Do Audit Fees Reflect Risk

Premiums for Control Risk?

Journal of Accounting,
Auditing & Finance
123
The Author(s) 2014
Reprints and permissions:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/0148558X14560896
jaf.sagepub.com

Wei Jiang1 and Myungsoo Son1

Abstract
This study examines whether audit fees reflect risk premiums in the presence of control
risk after controlling for audit effort through audit delay. Our results indicate that auditors
adjust risk premiums as well as audit efforts in response to altered control risk. Further
analysis shows that the extent of risk premium adjustment varies depending on the severity
of the underlying internal control problems. Overall, these findings provide insights into the
distinct role of audit effort and risk premium in audit pricing decisions.
Keywords
risk premium, audit fee, audit delay, Section 404, internal control

Introduction
Recent research (Hoag & Hollingsworth, 2011; Hogan & Wilkins, 2008; Hoitash, Hoitash,
& Bedard, 2008; Munsif, Raghunandan, Rama, & Singhvi, 2011; Raghunandan & Rama,
2006) has examined the relationship between audit fees and internal control risk using the
publicly available data on effectiveness of internal control, as mandated by the Sarbanes
Oxley Act of 2002 (SOX). These studies find that audit fees are significantly higher for
firms that disclose material weakness in internal control, and the remediation of a material
weakness leads to a reduction in audit fee. The consistent findings from these studies are
interpreted as auditors managing control risk by exerting more effort in audits of firms with
weak internal controls, resulting in increased audit fees.
However, as noted in Hogan and Wilkins (2008), the documented fee increase could
also be due partially to the risk premium associated with internal control deficiencies. The
audit fee model developed by Simunic (1980) suggests that audit fee is a function of audit
effort and the auditors expected future losses arising primarily from the litigation risk. In
other words, auditors may respond to increased client risks by charging a premium to
insure against potential litigation as well as exerting more engagement effort.
Prior studies show that the disclosure of internal control weaknesses is often associated
with lower accruals quality and higher frequency of restatements (Ashbaugh-Skaife,
Collins, Kinney, & LaFond, 2008; Doyle, Ge, & McVay, 2007), which, in turn, increases
1

California State UniversityFullerton, USA

Corresponding Author:
Wei Jiang, California State UniversityFullerton, 800 N. State College Blvd., Fullerton, CA 92834-6848, USA.
Email: wjiang@fullerton.edu

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Journal of Accounting, Auditing & Finance

the likelihood of litigation against auditors (Abbott, Parker, & Peters, 2006; Barron, Pratt,
& Stice, 2001; Heninger, 2001). This suggests that in the presence of increased control
risk, although auditors can manage the risk by performing additional substantive tests, such
tests cannot uncover all possible misstatements, especially frauds perpetrated through management collusion (Hoitash et al., 2008). To the extent that auditors cannot reduce control
risk completely by additional auditing, they may charge clients a premium to compensate
them for possible future litigation losses.
The purpose of this study is to empirically analyze whether audit fees reflect a premium
for control risk. Our sample comprises all firms that filed Section 404 reports over the
period of 2004-2011. We employ a research design that enables us to disentangle the two
alternative explanations for the audit fee effect.1 After controlling for auditor effort (proxied by audit delay), we find a positive association between changes in audit fees and
changes in the quality of internal control, indicating that control risk explains a significant
amount of variance in audit fees over and above that explained by audit effort. This result
supports the view that auditors adjust the risk premium charged to clients for the assessed
control risk in addition to the scope and nature of audit work. Further analysis shows that
changes in audit fees due to the risk premium vary depending on the severity of the underlying problems pertaining to internal control. We find that auditors consider the type of
weakness (i.e., general weaknesses [GWs] vs. account-specific problems) when adjusting
the risk premium upward in response to increased control risk. However, this is not the
case when they adjust the risk premium downward for decreased control risk. This result is
consistent with auditors taking a more conservative approach when pricing the control risk
for clients who have previously experienced internal control problems, irrespective of the
type of weakness involved.
The above conclusions are drawn from the changes analysis. We adopt a changes analysis approach for two reasons. First, results based on a cross-sectional association are vulnerable to correlated omitted variables and endogeneity problems (Johnstone, Li, & Rupley,
2011). As a result, it is difficult to interpret the results because the reported parameter estimates from levels regression may be biased. A changes model allows us to control for
unobservable company characteristics assumed to be constant over time and mitigate potential endogeneity bias.2 Second, while levels analysis tests the cross-sectional differences in
audit fee levels across firms, changes analysis examines fee changes over time within the
same firm. Thus, the fee change design allows us to directly measure auditors responses to
changes in control risks, and increases our ability to draw a causality relationship between
material weaknesses in internal control and audit efforts/risk premiums.
Our study corroborates and extends prior research on audit fees and control risk. While
much of the prior work in this area consistently finds higher audit fees in the presence of
internal control weaknesses, little is known about whether the documented fee increment is
attributable to extended audit effort and/or a higher risk premium. An exception is the
study by Bedard and Johnstone (2004), which finds that auditors charge a higher hourly
billing rate to clients with higher control risk as a premium to cover costs related to potential future litigation. However, our study differs from theirs in two important ways. First,
we focus on auditors assessments of control risks and their pricing decisions in the presence of the identified risks, while their study focuses on auditors assessments of earnings
manipulation risk and corporate governance risk, with control risk being included as a control variable in the model. Second, their study relies on data derived from engagement partners assessments of their clients internal control quality from a single year preceding the
mandatory SOX 404 disclosure requirements, and thus is limited to one audit firm and a

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Jiang and Son

small sample size. In comparison, our study uses a much broader sample based on publicly
available data, which enables us to conduct a more comprehensive analysis of and draw
more reliable inferences about the relationship between internal control quality and risk
premium. Overall, the results of this study provide new insights into the distinct roles of
audit effort and risk premium in explaining the variations in audit fees.
This research is closely related to Hoag and Hollingsworth (2011). Yet, our study differs
from theirs in several aspects. First, while their study examines the relationship between
audit fees and internal control weakness, we focus on disentangling the two components of
the audit fees and their distinct roles in the audit pricing process. Second, we provide a
more comprehensive analysis of the impact of severity of internal control problems on
audit fees. Third, our sample covers a much longer and more recent period, which increases
our ability to draw more robust conclusions. Collectively, the articles in this area shed light
on how internal controls affect audit pricing behavior.
The remainder of this article proceeds as follows: The section Background, Literature
Review, and Hypothesis Development provides a literature review and develops our hypotheses; Research Design section describes the research design; Empirical Results section
presents descriptive statistics and multivariate results; Additional Analysis section presents
additional analysis; and the final section contains conclusions of this study.

Background, Literature Review, and Hypothesis Development

Background of Section 404
Internal control is a major focus of recent SOX regulatory changes. Prior to SOX, disclosure
of significant deficiencies in internal control was publicly required only when companies changed their auditors (Securities and Exchange Commission [SEC], 1988). Public firms could
voluntarily report on the effectiveness of internal controls, but few firms actually did so
(Sankaraguruswamy & Whisenant, 2004). SOX Section 404 explicitly requires an annual
report detailing internal controls over financial reporting to be filed with the SEC Form 10-K.
More specifically, Section 404 requires (a) management to establish, monitor, and sustain
adequate internal controls and assess the effectiveness of these controls; (b) auditors to attest
to managements assessments and provide a separate opinion about internal control effectiveness based on their own review of the firms internal controls.3 The auditors report must
include disclosure of any material internal control weaknesses, and procedures and corrective
actions taken to address the weaknesses. When one or more such weaknesses exist, the auditor is required to issue an adverse opinion on the effectiveness of internal controls.

Literature Review and Hypothesis Development

Disclosures about internal control, as entailed in Section 404 of SOX, provide an objective
and reliable measure of auditors assessments of internal control risk. The audit risk model,
discussed in the Statement on Auditing Standards (SAS) No. 107 (American Institute of
Certified Public Accountants, 2006, Paragraph 26) and Auditing Standard No 8 (Public
Company Accounting Oversight Board [PCAOB], 2010), suggests that as control risk
increases, auditors can reduce detection risk to achieve a desirable level of overall audit
risk by increasing substantive testing, resulting in increased audit fees for the client.
Recent research has provided strong evidence that, in the post-SOX period, firms disclosing material weaknesses in internal control pay higher audit fees than firms with an
unqualified SOX 404 opinion. Raghunandan and Rama (2006) find that firms receiving

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Journal of Accounting, Auditing & Finance

adverse SOX 404 opinions paid 43% higher audit fees than firms that received clean audit
reports. Hoitash et al. (2008) expand on Raghunandan and Rama (2006) by using more
comprehensive samples, and confirm that firms with reported internal control problems
under SOX Section 404 have substantially higher audit fees than firms without such disclosures. Hogan and Wilkins (2008) also document that audit firms increased fees for clients
who disclosed internal control deficiencies under SOX Section 302.4 To the extent that
audit fees act as a proxy for audit effort, the evidence is consistent with the audit risk
model and suggests that auditors increase their effort in response to increased control risk.5
However, an important question that has not been addressed in these studies is whether
the increase in audit fees is also attributable to the risk premium charged by the auditor to
cover potential litigation risks. This issue was raised by Hoag and Hollingsworth (2011) in
their inter-temporal analysis of audit fees and Section 404 internal control opinions. They
find that although audit fees decline for companies that remediate a material weakness, the
corresponding fee reduction does not occur instantaneously. According to them, a possible
explanation for the slow decrease could be that clients who report a material weakness pose
an additional business or litigation risk to their auditors, and this additional risk is persistently
priced in the clients current and future audit fees (Hoag & Hollingsworth, 2011, p. 198).
According to the audit pricing model developed by Simunic (1980), an audit fee is a
function of two key elements: a resource cost component that is increasing the level of
auditor effort and an expected future loss component, which arises primarily from litigation
risk. If the existence of control risk is perceived to increase the likelihood of auditor litigation, and the litigation risk cannot be reduced by additional auditing, auditors may charge
clients a form of insurance premium for possible future litigation losses.
The link between internal control risk and business/litigation risk may be established
through the potential impact of earnings management.6 The presence of material weaknesses in internal control is often associated with lower earnings quality. Doyle et al.
(2007) find that material weaknesses are generally associated with accruals that are not realized as cash flows, whereas Ashbaugh-Skaife et al. (2008) find that companies who remediated previously reported internal control deficiencies exhibit an increase in accrual
quality. Lax internal controls facilitate earnings management and opportunistic managerial
behavior, thus reducing the reliability of financial reporting and increasing the possibility
that material misstatements will not be detected by normal audit procedures.
Not surprisingly, prior research has provided evidence that links earnings management to
auditor litigation risk. For example, Heninger (2001) documents a positive association
between income-increasing abnormal accruals and ex post auditor litigation. Barron et al.
(2001) find that auditor assessments of litigation risk are higher when potential errors overstate financial performance. Palmrose and Scholz (2004) show that 83% of restatements that
elicit litigation involve a reversal of previously income-increasing accounting. Accordingly,
auditors consider the risks of earnings management in their assessment of litigation risks and
incorporate these risks into their pricing decisions. In particular, Houston, Peters, and Pratt
(1999) show that when the risk of irregularities is high, auditors add a risk premium to the
audit fee to compensate them for bearing the business risk that cannot be controlled by gathering additional audit evidence. Johnstone and Bedard (2001) and Bedard and Johnstone
(2004) both find that, in addition to planning more audit hours, auditors charge higher billing
rates for clients who resort to earnings manipulation and fraud as a premium to compensate
them for costs related to potential litigation and reputational damage.
In light of the above discussion, it is reasonable to assume that intensified control risk
gives rise to higher litigation risk due to the prevalence of earnings management. This

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Jiang and Son

results in auditors charging a higher risk premium to reduce their exposure to future litigation as well as applying more audit effort toward detecting misstatements.
Focusing on the risk premium component of audit fees, we posit that risk premiums
increase (decline) in a manner consistent with weakened (strengthened) internal controls.
More specifically, we expect that when a companys internal controls deteriorate (i.e.,
receiving a clean 404 opinion in the prior year and an adverse opinion in the current year),
auditors respond to heightened control risk by increasing the risk premium.7 By similar
logic, the remediation of internal control weakness (i.e., receiving an adverse 404 opinion
in the prior year and a clean opinion in the current year) should result in lower risk premiums, thus reflecting reduced control risk. Our first hypothesis is stated in an alternative
form as follows:
Hypothesis 1: Changes in risk premiums are associated with changes in the internal
control quality as evidenced by the SOX 404 internal control opinions.
Our next hypothesis considers the impact of the severity of internal control weakness on
risk premium adjustments. General (company-level) material weaknesses, such as weakness
in the control environment or the overall financial reporting process, represent a more serious concern regarding the reliability of financial reporting than account-specific weaknesses (SWs) that are related to specific account balances or transaction-level processes.
While account-specific material weaknesses are identifiable and correctible by auditors
through substantive testing, company-level weaknesses are more difficult to audit
around (Ettredge, Li, & Sun, 2006). In other words, GWs are more likely to be associated
with irregularities that are difficult to detect and correct.
Consistent with this view, Doyle et al. (2007) find that only GWs are significantly associated with lower accrual quality. Hammersley, Myers, and Shakespeare (2008) find that
stock returns are significantly more negative when internal control weaknesses are company level as investors are more concerned about the potential for material misstatements.
Management fraud is also more likely to occur at firms with company-wide internal control
problems because management is more capable of overriding control procedures (Ettredge,
Heintz, Li, & Scholz, 2011). Hence, compared with account-specific material weaknesses,
general internal control problems pose greater audit and litigation risks to auditors.
Accordingly, we hypothesize that auditors recognize the difference in the level of risks represented by these two types of material weaknesses and adjust the risk premium in a
manner consistent with the degree of severity of internal control weaknesses. We state our
second hypothesis in an alternative form as follows:
Hypothesis 2: Changes in risk premiums vary depending on the severity of the
underlying internal control problems.

Research Design
Regression Model
Our hypotheses relate changes in internal control opinions to changes in risk premiums. To
test the hypotheses, we estimate the following audit fee model:

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Journal of Accounting, Auditing & Finance

DAF = b0 + b1 GB + b2 BG + b3 BB + b4 DAD + b5 DSIZE + b6 DFRGN
+ b7 DSEGNO + b8 DABACC + b9 DRECV + b10 DINVT
+ b11 DOPIN + b12 DDY + b13 DCHG + b14 DLVRG + b15 DLOSS
+ b16 DCATA + b17 DEBIT + b18 DQUICK + b19 DREST + b20 DERROR
+ b21 DLIT + b22 DLAAC + industry dummies + yeardummies + e:
1

The dependent variable is the change in the natural logarithm of audit fees (DAF).8 Our
variables of primary interest are a group of indicator variables representing the change in the
level of control risk. We classify our sample firms into four groups based on the Section 404
disclosures in two successive years: (a) good internal controls in both years (GG); (b) good
internal controls in a given year, followed by bad controls next year (GB); (c) bad internal
controls in a given year, followed by good controls next year (BG); and (d) bad internal controls in both years (BB). The terms good and bad refer to unqualified and qualified SOX 404
opinions, respectively.9 As GG serves as the reference group for the test, the other three test
variables measure the change in audit fees relative to that of GG firms.
To test whether audit firms adjust risk premiums in addition to altering their audit effort,
we include in our model the audit delay measure (DAD) to control for the effect of audit
effort. In its change form, DAD is calculated as the change in the number of days between
the fiscal year-end and the date of audit report.10 There exists a logical relationship
between audit delay and the scope and extent of audit work as the more (fewer) hours an
engagement consumes, the longer (shorter) the audit report lag will be. Knechel and Payne
(2001) provide direct evidence that a lengthy audit delay is due to more audit hours and
efforts being expended on an engagement. Ettredge et al. (2006) find that the reporting of
internal control weaknesses is strongly associated with longer audit delay, indicating that
the audit delay measure captures a significant portion of the extended audit work.11 The
inclusion of DAD in the audit fee model thus enables us to isolate the effects of audit
effort and risk premium, two distinct components of the audit fee. With DAD capturing the
effect of audit effort, the control risk indicator variables (i.e., GB, BG, and BB) represent
the portion of audit fee change attributable to risk premium adjustment beyond that
explained by altered audit effort. A risk premium for control risk in the audit fee can be
inferred if the coefficients on GB, BG, and BB are statistically significant after audit effort
(DAD) is controlled.
For completeness, we also estimate the audit delay model where the change in audit
delay (DAD) serves as the dependent variable, as depicted below:
DAD = Explanatory variables in Equation 1 excluding DAD:

In light of the finding by Ettredge et al. (2006) that the presence of material weaknesses
is associated with longer audit delays, we expect the coefficients on GB, BG, and BB to be
highly significant if the audit delay measure captures the level of audit effort associated
with managing control risk.
We also include in our models as control variables a comprehensive set of firm and auditor
characteristics that have been identified as determinants of audit fees from prior studies. All
control variables are measured as the change from year t 2 1 to year t. Hay, Knechel, and
Wong (2006) show that more complexity in clients business leads to higher audit fees. We
control for audit complexity by including firm size (DSIZE), foreign sales (DFRGN), number

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Jiang and Son

of segments (DSEGNO), accounts receivables (DRECV), and inventory (DINVT). We also

include the absolute value of discretionary accruals (DABACC) and accruals quality
(DERROR)12 to control for information risk, which has been shown in prior literature to be
positively associated with audit fees (Simunic, 1980). Prior research has also found that audit
fees are an increasing function of clients litigation risks (Simunic & Stein, 1996). We thus
include leverage (DLVRG), loss (DLOSS), liquidity (DCATA), earnings before interest and
taxes (DEBIT), quick ratio (DQUICK), and a direct measure of litigation risk (DLIT).13
Furthermore, we control for financial statement audit opinion (DOPIN) because firms receiving
modified audit opinions require more audit efforts (Simunic & Stein, 1996). As staff constraints
usually occurring in the peak season may lead to increased audit fees and delays (Francis,
Reichelt, & Wang, 2005), an indicator variable (DDY) is included to capture the effect of
December fiscal year-end. We include auditor switch (DCHG) to account for the possible
low-ball practice (DeAngelo, 1981) in initial audits. We also include an indicator variable
for restatement (DREST) because it has been shown to be associated with longer audit delay,
which, in turn, could affect audit fees (Feldmann, Read, & Abdolmohammadi, 2009). To
account for differential filing deadlines, we include a dummy variable for large accelerated
filers (DLACCL).14 Finally, we control for industry (industry dummies) and year effects (year
dummies).15 Definitions and data sources of the variables are summarized in Table 1.

Data
We begin with a sample of firms for which SOX 404 opinions are available in Audit
Analytics for the years 2004-2011. First, we require all firms to have internal control data
for 2 consecutive years, which eliminates 20,014 observations. Next, we exclude 2,016
observations with missing data on audit fees or audit report dates. To remain in the sample,
firms must also have available Compustat data required for our tests. These restrictions
result in a final sample of 26,940 firm-year observations.16 Panel A of Table 2 summarizes
the procedure of sample selection.
Panel B of Table 2 displays the sample distribution by year-to-year change in Section 404
internal control opinions. Most firms (89.19%) in our sample report effective internal control in
2 successive years (GG firms). As shown in column BG, about 4.77% of the sample firms
remedied internal control weaknesses in the year following the prior material weakness disclosure,17 whereas approximately 3.30% of the firms reported a deterioration in their internal controls as revealed in the GB column.18 Untabulated results show that non-remediation firms are
typically those reporting a large number of weaknesses and certain types of weaknesses (such
as weak control environment) that are more difficult to resolve. We also observe that the percentage of firms reporting consecutive clean opinions (GG) increases steadily over the sample
period, with the percentage of GB, GB, and BB firms exhibiting a downward trend.
Panel C reports sample distribution by industry. The industry group with the largest representation in our sample is manufacturing, followed by financials and services. Our samples
industry composition is closely aligned to the industry composition in the Compustat database.

Empirical Results
Descriptive Statistics
Panel A of Table 3 presents descriptive statistics for the sample data.19 The mean and
median of changes in audit fees (DAF) are US$31,192 and US$6,471, respectively, indicating that, on average, audit fees increased over the period 2004-2011. Interestingly, the

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Journal of Accounting, Auditing & Finance

Table 1. Variable Definitions.

Variable

Definition

Dependent variable
DAF
Changes in the natural logarithm of audit fees (AUDIT_FEES; source Audit
Analytics)
DAD
Changes in audit delays, where audit delay is measured as the number of days
between the fiscal year-end and the audit report date (SIG_DATE_OF_OP_S;
source Audit Analytics)
Test variables
GB
1 if a firm receives an unqualified SOX 404 opinion in t 2 1 and an adverse
opinion in t, and 0 otherwise (IC_IS_EFFECTIVE; source Audit Analytics)
BG
1 if a firm receives an adverse SOX 404 opinion in t 2 1 and an unqualified
opinion in t, and 0 otherwise (IC_IS_EFFECTIVE; source Audit Analytics)
BB
1 if a firm receives an adverse SOX 404 opinion in both t 2 1 and t, and 0
otherwise (IC_IS_EFFECTIVE; source Audit Analytics)
GG
1 if a firm receives an unqualified SOX 404 opinion in both t 2 1 and t, and 0
otherwise (IC_IS_EFFECTIVE; source Audit Analytics)
Control variables
DSIZE
Change in the natural logarithm of the firms market value of equity (PRCC_F 3
CSHO; source Compustat)
DFRGN
Change in the ratio of sales made by foreign subsidiaries to total sales (PIFO /
[PIFO + PIDOM]; source Compustat)
DSEG
Change in the number of business segments (BUSSEG; source Compustat)
DABACC
Change in the absolute value of total accruals divided by total assets ([IBC
OANCF] / AT; source Compustat)
DERROR
Changes in accruals quality, where accruals quality is the absolute value of the
error terms estimated from regressions of current accruals on prior, current,
and future operating cash flows, changes in sales revenue, and property, plant,
and equipment. (Francis, LaFond, et al., 2005)
DRECV
Change in the ratio of total receivables to total assets (RECT / AT; source
Compustat)
DINVT
Change in the ratio of total inventory to total assets (INVT / AT; source
Compustat)
DOPIN
1 if a firm receives an unqualified opinion in t but not in t 2 1, and 0 otherwise
(AUOP; source Compustat)
DDY
1 if a firm has a December fiscal year-end in t but not in t 2 1, and 0 otherwise
(FYR; source Compustat)
DBIGN
1 if a firm is audited by a Big N in t but not in t 2 1, and 0 otherwise (AU; source
Compustat)
DCHG
1 if a firm switches its auditor in t, and 0 otherwise (source Audit Analytics)
DLIT
Litigation score (Rogers & Stocken, 2005)
DLVRG
Change in the ratio of debt to total assets (LT / AT; source Compustat)
DLOSS
1 if earnings before extraordinary items is less than 0 in t but not in t 2 1, and 0
otherwise (IB; source Compustat)
DCATA
Change in the ratio of current assets to total assets (ACT / AT; source Compustat)
DEBIT
Change in the ratio of earnings before interest and tax to total assets (EBIT / AT;
source Compustat)
DQUICK
Change in the ratio of current assets (less inventory) to current liabilities
([ACT INVT] / LCT; source Compustat)
DREST
1 if a firm announces a restatement in t, but not in t 2 1, and 0 otherwise
(RES_ACCOUNTING; source Audit Analytics) and
DLACCL
1 if a firm is a large accelerated filer, and 0 otherwise (ACCEL_FILER_LARGE;
source Audit Analytics)
Note. The data item and data source for each variable are included in square brackets.

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Jiang and Son

Table 2. Sample Selection Procedure and Sample Distribution.

Panel A: Sample Selection Procedures.
No. of firm years
All firms with IC on Audit Analytics (2005-2011)
Less:
Firms without a prior year internal control opinion
Firms with missing data on audit fee or audit report date
Firms with missing data on Compustat
Final sample

53,138
20,014
2,016
4,168
26,940

Panel B: Sample Distribution by the Type of Change in Internal Control Effectiveness.

Year
2005
2006
2007
2008
2009
2010
2011
Total

GG
1,873
2,714
3,152
3,884
4,231
4,130
4,049
24,033

GB

(80.73%)
(83.64%)
(87.51%)
(89.16%)
(91.74%)
(92.68%)
(93.10%)
(89.19%)

118
162
139
124
122
107
116
888

(5.09%)
(4.99%)
(3.86%)
(2.85%)
(2.65%)
(2.40%)
(2.67%)
(3.30%)

BG
222 (9.57%)
246 (7.58%)
193 (5.36%)
234 (5.37%)
162 (3.51%)
128 (2.87%)
100 (2.30%)
1,285 (4.77%)

BB
107
123
118
114
97
91
84
734

(4.61%)
(3.79%)
(3.28%)
(2.62%)
(2.10%)
(2.04%)
(1.93%)
(2.72%)

Total
2,320
3,245
3,602
4,356
4,612
4,456
4,349
26,940

Note. Sample firms are classified into four groups based on the Section 404 disclosures in two successive years: (a)
good internal controls in both years (GG); (b) good internal controls at t 2 1 and bad internal controls at t (GB); (c)
bad internal controls at t 2 1 and good internal controls at t (BG); and (d) bad internal controls in both years (BB).

Panel C: Sample Distribution by Industry.

SIC codes

Industry

1000-1999
2000-2999
3000-3999
4000-4999
5000-5999
6000-6999
7000-9999
Total

Mining, construction
ManufacturingFood, textiles, lumber, chemicals
ManufacturingRubber, metal, machinery, equipment
Transportation, communication, utilities
Wholesale, retail
Financials
Services

No. of Observations
1,833
3,990
6,117
2,677
2,052
6,164
4,107
26,940

Note. SIC = Standard Industrial Classification.

change in audit delay (DAD) has a negative mean of 20.5556 and a median of 0, suggesting a small decrease in the average audit delay over the sample period. This is probably
due to the learning effects as auditors became more familiar with audits of internal controls,
which started in 2004, when a substantial increase in audit delay compared with 2003 was
observed (Ettredge et al., 2006).20
Panel B shows the mean values of changes in audit fees and audit delays separately for
the four groups. We observe a large increase in audit fees for GB firms (21.12%) and a
moderate decline for BG firms (25.98%). Note that the magnitude of fee decline for BG
firms is considerably smaller than the magnitude of increase for GB firms, suggesting that
there is some stickiness in audit fees (Munsif et al., 2011). Once an auditor increases
the audit fee in response to ineffective internal controls, there is some resistance against
reducing the fees to a great extent.

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

10

Journal of Accounting, Auditing & Finance

Table 3. Descriptive Statistics.

Panel A: Descriptive Statistics.
Variable
DAF (in US$000s)
DAF (in log)
DAD
DSIZE
DFRGN
DSEGNO
DABACC
DRECV
DINVT
DOPIN
DDY
DBIGN
DCHG
DLVRG
DLOSS
DCATA
DEBIT
DQUICK
DREST
DERROR
DLIT
DLACCL

SD

25th

Median

75th

31.1915
0.0221
20.5556
0.0587
0.0098
20.0177
0.0002
20.0010
0.0006
0.1605
0.0016
0.0163
0.0556
0.0108
0.0986
20.0033
20.0052
20.0717
0.0677
20.0808
0.0163
0.4285

868.9301
0.3009
9.3802
0.2436
0.6038
0.5909
0.1167
0.0383
0.0253
0.3670
0.0395
0.1267
0.2292
0.1031
0.2982
0.0719
0.1009
1.3422
0.2512
0.1632
0.3328
0.4949

296.0000
20.1014
23.0000
20.0399
0.0000
0.0000
20.0279
20.0141
20.0022
0.0000
0.0000
0.0000
0.0000
20.0289
0.0000
20.0195
20.0234
20.2140
0.0000
20.1904
20.1557
0.0000

6.4710
0.0012
0.0000
0.0457
0.0000
0.0000
0.0002
0.0000
0.0000
0.0000
0.0000
0.0000
0.0000
0.0005
0.0000
0.0000
20.0004
0.0000
0.0000
20.0158
0.0311
0.0000

118.4000
0.1211
2.0000
0.1450
0.0000
0.0000
0.0275
0.0135
0.0041
0.0000
0.0000
0.0000
0.0000
0.0367
0.0000
0.0214
0.0191
0.1809
0.0000
0.0154
0.1967
0.0000

Panel B: Mean Values of Audit Fee and Audit Delay Changes.

GG

GB

BG

BB

DAF in US$000s 27.6564 (1.21%)

310.3077 (21.12%) 2118.4702 (25.98%) 81.0936 (3.27%)
(% change)
DAD (% change) 20.4667 (20.71%)
6.7804 (9.52%)
26.8031 (28.75%) 21.4019 (21.71%)
Panel C: Correlation Matrix for Test Variables.

DAF
DAD

GG

GB

BG

BB

.0449 ( \ .0001)
2.0272 ( \ .0001)

.1005 ( \ .0001)
.1444 ( \ .0001)

2.0430 ( \ .0001)
2.1491 ( \ .0001)

.0318 ( \ .0001)
2.0151 (.0132)

Note. See Table 1 for variable definitions.

We also find that BB firms experience a significant increase in audit fees (3.27%), suggesting the auditors are particularly concerned about the heightened control risk associated
with firms failing to remediate internal control weaknesses, and hence seek to reduce the
risk by assessing higher fees. Audit firms continue to increase fees for GG firms (1.21%),
probably reflecting the overall upward trend in audit fees during our test period.
Turning to the audit delay measure, GB and BG firms have, on average, a 9.52%
increase and an 8.75% decrease in audit delay, respectively. The results are generally consistent with those for audit fees, supporting the notion that a significant portion of fee
change is attributable to the variation in audit effort.

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Jiang and Son

11

Panel C presents partial correlations between audit fees, audit delay, and the four internal control groups. GB has a positive association with both DAF (.1005) and DAD (.1444),
whereas BG is negatively correlated with DAF (2.0430) and DAD (2.1491). We also find
that both GG and BB are positively (negatively) correlated with DAF (DAD). These results
are generally consistent with those reported in Panel B. In the appendix, we present the correlation matrix for all control variables used in our tests.
Overall, the univariate results provide evidence consistent with prior studies (Ettredge
et al., 2006; Hoag & Hollingsworth, 2011). We next present a multivariate regression analysis to provide a direct test of our risk premium hypotheses.

Regression Results of Changes Model

Table 4 reports the ordinary least squares (OLS) regression results from the changes
model.21 For all regressions, the statistical inferences are based on the heteroskedasticityconsistent variancecovariance matrix (White, 1980).22 The first column presents the audit
delay model. We find that GB firms experience an average increase of 6.7113 days in audit
delay, whereas BG firms have their average reporting lag shortened by 6.1872 days, after
controlling for other determinants. To the extent that audit delay proxies for audit effort,
these results suggest that auditors alter the scope and extent of audit work in response to
the assessed control risk.23
To investigate whether auditors also manage control risk by adjusting risk premiums, we
estimate an audit fee model where the change in audit delay is included to control for audit
effort. Column 2 presents the estimated results. As expected, the change in audit fees is an
increasing function of the change in audit effort, as evidenced by the significantly positive
coefficient on DAD. Importantly, after controlling for changes in audit effort, we find significant coefficients for all the internal control opinion groups. Specifically, the positive
and significant coefficient on GB indicates that firms reporting deterioration in internal
control incur additional audit fees beyond the costs associated with extended audit effort.
In contrast, the negative and significant coefficient on BG indicates that there is an incremental reduction in audit fees over and above the fee decrease due to the reduced amount
of audit work for firms remediating previously disclosed internal control weaknesses. Note
that the size of fee decline for BG firms is considerably smaller than that of fee increase
for GB firms (20.012 vs. 0.1344), suggesting that auditors are more cautious about lowering fees for remediation clients. These results provide evidence that auditors manage control risk by adjusting risk premiums as well as audit efforts.
We also observe that the coefficient on BB is negative in the audit delay model but positive in the audit fee model. This implies that the fee increase for firms failing to remediate
previously disclosed material weaknesses stems primarily from a rise in the risk premium
rather than greater audit efforts. This might be because auditors become more efficient at
dealing with the same internal control problems for these clients, and thus the actual audit
hours could decrease (Hoag & Hollingsworth, 2011). At the same time, the continued existence of material weaknesses raises serious concerns about the reliability of financial
reporting, and hence heightens the litigation risk. Auditors might resort to charging a
higher risk premium to cover potentially higher incremental costs associated with such
clients.24
Collectively, these findings lend support to Hypothesis 1 and provide new insights into
the driving factors underlying the previously documented relationship between changes in
audit fee and changes in quality of internal control.25

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

12

Journal of Accounting, Auditing & Finance

Table 4. OLS Regression Results of Changes in Audit Delays and Audit Fees on Changes in Internal
Control Quality.
Audit delay model (1)
Variable
Intercept
GB
BG
BB
DAD
DSIZE
DFRGN
DSEGNO
DABACC
DRECV
DINVT
DOPIN
DDY
DBIG
DCHG
DLVRG
DLOSS
DCATA
DEBIT
DQUICK
DREST
DERROR
DLIT
DACCL
Industry dummies
Year dummies
F value
Adjusted R2
N

Audit fee model (2)

Expected sign

Coefficient

p value

Coefficient

p value

+ /2
+
2
+ /2
+
+
+
+
+
+
+
2
+
+
2
+
+
+
2
2
+
+
+
2

21.1968
6.7113
26.1872
21.0479

.0100
\ .0001
\ .0001
.0030

0.9881
0.0188
20.1745
2.0085
22.6782
1.0349
20.3460
22.8192
0.9032
20.8829
4.7136
1.1569
23.6598
23.3235
0.0381
0.1942
0.0199
0.5729
1.5529
Included
Included
21.53

.0001
.8382
.0637
.0001
.0771
.6445
.0254
.0450
.0469
.0003
\ .0001
\ .0001
\ .0001
\ .0001
.4303
.4051
.9637
.0020
\ .0001

0.0019
0.1344
20.0212
0.0895
0.0025
0.3257
0.0053
0.0065
0.1222
0.2051
0.2426
20.0197
0.2561
0.1058
20.1969
0.0504
0.0266
20.1598
20.1614
20.0067
0.0455
0.0073
0.0340
20.0012
Included
Included
44.45

.8956
\ .0001
.0125
\ .0001
\ .0001
\ .0001
.0630
.0254
.0001
\ .0001
.0005
\ .0001
\ .0001
\ .0001
\ .0001
.0077
\ .0001
\ .0001
\ .0001
\ .0001
\ .0001
.5910
\ .0001
.7550

\ .0001
.0655
26,940

\ .0001

.1304
26,940

Note. All p values are based on two-tailed tests. See Table 1 for variable definitions. OLS = ordinary least squares.

Severity of Internal Control Weaknesses

Next, we test Hypothesis 2 by considering the severity of internal control weakness in our
analyses. In particular, we investigate the extent to which the risk premium adjustment differs by the type of material weakness. Following Munsif et al. (2011), we classify an internal control weakness into GW or SW based on the material weakness categories compiled
by Audit Analytics. Depending on the Section 404 opinions received in 2 consecutive
years, a company is classified into one of these nine distinct groupsGWGW, GWSW,
GWNW, SWSW, SWGW, SWNW, NWSW, NWGW, and NWNW, where GW, SW, and NW
represent general weakness, account-specific weakness, and no weakness, respectively. For
example, a GWNW firm is one that discloses a GW in year t 2 1, and subsequently
receives a clean 404 opinion in year t. Such a classification scheme allows for cases,
wherein the internal control disclosure shifts from a more severe type to a less severe one
(GWSW) or vice versa (SWGW).

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Jiang and Son

13

Table 5 displays the regression results of the change in audit delay or audit fee on the
change in Section 404 internal control opinions categorized by the level of severity. As
NWNW (firms reporting no internal control weakness in both t and t 2 1) serves as the reference group, the parameter estimates for all other groups are measured relative to NWNW
firms.
The results from the audit delay model are presented in column 1. We make several
important observations. First, the coefficients for both NWGW and NWSW are significantly
positive, indicating that a breakdown in the firms internal control leads to an increase in
audit delay. A further test of the difference between these two parameter estimates shows
that NWGW is significantly more positive than NWSW (7.0671 vs. 5.4913, p value \
.0001), suggesting that auditors exert more effort in audits of firms with company-level
control weaknesses. Second, we find that the coefficients on GWNW and SWNW are both
negative and significant, indicating that remediation of material weaknesses results in a
decrease in audit delay. The F test shows that the reduction in audit effort is greater for
firms correcting general internal control problems than account-specific problems
(26.7897 vs. 24.3492, p value \ .0001). Third, the coefficient on GWSW (SWGW) is significantly negative (positive). This suggests that auditors adjust the scope and extent of
audit work even when there is only a partial remediation (degradation) of internal control
deficiencies. Finally, we observe a marginally significant coefficient on GWGW and an
insignificant coefficient on SWSW, indicating little change in audit effort when the severity
of control weaknesses remains unchanged.
We next discuss the results from the audit fee model presented in column 2, where audit
delay is included to control for the effect of audit effort. First, as with the audit delay
model, the coefficients on NWGW and NWSW are both significantly positive, and the difference between the two coefficient estimates is statistically significant (0.1662 vs. 0.0639,
p value \ .0001). The result suggests that auditors charge higher risk premiums to clients
who report GWs than to those disclosing account-specific problems, consistent with the
notion that GWs are more difficult to audit around and rectify and thus reflect a higher
level of risk. Second, we find negative and significant coefficients on GWNW and SWNW,
suggesting that auditors lower risk premiums in response to reduced control risk. However,
we find no significant difference between GWNW and SWNW. This seems to suggest that
auditors consider the type of weakness when adjusting the risk premium upward, but do
not do so in the case of a downward adjustment. While only a conjecture on our part, a
possible explanation for the asymmetric pattern of risk premium adjustment is that auditors
are not confident about clients who have previously disclosed internal control problems and
thus do not distinguish firms by the type of weakness when reducing the risk premium.
This is consistent with auditors taking a conservative approach in their pricing decisions.
Third, the coefficient on SWGW is significantly positive, whereas the coefficient on GWSW
is insignificant, indicating that auditors raise the risk premium when the internal control
disclosure changes from a less severe type to a more severe one, but are reluctant to lower
the premium in cases of partial remediation. The evidence again supports the view of auditor conservatism. Finally, we find positive and significant coefficients on both GWGW and
SWSW. Contrasting this with the results for GWGW and SWSW in the audit delay model,
the continued existence of material weaknesses seems to raise serious concerns about the
reliability of financial reporting, and hence heightened litigation risk. However, instead of
performing additional substantive tests, auditors rely on charging higher risk premiums to
compensate themselves for bearing a higher risk.

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

14

Journal of Accounting, Auditing & Finance

Table 5. OLS Regression Results of Changes in Audit Delays and Audit Fees on Changes in Internal
Control Disclosures Categorized by the Severity of Internal Control Weakness.
Audit delay model (1)
Variable
INTERCEPT
NWSW
NWGW
GWNW
SWNW
GWSW
SWGW
GWGW
SWSW
DAD
DSIZE
DFRGN
DSEGNO
DABACC
DRECV
DINVT
DOPIN
DDY
DBIG
DCHG
DLVRG
DLOSS
DCATA
DEBIT
DQUICK
DREST
DERROR
DLIT
DACCL
Industry dummies
Year dummies
F value
Adjusted R2
n

Audit fee model (2)

Expected sign

Coefficient

p value

Coefficient

p value

+ /2
+
+
2
2
2
+
+
+
+
+
+
+
+
+
+
2
+
+
2
+
+
+
2
2
+
+
+
2

21.2180
5.4913
7.0671
26.7897
24.3492
26.1703
2.8114
20.7318
1.2677

.0089
\.0001
\.0001
\.0001
\.0001
\.0001
.0067
.0784
.2180

0.9559
0.0108
20.1939
1.9368
22.7283
1.0136
20.3792
22.7143
0.9266
20.9554
4.7242
1.1380
23.6856
23.2451
0.0334
0.2922
20.0148
0.5752
21.5484
Included
Included
20.28

.0002
.9066
.0395
.0002
.0718
.6515
.0144
.0537
.0416
.0001
\.0001
\.0001
\.0001
\.0001
.4898
.2177
.9731
.0020
\.0001

0.0009
0.0639
0.1662
20.0118
20.0259
0.0352
0.1132
0.1006
0.0888
0.0025
0.3250
0.0054
0.0064
0.1207
0.20082
0.2357
20.0198
0.2577
0.1050
20.1976
0.0504
0.0264
20.1595
20.1597
20.0067
0.0461
0.0075
0.0341
20.0012
Included
Included
42.23
.1304
26,940

.9478
\.0001
\.0001
.0648
.0326
.2082
.0004
\.0001
.0053
\.0001
\.0001
.0598
.0275
\.0001
\.0001
.0007
\.0001
\.0001
\.0001
\.0001
.0076
\.0001
\.0001
\.0001
\.0001
\.0001
.5809
\.0001
.7521

\.0001
.0649
26,940

\.0001

Note. NW, GW, and SW refer to no weakness, general weakness, and account-specific weakness, respectively. A
combination of any two of the three internal control opinions indicates the change in internal control status. For
example, GWNW represents a firm reporting GW in year t 2 1 and subsequently receiving a clean 404 opinion in
year t. p values are based on two-tailed tests. See Table 1 for the definition of other variables. OLS = ordinary
least squares.

To summarize, the findings reported in Tables 4 and 5 jointly present a fairly consistent
picture of how changes in the effectiveness of internal control relate to changes in auditor
efforts and risk premiums. We find strong evidence that the audit fee bears a risk premium
for control risk. It is worth noting that although we find some evidence that the adjustment
of risk premium differs by the type of material weakness, such an adjustment is more sensitive to increases in control risk, reflecting conservatism on the part of the auditor.

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Jiang and Son

15

Additional Analyses
FamaMacBeth Regressions
This study covers a much longer sample period (2004-2011) than those used in prior
research. The study period, however, was characterized by increasing regulatory scrutiny
on auditing practices and the issuance of several new auditing standards and SEC rules. To
assess the sensitivity of our results to these regulatory changes, we rerun our main analyses
by estimating annual regressions. Following Fama and MacBeth (1973), we average out the
regression coefficients across the 7 yearly regressions for years 2005-2011, and compute t
statistics using the standard error of the coefficients. The results are presented in Table 6.
For brevity, we report only test variables. The estimated results are generally consistent
with the results obtained from the pooled sample.

Size Analysis
The documented relation between audit fees and quality of internal control could vary
across different sized firms. Prior research has shown that large clients (or Big N auditors)
have greater negotiation power than small clients (or non-Big N auditors). The relation
could also vary depending on the SEC filer type. Effective November 15, 2006, large accelerated filers have been subject to the newly instituted 60-day Form 10-K filing deadline,
whereas small accelerated filers continue to comply with the 75-day deadline. The different
requirements regarding the filing deadline might yield differential results across filer types.
We therefore perform separate tests to investigate whether size could affect our results.
First, we partition the sample based on the auditor type (Big4 vs. non-Big4). In our second
test, we partition our sample based on whether the firm is a large or small accelerator filer,
where a large accelerated filer is defined as a company that has a public float of US$700
million or more. Our final test partitions firms into two subgroups based on their fiscal
year-end total assets.
The results are presented in Panels A, B, and C of Table 7, respectively. In general, the
inferences drawn from size analyses are qualitatively similar to our primary analysis. In
combination, these three tests suggest that our results are not driven by any subgroup of the
sample.

Reporting Periods
The PCAOB issued Auditing Standard No. 5 (AS5) effective for fiscal year-ends on or
after November 15, 2007 (PCAOB, 2004, 2007). AS5 prescribes a top-down, risk-based
audit approach, which the PCAOB believes will increase audit efficiency and result in significant cost savings. Consistent with this expectation, recent research has documented a
decrease in audit fees, following the implementation of AS5 (Doogar, Sivadasan, &
Solomon, 2010; Krishnan, Krishnan, & Song, 2011).
To assess the potential impacts of AS5 on audit pricing behavior, we partitioned our
sample period into AS2 versus AS5 and reran our analyses separately for these two subperiods. As shown in Panel D of Table 7, the results are qualitatively similar across the two
periods and our inferences remain unchanged.

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

16

Journal of Accounting, Auditing & Finance

Table 6. FamaMacBeth Regression Results of Changes in Audit Delays and Audit Fees on Changes
in Internal Control Quality.
Panel A: Regression Results for Hypothesis 1.
Audit delay model (1)
Variable

Audit fee model (2)

Expected sign

Coefficient

p value

Coefficient

p value

+
2
+ /2

6.4122
25.5034
0.7568

.0008
.0004
.4496

0.1274
20.0146
0.0809

.0005
.0512
.0044

GB
BG
BB

Panel B: Regression Results for Hypothesis 2.

Audit delay model (1)

Audit fee model (2)

Variable

Expected sign

Coefficient

p value

Coefficient

p value

NWGW
NWSW
GWNW
SWNW
GWSW
SWGW
GWGW
SWSW

+
+
2
2
2
+
+
+

6.4549
5.3032
26.1180
23.6982
23.1804
3.0080
20.5718
1.1887

.0021
.0012
.0009
.0002
.0913
.0950
.6377
.1168

0.1538
0.0606
20.0060
20.0187
0.0520
0.0923
0.0908
0.0883

.0019
.0144
.0811
.0315
.4023
.0721
.0134
.0398

Note. This table presents the regression results using the FamaMacBeth procedure. The reported coefficient
estimates are obtained by averaging the coefficient estimates from annual regressions for years 2005-2011, and
p values are calculated using the standard errors of the annual coefficient estimates. See Table 1 for variable
definitions.

Conclusion
This study addresses the important question as to whether audit fees reflect risk premiums
in the presence of control risk. We hypothesize that increased control risk gives rise to
heightened risk of earnings management, which in turn increases the likelihood of litigation
against auditors. To the extent that the risk of errors and irregularities are high and normal
audit procedures cannot reduce the risks to tolerable levels, a premium is added to the fee
to compensate auditors for potential future litigation losses. We test this hypothesis by
investigating whether control risk provides additional explanatory power over and above
that provided by audit effort in explaining audit pricing.
Employing a changes analysis approach, we find strong evidence of auditors responding
to increased control risk by charging fees above the cost of conducting additional audits.
This suggests that auditors adjust risk premiums as well as audit procedures in face of
increasing control risk. In further analysis, we find some evidence that changes in risk premiums are also associated with changes in the severity of the underlying internal control
problems. Auditors assess a higher risk premium to clients reporting GWs compared with
firms reporting account-specific problems. However, reduction in risk premiums is indistinguishable between these two types of weaknesses when clients remediate their internal control problems. This result is consistent with auditors taking a more conservative approach

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Jiang and Son

17

Table 7. Sensitivity Analyses.

Panel A: Big N Versus Non-Big N.
Big N
Audit delay

Non-Big N
Audit fees

Audit delay

Audit fees

Variable Coefficient

p value

Coefficient

p value

Coefficient

p value

Coefficient

p value

GB
BG
BB

\.0001
\.0001
\.0001

0.1236
20.0154
0.1054

\.0001
.1139
\.0001

4.8006
26.0228
20.2635

\.0001
\.0001
0.6807

0.1550
20.0330
0.0723

\.0001
.0566
.0003

7.8146
26.4172
21.7778

Panel B: Large Accelerated Filers Versus Accelerated Filers.

Large accelerated filers
Audit delay

Small accelerated filers

Audit fees

Audit delay

Audit fees

Variable Coefficient

p value

Coefficient

p value

Coefficient

p value

Coefficient

p value

GB
BG
BB

\.0001
\.0001
\.0001

0.1252
20.0259
0.0474

\.0001
.0611
.0205

6.4288
25.9937
20.5830

\.0001
\.0001
.1827

0.1398
20.0192
0.1002

\.0001
.0813
\.0001

7.6441
26.7589
23.0927

Panel C: Large Versus Small Firms.

Assets  Median
Audit delay

Assets \ Median

Audit fees

Audit delay

Audit fees

Variable Coefficient

p value

Coefficient

p value

Coefficient

p value

Coefficient

p value

GB
BG
BB

\.0001
\.0001
\.0001

0.1262
20.0249
0.0479

\.0001
.0450
.0073

6.4059
26.3249
20.5928

\.0001
\.0001
0.1972

0.1409
20.0222
0.1046

\.0001
0.0624
\.0001

7.0715
26.1849
22.7381

Panel D: AS2 Versus AS5.

Pre-AS5
Audit delay

Post-AS5
Audit fees

Audit delay

Audit fees

Variable Coefficient

p value

Coefficient

p value

Coefficient

p value

Coefficient

p value

GB
BG
BB

\.0001
\.0001
.9675

0.1485
20.0383
0.1016

\.0001
.0161
\.0001

5.1620
24.8393
21.5478

\.0001
\.0001
.0002

0.1208
20.0107
0.0748

\.0001
.0985
\.0001

9.2427
27.8302
20.0280

Note. See Table 1 for variable definitions.

when pricing the control risk for clients that have previously experienced internal control
problems, regardless of the type of weakness involved.
Hogan and Wilkins (2008) call for additional research to investigate whether higher
audit fees in the presence of internal control weakness during the new SOX 404 reporting
regime are due to extended audit effort or a higher risk premium. Our study thus can be
viewed as a direct response to this call. The findings in this study lead us to conclude that
the positive relation between internal control disclosures and audit fees documented in

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

18

Journal of Accounting, Auditing & Finance

prior research (Hoag & Hollingsworth, 2011; Hogan & Wilkins, 2008; Hoitash et al., 2008;
Munsif et al., 2011; Raghunandan & Rama, 2006) is driven by changes in both risk premiums and auditor efforts.
Overall, this study provides a more comprehensive analysis of the dynamics between
audit fee responses and changes in internal control quality, and of the distinct role of audit
effort versus risk premium in explaining the variance in audit fees. The findings reported in
our study further add to our understanding of the impact of SOX 404 requirements on the
cost of internal control auditing.
One limitation of this study is our use of the audit delay measure as a proxy for the
scope and extent of audit work. We acknowledge that audit delay measures audit effort
with some noise. This is because audit delays can be influenced by many other unobservable determinants (e.g., auditors favor to a particular client in scheduling audit time).
Furthermore, we estimate audit delay by counting the number of days from the fiscal yearend to the audit report date because actual audit delaysthe number of days between the
start and finish of audit workare not publicly available. Ideally, such a study should be
conducted using audit hours and rates data obtained directly from audit firms working
papers and internal billing records. Therefore, our conclusions regarding audit effort versus
risk premium are valid to the extent that the audit delay measure adequately captures the
effects of audit effort.

Appendix
Pearsons Correlations for Control Variables.
(2) (3)

(4) (5) (6)

(7)

(8)

(9)

(10) (11) (12) (13) (14) (15) (16) (17) (18) (19) (20)

(1) .10 .24 .01 .03 .03 .00 .01 2.05 .04 2.13 .06 .03 2.09 2.02 2.05 .04 2.02 .09 .02
(2)
2.01 2.01 .00 .05 2.02 .01 2.03 2.01 2.02 .07 .07 2.04 2.07 2.03 2.02 2.04 .03 2.09
(3)
.02 .06 2.23 2.13 2.11 2.03 .02 .00 2.09 2.12 2.09 .30 .10 2.01 .03 .04 .13
(4)
.00 2.01 .00 .01 .01 2.01 .00 2.01 2.07 .00 .01 2.01 .01 .00 2.01 .00
(5)
2.01 .02 .01 .00 .00 .00 .02 .00 2.03 .00 2.02 2.01 2.01 2.01 2.02
(6)
.02 .01 2.01 .00 .01 .27 .21 .05 2.28 2.10 .00 .01 .12 2.02
(7)
.04 .02 .01 .00 .14 .00 .13 .04 2.11 .00 .02 .00 2.01
(8)
2.02 .00 2.01 .09 .03 .05 2.02 2.10 .00 2.01 .07 2.03
(9)
.00 .03 2.03 2.01 .03 .03 .01 .03 .00 2.02 .02
(10)
.00 .01 .01 2.02 .00 2.01 .00 .00 2.01 .00
(11)
.02 .01 2.01 2.01 2.01 .07 2.03 .00 2.09
(12)
.11 2.09 2.27 2.34 .01 .01 .09 2.06
(13)
2.01 2.21 2.02 .03 2.01 .08 2.09
(14)
.14 .38 .00 .00 2.07 .00
(15)
.20 2.02 .00 2.10 .05
(16)
.00 .00 2.03 .04
(17)
.01 .00 2.04
(18)
2.01 .09
(19)
.02
Note. (1) DAF (2) DAD (3) DSIZE (4) DFRGN (5) DSEGNO (6) DABACC (7) DRECV (8) DINVT (9) DOPIN (10) DDY
(11) DCHG (12) DLVRG (13) DLOSS (14) DCATA (15) DEBIT (16) DQUICK (17) DREST (18) DERROR (19) DLIT (20)
DLACCL. Bold text indicates significance at the .10 level or better (two-tailed).

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Jiang and Son

19

Acknowledgments
The authors thank Bharat Sarath, the Editor, and two anonymous referees for their careful reviews
and constructive suggestions. They also thank participants and discussants at the 2011 American
Accounting Association Annual Meeting for helpful comments and suggestions in the early phase of
the paper.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/
or publication of this article.

Funding
The author(s) received no financial support for the research, authorship, and/or publication of this
article.

Notes
1. Note that the two components of audit fees have differential impact on the quality of financial

2.
3.

4.
5.
6.

7.

reporting. While increasing the amount of audit work leads to improved audit quality, assessing
risk premiums does not affect audit quality (Carcello, Hermanson, & Riley, 2002; Lobo & Zhao,
2013).
For example, Ghosh and Lustgarten (2006) argue that changes models yield more accurate
results because they are overall less subject to biases.
Section 404 became effective for fiscal years ending on or after November 15, 2004, for accelerated filers (i.e., Securities and Exchange Commission [SEC] registrants who have market capitalization more than US$75 million). Non-accelerated filers have been required to be compliant
with Section 404(a) since December 15, 2007. The SEC allowed non-accelerated filers to defer
their compliance with Section 404(b) through multiple extensions until fiscal year ending on or
after June 15, 2010. However, the Dodd-Frank Wall Street Reform and Consumer Protection
Act, which was signed into law on July 21, 2010, permanently exempts non-accelerated filers
from the auditor attestation requirement of Section 404(b).
Section 302, which became effective in August 2002, mandates that management evaluate disclosure control processes and procedures, and certify the effectiveness of the controls. However, an
independent audit of the effectiveness of internal control is not required.
The findings are also consistent with SAS55 and SAS78 that prescribe increased substantive testing when controls cannot be relied upon.
It is also possible that firms with internal control deficiencies exhibit greater idiosyncratic risk,
which in turn increases the chances of a dramatic drop in stock price or unexpected corporate
failure that typically triggers third-party litigation (Ashbaugh-Skaife, Collins, Kinney, & LaFond,
2009).
Before the start of the audit, the auditor obtains an understanding of the client-specific business
risk, including control risk, and sets the audit fee at a level that is expected to cover the premium
for risks that cannot be adequately addressed by normal audit procedures as well as the costs of
conducting audits. Although the negotiated audit fee is sticky, it can be altered in response to significant changes in the scope of audit work, such as extra work associated with an internal control issue (Abbott et al., 2006; Hackenbrack, Jenkins, & Pevzner, 2014). Wang, Raghunandan,
and McEwen (2013) also note that, since 2002, fee contracts typically include an adjustment
range of 10% to 15%, which makes it more flexible for the auditor to renegotiate the fee with
the client. Ettredge, Heintz, Li, and Scholz (2011) find that the issuance of an adverse internal
control opinion substantially increases the probability of the auditor dismissal. This provides the
auditor an additional incentive to charge a risk premium to protect against potential loss of fee.

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

20

8.
9.
10.
11.

12.
13.

14.
15.
16.
17.
18.
19.
20.

21.

22.
23.

Journal of Accounting, Auditing & Finance

However, a counterargument is that a qualified opinion can be used as a defense in the event of
future litigation, thus reducing the litigation risk. It thus becomes an empirical question as to
which effect dominates.
We take the natural log of the audit fee variable to correct for violations of normality. The
results, however, are similar when the raw measure of audit fee is used.
Auditors can issue three types of internal control opinions: (a) adverse (or qualified), where at
least a material weakness is present; (b) unqualified (or clean), where no material weaknesses
are present; and (c) disclaimer, where the auditor could not issue an opinion.
Following Sengupta (2004), we retain audit report dates that fall between 7 days and 90 days
after the fiscal year-end to eliminate outliers and/or potential errors in the report dates.
It is worth noting that audit delay measures audit effort with some noise. This is because audit
delays can be influenced by many other unobservable determinants (e.g., auditors favor to the
particular client in scheduling audit time). However, employing a changes model allows us to
effectively control for these unobservable characteristics within the same firm. If any measure
error exists, it would be held constant across time and is not likely to bias our results.
The correlation between DABACC and DERROR (see the appendix) is only .01, indicating that
they are two distinct measures of earnings quality.
The litigation score is calculated based on the regression results from a probit model in Rogers
and Stocken (2005). Litigation score = 25.738 + 0.141 3 market values + 0.284 3 average
daily stock turnover + 0.012 3 market beta 2 0.237 3 buy and hold returns 2 1.340 3 standard deviation of returns + 0.011 3 skewness of returns 2 3.161 3 minimum returns 2 0.025
3 Bio-Technology industry + 0.378 3 Computer hardware industry + 0.075 3 Electronics
industry 2 0.034 3 Retailing industry + 0.211 3 Computer software industry. A higher litigation score indicates a higher probability of litigation.
The SEC implemented the 60-day 10-K filing deadline for large accelerated filers for fiscal years
ending on or after December 15, 2006 (Bronson, Hogan, Johnson, & Ramesh, 2011; Krishnan &
Yang, 2009).
Industry is defined based on the 2-digit Standard Industrial Classification (SIC) code.
Because year 2004 is the first year of Section 404 implementation, the data from this year are
used to calculate the change variables for year 2005 but are not included in the final sample.
To put the remediation rate in perspective, 63.45% (i.e., 1,285 / (1,285 + 734)) of the weakness
firms corrected their internal control deficiencies in the following year.
Ashbaugh-Skaife et al. (2009) note that when a firm is involved in a merger or acquisition, control problems associated with integrating systems of newly acquired entities might lead to deterioration in internal controls.
All continuous variables were winsorized at 1% and 99% levels to reduce the effects of extreme
values.
The statistic could also be skewed by the one-time significant decrease in audit delay from 2005
to 2006 as the accelerated 10-K filing deadline (shortened from 75 to 60 days) for large accelerated filers (i.e., companies with public float of US$700 million or more) became effective for
fiscal years ending on or after December 15, 2006. As discussed in the prior section, we include
a dummy variable for large accelerated filers to control for the differential filing deadline.
The R2 for Models 1 and 2 are .0655 and .1304, respectively, compared with .39 reported in
Hoag and Hollingsworth (2011). Their changes model includes both change and level variables,
resulting in a relatively high explanatory power. We are also able to obtain a similar R2 when we
include the level variables in our model.
An examination of the variance inflation factors (VIFs) for all regressions indicates that the highest value is well below 5.00, suggesting that multicollinearity is not a problem.
It is possible that a longer audit delay is due to a later starting date of the audit. For example,
firms with internal control weaknesses may have the incentive to delay audits to delay the earnings announcements. Consistent with this conjecture, we find that the correlation between

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Jiang and Son

21

changes in audit delay and changes in earnings announcement delay is 19% and statistically significant. However, our inferences remain unchanged after controlling for changes in earnings
announcement delay.
24. We also perform sensitivity tests using alternative audit delay measures. First, we calculate
abnormal audit delay as the residual value from the audit delay model and replace the audit
delay variable with this abnormal measure in the audit fee model. Untabulated results show that
the coefficients on BG, BG, and BB are 0.1507, 20.0367, and 0.0869, respectively, all of which
are statistically significant at the 1% level. Second, we rerun the audit fee regression using the
predicted value of audit delay. Again, our results remain qualitatively the same. The results for
the control variables are generally as expected. We find that audit fees are positively associated
with larger firm size (DSIZE), higher foreign sales (DFRGN), more segments (DSEGNO), higher
information risk (DABACC), higher receivables and inventory (DRECV and DINVT), Big N auditor (DBIG), higher leverage (DLVRG), loss (DLOSS), a restatement in the current year (DREST),
a December fiscal year-end (DDY), and litigation risk (DLIT). In addition, audit fees decrease
when firms increase their liquidity (DCATA and DQUICK), report higher earnings (DEBIT),
receive unqualified opinions (DOPIN), and change their auditors (DCHG).
25. We also conduct sensitivity tests after excluding financial companies (SIC codes 6000-6999)
from our sample and find qualitatively similar results.

References
Abbott, L. J., Parker, S., & Peters, G. F. (2006). Earnings management, litigation risk, and asymmetric audit fee responses. Auditing: A Journal of Practice & Theory, 25(1), 85-98.
American Institute of Certified Public Accountants. (2006). Audit risk and materiality in conducting
an audit (Statement on Auditing Standards No. 107). New York, NY: Author.
Ashbaugh-Skaife, H., Collins, D., Kinney, W., & LaFond, R. (2008). The effect of SOX internal
control deficiencies and their remediation on accrual quality. The Accounting Review, 83,
217-250.
Ashbaugh-Skaife, H., Collins, D., Kinney, W., & LaFond, R. (2009). The effect of SOX internal control deficiencies on firm risk and cost of equity capital. Journal of Accounting Research, 47, 1-43.
Barron, O., Pratt, J., & Stice, J. D. (2001). Misstatement direction, litigation risk, and planned audit
investment. Journal of Accounting Research, 39, 449-462.
Bedard, J. C., & Johnstone, K. M. (2004). Earnings manipulation risk, corporate governance risk, and
auditors planning and pricing decisions. The Accounting Review, 79, 277-304.
Bronson, S. N., Hogan, C. E., Johnson, M. F., & Ramesh, K. (2011). The unintended consequences of
PCAOB auditing standard Nos. 2 and 3 on the reliability of preliminary earnings releases. Journal
of Accounting & Economics, 51, 95-114.
Carcello, J. V., Hermanson, D. R., & Riley, R. A. (2002). Board characteristics and audit fees.
Contemporary Accounting Research, 19, 365-384.
DeAngelo, L. (1981). Auditor independence, low balling, and disclosure regulation. Journal of
Accounting & Economics, 3, 113-127.
Doogar, R., Sivadasan, P., & Solomon, I. (2010). The regulation of public company auditing:
Evidence from the transition to AS5. Journal of Accounting Research, 48, 795-814.
Doyle, J., Ge, W., & McVay, S. (2007). Accrual quality and internal control over financial reporting.
The Accounting Review, 82, 1141-1170.
Ettredge, M. L., Heintz, J., Li, C., & Scholz, S. (2011). Auditor realignments accompanying implementation of SOX 404 reporting requirements. Accounting Horizons, 25, 17-39.
Ettredge, M. L., Li, C., & Sun, L. (2006). The impact of SOX Section 404 internal control quality
assessment on audit delay in the SOX era. Auditing: A Journal of Practice & Theory, 25(2), 1-23.
Fama, E. F., & MacBeth, J. D. (1973). Risk, return, and equilibrium: Empirical tests. Journal of
Political Economy, 81, 607-636.

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

22

Journal of Accounting, Auditing & Finance

Feldmann, D. A., Read, W. J., & Abdolmohammadi, M. J. (2009). Financial restatements, audit fees,
and the moderating effect of CFO turnover. Auditing: A Journal of Practice & Theory, 28(1), 205223.
Francis, J., LaFond, R., Olsson, P. M., & Schipper, K. (2005). The market pricing of accruals quality.
Journal of Accounting & Economics, 39, 295-327.
Francis, J., Reichelt, K., & Wang, D. (2005). The pricing of national and city-specific reputation for
industry expertise in the U.S. audit market. The Accounting Review, 80, 113-136.
Ghosh, A., & Lustgarten, S. (2006). Pricing of initial audit engagements by large and small audit
firms. Contemporary Accounting Research, 23, 333-368.
Hackenbrack, K., Jenkins, N. T., & Pevzner, M. (2014). Relevant but delayed information in negotiated audit fees. Auditing: A Journal of Practice & Theory, 33(4), 95-117.
Hammersley, J. S., Myers, L. A., & Shakespeare, C. (2008). Market reactions to the disclosure of
internal control weaknesses and to the characteristics of those weaknesses under section 302 of the
Sarbanes Oxley Act of 2002. Review of Accounting Studies, 13, 141-165.
Hay, D., Knechel, W. R., & Wong, N. (2006). Audit fees: A meta-analysis of the effect of supply and
demand attributes. Contemporary Accounting Research, 23, 141-192.
Heninger, W. G. (2001). The association between auditor litigation and abnormal accruals. The
Accounting Review, 76, 111-126.
Hoag, M. L., & Hollingsworth, C. W. (2011). An intertemporal analysis of audit fees and section 404
material weaknesses. Auditing: A Journal of Practice & Theory, 30(2), 173-200.
Hogan, C. E., & Wilkins, M. S. (2008). Evidence on the audit risk model: Do auditors increase audit
fees in the presence of internal control deficiencies? Contemporary Accounting Research, 25, 219242.
Hoitash, R., Hoitash, U., & Bedard, J. C. (2008). Internal control quality and audit pricing under the
Sarbanes-Oxley Act. Auditing: A Journal of Practice & Theory, 27(1), 105-126.
Houston, R. W., Peters, M. F., & Pratt, J. H. (1999). The audit risk model, business risk, and audit
planning decisions. The Accounting Review, 74, 281-298.
Johnstone, K., & Bedard, J. C. (2001). Engagement planning, bid pricing, and client response in the
market for initial attest engagements. The Accounting Review, 76, 199-220.
Johnstone, K., Li, C., & Rupley, K. H. (2011). Changes in corporate governance associated with the
revelation of internal control material weaknesses and their subsequent remediation.
Contemporary Accounting Research, 28, 331-383.
Knechel, W., & Payne, J. (2001). Additional evidence on audit report lag. Auditing: A Journal of
Practice & Theory, 20(1), 137-146.
Krishnan, J., Krishnan, J., & Song, H. (2011). The effect of auditing standard No. 5 on audit fees.
Auditing: A Journal of Practice & Theory, 30(4), 1-27.
Krishnan, J., & Yang, J. S. (2009). Recent trends in audit report and earnings announcement lags.
Accounting Horizons, 23, 265-288.
Lobo, G. J., & Zhao, Y. (2013). Relation between audit effort and financial report misstatement:
Evidence from quarterly and annual restatements. The Accounting Review, 88, 1385-1412.
Munsif, A., Raghunandan, K., Rama, D. V., & Singhvi, M. (2011). Audit fees after remediation of
internal control weaknesses. Accounting Horizons, 25, 87-105.
Palmrose, Z.-V., & Scholz, S. (2004). The circumstances and legal consequences of non-GAAP
reporting: Evidence from restatements. Contemporary Accounting Research, 21, 139-180.
Public Company Accounting Oversight Board. (2004). Auditing Standard No. 2: An audit of internal
control over financial reporting performed in conjunction with an audit of financial statements.
Washington, DC: Author.
Public Company Accounting Oversight Board. (2007). Auditing Standard No.5: An audit of internal
control over financial reporting that is integrated with an audit of financial statement.
Washington, DC: Author.
Public Company Accounting Oversight Board. (2010). Auditing Standard No 8: Audit risk (PCAOB
Release no. 2010-2004). Washington, DC: Author.

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015

Jiang and Son

23

Raghunandan, K., & Rama, D. V. (2006). SOX Section 404 material weakness disclosures and audit
fees. Auditing: A Journal of Practice & Theory, 25(1), 99-114.
Rogers, J. L., & Stocken, P. C. (2005). Credibility of management forecasts. The Accounting Review,
80, 1233-1260.
Sankaraguruswamy, S., & Whisenant, J. S. (2004). An empirical analysis of voluntarily supplied
client-auditor realignment reasons. Auditing: A Journal of Practice & Theory, 23(1), 107-121.
Securities and Exchange Commission. (1988). Disclosure amendments to Regulation S-K, Form 8-K
and Schedule 14A regarding changes in accountants and potential opinion shopping situations
(Financial Reporting Release No. 31). Washington, DC: Government Printing Office.
Sengupta, P. (2004). Disclosure timing: Determinants of quarterly earnings release dates. Journal of
Accounting and Public Policy, 23, 457-482.
Simunic, D. A. (1980). The pricing of audit services: Theory and evidence. Journal of Accounting
Research, 18, 161-190.
Simunic, D. A., & Stein, M. T. (1996). The impact of litigation risk on audit pricing: A review of the
economics and the evidence. Auditing: A Journal of Practice & Theory, 15(2), 119-134.
Wang, C., Raghunandan, K., & McEwen, R. A. (2013). Non-timely 10-K filings and audit fees.
Accounting Horizons, 27, 737-755.
White, H. (1980). A heteroskedasticity-consistent covariance matrix estimator and a direct test for
heteroskedasticity. Econometrica, 48, 817-838.

Downloaded from jaf.sagepub.com at University of New England on May 30, 2015