Does a Firm’s Business Strategy Influence its Level of Tax Avoidance?
Danielle M. Higgins
University of Connecticut
Thomas C. Omer
Texas A&M University
John D. Phillips*
University of Connecticut
February 14, 2011
We appreciate helpful comments received from Amy Dunbar, Dave Weber, participants
at the American Accounting Association Northeast Regional Meeting, the discussant,
Michaele Morrow, and workshop participants at the University of Connecticut. Danielle
Higgins and John Phillips appreciate the research support received from the University of
Connecticut School of Business. Tom Omer appreciates the research support received
from Ernst & Young.
Electronic copy available at: http://ssrn.com/abstract=1761990
Does a Firm’s Business Strategy Influence its Level of Tax Avoidance?
Abstract: A firm’s business-level strategy dictates how the firm competes in its chosen
line of business. Porter (1996) argues that the best way for a firm to achieve a sustainable
competitive advantage in its chosen market is to reinforce its business-level strategy with
a host of activities, including functional policies, organization structure, etc. The choices
made by each firm are likely to determine, to some extent, the level of tax avoidance
because the firm strategies are, in part, based on firms’ willingness to deal with risk and
uncertainty. Thus, we examine whether a firm’s business strategy, as defined by the
Miles and Snow typology (1978), influences its level of tax avoidance. We find that firms
engaging in a strategy that focuses on minimizing and reducing the uncertainty of costs
(defenders) avoid fewer taxes than firms that not only follow a strategy focusing on
product differentiation and the aggressive pursuit of opportunities, but are more willing to
deal with uncertainty (prospectors). Our results suggest that even though defenders could
engage in tax avoidance activities to lower costs, prospectors are more willing to engage
in tax planning efforts that have uncertain outcomes.
Electronic copy available at: http://ssrn.com/abstract=1761990
Does a Firm’s Business Strategy Influence its Level of Tax Avoidance?
1. Introduction
A firm’s business-level strategy dictates how the firm will compete in its chosen
market.1 Porter (1996) argues that the best way for a firm to achieve a sustainable
competitive advantage in its chosen market is to reinforce its business-level strategy with
a host of activities, including functional policies, organization structure, etc. The choices
made by each firm are likely to determine, to some extent, the level of tax avoidance
because the firm strategies are, in part, based on firms’ willingness to deal with risk and
uncertainty. Thus, we investigate whether a firm’s business-level strategy influences it
tax planning activities (i.e., tax avoidance).2
Investigating whether a firm’s business strategy influences its level of tax avoidance
is interesting for at least two reasons. First, prior literature finds that there is substantial
variation in firms’ propensities to avoid income taxes (Dyreng et al. 2008). This variation
not only remains largely unexplained but also may be highly idiosyncratic and
determined by a number of factors and their interactions (Hanlon and Heitzman 2010).
Prior literature has documented an association between tax avoidance and several firmlevel characteristics (i.e., research and development activities, growth, capital intensity,
etc). The theoretical work of Miles and Snow (1978) identifies several of these firmlevel characteristics as reflective indicators of a firm’s business strategy. Therefore,
examining the association between tax avoidance and a combination of these firm-level
characteristics, reflective of a firm’s overall business strategy, should lead to a more
comprehensive understanding of firms’ tax avoidance activities.
Second, it remains unclear in the literature why seemingly similar firms engage in
varying degrees of tax avoidance. For example, firms within the same industry and with
similar incentives (and potentially similar opportunities) to engage in tax planning,
1
Business-level strategy is not the same as corporate strategy. Corporate strategy is concerned with the
overall purpose and scope of the business (Hambrick 1983) whereas business-level strategy is concerned
with how the firm competes in its chosen field of business.
2
Following Hanlon and Heitzman (2010), we view tax avoidance as a continuum that ranges from clearly
legal (e.g. investments in municipal bonds) to those of questionable legality (e.g. tax shelters). Thus, tax
avoidance refers to any activity that reduces a firm’s explicit tax liability relative to its pretax income.
1
Electronic copy available at: http://ssrn.com/abstract=1761990
exhibit substantial variation in their efforts to avoid income taxes. Snow and Hrebiniak
(1980) and Hambrick (1983) document that business strategies can vary within a single
industry. Therefore, if there is a relation between firms’ business strategy and tax
avoidance, it may explain the intra-industry variation in firms’ tax avoidance activities.
Although several influential business strategy typologies exist, (e.g., Abell 1980;
Porter 1980), we use the Miles and Snow (1978) typology to define firms’ business
strategies because of the advantages it possesses over other typologies.3 Specifically,
employing the Miles and Snow (1978) typology is advantageous because of its detailed
theoretical orientation and its generalizability across industry settings. Miles and Snow
(1978) contend that four basic patterns or strategies emerge as firms attempt to best adapt
to their competitive environment, and that three of these strategies may be fitted on a
continuum. At one end of the continuum are defenders, which are firms that follow a
cost leadership strategy. Defenders are firms that tend to have a narrow product domain, a
focus on efficiency and avoiding uncertainty, and a stable organizational structure. At
the opposite end of the spectrum are prospectors, which are firms that follow a
differentiation strategy.4 Prospectors have a very broad product domain, a focus on
innovation and change, a more flexible organization structure, and less emphasis on
avoiding uncertainty. Although Miles and Snow (1978) do not consider the tax avoidance
behavior of their topology of firms, the attributes of these groups may lend themselves to
an increased or decreased emphasis on tax planning.
To determine the extent of their tax avoidance activities, firms trade off the marginal
benefits against the marginal costs of managing tax expense. Tax avoidance activities not
only increase cash flow, but increase accounting earnings in certain circumstances (e.g.,
tax credits and permanent book-tax differences). Tax avoidance costs include planning
and implementation expenses, potential penalties imposed by tax authorities and
3
For example, Porter’s (1980) strategy has been criticized as for its lack of detailed theoretical orientation
and its inability to generalize to large firms, while Abell’s (1980) typology has been criticized for its
inability to differentiate among firms that follow cost leadership strategies, firms that fall in the middle of
the strategy spectrum and firms lacking any apparent strategy (Chrisman et al. 1988, Smith et. al. 1989).
4
Miles and Snow (1978) also identify two additional strategies, analyzers and reactors. Analyzers have
attributes of both defenders and prospectors and fall somewhere in between these two strategies on the
continuum. Reactors do not follow any consistent strategy but rather only respond to environmental change
(Miles and Snow 1978).
2
reputation costs if the firm’s tax avoidance activities are deemed to be aggressive and
subject to public scrutiny. The difference between defender and prospector firms’ tax
avoidance levels depends on the impact of the differential characteristics of these two
strategies on the benefits and costs of tax avoidance. For example, because defenders are
interested in minimizing costs, defender firms may benefit more from the increased tax
savings generated by tax avoidance activities than prospector firms. Yet, at the same
time, the cost of devising and implementing tax planning strategies can be expensive and
thus, for cost conscious defender firms, the benefit of additional tax savings may not
outweigh the costs of implementing the tax avoidance strategies. In addition, defenders’
emphasis on stability and their need to reduce uncertainty suggests that these firms could
be more concerned with 1) the uncertainty associated with tax avoidance that makes cost
minimization more difficult, 2) the potential penalties imposed by tax authorities and 3)
the reputation damage associated with public disclosure of involvement with particularly
egregious avoidance activities.
Relative to defenders, prospectors are less concerned with minimizing costs, and
more focused on growth and innovation, possibly making tax minimization less of a
concern for these types of firms. Nevertheless, prospectors are known for aggressively
pursuing new opportunities and it is possible that this aggressive culture influences their
level of tax avoidance. In addition, prospectors tend to have a greater propensity for risk,
and thus may be more willing to engage in more aggressive tax positions, relative to
defenders. Thus, because it is unclear whether defenders engage in more or less tax
avoidance activities than prospectors, we investigate this issue empirically.
To address our research question we begin by creating a measure (STRATEGY)
designed to provide a score that reflects the most likely business strategy adopted by
sample firms. The score is based on variables identified in prior literature as
characteristics of firms’ business strategies. Because Miles and Snow (1978) suggests
that competitive strategy can be classified as a continuum between two different strategic
orientations, with prospectors at one end and defenders at the other end, we identify
defenders as those firms having the lowest value and prospectors as those having the
highest value of our STRATEGY measure. All other firms are classified as analyzers.
3
Analyzers have attributes of both prospectors and defenders and fall somewhere in
between these two strategies on the continuum.
To examine firms’ tax avoidance activities, we use two different effective tax rate
measures drawn from prior literature. Our first measure, the book effective tax rate
defined as total tax expense divided by pre-tax book income (adjusted for special items),
captures tax avoidance activities that directly affect net income through total tax expense.
Our second measure, the cash effective tax rate, defined as cash taxes paid divided by
pre-tax book income (adjusted for special items), reflects the level of cash taxes paid to
tax authorities. Lower effective tax rates reflect increased tax avoidance.
Using data from 1998 to 2008, we regress each of our measures of tax avoidance
on our continuous measure, STRATEGY, and control for other factors that prior literature
suggests are associated with tax avoidance. We find that defender firms engage in less tax
avoidance than prospector firms, as demonstrated by higher book and cash effective tax
rates. These results suggest that defender firms’ concern with the costs and uncertainties
associated with tax avoidance outweighs the benefit of additional tax savings.
Our study makes several contributions to the literature. First, our results
contribute to the stream of research that investigates the variation in firms’ tax avoidance
activities. Prior research has documented an association between tax avoidance and firmlevel characteristics, executive characteristics, and managerial incentives (e.g., Dyreng et
al. 2009; Phillips 2003; Rego 2003; Rego and Wilson 2009). We extend this line of
research with evidence that a firm’s business strategy influences its level of tax
avoidance. In doing so, we respond to Hanlon and Heitzman’s (2010) conjecture that tax
avoidance is most likely explained by a number of factors and interactions by providing
evidence that a firm’s strategy, which represents a broader view of the firm and is
identified based on a range of determinants, is a determinant of level of tax avoidance.
Furthermore, our results contribute toward a better understanding of the impact of
different competitive strategies on firms’ tax planning and reporting practices.
Our study also links two streams of literature: organizational theory from the
management literature and tax avoidance from the accounting literature. Our study
contributes to these two streams of literature in several ways. First, we develop a proxy of
organizational business strategy that is constructed using publicly available data, and is
4
more comprehensive than existing proxies in accounting5 or management. Thus, our
measure can be used in future research in both management and accounting settings.6
Second, we contribute to the management literature by providing evidence that firms’ tax
planning activities are associated with their organizational strategy. Thus, our results
empirically support Porter’s (1996) argument that firms should reinforce their businesslevel strategy with support activities in order sustain a competitive advantage.
The remainder of the paper is organized as follows: Section two discusses prior
research and develops the hypotheses, Section three presents the research design, Section
four explains the sample selection process and provides descriptive statistics, Section five
provides the results of the main analyses, and Section 6 presents our sensitivity analysis
and Section 7 concludes.
2. Prior Literature and Hypothesis Development
2.1 Prior Literature on the Determinants of Tax Avoidance
Prior tax avoidance studies generally focus on firm-level characteristics, including
but not limited to an association between tax avoidance and firm-level characteristics
such as size, the magnitude of foreign operations, capital intensity, leverage, and research
and development expense (e.g. Armstrong et al. 2009; Dyreng et al. 2009; Phillips 2003;
Rego 2003). 7 In addition, several studies provide evidence of differing investments in tax
planning to lower effective tax rates. For example, Mills et al. (1998) provide evidence
that firms’ tax planning expenditures are associated with lower effective tax rates.8 Cook
et al. (2008) provide evidence that the magnitude of tax fees paid to the external auditor
5
The exception is a concurrent working paper (Bentley, Omer and Sharp 2010) that tests whether certain
business strategies are associated with more financial reporting irregularities. They use similar measures to
construct their measure of firms’ business strategies. They find that prospector firms are more likely to
experience financial reporting irregularities. Thus, given the findings in Frank et al. (2009), their results are
complementary to our results.
6
Thus far, the accounting literature has examined organizational strategy as a determinant of compensation
(Ittner et al. 1997), the outsourcing of the corporate tax function (Dunbar and Phillips 2001), and audit fees
and accounting irregularities (Bentley, Omer and Sharp 2010).
7
See Zimmerman (1983), Gupta and Newberry (1997), Mills et al. (1998), Rego (2003), Stickney and
McGee (1982), and Shevlin and Porter (1992). In addition, see Callihan (1994) for a review of the annual
effective tax rate literature.
8
Mills et al. measures ETR as the ratio of current tax expense to pre-tax income
5
is associated with greater reductions in fourth quarter effective tax rates. Furthermore,
McGuire et al. (2010) find that firms that engage auditors with industry expertise
continuously focus more on tax avoidance that lowers book effective tax rates.
In addition to firm-level characteristics, prior studies also examine whether
managerial incentives are associated with tax avoidance. After controlling for selection
bias, Phillips (2003) finds that business-unit manager, but not chief executive officer
(CEO), after-tax bonus compensation is associated with lower effective tax rates. Rego
and Wilson (2008) finds a positive relation between tax avoidance and CEO and chief
financial officer (CFO) compensation. Meanwhile, Armstrong et al. (2010) find that tax
directors’ incentive compensation is related to lower ETRs, but not cash ETRs suggesting
that tax directors focus on tax strategies that lower income tax expense and thus increase
earnings. Finally, Robinson et al. (2010) find that firms organizing their tax departments
as profit centers have lower book ETRs, but not lower cash ETRs.
Although prior studies have linked tax avoidance to firm-level characteristics and
managerial incentives, the literature is still unable to explain much of the variation in
measures of tax avoidance. For example, it remains unclear why firms within the same
industry, with seemingly similar incentives and opportunities, engage in different levels
of tax avoidance. Hanlon and Heitzman (2010) suggest that tax avoidance may be highly
idiosyncratic and determined by a number of factors and their interactions. Accordingly,
prior literature has documented an association between tax avoidance and several firmlevel characteristics (i.e., research and development activities, growth, capital intensity,
etc). The theoretical work of Miles and Snow (1978) identifies several of these firmlevel characteristics as reflective indicators of a firm’s business strategy. Examining the
association between tax avoidance and a combination of these firm-level characteristics
should therefore lead to a more comprehensive understanding of firms’ tax avoidance
activities. Therefore, we provide evidence on whether business-level strategy influences
firms’ tax avoidance activities.
2.2 Business Strategy and Tax Avoidance
Within the strategic management literature, the Miles and Snow (1978) typology
is one of the most popular and well cited theories of strategic types. Miles and Snow
6
(1978) argue that different competitive strategies arise from the way companies decide to
address three fundamental problems: entrepreneurial, engineering, and administrative
problems.9 The entrepreneurial problem relates to how a firm should manage its market
share. The engineering problem involves how a company should implement its solution
to the entrepreneurial problem. The administrative problem considers how a company
should structure itself to manage the implementation of the solutions to the first two
problems. Miles and Snow (1978) contend that four basic patterns or strategies emerge as
firms attempt to solve these problems: prospector, defender, analyzer, and reactor
strategies. Of these four strategies, three are viable and can be fitted on a continuum.10 At
one end of the continuum are defenders. Defenders are firms that stress cost efficiency
and certainty as the basis of competition. Defenders tend to have a narrow product
domain, a focus on efficiency, and a stable organizational structure. At the opposite end
of the continuum are prospectors. Prospectors have a very broad product domain, a focus
on innovation and change, and more flexible organization structure. Analyzers are an
intermediate type, and tend to exhibit traits of both prospectors and defenders. See
Appendix A for a detailed description of the key features of each strategic type.
Several studies have examined the association between business strategies and
certain firm characteristics (e.g., Snow and Hrebiniak 1980; Ittner et al. 1997; Simons
1987). Hambrick (1983) classifies firms as either prospectors or defenders using productmarket data and examines the relation between organizational strategy and firm
performance. He finds that defenders outperformed prospectors in terms of current
profitability and cash flows. Ittner et al. (1997) examine whether firms place more weight
on non-financial measures in chief executive officer (CEO) bonus contracts when they
follow a prospector strategy. They construct a measure of firm strategy and use this
measure to classify firms as either prospectors or defenders. They find that CEOs of
9
Miles and Snow (1978) refer to the process by which firms align themselves with their environment as the
adaptive cycle
10
Miles and Snow (1978) identify a fourth group of firms, reactors. These firms do not follow any
consistent strategy but rather only respond to environmental change and thus are expected to change over
time. Miles and Snow (1978) note that this strategy is not viable in the long-run. Organizations are often
forced into this strategy when top management is unwilling (or unable) to develop distinctive
competencies, organizational structures or management processes required by a particular strategy.
7
prospector firms are compensated more on the basis of non-financial measures than
defender firm CEOs.
Phillips and Dunbar (2001) is the only study that examines the association
between business strategy and corporate tax planning. The authors classify firms as either
defenders (nongrowth firms) or prospectors (growth firms), and examine the relation
between these two strategies and the outsourcing of the corporate tax function activities.
The authors argue that prospectors (growth firms) focus less on minimizing income tax
expense, and thus will outsource more of their tax-planning and -compliance activities.
Their results are consistent with this hypothesis. However, this study does not explicitly
address whether prospectors engage in more or less tax avoidance than defender firms.
Furthermore, Phillips and Dunbar classify firms as prospectors and defenders based on
single variable (i.e.,, growth). Prior theoretical and empirical research in management
provides evidence that strategy is multidimensional, however, and should not be defined
on the basis of a single variable (e.g. Miles and Snow 1978, Hambrick 1983).
Although Miles and Snow (1978) do not consider the tax avoidance behavior of
their topology of firms, the attributes of these groups should lead to an increased or
decreased emphasis on tax avoidance. Thus, we examine whether the business strategies
defined by the Miles and Snow topology are associated with more or less tax avoidance,
which depends on the impact of the differential characteristics of these two strategies on
the benefits and costs of tax avoidance. The benefits of tax avoidance include greater tax
savings, while the costs include planning and implementation costs, potential penalties
imposed by the IRS and reputation costs (Chen et al. 2010).
On one hand, the Miles and Snow (1978)’s topology suggests that the potential
benefits of tax avoidance are greater for defender firms than prospector firms because
defender firms emphasize cost efficiency as the cornerstone of competitive advantage
whereas prospectors are focused more on innovation and growth and less on cost
minimization. Therefore, because defenders are primarily interested in minimizing costs,
and because income tax expense is a major cost for most firms, firms following a
defender strategy should benefit more from the increased tax savings generated by
increased tax avoidance activities than firms following a prospector strategy.
8
Alternatively, although defenders should benefit more from tax planning than
prospectors, the costs and uncertainty associated with tax avoidance may be greater than
the tax benefits for defender firms. Although tax planning strategies may create
additional tax savings, the cost of devising and implementing these strategies can be
extensive. For example, Mills et al. (1998) contend that while all firms may engage in
some baseline amount of planning to comply with tax laws, some firms may find that the
cost of additional investments in tax planning outweigh the benefits. Thus, for cost
conscious defender firms, the benefit of additional tax savings may not outweigh the cost
of implementation.
Moreover, another important cost of tax avoidance is due to uncertainty regarding
the long-term success of sustaining aggressive tax strategies and the potential penalties
imposed by tax authorities. This cost is more likely to have a substantial effect on
defender firms’ competitive position relative to prospectors. The structure and processes
of the defender organization are all designed to reduce uncertainty within the
organization. Therefore, the inherent uncertainty associated with sustainability, an audit
by a tax authority and potential penalties threatens defender firms’ need for stability and
predictability (Miles and Snow 1978). Specifically, these penalties create an additional
and potentially significant unexpected cost for defender firms, which could threaten their
competitive position because cost efficiency is an important prerequisite for their success.
Alternatively, the prospector strategy is more conducive to uncertainty and thus taking on
more risk. Because these firms often aggressively pursue new opportunities before
detailed planning is completed, outcome certainty is not central to their competitive
strategy. Thus, for these firms, engaging in particularly risky tax avoidance strategies
would not, based on the Miles and Snow description of these firms, appear to threaten
their competitive position as significantly as defender firms.
Finally, another potential non-tax cost for defender firms is the reputation damage
from being associated with a particular egregious tax avoidance strategy. Defender firms’
operate in very narrow product domains, offer limited products (relative to prospectors),
and compete through cost leadership and quality (Miles and Snow 1978). Because these
firms compete on the basis of cost and not innovation, substitutes most likely exist for
these firms’ products or services. Thus, the negative publicity associated with being a
9
“bad” corporate citizen could impose a greater cost on defender firms’ competitive
position because customers can easily find substitutes for their product or services.
Alternatively, because prospector firms generate more unique products, it may be
difficult for consumers of such products to find an appropriate substitute.
Overall, both the benefits and costs of tax avoidance appear to be greater for
defender firms than for prospector firms. Thus, it is unclear whether defenders will
exhibit higher or lower levels of tax avoidance relative to prospectors. Based on these
arguments, we state the following null hypothesis:
H1: Organizational business strategies are not associated with firms’ levels of tax
avoidance
3. Research Methodology
3.1 Measure of Strategy
To categorize firms into their business strategy groups, we create a measure,
STRATEGY, designed to provide a score that reflects the most likely business strategy
adopted by sample firms. To construct the STRATEGY score, we follow prior literature
and identify variables which reflect different facets of this unobserved construct. We first
follow Ittner et al. (1997) and include the following variables: (1) the ratio of research
and development to sales, (2) the ratio of employees to sales and (3) the market-to-book
ratio.11, 12 We then expand upon this construct to include the following additional
measures based on the theoretical work of Miles and Snow (1978) and the empirical work
of Hambrick (1983): (4) the ratio of advertising expenses to total sales and (5) fixed asset
intensity. Consistent with Ittner et al. (1997), all variables are computed using a rolling
average of the respective yearly ratios over the prior five years.
11
We also consider an alternative measure of growth in order to capture the sporadic growth of prospectors
and the steady growth of defenders. To capture the rate of growth, we compute the coefficient of variation
on the change in total assets. We use this measure in place of market-to-book in the STRATEGY score. Our
results are robust to this measure.
12
Ittner et al. (1997) use a fourth measure, the number of new products and services, which requires access
to a proprietary database. Our goal in creating the STRATEGY score is to create a measure the can be easily
reproduced using publicly available data.
10
Each of the five measures is intended to capture different elements of a firm’s
organizational strategy. The ratio of research and development to sales (RD/SALES5)
serves as a measure of a firm’s propensity to seek new products. Because prospectors
engage in greater amounts of innovative activity, they are expected to have higher
research and development costs than defenders (Hambrick, 1983). The ratio of employees
to sales (EMP/SALE5) is a proxy for a firm’s ability to produce and distribute its goods
and services efficiently (Thomas et al. 1991). Because defenders focus on organizational
efficiency, we expect firms following this strategy to have fewer employees per dollar of
sales. The ratio of market value to book value (MtoB5) is a proxy for a firm’s growth or
investment opportunities (e.g., Smith and Watts 1992; Bushman et. al 1996). Consistent
with Ittner et al. (1997), we expect prospectors to have greater growth opportunities than
defenders. We include the ratio of advertising expense to total sales (MARKET5) as a
proxy for firms’ emphasis on marketing and sales.13 Prospectors spend more time
motivating, educating and informing their customers, thus we expect them to have higher
advertising expense than defenders. Furthermore, Hambrick (1983) finds that prospectors
have higher marketing expenses relative to defenders. Last, we include a measure of
fixed asset intensity (PPEINTENSITY5), measured as the ratio of gross property, plant
and equipment to lagged total assets. This measure is designed to capture a firms’ focus
on production assets, and thus higher ratios would be representative of defender firms
(Hambrick, 1983).
Next, we rank each of the five variables by forming quintiles within each 2-digit
SIC industry-year.14 For each of the first four variables (RD/SALE5, EMP/SALE5,
MTOB5, MARKET5), firms in the top quintile receive a score of 5, those in the next
quintile receive a score of 4, etc. The scoring for PPEINTENSITY5 is inverted because
defenders are expected to have greater fixed asset intensity. Therefore, firms in the top
PPEINTENSITY5 quintile receive a score of 1, those in the next quintile receive a score
of 2, etc.15 Then, for each firm-year, we sum the scores across the five variables such that
13
Following Dyreng et al. (2008, 2009), we replace advertising expense with zero if it is missing.
Hambrick (1983) notes that strategy is a relative phenomenon and should be defined within industry.
15
In constructing the STRATEGY variable, we assume equal weighting of each of the five variables. This
methodology sacrifices precision for the simplicity necessary to build a strategy index, and thus is similar
to the approach taken by Gompers et al. (2003) in constructing their governance index.
14
11
the maximum STRATEGY score a firm could receive is 25, and the minimum score a firm
could receive is 5. Higher STRATEGY scores represent prospector firms, while lower
STRATEGY scores represent defender firms.
3.2 Measures of Tax Avoidance
The tax avoidance literature has developed several proxies of tax avoidance.16 To
proxy for firms’ tax avoidance activities, we estimate firms’ book and cash effective tax
rates. We use two different proxies of tax avoidance because the proxies, although
correlated, capture different types of tax avoidance.
Our first measure of tax avoidance is the book effective tax rate (BOOK_ETR).
Following Dyreng et al. (2009), we define BOOK_ETR as total tax expense divided by
pre-tax book income (adjusted for special items).17 The ETR is a commonly used
measure of a firm’s tax burden (e.g. Phillips 2003; Rego 2003; Rego and Wilson 2009;
Robinson et al. 2010). It is designed to capture avoidance activities that directly affect net
income via the tax expense.18 Prior literature suggests that a lower ETR reflects higher
incidences of tax avoidance (e.g., Gupta and Newberry 1992; Rego 2003).
Our second measure of tax avoidance is the cash effective tax rate (CASH_ETR).
Following Dyreng et al. (2008, 2009), we define CASH_ETR as cash taxes paid divided
by pre-tax book income (adjusted for special items). The cash ETR captures all tax
avoidance activities that reduce cash taxes paid to the authorities. Therefore, the cash
ETR reflects avoidance activities that defer income taxes to later periods (i.e., activities
that create temporary differences and do not affect net income), as well as those that
avoid tax entirely (i.e., activities that create permanent differences and affect net income).
This measure’s numerator excludes changes in the tax contingency reserve and the
deferred tax asset valuation allowance, which affect current and deferred tax expense,
16
Hanlon and Heitzman (2010) identify 12 proxies of tax avoidance.
Dyreng et al. (2008, 2009) adjust pretax income for special items because these items can be quite large
in magnitude and introduce volatility in one-year ETR measures.
18
The book ETR has several limitations. In addition to tax avoidance activities, the book ETR also reflects
non-avoidance activities such as changes in the valuation allowance and changes in tax reserves.
Furthermore, the numerator is total tax expense, which represents the sum of current and deferred taxes.
Deferred taxes represent future tax payments or refunds that are associated with temporary book-tax
differences. Therefore, tax avoidance activities that generate temporary differences are not reflected in ETR
(Hanlon and Heitzman 2010).
17
12
respectfully. Cash payments relating to prior years, however, are included in CASH_ETR.
Dyreng et al. (2008) suggest that lower values of CASH_ETR represent higher levels of
tax avoidance.
Several measurement issues arise when using ETRs to measure tax avoidance.
One particular measurement issue regarding ETRs concerns firms with negative pretax
income or tax refunds (i.e., negative tax expense). Although we retain loss firms and
firms with negative tax expense in our sample, their ETRs are distorted and, thus,
difficult to interpret.19 To address this problem, we follow Gupta and Newberry (1997)
and set the ETR: (1) to zero for firms with negative tax expense, and (2) to 100 percent
for firms with positive tax expense and negative (or zero) pretax income.20 Another
measurement issue concerns the effect of relatively small values in the denominator of
the ETR, thus giving rise to unreasonably large ETRs. To address this problem, we
follow Gupta and Newberry (1997) and constrain ETRs to lie between zero and 100
percent.
3.3 Empirical Model
To examine the association between tax avoidance and organizational strategy,
we estimate the following model using ordinary least squares (OLS):
,
,
,
,
,
,
&
,
,
,
,
,
,
,
Where:
ETR
= BOOK_ETR or CASH_ ETR as defined above;
19
For example, the ETR of a firm with negative tax expense and negative pretax income will be positive,
even though the firm paid no taxes. Similarly, the ETR of a firm with positive tax expense and negative
pretax income will be negative, even though the firm paid taxes.
20
There is no clear solution for dealing with the measurement issues created by the inclusion of these firms.
Therefore, we perform a sensitivity analysis (discussed later) where we exclude loss firms. The results are
qualitatively similar.
13
STRATEGY
= Continuous strategy score, ranging from 5 (defender-type
firm) to 25 (prospector-type firm);
SIZE
= natural log of total assets (Compustat AT) for year t;
ROA
= the ratio of pre-tax income (Compustat PI) for year t to total
assets (Compustat AT) at the beginning of the year;
DEBT
= Long term debt (Compustat DLCC) for year t scaled by total
assets (Compustat AT) at the beginning of the year;
FOREIGN
= Indicator variable equal to 1 if a firm had positive pretax
foreign income (Compustat PIFO) in year t, and 0 otherwise;
INTAN
= Intangibles (Compustat INTAN) for year t scaled by total
assets (Compustat AT) at the beginning of the year;
R&D
= Total research and development expense (Compustat XRD)
in year t, scaled by beginning of the year total assets
(Compustat AT);
PPEINTENISTY = Gross property, plant and equiptment (Compustat PPEGT) in
year t, scaled by beginning of the year total assets (Compustat
AT) in year t;
NOL
= indicator variable equal to one if there was a tax-loss
carryforward (Compustat TLCF) as of the end of year t, zero
otherwise;
We estimate equation (1) for each measure of tax avoidance (BOOK_ETR and
CASH_ETR). Firms that engage in more tax avoidance will have lower book ETRs
and/or cash ETRs. A negative coefficient on STRATEGY would suggest that prospector
firms engage in more tax avoidance than defender firms, while a positive coefficient
would suggest than defender firms engage in more tax avoidance than prospectors.
In addition to our variable of interest, we also include several control variables
drawn from the effective tax rate literature. We include a variable for firm size (SIZE)
because prior literature documents that it is an important determinant of ETRs (Gupta and
Newberry 1997; Mills et al. 1998; Rego 2003). Prior research finds mixed results
between size and ETRs, and thus, we do not predict a sign for , the coefficient on SIZE.
14
We include a control for the presence of foreign operations (FOREIGN) because
multinational firms may have more opportunities to avoid income taxes than domestic
firms (Mills et al. 1998; Rego 2003). Multinational firms also face taxation on profits in
multiple jurisdictions or may have foreign tax credit limitations (Markle and Shackelford
2009). Therefore, because prior research finds mixed results between the presence of
foreign operations and ETRs, we do not predict a sign for , the coefficient on
FOREIGN. We include a firms’ financial leverage (DEBT) as a control for the
complexity of firms’ financial transactions because firms have the ability to minimize
taxes through financing transactions (Mills et al. 1998). Yet, firms with higher leverage
may have less need for other non-debt tax shields, and thus engage in less tax planning to
reduce taxes (Graham and Tucker, 1996). Because prior research finds mixed results
between debt and the ETR, we do not predict a sign for
, the coefficient on DEBT. In
addition, we also control for firm profitability (ROA) and the presence of a net operating
loss (NOL) to proxy for a firm’s need to avoid taxes (Rego 2003; Wilson 2009).
Consistent with Chen et al. (2010), we control for firms’ levels of intangible
assets (INTAN) due to the differing book and tax treatments of intangible assets.
However, opportunities to shift income could also be represented by INTAN (Grubert and
Slemrod, 1998; Hanlon et al. 2007). Thus, we do not predict a sign for , the coefficient
on INTAN. Finally, we include industry and year fixed effects because prior research
documents that tax avoidance varies by industry and year (Rego, 2003). All variables are
defined in Appendix C.
Prior research has documented an association between R&D, capital intensity and
tax avoidance (e.g. Mills et al. 1998), and therefore, we add two additional controls to
equation (1), the ratio of property, plant, and equipment to total assets (PPEINTENSITY)
and total research and development expense divided by beginning of the year total assets
(R&D). Although we use these two variables to construct our STRATEGY measure, we
include them as control variables in the main model to alleviate the concern that our
results may be driven by firms with high R&D expense or firms with significant property,
plant and equipment (PPE).
Last, in estimating equation (1), endogeneity is unlikely to be of critical concern
in our setting, as it is unlikely that a firm’s level of tax avoidance prompts the firm to
15
follow one strategy over the other. However, assuming that a firm’s strategy is not
entirely preordained by its operating environment and is partially shaped by management,
there is an issue of why some firms follow one strategy over the other.21 This self
selection bias introduces a potential omitted correlated variable problem as some of the
determinants may affect a firm’s level of tax avoidance. To address this problem, we
include a comprehensive list of controls in our empirical model.
4. Sample Selection and Descriptive Statistics
4.1 Sample
Table 1 outlines our sample selection process. The sample selection begins with
all Compustat firm-years beginning in 1998 and extends to 2008. We require five years
of data to compute STRATEGY.
To justify the use of the Miles and Snow typology (1978), which is most
applicable at the business unit level, we exclude firms that earned less than 70 percent of
their sales from a single industry. This criterion was originally suggested by Rumelt
(1974), and helps to establish comparability across firms. To identify firms with less than
70 percent of their sales in a single industry, we use Compustat Segment data to identify
the various industries in which the firm operates. We then total segment sales by industry,
and divide by the firm’s total sales. As indicated in Table 1, these criteria result in the
exclusion of 54,821 firm-year observations.
Next, we delete 14,092 firm-year observations associated with a subsidiary or
foreign incorporated unit, as these firms face different reporting incentives or regulatory
scrutiny. Consistent with prior literature, we also delete 13,932 firm-year observations
associated with firms in financial services and regulated industries (2-digit SIC codes 49,
60-69). All data used to construct the STRATEGY score requires a five-year rolling
average, therefore we delete 48,505 firm-year observations with insufficient data to
21
Management’s ability to determine a firm’s business-level strategy appears to be more of a concern in the
early stages of the firm’s life cycle. Miles and Snow (1978) contend that the decisions made by
management during one adaptive cycle constrain their choice in the next cycle. Furthermore, once a
particular business-level strategy is adopted, it is often difficult to change (Hambrick 1983). Therefore,
computing strategy over the current and prior four years should also address the concern regarding
managerial discretion over business-level strategy
16
construct the 5-year rolling average. We delete 9,896 firm years with insufficient data,
leaving 24,952 observations to construct the STRATEGY score. Next, we exclude 5,153
firm-year observations with insufficient data to compute CASH_ETR or BOOK_ETR.
Finally, we exclude 202 firm-year observations with insufficient data to construct control
variables. Our final sample consists of 19,597 firm-year observations representing 4,184
unique firms over the period 1998 to 2008.
Because Miles and Snow (1978) suggests that competitive strategy can be
classified as a continuum between two different strategic orientations, with prospectors at
one end and defenders at the other end, we define defenders as those firms having a
STRATEGY score ranging from 5 to 10 and prospectors as those firms having a
STRATEGY score ranging from 20 to 25. We create an indicator variable, DEFEND,
which equals one if the firm is classified as a “defender,” and zero otherwise. We also
create an indicator variable, PROSPECT, which equals one if the firms is classified as a
“prospector,” and zero otherwise. Firms having STRATEGY scores 11 through 19 are
classified as analyzers. Under this classification scheme, 1,546 firm-years are classified
as “defenders,” 1,636 firm-years are classified as “prospectors,” and 16,415 firm-years as
“analyzers.”
[Insert Table 1 here]
Table 2, Panel A presents the full sample descriptive statistics for the variables
used to construct STRATEGY and Panel B of Table 2 presents statistics comparing
defenders to prospectors. As expected, the comparisons reported in Panel B indicate that
defenders have lower advertising expense, research and development expense, market-tobook ratios and fewer employees to sales, but have greater fixed assets, Table 2, Panel C
reports the descriptive statistics for variables used in our regression analysis. Focusing on
panel C, the mean (median) BOOK_ETR for the full sample is 28.43 (29.14) percent,
while the mean (median) CASH_ETR of 31.73 (21.25) percent. Our median cash ETR is
lower than the median book ETR, which is consistent with prior research (e.g., Dyreng et
al. 2009; Robinson et al. 2010); however our mean cash ETR is higher than the mean
book ETR. This result appears to be driven by the inclusion of loss firms and firms with
negative tax expense. When we restrict the sample to strictly profitable firms (results not
17
tabulated), the mean book ETR (32.57 percent) is higher than the mean cash ETR (26.82
percent), which is consistent with prior research.
Table 2, Panel D presents the descriptive statistics for all variables included in
equation (1) for both prospectors and defenders. The mean (median) BOOK_ETR for
defenders is 27.99 (29.67) percent, which is higher than the mean (median) BOOK_ETR
for prospectors, which is 22.28 (0.00) percent, and thus is consistent with prospector
firms engaging in more tax avoidance. Defenders have a mean (median) CASH_ETR of
31.00 (20.51) percent, while prospectors have a mean (median) CASH_ETR of 26.24
(1.94) percent. This is consistent with prospectors engaging in more tax avoidance than
defenders when the cash effective tax rate is used as a proxy for tax avoidance. The 0.00
percent median for BOOK_ETR and the 1.94 percent median for CASH_ETR for
prospectors appears to be due to the presence of loss firms, and the method we use to
bound our ETRs between 0 and 1. When we restrict the sample to strictly profitable firms
(results not tabulated), the median book and cash effective tax rates for prospectors are
high, specifically, 34.85 percent for book, and 26.64 percent for cash. These medians are
still significantly lower than median book ETR (35.16 percent) and the median cash ETR
(27.25 percent) for defenders.
Table 2, Panel D also shows that defenders have more debt relative to
prospectors. Specifically, defenders have a mean (median) DEBT value of 0.2387
(0.1506), as compared with 0.1872 (0.0163) for prospectors, which is consistent with
defenders having more fixed assets in place, and the ability to utilize these assets as
collateral. We also find that defender firms are more profitable (ROA), have fewer
intangible assets (INTAN) and tend to be larger in size (SIZE), relative to prospector
firms.
[Insert Table 2 here]
In Table 3, Panel A, we present the correlations between the five variables used
to measure STRATEGY. MARKET5 is positively and significantly correlated with
RD/SALES5 and EMP/SALES5, which is consistent with the argument that firms that
incur costs to inform, educate and motivate their expanding customer base (i.e.,
prospectors), tend to spend more on research and development and tend to be less
18
efficient. The significant and negative correlation between PPEINTENSITY5 and
RD/SALES5 and MtoB is consistent with the argument that firms that place greater
emphasis on production assets (i.e., defenders) spend less money on research and
development, and have fewer investment opportunities. Firms with lower ratios of
employee to sales are assumed to be more efficient. The significant and positive relation
between RD/SALES5 and EMP/SALES suggests that the most efficient firms (i.e.,
defenders) spend less on research and development activities, as these activities are
viewed as an inefficient use of resources. The significant and positive association
between MtoB5 and EMP/SALES5 suggests that more efficient firms have more
investment opportunities, which is inconsistent with our expectations; however the
correlation is not very high (0.01). MtoB5 is also significantly and positively correlated
with RD/SALES5, which is consistent with our expectations. The remaining correlations
are not statistically significant.
Table 3, Panel B presents the correlations among our regression variables. As
expected, BOOK_ETR is significantly and positively correlated with CASH_ETR.
STRATEGY is significantly and negatively correlated with both BOOK_ETR and
CASH_ETR, suggesting that firms classified as prospectors (i.e., those firms with the
highest STRATEGY scores) tend to have lower effective rate rates relative to firms
classified as defenders (i.e., those firms with the lowest STRATEGY scores). Additionally,
based on the Pearson correlation coefficients, lower book ETRs are also reported for
smaller firms (SIZE), firms with fewer intangible assets (INTAN), firms with greater
leverage (DEBT), firms without foreign operations (FOREIGN), firms with higher
research and development expenditures (R&D) and firms with greater capital intensity
(PPEINTENSITY). In comparison, lower cash ETRs are reported for smaller firms
(SIZE), firms with greater leverage (DEBT), and firms with higher research and
development expenditures (R&D).
[Insert Table 3 here]
19
5. Multivariate Results
5.1 Main Results
Table 4 presents the results of our multivariate analyses.22 In column (A), we
report the results of estimating equation (1) using BOOK_ETR as the dependent variable
using the full sample. We find that the coefficient on STRATEGY is -0.0032 (p-value =
<0.001), which suggests prospector firms avoid more taxes than defender firms. In
column (B), we report the results of re-estimating equation (1) using CASH_ETR as the
dependent variable. Again, we find that prospector firms avoid more taxes than
defenders. Specifically, the coefficient on STRATEGY is -0.0042 (p-value = <0.001).
The economic significance between firms’ organizational strategy and the book
and cash effective tax rates is based on the interpretation of the coefficient on
STRATEGY. The coefficient on a continuous independent variable reflects the effect of a
one-unit increase in that variable on the dependent variable. For example, for every one
point increase in STRATEGY, there is a .32 percent decrease in the book effective tax
rate. Therefore, moving from a score of 5 (defender-type firm) to a score of 25
(prospector-type firm) is associated with a 6.4 percent decrease in the book effective tax
rate. Similarly, for every one point increase in STRATEGY, there is a .42 percent
decrease in the cash effective tax rate. Therefore, moving from a score of 5 (defendertype firm) to a score of 25 (prospector-type firm) is associated with a 8.4 percent
decrease in the cash effective tax rate. This evidence is consistent with the argument that
following a prospector strategy versus a defender strategy is not only statistically
associated with lower effective tax rates, but also results in economically significant tax
savings.
Overall, the above analyses indicate that firms that follow a defender strategy
exhibit a lower level of tax avoidance compared with firms that follow a prospector
strategy. These results suggest that even those tax avoidance activities generate tax
22
In both Tables 5 and 6, the p-values are two-tailed and based on standard errors that are clustered by
firm. Petersen (2009) suggests clustering by firm to address potential serial dependence in the data. See
Petersen (2009) for a more detailed discussion.
20
savings for these firms, the costs and uncertainty associated with increased avoidance
outweighs the benefit for defender firms.
[Insert Table 4 here]
In addition to these primary results, several of the estimated coefficients for the
control variables are statistically significant. Focusing on column (A), when the
BOOK_ETR is used as a measure of tax avoidance, the coefficients on SIZE, ROA,
INTAN and FOREIGN are positive and significant, while the coefficients on DEBT and
NOL are negative and significant. In column (B), when the CASH_ETR is used as a
measure of tax avoidance, only the coefficients on SIZE and ROA are positive and
significant, while only the coefficient on FOREIGN and negative and significant. The
coefficients on the remaining variables are not significant. The coefficients on R&D,
and PPEINTENSITY are negative, but insignificant, when both the BOOK_ETR and
CASH_ETR are as measures of tax avoidance, suggesting that our results are not being
driven by firms with higher research and development expenditures (e.g., prospectors)
or firms with more fixed assets in place (e.g., defenders).
To examine the robustness of the full sample results, and to ensure our results are
not being driven by loss firms, we restrict the sample to include only firms with strictly
positive income during the period from 1998 through 2008. We exclude loss firms
because these firms may have different incentives to engage in tax avoidance. We choose
to exclude loss firms, as opposed to loss firm-years, because loss years not only affect the
current year, but past and future years because of NOL carryfowards and carrybacks.
Therefore, only dropping the loss years could lead to the conclusion that a firm’s tax
burden was less than it was in reality. Restricting the sample to strictly profitable firms
reduces the sample to 6,334 firm-year observations.
Table 4, Columns (C) and (D) presents the results for the reduced sample. The
results are qualitatively similar to the full sample results, and thus lead to similar
inferences. Again, we find that prospectors engage in more tax avoidance than
defenders.
21
5.2. The Consistency of Firm Strategy
As an additional test of our STRATEGY measure, we consider whether the
particular strategy adopted by a firm is consistent over time, as suggested by Miles and
Snow (1978). Therefore, we expect that prospector, defender or analyzer firms should
have relatively stable STRATEGY scores from year-to-year.23 Specifically, we expect
firms classified as defenders to exhibit low STRATEGY scores and firms classified as
prospectors to exhibit high STRATEGY scores over our sample period. To test the
consistency of firms’ strategies, we compute first differences for each firm-year
STRATEGY score to determine how many total times a firm changed its STRATEGY
score. We find that 24 percent of firms never changed their score from year-to-year while
36 percent of firms changed their score by one value (e.g. changing from 25 to 24). Only
5 percent of firms changed their score by more than 3 values. We thus conclude that our
measure of strategy is relatively consistent throughout time.
6. Sensitivity Analysis
6.1. Defender/Prospector Indicator Variables
In this section, we discuss the results from estimating equation (1) after replacing
our continuous STRATEGY measure in model (1) with indicator variables for defenders
and prospectors. To examine whether defenders exhibit more or less tax avoidance than
prospectors, we perform an F-test to compare the difference between the coefficients on
DEFEND and PROSPECT. In untabulated results, we find that prospectors avoid more
taxes than defenders when the BOOK_ETR is used as the dependent variable (F-Statistic
= 9.26; Prob>F = 0.0024) and when the CASH_ETR is used as the dependent variable (FStatistic = 4.16; Prob>F = 0.0414). Thus, our main results are robust to this specification.
6.2. Is STRATEGY Capturing Something Different?
In this section, we examine whether our measure of strategy captures something
different and over and above the other explanatory variables in our model. We begin by
examining the correlations between STRATEGY and the other explanatory variables in
23
We would expect that firms classified as “reactors” would not exhibit relatively stable STRATEGY
scores. We do not attempt to classify reactors,nor do we make any predictions for these firms.
22
Table 3, Panel B. Although the correlations between STRATEGY and the other
explanatory variables are statistically significant, none of the correlations are greater than
0.05, with the exception of R&D (0.1199) and PPEINTENSITY (-0.1287), which are used
to construct STRATEGY.
Next, we compute the canonical correlations between the variables used to
construct the STRATEGY score and the other explanatory variables in the model.
Examining the correlations between the first set of variables (RD/SALES5, EMP/SALES5,
PPEINTENSITY5, MARKET5, MtoB5) and the second set of variables (SIZE, ROA,
INTAN FOREIGN, DEBT, R&D, PPEINTENSITY NOL), we find that of the resulting five
canonical correlation coefficients, four are statistically significant. The largest significant
coefficient is 61 percent, and the smallest significant coefficient is 2.7 percent. Although
the 61 percent appears high this is offset by a low redundancy index. Specifically, the
maximum shared variation between the two groups of variables is 11 percent, which is
quite low and considered poor by analysis standards. Furthermore, the strongest shared
variance observed (7 percent) is between PPEINTENSITY and the other canonical
variates, which is not surprising because property, plant and equipment is in both groups.
This suggests that the combined strategy variable set captures related, but not
substantially overlapping, information about the firm. We suggest this supports the
notion that our STRATEGY variable represents aspects of the firm that are not represented
by the additional variables in the model that have been documented as affecting firms’
levels of tax avoidance.
6.3. Is Size Driving Our Results?
We conduct and additional sensitivity tests to confirm that our STRATEGY variable is
not a proxy for firm size. We rank firms on our variable SIZE by forming quintiles
within each 2-digit SIC industry-year. We create two indicator variables, Large, which
equals one if a firm is in the top SIZE quintile and 0 otherwise, and Small, which equals
one if the firm is in the bottom quintile. We then interact each of these indicator variables
with our STRATEGY variable, and re-estimate equation (1) by including Large, Small,
LargexSTRATEGY and SmallxSTRATEGY. If STRATEGY is not proxying for size, then
the coefficients on the interaction terms should not be significant. In untabulated results,
23
we find that of the two interactions only the Small*STRATEGY coefficient is negative
and significant which is opposite of that expected if STRATEGY is a proxy for firm size.
Given that our prospector firms are smaller on average than defender firms in the sample
the significant interaction is not unexpected. The coefficient on Large is negative and
significant indicating that large firms have lower effective tax rates while the
STRATEGY variable continues to be negative and significant in all estimates.
7. Conclusion
We examine whether a firm’s tax planning activities (i.e., tax avoidance) are
influenced by its business strategy. Our research question is motivated by the results of
prior literature, which finds that there is substantial unexplained variation in firms’
propensities to avoid income taxes.
To address our research question, we create a measure (STRATEGY) designed to
provide a score that reflects the most likely business strategy adopted by sample firms.
We use this score to categorize firms as prospectors, defenders or analyzers. Using data
from 1998 to 2008, we examine the association between firms’ business strategy and two
measures of tax avoidance, the book effective tax rate and the cash effective tax rate.
We find that defender firms engage in less tax avoidance than prospector firms, as
demonstrated by higher book and cash effective tax rates. These results suggest that
even though tax avoidance activities generate tax savings for these firms, the costs and
uncertainty associated with increased avoidance outweighs the benefit.
Our study contributes to both the tax avoidance literature in accounting and the
organizational theory literature in management in the following ways. First, we develop a
replicable proxy of a firm’s business strategy, which is constructed using publicly
available data, and is more comprehensive than existing proxies in the accounting
management literatures. Second, we contribute to the management literature by providing
evidence that firms’ tax planning activities are associated with their organizational
strategy. Thus, our results empirically support Porter’s (1996) argument that firms should
reinforce their business-level strategy with support activities in order sustain a
competitive advantage. Third, our results contribute to the stream of research that
investigates the variation in firms’ tax avoidance activities, by providing evidence that a
24
firm’s business strategy influences its level of tax avoidance. Furthermore, our results
contribute toward a better understanding of the impact of different competitive strategies
on firms’ tax planning and reporting practices.
25
Appendix A
Strategic Type Characteristics
Defenders
Prospectors
Analyzers
Definition
A firm that has a very
narrow product-market
domain, a focus on
cost efficiency and a
stable organizational
structure.
A firm that has a very
broad product-market
domain, a focus on
innovation and a flexible
organizational structure.
A firm that has the attributes
of both a prospector and
a defender. The firm operates
in two types of product
market domains, one is
relatively stable and the
is changing.
Competitive
Advantage
Cost efficiency and stability
Innovation and flexibility
Balances flexibility and
stability
Require certainty in future
Adapt to uncertainty, and
Require some amount of
outcomes, and often engage
often engage in new
certainty in future outcomes
in detailed planning before
opportunities before detailed
undertaking new opportunities
planning is complete
Growth
Cautious and incremental
growth and advances in
productivity
Growth occurs in spurts
through product and market
development
Steady growth through both
market penetration and
product and market
development
Research and
Development
Minimal R&D, which is
usually related to current
products
Extensive R&D to identify
new products and market
opportunities
Minimal R&D because
they adopt most promising
innovations from prospectors
Strong focus on marketing
Strong focus on marketing
in innovative sector
Planning
Marketing
Strong emphasis is on
financial and production
functions, and less on
marketing
Capital Intensity
Focus on production
assets
Technology is embedded
in people
Moderate focus on
production assets
Management
Characteristics
Finance and production
experts are the most
powerful members.
Tenure of executives is
lengthy and managers are
promoted from within.
Marketing and R&D
experts are the most
powerful members.
Tenure of executives
is not lengthy and managers
may be hired from outside
or promoted from within.
Marketing experts,
followed by production,
and planning staff most
important members.
This table was constructed based on a review of the three viable business strategies identified by Miles & Snow (1978).
Miles and Snow (1978) identify fourth strategy, Reactors, which are generally presented as a "residual" type
lacking any systematic strategy, design, or structure.
26
Appendix B
STRATEGY Score Composition and Example
The STRATEGY score is constructed based on the following five measures, drawn from Ittner et al. (1997)
and Hambrick (1983). Each of the variables are computed using a rolling average of the respective yearly ratios
over the previous five years. We then rank each of the five variables by forming quintiles based on 2-digit
SIC code and year. For each of the first four variables, firms in the top quintile receive a score of 5,
those in the second highest quintile receive a score of 4, etc. The scoring for PPEINTENSITY5 is inverted,
so that observations in the highest quintile receive a score of 1, those in the next highest quintile receive a score
of 2, etc. The scores are summed over the five measures per firm-year, such that the maximum score a firm
could receive is 25 (prospector), and the minimum score a firm could receive is 5 (defender).
Variable
Definition
RD/SALES5
Five year rolling average (year t-5 through year t-1) of the yearly ratio of total
research and development expense (Compustat XRD) to total sales
(Compustat SALE)
EMP/SALES5
Five year rolling average (year t-5 through year t-1) of yearly ratio of the
total number of employees (Compustat EMP) to total sales (Compustat SALE)
MtoB5
Five year rolling average (year t-5 through year t-1) of the ratio of market
value of equity (Compustat abs(PRCC_F)*CSHO) to book value of equity
(Compustat CEQ)
MARKET5
Five year rolling average (year t-5 through year t-1) of the yearly ratio of total
advertising expense (Compustat XAD) to total sales (Compustat SALE)
PPEINTENSITY5
Five year rolling average (year t-5 through year t-1) of the yearly ratio of
gross property, plant and equipment (Compustat PPEGT) to total assets
(Compustat AT)
27
Appendix C
Regression Variable Definitions
Variable
Definition
BOOK_ETR
Total tax expense (Compustat TXT) in year t divided by pre-tax book income
(Compustat PI) in year t adjusted for special items (Compustat SPI)
CASH_ETR
Total cash taxes paid (Compustat TXPD) divided by pre-tax book income
Compustat PI) adjusted for special items (Compustat item SPI). If total cash taxes
paid are missing, total current taxes are used (Compustat item TXC)
STRATEGY
Strategy score (for composition, see Appendix B)
SIZE
Natural log of total assets (Compustat AT) in year t
ROA
Pre-tax income (Compustat PI) in year t divided by beginning of the year total assets
(Compustat AT)
DEBT
Total long term debt (Compustat DLCC) for year t divided by beginning of the year
total assets (Compustat AT)
FOREIGN
Indicator variable equal to 1 if a firm had positive pretax foreign income
(Compustat PIFO) in year t , and 0 otherwise
INTAN
Total intangibles (Compustat INTAN) in year t scaled by beginning of the year total
assets (Compustat AT)
R&D
Total research and development expense (Compustat XRD) in year t , scaled
by beginning of the year total assets (Compustat AT)
PPEINTENSITY Gross property, plant and equipment (Compustat PPEGT) in year t , divided by
total assets (Compustat AT) in year t
NOL
Indicator variable equal to one if the firm has a tax-loss carry forward
(Compustat TLCF) during the year t
28
References
Abell, D.F. 1980. Defining the business: The starting point of strategic planning.
Englewood Cliffs, NJ: Prentice-Hall.
Armstrong, C., J. Blouin, and D. Larcker. 2010. The incentives for tax planning. Working
paper, University of Pennsylvania.
Bentley, K.A., T.C. Omer, and N.Y. Sharp. 2010. Business strategy, audit fees and
financial reporting irregularities. Working paper, Texas A&M University.
Bushman, R., R. Indjejikian, and A. Smith. 1996. CEO compensation: The role of
individual performance evaluation. Journal of Accounting and Economics 21:
161-193.
Callihan, D. 1994. Corporate effective tax rates: A synthesis of the literature. Journal of
Accounting Literature 13: 1-43.
Chen, S., X. Chen, Q. Cheng, and T. Shevlin. 2010. Are family firms more tax aggressive
than non-family firms? Journal of Financial Economics 95: 41-61.
Chrisman, J.J., C.W. Hofer, and W.R. Boulton. 1988. Towards a system for classifying
business strategy. Academy of Management Review 13(3): 413-428.
Cook, K. R. Huston, and T. Omer. 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.
Dunbar, A. E. and J. D. Phillips, 2001. The outsourcing of corporate tax function
activities. Journal of the American Taxation Association, 23(2): 35-49.
Dyreng, S., M. Hanlon, and E. Maydew. 2008. Long-run corporate tax avoidance. The
Accounting Review 83: 61-82.
Dyreng, S., M. Hanlon, and E. Maydew. 2009. The effects of executives on corporate tax
avoidance. Working paper, Duke University.
Frank, M.M., L. Lynch and S. Rego. 2009. Tax reporting aggressiveness and its relation
to aggressive financial reporting. The Accounting Review 84(2): 467-496.
Graham, J. and A. Tucker. 2006. Tax shelters and corporate debt policy. Journal of
Financial Economics 81: 563-594.
Grubert, H. and J. Slemrod. 1998. The effect of taxes on investment and income shifting
to Puerto Rico. Review of Economics and Statistics 80(3):365-373.
29
Gupta, S., and K. Newberry. 1992. Corporate average effective tax rates after the Tax
ReformAct of 1986. Tax Notes May 4: 689-702.
Gupta, S. and K. Newberry. 1997. Determinants of the variability in corporate effective
tax rates: Evidence from longitudinal study. Journal of Accounting and Public
Policy 16: 1-34.
Gompers, P., J. Ishii, and A. Metrick. 2003. Corporate governance and equity prices. The
Quarterly Journal of Economics 118 (1): 107-155
Hambrick, D.C. 1983. Some tests of the effectiveness and functional attributes of Miles
and Snow's strategic types. The Academy of Management Journal 26(1): 5-26.
Hanlon, M. and S. Heitzman. 2010. A review of tax research. Journal of Accounting and
Economics, 50 (2-3): 127-178.
Hanlon, M., Mills, L. and J. Slemrod. 2007. An empirical examination of big business tax
noncompliance. Book chapter, Slemrod (editor). Corporate Taxation.
Ittner, C.D. and D.F. Larker. 1997. The choice of performance measures in annual bonus
contracts. The Accounting Review 72(2): 231-255
Markle, K. and D. Shackelford. 2009. Does the domicile of a multinational affect its
worldwide corporate income tax liability? University of North Carolina working
paper.
McGuire, S.T, T.C. Omer and D. Wang. 2010. Tax avoidance: Do industry experts make
a difference? Working paper, Texas A&M.
Miles, R. E., and Snow, C. C. Organizational strategy, structure and process. New York:
McGraw-Hill, 1978.
Mills, L., M. Erickson, E. Maydew. 1998. Investments in tax planning. The Journal of the
American Taxation Association 20 (1): 1-20.
Petersen, M. 2009. Estimating standard errors in finance panel data sets: Comparing
approaches. Review of Financial Studies 22: 435-481
Phillips, J. 2003. Corporate tax-planning effectiveness: The role of compensation-based
incentives. The Accounting Review 78: 847-874.
Porter, M. E. 1980. Competitive Strategy. New York: Free Press.
Porter, M. E. 1996. What is strategy? Harvard Business Review 74(6): 61-78.
30
Rego, S. and R. Wilson. 2009. Executive compensation, tax reporting aggressiveness, and
future firm performance. Working paper, University of Iowa.
Rego, S. 2003. Tax avoidance activities of U.S. multinational corporations.
Contemporary Accounting Research 20: 805-833.
Robinson, J., S. Sikes, and C. Weaver. 2010. The impact of evaluating the tax department
as a profit center on effective tax rates. The Accounting Review, forthcoming.
Rumelt, R.P. 1974. Strategy, Structure, and Economic Performance. Cambridge, MA:
Harvard University Press.
Shevlin, T. and S. Porter. 1992. The corporate tax comeback in 1987: Some further
evidence. Journal of the American Taxation Association 14: 58-79.
Smith, K., J. Gutherie, and M.J. Chen. 1989. Strategy, Size and Performance.
Organizational Studies, 10:63-81.
Smith, C., and R. Watts. 1992. The investment opportunity set and corporate financing,
dividend, and compensation policies. Journal of Financial Economics 32: 263292.
Snow, C. C. and Hrebiniak, D. C. 1980. Measuring organizational strategies. Academy of
Management Review 5, 317-336.
Stickney, C., and V. McGee. 1982. Effective corporate tax rates: the effect of size, capital
intensity, leverage, and other factors. Journal of Accounting and Public Policy 1:
125-152.
Thomas, A., R. Litschert, and K. Ramaswamy. 1991. The performance impact of
strategy-manager coalignment: An empirical examination. Strategic Management
Journal 12: 509-522.
Wilson, R. 2009. An examination of corporate tax shelter participants. The Accounting
Review 84: 969-999.
Zimmerman, J. 1983. Taxes and firm size. Journal of Accounting and Economics 5: 119149.
31
Table 1
Sample Selection
Panel A: Sample Selection Procedure
Criteria
Number of
Firm-Year Observations
All Compustat Firms for Fiscal Years 1993-2008
166,198
Less:
Firms that earn less than 70 percent of sales from a single industry
(54,821)
Observations associated with a foreign subsidiary or foreign incorporated units
(14,092)
Observations in regulated industries (SIC Codes 49, 60-69)
(13,932)
Firms without required five year rolling average data for STRATEGY variable
(48,505)
Observations with insufficient data to measure variables used in the STRATEGY score
(9,896)
Total Observations for STRATEGY Score
24,952
Less:
Observations with insufficient data to measure BOOK_ETR or CASH_ETR
Observations with insufficient data for measure control variables
Final Sample for Fiscal Years 1998-2008
(5,153)
(202)
19,597
32
Table 2
Descriptive Statistics
Panel A: STRATEGY Variables
Variable
MARKET5
PPEINTENSITY5
RD/SALES5
EMP/SALES5
MtoB5
N
19597
19597
19597
19597
19597
Mean
0.0208
0.5418
0.3821
0.0178
3.0331
Std. Dev.
0.6734
0.7577
1.7035
0.1650
3.8111
25th Pctl
0.0000
0.2223
0.0000
0.0040
1.2932
Median
0.0000
0.4169
0.0073
0.0062
2.2092
75th Pctl
0.0080
0.7146
0.1211
0.0099
3.8698
*See Appendix B for variable definitions
Panel B: STRATEGY Variables by Prospectors and Defenders
Variable
a,b
MARKET5
a,b
PPEINTENSITY5
a,b
RD/SALES5
a,b
EMP/SALES5
a,b
MtoB5
Mean
0.0010
0.8609
0.0164
0.0066
0.9458
Defenders (N= 1546 )
Std. Dev.
25th Pctl
Median
0.0049
0.0000
0.0000
1.0221
0.4876
0.7376
0.0468
0.0000
0.0000
0.0119
0.0027
0.0047
2.2204
0.6599
1.1651
75th Pctl
0.0000
1.0522
0.0145
0.0070
1.7647
Mean
0.1151
0.3429
1.8055
0.0565
6.5022
Prospectors (N= 1636)
Std. Dev.
25th Pctl
Median
2.3126
0.0000
0.0048
0.3155
0.1375
0.2362
3.7636
0.0011
0.2105
0.3203
0.0064
0.0109
5.1875
3.0743
5.1375
75th Pctl
0.0310
0.4526
1.1698
0.0269
8.3890
a (b) indicates that the differences in the means (medians) between defenders and prospectors is significant at the .10 level using a t-test (Wilcoxon test) of differences in means (medians)
*See Appendix B for variable definitions
33
Table 2 (cont'd)
Descriptive Statistics
Panel C: Regression Variables
Variable
BOOK_ETR
N
19597
Mean
0.2843
Std. Dev.
0.3014
25th Pctl
0.0000
Median
0.2914
75th Pctl
0.3829
CASH_ETR
STRATEGY
SIZE
ROA
INTAN
DEBT
FOREIGN
R&D
PPEINTENSITY
NOL
19597
19597
19597
19597
19597
19597
19597
19597
19597
19597
0.3173
15.0254
5.0456
-0.3446
0.1549
0.1989
0.2447
0.0854
0.5735
0.3969
0.3565
3.2443
2.1411
27.6541
0.7092
0.4943
0.4299
0.3335
1.1633
0.4893
0.0000
13.0000
3.5026
-0.0886
0.0000
0.0000
0.0000
0.0000
0.2149
0.0000
0.2125
15.0000
5.0655
0.0420
0.0435
0.0771
0.0000
0.0057
0.4250
0.0000
0.4189
17.0000
6.5325
0.1287
0.1866
0.2721
0.0000
0.0976
0.7454
1.0000
*See Appendix C for variable definitions
Panel D: Regression Variables by Prospectors and Defenders
Variable
Mean
Defenders (N= 1546)
Std. Dev.
25th Pctl
Median
75th Pctl
Mean
0.2967
0.3873
0.2228
0.3272
0.0000
0.0000
0.3673
0.0028
3.4296
0.2051
4.8733
0.4151
6.1520
0.2624
4.2859
0.3753
2.0325
0.0000
2.8244
0.0194
4.0998
0.3700
5.7811
1.0578
-0.0347
0.0475
0.1132
-0.3881
2.0556
-0.4381
-0.0959
0.0878
0.1419
0.2387
0.1986
1.2839
0.3769
0.3991
0.0000
0.0105
0.0000
0.0167
0.1506
0.0000
0.1255
0.3312
0.0000
0.1947
0.1872
0.1559
0.8278
0.4346
0.3628
0.0000
0.0000
0.0000
0.0431
0.0163
0.0000
0.2137
0.2493
0.0000
a,b
0.0261
0.9115
0.1244
2.3431
0.0000
0.4395
0.0000
0.7163
0.0146
1.0742
0.1753
0.3968
0.3135
0.4051
0.0000
0.1433
0.0884
0.2671
0.2322
0.5244
a,b
0.3752
0.4843
0.0000
0.0000
1.0000
0.4401
0.4966
0.0000
0.0000
1.0000
BOOK_ETR
a,b
0.2799
0.2879
0.0000
CASH_ ETR
a,b
SIZE
a,b
0.3100
4.8502
0.3465
1.9874
-0.0368
INTAN
a,b
DEBT
a,b
FOREIGN
ROA
a,b
b
R&D
a,b
PPEINTENSITY
NOL
Prospectors (N=1636)
Std. Dev.
25th Pctl
Median
75th Pctl
a (b) indicates that the differences in the means (medians) between defenders and prospectors is significant at the .10 level using a t-test (Wilcoxon test) of differences in means (medians)
*See Appendix C for variable definitions
34
Table 3
Correlations
Panel A: STRATEGY Variables
MARKET5
PPEINT5
RD/SALES5 EMP/SALES5
MtoB5
MARKET5
1.0000
PPEINTENSITY5
-0.0048
1.0000
RD/SALES5
EMP/SALES5
0.0350
0.4711
0.0098
-0.0517
-0.0061
1.0000
0.3598
1.0000
-0.0284
0.0457
0.0144
1.0000
SIZE
ROA
INTAN
DEBT
FOREIGN
R&D
1.0000
0.0292
0.0393
0.0315
0.3608
-0.1777
-0.0763
-0.0738
1.0000
-0.0073
-0.0114
0.0088
-0.0467
-0.0083
-0.0109
1.0000
0.2085
0.0136
0.0343
-0.0482
0.0307
1.0000
-0.0441
0.0188
0.1398
0.0271
1.0000
-0.0335
-0.0578
0.0470
1.0000
-0.0045
0.0391
MtoB5
*See Appendix B for variable definitions
Panel B: Regression Variables
BOOK_ETR CASH_ETR STRATEGY
BOOK_ETR
CASH_ETR
STRATEGY
SIZE
ROA
INTAN
DEBT
FOREIGN
R&D
PPEINTENSITY
NOL
1.0000
0.3752
-0.0414
0.1396
0.0108
0.0208
-0.0267
0.0677
-0.0777
-0.0272
-0.0564
1.0000
-0.0437
0.0364
0.0078
0.0066
-0.0168
0.0078
-0.0499
-0.0089
-0.0220
1.0000
-0.0303
-0.0084
0.0182
-0.0350
-0.0016
0.1199
-0.1287
0.0410
Note: Bold indicates that the correlation is significant at p<0.05.
*See Appendix C for variable definitions
PPEINT.
1.0000
-0.0185
NOL
1.0000
Table 4
Regression Results
Full Sample
Variables
(A)
BOOK_ETR
Coeff. Est. p-value
(B)
CASH_ETR
Coeff. Est. p-value
Strictly Profitable Firms Only
(C)
(D)
BOOK_ETR
CASH_ETR
Coeff. Est. p-value
Coeff. Est. p-value
Intercept
0.3761
<0.001
0.3810
<0.001
0.3523
<0.001
0.4065
<0.001
STRATEGY
-0.0032
0.001
-0.0042
<0.001
-0.0024
0.006
-0.0028
0.009
SIZE
0.0156
<0.001
0.0063
0.003
0.0040
0.024
-0.0033
0.116
ROA
0.0000
0.001
0.0001
0.001
0.0921
0.129
-0.0073
0.740
INTAN
0.0082
0.051
0.0043
0.353
0.0041
0.727
-0.0161
0.260
DEBT
-0.0171
0.001
-0.0092
0.117
-0.0167
0.179
-0.0543
0.003
FOREIGN
0.0155
0.022
-0.0160
0.054
-0.0175
<0.001
0.0068
0.355
R&D
-0.0337
0.135
-0.0333
0.109
-0.3639
<0.001
-0.3482
<0.001
PPEINTENSITY
-0.0032
0.435
0.0013
0.855
-0.0315
<0.001
-0.0193
0.006
NOL
-0.0273
<0.001
-0.0085
0.288
-0.0162
0.007
-0.0349
<0.001
Adjusted R-Squared
N (firm-years)
0.0385
19,597
0.0252
19,597
0.1057
6,334
0.1022
6,334
*p-values are two-tailed and based on standard errors clustered by firm (Petersen 2009).
*Coefficient estimates for fiscal year and industry dummies are not reported for brevity
*See Appendix B for variable definitions
36