INVESTING IN THE CORPORATE TAX FUNCTION: THE EFFECTS OF
REMEDIATING MATERIAL WEAKNESSES IN INTERNAL CONTROL ON TAX
AVOIDANCE
By
Daniel Lynch

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
Business Administration – Doctor of Philosophy
2014

ABSTRACT
INVESTING IN THE CORPORATE TAX FUNCTION: THE EFFECTS OF REMEDIATING
MATERIAL WEAKNESSES IN INTERNAL CONTROL ON TAX AVOIDANCE
By
Daniel Lynch
This study investigates the effect of increases in firm-level investments in the corporate
tax function to address financial reporting concerns on the magnitude and timing of firm’s tax
avoidance. I exploit the novel setting of the remediation of tax-related material weaknesses in
internal control (MWs) as a shock that motivates firms to increase investments in the corporate
tax function. I examine the impact of increases in these investments on the magnitude and timing
of future payoffs in the form of increased tax avoidance. I find no evidence of increases in tax
function investments contemporaneously impacting tax avoidance. However, I do find that
investments in the income tax function are associated with greater levels of future tax avoidance.
Specifically, I find firms making investments in year t have 4-8 percentage point lower 3-year
cash and GAAP effective tax rates (ETR) measured at time t+1 to t+3 compared to their industry
peers. Descriptive results suggest the primary drivers of this increase in future tax avoidance are
lower foreign and state tax expense. I also hand-collect internal control disclosures to investigate
how firms remediate tax-related MWs. The results suggest the most common types of
investments are in outside consulting and new personnel, including hiring a new tax director. I
find that both external and internal tax function investments are positively associated with future
tax avoidance, with no differential impact on future ETRs. Results also indicate that risk,
complexity, and a lack of resources are predictors of engaging an external service provider.
These results demonstrate the magnitude and timing lag of payoffs on tax function investments

 

and illustrate how strengthening internal controls can improve firm performance via increased
tax avoidance.

 

ACKNOWLEDGMENTS

This dissertation has been the culmination of a five year journey at Michigan State
University. There have been many people who have worked so hard to train and support me
throughout this process. I want to thank my dissertation chair and advisor Professor Sanjay
Gupta, who provided the continuous support and guidance that I needed throughout the Ph.D.
program. He challenged me to be the best academic I can be, and I am forever grateful for his
trust, support and guidance. In addition to Sanjay, I am very fortunate and thankful to have
Professors Ed Outslay and Kathy Petroni serving on my dissertation committee. Their comments
and guidance throughout the various stages of the PhD program and especially the dissertation
process were invaluable. Furthermore, I appreciate how they took interest in my career at an
early stage and appreciate having been able to work with them in co-authored projects. I am also
thankful to Professor Charlie Hadlock from the finance department for serving on my committee
and providing feedback and thoughtful outside perspective that greatly benefited this
dissertation.
I want to thank several special people who have had a major impact on my academic
career. First, Professor Charlie Bokemeier for inspiring me to become an accounting major as an
undergraduate at MSU and for shaping me as teacher. I will be forever indebted to him for
allowing me to serve as a student assistant in his course and encouraging me to pursue my PhD. I
want to thank my parents, Alfred and Bozena Lynch, for instilling in me at an early age a strong
work ethic and a passion for school. They are two of the hardest working people I know, and I
am so grateful to them for everything they have done for me.

iv
 

Last, but certainly not least, I want to thank my wonderful wife Laura and our daughter
Eleonora. Laura was the first person to suggest that I return to school and pursue a PhD. I am
eternally grateful for her recognizing that academia would be a great fit for me. She has been so
supportive and was always there during the stressful times even while pursuing and completing
her own PhD program. I appreciate all of the dinner table conversations about statistical analyses
and look forward to the many conversations to come. I am especially thankful for our daughter
Elli who was born during the program, and forever changed my outlook on life and my
perspective of what truly matters. I am so lucky to have so many amazing people in my life, and
I look forward to a long and prosperous career in the years to come.

v
 

TABLE OF CONTENTS

LIST OF TABLES

viii

CHAPTER 1: INTRODUCTION

1

CHAPTER 2: LITERATURE REVIEW AND HYPOTHESES
2.1. Disclosure and Remediation of Internal Control Weaknesses
2.2. Tax Avoidance Determinants
2.3. Investments in the Tax Function
2.4. Determinants of Outsourcing
2.5. Hypothesis Development
2.5.1 Tax Function Investments and Tax Avoidance
2.5.2 Internal Versus External Investments and Future Tax Avoidance
2.5.3 Determinants of External Versus Internal Tax Function Investments

8
8
11
12
15
16
16
19
20

CHAPTER 3: RESEARCH DESIGN AND SAMPLE SELECTION
3.1 Measures
3.1.1 Measure of Tax Avoidance (Dependent Variable)
3.1.2 Measure of Tax Function Investment
3.1.3 Control Variables for Tax Avoidance
3.2. Research Design
3.2.1. Tax Function Investments and Tax Avoidance
3.2.2. Tax Avoidance Payoffs on External Versus Internal Investments
3.2.3. Determinants of External Tax Function Investments
3.3. Sample Selection and Data

23
23
23
24
25
26
26
28
29
30

CHAPTER 4: MAIN RESULTS
4.1. Descriptive Statistics and Univariate Results
4.1.1. Descriptive Statistics
4.1.2. Univariate Results
4.2. Multivariate Results
4.2.1. Tax Function Investments and Tax Avoidance
4.2.2. How Firms Increase Future Tax Avoidance
4.2.3. External Versus Internal Investments and Future Tax Avoidance
4.2.4. Determinants of External Tax Function Investments

33
33
33
34
36
36
38
40
41

CHAPTER 5: SUPPLEMENTAL ANALYSES AND SENSITIVITY TESTS
5.1. Tax Function Investment Firm Characteristics
5.2. Timing of Payoffs on Tax Function Investments
5.3. Additional Control Variables
5.4. Alternative Measures of Tax Avoidance

42
42
44
45
46

vi
 

CHAPTER 6: CONCLUDING REMARKS

47

APPENDICES
APPENDIX A: Tax-Related MW Disclosures
APPENDIX B: Variable Definitions
APPENDIX C: Tables

49
50
58
61

REFERENCES

80

vii
 

LIST OF TABLES

TABLE 1:

MWs Disclosed by Year and Time to Remediate

62

TABLE 2:

Sample Selection

63

TABLE 3:

Descriptive Statistics Type of Investment by Year

64

TABLE 4:

Descriptive Statistics Main and Matched Samples

66

TABLE 5:

OLS ETR Regressions Main Analysis

69

TABLE 6:

Disclosed Factors for Increase in Tax Avoidance

71

TABLE 7:

OLS 3-year ETR Regressions External and Internal Investments

72

TABLE 8:

Logistic Regression: Determinants of External Investments

73

TABLE 9:

OLS Changes and Firm Fixed Effects Specifications

74

TABLE 10:

OLS 3-year ETR Regressions-Additional Controls

75

TABLE 11:

OLS 3-year ETR Regressions MW Subsamples

76

TABLE 12:

OLS 1-year ETR Regressions (t+1,t+2,t+3)

77

TABLE 13:

OLS 3-year ETR Regressions-Unadjusted Tax Avoidance Mean Controls

79

viii
 

CHAPTER 1: INTRODUCTION

The Internal Revenue Service (IRS) estimates that corporate taxpayers spend over $25
billion dollars annually to prepare and file federal corporate income tax returns (Contos et al.
2012). In addition, firms face costs related to tax filings in foreign and state jurisdictions, tax
planning, and complying with financial reporting requirements related to the income tax
provision. I define the sum of all internal and external firm resources devoted to these activities
(i.e., preparation, filing, compliance, planning, and financial reporting) as a firm’s investment in
the corporate tax function. Despite the magnitude of these investments and the materiality of
income tax expense for most companies, we know surprisingly little about how they influence
tax avoidance behavior.
I use a novel setting in which I observe a firm’s investment in its tax function made to
address a financial reporting concern to examine the impact of increases in these investments on
both the magnitude and timing of future payoffs in the form of increased tax avoidance.
Specifically, I exploit disclosures about the remediation of tax-related material weaknesses in
internal control (MW) under the Sarbanes Oxley Act of 2002 (SOX) as a setting in which firms
increase investments in the corporate tax function. Hand-collected tax-related MW remediation
disclosures provide information on the specific types of investments being made (e.g., hiring new
personnel/tax director, engaging an outside consultant, and changing procedures). Similar to
research using the SOX-mandated disclosures to directly test the causal link between financial
reporting quality and investment efficiency (Cheng et al. 2013), I use the disclosure of a taxrelated MW and its subsequent remediation as a strong setting to test the effects of increases in
investments in the tax function on tax avoidance. Additionally, using data on the different types
1
 

of investments being made, I investigate why firms engage an external service provider versus
investing internally and examine the impact of this choice on tax avoidance.
Because a primary goal of internal control is to provide assurance regarding the
effectiveness of operations (COSO 2011), a firm’s disclosure of a tax-related MW is indicative
of insufficient internal control systems in place to provide assurance about the effectiveness of
the tax function, which is often evaluated on its ability to avoid taxes (Armstrong et al. 2012).
Due to the complexity of the tax accounts and some firms’ apparent lack of resources/expertise
in the area, tax-related MWs are the most common type of MW disclosed (Deloitte 2011).1
Remediation of tax-related MWs therefore requires a firm to make an investment in the tax
function if it chooses to address the disclosed MW. Thus, I use the remediation of tax-related
MWs as a setting in which firms increase their tax function investment to provide evidence on
the how these investments impact tax avoidance. Additionally, by examining the impact of the
remediation of MWs on tax avoidance I add to literature examining the impact of internal control
improvements due to SOX on firm performance (Feng et al. 2013; Cheng et al. 2013).
Bauer (2013) finds the disclosure of a tax-related MW is negatively associated with the
level of contemporaneous and prior tax avoidance. This result is consistent with deficiencies in
tax-related internal controls indicating reduced tax function effectiveness and suggests
investments in the corporate tax function to remediate MWs could increase tax avoidance.
Further, because it is likely that these investments go beyond simply addressing internal control
deficiencies related to accounting for income taxes, I hypothesize a spillover effect of these
investments onto a firm’s tax avoidance activities. Specifically, I conjecture that the improved
information environment following a tax function investment to address internal control issues
                                                            
1

Additionally, SOX independence rules led many firms to disclose a tax-related MW due to their reliance on their
independent auditor for tax provision services. http://www.nysscpa.org/printversions/cpaj/2006/706/p38.htm

2
 

will facilitate greater tax planning. However, such investments might not yield tax avoidance
benefits if the remediation of a tax-related MW simply corrects issues related to the financial
reporting of the income tax provision. Additionally, given the complex nature of most tax
planning, investments in the income tax function may not allow a firm to contemporaneously
realize the benefits of a tax planning spillover effect.
Financial statement disclosures indicate that the most common types of tax function
investments are engaging an external advisor (58.1%), hiring additional internal personnel
(55.2%), and hiring a new tax director (34.3%). These disclosures suggest the majority of
remediation investments go beyond simply changing procedures to address accounting for the
income tax issue(s) that led to the disclosed tax-related MW, supporting a spillover effect onto
tax avoidance. However, because it could take time for new consultants, personnel, and tax
directors to learn the business and correct internal control issues before being able to increase tax
avoidance, these investments likely will not yield an immediate payoff. I therefore predict that
any tax avoidance benefits lag tax function investments.
Prior literature documents a positive association between the level of tax planning
expenditures and tax avoidance (Mills et al. 1998; Omer et al. 2006).2 Additionally, Cheng et al.
(2012) examine the impact of hedge fund activism, which presumably is accompanied by an
increased tax function investment, on corporate tax avoidance. The results suggest increases in
hedge fund ownership are positively associated with contemporaneous tax avoidance. Although I
cannot directly observe tax function expenditures, I use the disclosure of a tax-related MW and
subsequent remediation as a shock that provides an observable setting in which to test the effects

                                                            
2

Mills et al. (1998) uses survey data from 1991 on the level of tax planning expenditures to examine the association
with effective tax rates. Omer et al. (2006) uses the level of purchased auditor provided tax services at time t as a
proxy for tax planning expenditures to examine the impact on effective/marginal tax rates at time t+1.

3
 

of increases in investments in the tax function on tax avoidance. Hand-collected tax-related MW
remediation disclosures provide information on the specific types of tax function investments
(e.g., hiring a tax director, engaging an outside consultant, and changing procedures). By using
data on the different types of investments being made, I also investigate the impact of external
investments versus internal investments on tax avoidance and the determinants of external
investments.
To account for the specialized setting of the tax function investments examined in this
study (i.e., the remediation of a tax-related MW) and to control for industry, year, firm size, and
other characteristics that impact the opportunities to engage in tax avoidance, I measure tax
avoidance using mean-adjusted cash and GAAP effective tax rates (ETR) (Balakrishnan et al.
2012). These relative measures allow for a comparison of each firm’s ETR to its year and sizeadjusted industry mean to examine the effects of investments in the tax function on how a firm’s
tax avoidance compares to its competitors. I define these tax avoidance measures as each firm’s
relative ETR.
The results using a sample of 105 firm-year observations remediating tax-related MW
and over 7,000 control firm-years during the period 2005- 2009, and after controlling for the
presence of other types of MWs and their remediation, indicate tax function investments in year t
are negatively associated with 3-year relative ETRs measured from time t+1 to t+3. Specifically,
firms that invest in the corporate income tax function in year t have 3-year future cash (GAAP)
ETRs that are 8.1 percentage points (4.3 percentage points) lower than their industry peers.
Additional analyses suggest that this result implies a decrease of between $43.4 and $104.2
million in cash taxes paid (between $29.0 and $57.9 million in total tax expense) over the three

4
 

year period.3 In contrast, I find no evidence that tax function investments in year t are associated
with contemporaneous tax avoidance.
I also hand-collect MD&A and tax footnote data for firms with the largest increases in
future tax avoidance to investigate how firms increase their tax avoidance subsequent to
remediating a tax-related MW.4 Results suggest the largest increases in tax avoidance following
a tax function investment are associated with the use of foreign (58%) and state (29%) tax
planning strategies. Other disclosed factors include changes in tax accruals, increased research
and development credits, and tax-advantaged mergers and acquisitions.5 These descriptive
statistics suggest firms engage in more sophisticated tax planning strategies following an
increased investment in the corporate tax function.
I also document a positive association between both external and internal tax function
investments and future tax avoidance. Additionally, I find no evidence of differential tax
avoidance payoffs on external versus internal tax function investments. Finally, results for the
determinants of external tax function investments suggest resource constraints and
risk/complexity are positively associated with the decision to engage an external service
provider.
To address sample selection and endogeneity concerns related to the type of firm that
discloses and remediates tax-related MWs, I employ several alternative research design

                                                            
3

Prior literature estimates the costs of complying with SOX to be 0.06% of revenues (Jeffrey and Lourens 2008).
Mean revenues of $1.75 billion in the tax function investment sample indicates total SOX compliance costs of $1.05
million, on average. Although I cannot estimate a return on investment as most firms do not disclose the dollar
amount spent on remediation of MWs, the coefficient estimates imply economically significant returns in
comparison to SOX compliance costs.
4
Specifically, I examine the top quintile of cash ETR changes following a tax function investment (n = 24).
5
I examine the discussion of the GAAP ETR in the MD&A and tax footnotes for the three years following an
increase in a tax function investment for explanations of why the GAAP ETR decreased during that period and then
code (1/0 indicator) for the disclosure of certain reasons (e.g., lower foreign tax expense due to a better geographic
mix of income).

5
 

approaches. First, I confirm the results using a matched sample design. Second, I employ
simulation analyses to randomly assign the tax function investment indicator variable to
alternative years of firms that have a MW during my sample period. The results using this
simulation test suggest my primary findings are not driven by chance or other characteristics of
firms that disclose a MW. Finally, I use various sub-samples of firms including: 1) large firms
(S&P 1500), 2) firms that disclose any MW at some point in the sample period, 3) firms that
disclose only a tax-related MW at some point in the sample period, and 4) firms that disclose and
remediate a tax-related MW at some point in the sample period. Results are qualitatively similar
using these sub-samples, reinforcing the main results that increases in tax function investments
are associated with higher future tax avoidance. These alternative sub-sample specifications help
alleviate concerns that the results are driven by characteristics unique to firms that disclose MWs
or a time trend in tax avoidance
Research surrounding the implementation of SOX suggests that disclosing MWs provides
information about prior firm actions and is positively associated with poor accounting quality
(Ashbaugh-Skaife et al. 2008), restatements (Doyle et al. 2007a; Ashbaugh-Skaife et al. 2007),
audit delays (Ettredge et al. 2006), and rent extraction (Skaife et al. 2013). Recent literature
suggests that the remediation of MWs not only improves financial reporting quality (AshbaughSkaife et al. 2008), but also increases firm performance by improving inventory management
(Feng et al. 2013) and investment efficiency (Cheng et al. 2013). By examining tax avoidance
following remediation of tax-related MWs, I provide evidence on how improvements in internal
control under SOX can improve firm performance via increased tax avoidance.
This study also contributes to our understanding of the magnitude and timing of tax
avoidance payoffs on tax function investments. By documenting an economically significant

6
 

future payoff on investments in the income tax function, I provide insights for managers and
practitioners regarding tax function investment decisions. Omer et al. (2006) find that the
positive association between investments in auditor-provided tax services and tax avoidance is
mitigated following the passage of SOX. In contrast, I document significant future tax avoidance
payoffs on tax function investments during the post-SOX era illustrating that certain firms are
able to realize tax avoidance payoffs on their investments during this time period.
The remainder of the study is organized as follows. Chapter 2 provides a discussion of
related literature and develops the hypotheses. Chapter 3 discusses the research design and
sample selection. Chapter 4 presents the empirical results of the main analysis. Chapter 5
provides the results of additional analyses and sensitivity tests to provide further perspective
about the main findings reported in Chapter 4. Chapter 6 provides concluding remarks about the
dissertation.

7
 

CHAPTER 2: LITERATURE REVIEW AND HYPOTHESES

I begin this chapter by discussing the literature that explores the consequences of the
disclosure and remediation of internal control weaknesses. I follow this by examining the
literature on tax avoidance determinants. I then discuss the prior literature on the impact of
investments in the tax function on tax avoidance. Finally, I review the literature on the
determinants of outsourcing and how these studies apply to the decision regarding the
outsourcing of tax function investments.

2.1. Disclosure and Remediation of Internal Control Weaknesses
Following the corporate accounting failures of Enron and WorldCom, Congress enacted
SOX in 2002 with the goal of protecting investors through multiple internal control and financial
reporting requirements, as well as establishing the Public Company Accounting Oversight Board
(PCAOB) (SOX: Section 101). SOX requires firms’ management (Section 404(a)) and auditors
(Section 404(b)) subject to its provisions to annually test internal controls over financial
reporting, report on control effectiveness, and disclose any unremediated MW at year-end in the
annual financial report (10-K).6
An extensive literature examines the consequences of the disclosure of internal control
weaknesses under SOX. Beneish et al. (2008) and Hammersley et al. (2008) document a negative
market reaction to the initial disclosure of internal control weaknesses (ICWs) under Section

                                                            
6

Section 302 of SOX requires quarterly reporting by management of accelerated filers on the effectiveness of
financial reporting controls. In this study I focus on only Section 404(b) disclosed MWs and their remediation
because the dependent variable of interest is tax avoidance measured on an annual basis due to quarterly data
limitations.

8
 

302, suggesting these disclosures contain value-relevant information.7 ICWs also are associated
with lower accruals quality (Ashbaugh-Skaife et al. 2008; Doyle et al. 2007b), restatements
(Ashbaugh-Skaife et al. 2007), less accurate management forecasts (Feng et al. 2009) and
earnings management (Chan et al. 2008; Gleason et al. 2013). Additional consequences include
increased audit fees (Hogan and Wilkins 2008; Hoag and Hollingsworth 2011); auditor changes
(Ettredge et al. 2011); financial reporting lags (Ettredge et al. 2006), credit rating downgrades
(Dhaliwal et al. 2011) and an increased cost of equity (Ashbaugh-Skaife et al. 2009). Several
shareholder lawsuits also have been triggered by the disclosure of an ICW.8 Overall, these
results suggest that the disclosure of an ICW has significant economic consequences to a firm.
Regulators claim that the presence of ICWs reduces investor confidence in reported
financial results and that the remediation of such weaknesses will lead to improved financial
reporting quality (Niemeier 2004; Countryman 2005; Pickard 2005). Hamerserley et al. (2012)
document that firms that do not remediate ICWs pay higher audit fees, have a greater likelihood
of an adverse audit opinion, experience filing delays, and incur higher interest rates, suggesting
there are economic consequences to not remediating ICWs. Additionally, evidence suggests the
remediation of ICWs, in general, is associated with improved accounting quality (AshbaughSkaife et al. 2008), a reduction in the cost of equity capital (Ashbaugh-Skaife et al. 2009), and
improved investment efficiency (Cheng et al. 2013). Taken together, the results on the

                                                            
7

Prior studies define ICWs as any MW disclosed under Section 404, Section 302 of SOX, significant deficiencies,
other control weaknesses or some subset of these categories. I use the term ICW to describe internal control issues
examined by these studies. I use the term MW only for annually disclosed MW under Section 404 of SOX.
8
See http://news.cnet.com/8301-1023_3-57411560-93/new-shareholder-lawsuit-targets-groupon-execs/ for a
discussion of Groupon Inc.'s disclosed ICW leading to a shareholder lawsuit. Also, see
http://www3.cfo.com/article.cfm/3396587/c_2984303/ for a more general discussion.

9
 

consequences of remediating versus not remediating ICWs suggest that firms have an economic
incentive to remediate.9
Literature examining the consequences of remediating specific types of ICWs is less
prevalent. With the exception of classifying ICWs based on severity and account-level versus
entity-level issues, this research tends to focus on all ICWs rather than on specific account types
of ICWs (Bedard et al. 2011).10 An exception is Feng et al. (2013), who find that the remediation
of inventory-related MWs is positively associated with improved gross margin, sales, and
inventory turnover. These results suggest that investments to remediate MWs go beyond
addressing financial reporting quality and have a spillover effect on firm performance.
Initial disclosures under SOX show tax-related issues are the most common source of
MWs.11 Although the number of disclosed MWs has recently declined, tax-related issues
continue to be the leading cause of MW disclosures (Deloitte 2011). Gleason et al. (2013) find
that the disclosure (remediation) of tax-related ICWs is positively (negatively) associated with
upward earnings management via the tax expense account. These results suggest that
investments to remediate tax-related MWs constrain management’s ability to use the tax expense
account to meet earnings targets.

                                                            
9
I acknowledge that a firm has an incentive to remediate ICWs only to the extent the benefits of remediating exceed
the costs of taking such actions. Literature on the decision to remediate ICWs suggests that firms with enhanced
corporate governance (Goh 2009; Johnstone et al. 2011), stronger CFOs (Li et al. 2010) or CFOs with public
accounting experience are more likely to remediate. In addition, firms are less likely to remediate when the
weakness disclosed is at the entity-level and when firm-level operations are more complex (Jonas 2005;
Hammersley et al. 2012).
10
Entity-level weaknesses are the most severe type of MW and relate to the overall control environment
(Hammersley et al. 2012). For a MW to be classified as an entity-level weakness it must include a weakness in one
or more of the five components of the COSO control framework. The five components are control environment, risk
assessment, control activities, information and communication, and monitoring (COSO 1994).
11
Approximately one third of the initial MW disclosures for 2004 listed tax as a contributing factor
http://www.cfo.com/article.cfm/5077959 . Also, see Ge and McVay (2005) for statistics on the number of initial
disclosed weaknesses that are tax-related.

10
 

Bedard and Graham (2011) use a proprietary dataset of internal control disclosures
(including significant deficiencies and MWs) and find that resource constraints are associated
with the disclosure of tax-related ICWs. Additionally, Bauer (2013) finds that the disclosure of
tax-related MWs can be predicted by lower levels of tax avoidance. This result is consistent with
firms that disclose a tax-related MW choosing not to invest significant resources into their
income tax function. I build on this prior literature by examining the association between the
remediation of tax-related MWs and both current and future tax avoidance. Additionally, I
examine how firms increase future tax avoidance and whether payoffs (in terms of increases in
future tax avoidance) on external investments are greater than those on internal investments.

2.2. Tax Avoidance Determinants
Tax avoidance can be defined as the reduction of all explicit taxes (Hanlon and Heitzman
2010). Tax avoidance is a broad term that captures activities that range from low-risk tax
planning such as investing in tax-free municipal bonds or depreciating assets, to more aggressive
activities such as tax sheltering or even tax evasion (Dyreng et al. 2008). Research in accounting
and economics has documented a large array of firm characteristics that are associated with
corporate tax avoidance (Gupta and Newberry 1997). For example, leverage, profitability, size,
capital expenditures, research and development expenditures and foreign operations are
documented determinants of tax avoidance (Gupta and Newberry 1997; Rego 2003).
Ownership structure is another important determinant of tax avoidance. Chen et al.
(2010) find that concentrated ownership in family firms leads to reduced tax avoidance
presumably to avoid negative reputation effects and suspicions of rent extraction via the tax
accounts. Khurana and Moser (2011) document a positive association between short term

11
 

institutional ownership and corporate tax avoidance. Finally, Cheng et al. (2012) find that an
increase in ownership by activist hedge funds is associated with increased tax avoidance.

2.3. Investments in the Tax Function
Prior literature examining the impact of investments in the tax function on tax avoidance
is sparse and typically focuses on tax planning investments. The most common proxy for tax
planning investments is the purchase of auditor-provided tax services (APTS). Using this proxy,
Omer et al. (2006) document an association between the level of APTS in year t and reductions
in ETRs in year t+1 prior to SOX. Additionally, APTS are correlated with decreases in ETRs
from the third to fourth quarter (Cook et al. 2008).12 Finally, Donohoe and Knechel (2013) find
that prior to FIN 48, the provision of APTS reduces the documented audit fee premium for tax
aggressive firms.
One exception to using APTS as a proxy for tax planning investments is Mills et al.
(1998). Based on confidential survey data on the level of tax department expenditures during
1991 for a sample of large firms, the authors find a significant positive association between these
expenditures and the level of tax avoidance measured as the 3-year GAAP ETR for the years
1990-1992. These results suggest that for every $1 spent on tax planning (internal and external
payments combined) a firm can expect a $4 reduction in federal tax liabilities.13 Importantly,
due to the single year of data, the authors are unable to completely rule out potential correlated
omitted variables or generalize their results to alternative time periods and tax regimes.

                                                            
12

In a multivariate framework, auditor-provided tax services are only associated with third to fourth quarter
decreases in ETRs when interacted with an indicator variable for whether the firm would have met the consensus
forecasts absent the decrease in ETR. This is consistent with earnings management incentives playing a role.
13
Using the same data set Gupta and Mills (2002) document a $160 reduction in state tax liabilities for each
additional dollar spent on tax planning.

12
 

Additionally, Mills et al. (1998) cannot examine the timing of returns on these expenditures,
because they lack a time series of tax planning expenditures and thus cannot identify changes in
such expenditures.
Another exception to using the APTS for an investment in the tax function is Cheng et al.
(2012). In this study the authors examine the impact of hedge fund activism on corporate tax
avoidance. After controlling for alternative explanations via a matched sample design, they find
that firms targeted by hedge fund activists have lower levels of tax avoidance prior to being
targeted and subsequently increase tax avoidance in the year of a hedge fund activist event.
Additionally, results are stronger for hedge funds that have a history of implementing tax related
changes at firms they have targeted in the past. Although the authors do not discuss how the
change in ownership structure during one of these events impacts tax function investments, it is
reasonable to assume that these firms increase their tax function investment in response to these
changes.14 This leads to the conclusion that increases in tax function investments due to hedge
fund activism are positively associated with contemporaneous tax avoidance and additional
evidence suggests this effect persists into future tax avoidance activities.15
In contrast to Cheng et al. (2012), I operationalize increases in tax function investments
by examining the response to a regulatory event rather than a change in ownership structure.
Specifically, I use the remediation of a disclosed tax-related MW under Section 404(b) of SOX
as a setting to observe increases in tax function investments.16 Internal control is defined as a
                                                            
14

Hedge fund activist events are most likely endogenous to other firm characteristics and changes in those
characteristics make it difficult to rule out alternative explanations and establish temporal precedence which reduces
the ability to make causal inferences (Campbell and Cook 1979).
15
A potential explanation for the documented contemporaneous association is that hedge fund activists demand
immediate improvement in tax avoidance activities and therefore, tax function investments are directly targeted at
improving tax planning. In contrast, the setting used in this study examines an investment in the tax function to
address financial reporting issues and hypothesizes a spillover effect onto tax avoidance as further described below.
16
Although I cannot distinguish between management discovered versus auditor discovered MWs, Bedard and
Graham (2011) find that over 80% of tax-related ICWs were auditor identified.

13
 

process that is designed to provide assurance regarding the achievement of 1) effectiveness and
efficiency of operations; 2) reliability of financial reporting; and 3) compliance with applicable
laws (COSO 2011). A primary measure of a corporate tax function’s effectiveness is the level of
tax avoidance achieved (Armstrong et al. 2012). Prior to SOX, many firms engaged audit and tax
personnel from their independent audit firm to assist in the preparation of the tax provision for
financial reporting purposes.17 However, following the passage of SOX, auditor independence
rules now make it much more difficult to engage the independent auditor for these types of
services.18 Additionally, in a recent survey, over 50 percent of tax executives list financial
reporting risks as an important factor in tax planning implementation decisions (Graham et al.
2013). Firms disclosing a tax-related MW therefore likely do not have sufficient internal control
systems in place to provide assurance and aid in complying with applicable financial reporting
standards, tax regulations, and achieving the effectiveness objective from a tax avoidance
perspective.
Bauer (2013) finds the disclosure of a MW is associated with a lack of managerial control
and resource constraints consistent with potential underinvestment. The disclosure of a taxrelated MW under SOX is a shock that incentives a firm to make an investment in its income tax
function if it chooses to remediate the deficiency. Given the substantial costs associated with the
disclosure and non-remediation of MWs demonstrated in the prior literature, there are significant
incentives to make these investments to address the internal control issue that caused the MW. I
argue that these investments have spillover effects that impact the resources devoted to and the
effectiveness of the firm’s tax function.

                                                            
17

http://www.nysscpa.org/printversions/cpaj/2006/706/p38.htm
Specifically, Section 201 precludes auditors from providing bookkeeping services once they have audited a public
business.

18

14
 

2.4. Determinants of Outsourcing
Coase (1937) conjectures that a firm is a nexus of contracts and argues that the reason
why firms exist is because the benefits to having these contracts organized within a firm rather
than just having contracts on an open market place exceed the marginal costs. Transaction cost
economics (TCE) theory suggests that firms will seek to minimize transaction costs and predicts
that as these costs increase a firm is more likely to vertically integrate (i.e., in-source) (Coase
1937; Williamson 1975). Transaction costs are defined as the costs of contracting and the
opportunity costs of inefficient contracts, as well as, any computational costs (Williamson 1975).
Specifically, TCE predicts as the asset specificity, uncertainty, and frequency of transactions
increases firms will be less likely to contract with an external party and more likely to develop
the good in-house.
Another theory of outsourcing is the resource-based view of the firm (RBV). RBV views
a firm as a collection of resources and capabilities (Grant 1991; Espino-Robriguez and PadronRobaina 2006). RBV takes a strategic prospective and examines how the decision to procure
something in-house is related to a firm’s competitive advantage. By devoting scarce resources to
an activity that is not part of a firm’s core strategy a firm reduces the resources that are allocated
to core business units (Quelin and Duhamel 2003). Thus, there is an incentive to out-source
activities that are not a part of the firm’s core business strategy. When a firm faces significant
resource constraints it is even more likely to outsource these support-type activities (EspinoRodriguez and Padron-Robaina 2006).
Empirical research examining the decision to outsource accounting services is limited
and uses TCE theory to focus on the decision to outsource internal auditing activities. Widener
and Selto (1999) use survey data to document a negatively association between asset specificity

15
 

and frequency of transactions with outsourcing internal audit activities. Caplan and
Kirschenheiter (2000) find that the incentive to outsource internal audit activities increases in
various measures of risk including the risk that an internal control weakness exists.

2.5. Hypothesis Development
2.5.1. Tax Function Investments and Tax Avoidance
A firm’s tax strategy is a form of investment (McGuire et al. 2011). To compute the net
present value of a tax project investment it is important to determine the magnitude and timing of
payoffs on capital outlays (Brealey and Myers 2002). As expected payoffs take longer to realize,
the uncertainty surrounding those payoffs also increases (Busby and Pitts 1997). Thus,
determining if tax function investments have a substantial and immediate impact on tax
avoidance is an important empirical question.
Prior literature suggests the remediation of MWs has an immediate impact on firm
performance. Feng et al. (2013) document that the remediation of inventory-related MWs has an
immediate impact on inventory management efficiency Additionally, Cheng et al. (2013) finds
that the remediation of MWs increases investment efficiency in the year of remediation, as well
as future years. Thus, prior research suggests that remediating MWs has an immediate impact on
firm performance.
Related to this study, there are several reasons why an investment in the tax function
would not be contemporaneously associated with higher levels of tax avoidance. First, if the
investment simply addresses internal control issues related to the income tax provision and has
no impact on tax function resources or effectiveness, there should not be an impact on the tax
avoidance activities of a firm at any point in time. Second, if the tax function investments enable

16
 

tax planning activities that can only be realized in future taxable years, there would be no
contemporaneous association. Third, the internal control disclosures used to identify tax function
investments are based on year-end assessments (e.g., as of December 31st for a calendar year
firm). Thus, the remediation of an MW can occur at any time during the year. If the investment is
not made until late in the reporting period, it may not have an immediate impact on tax
avoidance.
In contrast, tax function investments can affect future levels of tax avoidance for several
reasons. First, many tax avoidance activities involve significant uncertainty (Rego and Wilson
2012) and provide tax avoidance “annuities” that are recognized over future taxable years.
Several tax planning activities require significant changes to firm structure or operations and take
time to implement. For example, creating a foreign subsidiary in a favorable taxing jurisdiction
and transferring or co-developing intellectual property to or with that subsidiary takes time and
resources. Furthermore, benefits recognized from transfer pricing strategies that allow income to
be shifted to a low-taxed foreign subsidiary require future taxable income streams over multiple
periods.
Second, any effects on tax avoidance likely would be due to spillover effects.19 MWs are
indicative of poor internal controls which can reduce the amount and quality of information
available to managers (Feng et al. 2009). An important piece of information for tax planning is
an accurate and reliable estimate of future taxable income (Kubick et al. 2014). Consistent with
this conjecture, Mayberry et al. (2013) finds that firms with more persistent or smooth earnings
are better able to realize the benefits of tax planning. Therefore, one possible effect of a tax
                                                            
19

Prior literature documents spillover effects of investing in tax planning via auditor-provided tax services on
financial reporting quality (Larcker and Richardson 2004; De Simone, Ege, Stomberg 2013). However, to my
knowledge no prior study examines the timing of these spillover effects or how an investment to improve financial
reporting quality impacts tax avoidance.

17
 

function investment that addresses internal control issues is an improvement in the information
available to managers regarding the tax accounts. This information will then allow managers to
improve their forecasts of taxable income enabling increased tax planning opportunities.
The most common investments being made in my sample are engaging an external
advisor (58.1%) and hiring additional personnel (55.2%).20 A firm that engages an external
advisor would prefer an immediate payoff on its investment. However, some remediation
disclosures state that these advisors are engaged to remediate the internal control issue that led to
the MW. Thus, any spillover effects onto tax avoidance may occur predominantly in future
periods as the firm’s immediate attention, as well as that of the firm’s advisors, is on remediating
the MW.
A similar argument can be made for an investment in new personnel. In the short-run the
attention of new personnel likely is devoted to internal control issues as opposed to tax
avoidance. Additionally, human resource investments may not be immediately recognized due to
the time needed for new employees to familiarize themselves with the company’s operations,
procedures, and to receive training (Huselid 1995). Given the complex nature of tax planning
activities, new employees within the tax department are likely to require significant time to
assess a firm’s current tax situation before implementing new tax planning initiatives. Because
prior literature documents a positive association between tax planning investments and tax
avoidance (Mills et al. 1998; Omer et al. 2006), I expect that investments in the tax function will
increase the level of tax avoidance. However, the above arguments suggest a lag between the
initial investment and tax planning activities. Thus, I offer the following hypothesis in the
alternative form:

                                                            
20

See Table 3 for a break-down of the different types of investments.

18
 

H1: Increases in investments in the corporate tax function are positively associated
with future tax avoidance.

2.5.2. Internal Versus External Investments and Future Tax Avoidance
Mills et al. (1998) bifurcate tax planning expenditures into external and internal
components and provide some evidence that external tax planning expenditures result in larger
tax avoidance payoffs when compared to internal investments.21 This finding suggests that
different types of tax function investments could have different impacts on a firm’s level of tax
avoidance. This finding also is consistent with an external investment being more substantial
than an internal investment. In addition to potentially being more substantial, external
investments could result in larger increases in tax avoidance if the external advisors are engaged
to and evaluated on their ability to increase tax avoidance. If external advisors have an expertise
and knowledge advantage over potential additional internal personnel, I would expect a higher
payoff on external investments. These arguments lead to the prediction of external investments
resulting in higher future tax avoidance payoffs.
However, if the external advisors are engaged and evaluated only on their ability to
remediate the internal control issue, the payoff (if any) on external investments may not be as
high. Additionally, investing internally in new personnel and/or a new tax director could result in
larger payoffs to long-run tax avoidance if these types of investment allow for a greater
propensity to recognize benefits over multiple years as compared to a potentially short-term
payoff for an external investment. Finally, if each firm invests optimally, an external investment

                                                            
21

Specifically, external investments are associated with lower ETRs but internal investments are not. However,
when total assets are used as the ETR denominator rather than pre-tax income internal investments are associated
with increased tax avoidance (Mills et al. 1998).

19
 

could yield a similar payoff when compared to another firm that has certain characteristics that
make an internal investment the economically rational choice. Consistent with external
investments leading to higher tax avoidance payoffs I state H2 in the alternative form:

H2: Increases in external tax function investments are associated with higher levels of
future tax avoidance when compared to increases in internal tax function
investments.

2.5.3. Determinants of External Tax Function Investments
Tax provision preparation and tax planning are a service good that a firm must internally
invest in, outsource to an external party, or procure through a combination of both internal and
external investments. Firms face resource constraints when determining what activities to
perform internally (Laios and Moschuris 1999). Firms facing significant resource constraints
likely have fewer internal resources available to devote to support activities. RBV predicts that
firms will be more likely to outsource non-core competency activities and that this effect will be
intensified in the presence of greater resource constraints (Espino-Rodriguez and PadronRobaina 2006).
Additionally, a firm facing resource constraints (i.e., a firm in a loss position) could
potentially need to reassure investors that significant resources are being devoted to addressing
the internal control issue. The engagement of an external service provider could provide a signal
to investors that the company is taking the issue seriously, thereby increasing the probability that
it will be remediated quickly. Given these arguments, I offer the following hypothesis in the
alternative form:

20
 

H3a: Resource constrained firms are more likely to outsource investments in their
income tax function.

Tax planning and preparation of the tax provision can be complex and risky processes
(Hanlon and Heitzman 2010). The supply chain management and RBV literature suggests that as
the complexity of a support task increases, a firm is more likely to outsource that activity (Laios
and Moschuris 1999; Espino-Rodriguez and Padron-Robaina 2006). The tax function is a support
activity, suggesting that as the complexity of a firm’s tax function increases a firm is more likely
to outsource investments. Additionally, as complexity and risk increase, a firm could be forced to
provide a signal to shareholders that they are engaging external help. The disclosure of a taxrelated MW is an admission that the firm does not have adequate tax accounting expertise and
thus disclosing an internal investment may not provide as much confidence to investors when
compared to disclosing an external investment.
TCE theory suggests that as the contracting costs increase a firm is more likely to
vertically integrate (i.e., invest internally) (Williamson 1975; Tadelis 2002; Barthelemy and
Quelin 2006). TCE theory conjectures that contracting costs increase as uncertainty regarding the
future/external party’s actions and the specificity of assets used in production increases
(Williamson 1975). Consistent with this theory, the internal audit literature finds a negative
relation between outsourcing the internal audit function and firm complexity (Widener Selto
1999).
Applied to the tax function, TCE predicts that as the complexity of a firm’s tax function
increases so does the uncertainty/risk and the specificity of assets required to perform that

21
 

function; leading to a greater propensity to invest internally. In contrast, in this setting it could be
relatively easy to observe the external party’s action, thereby reducing contracting costs and
allowing a firm to use high powered incentives. Specifically, a firm could write a contract with
an external service provider that only remits payment on remediation of the disclosed MW.
Consistent with this argument, Caplan and Kirschenheiter (2000) find that the risk of the
presence of an internal control weakness leads to more internal audit outsourcing suggesting the
specific type of risk examined in my study leads to an increase in out-sourcing. Given the
competing theories I offer the following hypothesis in the null form.

H3b: The tax complexity and risk of a firm is not associated with the decision to
outsource investments in its income tax function.

22
 

CHAPTER 3: RESEARCH DESIGN AND SAMPLE SELECTION

Chapter 3 discusses the research design and sample selection. I begin by presenting
details on the tax avoidance measures, the tax function investment proxy, and control measures
followed by a discussion of the empirical model used to test H1, which predicts a positive
association between tax function investments and future tax avoidance. I then discuss the model
used to examine H2, which investigates tax avoidance payoffs on external versus internal
investments. Finally, I discuss the model for the determinants of external investment. I follow
with a description of the sample and data.

3.1. Measures
3.1.1. Measure of Tax Avoidance (Dependent Variable)
Hanlon and Heitzman (2010) describe and discuss various measures of tax avoidance and
conclude that many of the measures capture different forms of tax planning activities. In this
study, I do not distinguish between the different types of tax planning activities. Rather, the tax
avoidance proxies I employ capture tax planning activities that vary across the aggressiveness
spectrum.
Two commonly used measures of tax avoidance are the GAAP and cash ETRs. GAAP
ETRs are computed using total reported income tax expense and reflect tax avoidance activities
and tax accruals that generate permanent book tax differences (BTDs), thereby affecting reported
accounting earnings. However, the GAAP ETR does not reflect deferral strategies that generate
temporary BTDs. In contrast, the cash ETR measures taxes paid to the revenue authorities during

23
 

that period and captures tax planning activities that generate most permanent and temporary
BTDs.22
Private conversations with several Big Four professionals and a review of the practitioner
literature reveals that firms benchmark their effective tax rate measures against industry
competitors to infer information regarding tax function performance.23 Additionally,
opportunities for tax planning activities differ across industries. For example, an international
intellectual property-intensive software company has access to transfer pricing tax planning
strategies that a domestic capital-intensive manufacturing company does not. Using ETR
measures without adjusting for these factors could lead to incorrect inferences.
To address this concern, I follow Balakrishnan et al. (2012) and adapt the GAAP and
cash ETR measures by computing and subtracting the mean industry-year ETR for a firm’s given
asset quintile. This measure adjusts for opportunities to engage in tax planning activities afforded
to certain industries and implicitly controls for macro-economic shocks (e.g., 2007-2010
economic recession) that could impact tax avoidance activities. Additionally, by comparing each
firm’s deviation from a relevant benchmark I am able to focus on the tax avoidance measure to
which firms most likely are devoting resources.24

3.1.2. Measure of Tax Function Investment
Following Ashbaugh-Skaife et al. (2008) and Bauer (2013), I include indicator variables
for the presence of a tax-related (non-tax related) MW in the current or prior year, WEAK_TAX
                                                            
22
The cash ETR will usually not reflect changes in the tax accrual such as the valuation allowance account or the
unrecognized tax benefits account (UTB). One exception is the settlement of a tax position that results in a cash
payment/refund and an adjustment in the portion of UTB that impacts earnings.
23
http://www.pwc.com/us/en/industrial-products/publications/tax-rate-benchmarking-study.jhtml
24
I confirm results are robust to using traditional mean unadjusted tax avoidance measures in the supplemental
analyses section.

24
 

(WEAK_OTHER).25 My primary variable of interest, TAX_FIXED, is equal to one if a firm
remediates a previously disclosed tax-related MW in the current year. By construction,
TAX_FIXED is essentially an interaction term between WEAK_TAX and another indicator equal
to one for firms that remediate tax-related MW in the current year. Thus, WEAK_TAX captures
the tax avoidance of firms that disclose a tax-related MW but fail to remediate. I expect a
positive coefficient on WEAK_TAX consistent with the disclosure of a tax-related MW being
associated with lower levels of tax avoidance (Bauer 2013). The coefficient on TAX_FIXED
captures the incremental tax avoidance of a firm that invests in its tax function to remediate a
tax-related MW relative to a firm that does not remediate (WEAK_TAX).
A negative coefficient on TAX_FIXED when CurrentCash1yrETR or
CurrentGAAP1yrETR are used as the dependent variable would suggest investments in the tax
function are contemporaneously associated with greater levels of tax avoidance. A negative
coefficient on TAX_FIXED when FutureCash3yrETRi,t+1-t+3, or FutureGAAP3yrETRi,t+1-t+3 are
employed would provide evidence consistent with investments in the tax function being
associated with greater levels of future tax avoidance Finally, I include OTHER_FIXED, which
is equal to one if a firm remediates a previously disclosed non-tax MW.

3.1.3. Control Variables for Tax Avoidance
Additional previously documented determinants of tax avoidance are included as control
variables and are defined in Appendix B. Pre-tax return on assets (ROA) is included to control
for the positive association between tax avoidance and profitability (Gupta and Newberry 1997;
Rego 2003). I control for leverage (LEV), foreign income (FORINC), and include a tax haven
                                                            
25

WEAK_TAX and WEAK_OTHER are mutually exclusive so a firm that discloses a tax-related MW will be coded
one for WEAK_TAX and zero for WEAK_OTHER. Inferences are unchanged if this restriction is not imposed.

25
 

indicator variable set to one for firms that disclose a subsidiary in a known tax haven
(TAXHAVEN) (Dyreng and Lindsey 2009). I also include controls for intangibles (INTAN), firm
size (SIZE), the presence of a net operating loss (NOL), research and development intensity
(RD), and lagged book-to-market ratio (BM).
Prior literature finds that resource constraints are positively associated with the disclosure
of ICWs (Krishnan 2005; Doyle et al. 2007b). Bauer (2013) shows that the negative association
between the disclosure of a tax-related MW and the level of tax avoidance is stronger for
resource constrained firms. Thus, to control for resource constraints, I include AGGR_LOSS,
which is equal to one for firms with aggregate earnings before extraordinary items for the current
and prior year less than zero, and zero otherwise. Additionally, I include CONSTR_CF, which is
equal to one minus operating cash flows.26 I also include industry (two-digit NAICS level) and
year fixed effects and use robust standard errors clustered at the firm-level in all model
specifications. Industry fixed effects control for time-invariant industry characteristics, which
could impact tax planning opportunities. Year fixed effects control for macro-economic shocks
and potential trends in tax avoidance levels. Finally, by clustering standard errors at the firm
level I address serial correlation concerns which could impact the estimation of standard errors
and increase the risk of incorrectly detecting statistically significant associations.

3.2. Research Design
3.2.1 Tax Function Investments and Tax Avoidance
To examine the impact of tax function investments on tax avoidance, I adapt a recently
developed measure of a firm’s ETR relative to those of its industry peers as my tax avoidance

                                                            
26

Inferences remain unchanged if CONSTR_CF is scaled by total sales or total assets.

26
 

proxy. Following Balakrishnan et al. (2012) I rank all firms by asset quintiles for each year and
Fama French industry classification (48 industries). I then compute the mean effective tax rate
measure for each industry-year asset quintile and subtract that value from each firm’s effective
tax rate measure.
To examine the contemporaneous association between increases in tax function
investments and tax avoidance, I employ the 1-year cash and GAAP relative ETR measures,
CurrentCash1yrETR and CurrentGAAP1yrETR, respectively. To investigate the impact of tax
function investments on future tax avoidance, I use FutureCash3yrETRi,t+1-t+3,
(FutureGAAP3yrETRi,t+1-t+3), which are the 3 year cash (GAAP) relative ETRs, respectively. I
compute these two measures by summing cash taxes paid (total tax expense) over years t+1, t+2,
and t+3 following the tax function investment and dividing that total by the sum of pretax
income less special items over that same period. I then subtract the mean industry-year sizeadjusted value as described above. Lower values of these tax avoidance measures can be viewed
as a firm disclosing ETRs below their industry peers, consistent with greater tax avoidance.
Following prior research, all of the ETR measures are winsorized between the [0,1] interval
(Dyreng et al. 2008). I estimate the following OLS regression model where TAXi, is defined as
one of the four tax avoidance proxies.
TAXi, =

β0 + β1TAX_FIXEDi,t + β2WEAK_TAXi,t + β3 WEAK_OTHERi,t +
β4OTHER_FIXEDi,t + β5ROAi,t + β6LEVi,t + β7FORINCi,t +
β8TAXHAVENi,t + β9INTANi,t + β10SIZEi,t + β11NOLi,t + β12RDi,t +
β13BMi,t-1 + β14AGGR_LOSSi,t + β15CONSTR_CFi,t +
industry fixed effects + year fixed effects + εi,t
(1)

27
 

3.2.2 Tax Avoidance Payoffs on External Versus Internal Investments
To examine H2 and investigate whether external investments yield higher tax avoidance
payoffs when compared to internal tax function investments, I modify equation (1) bifurcating
the variable TAX_FIXED into firms that disclose an external investment (EXTERNAL_FIX:
n=61) versus firms that only invest internally (INTERNAL_FIX: n=54). A negative and
significant coefficient on these variables suggests that the type of tax function investment being
examined is positively associated with future tax avoidance. An F-test of the null hypothesis that
β2 > β3 provides evidence on whether external tax function investments lead to higher tax
avoidance payoffs when compared to internal investments. Finally, I also focus on the impact of
tax function investments on future tax avoidance given the hypothesized relationship.
FutureTAX3yri,t+1-t+3 is defined as either FutureCash3yrETRi,t+1-t+3, or
FutureGAAP3yrETRi,t+1t+3, respectively.

FutureTAX3yri,t+1-t+3 =

β0 + β1WEAK_TAXi,t + β2EXTERNAL_FIXi,t + β3INTERNAL_FIXi,t +
β4WEAK_OTHERi,t + β5OTHER_FIXEDi,t + β6ROAi,t + β7LEVi,t +
β8FORINCi,t + β9TAXHAVENi,t + β10INTANi,t + β11SIZEi,t + β12NOLi,t+
β13RDi,t + β14BMi,t-1 + β15AGGR_LOSSi,t + β16CONSTR_CFi,t +
industry fixed effects + year fixed effects + εi,t
(2)

28
 

3.2.3 Determinants of External Tax Function Investments
To investigate the determinants of making an external investment I estimate the following
logistic regression model for the primary sample of firms identified as making a tax function
investment (105 firm-years).
Externali,t =

β0 + β1RESTATEMENTi,t + β2FOROPSi,t + β3ROAi,t + β4SIZEi,t +
β5LOSSi,t + β6LEVi,t + β7RDi,t + β8BMi,t-1 + β9INVENTORYi,t +
year fixed effects + εi,t
(3)

External is equal to 1 if a firm discloses the investment in an external service provider
related to the remediation of the tax-related MW and zero otherwise. As detailed in Table 3, 61
firms of the 105 observations disclose engaging an external service provider.
To examine the effects of resource constraints on investing in an external service
provider I include ROA, SIZE, and LOSS. With the exception of LOSS, that is equal to 1 for firms
reporting negative income before extraordinary items in the current year, zero otherwise, all
variables are as previously defined. A negative and significant coefficient on ROA and SIZE
would be consistent with less profitable and smaller firms that most likely have fewer resources,
choosing to invest externally. A positive and significant coefficient on LOSS would suggest firms
with negative income choose to invest externally.
To examine the impact of risk and complexity on the decision to engage an external
provider, I include two variables, RESTATEMENT and FOROPS. RESTATEMENT is an
indicator variable equal to 1 for tax-related MWs that led to a restatement, and zero otherwise.
This variable captures the severity of the disclosed accounting issue. A positive coefficient

29
 

would be consistent with risk and complexity being positively associated with the decision to
invest in an external service provider. FOROPS is an indicator variable equal to 1 for all firms
with non-zero values of pre-tax foreign income, and zero otherwise. Firms with foreign
operations have more complex tax issues to consider due to multiple jurisdictions and transfer
pricing rules. Thus, a positive coefficient on FOROPS would provide evidence of firms with
more complex tax situations being more likely to engage an external service provider.
I include leverage (LEV), research and development intensity (RD), lagged to book to
market (BM) and inventory (INVENTORY) as other control variables27 Mills et al. (1998) find
the level of internal investments in tax planning to be negatively correlated with the level of
inventory; thus I control for this potential explanatory factor.

3.3. Sample Selection and Data
Table 1 provides summary statistics on the typical time to remediate for all MWs
disclosed in Audit Analytics from 2004 to 2011 bifurcated between the tax and other typologies.
Panel A provides a break-down by year and the number of years to remediate. Although the
frequency of all other MWs combined is larger than tax-related MWs, tax-related MWs still are
the leading cause of MWs. Examining the Table it appears the majority of the timing of
remediation is similar across the different types. Panel B summarizes the number of MWs by
year during the sample period employed in this study (2004-2009). The majority of both tax and
other MWs are remediated in the year following disclosure with other MWs having a slightly
greater probability of being remediated within the 1 year post-disclosure window.

                                                            
27

Inventory is equal to total inventories scaled by total assets.

30
 

Table 2 details sample selection procedures for the observations used in the primary tests
of the effects of tax function investments on future tax avoidance. To operationalize tax function
investments I use Audit Analytics to identify firms that disclose a new tax-related MW during the
2004-2008 time period.28 I then manually examine each firm’s 10-K to confirm the presence of
the tax-related MW and determine if the tax-related MW was remediated. This results in a
sample of 303 disclosed tax-related MWs, before removing observations that have missing
required Compustat and Audit Analytics data (128 firms) or have a negative sum of pretax
income for years t+1 through t+3 following remediation (70 firms). My final sample of tax
function investments is 105 firm-years.
For this sample of firms I then classify the disclosed type of investment made to
remediate the MW into the following categories 1) Hiring additional personnel (non-tax
director); 2) Hiring a tax director; 3) Engaging an external provider; and 4) Only investing in
new procedures. Appendix A provides examples of the different types of investments. In
example 1, H&R Block Inc. discloses a tax-related MW related to “non-routine and complex”
tax transactions and ETR/deferred tax reconciliations and balances. The firm goes on to describe
remediation steps including hiring additional personnel, changing procedures, and engaging an
external service provider. Example 2 describes Symantec Corp.’s tax-related MW related to an
acquisition and the subsequent missed election and extension filing due date. The firm
subsequently hired a new tax director. Example 3 details Netflix Holdings Inc.’s MW disclosure,
which discusses a lack of management review. The only remediation step discussed and
implemented was a change in internal procedures. Finally, example 4 is PAR Pharmaceuticals
Companies, which disclosed a tax-related MW that led to a restatement. Remediation procedures
                                                            
28

I limit my sample to pre-2009 observations to allow for the computation of future effective tax rates following
remediation from t+1 to t+3 (i.e., 2010-2012 for a 2008 disclosed tax-related MW that is remediated in 2009).

31
 

included hiring additional personnel, changing procedures, and engaging an external service
provider.
These examples illustrate the diversity in the different types of investments being made in
the tax function. The aggregated results of my classifications are displayed in Table 3. In
addition, to the categories listed above, I include an additional designation for firms that only
made an internal tax function investment and did not disclose the involvement of any third
parties in remediating the MW. Panel A displays the results for all remediated tax MWs in Audit
Analytics. Panel B presents the results for the 105 firm sample used in the multivariate tests. The
results show that for the 105 firms, the most common forms of investment are engaging an
external party (58.1%) and hiring additional personnel (non-tax director) (55.2%). Just over 40%
percent of firms examined chose to only invest internally. Examining time trends in the type of
investment, reveals that firms initially relied more heavily on internal rather than external
investments (i.e., 2005 versus subsequent years).

32
 

CHAPTER 4: MAIN RESULTS

In Chapter 4, I begin by providing descriptive statistics on the firms that made tax
function investments compared to various control groups. I then discuss univariate results for
impact of tax function investments on tax avoidance. These results are followed by a detailed
discussion of the main empirical results testing H1, H2, H3a, and H3b, respectively.

4.1. Descriptive Statistics and Univariate Results
4.1.1. Descriptive Statistics
Table 4 Panel A displays descriptive statistics for the primary sample of firm-years used
to estimate equation (1). Column 1 describes the variables for firms that make a tax function
investment at some point during the sample period (i.e., TAX_FIXED = 1 in at least one year).
Column 2 offers descriptives of all other firm-years. Univariate tests of means and medians
suggest firms that make tax function investments have higher ETRs throughout the sample
period. Importantly, this test does not examine the effect of tax function investments on tax
avoidance as column 1 includes firm-years both before and after remediation. Tax function
investing firms have higher levels of foreign income and are more likely to have a tax haven
subsidiary. Finally, these firms also are less profitable, smaller, and more likely to be in a loss
position consistent with prior research on the disclosure of MWs (Doyle et al. 2007b). The
significant differences in the type of firms making tax function investments versus the control
firms make it important to include these variables as independent variables in my multivariate
models.

33
 

4.1.2. Univariate Results
Panel B of Table 4 displays descriptive statistics for all firm-years of the tax function
investment firms bifurcated in observations prior to the investment (column 3) and observations
in the year of and following the investment (column 4). Consistent with Bauer (2013), tax
function investment firms exhibit future 3-year cash (GAAP) ETRs measures that are 7.7
percentage points (6.4 percentage points) higher than their industry peers prior to the tax function
investment. In the year of and following a tax function investment these firms display future 3year cash (GAAP) ETRs measures that are 1.8 percentage points (2.9 percentage points) higher
than their industry peers. The 3-year cash (GAAP) ETRs measures are significantly lower in the
year of and following a tax function investment than firm-years prior to the investment (i.e.,
FutureCash3yrETRt+1-t+3 is 5.9% lower; p-value <0.01 and FutureGAAP3yrETRt+1-t+3 is 3.5%
lower; p-value = 0.084) supporting a positive association between tax function investments and
future tax avoidance.
Additionally, to address endogeneity concerns I use a matched sample design where I
match each tax function investment firm-year (i.e., 105 firm-years) to a control firm based on
two-digit NAICS industry code, year, asset size, and pre-tax income. Descriptive statistics for tax
function investment firm-years and the matched sample firm-years are displayed in Panel C of
Table 4. Univariate tests of differences suggest a relatively accurate match on firm
characteristics.29 Importantly there are no significant differences across the contemporaneous tax
avoidance measures suggesting the firms have a similar starting point in terms of tax avoidance

                                                            
29

Exceptions include firm-years that disclose a tax function investment are more likely to be in a net operating loss
position and have lower profitability (i.e., ROA median is significantly lower). I include the net operating loss
indicator variable and ROA in all regression specifications to control for this potential issue.

34
 

behavior prior to tax function investments.30 This helps alleviate concerns that the tax function
investment firms have systematically different tax avoidance behavior prior to the investment
when compared to the control firms, which could impact the ability to change tax avoidance
behavior.
Given that I have matched the tax function investment firms to firms with similar
characteristics including tax avoidance behavior prior to the investment, I now examine the
impact of tax function investments on tax avoidance. Univariate results that examine the relation
between tax function investments and tax avoidance for the tax function investment firm-years
and the matched sample firm-years are also displayed in Panel C of Table 4. Specifically, when
examining the contemporaneous association the mean of CurrentCash1yrETRt
(CurrentGAAP1yrETRt) in tax function investment firm-years are 3.1% (4.9%), respectively.
However, comparing these observations to the matched firm-years does not yield statistically
different mean cash (0.004%; p-value = 0.47) or GAAP (6.9%; p-value = 0.55) relative ETRs,
respectively. These univariate results suggest that an investment in the tax function is not
contemporaneously associated with tax avoidance.
Examining the means of FutureCash3yrETRt+1-t+3 and FutureGAAP3yrETRt+1-t+3 of tax
function investment firm-years and the matched sample firm-years allows for a univariate test of
the impact of increases in tax function investments in year t on future tax avoidance levels,
measured as 3-year relative ETRs from years t+1 to t+3. Firms making a tax function investment
have future 3-year cash (GAAP) ETRs that are 0.008 percentage points (2.1 percentage points)

                                                            
30

In untabulated analyses I confirm that the 3-year tax avoidance measures measured from t-3 to t-1 are not
significantly different.

35
 

higher than their industry peers.31 The matched sample firm-years exhibit future 3-year cash
relative ETRs that are significantly higher (8.2%; p-value <0.01), while the future 3-year relative
GAAP ETR measures do not (5.0%; p-value = 0.36). These results combined with those
presented in Panel B provide some evidence that tax function investments in year t are associated
with higher levels of tax avoidance in years t+1 to t+3.

4.2. Multivariate Results
4.2.1. Tax Function Investments and Tax Avoidance
Table 5 displays results from estimating model (1), which examines the contemporaneous
association between tax function investments and 1-year relative ETRs (columns 1 & 2) and the
association between increases in current period tax function investments and future 3-year
relative ETRs (columns 3 through 6). Consistent with prior research, the coefficient on
WEAK_TAXt is positive and significant at the 5% level (two-tailed) in all specifications
suggesting that lower levels of tax avoidance are associated with the disclosure of a tax-related
MW (Bauer 2013). Specifically, a firm that discloses and does not remediate a tax-related MW
in year t has, on average, 6.3 percentage point (6.2 percentage point) higher 3-year future cash
(GAAP) ETRs, when compared to its industry peers (p-values < 0.01).
In the contemporaneous model, the coefficient on the primary variable of interest,
TAX_FIXED, is not statistically significant (column 1 coefficient = 0.005, p-value =0.88; column
2 coefficient = -0.012, p-value = 0.64). This result is consistent with tax function investments at

                                                            
31

A t-test of whether the FutureCash3yrETRt+1-t+3 and FutureGAAP3yrETRt+1-t+3 are significantly different from
zero in the year of the tax function investment fail to reject the null (p-value>0.30). This suggests tax function
investment firms exhibit future tax avoidance equal to the mean level of their competitors in years following a tax
function investment.

36
 

time t not leading to increased or decreased tax avoidance during that same time period.32
However, the coefficient on OTHER_FIXEDt is negative and significant when the cash ETR is
employed as the dependent variable, suggesting the remediation of non-tax-related MW may
have spillover effects on tax avoidance in the current period.33
In the future tax avoidance specification, the primary variable of interest, TAX_FIXED, is
negative and significant across all four specifications (Columns 3 through 6). The coefficient
estimate when the future 3-year relative ETRs are employed shows that firms that make a tax
function investment in year t have 8.1 percentage point (4.3 percentage point) lower cash
(GAAP) relative ETRs measured across t+1 to t+3 (FutureCash3yrETR: p-value < 0.01;
FutureGAAP3yrETR: p-value < 0.10). These results support H1 that current period increases in
tax function investments are positively associated with future tax avoidance. To assess economic
significance I re-estimate model (1) using unadjusted 3-year tax avoidance measures. The
coefficient on TAX_FIXED continues to be negative and significant (Cash3yrETRUnadjusted: 0.072, p-value <0.001; GAAP3yrETRUnadjusted: -0.040, p-value < 0.10). Interpreted at the
mean pre-tax income level, these coefficient estimates imply a reduction of $104.2 ($57.9)
million of cash taxes paid (total tax expense) over the 3-year period for the mean firm.34
To address concerns regarding the relatively small proportion of observations making tax
function investments (i.e., 105 firm-years of 7,687 observations), I re-estimate model (1) using a
matched sample design employing all firm-years of the firms obtained through the matching

                                                            
32

These results should be interpreted with caution as one year ETRs are noisy proxies for tax avoidance (Dyreng et
al. 2008).
33
Another potential explanation is that firms remediating other types of MWs could be reducing current or prior
earnings due to a restatement that leads to an amended tax filing and subsequent tax refund.
34
Univariate tests of changes in unadjusted 3-year tax avoidance measures following a tax function investment
imply reductions of $57.9 million in cash taxes paid and $29.0 in total tax expense, respectively, for the mean firm.

37
 

procedure.35 Additionally, as the matching criteria for the sample include industry, year, size,
and profitability, I do not adjust the tax avoidance measures using the mean peer group
measure.36 Results are displayed in Columns 5 and 6 of Table 5. The coefficient on TAX_FIXED
confirms the prior results for both the FutureCash3yrETR (Column 5 coefficient = -0.075, pvalue < 0.01) and FutureGAAP3yrETR (Column 6 coefficient = -0.044; p-value <0.10),
respectively.

4.2.2. How Firms Increase Future Tax Avoidance
Given the observed association between tax function investments and future tax
avoidance, it begs the question of what types of tax planning strategies firms are engaging in to
increase future tax avoidance. To address this issue, I hand-collect MD&A and tax footnote
disclosures for tax function investment firms in the largest quintile of decreases in relative cash
ETRs in years t+1 to t+3 (24 firm-years). I then categorize these disclosures based on the factors
for why the ETR decreased.37 Examples 5 and 6 in Appendix A illustrate the analyzed
disclosures.
In example 5, Shaw Group Inc. discloses a tax-related MW for the year ended 8/31/2008
due to insufficient numbers of tax professionals with income tax accounting expertise.
Remediation efforts included hiring a tax director and engaging external tax resources. The
discussion of the ETR in the 2010 10-K shows a dramatic decrease in the amount of tax expense
from 2008 to 2009 and 2010. The cited reasons for this decrease in income tax expense are due
                                                            
35
Inferences are qualitatively similar if the sample is limited to 210 firm-years rather than all firm-year observations
from the matched sample firms.
36
Inferences are qualitatively similar if relative ETR measures are employed.
37
The disclosed factors usually relate to GAAP ETRs, but I choose to examine the largest cash ETR changes as this
is the most robust result. Additionally, the factors are not mutually exclusive. For example, many firms cite both
foreign and state tax planning as reasons why the ETR decreased.

38
 

to a beneficial earnings mix between domestic and foreign operations (i.e., larger amounts of
income in low-tax rate countries).38 Additional disclosed factors include beneficial adjustments
to the tax contingency account and R&D tax credits. This observation would be coded as having
three disclosed reasons in my classification schema, 1) foreign taxes; 2) unrecognized tax
benefits account; and 3) other- R&D.
In example 6, True Religion Apparel Inc. discloses a tax-related MW in its 2007 10-K
related to the federal and state income tax consequences of their executive compensation plans.
Remediation efforts included hiring new internal and external personnel, as well as, changing
procedures. The 2009 10-K disclosures indicate substantial decreases in the reported ETR citing
changes in filing status in certain states. This observation is coded as a state tax expense
disclosure in my classification.
The results of these classifications are presented in Table 6 and suggest, for the firms
with the largest increases in future tax avoidance following a tax function investment, that the
most commonly disclosed factor for the decrease in the ETR is a reduction in foreign tax expense
(58.3%). State tax planning is also frequently disclosed as a factor (29.2%). The valuation
allowance account is mentioned 20.8% of the time and the unrecognized tax benefits account is
disclosed as a factor in 16.7% of the observations. Other issues such as research and
development tax credits, the domestic producer’s deduction (Section 199), tax-favored mergers
and acquisitions, etc. are mentioned in 29.2% of the observations. These results suggest firms are
undertaking potentially sophisticated tax planning strategies to increase future tax avoidance.

                                                            
38

Examining exhibit 21 of the 10-K it appears that Shaw Group Inc. did not set up any new foreign subsidiaries
during this time period. This suggests the decrease in foreign tax expense was due to more profits being allocated to
lower tax rate jurisdictions where subsidiaries existed prior to the tax-related MW.

39
 

4.2.3. External versus Internal Tax Function Investments and Future Tax Avoidance
The results from estimating model (2), which examines the impact of external versus
internal tax function investments on future tax avoidance, are presented in Table 7. The
coefficient on EXTERNAL_FIX in Column (1) results where FutureCash3yrETR is used as the
dependent variable suggest that firms making some type of external investment in year t have 7.9
percentage point lower future cash ETRs relative to their industry peers (p-value < 0.01). The
coefficient on INTERNAL_FIX suggests firms that choose to only internally invest in their tax
function have 8.5 percentage point lower future cash ETRs relative to their industry peers (pvalue < 0.05). An F-test of whether the coefficient estimate on EXTERNAL_FIX equals the
coefficient estimate on INTERNAL_FIX fails to reject the null, (p-value = 0.892).
Column (2) displays results for when FutureGAAP3yrETR is used as the dependent
variable. Consistent with the cash ETR results, firms making an external investment have 6.0
percentage point lower future GAAP ETRs relative to their industry peers (p-value < 0.05). The
coefficient estimate on internal tax function investments suggests internal investments result in
5.3 percentage point lower future GAAP ETRs relative to their industry peers, but is not
significant (p-value = 0.113). Although the external investment coefficient is higher, an F-test of
equality of the coefficients fails to reject the null that the external and internal investment
coefficients are equal (p-value = 0.848). These results are confirmed using the matched sample
design in Columns 3 and 4.39 In summary, I find evidence consistent with both external and
internal current period tax function investments being positively associated with future tax

                                                            
39

Results for the GAAP ETR are not statistically significant (Columns (2) and (4)). This could be attributed to the
limited sample size. An F-test of the equality of coefficients on external and internal tax function investments fails to
reject the null, providing no evidence of a differential impact of investments on future tax avoidance.

40
 

avoidance. However, I find no evidence of external investments resulting in higher future tax
avoidance payoffs when compared to internal investments.

4.2.4. Determinants of External Tax Function Investments
Results from estimating the logistic regression of the determinants of external
investments in the tax function are presented in Table 8. Consistent with H3a, firms facing
resource constraints due to being in a loss position are more likely to invest externally (LOSS
coefficient estimate = 1.376; p-value <0.05). Examining H3b, I find that risk and complexity are
positively associated with the decision to invest externally. Specifically, the coefficient on the
RESTATEMENT indicator is positive and significant (p-value <0.10). However, I find no
evidence of the complexity added by foreign operations contributing to the decision to engage an
external service provider. Finally, there is a time trend in the decision to engage an external
service provider. The results suggest that in 2007 firms were more likely to engage an external
service provider than in 2005 (p-value < 0.10). Model diagnostics suggest a reasonably good fit
with a pseudo r-squared of 13.2% and area under the ROC curve of 0.74.40

                                                            
40

Given the small sample size, I also investigate the potential impact of multicollinearity on the ability to detect
statistically significant determinants. All variance inflation factors (VIF) are below 4 with the exception of SIZE,
which has a VIF of 9.87, which could explain the lack of significance on some variables.

41
 

CHAPTER 5: SUPPLEMENTAL ANALYSES AND SENSITIVITY TESTS

In Chapter 5, I discuss several additional analyses to provide further perspective about the
main findings reported in Chapter 4. I begin by focusing on the potential endogeneity issue of a
certain type of firm disclosing tax-related MWs. I next examine additional specifications to
examine the specific timing of future tax avoidance payoffs. I then examine the impact of
additional control variables which are shown to be correlated with either the dependent variable
or the primary variables of interest. I conclude by re-estimating the main models using
unadjusted tax avoidance measures.

5.1. Tax Function Investment Firm Characteristics
One possible explanation for the primary results is the sample of firms making tax
function investments have certain characteristics which are correlated with the likelihood of
increasing future tax avoidance. This section presents the results of alternative tests, which seek
to alleviate concerns regarding sample selection, time trends in tax avoidance, and endogeneity.
To address concerns that my results are driven by characteristics unique to firms that
disclose MWs or a time trend in tax avoidance, I employ a simulation technique that randomly
assigns the tax function investment indicator to other firm-years within the sample of firms that
disclose at least one MW during my sample period. I then estimate model (1) employing
FutureCash3yrETR as the dependent variable and using these alternative assignments. I repeat
this procedure for 10,000 iterations and obtain coefficient and standard error estimates for each
of these simulations. The mean coefficient estimate on TAX_FIXED using the simulation
technique is -0.001 (t-stat = -0.152; p-value = 0.879). Additionally, only 35 of the 10,000

42
 

simulations have a negative and significant coefficient estimate on the TAX_FIXED indicator
variable equal to the magnitude of my primary result.41 This implies there is a 0.35% chance of
obtaining my results by randomly assigning the tax function investment indicator variable to
alternative years in the sample period, alleviating concerns that the primary result is driven by
chance or a systematic increase in tax avoidance across time. These tests also help rule out the
possibility that my results are driven by other characteristics of firms which disclose MWs.
Specifically, if the characteristics of tax function investment firms lead to constantly increasing
tax avoidance I would expect to find a significant association between the randomly assigned tax
function investment indicator and future tax avoidance.
To further address the issue endogeneity I employ two alternative econometric
specifications including 1) a changes model including lagged 3-year cash ETR measures a time t,
t+1, and t+2 as additional controls and 2) a model including firm fixed effects. A changes model
which includes lagged measures of the level of tax avoidance controls for the starting level of tax
avoidance prior to the investment. Employing a changes specification also improves the ability
to make causal inferences by controlling for alternative explanatory factors and establishing
temporal precedence (Campbell and Cook 1979). The firm fixed effects model subtracts the
mean value of each firm’s tax avoidance across the sample period, transforming the model to
explain deviations of each firm’s tax avoidance from its mean level over the sample period. This
specification controls for all time-invariant firm characteristics providing a strong test of the
impact of tax function investments on tax avoidance. Results are tabulated in Table 9. Inferences
are qualitatively similar for the cash ETR specifications, but I am unable to confirm the GAAP
ETR results using these alternative models.
                                                            
41

Results are also confirmed if a simulation test is run for 10,000 iterations on only firms that disclose an MW
during the sample period.

43
 

Finally to address sample selection concerns I adjust the control group of firms by
confirming my results using various subsamples. These subsamples include all firm-years
consisting of 1) only firms that disclose a MW during the sample period; 2) only firms that
disclose a tax-related MW during the sample period; and 3) only firms that disclose and
remediate a tax-related MW during the sample period. Results are presented in Table 10. The
cash ETR results are also robust to the smaller subsamples described in 1), 2) and 3); however,
the GAAP ETR specifications do not yield statistically significant coefficient estimates on the
tax function investment indicators using these subsamples.42

5.2. Timing of Payoffs on Tax Function Investments
The main results suggest a tax function investment at time t is associated with higher
levels of tax avoidance measured across times t+1 to t+3. To more precisely determine the
timing of the impact of tax function investments on future tax avoidance I re-estimate model (1)
replacing the dependent variable with the 1-year relative cash ETR measured at time t+1, t+2,
and t+3, respectively, and report in Table 11. The coefficient estimate on TAX_FIXED
employing tax avoidance at time t+1 as the dependent variable is negative and insignificant
(coefficient = -0.048, p-value = 0.105). Tax function investments in year t appear to result in 5.1
percentage point lower 1-year relative cash ETRs in period t+2 (p-value < 0.05), consistent with
a two-year time lag between tax function investment and payoffs via increased tax avoidance.
However, the coefficient estimate when the 1-year relative cash ETR at time t+3 is examined is
positive and marginally significant (coefficient = 0.054, p-value = 0.068). These results suggest
that the timing of payoffs on tax function investments examined in this study is approximately
                                                            
42

Results for the FutureCash3yrETRt+1-t+3 and FutureGAAP3yrETRt+1-t+3 are corroborated using a subsample of
large firms (S&P 1500) following Bauer (2013).

44
 

two years. This result provides useful information for managers evaluating tax function
investment decisions.

5.3. Additional Control Variables
Prior research documents a number of variables that are correlated with tax avoidance or
the disclosure of MWs. If these variables are correlated with both tax avoidance and the
remediation of MWs there is a significant threat to the internal validity of this study. In this
section I discuss the additional control variables I include to address these concerns. First, to
control for the possibility that shareholder monitoring impacts the level of tax avoidance and the
likelihood of disclosing a tax-related MW, I include institutional ownership (Khurana and Moser
2013). Second, I include an indicator variable for the presence of a restatement linked to the
disclosed MW to control for the severity of accounting issues leading to the MW. Finally, I
include a measure of firm governance, which has been shown to moderate the association
between ICWs and CFO compensation (Hoitash et al. 2012) . Specifically, I use the g-index from
Gompers et al. (2003) as a control variable in my primary model. Results from the alternative
specifications which include all three of these additional control variables are tabulated in Table
12. Due to a lack of data for these variables for many of the sample firms, I lose roughly 30
percent of the tax investment observations. Even with this reduced sample size, however, I
continue to find a positive and significant association between tax function investments and
future tax avoidance using cash ETRs. However, the GAAP ETR results are not confirmed
using this truncated sample.

45
 

5.4. Alternative Measures of Tax Avoidance
I replicate model (1) using unadjusted cash and GAAP ETRs to illustrate that the findings
are not driven by the industry-year size adjusting procedure. I also modify model (1) by using
average values for the control variables measured across t+1 to t+3. Given that the dependent
variable is measured across that same period, this alternative specification controls for alternative
factors over that same measurement period. Results for these additional tests are presented in
Table 13. Results for both the cash and GAAP ETR specifications are confirmed using these
alternative research designs.
Overall, these sensitivity tests confirm the primary findings that tax function investments
are positively associated with higher levels of cash tax avoidance, and in certain circumstances,
GAAP tax avoidance. These results suggest the timing of payoffs is approximately two years.
The alternative tests alleviate concerns that the observed association is driven by the sample of
tax function investment firms or other potentially endogenous factors.

46
 

CHAPTER 6: CONCLUDING REMARKS

This study uses the disclosed remediation of tax-related MW to identify firms that
increase tax function investments to provide evidence on the magnitude and timing of payoffs on
such investments. Results are consistent with increases in current period tax function investments
being positively associated with future tax avoidance measured over time t+1 to t+3. To control
for sample selection issues related to the tax function investment firms and other endogeneity
concerns these results are confirmed using several alternative economic specifications including
a changes specification, firm fixed effects, and various subsamples. Descriptive evidence
suggests firms engaged in foreign and state tax planning activities to increase tax avoidance
consistent with these tax function investments enabling firms to implement potentially
sophisticated tax planning activities. Supplemental analyses suggest the timing lag between
investments and tax avoidance payoffs is approximately two years. I provide evidence that both
external and internal tax function investments are associated with higher levels of future tax
avoidance, but find no evidence that external investments yield higher tax avoidance payoffs.
Results indicate that resource constraints and the complexity and risk of the disclosed MW are
positively associated with the decision to engage an external advisor to assist in remediating a
MW.
This study contributes to the literature that examines the impact of improvements in
internal control due to SOX on firm performance (Feng et al. 2013; Cheng et al. 2013). By
documenting significant tax avoidance payoffs, this study demonstrates how internal control
improvements can have spillover effects onto other aspects of firm operations. This study also
enhances our understanding of tax avoidance payoffs on investments in tax planning. Due to the

47
 

time value of money, determining the magnitude and timing of payoffs on capital outlays is an
important part of evaluating investment alternatives (Brealey and Myers 2002; Busby and Pitts
1997). These findings further our understanding of the economic payoffs on increases in tax
function investments providing useful information for managers and practitioners.
This study is subject to limitations. The specialized setting used to operationalize tax
function investments (i.e., remediation of tax-related MWs) could limit external validity. Results
using simulation analyses and various subsamples alleviate concerns regarding endogeneity and
spurious correlation, but I cannot rule out the possibility that other types of tax function
investments in an alternative sample of firms could impact tax avoidance in a different manner.

48
 

APPENDICES

49
 

APPENDIX A

Tax-Related MW Disclosures

50
 

1. H&R Block, Inc. 2005/2006 10-K
Description of Weakness
Based on our assessment, management determined that a material weakness existed in the Company’s internal
controls over accounting for income taxes as of April 30, 2005. Specifically, the Company did not maintain
sufficient resources in the corporate tax function to accurately identify, evaluate and report, in a timely manner, nonroutine and complex transactions. In addition, the Company had not completed the requisite historical analysis and
related reconciliations to ensure tax balances were appropriately stated prior to the completion of the Company’s
internal control activities. These deficiencies resulted in errors in the Company’s accounting for income taxes. These
errors were corrected prior to issuance of the consolidated financial statements as of and for the year ended April 30,
2005. In the aggregate, these deficiencies represent a material weakness in internal control over financial reporting
on the basis that there is a more than remote likelihood that a material misstatement of the Company’s annual or
interim financial statements will not be prevented or detected by its internal control over financial reporting.
Because of this material weakness in internal control over financial reporting, management concluded that, as of
April 30, 2005, the Company’s internal control over financial reporting was not effective based on the criteria set
forth by COSO.

Remediation steps-2005 10K
We have dedicated substantial resources to the review of our control processes and procedures specifically related to
accounting for income taxes. Based on the results of this review, during the fourth quarter, management completed
numerous enhancements to improve our internal controls over financial reporting, specifically those related to
accounting for income taxes, including the following actions:
▪ Implemented a comprehensive set of policies and procedures related to accounting for income taxes.
▪ Filled senior-level positions in the corporate tax department with experienced individuals focusing on corporate
tax, state/local tax, and mortgage accounting.
▪ Engaged a qualified third-party firm to provide supplementary assistance, REMIC transaction tax expertise, and
to assess the tax implications of select historical and future securitizations and the adequacy of the model used by
Mortgage Services to track the related book/tax basis adjustments.
▪ Increased the formality and rigor around the operation of key controls.
Other than the changes outlined above, there were no changes that materially affected, or are reasonably likely to
materially affect, our internal control over financial reporting.
In order to remediate the material weakness identified by management as of April 30, 2005, and continuing
thereafter, management completed the requisite historical analysis including creation of the necessary tax basis
balance sheets and current and deferred reconciliations required and related internal control testing to ensure
propriety of all tax related financial statement account balances as of this Form 10-K filing date. The Company
believes it has established appropriate controls and procedures and created the appropriate tax account analysis and
support subsequent to April 30, 2005. In addition to the above actions, management will conduct a comprehensive
evaluation of the corporate tax function, including resource requirements, during the current fiscal year to identify
and implement additional improvements to ensure compliance with the controls and procedures that have been put
in place to remediate deficiencies previously identified.
Remediation steps-2006 10K
CHANGES IN INTERNAL CONTROL OVER FINANCIAL REPORTING –

51
 

During the fourth quarter of fiscal year 2006, we completed remediation efforts relating to a material weakness in
our controls over accounting for income taxes that was reported as of April 30, 2005. In addition to control
enhancements identified in our previously filed reports on Form 10-Q, management implemented additional
improvements to controls in the state income tax rate calculation process to incorporate the use of current period pro
forma federal and state taxable income calculations and the use of current and projected state apportionment factors,
among other data inputs.
Other than the changes outlined above, there were no changes that materially affected, or are reasonably likely to
materially affect, our internal control over financial reporting.

2. Symantec Corporation 2005/2006 10-K
Description of Weakness
Based on its evaluation, our management has identified a material weakness in internal control over financial
reporting related to accounting for income taxes as of March 31, 2006…
Management has determined that we had insufficient personnel resources with adequate expertise to properly
manage the increased volume and complexity of income tax matters associated with the acquisition of Veritas
Software Corporation. This lack of resources resulted in inadequate levels of supervision and review related to the
our IRS filings and our accounting for income taxes. This material weakness resulted in our failure to follow
established policies and procedures designed to ensure timely income tax filings. Specifically, we did not complete
the timely filing of an extension request with the IRS for the final pre-acquisition income tax return for Veritas and,
accordingly, did not secure certain income tax related elections. In addition, this material weakness resulted in errors
in our annual accounting for income taxes. These errors in accounting were corrected prior to the issuance of our
2006 consolidated financial statements. The aforementioned material weakness results in more than a remote
likelihood that a material misstatement of our annual or interim financial statements, due to a failure to complete
income tax filings consistent with management’s intentions, and due to errors in accounting for income taxes, would
not be prevented or detected.

Remediation steps-2005 10K
At the end of February 2006, we hired a new Vice President of Tax and Treasury to help manage the increased
complexity of our income tax matters. During the quarter ended March 31, 2006, there were no other changes in our
internal control over financial reporting that have materially affected, or are reasonably likely to materially affect,
our internal control over financial reporting.
Since April 1, 2006, we have implemented additional controls in our internal control over financial reporting that
serve to remediate the material weakness described above, including the addition of resources dedicated to financial
reporting for income taxes and the implementation of processes to identify and calendar all incremental tax
compliance and financial accounting for income tax requirements arising from acquisitions. In addition, we intend to
automate key elements of our processes to enhance the analysis and calculation of the income tax provision and the
reconciliation of the tax accounts.

Remediation steps-2006 10K
Our annual report on Form 10-K for the fiscal year ended March 31, 2006 disclosed a material weakness relating to
accounting for income taxes. In order to remediate this material weakness, during the first three quarters of fiscal
2007, we:
• Completed our restructuring of personnel dedicated to financial reporting for income taxes;

52
 

• More specifically defined existing key controls, and developed additional controls, applicable to our interim
accounting for income taxes;
• Automated certain elements of our processes to enhance the analysis and calculation of the income tax
provision and the reconciliation of the tax accounts;
• Enhanced the documentation regarding conclusions reached in the implementation of generally accepted
accounting principles; and
• Added additional levels of review by qualified personnel of the application of each key control.
As a result of these actions, management has concluded that Symantec has remediated the material weakness as of
March 31, 2007. Although certain steps were taken in the fourth quarter of fiscal 2007 to address the material
weakness relating to accounting for income taxes, there were no changes in Symantec’s internal control over
financial reporting during the quarter ended March 31, 2007 that have materially affected, or are reasonably likely to
materially affect, Symantec’s internal control over financial reporting.

3. Netflix Holdings Inc. 2005/2006 10-K
Description of Weakness
Management identified a material weakness in our internal control over financial reporting as of December 31, 2005
related to our accounting for income taxes. Specifically, our policies and procedures do not include adequate
management review of the calculations and related supporting documentation to ensure that its accounting for
income taxes is in accordance with generally accepted accounting principles. This material weakness resulted in a
material error in the Company’s consolidated financial statements related to the understatement of Deferred Tax
Assets in the consolidated balance sheet and the understatement of the Benefit from Income Taxes in the
consolidated statement of income. This error was corrected prior to the filing of our 2005 consolidated financial
statements included in Item 8 of this Form 10-K.

Remediation steps-2005 10K
We are taking the following action to remediate the material weakness described above: implementing additional
review procedures to ensure complete supporting documentation is available to ensure compliance with generally
accepted accounting principles; this action will be in place in connection with the preparation of our financial
statements for the first quarter of 2006.
Remediation steps-2006 10K
Remediation of Material Weakness
In the first quarter of 2006, we implemented additional review procedures to ensure complete supporting
documentation is available to ensure that our accounting for income taxes is in accordance with generally accepted
accounting principles; this action was in place in connection with the preparation of our financial statements for the
first quarter of 2006. As such, we believe that the remediation initiative outlined above was sufficient to eliminate
the material weakness in internal control over financial reporting as discussed above.

53
 

4. PAR Pharmaceutical Companies 2007/2008 10-K
Description of Weakness
The Company’s management has concluded that the Company did not maintain effective internal controls over
financial reporting as of December 31, 2007 due to the restatement of the consolidated financial statements for the
years ended December 31, 2006 and 2005 to properly determine the income tax benefits from discontinued
operations and related interest expense, as a result of a material weakness that existed at that time. The Company
lacked a formal process to identify, analyze and communicate non-routine tax matters and lacked sufficient
oversight by management for non-routine tax matters.
Remediation steps-2007 10K
The Company has taken the following steps to remediate the weakness mentioned above: 1) replaced members of
senior management and managers responsible for the oversight of income tax matters, 2) formalized a policy and
procedure for the communication and review of non-routine tax matters by senior management, and 3) as
appropriate the Company will engage external tax advisors for advice with respect to non-routine tax matters.
Remediation steps-2008 10K
As of the date of the filing of our 2007 Annual Report on Form 10-K, Par did not maintain effective controls due to
the restatement of the consolidated financial statements for the years ended December 31, 2006 and 2005 to properly
determine the income tax benefits from discontinued operations and related interest expense, as a result of a material
weakness that existed at that time. Par lacked a formal process to identify, analyze and communicate non-routine tax
matters and lacked sufficient oversight by management for non-routine tax matters

The following measures have been taken to address the 2007 material weaknesses mentioned above:
Par had replaced members of senior management subsequent to the identification of the transaction that gave rise
to the material weakness as of the filing of the 2007 Annual Report on Form 10-K.
Formalized a policy and procedure for the communication and review of non-routine tax matters by senior
management.
As appropriate Par will engage external tax advisors for advice with respect to non-routine tax matter

5. Shaw Group Inc. 2008/2009 10-K

Description of Weakness
We identified the following material weaknesses in internal controls over financial reporting that continued to exist
as of August 31, 2008:
(2)

Accounting for Income Taxes

54
 

We did not maintain a sufficient number of tax professionals with adequate experience in the application of
Financial Accounting Standards Board Statement No. 109, Accounting for Income Taxes (FAS 109). As a result, our
policies and procedures for the identification and analysis of the appropriate accounting treatment of routine and
non-routine income tax matters were not effective to ensure that the our income tax accounting was consistent with
generally accepted accounting principles. These control deficiencies give rise to a reasonable possibility of a
material misstatement in our annual or interim financial reporting not being prevented or detected on a timely basis.
This material weakness resulted in errors in the accounting for both our current and deferred income tax amounts
and related disclosures, which were corrected prior to issuance of this Form 10-K.

Remediation steps-2008 10K
Our planned remedial measures related to the material weaknesses identified above include:
For remedial measures related to Accounting for Income Taxes, we recently hired a new Vice President of
Tax to lead our global tax function who will be developing a plan to ensure we have an adequate number of
experienced tax accounting professionals. We will further enhance our policies and procedures related to the
application of FAS 109. We will continue to engage external tax resources as necessary to assist us until the
remedial measures can be designed, implemented and tested.
In light of the material weaknesses described above, we performed additional procedures that provided us with
reasonable assurance regarding the reliability of: (1) our financial reporting and (2) the preparation of the
consolidated financial statements contained in this Form 10-K. Accordingly, management believes that the
consolidated financial statements included in this Form 10-K fairly present, in all material respects, our financial
position, results of operations and cash flows for the periods presented.
We are committed to finalizing our remediation action plans and implementing the necessary enhancements to
remediate the material weaknesses described above. These material weaknesses will not be considered remediated
until (1) the new processes are designed, appropriately controlled and implemented for a sufficient period of time
and (2) we have sufficient evidence that the new processes and related controls are operating effectively.

Remediation steps-2009 10K
During the three months ended August 31, 2009, we completed testing of the control improvements that were
implemented in fiscal 2009 to remediate our material weaknesses that existed as of August 31, 2008. Based on the
results of our testing of the remedial actions taken during fiscal 2009, we believe the following material weaknesses
no longer existed as of August 31, 2009:

2)

Accounting for Income Taxes
As previously reported, we did not maintain a sufficient number for tax professionals with adequate
experience in the application of Financial Accounting Standards Board No. 109, Accounting for Income Taxes
(FAS 109) as of August 31, 2008. As a result, our policies and procedures for the identification and analysis of the
appropriate accounting treatment of routine and non-routine income tax matters were not effective to ensure that our
income tax accounting was consistent with generally accepted accounting principles. Based upon the remediation
actions we took during the first, second and third quarters of fiscal 2009 and our testing that was completed in the
fourth quarter of fiscal 2009 on the control improvements implemented during the year, we believe this material
weakness no longer existed as of August 31, 2009.

55
 

Discussion of Effective Tax Rate Changes in MDA-2010 10K
Consolidated Provision for Income Taxes:
Consolidated effective tax rate for fiscal year 2010 was 30% as compared to 37% for fiscal year 2009. The
decrease in our effective tax rate was primarily due to the mix of earnings between our domestic and foreign
operations, and a lower provision for uncertain tax positions in fiscal year 2010 as compared to fiscal year 2009. The
consolidated effective tax rate for fiscal year 2010 also includes a benefit of research and experimentation tax credits
in submitted claims or filed returns that were included in the provision for uncertain tax positions. See Note 10 —
Income Taxes to our consolidated financial statements for a reconciliation of the federal statutory rate to the
consolidated effective tax rate.

6. True Religion Apparel Inc. 2007/2008 10-K
Description of Weakness

Based on the COSO criteria, management identified control deficiencies that constitute material weaknesses. A
"material weakness" is a deficiency, or combination of deficiencies, in internal control over financial reporting, such
that there is more than a reasonable possibility that a material misstatement of the Company's financial statements
will not be prevented or detected on a timely basis. The following material weaknesses were identified:
• Accounting for income taxes: we did not properly determine the federal and state income tax
consequences of our executive compensation because our relevant controls were not designed or operating
effectively. These deficiencies resulted in errors that were material to our interim and annual financial
statements;

Remediation steps-2007 10K
We have taken actions that are intended to remediate our material weaknesses in internal control over financial
reporting, including the following:
(1) Hired finance and accounting department management that is more experienced in selecting the correct
accounting policies for, and establishing policies, procedures and controls to record and report, the
Company's business transactions;
(2) Implemented formal procedures and controls to record and report all material business transactions
timely (this effort is on-going);
(3) Hired additional accounting department personnel;
(4)Engaged new income tax and internal-control-over-financial-reporting advisors.
Remediation steps-2008 10K
No mention of prior year weakness or remediation.

Discussion of Effective Tax Rate Changes in MDA-2009 10K

56
 

Provision for Income Taxes- 2009 vs. 2008
The effective tax rate was 39.1% for 2009 compared to 36.6% in 2008. In 2008, we implemented a tax planning
strategy that retroactively changed our filing status in certain states, which reduced our cumulative tax provision by
$1.5 million. We also finalized our 2007 tax returns during the 2008 period, which included additional analysis of
our federal and state tax obligations; as a result of this analysis, we reduced our income tax provision in 2008 by
$0.6 million. These factors were the primary drivers resulting in a lower effective tax rate of 36.6% in 2008.
Provision for Income Taxes- 2008 vs. 2007
The effective tax rate was 36.6% for 2008 compared to 43.1% in 2007. The 2007 effective tax rate was higher than
the 2008 rate because in 2008 we implemented a tax planning strategy that retroactively changed our filing status in
certain states, which reduced our tax provision by $1.6 million and increased our diluted earnings per share by
$0.07. Additionally, in 2007 a larger portion of our executive compensation was nondeductible for income tax
purposes resulting in a higher effective income tax rate.

57
 

APPENDIX B

Variable Definitions

58
 

Variable
Dependent Variables

Definition

CurrentCash1yrETRt

Firm-level deviation from the year industry-size adjusted 1-year
mean of the cash ETR: Computed as (txpd/pi-spi) winsorized
between 0 and 1, less the mean (txpd/pi-spi) for all firms in the
same two digit SIC industry asset quintile for the given year where
each observation to compute the mean is winsorized between 0 and
1. All negative values of pi-spi are set to missing (Balakrishnan et
al. 2012).

CurrentGAAP1yrETRt

Firm-level deviation from the year industry-size adjusted 1-year
mean of the GAAP ETR: Computed as (txt/pi-spi) winsorized
between 0 and 1, less the mean (txt/pi-spi) for all firms in the same
two digit SIC industry asset quintile for the given year where each
observation to compute the mean is winsorized between 0 and 1.
All negative values of pi-spi are set to missing (Balakrishnan et al.
2012).

FutureCash3yrETRt+1-t+3

Firm-level deviation from the year industry-size adjusted 3-year
mean of the cash ETR centered on year t+2: Computed as Σtxpdt+1t+3/ (Σpi-spit+1-t+3) winsorized between 0 and 1, less the mean
Σtxpdt+1-t+3/ (Σpi-spit+1-t+3) for all firms in the same two digit SIC
industry asset quintile for the given year where each observation to
compute the mean is winsorized between 0 and 1. All negative
values of Σpi-spit+1-t+3 are set to missing (Balakrishnan et al. 2012).

FutureGAAP3yrETR t+1-t+3

Firm-level deviation from the year industry-size adjusted 3-year
mean of the GAAP ETR centered on year t+2: Computed as
Σtxtt+1-t+3/ (Σpi-spit+1-t+3) winsorized between 0 and 1, less the mean
Σtxtt+1-t+3/ (Σpi-spit+1-t+3) for all firms in the same two digit SIC
industry asset quintile for the given year where each observation to
compute the mean is winsorized between 0 and 1. All negative
values of Σpi-spit+1-t+3 are set to missing (Balakrishnan et al. 2012).

Internal Control Variables
WEAK_TAXt
Indicator variable for the presence of a tax-related MW: set to 1 if a
firm discloses a tax-related MW in the current or prior year, zero
otherwise (Ashbaugh-Skaife et al. 2008).
TAX_FIXEDt

Indicator variable for the remediation of a tax-related MW: set to 1
if a firm discloses a tax-related MW in the prior year but not the
current year, zero otherwise (Ashbaugh-Skaife et al. 2008).

WEAK_OTHERt

Indicator variable for any other type of MW: set to 1 if a firm has
an MW in the current or prior year and WEAK_TAXt is equal to zero,
zero otherwise (Ashbaugh-Skaife et al. 2008).
59

 

OTHER_FIXEDt

Indicator variable for the remediation of any other type of MW: set
to 1 if a firm discloses a non-tax-related MW in the prior year but
not the current year and WEAK_OTHERt is equal to zero, zero
otherwise (Ashbaugh-Skaife et al. 2008).

Control Variables
ROAt

Return on assets: pi/att-1

LEVt

Leverage: dltt/att-1

FORINCt

Foreign income: pifo/pi, missing values of pifo set to zero.

TAXHAVENt

Tax haven indicator: set to 1 if a firm discloses material operations
in a tax haven country on Exhibit 21 of Form 10-K (source: Dyreng
and Lindsey 2009).

INTANt

Intangibles: intan/at, missing values of intan set to zero.

SIZEt

Natural log of assets: ln(at)

NOLt

Net operating loss indicator: set to 1 if tlcf is greater than zero,
missing values of tlcf are set to zero.

RDt

Research and development intensity: xrd/revt, missing values of
xrd set to zero.

BMt-1

Lagged book to market: ceqt-1/(prcc_ft-1*cshot-1).

AGGR_LOSSt

Indicator variable for consecutive losses: set to 1 if the sum of
earnings before extraordinary items for the prior year and current
year are less than zero. (ibt-1 + ibt < 0).

CONSTR_CFt

Cash flow constraints: Equal to 1 minus operating cash flows
(oancf).

60
 

APPENDIX C

Tables

61
 

TABLE 1
MWs Disclosed by Year and Time to Remediate
Panel A: Number of Tax MWs and Other MWs by Year
Tax MWs
Remediated following year
Remediated in 2 years
Remediated in 3 years +
Unremediated or missing

2004
69
36
11
7
12

2005
92
67
11
1
13

2006
64
33
14
6
11

2007
38
24
7
0
7

2008
18
10
1
0
7

2009
22
13
2
0
7

2010
12
7
0
N/A
N/A

2011
17
N/A
N/A
N/A
N/A

Total
332
190
46
14
57

Other MWs
Remediated following year
Remediated in 2 years
Remediated in 3 years +
Unremediated or missing

136
99
9
3
25

161
118
12
2
29

135
86
16
3
30

109
72
10
0
27

78
42
6
1
29

50
31
4
0
15

40
22
2
N/A
N/A

60
N/A
N/A
N/A
N/A

769
470
59
9
155

Panel B: Remediation of Tax MWs and Other MWs during sample period
Tax MWs
Tax MWs
# Tax MW remediated next year
% of Tax MWs remediated next year

2004
69
36
52.2%

2005
92
67
72.8%

2006
64
33
51.6%

2007
38
24
63.2%

2008
18
10
55.6%

2009
22
13
59.1%

Total
303
183
61.4%

Other MWs
Other MWs
# Other MW remediated next year
% of Other MWs remediated next year

2004
136
99
72.8%

2005
161
113
70.2%

2006
135
93
68.9%

2007
109
85
78.0%

2008
78
51
65.4%

2009
50
36
72.0%

Total
669
477
71.3%

Note: Table 1 provides descriptive statistics for the number and time to remediate both tax and non-tax MWs. Panel A provides statistics on all ICWs in Audit Analytics from 2004-2009. Panel B
provides descriptive statistics on the MWs during my sample period of 2004-2009.

62
 

TABLE 2
Sample Selection
Audit Analytics Opinion File with Auditor IC Opinion firm years from 2004 to 2008
Less: Observations missing Compustat data to compute variables
Less: Observations with negative pretax income less special items t+1 to t+3
Observations available for t+1 to t+3 3-year Cash ETR analysis

Firm years that disclose a new tax-related MW from 2004 to 2008
Less: Observations missing Compustat/Audit Analytics data to compute variables
Less: Observations with negative pretax income less special items t+1 to t+3
Tax-related MW remediation observations available for t+1 to t+3 3-year Cash ETR analysis

63
 

16,718
(7,526)
(1,505)
7,687

303
(128)
(70)
105

TABLE 3
Descriptive Statistics Type of Investment by Year
Panel A: All Remediated Tax MWs
2005
36

2006
68

2007
46

2008
35

2009
19

2010
17

2011
9

Total

30
83.3%

42
61.8%

31
67.4%

20
57.1%

11
57.9%

9
52.9%

6
66.7%

149
64.8%

Hiring new tax director

19
52.8%

45
66.2%

27
58.7%

11
31.4%

8
42.1%

9
52.9%

2
22.2%

121
52.6%

Engaging outside consultant

22
61.1%

39
57.4%

26
56.5%

22
62.9%

16
84.2%

12
70.6%

5
55.6%

142
61.7%

Engaging auditor tax services

8
22.2%

22
32.4%

26
56.5%

21
60.0%

11
57.9%

13
76.5%

5
55.6%

106
46.1%

Only additional procedures/review

6
16.7%

14
20.6%

12
26.1%

6
17.1%

3
8.6%

4
23.5%

2
22.2%

47
20.4%

Total Remediated Tax MWs
Type of Tax Function Investment
Hiring additional personnel

64
 

230

 

TABLE 3 (cont’d)
Panel B: All Remediated Tax-related MWs in Sample

 
 

Total Remediated Tax-related MWs

2005
16

2006
36

2007
22

2008
25

2009
6

Total
105

 

Type of Tax Function Investment

 

Hiring additional personnel

6
37.5%

18
50.0%

16
72.7%

14
56.0%

4
66.7%

58
55.2%

Hiring new tax director

9
56.3%

11
30.6%

6
27.3%

9
36.0%

1
16.7%

36
34.3%

Internal Investment Only

10
62.5%

12
33.3%

8
36.4%

12
48.0%

2
33.3%

44
41.9%

Engaging External Party

6
37.5%

24
66.7%

14
63.6%

13
52.0%

4
66.7%

61
58.1%

External Investment Only

1
6.3%

15
41.7%

8
36.4%

6
24.0%

1
16.7%

31
29.5%

Only additional procedures/review

5
31.3%

9
25.0%

4
18.2%

6
24.0%

0
0.0%

24
22.9%

 
  

Note: Table 3 provides descriptive statistics for the number of remediated tax-related MWs and the tax function investment made. With the exception of “only additional
procedures/review” each investment is not mutually exclusive. For example, a firm could disclose the hiring of additional personnel and the engagement of an outside
consultant. Panel A provides descriptive statistics for all remediated tax-related MWs in Audit Analytics (2004-2011). Panel B provides descriptive statistics for remediated
tax-related MWs that are used in the multivariate analyses where the dependent variable is FutureCash3yrETRt+1-t+3.

65
 

TABLE 4
Descriptive Statistics Main and Matched Samples
Panel A: Full Sample n = 7,687
Column 1: Tax Function Investment Firms
Dependent Variables

CurrentCash1yrETRt
CurrentGAAP1yrETRt
FutureCash3yrETRt+1-t+3
FutureGAAP3yrETR t+1-t+3

n

Mean

25%

Column 2: Control Firms

Median

75%

SD

n

Mean

25%

Median

75%

SD

473
473
473
473

0.006
0.046***
0.049**
0.048***

-0.161
-0.130
-0.104
-0.086

-0.067***
0.044
0.011
0.041***

0.108
0.144
0.129
0.124

0.248
0.238
0.234
0.222

7214
7214
7214
7214

0.012
0.024
0.026
0.018

-0.117
-0.068
-0.115
-0.096

-0.019
0.032
0.000
0.022

0.097
0.098
0.098
0.097

0.197
0.165
0.194
0.175

473
473
473
473
473
473
473
473
473
473
473

0.056***
0.179
0.021**
0.581***
0.000
6.716***
0.507***
0.052
0.431**
0.271***
0.898

0.000
0.000
0.000
0.000
0.000
5.494
0.000
0.000
0.221
0.000
0.862

0.043***
0.092
0.003***
1.000***
0.000***
6.529***
1.000***
0.009***
0.396***
0.000***
0.919

0.114
0.265
0.034
1.000
0.000
7.836
1.000
0.079
0.598
1.000
0.966

0.131
0.244
0.048
0.494
0.001
1.602
0.500
0.077
0.378
0.445
0.126

7214
7214
7214
7214
7214
7214
7214
7214
7214
7214
7214

0.074
0.193
0.016
0.396
0.000
7.224
0.34
0.046
0.462
0.128
0.904

0.017
0.018
0.000
0.000
0.000
6.043
0.000
0.000
0.268
0.000
0.856

0.059
0.127
0.000
0.000
0.000
7.114
0.000
0.000
0.434
0.000
0.919

0.132
0.288
0.013
1.000
0.000
8.308
1.000
0.015
0.616
0.000
0.975

0.148
0.247
0.047
0.489
0.003
1.813
0.474
0.373
0.310
0.335
0.203

Control Variables

ROA
LEV
FORINC
TAXHAVEN
INTAN
SIZE
NOL
RD
BM
AGGR_LOSS
CONSTR_CF

Note: Table 4 Panel A provides descriptive statistics for the dependent, independent and control variables used in the primary analyses where the dependent variable is FutureCash3yrETRt+1-t+3 for all
firm-years of the 105 tax function investment firms (Column 1) and all other firm-years in the primary sample (column 2). All variables are defined in Appendix B. *** p<0.01, ** p<0.05, * p<0.1
(two-tailed) for t-tests and Wilcoxon rank-sum test across columns 1 and 2.

66
 

TABLE 4 (cont’d)
Panel B: Tax Function Investment Firms Pre- and Post-Investment n =473
Column 3: Pre-Investment
Dependent Variables

CurrentCash1yrETRt
CurrentGAAP1yrETRt
FutureCash3yrETRt+1-t+3
FutureGAAP3yrETR t+1-t+3

N

Mean

25%

Median

75%

SD

N

Mean

25%

75%

SD

248
248
248
248

-0.006
0.037
0.077***
0.064*

-0.167
-0.174
-0.096
-0.092

-0.080
0.033
0.020**
0.051

0.094
0.156
0.153
0.141

0.232
0.252
0.254
0.246

0.020
0.056
0.018
0.029

-0.155
-0.074
-0.115
-0.082

-0.053
0.049
0.000
0.031

0.113
0.133
0.106
0.102

0.265
0.220
0.206
0.191

248
248
248
248
248
248
248
248
248
248
248

0.057
0.176
0.018
0.605
0.000
6.686
0.512
0.057
0.429
0.266
0.894

0.004
0.000
0.000
0.000
0.000
5.438
0.000
0.000
0.225
0.000
0.869

0.040
0.100
0.004
1.000
0.000*
6.465
1.000
0.014
0.374
0.000
0.920

0.102
0.262
0.037
1.000
0.000
7.837
1.000
0.099
0.583
1.000
0.964

0.122
0.264
0.046
0.490
0.001
1.587
0.501
0.081
0.385
0.443
0.133

225
225
225
225
 
 
225
225
225
225
225
225
225
225
225
225
225

0.054
0.183
0.023
0.556
0.000
6.748
0.502
0.047
0.433
0.276
0.901

-0.003
0.000
0.000
0.000
0.000
5.605
0.000
0.000
0.219
0.000
0.858

0.045
0.081
0.002
1.000
0.000
6.586
1.000
0.006
0.420
0.000
0.917

0.122
0.285
0.030
1.000
0.000
7.821
1.000
0.070
0.625
1.000
0.968

0.140
0.220
0.049
0.498
0.001
1.622
0.501
0.072
0.370
0.448
0.119

Control Variables

ROA
LEV
FORINC
TAXHAVEN
INTAN
SIZE
NOL
RD
BM
AGGR_LOSS
CONSTR_CF

Column 4: Year of and Post-Investment
Median

Note: Table 4 Panel B provides descriptive statistics for the dependent, independent and control variables used in the primary analyses where the dependent variable is FutureCash3yrETRt+1-t+3 for
all firm-years of the 105 tax function investment firms. Column 3 provides descriptives for all firm-years prior to the tax function investment. Column 4 provides descriptives for all firm-years postinvestment (including the year of investment). All variables are defined in Appendix B. *** p<0.01, ** p<0.05, * p<0.1 (two-tailed) for t-tests and Wilcoxon rank-sum test across columns 3 and 4.

67
 

TABLE 4 (cont’d)
Panel C: Matched Sample n = 210
Column 5: TAX_FIXED = 1
Dependent Variables

CurrentCash1yrETRt
CurrentGAAP1yrETRt
FutureCash3yrETRt+1-t+3
FutureGAAP3yrETR t+1-t+3

N

Mean

Column 6: TAX_FIXED = 0

25%

Median

75%

SD

N

Mean

25%

Median

75%

SD

105
105
105
105

0.031
0.049
0.008***
0.021

-0.186
-0.098
-0.143
-0.127

-0.074
0.049
0.000*
0.024

0.126
0.116
0.118
0.102

0.310
0.235
0.197
0.207

105
105
105
105

0.004
0.069
0.082
0.050

-0.131
-0.074
-0.078
-0.124

-0.047
0.034
0.025
0.029

0.115
0.127
0.170
0.123

0.215
0.238
0.242
0.245

105
105
105
105
105
105
105
105
105
105
105

0.038
0.187
0.017
0.543
0.001
6.615
0.505**
0.052
0.455
0.324
0.904

-0.017
0.000
0.000
0.000
0.000
5.486
0.000
0.000
0.236
0.000
0.863

0.030**
0.075
0.002
1.000
0.000
6.408
1.000**
0.008*
0.416
0.000
0.919

0.087
0.276
0.025
1.000
0.001
7.564
1.000
0.089
0.628
1.000
0.966

0.136
0.234
0.038
0.501
0.001
1.531
0.502
0.078
0.402
0.470
0.118

105
105
105
105
105
105
105
105
105
105
105

0.062
0.155
0.009
0.457
0.000
6.695
0.343
0.055
0.415
0.229
0.909

0.011
0.000
0.000
0.000
0.000
5.652
0.000
0.000
0.226
0.000
0.857

0.049
0.087
0.000
0.000
0.000
6.672
0.000
0.000
0.387
0.000
0.922

0.119
0.253
0.021
1.000
0.000
7.774
1.000
0.028
0.564
0.000
0.985

0.153
0.188
0.061
0.501
0.001
1.672
0.477
0.194
0.239
0.422
0.130

Control Variables

ROA
LEV
FORINC
TAXHAVEN
INTAN
SIZE
NOL
RD
BM
AGGR_LOSS
CONSTR_CF

Note: Table 4 Panel C provides descriptive statistics for the dependent, independent and control variables used in the primary analyses where the dependent variable is FutureCash3yrETRt+1-t+3 for
the 105 tax function investment firm-years (column 5) and the matched sample firm-years (column 6). All variables are defined in Appendix B. *** p<0.01, ** p<0.05, * p<0.1 (two-tailed) for t-tests
and Wilcoxon rank-sum test across columns 5 and 6.

68
 

VARIABLES

WEAK_TAXt
TAX_FIXEDt
WEAK_OTHERt
OTHER_FIXEDt
ROAt
LEVt
FORINCt
TAXHAVENt
INTANt
SIZEt
NOLt
RDt
BMt-1

TABLE 5
OLS ETR Regressions Main Analysis
Contemporaneous
Future
Tax Avoidance at t
Tax Avoidance at t+1 to t+3
Full Sample
Full Sample
(1)
(2)
(3)
(4)
Current
Current
Future
Future
Cash1yrETR GAAP1yrETR Cash3yrETR GAAP3yrETR
(t)
(t)
(t+1 to t+3)
(t+1 to t+3)
0.040**
(2.009)
0.005
(0.157)
0.037***
(2.782)
-0.068***
(-4.052)
0.061**
(2.057)
-0.024*
(-1.926)
-0.032
(-0.422)
0.012**
(2.207)
13.037***
(3.492)
-0.007***
(-3.902)
-0.014***
(-2.624)
-0.173***
(-4.646)
-0.005
(-0.705)

0.042**
(2.102)
-0.012
(-0.464)
0.004
(0.366)
0.001
(0.100)
0.159***
(6.389)
-0.012
(-1.115)
-0.209***
(-3.054)
-0.000
(-0.088)
13.813***
(3.757)
-0.005***
(-3.621)
0.005
(1.038)
-0.046*
(-1.745)
0.003
(0.470)

0.063***
(2.936)
-0.081***
(-3.499)
-0.012
(-0.899)
0.001
(0.034)
0.193***
(7.141)
-0.050***
(-3.355)
0.006
(0.062)
-0.001
(-0.176)
0.392
(1.046)
-0.001
(-0.527)
-0.013*
(-1.916)
-0.007*
(-1.825)
-0.011
(-0.957)

69
 

0.062***
(3.179)
-0.043*
(-1.732)
-0.007
(-0.540)
0.020
(1.144)
0.181***
(9.166)
-0.035***
(-3.073)
-0.220**
(-2.377)
-0.001
(-0.108)
0.674*
(1.708)
-0.007***
(-3.601)
0.010
(1.637)
0.003*
(1.856)
-0.022**
(-2.223)

Future
Tax Avoidance at t+1 to t+3
Matched Sample
(5)
(6)
Future
Future
Cash3yrETR GAAP3yrETR
Unadjusted
Unadjusted
(t+1 to t+3)
(t+1 to t+3)
0.017
(0.573)
-0.075***
(-2.975)
-0.048
(-1.283)
0.080
(1.517)
0.164
(1.647)
-0.045
(-0.659)
0.140
(0.643)
0.010
(0.385)
-24.246***
(-2.603)
-0.008
(-0.695)
-0.020
(-0.785)
-0.220**
(-2.052)
0.061
(1.366)

0.025
(0.905)
-0.044*
(-1.672)
0.020
(0.521)
-0.035
(-0.885)
0.310**
(2.509)
-0.059
(-0.905)
-0.337*
(-1.823)
-0.003
(-0.121)
-15.041*
(-1.655)
-0.004
(-0.352)
0.005
(0.189)
-0.240*
(-1.783)
0.001
(0.038)

TABLE 5 (cont’d)
-0.061***
(-6.468)
0.016
(0.642)
-0.117*
(-1.670)

-0.014*
(-1.698)
-0.016
(-0.720)
-0.072
(-1.133)

0.006
(0.522)
0.072***
(3.606)
-0.115
(-1.121)

-0.021**
(-2.061)
0.004
(1.128)
-0.074
(-0.840)

0.041
(0.914)
0.117***
(5.693)
0.084
(0.764)

0.009
(0.215)
0.118***
(4.867)
0.182
(1.539)

Year Effects
Industry Effects

Yes
Yes

Yes
Yes

Yes
Yes

Yes
Yes

Yes
Yes

Yes
Yes

# TAX_FIXED = 1
# OTHER_FIXED = 1

139
277

142
287

105
186

109
195

105
28

105
28

14,932
0.052

7,687
0.053

8,211
0.065

707
0.131

710
0.148

AGGR_LOSSt
CONSTR_CFt
INTERCEPT

Observations
R-squared

14,399
0.048

Note: Table 5 displays results from the estimation of the OLS regression model (1) for the full sample of firms (columns 1-4) and the matched sample of firms (columns 5-6).
The primary variable of interest is TAX_FIXED. Robust t-statistics in parentheses. Standard errors are clustered by firm. All variables are defined in Appendix B. *** p<0.01,
** p<0.05, *p<0.10 (two-tailed).

70
 

TABLE 6
Disclosed Factors for Increase in Tax Avoidance
Largest Quintile of Increases in Future Tax Avoidance: N = 24
16
14

14 (58.3%)

12
10
8

7 (29.2%)

6

7 (29.2%)
5 (20.8%)
4 (16.7%)

4
2
0
Foreign Tax Expense State Tax Expense

VAA

UTB

Other (R&D, DPAD,
M&A)

Note: Table 6 provides descriptive statistics for the tax function investment firms in the largest quintile of decreases in FutureCash3yrETRt+1-t+3.. Disclosed
factors for these decreases are hand-collected from each firm’s 10-Ks from time t+1, t+2, and t+3, respectively. A disclosed reason is coded as present if
management discloses the factor either in the MDA or tax footnote as being a factor leading to the decrease in ETR in any of those years. Valuation allowance
account is abbreviated as VAA. Unrecognized tax benefit account is abbreviated as UTB. The Other category includes research and development credits (R&D),
domestic producer’s activities deduction (DPAD), and mergers and acquisitions (M&A).Disclosed factors are not mutually exclusive.

71
 

TABLE 7
OLS 3-year ETR Regressions External and Internal Investments
3-year
3-year
Cash ETRs
Cash & GAAP ETRs
Full Sample
Matched Sample
(1)
(2)
(3)
(4)
VARIABLES
Future
Future
Future
Future
Cash3yrETR
GAAP3yrETR
Cash3yrETR
GAAP3yrETR
(t+1 to t+3)
(t+1 to t+3)
Unadjusted
Unadjusted
(t+1 to t+3)
(t+1 to t+3)
β1WEAK_TAXt

0.063***
(2.936)
-0.079***
(-2.783)
-0.085**
(-2.527)
-0.012
(-0.898)
0.001
(0.034)

0.066***
(3.368)
-0.060**
(-2.080)
-0.053
(-1.587)
-0.007
(-0.524)
0.020
(1.113)

0.017
(0.570)
-0.084***
(-3.018)
-0.064*
(-1.681)
-0.049
(-1.293)
0.081
(1.521)

0.025
(0.902)
-0.057*
(-1.767)
-0.027
(-0.863)
0.020
(0.503)
-0.035
(-0.876)

Controls
Year Effects
Industry Effects

Yes
Yes
Yes

Yes
Yes
Yes

Yes
Yes
Yes

Yes
Yes
Yes

F-test: β2 > β3
Prob > F =

0.02
0.892

0.04
0.848

0.23
0.630

0.66
0.418

105

109

105

105

7,687
0.053

8,211
0.065

707
0.133

710
0.149

β2EXTERNAL_FIXt
β3INTERNAL_FIXt
β4WEAK_OTHERt
β5OTHER_FIXEDt

# TAX_FIXED = 1
Observations
R-squared

Note: Table 7 displays results from the estimation of the OLS regression model (1) for the full sample of firms (columns 1 & 2) and the
estimation of the OLS regression model (1) using unadjusted future 3-year cash/GAAP ETRs as the dependent variable for the matched sample
(columns 3 & 4). The primary variables of interest are EXTERNAL_FIX and INTERNAL_FIX. EXTERNAL_FIX is equal to 1 if a firm discloses
using an external advisor to assist in the remediation of the disclosed tax-related MW, zero otherwise. INTERNAL_FIX is equal to one of a firm
discloses only investing internally to remediate the disclosed tax-related MW, zero otherwise. Robust t-statistics in parentheses. Standard errors
are clustered by firm. All variables are defined in Appendix B. *** p<0.01, ** p<0.05, *p<0.10 (two-tailed). An f-test is also performed to test
for differential payoffs on external versus internal tax function investments (i.e., β2 > β3) (Prob: F > 0: two-tailed).

72
 

Table 8
Logistic Regression:
Determinants of External Investments
VARIABLES

Externalt

Marginal
Effects

RESTATEMENTt

1.178*
(1.87)
0.738
(1.31)
3.224
(1.58)
-0.010
(-0.06)
1.376**
(2.17)
-0.799
(-0.72)
-3.382
(-1.13)
0.424
(0.68)
-2.393
(-0.95)
1.006
(1.38)
1.355*
(1.72)
0.259
(0.36)
0.945
(0.97)
-1.247
(-0.86)

28.2%*

FOROPSt
ROAt
SIZEt
LOSSt
LEVt
RDt
BMt-1
INVENTORY
2006
2007
2008
2009
CONSTANT
Year Effects

Yes

Observations

105

Pseudo R-squared
ROC Curve:

 

-0.02%
33.0%**
-19.1%
-81.0%
10.2%
-57.3%
24.5%
31.8%
6.4%
23.1%

0.132
0.74

 
 
73
 

77.2%

Note: Table 8 displays results from the estimation of the logistic regression model (2) using
External_FIX as the dependent variable for the subsample of firms which are making a tax
function investment. External_FIX is equal to 1 for firms which engage an external service
provider to remediate their tax-related MWIC, zero otherwise. Robust t-statistics in parentheses.
Standard errors are clustered by firm. All variables are defined in Appendix B and the text of the
manuscript. *** p<0.01, ** p<0.05, *p<0.10 (two-tailed). Marginal effects are calculated for a one
unit change of the variable of interest holding all other variables constant at their respective means.

 

 

17.7%

VARIABLES

WEAK_TAXt
TAX_FIXEDt
WEAK_OTHERt
OTHER_FIXEDt
Cash3yrETRt+2
Cash3yrETRt+1
Cash3yrETRt

TABLE 9
OLS Changes and Firm Fixed Effects Specifications
Changes Analysis
Firm Fixed Effects
(1)
(2)
(3)
(4)
Change
Change
Future
Future
Future
Future
Cash3yrETR GAAP3yrETR Cash3yrETR GAAP3yrETR
(t+1 to t+3)
(t+1 to t+3)
(t+1 to t+3)
(t+1 to t+3)
0.033*
(1.656)
-0.064**
(-2.488)
-0.003
(-0.262)
-0.018
(-1.041)
0.546***
(20.835)
0.085***
(2.932)
-1.017***
(-53.171)

0.021
(1.236)
-0.031
(-1.215)
-0.010
(-0.833)
0.017
(0.949)

0.028
(0.859)
-0.053**
(-2.033)
-0.011
(-0.566)
-0.002
(-0.080)

0.012
(0.465)
-0.021
(-0.753)
-0.023
(-1.084)
0.020
(0.944)

0.460***
(18.663)
0.099***
(4.565)
-1.012***
(-53.648)

GAAP3yrETRt+2
GAAP3yrETRt+1
GAAP3yrETRt
Controls
Firm Fixed Effects
Industry Fixed Effects
Year Fixed Effects

Yes
No
Yes
Yes

Yes
No
Yes
Yes

Yes
Yes
No
Yes

Yes
Yes
No
Yes

# TAX_FIXED = 1

95

107

105

109

6,741
0.528

7,994
0.530

7,687
0.663

8,211
0.685

Observations
R-squared

Note: Table 9 displays results from the estimation of the OLS regression model (1) replacing the dependent variable with the change in ETR
computed as the difference in 3 year ETR from (t-2 to t) to the 3 year ETR from (t+1 to t+3) and including lagged measures of 3 year ETRs
(columns 1 & 2) and employing firm fixed effects (columns 3 & 4). The primary variable of interest is TAX_FIXED. Robust t-statistics in
parentheses. Standard errors are clustered by firm. All variables are defined in Appendix B. *** p<0.01, ** p<0.05, *p<0.10 (two-tailed).

74
 

TABLE 10
OLS 3-year ETR Regressions-Additional Controls
(1)
(2)
VARIABLES
Future
Future
Cash3yrETR
GAAP3yrETR
(t+1 to t+3)
(t+1 to t+3)
0.089***
(3.463)
-0.084***
(-2.795)
-0.005
(-0.265)
-0.025
(-1.409)
-0.001
(-0.558)
-0.040**
(-2.112)
-0.003
(-0.239)
-0.062
(-0.485)

0.082***
(3.073)
-0.042
(-1.146)
-0.005
(-0.288)
0.002
(0.111)
-0.002
(-1.449)
-0.035**
(-2.106)
-0.012
(-1.551)
-0.142
(-1.525)

Controls
Year Effects
Industry Effects

Yes
Yes
Yes

Yes
Yes
Yes

# TAX_FIXED = 1

52

55

3,690
0.065

3,772
0.079

WEAK_TAXt
TAX_FIXEDt
WEAK_OTHERt
OTHER_FIXEDt
GINDEX
IO
RESTATEMENT
CONSTANT

Observations
R-squared

Note: Table 10 displays results from the estimation of the OLS regression model (1) including
additional controls for governance (GINDEX: G-index score for year t (Gompers et al. 2003)),
institutional ownership (IO: equal to percent of firm owned by institutional owners in year t), and
RESTATEMENT as previously defined. The primary variable of interest is TAX_FIXED. Robust tstatistics in parentheses. Standard errors are clustered by firm. All variables are defined in Appendix B.
*** p<0.01, ** p<0.05, *p<0.10 (two-tailed).

 

75
 

 

VARIABLES

TABLE 11
OLS 3-year ETR Regressions MW Subsamples
Firms with
Firms with
any type of MW
a tax MW
(1)
(2)
(3)
(4)
Future
Future
Future
Future
Cash3yrETR GAAP3yrETR Cash3yrETR GAAP3yrETR
(t+1 to t+3)
(t+1 to t+3)
(t+1 to t+3)
(t+1 to t+3)

Firms who remediate
a tax MW
(5)
(6)
Future
Future
Cash3yrETR GAAP3yrETR
(t+1 to t+3)
(t+1 to t+3)

0.041*
(1.803)
-0.084***
(-3.548)
-0.020
(-1.206)
-0.001
(-0.082)

0.048**
(2.461)
-0.041
(-1.629)
-0.010
(-0.597)
0.018
(1.000)

0.036
(1.223)
-0.080***
(-3.135)
-0.027
(-0.804)
-0.047*
(-1.703)

0.026
(1.132)
-0.032
(-1.256)
0.002
(0.050)
-0.002
(-0.047)

0.067*
(1.971)
-0.102***
(-3.456)
-0.053
(-1.634)
-0.033
(-1.000)

0.026
(0.930)
-0.037
(-1.301)
-0.006
(-0.142)
-0.001
(-0.028)

Controls
Year Effects
Industry Effects

Yes
Yes
Yes

Yes
Yes
Yes

Yes
Yes
Yes

Yes
Yes
Yes

Yes
Yes
Yes

Yes
Yes
Yes

# TAX_FIXED = 1
# OTHER_FIXED = 1

105
186

109
195

105
41

109
42

105
33

109
34

1,432
0.114

1,514
0.088

598
0.142

623
0.142

473
0.171

493
0.155

WEAK_TAXt
TAX_FIXEDt
WEAK_OTHERt
OTHER_FIXEDt

Observations
R-squared

Note: Table 11 displays results from the estimation of the OLS regression model (1) for firms which disclose any type of MW during the sample period (columns 1 & 2), for
firms which disclose a tax-related MW during the sample period (columns 3 & 4), and for firms which disclose and remediate a tax-related MW during the sample period
(columns 5 & 6). Primary variable of interest is TAX_FIXED. Robust t-statistics in parentheses. Standard errors are clustered by firm. All variables are defined in Appendix B.
*** p<0.01, ** p<0.05, *p<0.10 (two-tailed).

76
 

Table 12
OLS 1-year ETR Regressions (t+1,t+2,t+3)
Panel A: All tax function investments
(1)
(2)
(3)
VARIABLES
Cash1yrETR Cash1yrETR Cash1yrETR
(t+1)
(t+2)
(t+3)
0.048**
(2.009)
-0.048
(-1.625)
-0.024*
(-1.851)
0.033
(1.566)

0.027
(1.368)
-0.051**
(-1.957)
0.001
(0.051)
0.033
(1.549)

-0.013
(-0.730)
0.054*
(1.829)
-0.005
(-0.320)
-0.002
(-0.088)

Controls
Year Effects
Industry Effects

Yes
Yes
Yes

Yes
Yes
Yes

Yes
Yes
Yes

# Tax_Fixed = 1

105

105

105

7,687
0.058

7,687
0.047

7,687
0.044

WEAK_TAXt
TAX_FIXEDt
WEAK_OTHERt
OTHER_FIXEDt

Observations
R-squared

77
 

TABLE 12 (cont’d)
Panel B:External versus Internal Tax Function Investments
(1)
(2)
(3)
VARIABLES
Cash1yrETR Cash1yrETR Cash1yrETR
(t+1)
(t+2)
(t+3)
β1WEAK_TAXt

0.048**
(2.009)
-0.028
(-0.823)
-0.076**
(-2.041)
-0.024*
(-1.849)
0.033
(1.566)

0.027
(1.368)
-0.045
(-1.355)
-0.059*
(-1.902)
0.001
(0.051)
0.033
(1.548)

-0.013
(-0.730)
0.056
(1.411)
0.052
(1.267)
-0.005
(-0.319)
-0.002
(-0.088)

Controls
Year Effects
Industry Effects

Yes
Yes
Yes

Yes
Yes
Yes

Yes
Yes
Yes

F-test: β2 = β3
Prob > F =

1.40
0.236

0.13
0.716

0.01
0.942

# Tax_Fixed = 1

105

105

105

7,687
0.058

7,687
0.047

7,687
0.044

β2EXTERNAL_FIXt
β3INTERNAL_FIXt
β4WEAK_OTHERt
β5OTHER_FIXEDt

Observations
R-squared

Note: Table 12 Panels A and B displays results from the estimation of the OLS regression model (1) for
the full sample of firms where the dependent variable is the 1-year industry adjusted ETR measured at
time t+1(column 1), t+2(column 2), or t+3(column 3), respectively. In Panel B an f-test is also performed
to test for differential returns on external versus internal tax function investments (β2 = β3) ( Prob > F:
two-tailed).Robust t-statistics in parentheses. Standard errors are clustered by firm.

78
 

TABLE 13
OLS 3-year ETR Regressions-Unadjusted Tax Avoidance Mean Controls
(1)
(2)
VARIABLES
Future
Future
Unadjusted
Unadjusted
Cash3yrETR
GAAP3yrETR
(t+1 to t+3)
(t+1 to t+3)
0.064***
(2.905)
-0.070***
(-3.148)
-0.021
(-1.507)
0.013
(0.785)
0.236***
(4.551)
-0.040
(-1.232)
0.105
(0.808)
-0.004
(-0.498)
-1.566
(-1.571)
0.001
(0.253)
-0.019***
(-2.692)
-0.030***
(-3.158)
0.039***
(2.907)
-0.031***
(-2.908)
0.116**
(2.147)
0.157
(1.209)

0.070***
(3.537)
-0.040*
(-1.670)
-0.006
(-0.469)
0.030*
(1.687)
0.323***
(11.101)
-0.037**
(-2.239)
-0.221*
(-1.801)
-0.006
(-0.828)
0.882
(1.478)
0.002
(0.889)
0.011*
(1.784)
-0.001
(-0.262)
-0.002
(-0.215)
-0.041***
(-4.002)
0.008
(1.189)
0.113
(1.142)

Year Effects
Industry Effects

Yes
Yes

Yes
Yes

# TAX_FIXED = 1

107

112

7,845
0.120

8,387
0.148

WEAK_TAXt
TAX_FIXEDt
WEAK_OTHERt
OTHER_FIXEDt
AVGROAt to t+3
AVGLEVt to t+3
AVGFORINCt to t+3
TAXHAVENt
AVGINTANt to t+3
AVGSIZEt to t+3
NOLt
AVGRDt to t+3
AVGBMt-1to t+2
AGGR_LOSSt
AVGCONSTR_CFt to t+3
INTERCEPT

Observations
R-squared

Note: Table 13 displays results from the estimation of the OLS regression model (1) including unadjusted tax avoidance measures and
average control variables over the period in each subscript (e.g., t to t+3). The primary variable of interest is TAX_FIXED. Robust t-statistics
in parentheses. Standard errors are clustered by firm. All variables are defined in Appendix B. *** p<0.01, ** p<0.05, *p<0.10 (two-tailed).

79
 

REFERENCES

80
 

REFERENCES

Armstrong, C., Blouin, J., Larcker, D., 2012. The incentives for tax planning. Journal of
Accounting and Economics, 53(1): 391–411.
Ashbaugh-Skaife, H., Collins, D., Kinney, W., 2007. The discovery and reporting of internal
control deficiencies prior to SOX-mandated audits. Journal of Accounting and Economics,
44(1): 166–192.
Ashbaugh-Skaife, H., Collins, D., Kinney, W., LaFond, R. 2008. Internal control deficiencies,
remediation and accrual quality. The Accounting Review 83(1): 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. Journal of Accounting Research 47(1):
1-43.
Balakrishnan, K., Blouin, J., Guay, W., 2012. Does Tax Aggressiveness Reduce Financial
Reporting Transparency? Working Paper, University of Pennsylvania.
Barthélemy, J. & Quelin, B., 2006. Complexity of outsourcing contracts and ex post transaction
costs: an empirical investigation. Journal of Management Studies, 43(8): 1775–1797.
Bauer, A., 2013. Tax avoidance and the implications of weak internal controls. Working Paper,
University of Illinois.
Bedard, J., Hoitash, R., Hoitash, U., Westermann, K., 2011. Material weakness remediation and
earnings quality: A detailed examination by type of control deficiency. Working Paper,
Bentley University.
Bedard, J., Graham, L., 2011. Detection and severity classifications of Sarbanes-Oxley section
404 internal control deficiencies. The Accounting Review, 86(3): 825-855.
Beneish, M., Billings, M., Hodder, L., 2008. Internal control weaknesses and information
uncertainty. The Accounting Review, 83(3): 665–703.
Brealey, R., Myers, S., 2002. Corporate finance, McGraw-Hill/Irwin. Boston, MA.
Busby, J., Pitts, C., 1997. Real options in practice: an exploratory survey of how finance officers
deal with flexibility in capital appraisal. Management Accounting Research, 8(2): 169–186.
Campbell, D., Cook, T., 1979. Quasi-experimental Design, Houghton Mifflin. Boston, MA.
Caplan, D., Kirschenheiter, M., 2000. Outsourcing and audit risk for internal audit services.
Contemporary Accounting Research, 17(3): 387–428.
81
 

Chan, K., Farrell, B., Lee, P., 2008. Earnings management of firms reporting material internal
control weaknesses under Section 404 of the Sarbanes-Oxley Act. Auditing: A Journal of
Practice & Theory, 27(2): 161–179.
Chen, S., Chen, X., Cheng, Q., Shevlin, T. 2010. Are family firms more tax aggressive than nonfamily firms? Journal of Financial Economics, 95(1): 41-61.
Cheng, C., Hueang, H., Li, Y., Stanfield, J., 2012. The effect of hedge fund activism on
corporate tax avoidance. The Accounting Review, 87(5): 1493–1526.
Cheng, M., Dhaliwal, D., Zhang, Y., 2013. Disclosure of internal control over financial reporting
effectiveness and investment efficiency. Journal of Accounting and Economics, 56(1): 1-18.
Coase, R., 1937. The nature of the firm. Economica, 4(16): 386–405.
Contos, G., Guyton, J., Langeteig, P., Lerman, A., Nelson, S., 2012. Taxpayer compliance costs
for corporations and partnerships: A New Look. 2012 IRS-TPC Research Conference.
http://www.irs.gov/uac/SOI-Tax-Stats-2012-IRS-TPC-Research-Conference
Committee of Sponsoring Organizations of the Treadway Commission (COSO), 2011. Internal
Control – Integrated Framework. www.coso.org.
Cook, K., Huston, G., Omer, T., 2008. Earnings management through effective tax rates: The
effects of tax-planning investment and the Sarbanes-Oxley Act of 2002. Contemporary
Accounting Research, 25(2): 447–471.
Countryman, A., 2005. Compliance law changes urged. SEC weighs effect on smaller firms.
Chicago Tribune (January 3).
De Simone, L., Ege, M., Stomberg, B., 2013. Tax internal control quality: The role of auditorprovided tax services. Working Paper, Stanford University.
Dhaliwal, D., Hogan, C., Trezevant, R., Wilkins, M., 2011. Internal control disclosures,
monitoring, and the cost of debt. The Accounting Review, 86(4): 1131–1156.
Deloitte Practitioner Release. 2011. Material weaknesses and restatements. Is tax still in the hot
seat? http://www.deloitte.com/assets/DcomUnitedStates/Local%20Assets/Documents/Tax/us_tax_materialweaknesses_012011.pdf .
Donohoe, M., Knechel, W., 2013. Does corporate tax aggressiveness influence audit pricing?
Contemporary Accounting Research, forthcoming.
Doyle, J., Ge, W., McVay, S., 2007a. Accruals quality and internal control over financial
reporting. The Accounting Review, 82(2): 1141-1170.

82
 

Doyle, J., Ge, W., McVay, S., 2007b. Determinants of weaknesses in internal control over
financial reporting. Journal of Accounting and Economics, 44(1-2): 193–223.
Dyreng, S., Hanlon, M., Maydew, E., 2008. Long-run corporate tax avoidance. The Accounting
Review, 83(1): 61–82.
Dyreng, S., Lindsey, B., 2009. Using financial accounting data to examine the effect of foreign
operations located in tax havens and other countries on US multinational firms’ tax rates.
Journal of Accounting Research, 47(5): 1283–1316.
Espino-Rodríguez, T., Padrón-Robaina, V., 2006. A review of outsourcing from the resourcebased view of the firm. International Journal of Management Reviews, 8(1): 49–70.
Ettredge, M., 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.
Ettredge, M., Heintz, J., Li, C., Scholz, S., 2011. Auditor realignments accompanying
implementation of SOX 404 ICFR reporting requirements. Accounting Horizons, 25(1): 17–
39.
Feng, M., Li, C., McVay, S., 2009. Internal control and management guidance. Journal of
Accounting and Economics, 48(2-3): 190–209.
Feng, M., Li, C. McVay, S., Skaife, H. 2013. Ineffective internal control over financial reporting
and firm operations. Working Paper, University of Pittsburgh.
Ge, W., McVay, S., 2005. The disclosure of material weaknesses in internal control after the
Sarbanes-Oxley Act. Accounting Horizons, 19(3): 137-158.
Gleason, C., Pincus, M. Rego, S., 2013. Consequences of material weaknesses in tax-related
internal controls for financial reporting and earnings management. Working Paper,
University of Iowa.
Goh, B., 2009. Audit committees, boards of directors, and remediation of material weaknesses in
internal control. Contemporary Accounting Research, 26(2): 549–579.
Gompers, P., Ishii, J., Metrick, A., 2003. Corporate governance and equity prices. The Quarterly
Journal of Economics, 118(1): 107-55.
Graham, J., Hanlon, M., Shevlin, T., Shroff, N. 2013. Incentives for tax planning and avoidance:
evidence from the field, The Accounting Review, forthcoming.
Grant, R. 1991. The resource-based theory of competitive advantage: implications for strategy
formulation. California Management Review, 33(1): 114-135.

83
 

Gupta, S., Mills, L., 2002. Corporate multistate tax planning: benefits of multiple jurisdictions.
Journal of Accounting and Economics, 33(1): 117–139.
Gupta, S., Newberry, K., 1997. Determinants of the variability in corporate effective tax rates:
Evidence from longitudinal data. Journal of Accounting and Public Policy, 16(1): 1–34.
Hanlon, M., Heitzman, S., 2010. A review of tax research. Journal of Accounting and
Economics, 50(2): 127–178.
Hammersley, J., Myers, L., 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(1): 141–165.
Hammersley, J., Myers, L., Zhou, J., 2012. The failure to remediate previously-disclosed
material weaknesses in internal controls. Auditing: A Journal of Practice & Theory, 31(2):
73–111.
Hoag, M., Hollingsworth, C., 2011. An intertemporal analysis of audit fees and Section 404
material weaknesses. Auditing: A Journal of Practice & Theory, 30(2): 173–200.
Hogan, C., Wilkins, M., 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.
Hoitash, R., Hoitash, U., Johnstone, K., 2012. Internal control material weaknesses and CFO
compensation. Contemporary Accounting Research, 29(3): 768-803.
Huselid, M., 1995. The impact of human resource management practices on turnover,
productivity, and corporate financial performance. Academy of Management Journal, 38(3):
635–672.
Jeffrey, C., Lourens, C., 2008. Lessons from SOX can improve internal controls for private
companies. Practical Tax Strategies, 80(2): 100-105.
Johnstone, K., Li, C., Rupley, K., 2011. Changes in corporate governance associated with the
revelation of internal control material weaknesses and their subsequent remediation.
Contemporary Accounting Research, 28(1): 331–383.
Jonas, G. 2005. Section 404 report on internal control: Our early experience. New York, NY:
Moody’s Investors Service, Global Credit Research.
Khurana, I., Moser, W. 2013. Shareholder investment horizons and tax aggressiveness. Working
Paper, University of Missouri.
Krishnan, J., 2005., Audit committee quality and internal control: An empirical analysis. The
Accounting Review, 80(2): 49–675.
84
 

Kubick, T., Mayberry, M., Lynch, D., Omer, T., 2014. Product market power and tax avoidance:
market leaders, mimicking strategies, and stock returns. Working Paper, University of
Wisconsin-Madison.
Laios, L., Moschuris, S., 1999. An empirical investigation of outsourcing decisions. Journal of
Supply Chain Management, 35(1): 33–41.
Larcker, D., Richardson, S., 2004. Fees paid to audit firms, accrual choices, and corporate
governance. Journal of Accounting Research, 42(3): 625–658.
Li, C., Sun, L., Ettredge, M., 2010. Financial executive qualifications, financial executive
turnover, and adverse SOX 404 opinions. Journal of Accounting and Economics, 50(1): 93–
110.
Mayberry, M., Mcguire, S., Omer, T. 2013. Smoothness and the information content of taxable
income. Working Paper. University of Florida.
McGuire, S., Neuman, S., Omer, T., 2011. Through the looking glass: Are sustainable tax
strategies reflected in earnings persistence. Working Paper, Texas A&M University.
Mills, L., Erickson, M., Maydew, E., 1998. Investments in tax planning. Journal of the American
Taxation Association, 20(1): 1–20.
Niemeier, C., 2004. The Value of Integrity. Speech Delivered at the AICPA National Conference
on Current SEC and PCAOB Developments. Washington, D.C. (December 7). Available at:
http://pcaobus.org/News/Speech/Pages/12072004_NiemeierValueIntegrity.aspx
Omer, T., Bedard, J., Falsetta, D., 2006. Auditor-provided tax services: The effects of a changing
regulatory environment. The Accounting Review, 81(5): 1095–1117.
Pickard, G., 2005. In the public interest, part one: A conversation with the chief accountant of
the SEC. Journal of Accountancy, 199 (1). Available at:
http://www.journalofaccountancy.com/Issues/2005/Jan/InThePublicInterestPartOne.htm
Quélin, B. & Duhamel, F., 2003. Bringing together strategic outsourcing and corporate strategy:
Outsourcing motives and risks. European Management Journal, 21(5): 647–661.
Rego, S., 2003. Tax-avoidance activities of U.S. multinational corporations. Contemporary
Accounting Research, 20(4): 805-833.
Rego, S., Wilson, R., 2012. Equity risk incentives and corporate tax aggressiveness. Journal of
Accounting Research, 50(3): 75–810.

85
 

Skaife, H., Veenman, D.,Wangerin, D., 2013. Internal control over financial reporting and
managerial rent extraction: Evidence from the profitability of insider trading. Journal of
Accounting and Economics, 55(1): 91–110.
Tadelis, S., 2002. Complexity, flexibility, and the make-or-buy decision. The American
Economic Review, 92(2): 433–437.
Widener, S., Selto, F., 1999. Management control systems and boundaries of the firm: why do
firms outsource internal auditing activities? Journal of Management Accounting Research,
11(1): 45–73.
Williamson, O., 1975. Markets and hierarchies. The Free Press, New York, NY.
 
 

 

86