2.502** 0.012

%_LeadPartner 0.048 0.048 -0.345 0.730

%_EQR 0.010 0.011 -3.994*** 0.000

%_OtherExperienced 0.183 0.180 0.830 0.407

%_LeadPartner_Pre-final 0.027 0.027 0.448 0.654

%_EQR_Pre-final 0.005 0.006 -3.251*** 0.001

%_OtherExperienced_Pre-

final 0.106 0.102 1.738* 0.082

Avg_ClientExp 3.109 3.119 -0.169 0.866

LeadPartner_ClientExp 3.388 3.536 -1.248 0.212

EQR_ClientExp 2.523 2.461 0.990 0.322

OtherExperienced_ClientExp 3.207 3.229 -0.266 0.791

Avg_IndustryExp 12.569 13.133 -2.624*** 0.009

LeadPartner_IndExp 17.805 18.969 -2.858**** 0.004

EQR_IndExp 18.767 19.670 -2.125** 0.034

OtherExperienced_IndExp 7.892 8.215 -1.932* 0.053

Avg_Seniority 17.300 17.340 -0.285 0.775

LeadPartner_Seniority 23.733 24.043 -1.030 0.303

EQR_Seniority 25.603 25.108 1.667* 0.096

OtherExperienced_Seniority 11.258 11.659 -2.919*** 0.004

Count_OtherExperienced 3.023 3.267 -2.361** 0.018

IntegratedAudit 0.734 0.859 -6.486*** 0.000

FirstYear 0.076 0.084 -0.635 0.525

ForeignPifo 0.278 0.297 -0.676 0.499

Logat 7.337 7.560 -2.702*** 0.007

Geoseg 2.539 2.210 2.982*** 0.003

Busseg 2.076 2.113 -0.465 0.642

DecYE 0.734 0.733 0.035 0.972

StdCFOat 0.062 0.048 4.338*** 0.000

CFOat 0.093 0.066 5.293*** 0.000

Leverage 0.362 0.399 -2.585** 0.010

Loss 0.254 0.253 0.061 0.951

BTM 0.571 0.716 -5.013*** 0.000

Litigation 0.303 0.212 4.463*** 0.000

SaleGrowth 0.158 0.131 1.452 0.147

Weakness 0.049 0.055 -0.593 0.553

Page 56: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

54

Table 4: Role of Pre-final Hours

This table presents the results of Model (1) when using %_Pre-final, the proportion of pre-final hours, as the main

explanatory variable of interest in Panel A; and in Panel B the split of this variable between %_Audit_Pre-final and

%_NonAudit_Pre-final, the proportion of pre-final hours for the core audit team and IS, tax and specialist members,

respectively. Variable definitions are provided in the Appendix. The robust standard error, clustered at the client

level (in parenthesis), is below the coefficient. Significance levels are * 10%, ** 5% and *** 1%.

Panel A: Pre-final Hours

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

%_Pre-final -0.182* -0.247*** -0.159*** 0.460*** 5.703 0.633***

(0.0976) (0.0808) (0.0502) (0.115) (4.069) (0.132)

IntegratedAudit 0.0249 -0.0318 0.0121 -0.0366 -1.037 -0.0727*

(0.0281) (0.0247) (0.0157) (0.0331) (1.249) (0.0378)

FirstYear 0.0324 -0.00922 -0.00720 0.222*** -17.28*** -0.114***

(0.0340) (0.0262) (0.0165) (0.0332) (1.458) (0.0378)

ForeignPifo 0.0216 -0.00119 0.00890 0.0848*** -0.297 0.141***

(0.0176) (0.0126) (0.00946) (0.0213) (0.556) (0.0221)

Logat 0.00535 -0.00867 -0.00537* 0.377*** 1.663*** 0.460***

(0.00675) (0.00529) (0.00315) (0.00913) (0.272) (0.0105)

Geoseg -0.0103** -0.000739 -0.00278 0.0387*** -0.310** 0.0402***

(0.00407) (0.00309) (0.00184) (0.00522) (0.142) (0.00536)

Busseg 0.00432 -0.00133 0.00147 0.0603*** 0.213 0.0635***

(0.00531) (0.00422) (0.00239) (0.00666) (0.200) (0.00715)

DecYE -0.0355 -0.0225 -0.0181 -0.00365 -0.343 -0.0121

(0.0223) (0.0183) (0.0121) (0.0253) (0.904) (0.0296)

StdCFOat -0.385*** -0.0459 0.133* 0.410** 1.200 0.651***

(0.133) (0.0997) (0.0774) (0.181) (5.850) (0.217)

CFOat -0.250*** -0.0714 -0.0102 -0.0244 6.394* -0.0287

(0.0874) (0.0688) (0.0497) (0.106) (3.477) (0.124)

Leverage 0.0574* 0.0736*** 0.0462** -0.0164 -1.203 -0.0762*

(0.0334) (0.0274) (0.0201) (0.0384) (1.246) (0.0455)

Loss -0.0422* -0.0149 -0.00530 0.134*** 1.706** 0.152***

(0.0244) (0.0200) (0.0134) (0.0271) (0.861) (0.0306)

BTM 0.0512*** 0.0134 0.00492 -0.0530*** -0.586 -0.0782***

(0.0162) (0.0128) (0.00788) (0.0183) (0.515) (0.0207)

Litigation -0.0140 -0.00189 -0.00468 -0.0798** -0.0798 -0.103***

(0.0293) (0.0238) (0.0152) (0.0365) (1.140) (0.0392)

SaleGrowth 0.00910 0.00503 0.0194 -0.0518* -0.722 -0.0447

(0.0246) (0.0176) (0.0143) (0.0282) (0.867) (0.0296)

Weakness 0.0160 0.133*** 0.0517* 0.393*** 0.599 0.362***

(0.0420) (0.0394) (0.0279) (0.0457) (1.913) (0.0543)

Audit Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,339 2,339 2,339 2,339 1,399 2,339

Adjusted R-squared 0.118 0.035 0.027 0.750 0.738 0.771

Page 57: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

55

Panel B: Partition of Pre-final Hours Between Core Audit Team and Other Members

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

%_Audit_Pre-final -0.252** -0.250*** -0.146*** 0.322** 6.045 0.389***

(0.104) (0.0861) (0.0527) (0.127) (4.338) (0.142)

%_NonAudit_Pre-final 0.0232 -0.224 -0.203** 0.840*** 4.339 1.458***

(0.185) (0.142) (0.0902) (0.218) (7.991) (0.257)

IntegratedAudit 0.0258 -0.0317 0.0121 -0.0354 -1.047 -0.0710*

(0.0281) (0.0247) (0.0158) (0.0331) (1.247) (0.0377)

FirstYear 0.0317 -0.00943 -0.00710 0.221*** -17.26*** -0.117***

(0.0340) (0.0262) (0.0165) (0.0332) (1.453) (0.0377)

ForeignPifo 0.0208 -0.00128 0.00902 0.0834*** -0.292 0.139***

(0.0176) (0.0126) (0.00946) (0.0215) (0.557) (0.0216)

Logat 0.00324 -0.00896* -0.00491 0.373*** 1.681*** 0.451***

(0.00685) (0.00540) (0.00323) (0.00927) (0.285) (0.0104)

Geoseg -0.0102** -0.000718 -0.00281 0.0390*** -0.311** 0.0408***

(0.00406) (0.00309) (0.00184) (0.00520) (0.142) (0.00532)

Busseg 0.00416 -0.00138 0.00147 0.0601*** 0.214 0.0631***

(0.00531) (0.00422) (0.00239) (0.00668) (0.199) (0.00716)

DecYE -0.0335 -0.0223 -0.0184 -0.000110 -0.354 -0.00510

(0.0224) (0.0182) (0.0120) (0.0252) (0.905) (0.0293)

StdCFOat -0.377*** -0.0449 0.132* 0.423** 1.081 0.677***

(0.133) (0.0998) (0.0773) (0.179) (5.855) (0.212)

CFOat -0.258*** -0.0725 -0.00880 -0.0380 6.437* -0.0576

(0.0874) (0.0689) (0.0497) (0.106) (3.489) (0.123)

Leverage 0.0617* 0.0745*** 0.0456** -0.00968 -1.248 -0.0612

(0.0336) (0.0276) (0.0204) (0.0385) (1.269) (0.0456)

Loss -0.0425* -0.0149 -0.00522 0.134*** 1.708** 0.151***

(0.0244) (0.0200) (0.0134) (0.0271) (0.862) (0.0306)

BTM 0.0526*** 0.0137 0.00471 -0.0508*** -0.599 -0.0731***

(0.0162) (0.0128) (0.00798) (0.0182) (0.520) (0.0206)

Litigation -0.0123 -0.00203 -0.00519 -0.0757** -0.0906 -0.0959**

(0.0295) (0.0238) (0.0152) (0.0367) (1.142) (0.0392)

SaleGrowth 0.0100 0.00507 0.0191 -0.0499* -0.732 -0.0405

(0.0246) (0.0176) (0.0144) (0.0281) (0.870) (0.0293)

Weakness 0.0140 0.132*** 0.0518* 0.390*** 0.608 0.358***

(0.0419) (0.0394) (0.0278) (0.0455) (1.912) (0.0538)

Audit Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,339 2,339 2,339 2,339 1,399 2,339

Adjusted R-squared 0.119 0.035 0.027 0.750 0.737 0.772

Page 58: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

56

Table 5: Role of Experienced Hours

This table presents the results of Model (1) when considering the proportion of hours spent by the lead engagement

partner, the EQR, and the other experienced audit team members, respectively. Panel A focuses on all phases of the

audit, whereas Panel B focuses on the pre-final phase of the audit. Variable definitions are provided in the

Appendix. The robust standard error, clustered at the client level (in parenthesis), is below the coefficient.

Significance levels are * 10%, ** 5% and *** 1%.

Panel A: Proportion of Hours Spent at Different Auditor Levels

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

%_LeadPartner 0.0373 0.490 0.415* -4.614*** 24.51 -3.337***

(0.500) (0.376) (0.238) (0.541) (20.40) (0.628)

%_EQR -0.688 -2.274* -0.927 -15.62*** 1.376 -15.98***

(1.787) (1.289) (0.824) (1.911) (66.48) (2.117)

%_OtherExperienced -0.358** -0.0403 0.0254 1.350*** -9.605 1.491***

(0.147) (0.106) (0.0680) (0.171) (5.988) (0.194)

IntegratedAudit 0.0175 -0.0395 0.00834 -0.0422 -0.989 -0.0716*

(0.0280) (0.0246) (0.0157) (0.0310) (1.239) (0.0367)

FirstYear 0.0334 -0.0118 -0.00794 0.152*** -16.79*** -0.173***

(0.0345) (0.0268) (0.0171) (0.0313) (1.459) (0.0373)

ForeignPifo 0.0244 -0.000678 0.00881 0.0830*** -0.272 0.135***

(0.0176) (0.0128) (0.00947) (0.0218) (0.555) (0.0219)

Logat 0.00343 -0.0151*** -0.00891*** 0.343*** 1.941*** 0.432***

(0.00696) (0.00527) (0.00320) (0.00861) (0.290) (0.0101)

Geoseg -0.00992** -0.000933 -0.00294 0.0357*** -0.306** 0.0369***

(0.00411) (0.00311) (0.00185) (0.00504) (0.143) (0.00518)

Busseg 0.00448 -0.00155 0.00147 0.0526*** 0.236 0.0565***

(0.00531) (0.00424) (0.00239) (0.00640) (0.200) (0.00695)

DecYE -0.0362 -0.0240 -0.0194 0.00743 -0.242 -0.000722

(0.0223) (0.0183) (0.0123) (0.0237) (0.899) (0.0284)

StdCFOat -0.373*** -0.0545 0.124 0.403** 2.044 0.629***

(0.133) (0.101) (0.0780) (0.163) (5.836) (0.212)

CFOat -0.251*** -0.0766 -0.0133 -0.0378 6.387* -0.0402

(0.0878) (0.0697) (0.0505) (0.101) (3.479) (0.121)

Leverage 0.0617* 0.0793*** 0.0495** -0.0211 -1.282 -0.0833*

(0.0335) (0.0271) (0.0199) (0.0367) (1.259) (0.0448)

Loss -0.0397 -0.0138 -0.00490 0.128*** 1.747** 0.144***

(0.0243) (0.0200) (0.0135) (0.0254) (0.859) (0.0295)

BTM 0.0539*** 0.0167 0.00645 -0.0431** -0.641 -0.0712***

(0.0163) (0.0127) (0.00792) (0.0168) (0.512) (0.0196)

Litigation -0.0187 -0.00666 -0.00762 -0.0665* -0.0910 -0.0852**

(0.0292) (0.0236) (0.0152) (0.0347) (1.128) (0.0371)

SaleGrowth 0.0120 0.00653 0.0196 -0.0398 -0.790 -0.0379

(0.0244) (0.0179) (0.0145) (0.0257) (0.874) (0.0280)

Weakness 0.0260 0.141*** 0.0569** 0.350*** 0.500 0.309***

(0.0419) (0.0401) (0.0284) (0.0421) (1.942) (0.0507)

Audit Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,339 2,339 2,339 2,339 1,399 2,339

Adjusted R-squared 0.119 0.032 0.024 0.776 0.738 0.786

Page 59: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

57

Panel B: Proportion of Pre-final Hours Spent at Different Auditor Levels

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

%_LeadPartner_Pre-final 0.349 0.391 0.198 -6.253*** 25.43 -4.053***

(0.791) (0.550) (0.358) (0.820) (30.07) (0.979)

%_EQR_Pre-final -0.740 -4.023* -0.941 -27.73*** 165.3 -28.41***

(3.013) (2.260) (1.537) (3.288) (110.8) (3.610)

%_OtherExperienced

_Pre-final

-0.590*** -0.152 -0.0461 2.406*** -8.519 2.743***

(0.218) (0.160) (0.107) (0.266) (8.536) (0.283)

IntegratedAudit 0.0197 -0.0392 0.00812 -0.0350 -0.916 -0.0663*

(0.0280) (0.0246) (0.0158) (0.0305) (1.249) (0.0359)

FirstYear 0.0364 -0.0145 -0.00945 0.143*** -16.84*** -0.185***

(0.0345) (0.0267) (0.0169) (0.0316) (1.454) (0.0376)

ForeignPifo 0.0244 0.000281 0.00978 0.0788*** -0.293 0.131***

(0.0176) (0.0127) (0.00957) (0.0219) (0.548) (0.0213)

Logat 0.00638 -0.0150*** -0.00884*** 0.342*** 2.005*** 0.427***

(0.00703) (0.00540) (0.00332) (0.00882) (0.289) (0.0102)

Geoseg -0.00990** -0.000859 -0.00283 0.0360*** -0.310** 0.0370***

(0.00412) (0.00311) (0.00184) (0.00504) (0.143) (0.00517)

Busseg 0.00463 -0.00162 0.00142 0.0541*** 0.243 0.0577***

(0.00531) (0.00424) (0.00240) (0.00643) (0.199) (0.00693)

DecYE -0.0352 -0.0230 -0.0191 0.00996 -0.278 -0.000628

(0.0224) (0.0184) (0.0124) (0.0238) (0.899) (0.0286)

StdCFOat -0.366*** -0.0437 0.130* 0.416** 1.781 0.626***

(0.133) (0.101) (0.0776) (0.166) (5.794) (0.213)

CFOat -0.252*** -0.0750 -0.0125 -0.0188 6.527* -0.0211

(0.0878) (0.0695) (0.0504) (0.101) (3.451) (0.121)

Leverage 0.0607* 0.0796*** 0.0494** -0.0172 -1.438 -0.0775*

(0.0335) (0.0271) (0.0199) (0.0369) (1.253) (0.0446)

Loss -0.0404* -0.0138 -0.00459 0.126*** 1.734** 0.142***

(0.0243) (0.0200) (0.0134) (0.0254) (0.858) (0.0292)

BTM 0.0523*** 0.0167 0.00652 -0.0420** -0.697 -0.0689***

(0.0163) (0.0128) (0.00799) (0.0170) (0.514) (0.0195)

Litigation -0.0147 -0.00495 -0.00756 -0.0664* -0.118 -0.0894**

(0.0293) (0.0237) (0.0152) (0.0346) (1.132) (0.0372)

SaleGrowth 0.0108 0.00774 0.0206 -0.0408 -0.833 -0.0376

(0.0244) (0.0179) (0.0146) (0.0257) (0.854) (0.0279)

Weakness 0.0208 0.138*** 0.0577** 0.332*** 0.666 0.300***

(0.0421) (0.0401) (0.0285) (0.0424) (1.934) (0.0518)

Audit Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,339 2,339 2,339 2,339 1,399 2,339

Adjusted R-squared 0.119 0.032 0.022 0.775 0.738 0.788

Page 60: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

58

Table 6: Role of Client-Specific Expertise

This table presents the results of Model (1) when using client expertise as the main explanatory variable of interest.

Panel A considers average client expertise for the experienced audit team members, and Panel B splits this expertise

between lead partner, EQR, and other experienced team members. Panel C transforms each continuous variable of

interest into an indicator variable for when the variable is above the median of the distribution. Variable definitions

are provided in the Appendix. The robust standard error, clustered at the client level (in parenthesis), is below the

coefficient. Significance levels are * 10%, ** 5% and *** 1%.

Panel A: Average Client Expertise

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

Avg_ClientExp -0.00709 -0.0147** -0.00716** -0.0237*** 2.049*** 0.0180*

(0.00788) (0.00597) (0.00325) (0.00785) (0.264) (0.00936)

Count_OtherExperienced 0.00393 -0.00131 0.000847 0.109*** -0.633*** 0.106***

(0.00507) (0.00398) (0.00266) (0.00653) (0.182) (0.00710)

IntegratedAudit 0.0225 -0.0374 0.00906 0.0148 -0.923 -0.0200

(0.0281) (0.0246) (0.0159) (0.0302) (1.236) (0.0356)

FirstYear 0.0133 -0.0435 -0.0250 0.144*** -12.77*** -0.101**

(0.0378) (0.0292) (0.0185) (0.0359) (1.529) (0.0427)

ForeignPifo 0.0222 -0.000269 0.00951 0.0762*** -0.211 0.134***

(0.0176) (0.0128) (0.00962) (0.0211) (0.557) (0.0218)

Logat 0.0000618 -0.00972 -0.00782** 0.310*** 1.820*** 0.390***

(0.00752) (0.00611) (0.00393) (0.00943) (0.304) (0.0109)

Geoseg -0.0103** -0.000774 -0.00280 0.0395*** -0.318** 0.0410***

(0.00409) (0.00310) (0.00185) (0.00476) (0.139) (0.00505)

Busseg 0.00375 -0.00112 0.00135 0.0449*** 0.325* 0.0484***

(0.00533) (0.00425) (0.00239) (0.00632) (0.195) (0.00675)

DecYE -0.0376* -0.0264 -0.0202* 0.00429 -0.257 0.00224

(0.0223) (0.0183) (0.0123) (0.0233) (0.886) (0.0275)

StdCFOat -0.399*** -0.0548 0.125 0.204 3.344 0.471**

(0.133) (0.100) (0.0779) (0.156) (5.764) (0.199)

CFOat -0.254*** -0.0785 -0.0140 -0.00896 6.232* -0.00158

(0.0874) (0.0693) (0.0503) (0.0957) (3.447) (0.116)

Leverage 0.0589* 0.0738*** 0.0471** -0.0214 -0.618 -0.0735*

(0.0334) (0.0274) (0.0199) (0.0363) (1.228) (0.0432)

Loss -0.0421* -0.0136 -0.00478 0.113*** 1.652** 0.130***

(0.0243) (0.0200) (0.0135) (0.0243) (0.835) (0.0285)

BTM 0.0529*** 0.0146 0.00605 -0.0441** -0.653 -0.0690***

(0.0162) (0.0128) (0.00791) (0.0188) (0.497) (0.0208)

Litigation -0.0178 -0.00709 -0.00801 -0.0688** 0.0480 -0.0887**

(0.0293) (0.0235) (0.0151) (0.0347) (1.109) (0.0374)

SaleGrowth 0.00935 0.00427 0.0193 -0.0538** -0.310 -0.0413

(0.0246) (0.0178) (0.0146) (0.0247) (0.836) (0.0278)

Weakness 0.0176 0.139*** 0.0552** 0.259*** 1.672 0.239***

(0.0417) (0.0400) (0.0280) (0.0385) (1.947) (0.0483)

Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,339 2,339 2,339 2,339 1,399 2,339

Adjusted R-squared 0.117 0.033 0.024 0.792 0.750 0.799

Page 61: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

59

Panel B: Client Expertise Split Between Lead Partner, EQR, and Other Experienced Team Members

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

LeadPartner_ClientExp 0.00575 -0.000764 -0.00110 -0.00648* 0.457*** -0.00452

(0.00364) (0.00281) (0.00136) (0.00353) (0.109) (0.00396)

EQR_ClientExp -0.00720 0.00550 -0.00237 -0.0167** 0.461** -0.00246

(0.00694) (0.00516) (0.00333) (0.00695) (0.222) (0.00793)

OtherExperienced_

ClientExp

-0.00827 -0.0120*** -0.00280 -0.0102* 1.136*** 0.0182***

(0.00583) (0.00422) (0.00251) (0.00556) (0.201) (0.00653)

Count_Other

Experienced

0.00372 -0.00170 0.000750 0.109*** -0.603*** 0.107***

(0.00508) (0.00401) (0.00268) (0.00649) (0.187) (0.00702)

Control Variables Yes Yes Yes Yes Yes Yes

Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,334 2,334 2,334 2,334 1,395 2,334

Adjusted R-squared 0.118 0.034 0.023 0.794 0.750 0.800

Panel C: Client Expertise Split with Indicator Variables for Above Median

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

LeadPartner_ClientExp

>median

0.0325* -0.0152 -0.00332 -0.0333* 2.854*** -0.0105

(0.0190) (0.0144) (0.00896) (0.0187) (0.609) (0.0207)

EQR_ClientExp

>median

-0.0168 0.0123 -0.0131 -0.0518*** 1.958*** -0.00931

(0.0186) (0.0141) (0.00886) (0.0185) (0.613) (0.0214)

OtherExperienced_

ClientExp>median

-0.0520*** -0.0238 0.00499 -0.00133 3.044*** 0.0837***

(0.0193) (0.0156) (0.00972) (0.0194) (0.666) (0.0222)

Count_Other

Experienced

0.00433 -0.00124 0.000823 0.109*** -0.636*** 0.106***

(0.00507) (0.00403) (0.00267) (0.00649) (0.188) (0.00696)

Control Variables Yes Yes Yes Yes Yes Yes

Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,334 2,334 2,334 2,334 1,395 2,334

Adjusted R-squared 0.120 0.0319 0.0227 0.793 0.750 0.800

Page 62: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

60

Table 7: Role of Industry Expertise

This table presents the results of Model (1) when using industry expertise as the main explanatory variable of

interest. Panel A considers average industry expertise for the experienced audit team members, and Panel B splits

this expertise between lead partner, EQR, and other experienced team members. Panel C transforms each continuous

variable of interest into an indicator variable for when the variable is above the median of the distribution. Variable

definitions are provided in the Appendix. The robust standard error, clustered at the client level (in parenthesis), is

below the coefficient. Significance levels are * 10%, ** 5% and *** 1%.

Panel A: Average Industry Expertise

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

Avg_IndustryExp 0.000479 -0.00223 -0.00195* -0.00277 0.0363 0.00130

(0.00213) (0.00170) (0.00110) (0.00221) (0.0791) (0.00247)

Count_OtherExperienced 0.00269 -0.00189 -0.000126 0.109*** -0.612*** 0.108***

(0.00518) (0.00401) (0.00263) (0.00654) (0.197) (0.00711)

IntegratedAudit 0.0220 -0.0389 0.00778 0.0129 -1.017 -0.0183

(0.0280) (0.0245) (0.0159) (0.0301) (1.235) (0.0356)

FirstYear 0.0289 -0.0136 -0.0105 0.197*** -17.09*** -0.134***

(0.0341) (0.0264) (0.0167) (0.0315) (1.448) (0.0372)

ForeignPifo 0.0225 0.000790 0.0102 0.0772*** -0.341 0.132***

(0.0176) (0.0127) (0.00953) (0.0212) (0.558) (0.0218)

Logat -0.000530 -0.0127** -0.00883** 0.304*** 2.282*** 0.392***

(0.00728) (0.00582) (0.00377) (0.00918) (0.302) (0.0106)

Geoseg -0.0104** -0.000786 -0.00281 0.0395*** -0.315** 0.0410***

(0.00408) (0.00311) (0.00184) (0.00478) (0.140) (0.00506)

Busseg 0.00396 -0.00138 0.00122 0.0445*** 0.306 0.0484***

(0.00533) (0.00426) (0.00240) (0.00635) (0.199) (0.00671)

DecYE -0.0366 -0.0252 -0.0199 0.00692 -0.250 0.000874

(0.0223) (0.0183) (0.0122) (0.0232) (0.898) (0.0275)

StdCFOat -0.395*** -0.0444 0.133* 0.209 2.712 0.450**

(0.133) (0.101) (0.0782) (0.157) (5.850) (0.199)

CFOat -0.252*** -0.0768 -0.0139 -0.0100 6.263* -0.00886

(0.0874) (0.0695) (0.0502) (0.0953) (3.498) (0.116)

Leverage 0.0611* 0.0758*** 0.0473** -0.0158 -1.275 -0.0753*

(0.0337) (0.0272) (0.0201) (0.0362) (1.254) (0.0429)

Loss -0.0420* -0.0135 -0.00460 0.111*** 1.805** 0.128***

(0.0244) (0.0200) (0.0134) (0.0244) (0.853) (0.0284)

BTM 0.0531*** 0.0155 0.00644 -0.0425** -0.724 -0.0696***

(0.0162) (0.0128) (0.00796) (0.0187) (0.510) (0.0208)

Litigation -0.0170 -0.0104 -0.0109 -0.0718** 0.0389 -0.0856**

(0.0295) (0.0239) (0.0152) (0.0348) (1.122) (0.0376)

SaleGrowth 0.0109 0.00434 0.0185 -0.0538** -0.770 -0.0437

(0.0246) (0.0176) (0.0142) (0.0250) (0.849) (0.0280)

Weakness 0.0218 0.144*** 0.0573** 0.266*** 0.972 0.232***

(0.0418) (0.0402) (0.0282) (0.0389) (1.951) (0.0482)

Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,339 2,339 2,339 2,339 1,399 2,339

Adjusted R-squared 0.117 0.031 0.024 0.792 0.740 0.800

Page 63: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

61

Panel B: Industry Expertise Split Among Lead Partner, EQR, and Other Experienced Team Members

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

LeadPartner_IndExp 0.00175 0.000410 0.000434 0.00146 0.00852 0.00262*

(0.00114) (0.000869) (0.000531) (0.00124) (0.0391) (0.00135)

EQR_IndExp -0.000594 -0.000912 -0.00115** 0.00165 -0.00257 0.00167

(0.00112) (0.000839) (0.000508) (0.00115) (0.0374) (0.00125)

OtherExperienced_

IndExp

-0.00240 -0.00283 -0.00231* -0.00857*** 0.0603 -0.00447

(0.00283) (0.00216) (0.00134) (0.00276) (0.102) (0.00316)

Count_Other

Experienced

0.00275 -0.000726 0.000996 0.112*** -0.630*** 0.108***

(0.00514) (0.00406) (0.00264) (0.00647) (0.195) (0.00700)

Control Variables Yes Yes Yes Yes Yes Yes

Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,333 2,333 2,333 2,333 1,395 2,333

Adjusted R-squared 0.117 0.032 0.026 0.794 0.739 0.801

Panel C: Industry Expertise Split with Indicator Variables for Above Median

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

LeadPartner_IndExp

>median

0.0514*** 0.0116 0.00603 0.0199 1.316** 0.0433**

(0.0188) (0.0145) (0.00894) (0.0192) (0.663) (0.0214)

EQR_IndExp

>median

-0.00775 -0.0186 -0.0159* 0.0205 -0.124 0.0341

(0.0195) (0.0146) (0.00842) (0.0198) (0.695) (0.0213)

OtherExperienced

_IndExp>median

-0.0186 -0.0361** -0.0255*** -0.0177 0.0229 0.0120

(0.0187) (0.0145) (0.00852) (0.0198) (0.723) (0.0222)

Count_OtherExperienced 0.00261 -0.0000922 0.00142 0.111*** -0.625*** 0.107***

(0.00515) (0.00408) (0.00264) (0.00646) (0.194) (0.00695)

Control Variables Yes Yes Yes Yes Yes Yes

Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,333 2,333 2,333 2,333 1,395 2,333

Adjusted R-squared 0.119 0.0337 0.0270 0.793 0.740 0.801

Page 64: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

62

Table 8: Role of Overall Expertise Proxied by Seniority

This table presents the results of Model (1) when using overall expertise proxied by seniority as the main

explanatory variable of interest. Panel A considers average industry expertise for the experienced audit team

members, and Panel B splits this expertise between lead partner, EQR, and other experienced team members. Panel

C transforms each continuous variable of interest into an indicator variable for when the variable is above the

median of the distribution. Variable definitions are provided in the Appendix. The robust standard error, clustered at

the client level (in parenthesis), is below the coefficient. Significance levels are * 10%, ** 5% and *** 1%.

Panel A: Average Overall Expertise

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

Avg_Seniority 0.00126 -0.000454 0.000103 -0.00460 -0.0313 -0.000988

(0.00311) (0.00244) (0.00163) (0.00345) (0.117) (0.00363)

Count_OtherExperienced 0.00315 -0.00166 0.000540 0.106*** -0.626*** 0.104***

(0.00528) (0.00402) (0.00264) (0.00686) (0.200) (0.00727)

IntegratedAudit 0.0219 -0.0380 0.00870 0.0134 -1.051 -0.0196

(0.0281) (0.0246) (0.0159) (0.0301) (1.237) (0.0357)

FirstYear 0.0287 -0.0127 -0.00986 0.198*** -17.09*** -0.134***

(0.0341) (0.0264) (0.0167) (0.0315) (1.446) (0.0373)

ForeignPifo 0.0224 0.000406 0.00981 0.0781*** -0.330 0.133***

(0.0176) (0.0127) (0.00958) (0.0212) (0.555) (0.0218)

Logat -0.000682 -0.0128** -0.00914** 0.307*** 2.285*** 0.395***

(0.00732) (0.00580) (0.00374) (0.00927) (0.300) (0.0107)

Geoseg -0.0104** -0.000790 -0.00281 0.0394*** -0.316** 0.0409***

(0.00408) (0.00311) (0.00185) (0.00478) (0.141) (0.00507)

Busseg 0.00395 -0.00118 0.00137 0.0450*** 0.298 0.0487***

(0.00533) (0.00426) (0.00240) (0.00636) (0.200) (0.00675)

DecYE -0.0368* -0.0245 -0.0193 0.00672 -0.250 -0.000657

(0.0223) (0.0183) (0.0122) (0.0232) (0.897) (0.0275)

StdCFOat -0.395*** -0.0486 0.128 0.209 2.731 0.459**

(0.133) (0.100) (0.0781) (0.158) (5.846) (0.199)

CFOat -0.253*** -0.0749 -0.0125 -0.00499 6.176* -0.00859

(0.0875) (0.0697) (0.0504) (0.0958) (3.490) (0.116)

Leverage 0.0604* 0.0777*** 0.0489** -0.0162 -1.278 -0.0796*

(0.0334) (0.0271) (0.0199) (0.0363) (1.251) (0.0430)

Loss -0.0419* -0.0137 -0.00475 0.112*** 1.800** 0.129***

(0.0244) (0.0200) (0.0134) (0.0245) (0.854) (0.0285)

BTM 0.0532*** 0.0154 0.00639 -0.0447** -0.714 -0.0714***

(0.0162) (0.0128) (0.00794) (0.0188) (0.511) (0.0208)

Litigation -0.0173 -0.00729 -0.00798 -0.0707** 0.00750 -0.0891**

(0.0294) (0.0237) (0.0151) (0.0348) (1.125) (0.0376)

SaleGrowth 0.0110 0.00642 0.0205 -0.0515** -0.841 -0.0442

(0.0245) (0.0178) (0.0145) (0.0251) (0.851) (0.0279)

Weakness 0.0211 0.145*** 0.0582** 0.269*** 0.948 0.232***

(0.0418) (0.0401) (0.0281) (0.0393) (1.946) (0.0484)

Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,339 2,339 2,339 2,339 1,399 2,339

Adjusted R-squared 0.117 0.031 0.022 0.791 0.740 0.799

Page 65: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

63

Panel B: Overall Expertise Split Between Lead Partner, EQR, and Other Experienced Team Members

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

LeadPartner_Seniority 0.000812 -0.000301 0.000320 0.00332** 0.0448 0.00440***

(0.00142) (0.00118) (0.000735) (0.00145) (0.0533) (0.00163)

EQR_Seniority -0.00246* -0.00115 -0.00147** 0.00293** 0.0433 0.00458***

(0.00142) (0.00101) (0.000627) (0.00148) (0.0515) (0.00158)

OtherExperienced_

Seniority

0.00450 0.0000248 -0.000614 -0.00803** -0.203* -0.00585*

(0.00321) (0.00242) (0.00171) (0.00314) (0.114) (0.00347)

Count_OtherExperienced 0.00155 -0.00147 0.000687 0.109*** -0.573*** 0.105***

(0.00518) (0.00409) (0.00268) (0.00644) (0.193) (0.00705)

Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,333 2,333 2,333 2,333 1,395 2,333

Adjusted R-squared 0.118 0.030 0.024 0.793 0.740 0.801

Panel C: Overall Expertise Split with Indicator Variables for Above Median

(1) (2) (3) (4) (5) (6)

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

LeadPartner_Seniority

>median

0.00713 -0.0104 -0.000433 0.0491** 0.363 0.0551***

(0.0182) (0.0147) (0.00904) (0.0191) (0.658) (0.0214)

EQR_Seniority

>median

-0.0348* -0.0178 -0.0136* 0.0242 0.803 0.0441**

(0.0180) (0.0136) (0.00815) (0.0189) (0.643) (0.0210)

OtherExperienced

_Seniority>median

0.00516 -0.000923 -0.00736 -0.0433** -0.470 -0.0328

(0.0182) (0.0146) (0.00914) (0.0188) (0.646) (0.0206)

Count_OtherExperienced 0.00264 -0.00130 0.000927 0.110*** -0.596*** 0.106***

(0.00521) (0.00409) (0.00265) (0.00644) (0.191) (0.00704)

Control Variables Yes Yes Yes Yes Yes Yes

Firm Fixed Effects Yes Yes Yes Yes Yes Yes

Industry Fixed Effects Yes Yes Yes Yes Yes Yes

Year Fixed Effects Yes Yes Yes Yes Yes Yes

Observations 2,333 2,333 2,333 2,333 1,395 2,333

Adjusted R-squared 0.118 0.0310 0.0230 0.793 0.740 0.800

Page 66: Which Audit Input Matters? An Analysis of the Determinants ... Audit Input Matters - Aobdia...1 Which Audit Input Matters? An Analysis of the Determinants of Audit Quality, Profitability,

64

Table 9: Analysis of the Impact of Potential Selection Bias

This table assesses the impact of potential selection bias following Heckman (1979), who reduces selection bias to

an omitted variable issue, and Frank (2000) and Larcker and Rusticus (2010), who propose an analysis of omitted

variables to understand their minimum impact necessary to overturn results. For each explanatory variable of

interest that loads in prior analyses, we compute the impact threshold for a confounding variable (ITCV), that is the

lowest product of the partial correlation between the dependent variable and the omitted variable, and the partial

correlation between the explanatory variable of interest and the omitted variable that makes the coefficient

statistically insignificant; the square root of the reported values is how much the correlation needs to be with X and

Y to overturn the results. We also evaluate the control variables that have a benchmark impact sufficiently large to

overturn the result. That is, an omitted variable would need to have an impact comparable to these control variables

to overturn the results. ITCV in bold (italics) have no control variable (only Logat) with impact sufficiently large to

overturn the results. Overall, the results are more likely to be robust to selection bias when ITCV is high, and there

are fewer control variables with sufficient impact to overturn the results (that is, the ITCV is either in bold or in

italics). -- denotes coefficients that were not statistically significant at conventional levels in prior analyses and

where an analysis of impact is not conducted.

(1) (2) (3) (4) (5) (6)

Variable

PartI

Finding Restatement

Material_

Restatement

Logaudit

hours

Realization

_Rate

Logaudit

fees

Table 4 - Panel B

%_Audit_Pre-final -0.0101 -0.0205 -0.0176 0.0124 -- 0.0168

%_NonAudit_Pre-final -- -- 0.0245 0.1291 -- 0.1953

Table 5 - Panel A

%_LeadPartner -- -- 0.0660 -0.1402 -- -0.0723

%_EQR -- -0.0109 -- -0.2572 -- -0.2594

%_OtherExperienced 0.0079 -- -- 0.0113 -- 0.0206

Table 5 - Panel B

%_LeadPartner_Pre-final -- -- -- -0.1213 -- -0.0472

%_EQR_Pre-final -- -0.0098 -- -0.2352 -- -0.2293

%_OtherExperienced_Pre-final -0.0075 -- -- 0.0683 -- 0.0890

Table 6 - Panel B LeadPartner_ClientExp -- -- -- 0.0024 0.0634 --

EQR_ClientExp -- -- -- -0.0301 0.0302 --

OtherExperienced_ClientExp -- -0.0334 -- -0.0198 -0.0047 -0.0213

Count_OtherExperienced -- -- -- 0.3832 -0.1109 0.3172

Table 7 - Panel B

LeadPartner_IndExp -- -- -- -- -- -0.0004

EQR_IndExp -- -- -0.0034 -- -- --

OtherExperienced_IndExp -- -- -0.0013 -0.0718 -- --

Count_OtherExperienced -- -- -- 0.3848 -0.1088 0.3219

Table 8 - Panel B

LeadPartner_Seniority -- -- -- 0.0072 -- 0.0163

EQR_Seniority -0.0039 -- -0.0067 0.0304 -- 0.0430

OtherExperienced_Seniority -- -- -- -0.0036 -0.0357 -0.0076

Count_OtherExperienced -- -- -- 0.3755 -0.1108 0.3121