THE ACCOUNTING REVIEW
Vol. 90, No. 4
2015
pp. 1469–1496
American Accounting Association
DOI: 10.2308/accr-50975
Internal Control Quality: The Role of
Auditor-Provided Tax Services
Lisa De Simone
Stanford University
Matthew S. Ege
Texas A&M University
Bridget Stomberg
The University of Georgia
ABSTRACT: We propose that auditor-provided tax services (tax NAS) improve internal
control quality by accelerating audit firm awareness of transactions material to the
financial statements. Using data from 2004 to 2012, we find robust evidence that
companies purchasing tax NAS are significantly less likely to disclose a material
weakness and that this result is not due to auditor independence impairment. A onestandard-deviation increase in tax NAS is associated with approximately a 13 percent
decrease in the rate of material weaknesses relative to the base rate. These results are
robust to tests addressing endogeneity concerns. Additional cross-sectional analyses
reveal expected increased effects of tax NAS on internal control quality (1) after
significant operational changes that require changes to the internal control structure, and
(2) earlier in the relationship with the financial statement audit firm, when there are fewer
established lines of communication between the audit team and client. This paper
contributes to the knowledge spillover literature by identifying a mechanism through
which tax NAS improve overall financial reporting quality.
Keywords: auditor fees; nonaudit services; auditor independence; internal controls; tax;
financial reporting quality.
We appreciate helpful comments and guidance from Andrew Bauer, Patrick Badolato, Linda Bamber, Greg Capps,
Michael Clement, Dain Donelson, Michael Donohoe, John Harry Evans III, David A. Guenther (editor), Steve
Kachelmeier, William Kinney, Robert Knechel, Jayanthi Krishnan (discussant), Phil Lamoreaux, Pete Lisowsky, Tracie
Majors, John McInnis, Lillian Mills, Jaime Schmidt, Laura Wang, and two anonymous reviewers, as well as participants
at the 2012 AAA Annual Meeting and the 2012 Illinois Audit Symposium. We appreciate the helpful research assistance
of Patrick Kielty. All authors gratefully acknowledge support from the Accounting Doctoral Scholars Program, as well as
the Red McCombs School of Business. Lisa De Simone acknowledges funding from the Stanford Graduate School of
Business, Matthew Ege acknowledges funding from Texas A&M University, and Bridget Stomberg acknowledges
funding from the Terry College of Business and the Tull School of Accounting.
Editor’s note: Accepted by David A. Guenther.
Submitted: November 2012
Accepted: October 2014
Published Online: October 2014
1469
De Simone, Ege, and Stomberg
1470
I. INTRODUCTION
T
his paper examines whether tax services provided to a company by the financial statement
audit firm (i.e., auditor-provided tax nonaudit services or ‘‘tax NAS’’) benefit internal
control quality, and examines factors expected to strengthen the magnitude of these
potential benefits. Evidence on this potentially favorable consequence of tax NAS is relevant to
auditing regulators and others charged with corporate governance who must weigh the benefits and
costs of tax NAS, such as threats to auditor independence. Our study is also important to researchers
because although the knowledge spillover literature links tax NAS to improved financial reporting
quality (e.g., Gleason and Mills 2011; Kinney, Palmrose, and Scholz 2004; Robinson 2008), it has
not yet identified a mechanism through which tax NAS affects non-tax financial reporting quality.
We posit that companies have greater opportunities to timely improve their internal controls
when they obtain tax services from their audit firm than when they do not.1 Identifying existing or
potential control deficiencies requires: (1) extensive knowledge of the company’s operations, (2)
knowledge of which transactions are material to the financial statements, (3) awareness of which
controls are key to properly recording transactions material to the financial statements, and (4)
expertise to evaluate the quality of existing internal controls. Audit firms providing tax NAS are
more likely to have knowledge across these four areas relative to other tax service providers
because they audit the financial statements and internal controls.
Tax planning and compliance are affected by numerous business and accounting processes
throughout an organization. For example, computing taxable income requires an understanding of
the firm’s revenue-generating transactions and revenue recognition policies, which are often a
source of internal control deficiencies. Therefore, performing tax services exposes the provider to
areas outside of the tax function that impact internal controls. Because the tax partner who is part of
the audit team is typically involved in or oversees tax services provided to the client, he or she
frequently communicates with the audit partner. This communication between the tax and audit
partners allows the audit team to bring their expertise to bear earlier than if they only performed the
financial statement and internal control audits, and allows more time to identify and remediate
control issues prior to year-end.2 Practitioners indicate that having the audit firm involved in tax
NAS helps companies avoid surprises at year-end by increasing the audit firm’s awareness of
transaction details and audit risks early in the year (Larsen 2011). Therefore, we predict that tax
NAS decrease the likelihood of a material weakness.
Relative to other tax service providers outside the company, audit firm personnel are better
positioned to determine how transactions affect a company’s internal controls. Other accounting firms
or tax consulting firms do not have the extensive knowledge of company operations that auditors
acquire during financial statement and internal control audits. These other providers, therefore, lack
awareness of how key controls and processes interact to affect the quality of the client’s financial
statements. Anecdotal evidence from practitioners suggests that other accounting firms, law firms, and
boutique tax consulting firms focus on tax reporting and often fail to contemplate associated internal
controls risks and their implications for financial reporting. Finally, these other tax service providers
have less access to the audit team than tax personnel within the audit firm.
1
2
Throughout this paper, we use ‘‘client’’ and ‘‘company’’ to refer to the entity being audited or purchasing tax services.
We use ‘‘auditor’’ or ‘‘audit firm’’ to refer to the accounting firm auditing the financial statements, and ‘‘accounting
firm’’ to refer to any accounting firm regardless of whether it is the company’s financial statement auditor. We use
‘‘tax services’’ to refer to tax compliance or planning services. We use ‘‘tax NAS’’ to refer to tax services a company
purchases from their audit firm.
Auditor independence rules prohibit the auditor from assuming the role of management or auditing their own work
(Public Company Accounting Oversight Board [PCAOB] 2007). However, audit firms can identify internal control
risks, assess the design and effectiveness of internal controls, and communicate concerns to the client without
violating independence rules.
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1471
The audit firm is also better positioned to benefit the client’s internal control quality relative to
tax service providers within the company (i.e., the tax department). Tax department personnel may
lack detailed knowledge of material non-tax transactions and often focus primarily on tax
compliance. Further, tax risks are not well integrated into the overall risk assessment of the
company (Deloitte 2011; Robinson, Sikes, and Weaver 2010). Compared to the company’s tax
personnel, the audit firm has a more holistic view of interactions between financial reporting and
internal controls, and greater ability to evaluate the quality of the internal controls.
The following anecdotal example illustrates how tax NAS can benefit internal control quality.
A multinational corporation engaged its audit firm to assist in calculating arm’s-length transfer
prices and to evaluate related updates to its accounting system to record these intercompany
transactions in real time.3 In working with company personnel, the project team (which included tax
and accounting information systems advisors from the audit firm, as well as client accounting
personnel) discovered that the cost of goods sold to an unrelated party as part of a new revenue
stream was not being recorded in accordance with company policies. Because the project team was
from the audit firm, it was easy for them to communicate with the audit team, who had knowledge
of the company’s financial reporting policies and control structure, as well as expertise in evaluating
internal control quality. The early involvement of the audit team enabled the timely identification
and remediation of an internal control weakness, and allowed the company to avoid a material
weakness related to cost of goods sold.
We test our prediction that tax NAS are associated with better internal control quality using
pooled logistic regressions that control for the client’s economic importance to the audit firm. Our
sample includes 32,048 company-years with auditor internal control opinions from 2004 through
2012. We estimate that, compared to companies purchasing an average amount of tax NAS,
companies at one standard deviation above the mean exhibit a 13.2 percent decrease in the
probability of having a material weakness compared to the base rate of 6.6 percent. These results
are consistent with tax NAS benefitting internal control quality and are robust to controlling for
economic bonding between the client and audit firm as a proxy for impaired auditor independence.
To further rule out impaired auditor independence as an alternative explanation, we test for an
association between tax NAS and restated internal control opinions that disclose new material
weaknesses upon a financial restatement. In these instances, the original internal control opinion was
incorrect either because the audit firm was unaware of existing material weaknesses or was aware of
existing material weaknesses but chose not to disclose them. If impaired auditor independence
explained our results, then we would expect a positive association between restated internal control
opinions and tax NAS, suggesting that greater amounts of tax NAS increase the likelihood of the
auditor choosing not to disclose known material weaknesses in the original internal control opinion.
However, we find no association between tax NAS and instances of incorrect internal control opinions.
To reduce the likelihood that our results are attributable to an unobservable company
characteristic that is associated with both poor internal control quality and the decision to not
purchase tax NAS, we conduct several additional analyses that address endogeneity concerns. First,
we delete consecutive instances of material weaknesses to address the concern that companies with
habitually poor internal controls are less likely to devote resources to taxes. Second, we limit the
sample to observations with non-zero tax NAS and continue to find a positive association with
internal control quality, suggesting that the amount of tax NAS matters incremental to the decision
of whether to purchase tax NAS. Third, results are robust to using propensity score matching to pair
3
According to the Ernst & Young (2013) Global Transfer Pricing Survey, 41 percent of surveyed corporate taxpayers’
accounting systems do not automatically record transfer pricing transactions and 58 percent rely on Microsoft Excel
to calculate transfer prices. Companies are trying to better integrate transfer pricing within the accounting system to
increase automation of transfer pricing accounting (Ernst & Young 2013).
The Accounting Review
July 2015
De Simone, Ege, and Stomberg
1472
company-years with significant tax NAS to company-years with zero tax NAS, but with similar
predicted likelihoods of purchasing tax NAS. Fourth, we use an instrumental variable probit model.
Inferences are robust to these specifications. Finally, we document that companies not purchasing
tax NAS in a given year have cash effective tax rates significantly lower than the U.S. statutory rate
of 35 percent. This descriptive analysis provides evidence that firms not purchasing tax NAS in a
particular year still engage in tax planning and that the sample of firm years with no tax NAS
purchases is not comprised exclusively of firms that do not have adequate resources to effectively
manage their tax function. Results are robust to restricting our analysis to the subsample of firms
with low cash effective tax rates (i.e., tax-planning firms).
We also find that the positive relation between tax NAS and internal control quality is stronger
in settings where the expected benefits of tax NAS are greater. First, we predict and find that the
relation is stronger when companies undergo a significant change in their operating environment,
such as creating a new business segment or expanding overseas for the first time. Changing
operations bring about new processes that may require modifications to the company’s internal
control structure, and tax NAS can alert the audit firm to the associated financial reporting and
internal control risks earlier in the transaction lifecycle. Second, we predict and find that tax NAS
are more beneficial earlier in the client-audit firm relationship, when the audit firm is less familiar
with the client environment (e.g., Johnson, Khurana, and Reynolds 2002). As the audit firm’s tenure
increases, the relationship between the client and audit firm strengthens and individuals from both
entities have more frequent communication. Thus, tax NAS afford more opportunity for earlier
audit firm involvement in material transactions during the initial years of the client-auditor
relationship. These findings on auditor tenure also provide further evidence contrary to the
alternative explanation of compromised auditor independence.4
This study makes several contributions to the literature. First, we provide evidence consistent
with tax NAS accelerating audit firm awareness of transactions material to the financial statements
by offering evidence that tax NAS are associated with a lower likelihood of a material weakness.
We provide further evidence that this result is unlikely due to the alternative explanation of
impaired auditor independence. In this way, we inform the debate about the trade-off between the
benefits and potential costs of allowing tax NAS. We also extend the literature on knowledge
spillover by demonstrating that tax NAS are associated with tax and non-tax internal control
quality, an important element of financial reporting quality (e.g., Gleason and Mills 2011; Kinney et
al. 2004). Further, we identify specific situations in which the benefits of tax NAS are stronger,
which informs regulators and boards charged with tax NAS purchasing decisions. Our findings also
have implications for tax planning. Bauer (2014) suggests that strong tax internal control quality
facilitates tax planning. Therefore, tax NAS can potentially help companies expand tax-planning
opportunities by improving tax internal control quality.
II. RELATED LITERATURE AND HYPOTHESIS DEVELOPMENT
Related Literature and Theory
Prior literature documents benefits of tax NAS for financial reporting and audit quality. For
example, Kinney et al. (2004) provide evidence that tax NAS are negatively associated with the
likelihood of restating the financial statements and conclude that banning the provision of tax NAS
could reduce financial reporting quality. Robinson (2008) finds that tax NAS are positively
4
If auditor tenure and tax NAS impaired auditor independence, then we would expect the likelihood of disclosing a
material weakness to decrease in the presence of longer tenure with larger amounts of tax NAS. We find a decrease in
the likelihood of disclosing a material weakness in the presence of shorter tenure with larger amounts of tax NAS.
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1473
associated with the audit firm correctly issuing a going-concern opinion. Gleason and Mills (2011)
and Gleason, Mills, and Nessa (2013) provide evidence that companies purchasing tax NAS more
accurately estimate income tax expense, specifically when accruing reserves for potential future tax
liabilities. These studies support the theory that tax NAS increase the information available to the
audit firm, leading to higher-quality audits and improved financial reporting quality.
These studies and the literature in general suggest that benefits to audit quality and financial
reporting quality from tax NAS outweigh threats to independence created by the economic bond tax
NAS generate. Francis (2006, 752) states, ‘‘In reviewing the NAS literature over the last 40 years,
one has to conclude there is no smoking gun evidence linking the provision of non-audit services
with audit failure.’’ Thus, the totality of evidence on the provision of tax NAS supports Congress’
decision to exclude tax NAS from the list of prohibited nonaudit services within the SarbanesOxley Act (SOX, U.S. House of Representatives 2002).
While the literature generally concludes that benefits of tax NAS for audit and financial
reporting quality outweigh costs associated with independence impairment, little is known about
how tax NAS provide these benefits. Indeed, Francis (2004) argues that the proposition that
knowledge spillover results from providing both audit and nonaudit services to the same company
is ‘‘highly suspect’’ because the audit team is often not the same team that provides nonaudit
services. In other words, it is unclear how nonaudit services provided by nonaudit personnel can
lead to higher-quality financial reporting. As we further describe below, one possibility is that tax
NAS improve financial reporting quality as reflected in overall internal control quality.
Harris and Zhou (2008) were the first to examine the association between tax NAS and internal
control quality, finding that tax NAS are associated with fewer reported tax and non-tax material
weaknesses from 2004–2006. In their revision, Harris and Zhou (2013) submit that tax NAS should
benefit tax and non-tax internal control quality through knowledge spillover, but that there should
be greater knowledge spillover benefits to tax internal control quality. Contrary to their prediction,
they find a greater benefit of tax NAS on non-tax internal control quality and conclude that this is
possibly because of independence impairment and/or because auditors ‘‘bring to the client’s
attention more non-tax internal control problems than other tax consultants’’ (Harris and Zhou
2013, 32).
Corporate income taxes are affected by almost all business and accounting processes outside of the
tax department. For example, the computation of taxable income begins with pre-tax income reported
in the financial statements and incorporates differences between generally accepted accounting
principles (GAAP) and tax law. Correctly calculating taxable income, therefore, requires an
understanding of the transactions and accounting policies incorporated in each line item of the general
ledger. Similarly, identifying and implementing tax-planning strategies requires at least some
understanding of the company’s overall business operations. Therefore, tax services naturally involve
both tax and non-tax processes that are material to the financial statements and require internal controls.
We predict that companies have greater opportunities to improve their internal controls and
avoid material weaknesses when they purchase tax services from their audit firm. We make this
prediction because the audit firm has the knowledge and expertise required to evaluate the
company’s internal controls, and being involved in tax services allows the audit firm to consider
internal control issues related to material transactions earlier than they would if not performing tax
NAS. Identifying existing or potential control deficiencies requires: (1) extensive knowledge of the
company’s operations, (2) knowledge of which transactions are material to the financial statements,
(3) awareness of which controls are key to properly recording transactions material to the financial
statements, and (4) expertise to evaluate the quality of internal controls. Because they are
responsible for opining on the company’s financial statements and internal controls, the audit firm is
most likely to have this knowledge and expertise.
The Accounting Review
July 2015
1474
De Simone, Ege, and Stomberg
Tax NAS also provide greater opportunities for earlier audit firm involvement, primarily through
the tax partner on the audit team. The tax partner on the audit team, who is responsible for auditing the
income tax provision, communicates with the audit partner throughout the year, such as at audit team
meetings. The tax partner can also be directly involved in tax services that are provided to the client. If
not directly involved, perhaps because the tax services are beyond the scope of his or her expertise,
then the tax partner usually oversees tax services provided by the audit firm to the client. The tax
partner’s oversight helps streamline the provision of services and allows for a single point of contact
within the tax group of the audit firm. The tax partner’s role in overseeing client services also affords
the tax partner opportunities to communicate tax service details to the audit partner and for the audit
team to consider any related effects on internal control and financial reporting quality.5 In summary, tax
NAS create opportunities for earlier audit firm involvement in transactions material to the financial
statements, which results in earlier identification of internal control weaknesses and a lower likelihood
of material weaknesses existing at year-end.
Practitioners with whom we spoke agree that tax NAS can improve internal control quality
because the audit team is made aware of transactions material to the financial statements earlier in
the transaction lifecycle. When the audit firm is not involved in tax services, the audit team is often
not made aware of significant changes until after fiscal year-end and, therefore, cannot help identify
potential control weaknesses.6 Although it is possible for the audit team to be involved in material
transactions before year-end even if not providing tax services, this takes a degree of organization
and coordination within the company, which practitioners said is not often in place.
Anecdotal Evidence
In addition to the example provided in the introduction, we present two more examples from
our informal discussions with several partners from public accounting firms and tax executives from
large, publicly traded companies, all of whom requested that we not disclose company names. First,
a company engaged their audit firm to assist in implementing a tax-planning strategy to reduce
foreign tax payments. The strategy prohibited foreign country personnel from making significant
business decisions, such as negotiating prices. The tax partner included the audit partner in planning
discussions to help the company identify the key processes and controls required to maintain the
integrity of the tax strategy, preserve the intended tax benefits, and ensure that related tax accounts
were fairly presented on the financial statements. As a result, the company implemented a control
requiring approval of sales orders generated by foreign personnel. The tax partner who shared this
example with us said that, as part of the audit team, he had seen situations where a provider other
than the audit firm performed similar tax-planning services without considering internal controls.
Other tax service providers do not have the same understanding of the company’s internal control
structure as the audit firm and, therefore, are not well positioned to evaluate its effectiveness. The
partner noted that external consulting firms tend to implement the tax strategy and ‘‘walk away’’
without considering what controls need to be in place to maintain the integrity of the strategy.
Practitioners also provided examples of how tax compliance work improved internal control
quality. Some accounting firm offices have an informal policy requiring tax teams performing tax
NAS to consult the audit team for any document requests. This strategy reduces duplicate requests
5
6
There are typically also key interactions between other tax personnel on the audit team and audit team members that
would result in improved internal controls. Further, there are many ways in which information about internal control
considerations flows back to the company (e.g., from the tax partner to the tax director at the company or from the
audit partner to the chief financial officer, etc.).
Internal control opinions are ‘‘as of’’ a specific date, as opposed to covering an entire period. Accordingly, if a
material weakness is discovered, but subsequently remediated prior to the ‘‘as of’’ date on the report, then the material
weakness does not have to be disclosed.
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1475
from the audit firm and creates efficiencies. It also has led to improved internal controls for clients.
For example, in preparing quarterly tax provision estimates for an audit client, the tax team
requested an organizational chart from the audit team and discovered a change in ownership of
which the client’s tax department was not aware. The tax team at the audit firm communicated with
the client’s tax director, thus prompting the client to put a control in place whereby the tax director
met with the legal department on a quarterly basis to discuss any changes in legal entities. Without
this control, the income tax accrual could have been materially misstated. This early identification
of the organizational change would not have occurred had the tax team and audit team not both
been from the same audit firm. This example speaks to the ease with which tax and audit personnel
can communicate when they are all from the same audit firm.
Hypothesis
We believe that tax NAS allow for the prevention or early detection and remediation of internal
control weaknesses due to earlier audit firm awareness of transactions material to the financial
statements. Therefore, we predict a positive association between tax NAS and internal control
quality. We formally state this prediction in the alternative form below.
H: There is a negative relation between tax NAS and the probability of a material weakness in
internal controls.
To provide further insight into this relation, we explore settings in which we believe the
benefits of tax NAS are stronger. First, we examine whether tax NAS are more beneficial to
companies experiencing a shock to their internal control environment. Once strong internal controls
are in place, it is likely that these controls continue to operate in an effective manner. However, a
major change in the operating environment could expose the company to a risk for which they do
not have appropriate controls in place. Hence, these changes provide more opportunities to observe
the effects of tax NAS on internal control quality. Based on examples provided in Sections I and II,
we expect new internal controls to be required when a new business segment is created or when the
company expands overseas. Therefore, we examine the effects of tax NAS on internal control
quality when there is an increase in the number of segments reported or when there is an expansion
into a foreign jurisdiction. This analysis serves as somewhat of a ‘‘changes’’ analysis that mitigates
concerns about unobserved firm characteristics that simultaneously drive decisions surrounding
internal control quality and tax NAS purchases. We expect that tax NAS incrementally improve
internal control quality in the presence of an increase in segments or a foreign expansion.
Second, we investigate how the relation between tax NAS and internal control quality is affected
by audit firm tenure. As the length of the relationship between the audit firm and client increases, it is
likely that the relationship becomes stronger and that individuals within the two entities have more
frequent communication. This strengthened relationship should increase the likelihood that client
personnel communicate material transactions to audit firm personnel earlier in the transaction lifecycle.
For example, a client with an established relationship with its auditor is more likely to informally ask
the audit partner for his or her opinion about the need for additional controls over the transaction. In
turn, the audit partner would likely be more comfortable providing such informal advice to a long-term
client. In short, because the client personnel and audit team personnel are more likely to communicate
informally as audit firm tenure increases, there is less need for tax personnel on the audit team to
facilitate this communication and, therefore, less opportunity for tax NAS to incrementally benefit
internal control quality. Thus, we expect that the positive effect that tax NAS have on internal control
quality is stronger (i.e., the negative relation between tax NAS and likelihood of material weakness
disclosure would be more negative) earlier in the auditor’s tenure.
The Accounting Review
July 2015
De Simone, Ege, and Stomberg
1476
Although we believe longer audit firm tenure promotes more communication between the audit
firm and the client, it is also possible that longstanding client-auditor relationships threaten
independence.7 Mandatory audit firm rotation has been considered on numerous occasions in the
United States, with proponents believing that mandatory audit firm rotation would benefit audit firm
independence by limiting the continuous revenue stream from a single client (PCAOB 2011). If tax
NAS and auditor tenure both impair independence, then we would expect to find fewer disclosed
material weaknesses in situations where companies purchase tax NAS from a long-tenured audit
firm (i.e., the negative relation between tax NAS and likelihood of material weakness disclosure
would be more negative).
III. SAMPLE AND RESEARCH DESIGN
Hypothesis Tests
Relation between Tax NAS and Internal Control Quality
To test our hypothesis, we begin with all SOX Section 404(b) auditor internal control opinions
included in the Audit Analytics database from 2004 through 2012. Section 404(b) applies to
accelerated filers, those companies with public float of $75 million or more within six months of the
fiscal year-end. We require each observation to have data available from both CRSP and Compustat
to calculate control variables shown in prior literature to affect the probability of a material
weakness in internal control over financial reporting (Ashbaugh-Skaife, Collins, and Kinney 2007;
Doyle, Ge, and McVay 2007). The full sample consists of 32,048 observations from 5,830 unique
companies. Of these companies, 4,987 purchase tax NAS at some point during the sample.
Table 1 provides descriptive statistics for the sample. The mean (median) amount of tax fees
paid to audit firms is approximately $330,000 ($52,000) for the entire sample. These tax services
can be for either tax compliance or consulting work.8 Within the subsample of observations with
non-zero tax NAS, the mean (median) amount is $451,000 ($115,000), suggesting that among
companies electing to engage their audit firm for tax services, the magnitude of the fees incurred is
not inconsequential (untabulated). For example, given an average blended hourly billing rate of
$300, these mean (median) fees correspond to 1,500 (300) hours of tax service work. The scope of
work represented by these amounts is sufficiently large to expose the tax team at the audit firm to
issues that are material to the financial statements.
Our hypothesis predicts a negative relation between tax NAS and the probability of a material
weakness. To test this hypothesis, we estimate the following logistic regression:
ProbðICW ¼ 1Þ ¼ Fðb0 þ b1 LNTAXNAS þ b2 LNAUDIT þ b3 LNOTHERNAS
þb4 INFLUENCE þ b5 LMARKETCAP þ b6 AGGLOSS þ b7 SHUMWAY
þb8 LSEGCOUNT þ b9 FORTRANS þ b10 MERGER þ b11 EXTREMESG
þb12 RESTRUCTURE þ b13 BIG4 þ b14 AUDITOR RESIGN
þb15 PCTFORSALES þ b16 TAXLOSS þ Year FEÞ:
ð1Þ
We define all variables in detail in Appendix A with references to Compustat variable names.
External auditors disclose material weaknesses in internal controls in an opinion on internal controls
7
8
Prior literature finds that auditor tenure is not associated with impaired auditor independence. In fact, there is
evidence of higher financial reporting quality with longer auditor tenure (Johnson et al. 2002; J. Myers, L. Myers, and
Omer 2003).
Audit Analytics does not provide detail about the nature of tax NAS. Therefore, we cannot readily observe what the
tax NAS specifically relate to. We acknowledge that some types of tax NAS, such as preparing individual tax returns
for expatriate employees, are not likely to provide benefits to internal control quality.
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1477
TABLE 1
Descriptive Statistics
Internal Control Quality
ICW
ICW_TAX
ICW_NONTAX
Fees
TAXFEES
AUDITFEES
n
Mean
Std. Dev.
25th Pctl.
Median
75th Pctl.
32,048
32,048
32,048
0.066
0.020
0.046
0.249
0.140
0.210
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
0.000
32,048
32,048
331,164
2,772,064
1,112,093
6,238,772
0.000
575,809
51,950
1,111,000
226,444
2,405,260
5.345
0.000
10.858
12.330
1.174
4.778
0.221
1.772
0.437
2.872
0.751
0.461
0.391
0.380
0.418
0.375
0.102
0.314
1.398
13.326
0.000
0.015
5.542
0.000
2.000
0.693
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
13.986
0.000
0.046
6.674
0.000
5.000
1.099
0.000
0.000
0.000
0.000
1.000
0.000
0.000
0.000
14.780
7.824
0.140
7.960
1.000
7.000
1.609
1.000
0.000
0.000
0.003
1.000
0.000
0.378
0.052
Variable of Interest
LNTAXNAS
32,048
8.561
Control Variables—Internal Control Quality
LNAUDIT
32,048
14.103
LNOTHERNAS
32,048
3.271
INFLUENCE
32,048
0.136
LMARKETCAP
32,048
6.853
AGGLOSS
32,048
0.256
SHUMWAY
32,048
4.502
LSEGCOUNT
32,048
1.017
FORTRANS
32,048
0.306
MERGER
32,048
0.188
EXTREMESG
32,048
0.175
RESTRUCTURE
32,048
0.010
BIG4
32,048
0.831
AUDITOR_RESIGN
32,048
0.011
PCTFORSALES
32,048
0.209
TAXLOSS
32,048
0.254
This panel provides descriptive statistics for variables of interest and control variables included in Equation (1).
Variable definitions are provided in Appendix A.
over financial reporting. We set ICW equal to 1 if a company’s auditor reports a material weakness
in internal controls for that year, and 0 otherwise. It is possible that some material weaknesses that
exist in year t are not disclosed in the auditor’s internal control opinion for year t, either because of a
lack of knowledge that they exist or because of impaired auditor independence. To address these
concerns and attempt to identify years in which material weaknesses existed, we (1) use the most
recently issued internal control opinion for year t, (2) include controls that relate to the economic
importance of the client to the audit firm, and (3) conduct additional tests to rule out impaired
independence as an explanation for our results. Using the most recently issued internal control
opinion for year t helps to ensure that we capture material weaknesses that existed in year t
regardless of when the external auditor reported it.
We operationalize tax NAS using the natural logarithm of tax fees reported in Audit Analytics
in year t (LNTAXNAS) following prior literature (e.g., DeFond, Raghunandan, and Subramanyam
2002).9 b1 is the coefficient of interest. Our hypothesis predicts that b1 is negative. Consistent with
9
Consistent with DeFond et al. (2002), in untabulated tests, we confirm that results are robust to measuring tax NAS as
the ratio of TAXFEES to TOTALFEES as reported in Audit Analytics in year t. This measure allows the effect of tax
NAS on internal control quality to vary with size.
The Accounting Review
July 2015
1478
De Simone, Ege, and Stomberg
DeFond et al. (2002), we control for the natural logarithm of total audit fees (LNAUDIT) and expect
the coefficient b2 to be positive because auditors increase their substantive procedures in the
presence of material weaknesses (Hogan and Wilkins 2008).
Because both knowledge spillover and auditor economic dependence predict a negative
coefficient on LNTAXNAS, we control for the client’s economic importance to its audit firm. Kinney
et al. (2004) link other non-specified nonaudit services to restatements, suggesting that they impair
auditor independence. However, Reynolds and Francis (2001) provide evidence that audit firms
provide higher-quality audits to their economically important clients. We control for both fees paid
to the financial statement audit firm for other NAS (LNOTHERNAS) and for the client’s relative
importance to the financial statement audit firm (INFLUENCE). The audit firm’s economic
dependence on the client is increasing in INFLUENCE. Because of mixed evidence regarding the
overall effect of fees paid to the audit firm on independence, we make no prediction as to the sign of
b3 or b4.
We also include several control variables shown in Ashbaugh-Skaife et al. (2007) and Doyle et
al. (2007) to be significant determinants of material weaknesses. LMARKETCAP controls for size,
as smaller companies exhibit an increased likelihood of material weaknesses. We include
AGGLOSS and SHUMWAY as controls for financial distress. These controls are also important
because it is possible that companies in financial distress are less likely to devote resources to tax
NAS. Next, we include variables to control for financial reporting complexity (LSEGCOUNT,
FORTRANS, and MERGER), all of which have been shown to increase the likelihood of material
weaknesses. We also control for rapid growth, which increases the likelihood of reporting a material
weakness, with EXTREMESG and RESTRUCTURE. We include BIG4 and AUDITOR_RESIGN to
capture auditor quality and disagreement between the audit firm and client, respectively.
We include two additional variables to control for income tax reporting complexity, which
could affect both internal control quality and the decision to purchase tax NAS. First, we include the
percentage of total sales made to foreign customers because companies are more likely to face
complex tax issues when they conduct business in multiple jurisdictions. Foreign operations are
also positively associated with the purchase of tax NAS (Bédard and Paquette 2011). Therefore, we
expect the coefficient on PCTFORSALES (b15 ) to be positive. Finally, because Bédard and Paquette
(2011) find some evidence that tax net operating losses reduce the likelihood of purchasing tax
NAS, we control for TAXLOSS. However, we make no prediction as to how TAXLOSS will affect
internal control quality.
Table 2 presents correlations of our main regression variables, with Pearson correlations above
the diagonal and Spearman correlations below the diagonal. This univariate analysis suggests a
negative correlation between ICW and LNTAXNAS.
We winsorize all continuous variables at 1 and 99 percent. In the tables, we present standard
errors clustered by company in parentheses under parameter estimates. In all tables where we
estimate a logistic regression, we include the max-rescaled pseudo R2 reported by SAS, also known
as the Cragg and Uhler pseudo R2.
Endogeneity
Prior studies present evidence that tax NAS purchases are not random (e.g., Lassila, Omer,
Shelley, and Smith 2010). Therefore, it is possible that companies choosing to engage their
financial statement audit firm for tax services have unobservable characteristics that are also
associated with strong internal control quality. Although we are not aware of any theory or
empirical evidence suggesting that companies purchasing tax NAS are also more committed to
strong internal controls, we nonetheless conduct several additional analyses to help allay concerns
that our results are confounded by the potential endogeneity of tax NAS purchases.
The Accounting Review
July 2015
The Accounting Review
July 2015
ICW
LNTAXNAS
LNAUDIT
LNOTHERNAS
INFLUENCE
LMARKETCAP
AGGLOSS
SHUMWAY
LSEGCOUNT
FORTRANS
MERGER
EXTREMESG
RESTRUCTURE
BIG4
AUDITOR_RESIGN
PCTFORSALES
TAXLOSS
SEGMENT_INC
FOREIGN_EXP
TENURE
1.000
0.040
0.018
0.016
0.046
0.119
0.100
0.107
0.007
0.019
0.031
0.004
0.016
0.061
0.097
0.022
0.056
0.032
0.088
0.092
3
0.050
0.624
0.104
0.074
0.004
0.125
0.082
0.093
0.104
0.058
0.049
0.034
0.074
0.136
0.053
0.105
0.003
0.006
0.038
0.175
2
0.037
1.000
0.495
0.179
0.114
0.411
0.128
0.221
0.275
0.156
0.102
0.061
0.198
0.290
0.064
0.247
0.031
0.048
0.065
0.265
1. ICW
2. LNTAXNAS
12
0.004
0.052
11
0.031
0.071
0.003
0.009
13
Panel B: Correlation Variables MERGER to TENURE
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
1
Panel A: Correlation Variables ICW to FORTRANS
0.061
0.231
14
0.011
0.352
1.000
0.220
0.238
0.705
0.152
0.315
0.433
0.255
0.157
0.073
0.311
0.444
0.061
0.358
0.075
0.103
0.075
0.273
3
0.097
0.066
15
0.017
0.158
0.249
1.000
0.076
0.193
0.056
0.087
0.104
0.101
0.066
0.010
0.059
0.115
0.024
0.101
0.004
0.025
0.027
0.106
4
Correlations
TABLE 2
0.033
0.030
0.177
0.095
1.000
0.195
0.100
0.051
0.099
0.115
0.069
0.016
0.050
0.350
0.054
0.114
0.012
0.034
0.019
0.082
5
0.024
0.142
16
0.115
0.297
0.735
0.218
0.175
1.000
0.375
0.622
0.274
0.151
0.101
0.020
0.071
0.394
0.068
0.194
0.096
0.045
0.016
0.284
6
0.011
0.053
17
0.100
0.102
0.147
0.057
0.062
0.366
1.000
0.380
0.040
0.018
0.026
0.016
0.154
0.058
0.038
0.020
0.265
0.018
0.012
0.112
7
0.032
0.027
18
0.088
0.052
19
0.005
0.191
0.415
0.106
0.055
0.282
0.041
0.297
1.000
0.429
0.095
0.060
0.299
0.207
0.011
0.788
0.224
0.247
0.213
0.122
9
0.060
0.193
20
0.019
0.115
0.241
0.104
0.108
0.157
0.018
0.168
0.419
1.000
0.066
0.008
0.180
0.117
0.008
0.478
0.161
0.108
0.287
0.003
10
(continued on next page)
0.107
0.151
0.316
0.090
0.039
0.620
0.380
1.000
0.300
0.168
0.024
0.023
0.020
0.212
0.042
0.242
0.051
0.048
0.075
0.152
8
Internal Control Quality: The Role of Auditor-Provided Tax Services
1479
LNAUDIT
LNOTHERNAS
INFLUENCE
LMARKETCAP
AGGLOSS
SHUMWAY
LSEGCOUNT
FORTRANS
MERGER
EXTREMESG
RESTRUCTURE
BIG4
AUDITOR_RESIGN
PCTFORSALES
TAXLOSS
SEGMENT_INC
FOREIGN_EXP
TENURE
0.147
0.067
0.020
0.099
0.026
0.024
0.092
0.066
1.000
0.067
0.132
0.030
0.014
0.110
0.067
0.062
0.036
0.048
13
0.013
0.006
0.006
0.023
0.042
0.020
0.011
0.014
0.006
0.001
1.000
0.129
0.015
0.299
0.201
0.057
0.043
0.122
12
0.072
0.010
0.003
0.022
0.016
0.023
0.059
0.008
0.067
1.000
0.116
0.043
0.009
0.035
0.001
0.057
0.051
0.096
0.427
0.118
0.216
0.368
0.058
0.212
0.205
0.117
0.030
0.043
0.004
1.000
0.139
0.142
0.046
0.040
0.047
0.369
14
0.063
0.024
0.041
0.061
0.038
0.042
0.011
0.008
0.014
0.009
0.001
0.139
1.000
0.006
0.009
0.012
0.023
0.150
15
0.275
0.101
0.096
0.162
0.004
0.196
0.641
0.435
0.082
0.011
0.014
0.111
0.001
1.000
0.225
0.191
0.269
0.073
16
0.117
0.022
0.039
0.166
0.243
0.153
0.038
0.011
0.033
0.033
0.006
0.027
0.017
0.011
1.000
0.063
0.070
0.028
17
0.096
0.025
0.013
0.046
0.018
0.048
0.257
0.108
0.062
0.057
0.010
0.040
0.012
0.166
0.011
1.000
0.096
0.011
18
0.069
0.027
0.018
0.019
0.012
0.075
0.199
0.287
0.036
0.051
0.004
0.047
0.023
0.217
0.001
0.096
1.000
0.083
19
0.315
0.101
0.029
0.301
0.121
0.148
0.142
0.013
0.024
0.094
0.005
0.232
0.071
0.056
0.049
0.001
0.045
1.000
20
This table provides correlations for variables contained in Equation (1) and other variables of interest. Pearson (Spearman) correlations are presented above (below) the diagonal.
Bold correlations denote significance at or below 10 percent.
Variable definitions are provided in Appendix A.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
11
TABLE 2 (continued)
1480
De Simone, Ege, and Stomberg
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1481
First, we restrict our full sample to include only non-consecutive instances of material weakness
disclosures. That is, if a company reports a material weakness over internal controls in two
consecutive years, then we retain only the first observation.10 This approach helps alleviate concerns
that our results are driven by companies with habitually poor internal control environments for which
engaging external tax-planning or compliance services is not likely to be a priority. Second, we
reestimate Equation (1) on the subsample of observations with non-zero tax NAS. We do this to
address the concern that it is the binary decision of whether to purchase tax NAS, and not the amount,
that drives our results. This analysis also addresses the concern that observing zero tax NAS
represents a decision not to purchase any tax services, perhaps due to a lack of resources or poor
performance, which might be correlated with weak internal controls. Third, we follow Cheng,
Dhaliwal, and Zhang (2013) and estimate Equation (1) using a propensity score matched sample. We
estimate a company’s propensity to purchase tax NAS with the following regression:
ProbðTAXNASIND ¼ 1Þ ¼ Fðb0 þ b1 LNAUDIT þ b2 LNOTHERNAS þ b3 INFLUENCE
þb4 LMARKETCAP þ b5 AGGLOSS þ b6 SHUMWAY
þb7 LSEGCOUNT þ b8 FORTRANS þ b9 MERGER
þb10 EXTREMESG þ b11 RESTRUCTURE þ b12 BIG4
þb13 AUDITOR RESIGN þ b14 PCTFORSALES þ b15 TAXLOSS
þb16 TENURE þ b17 EQINC þ b18 R&D þ b19 LEVERAGE
þb20 BTM þ b21 PPE þ b22 ROA þ b23 CASH þ b24 DEP
þb25 SECTIER þ Year FEÞ:
ð2Þ
We include determinants of tax NAS purchases using variables from Lassila et al. (2010) and
McGuire, Omer, and Wang (2012) and include all control variables from Equation (1). Our control
sample contains all observations with zero tax NAS, and our treatment sample includes all
observations in the highest quartile of the ratio of TAXFEES to TOTALFEES by year. After
calculating propensity scores for all observations in both samples, we match each treatment
company to a unique control company and retain only those pairs whose scores match within 0.01.
We then reestimate Equation (1) using this propensity score matched sample.11
Fourth, we estimate ICW as a function of LNTAXNAS using an instrumental variables probit
model. We estimate LNTAXNAS using an ordinary least squares (OLS) regression derived from
Bédard and Paquette (2011), Lassila et al. (2010), and McGuire et al. (2012). We simultaneously
estimate the following equation with Equation (1):
LNTAXNAS ¼ b0 þ b1 LNAUDIT þ b2 LNOTHERNAS þ b3 INFLUENCE
þ b4 LMARKETCAP þ b5 AGGLOSS þ b6 SHUMWAY þ b7 LSEGCOUNT
þ b8 FORTRANS þ b9 MERGER þ b10 EXTREMESG þ b11 RESTRUCTURE
þ b12 BIG4 þ b13 AUDITOR RESIGN þ b14 PCTFORSALES þ b15 TAXLOSS
þ b16 TENURE þ b17 EQINC þ b18 R&D þ b19 LEVERAGE þ b20 BTM
þ b21 PPE þ b22 ROA þ b23 CASH þ b24 DEP þ b25 SECTIER þ Year FE þ e:
ð3Þ
10
11
Because a company may report material weaknesses related to different issues over the course of the full sample
period (2004 to 2012), we allow more than one material weakness company-year for the same company in our sample
as long as they are not consecutive. However, results are robust to limiting material weakness company-years to only
the first instance in the sample period.
Results are robust to matching observations in the highest quartile of LNTAXNAS to observations with zero tax NAS.
However, we retain dramatically fewer pairs when we require scores to match within 0.01. Therefore, we tabulate
results using the ratio of TAXFEES to TOTALFEES to match companies.
The Accounting Review
July 2015
De Simone, Ege, and Stomberg
1482
IV. RESULTS
Hypothesis Tests
Table 3 provides results of testing our hypothesis. Column (1) of Panel A presents full sample
results, which are consistent with our hypothesis that, on average, auditor-provided tax services
decrease the probability of disclosing a material weakness. Specifically, we find a negative and
statistically significant coefficient on LNTAXNAS of 0.0295, suggesting that the probability of
disclosing an internal control weakness is decreasing in purchases of tax NAS. Additionally, these
results are economically significant. A one-standard-deviation increase in LNTAXNAS corresponds
to a marginal effect of 0.9 percent, which represents approximately a 13.2 percent decrease in the
6.6 percent unconditional probability of disclosing a material weakness.12 We interpret this
evidence as consistent with tax NAS improving internal control quality.
In Columns (2) and (3) of Table 3, Panel A we present results of two specifications that address
potential endogeneity. In Column (2), we estimate Equation (1) on the subsample of observations
disclosing non-consecutive material weaknesses. We estimate a negative and significant coefficient
on the measure of tax NAS of 0.0187, suggesting that results in Column (1), estimated on the full
sample, are not driven by companies with overall weak internal control environments not
prioritizing tax planning or compliance.13 In this sample, a one-standard-deviation increase in
LNTAXNAS corresponds to a 0.4 percent marginal effect that represents an approximate 8.8 percent
decrease in the 5.1 percent unconditional probability of disclosing a material weakness. In Column
(3), we estimate Equation (1) using a subsample of observations that purchase tax NAS. We
continue to find a significant negative association between the measure of tax NAS and the
likelihood of a material weakness disclosure in this sample, providing support that the amount, not
just the presence, of tax NAS affects the likelihood of material weakness.
In Panel B of Table 3 we estimate Equation (1) using a propensity score matched sample.
Column (1) presents results of estimating the propensity score model shown in Equation (2), and
Column (2) presents results of estimating the effect of tax NAS on this matched sample. We continue
to a find significant negative coefficient on the measure of tax NAS. In Panel C of Table 3 we
implement an instrumental variable probit approach to control for the endogenous choice to purchase
tax NAS. Results of the selection model are presented in Column (1), while results of estimating
Equation (1) are in Column (2). We continue to find a negative relation between tax NAS and material
weaknesses even after controlling for possible selection bias.14 Taken together, results presented in
Table 3 collectively suggest that tax NAS, on average, benefit internal control quality.
Finally, it is important to note that the company-year decision to not purchase tax NAS does
not necessarily reflect a lack of commitment to tax planning. In company-years without tax NAS,
we find an average GAAP effective tax rate (ETR) of approximately 28 percent and an average cash
ETR of approximately 22 percent, both well below the U.S. statutory tax rate of 35 percent. These
tax rates are consistent with some degree of tax planning in company-years without tax NAS
purchases. Furthermore, we find no significant difference in average GAAP ETRs between
12
13
14
Marginal effects are computed using the average of discrete or partial changes over all observations (Bartus 2005).
Results are also robust to controlling for material weaknesses disclosed in t1 and controlling for the time to
remediate the disclosed material weakness. We use the number of years from the original material weakness
disclosure until an unqualified internal control opinion is issued as a proxy for the time to remediate. These results
provide further comfort that our results are not due to lack of commitment to internal control quality or an inability to
have strong internal controls.
Because the instrumental variable probit methodology utilizes simultaneous equations, we do not obtain goodness-offit statistics for the selection model. However, when we estimate Equation (3) separately, we obtain an R2 of 20.7
percent. Lassila et al. (2010) report a pseudo R2 of 22.9 percent when modeling the likelihood of purchasing tax NAS.
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1483
TABLE 3
Effect of Tax NAS on Internal Control Weaknesses
Panel A: Pooled Logistic Regressions
ProbðICW ¼ 1Þ ¼ Fðb0 þ b1 LNTAXNAS þ Controls þ Year FEÞ:
Sample
Independent
Variables
Pred.
LNTAXNAS
LNAUDIT
þ
LNOTHERNAS
?
INFLUENCE
?
LMARKETCAP
AGGLOSS
þ
SHUMWAY
þ
LSEGCOUNT
þ
FORTRANS
þ
MERGER
þ
EXTREMESG
þ
RESTRUCTURE
þ
BIG4
þ
AUDITOR_RESIGN
þ
PCTFORSALES
þ
TAXLOSS
?
Obs. where ICW ¼ 1
Total Obs.
Pseudo R2
Full Sample
(1)
0.0295***
(0.006)
0.8441***
(0.053)
0.0005
(0.006)
0.6032***
(0.153)
0.6384***
(0.037)
0.3591***
(0.068)
0.0112
(0.013)
0.1566
(0.054)
0.2221***
(0.073)
0.0049
(0.077)
0.1435**
(0.066)
0.0589
(0.035)
0.5285
(0.084)
1.0723***
(0.156)
0.3682***
(0.108)
0.0581**
(0.028)
2,129
32,048
18.62%
Removed
Consecutive ICWs
(2)
0.0187***
(0.006)
0.7106***
(0.046)
0.0027
(0.006)
0.3020**
(0.143)
0.5975***
(0.033)
0.2585***
(0.070)
0.0052
(0.013)
0.1446
(0.050)
0.1962***
(0.069)
0.1039*
(0.077)
0.2119***
(0.069)
0.0505
(0.025)
0.4721
(0.082)
0.8944***
(0.179)
0.3943***
(0.096)
0.0558**
(0.025)
1,600
31,519
16.91%
Positive
Tax NAS
(3)
0.0889***
(0.024)
0.9603***
(0.056)
0.0016
(0.007)
0.3493**
(0.167)
0.6947***
(0.038)
0.3628***
(0.079)
0.0214
(0.015)
0.1106
(0.061)
0.2036***
(0.084)
0.0327
(0.091)
0.2503***
(0.078)
0.0615
(0.039)
0.6671
(0.105)
1.3439***
(0.222)
0.2340**
(0.123)
0.0633*
(0.035)
1,453
23,528
19.16%
(continued on next page)
The Accounting Review
July 2015
De Simone, Ege, and Stomberg
1484
TABLE 3 (continued)
*, **, *** Denote one-tailed significance at 10 percent, 5 percent, and 1 percent, respectively, when a prediction is made;
otherwise, significance denoted is based on two-tailed tests.
Panel A presents results of estimating Equation (1). Column (1) presents results from the full sample; Column (2)
presents results after removing consecutive ICWs from the sample; and Column (3) presents results after limiting the
sample to company-years with positive tax NAS. Standard errors clustered by company are in parentheses. Our variable
of interest, LNTAXNAS, is the natural logarithm of tax fees paid to the financial statement auditor in year t. ICW equals 1
if the company’s auditor reported an internal control weakness in year t.
Variable definitions are provided in Appendix A.
Panel B: Propensity Score Matched Sample
ProbðTAXNASIND ¼ 1Þ ¼ Fðb0 þ Controls þ Year FEÞ:
ProbðICW ¼ 1Þ ¼ Fðb0 þ b1 LNTAXNAS þ Controls þ YearFEÞ:
Dependent Variables
Independent Variables
Pred.
TAXNASIND
(1)
Variable of Interest
LNTAXNAS
Controls for Economic Dependence
LNAUDIT
þ
LNOTHERNAS
?
INFLUENCE
Other Controls
LMARKETCAP
þ
AGGLOSS
SHUMWAY
LSEGCOUNT
þ
FORTRANS
þ
MERGER
þ
EXTREMESG
RESTRUCTURE
þ
BIG4
?
AUDITOR_RESIGN
Pred.
ICW
(2)
0.0433***
(0.011)
0.2686***
(0.041)
0.0374***
(0.005)
0.2249
(0.870)
þ
0.7283***
(0.115)
0.0017
(0.015)
1.0803***
(0.269)
0.0667***
(0.028)
0.1216***
(0.052)
0.0088
(0.011)
0.0842**
(0.048)
0.1802***
(0.060)
0.1750***
(0.054)
0.1364***
(0.041)
0.0353
(0.059)
0.1128
(0.103)
0.8593***
(0.123)
?
?
þ
þ
þ
þ
þ
þ
þ
þ
þ
0.4893***
(0.077)
0.4756***
(0.150)
0.0330
(0.030)
0.3293
(0.138)
0.2699**
(0.156)
0.2671*
(0.197)
0.0998
(0.146)
2.6295
(1.000)
0.6011
(0.151)
1.3106***
(0.343)
(continued on next page)
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1485
TABLE 3 (continued)
Dependent Variables
Independent Variables
Pred.
PCTFORSALES
þ
TAXLOSS
þ
TENURE
þ
EQINC
R&D
?
LEVERAGE
þ
BTM
PPE
?
ROA
?
CASH
?
DEP
SECTIER
Obs. where ICW ¼ 1
Total Obs.
Pseudo R2
TAXNASIND
(1)
0.2194**
(0.112)
0.0035
(0.014)
0.0148***
(0.003)
1.6909
(1.660)
0.0003
(0.157)
0.0507
(0.120)
0.0000
(0.000)
0.2555***
(0.079)
0.0093
(0.010)
0.0824
(0.166)
1.1401**
(0.687)
0.7648***
(0.118)
NA
32,048
13.11%
Pred.
þ
?
ICW
(2)
0.6700**
(0.320)
0.0541
(0.046)
400
7,042
17.96%
*, **, *** Denote one-tailed significance at 10 percent, 5 percent, and 1 percent, respectively, when a prediction is made;
otherwise, significance denoted is based on two-tailed tests.
Panel B presents the results of using a propensity score matched sample to test the effect of tax NAS on internal control
quality. Column (1) presents the results from estimating companies’ propensity to purchase tax NAS (TAXNASIND
equals 1 if the company purchased tax NAS in year t, and 0 otherwise). After calculating propensity scores for all
observations, we match each treatment company (i.e., those in the highest quartile of the ratio of TAXFEES to
TOTALFEES) to a unique control company (i.e., those without tax NAS) and retain only those pairs whose scores match
within 0.01. We then reestimate Equation (1) using this propensity score matched sample. Column (2) presents the results
of reestimating Equation (1). Standard errors clustered by company are in parentheses. Our variable of interest,
LNTAXNAS, is the natural logarithm of tax fees paid to the financial statement auditor in year t. ICW equals 1 if the
company’s auditor reported an internal control weakness in year t.
Variable definitions are provided in Appendix A.
(continued on next page)
The Accounting Review
July 2015
De Simone, Ege, and Stomberg
1486
TABLE 3 (continued)
Panel C: Selection Model
LNTAXNAS ¼ b0 þ Controls þ Year FE þ e:
ProbðICW ¼ 1Þ ¼ Fðb0 þ b1 LNTAXNAS INST þ Controls þ Year FEÞ
Dependent Variables
Pred.
LNTAXNAS
(1)
Variable of Interest
LNTAXNAS_INST
Controls for Economic Dependence
LNAUDIT
þ
LNOTHERNAS
?
INFLUENCE
?
Other Controls
LMARKETCAP
þ
AGGLOSS
SHUMWAY
LSEGCOUNT
þ
FORTRANS
þ
MERGER
þ
EXTREMESG
RESTRUCTURE
þ
BIG4
?
AUDITOR_RESIGN
PCTFORSALES
þ
TAXLOSS
þ
TENURE
þ
EQINC
R&D
?
Pred.
ICW
(2)
0.0497***
(0.020)
1.0375***
(0.043)
0.0703***
(0.006)
0.5406***
(0.137)
þ
0.4286***
(0.022)
0.0021
(0.003)
0.3156***
(0.055)
0.1458***
(0.032)
0.3072***
(0.075)
0.0042
(0.015)
0.2636***
(0.053)
0.3239***
(0.068)
0.4889***
(0.074)
0.3317***
(0.073)
0.0925*
(0.065)
0.0707
(0.115)
2.0291***
(0.271)
0.5077***
(0.119)
0.0015
(0.022)
0.0246***
(0.002)
2.5497
(2.168)
0.0448
(0.246)
?
?
þ
þ
þ
þ
þ
þ
þ
þ
þ
þ
?
0.2975***
(0.017)
0.1833***
(0.031)
0.0055
(0.006)
0.0707
(0.023)
0.1198***
(0.029)
0.0106
(0.035)
0.0582**
(0.032)
0.0291
(0.018)
0.2162
(0.044)
0.5484***
(0.095)
0.2127***
(0.047)
0.0268**
(0.011)
(continued on next page)
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1487
TABLE 3 (continued)
Dependent Variables
Pred.
LEVERAGE
þ
BTM
PPE
?
ROA
?
CASH
?
DEP
SECTIER
Obs. where ICW ¼ 1
Total Obs.
LNTAXNAS
(1)
Pred.
ICW
(2)
0.3670***
(0.138)
0.0000
(0.000)
0.4057***
(0.058)
0.0289**
(0.014)
0.1612
(0.216)
4.1089***
(0.801)
1.8261***
(0.148)
32,048
2,129
32,048
*, **, *** Denote one-tailed significance at 10 percent, 5 percent, and 1 percent, respectively, when a prediction is made;
otherwise, significance denoted is based on two-tailed tests.
Panel C presents the results of an instrumental variables probit model design. We simultaneously estimate the amount of
tax NAS purchased using an OLS regression and Equation (1) with estimated LNTAXNAS as the independent variable of
interest. Column (1) presents the results when estimating a company’s amount of tax NAS purchased. Column (2)
presents the results of estimating Equation (1). Standard errors clustered by company are in parentheses. Our variable of
interest, LNTAXNAS_INST, is the estimated natural logarithm of tax fees paid to the financial statement auditor in year t.
ICW equals 1 if the company’s auditor reported an internal control weakness in year t.
Variable definitions are provided in Appendix A.
company-years with and without tax NAS purchases. Finally, we perform two additional
untabulated analyses. First, we estimate Equation (1) on a matched-pair sample of tax NAS and
non-tax NAS company-years with similar GAAP ETRs and again with similar cash ETRs. Second,
we estimate Equation (1) on a subsample of observations with GAAP or cash ETRs less than the
U.S. statutory rate of 35 percent, as ETRs below this rate reflect some degree of tax planning. In all
of these additional specifications, we continue to find a negative relation between LNTAXNAS and
ICW. These analyses further help to mitigate concerns that our results are driven by companies that
lack a commitment to internal control quality or that do not focus on tax matters.15
Triangulating Results of the Hypothesis
By Shock to Control Environment
We next examine the effect of tax NAS separately on samples of observations undergoing a
shock to their internal control environment. We expect tax NAS to have a stronger (more negative)
15
Descriptive statistics are qualitatively similar if we compare ETRs at the company level instead of by company-year.
The second untabulated regression is robust to limiting the sample to observations reporting GAAP or cash ETRs
below several alternative thresholds, including 30 percent, 25 percent, and 20 percent.
The Accounting Review
July 2015
1488
De Simone, Ege, and Stomberg
effect on the probability of disclosing a material weakness in the year of the shock. To test our
prediction, we estimate Equation (1) and include proxies aimed at capturing a new revenue stream,
new line of business, or foreign expansion, all of which likely require internal controls to be
implemented for proper financial reporting.
As a proxy for a new revenue stream or line of business, we set SEGMENT_INC to 1 for year t
when there has been an increase in the number of segments compared to year t1, as reported in the
Compustat Segments database. We expect tax NAS to have a stronger (more negative) effect on the
probability of disclosing a material weakness in the year of a segment increase because the internal
control environment must be changed to address financial reporting for the new segment. To test
our prediction, we add SEGMENT_INC and SEGMENT_INC LNTAXNAS to Equation (1).
Second, FOREIGN_EXP is set to 1 in year t if the company reports for the first time either
a change in foreign sales greater than or equal to 1 percent of total assets or a foreign currency
adjustment.16 We obtain foreign sales information from the Compustat Segments database,
which provides geographic segment information since 1976, and foreign currency adjustments
from the Compustat Annual database. We expect tax NAS to have a stronger (more negative)
effect on the probability of disclosing a material weakness in the year of a first-time foreign
expansion. To test this prediction, we add FOREIGN_EXP and FOREIGN_EXP LNTAXNAS
to Equation (1).
Table 4 presents results of these ‘‘changes’’ analyses. In all columns, we continue to find a
significantly negative main effect of LNTAXNAS, consistent with tax NAS benefiting internal
control quality, on average, even in company-years without a shock to the control environment.
Column (1) reports that the main effect of SEGMENT_INC is positive and significant, suggesting
that companies, on average, do not effectively make changes to their internal control environment
when adding a new segment. The coefficient on SEGMENT_INC LNTAXNAS is negative and
significant, suggesting an incremental effect of tax NAS on internal control quality in the presence
of a segment increase relative to company-years without a segment increase. Similarly, Column (2)
reports that the main effect of FOREIGN_EXP is positive and significant, suggesting that
companies experience a decrease in internal control quality associated with this shock to their
internal control environment. The coefficient on the interaction term is negative and significant,
suggesting that tax NAS are incrementally beneficial for companies expanding overseas relative to
other companies.17
Overall, these results are consistent with tax NAS having a more positive effect on internal
control quality when companies experience a significant shock to their internal control
environment, as predicted.
By Auditor Tenure
The second type of moderator we consider is auditor tenure. As auditor tenure increases, the audit
firm likely has more frequent communication with client personnel, suggesting that tax NAS are more
beneficial in the early years of the audit firm-client relationship. We set TENURE equal to the number
16
17
We intend for our proxy for foreign operations to capture company expansion into a new foreign jurisdiction.
However, it is difficult to capture this specific activity. Therefore, we tabulate results using a proxy that captures the
first time a company reports foreign activity. Results are robust to several alternative measures of foreign expansion,
including separately identifying the first material increase in foreign sales (where material can be defined as greater
than zero, greater than the average sample change in foreign sales, or greater than 1 or 10 percent of total assets) or
first foreign currency adjustment, as well as not limiting the indicator variable to the first instance of a material foreign
transaction. Results are also robust to using a continuous variable for the percent of foreign sales, which suggests that
as the extent of foreign operations increases, tax NAS are more beneficial.
The one-tailed p-value using the correction from Norton, Wang, and Ai (2004) is less than 3 percent for Column (1)
and less than 1 percent for Column (2).
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1489
TABLE 4
Effect of Tax NAS on Internal Control Quality
Triangulating Results
ProbðICW ¼ 1Þ ¼ Fðb0 þ b1 LNTAXNAS þ b2 MODERATOR
þb3 LNTAXNAS MODERATOR þ Controls þ Year FEÞ:
Independent Variables
Pred.
(1)
(2)
(3)
LNTAXNAS
0.0240***
(0.006)
0.0339***
(0.007)
SEGMENT_INC
þ
LNTAXNAS SEGMENT_INC
0.0262***
(0.006)
0.3993***
(0.115)
0.0181*
(0.011)
FOREIGN_EXP
þ
LNTAXNAS FOREIGN_EXP
TENURE
?
LNTAXNAS TENURE
þ
LNAUDIT
þ
LNOTHERNAS
?
INFLUENCE
?
LMARKETCAP
AGGLOSS
þ
SHUMWAY
þ
LSEGCOUNT
þ
FORTRANS
þ
MERGER
þ
EXTREMESG
þ
RESTRUCTURE
þ
BIG4
þ
AUDITOR_RESIGN
þ
PCTFORSALES
þ
0.4799***
(0.118)
0.0325***
(0.011)
0.8429***
(0.053)
0.0007
(0.006)
0.6047***
(0.153)
0.6372***
(0.037)
0.3584***
(0.067)
0.0119
(0.013)
0.1920
(0.055)
0.2233***
(0.073)
0.0128
(0.077)
0.1211**
(0.067)
0.0595
(0.036)
0.5277
(0.084)
1.0697***
(0.156)
0.3733***
(0.107)
0.8467***
(0.054)
0.0005
(0.006)
0.5966***
(0.153)
0.6350***
(0.037)
0.3544***
(0.068)
0.0091
(0.013)
0.1691
(0.054)
0.1692**
(0.075)
0.0066
(0.077)
0.1268**
(0.067)
0.0569
(0.034)
0.5338
(0.084)
1.0517***
(0.159)
0.3471***
(0.108)
0.0204***
(0.006)
0.0009**
(0.001)
0.8496***
(0.054)
0.0002
(0.006)
0.6251***
(0.151)
0.6276***
(0.037)
0.3456***
(0.067)
0.0096
(0.013)
0.1436
(0.054)
0.2048***
(0.073)
0.0131
(0.077)
0.1203**
(0.066)
0.0567
(0.034)
0.4521
(0.087)
0.9921***
(0.157)
0.3481***
(0.107)
(continued on next page)
The Accounting Review
July 2015
De Simone, Ege, and Stomberg
1490
TABLE 4 (continued)
Independent Variables
TAXLOSS
Pred.
?
Obs. where ICW ¼ 1
Total Obs.
Pseudo R2
(1)
0.0572**
(0.028)
2,129
32,048
18.74%
(2)
(3)
0.0582**
(0.028)
2,129
32,048
18.74%
0.0599**
(0.029)
2,129
32,048
18.90%
*, **, *** Denote one-tailed significance at 10 percent, 5 percent, and 1 percent, respectively, when a prediction is made;
otherwise, significance denoted is based on two-tailed tests.
This table presents the results of estimating Equation (1) including variables indicating a shock to the organizational
structure or auditor tenure. Standard errors clustered by company are in parentheses. Our variable of interest is the
interaction of LNTAXNAS and one of three moderator variables. ICW equals 1 if the company’s auditor reported an
internal control weakness in year t.
Variable definitions are provided in Appendix A.
of years that the audit firm has been engaged by the client to perform the audit. We add TENURE and
TENURE LNTAXNAS to Equation (1), and expect the coefficient on the interaction to be positive.
Results are presented in Table 4, Column (3). The coefficient on TENURE LNTAXNAS is
positive and significant, suggesting that the effect of tax NAS is greatest earlier in the audit firm
tenure.18 This result is also inconsistent with impaired auditor independence affecting the disclosure
of material weaknesses. If audit tenure and tax NAS impaired audit firm independence, then we would
expect a negative coefficient on TENURE LNTAXNAS, suggesting that an increase in tenure and tax
NAS would interact and incrementally result in fewer material weaknesses. However, the coefficient
on TENURE LNTAXNAS is positive, which is consistent with our expectations.
V. ADDITIONAL ANALYSES AND ROBUSTNESS TESTS
Independence
To further rule out economic bonding as an alternative explanation for our findings, we test for
an association between tax NAS and instances where we observe that an internal control opinion
was incorrect. Upon a financial statement restatement, if there is a new material weakness for the
restated period that needs to be disclosed, then the auditor restates the internal control opinion in
addition to the financial statement opinion. In these situations, the auditor is already admitting to
giving an incorrect financial statement opinion. Thus, admitting that the internal control opinion
was also incorrect should have low incremental reputational cost for the audit firm.
If economic bonding explained our results, then we would expect tax NAS to be positively
associated with restated internal control opinions that reveal new material weaknesses, suggesting
that the audit firm initially issued an incorrect internal control opinion because independence was
compromised and subsequently corrected that opinion because the financial statements had to be
restated. We reestimate Equation (1), but replace the dependent variable with RESTATED_ICO.
This variable is set equal to 1 if a company subsequently restates an internal control opinion in
conjunction with a financial restatement and this restated internal control opinion reports a new
material weakness. As such, this variable indicates that the original internal control opinion was
18
The one-tailed p-value using the correction from Norton et al. (2004) is less than 2 percent.
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1491
TABLE 5
Effect of Tax NAS on Internal Control Restatements
ProbðRESTATED ICO ¼ 1Þ ¼ Fðb0 þ b1 LNTAXNAS þ b2 LNOTHERNAS
þb3 INFLUENCE þ Controls þ Year FEÞ:
Independent Variables
Pred.
LNTAXNAS
?
LNAUDIT
LNOTHERNAS
?
INFLUENCE
?
Obs. where RESTATED_ICO ¼ 1
Total Obs.
Pseudo R2
0.0039
(0.013)
0.2260**
(0.097)
0.0058
(0.014)
0.0008
(0.322)
293
2,461
0.05
*, **, *** Denote one-tailed significance at 10 percent, 5 percent, and 1 percent, respectively, when a prediction is made;
otherwise, significance denoted is based on two-tailed tests.
This table presents the results of whether tax NAS have an effect on the probability of a restated internal control opinion.
Standard errors clustered by company are in parentheses. Our variable of interest, LNTAXNAS, is the natural logarithm of
tax fees paid to the financial statement auditor in year t. RESTATED_ICO equals 1 for restated company-years where the
number of disclosed material weaknesses increased (i.e., the original internal control opinion was incorrect). The sample
for this test includes company-years with misstated financial statements (i.e., those company-years where the financial
statements were restated due to material departures from GAAP).
Variable definitions are provided in Appendix A.
incorrect. In Table 5, the coefficient on LNTAXNAS is insignificant. Therefore, it does not appear
that our main results reflect impairment of auditor independence.
The Effect of Tax NAS on Tax and Non-Tax Internal Control Quality
Examples from practitioners presented above illustrate that tax NAS can improve internal
controls over all reporting processes, not only those over income tax reporting. Therefore, we
investigate whether tax NAS are associated with non-tax internal control quality by separately
testing for the effect of tax NAS on non-tax and tax internal control quality. In Table 6, we present
the results of estimating Equation (1) after replacing ICW with ICW_NONTAX and ICW_TAX in
separate regressions. We find a negative association between our measure of tax NAS and the
likelihood of disclosing both types of material weakness, consistent with the findings in Harris and
Zhou (2013). These results suggest that tax NAS provide benefits to both types of internal control
quality.19 However, we caution the reader in making conclusions about the effects of tax NAS on
19
Harris and Zhou (2013) hypothesize that if tax NAS focus on tax matters, then tax NAS should be more negatively
associated with tax than non-tax internal control weaknesses. However, they find that tax NAS are more negatively
associated with non-tax than tax internal control weaknesses and characterize this finding as ‘‘consistent with both a
reduction in auditor independence due to the consulting and/or auditor-tax consultants providing assistance that is
substantially related to non-tax issues and potentially beyond the scope of the engagement approved by the audit
committee’’ (Harris and Zhou 2013, 15–16). In contrast, we make no prediction about the relative magnitude of the
effect on different types of internal controls.
The Accounting Review
July 2015
De Simone, Ege, and Stomberg
1492
TABLE 6
Effect of Tax NAS on Non-Tax and Tax Internal Control Quality
ProbðICW ¼ 1Þ ¼ Fðb0 þb1 LNTAXNAS þ Controls þ Year FEÞ:
Independent
Variables
Pred.
LNTAXNAS
LNAUDIT
þ
LNOTHERNAS
?
INFLUENCE
?
Obs. where ICW_NONTAX or ICW_TAX ¼ 1
Total Obs.
Pseudo R2
ICW_NONTAX
(1)
0.0295***
(0.006)
0.6309***
(0.057)
0.0042
(0.007)
0.7474***
(0.152)
1,488
32,048
0.14
ICW_TAX
(2)
0.0241***
(0.010)
1.0945***
(0.077)
0.0052
(0.010)
0.0110
(0.303)
641
32,048
0.17
*, **, *** Denote one-tailed significance at 10 percent, 5 percent, and 1 percent, respectively, when a prediction is made;
otherwise, significance denoted is based on two-tailed tests.
This table presents the results of estimating Equation (1) using tax and non-tax internal control weaknesses as the
dependent variable. Standard errors clustered by company are in parentheses. Our variable of interest, LNTAXNAS, is the
natural logarithm of tax fees paid to the financial statement auditor in year t. ICW_NONTAX equals 1 if the company
disclosed a material weakness in internal controls unrelated to tax in year t. ICW_TAX equals 1 if the company disclosed
a material weakness in internal controls related to tax in year t.
Variable definitions are provided in Appendix A.
specific types of material weaknesses because auditors and managers are not required to follow a
particular reporting taxonomy when disclosing material weaknesses. Thus, auditors and managers
are free to describe material weaknesses in detail or in summary, potentially making it difficult to
determine whether a disclosed material weakness relates to tax or to non-tax issues.20
VI. CONCLUSIONS
Prior research reports a positive association between tax NAS and both financial reporting quality
and audit quality. We extend this literature by providing evidence that tax NAS improve internal
control quality, which is an important component of financial reporting quality. We propose that tax
NAS improve internal control quality by facilitating earlier audit firm awareness of material
transactions. Consistent with predictions, we find a negative relation between tax NAS and the
likelihood of a material weakness in internal control from 2004 through 2012. This finding is robust to
additional analyses intended to address concerns of auditor independence impairment and endogeneity.
20
Another issue related to classifying material weakness disclosures is that Audit Analytics appears to classify
disclosures based upon types of issues that are mentioned as opposed to the root cause of the issue. For example, the
material weakness disclosures from Symantec Corporation for the year ended March 31, 2006 are flagged as having
both tax- and acquisition-related material weaknesses. The description from Symantec Corporation’s 10-K (Securities
and Exchange Commission [SEC] 2006, 19) says, ‘‘Management has determined that we had insufficient personnel
resources with adequate expertise to properly manage the increased volume and complexity of tax matters associated
with the acquisition of Veritas Software Corporation.’’ It seems that the root cause of the material weaknesses is a tax,
not an acquisition, issue.
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1493
Additionally, we identify and test settings in which we expect the benefits of tax NAS to be
stronger. We find that the benefits of tax NAS are greatest when a company experiences a shock to
its internal control environment and earlier in the audit firm’s tenure. This latter result provides
additional evidence contrary to the argument that tax NAS impair auditor independence. Overall,
our findings suggest that tax NAS are an important determinant of internal control quality and are
consistent with tax NAS improving internal control quality by accelerating audit firm awareness of
transactions material to the financial statements. These results are of interest to shareholders and
audit committees assessing the costs and benefits of tax NAS, as well as to the various regulators
around the globe that either have banned tax NAS or are considering restrictions on tax NAS.
REFERENCES
Ashbaugh-Skaife, H., D. W. Collins, and W. R. Kinney, Jr. 2007. The discovery and reporting of internal
control deficiencies prior to SOX-mandated audits. Journal of Accounting and Economics 44: 166–192.
Bartus, T. 2005. Estimation of marginal effects using margeff. Stata Journal 5 (3): 309–329.
Bauer, A. 2014. Tax Avoidance and the Implications of Weak Internal Controls. Working paper, University
of Illinois.
Bédard, J., and S. Paquette. 2011. Perception of Auditor Independence, Audit Committee Characteristics,
and Auditor Provision of Tax Services. Working paper, Université Laval.
Cheng, M., D. Dhaliwal, and Y. Zhang. 2013. Does investment efficiency improve after the disclosure of
material weaknesses in internal control over financial reporting? Journal of Accounting and
Economics 56: 1–18.
DeFond, M. L., K. Raghunandan, and K. R. Subramanyam. 2002. Do non-audit service fees impair auditor
independence? Evidence from going concern audit opinions. Journal of Accounting Research 40 (4):
1247–1274.
Deloitte. 2011. Material Weaknesses and Restatements: Is Tax Still in the Hot Seat? New York, NY:
Deloitte.
Doyle, J. T., W. Ge, and S. McVay. 2007. Determinants of weaknesses in internal control over financial
reporting. Journal of Accounting and Economics 44: 193–223.
Ernst & Young (EY). 2013. Navigating the Choppy Waters of International Tax: Global Transfer Pricing
Survey. Available at: http://www.ey.com/Publication/vwLUAssets/EY-2013_Global_Transfer_
Pricing_Survey/$FILE/EY-2013-GTP-Survey.pdf
Francis, J. R. 2004. What do we know about audit quality? British Accounting Review 36: 345–368.
Francis, J. R. 2006. Are auditors compromised by nonaudit services? Assessing the evidence.
Contemporary Accounting Research 23 (3): 747–760.
Gleason, C. A., and L. F. Mills. 2011. Do auditor provided tax services improve the estimate of tax
reserves? Contemporary Accounting Research 28 (5): 1484–1509.
Gleason, C. A., L. F. Mills, and M. L. Nessa. 2013. Did SOX and FIN 48 Eliminate the Knowledge
Spillover Benefits of Auditor-Provided Tax Services? Working paper, University of Iowa.
Harris, D., and J. Zhou. 2008. Tax Consulting and Reported Weaknesses in Internal Control. Working
paper, Syracuse University and SUNY.
Harris, D., and J. Zhou. 2013. Auditor-Provided Tax Consulting, Knowledge Spillovers, and Reported
Weaknesses in Internal Control. Working paper, Syracuse University and University of Hawaii at
Manoa.
Hogan, C. E., and M. S. Wilkins. 2008. Evidence on the audit risk model: Do auditors increase audit fees in
the presence of internal control deficiencies? Contemporary Accounting Research 25 (1): 219–242.
Johnson, V. E., I. K. Khurana, and J. K. Reynolds. 2002. Audit-firm tenure and the quality of financial
reports. Contemporary Accounting Research 19 (4): 637–660.
Kinney, W. R., Z.-V. Palmrose, and S. Scholz. 2004. Auditor independence, non-audit services, and
restatements: Was the U.S. government right? Journal of Accounting Research 42 (3): 561–588.
The Accounting Review
July 2015
De Simone, Ege, and Stomberg
1494
Larsen, R. G. 2011. Communicating with the Board of Directors on Taxes. (July/August). Tax Executives
Institute, Inc. Available at: https://www.tei.org/news/articles/Documents/Communicating%20with%
20the%20Board%20of%20Directors%20on%20Taxes.pdf
Lassila, D., T. Omer, M. Shelley, and L. Smith. 2010. Do complexity, governance, and auditor
independence influence whether firms retain their auditors for tax services? Journal of the American
Taxation Association 32 (1): 1–23.
McGuire, S., T. Omer, and D. Wang. 2012. Tax avoidance: Does tax-specific industry expertise make a
difference? The Accounting Review 87 (3): 975–1003.
Myers, J., L. Myers, and T. Omer. 2003. Exploring the term of the auditor-client relationship and the quality
of earnings: A case for mandatory auditor rotation? The Accounting Review 78 (3): 779–799.
Norton, E. C., H. Wang, and C. Ai. 2004. Computing interaction effects and standard errors in logit and
probit models. Stata Journal 4: 154–167.
Public Company Accounting Oversight Board (PCAOB). 2007. Professional Standards—Rule 3525: Audit
Committee Pre-Approval of Non-Audit Services Related to Internal Control Over Financial
Reporting (July 27). Available at: http://www.pcaobus.org/Rules/PCAOBRules/Pages/Section_3.
aspx#rule3525
Public Company Accounting Oversight Board (PCAOB). 2011. Concept Release on Auditor Independence
and Audit Firm Rotation (August 16). Release No. 20122-006. Available at: http://www.pcaobus.
org/rules/rulemaking/docket037/release_2011-006.pdf
Reynolds, J. K., and J. R. Francis. 2001. Does size matter? The influence of large clients on office-level
auditor reporting decisions. Journal of Accounting and Economics 30: 375–400.
Robinson, D. 2008. Auditor independence and auditor-provided tax service: Evidence from going-concern
audit opinions prior to bankruptcy filings. Auditing: A Journal of Practice & Theory 27 (2): 31–54.
Robinson, J. R., S. A. Sikes, and C. D. Weaver. 2010. Performance measurement of corporate tax
departments. The Accounting Review 85 (3): 1035–1064.
Securities and Exchange Commission (SEC). 2006. Symantec Corporation Form 10-K. Available at: http://
www.sec.gov/Archives/edgar/data/849399/000089161806000254/f21002e10vk.htm
Shumway, T. 2001. Forecasting bankruptcy more accurately: A simple hazard model. Journal of Business
74: 101–124.
U.S. House of Representatives. 2002. The Sarbanes-Oxley Act of 2002. Public Law 107-204. Washington,
DC: Government Printing Office.
APPENDIX A
Variable Definitions
Dependent Variables
ICW ¼ indicator variable equal to 1 if the company’s auditor reported an internal control
weakness in year t, and 0 otherwise;
ICW_NONTAX ¼ indicator variable equal to 1 if the company’s auditor reported an internal
control weakness unrelated to tax in year t and did not report an internal control weakness
related to tax in year t, and 0 otherwise; and
ICW_TAX ¼ indicator variable equal to 1 if the company’s auditor reported an internal control
weakness related to tax in year t, and 0 otherwise.
Auditor Fees from Audit Analytics
TAXFEES ¼ total fees paid to the auditor for tax services in year t;
TAXNASIND ¼ indicator variable equal to 1 if the company purchased tax NAS in year t, and 0
otherwise;
AUDITFEES ¼ total fees paid to the auditor for audit services in year t;
The Accounting Review
July 2015
Internal Control Quality: The Role of Auditor-Provided Tax Services
1495
TOTALFEES ¼ total audit, audit-related, tax, and other fees paid to the auditor in year t; and
LNTAXNAS ¼ natural logarithm of TAXFEES.
Control Variables from Audit Analytics, Compustat, and CRSP
LNAUDIT ¼ natural logarithm of AUDITFEES;
LNOTHERNAS ¼ natural logarithm of other nonaudit fees paid to the auditor in year t;
INFLUENCE ¼ company-year ratio capturing TOTALFEES as a percentage of the total fees the
company’s audit firm received from all public companies it audits from the same practice
office. We identify the practice office based upon the office listed in the auditor’s signature
in the audit opinion as reported in Audit Analytics;
LMARKETCAP ¼ natural logarithm of market capitalization (PRCC_F CSHO) at the end of
year t;
AGGLOSS ¼ indicator variable equal to 1 if earnings before extraordinary items (IB) in years t
and t1 sum to less than 0, and 0 otherwise;
SHUMWAY ¼ decile rank of the percentage probability of bankruptcy in year t from the default
hazard prediction model based on Shumway (2001). Higher scores indicate a higher
probability of bankruptcy;
LSEGCOUNT ¼ natural logarithm of the number of operating and geographic segments in year
t as reported in the Compustat Segments database;
FORTRANS ¼ indicator variable equal to 1 if the company reports non-zero foreign currency
translation (FCA) in year t, and 0 otherwise;
MERGER ¼ indicator variable equal to 1 if the company reports non-zero acquisition expense
(AQP) in year t or t1, and 0 otherwise;
EXTREMESG ¼ indicator variable equal to 1 if year-over-year, industry-adjusted sales growth
(SALE) is in the top quintile in year t, and 0 otherwise;
RESTRUCTURE ¼ sum of restructuring amounts reported in years t1 and t, scaled by
LMARKETCAP. We multiply values by 1 such that net restructuring expense (income) is
positive (negative);
BIG4 ¼ indicator variable equal to 1 if the company is audited by Deloitte, Ernst & Young,
KPMG, or PricewaterhouseCoopers in year t, and 0 otherwise;
AUDITOR_RESIGN ¼ indicator variable equal to 1 if the company experienced an auditor
resignation in year t, as reported in Audit Analytics, and 0 otherwise;
PCTFORSALES ¼ percentage of total sales (SALE) attributable to foreign segments as
reported in the Compustat Segments database; and
TAXLOSS ¼ tax net operating loss carryforward reported at the end of year t, scaled by total
assets at the end of year t (TLCF/AT).
Moderator Variables
SEGMENT_INC ¼ indicator variable equal to 1 for year t when there has been an increase in
the number of business, operating, geographic, or state segments compared to year t1;
FOREIGN_EXP ¼ indicator variable equal to 1 if the company reports for the first time either
an increase in foreign sales (non-U.S. geographic SALES from the Compustat Segment
database) greater than or equal to 1 percent of total assets or a foreign currency adjustment
(FCA); and
TENURE ¼ length of the auditor’s tenure with the client as of year t.
Additional Control Variables used in Selection Model
EQINC ¼ equity income in year t, scaled by total assets at the end of year t1 (ESUB/AT);
The Accounting Review
July 2015
1496
De Simone, Ege, and Stomberg
R&D ¼ research and development expense in year t, scaled by total assets at the end of year t1
(XRD/AT). If XRD is missing, then XRD is set to 0;
LEVERAGE ¼ long-term debt at the end of year t, scaled by total assets at the end of year t
(DLTT/AT);
BTM ¼ book value of common equity at the end of year t, scaled by market value of equity at
the end of year t (CEQ/PRCC_F CSHO);
PPE ¼ net property, plant, and equipment at the end of year t, scaled by total assets at the end
of year t1 (PPENT/AT);
ROA ¼ return on assets, measured as the ratio of income before extraordinary items to average
total assets reported in years t1 and t (IB/AT);
CASH ¼ cash holdings at the end of year t, scaled by total assets at the end of year t1 (CH/
AT);
DEP ¼ depreciation expense for year t, scaled by total assets at the end of year t1 (DEP/AT);
SECTIER ¼ indicator variable equal to 1 if the company is audited by Grant Thornton or BDO
Seidman, and 0 otherwise; and
RESTATED_ICO ¼ indicator variable equal to 1 for restated company-years where the number
of disclosed material weaknesses increased (i.e., the original internal control opinion was
incorrect), and 0 otherwise.
The Accounting Review
July 2015