Author's Accepted Manuscript

Corporate governance, Incentives, and tax

avoidance
Christopher S. Armstrong, Jennifer L. Blouin,
Alan D. Jagolinzer, David F. Larcker

www.elsevier.com/locate/jae

PII: S0165-4101(15)00017-8
DOI: http://dx.doi.org/10.1016/j.jacceco.2015.02.003
Reference: JAE1054

To appear in: Journal of Accounting and Economics

Received date: 17 April 2013

Revised date: 25 February 2015
Accepted date: 25 February 2015

Cite this article as: Christopher S. Armstrong, Jennifer L. Blouin, Alan D.

Jagolinzer, David F. Larcker, Corporate governance, Incentives, and tax
avoidance, Journal of Accounting and Economics, http://dx.doi.org/10.1016/j.jacce-
co.2015.02.003

This is a PDF file of an unedited manuscript that has been accepted for
publication. As a service to our customers we are providing this early version of
the manuscript. The manuscript will undergo copyediting, typesetting, and
review of the resulting galley proof before it is published in its final citable form.
Please note that during the production process errors may be discovered which
could affect the content, and all legal disclaimers that apply to the journal
pertain.
 Corporate Governance, Incentives, and Tax Avoidance

Christopher S. Armstrong
The Wharton School, University of Pennsylvania
[email protected]

Jennifer L. Blouin*
The Wharton School, University of Pennsylvania
[email protected]

Alan D. Jagolinzer
Leeds School of Business, University of Colorado
[email protected]

David F. Larcker
Graduate School of Business, Rock Center for Corporate Governance, Stanford University
[email protected]

February 17, 2015

Abstract: We examine the link between corporate governance, managerial incentives, and corporate tax
avoidance. Similar to other investment opportunities that involve risky expected cash flows, unresolved
agency problems may lead managers to engage in more or less corporate tax avoidance than shareholders
would otherwise prefer. Consistent with the mixed results reported in prior studies, we find no relation
between various corporate governance mechanisms and tax avoidance at the conditional mean and
median of the tax avoidance distribution. However, using quantile regression, we find a positive relation
between board independence and financial sophistication for low levels of tax avoidance, but a negative
relation for high levels of tax avoidance. These results indicate that these governance attributes have a
stronger relation with more extreme levels of tax avoidance, which are more likely to be symptomatic of
over- and under-investment by managers.

JEL: G34, H25, H26, K34, M41

Keywords: Tax aggressiveness; FIN 48; tax avoidance; CEO incentives; corporate governance

1. Introduction

We examine the role of governance in tax planning decisions to help resolve the debate in

the governance and tax literatures about whether a link exists between firms’ corporate
 
*Corresponding author

We thank Shane Heitzman and workshop participants at University of Delaware, ESADE, Goethe University,
University of Iowa, the INSEAD research conference, University of Oregon, University of Toronto research
conference, Texas A&M University, Washington University, and Yale University for helpful comments. Jagolinzer
thanks the EKS&H Faculty Fellowship for financial support. Larcker thanks the Winnick Family Fund for financial
support.




governance structures, including managers’ incentive-compensation contracts, and corporate tax

avoidance. The debate exists, in part, because prior research relies on Desai and Dharmapala’s

(2006) theory that provides counterintuitive predictions about the link between governance and

tax avoidance decisions. In particular, Desai and Dharmapala (2006) argue that tax avoidance

and managerial rent extraction can be complementary if tax avoidance reduces corporate

transparency which, in turn, increases the opportunity for managers to divert corporate resources

for personal benefit. Hence, their theory suggests that reducing tax planning can simultaneously

reduce managerial diversion. Desai and Dharmapala further assume that well-governed firms are

more likely to have internal control mechanisms to prevent such diversion and argue that a

negative relation between managers’ equity incentives and tax avoidance will only manifest in

well-governed firms. In contrast, they assume that poorly governed firms will not use equity

incentives to encourage tax avoidance because they lack the governance mechanisms to prevent

managerial diversion.

While Desai and Dharmapala’s theory is provocative, it has a number of important

limitations. For example, Desai and Dharmapala fail to acknowledge that equity incentives are,

themselves, an important governance mechanism (i.e., managers’ equity incentives result from

endogenous decisions by the board of directors that are, in part, aimed at mitigating agency

problems). It is also unclear why well-governed firms would be more likely to rely on equity

incentives than poorly-governed firms. If anything, prior studies suggest that managerial equity

incentives can substitute for other governance mechanisms where more direct monitoring is

either too costly or infeasible (e.g., Demsetz and Lehn, 1985; Core and Guay, 1999). Prior

studies also suggest that poorly-governed firms (where managers have “control of the board”)

are more likely to over-pay managers in the form of equity (Bebchuk and Fried, 2004).




Moreover, these same managers should have strong incentives to reduce tax payments and

personally capture the resulting economic benefits (because of the lax oversight). Therefore,

similar to their well-governed counterparts, poorly-governed firms should also exhibit a negative

relation between equity incentives and tax avoidance. This chain of reasoning calls into question

Desai and Dharmapala’s conjecture that the relation between equity incentives and tax avoidance

should only exist in well-governed firms.

In contrast to Desai and Dharmapala, we adopt a more traditional view of the role of

governance on firms’ tax avoidance. Under our alternative agency-theoretic view, tax avoidance

is one of many risky investment opportunities available to management. Similar to other

investment decisions, unresolved agency problems can lead managers to select a level of tax

avoidance that differs from what shareholders would prefer. We do not assume that tax

avoidance necessarily results in opportunities for managerial diversion. Rather, as with other

agency problems, we assume that the various governance mechanisms in place, including

managers’ incentive-compensation contracts, can mitigate agency problems with respect to tax

avoidance.

Other papers directly examine the link between corporate governance and tax avoidance.

Minnick and Noga (2010) investigate whether several measures of corporate governance are

associated with a variety of proxies intended to capture firms’ level of tax avoidance, but find

little evidence of a link. Rego and Wilson (2012) find that firms at which managers have

relatively large risk-taking equity incentives engage in more tax avoidance. However, they fail to

find evidence of a relation between other governance mechanisms and tax avoidance. Finally, in

a concurrent study, Robinson et al. (2012) report evidence that audit committee financial

expertise is generally positively associated with tax planning, but that this association is negative




when tax planning is thought to be risky (i.e., aggressive). Overall, the relationships among

corporate governance, managerial equity incentives, and tax avoidance are mixed and result in

inconclusive inferences in the existing literature.

One common theme across prior studies is that inferences are based on estimates of how

governance relates to the conditional mean of the tax avoidance distribution. However, the

relationship between governance and the (conditional) average level of tax avoidance may not

accurately describe the relationship in other parts of the tax avoidance distribution. Rather than

rely solely on traditional econometric methods (i.e., ordinary least squares regression) that only

provide estimates of the average relationship, we also estimate a series of quantile regressions to

assess the relation across the entire tax avoidance distribution. This research design follows

naturally from our conjecture that the relationship between corporate governance and tax

avoidance will differ at relatively high and low levels of tax avoidance. In particular, boards that

are more knowledgeable about the net benefits of tax strategies should encourage more tax

planning at lower levels of tax avoidance because this improves cash flows with little

accompanying risk. Conversely, more knowledgeable boards should discourage additional tax

avoidance when the level is high because the increased costs (e.g., regulatory or reputational) are

more likely to outweigh the marginal benefit of additional tax savings.

We examine a sample of firms between 2007 and 2011 and find that CEOs’ risk-taking

equity incentives exhibit a positive relationship with the average level (i.e., conditional mean) of

tax avoidance. This result is analogous to the positive relationship between risk-taking equity

incentives and earnings management reported by Armstrong et al. (2013) and is consistent with

our characterization of tax avoidance as a risky positive expected net present value investment

from the perspective of CEOs. More importantly, we find that this relationship is stronger for




higher levels of tax avoidance, which suggests that managerial risk-taking incentives are an

important determinant of aggressive tax positions that are likely to entail more risk.

We also examine how other governance mechanisms relate to observed levels of tax

avoidance. In particular, we examine two important attributes of the board: financial expertise (to

measure knowledge of the costs and benefits of tax avoidance) and independence (to measure the

ability and incentive to monitor managers’ tax avoidance decisions). We find that the relation

between boards’ financial expertise and independence and the level of tax avoidance differs

considerably across the tax avoidance distribution. Specifically, we find that board financial

expertise and independence both have a positive relation with tax avoidance for low levels of tax

avoidance, which is consistent with under-investment in tax avoidance in the absence of

monitoring. We also find that board financial expertise and independence both have a negative

relation with tax avoidance for high levels of the tax avoidance, which is also consistent with

over-investment in tax avoidance in the absence of monitoring. Collectively, these findings

suggest that more financially sophisticated and more independent boards attenuate relatively

extreme levels of tax avoidance, which are likely to be symptomatic of unresolved agency

problems.

Because our findings are at odds with the predictions from Desai and Dharmapala’s

model, we attempt to reconcile the two sets of results. Consistent with Desai and Dharmapala

(2006), we find that an interaction between an indicator of “good” governance and top

executives’ stock option compensation exhibits no relation with tax avoidance based on ordinary

least squares estimates of the conditional mean. However, quantile regression estimates of their

specification indicate a strong negative relation for high levels of tax avoidance and no relation

for low levels. These estimates suggest that “good” corporate governance mitigates over-




investment in high levels of tax avoidance that would otherwise occur when executives receive

large stock option grants. Thus, in contrast to Desai and Dharmapala (2006), we find that

corporate governance appears to be related to managers’ tax avoidance decisions, but only for

high levels of tax avoidance. Importantly, this relationship is not apparent from examining the

average level of tax avoidance. Our findings call into question whether tax avoidance and

managerial rent extraction are, in fact, complementary activities as predicted by Desai and

Dharmapala, and whether boards view them as such. If tax avoidance is instead better

characterized as one of many alternative risky investment opportunities as our results suggest,

then studies that interpret low levels of (i.e., underinvestment in) tax avoidance as a symptom of

“good governance” should be reevaluated (e.g., Chen et al., 2010).

The remainder of the paper proceeds as follows. Section two discusses competing

theories for the relation between corporate governance and tax avoidance and provides

arguments for why the relation may vary for different levels of tax avoidance. Section three

describes our sample. Section four discusses our research design and explains our choice of

quantile regression estimation. Section five presents our primary empirical results and

inferences. Section six discusses additional supplemental analysis and our reexamination of

Desai and Dharmapala’s (2006) findings. Section seven provides concluding remarks.

2. Prior Literature and Hypothesis Development

2.1 Prior Literature




 A mature stream of corporate tax research examines the determinants of effective tax

rates and book-tax differences (e.g., Gupta and Newberry, 1997). Subsequent studies (e.g., Mills

and Newberry, 2001 and Cloyd et al., 1996) focus on the book-tax tradeoffs that are associated

with various tax avoidance opportunities (i.e., some tax avoidance strategies reduce both taxable

and financial statement income, whereas others affect only taxable income). Although this prior

research is useful, it provides little insight into why some firms engage in more tax planning than

others (Shackelford and Shevlin, 2001).

This gap in the literature spurred a series of papers that more directly examine the

determinants of corporate tax avoidance.1 For example, Dyreng et al. (2010) report evidence that

executives who were previously employed by firms that are characterized as tax aggressive seem

to import this aggressiveness to their new employer. Slemrod (2004), Crocker and Slemrod

(2005), and Chen and Chu (2005) suggest that corporate tax noncompliance (i.e., extreme tax

avoidance) could result from the tax reporting incentives provided by managers’ incentive-

compensation contracts. Consistent with this notion, there is empirical evidence that tax

avoidance is associated with greater levels of incentive compensation (e.g., Phillips, 2003;

Armstrong et al., 2012; Rego and Wilson, 2012).

There is little research that directly examines whether (or how) corporate governance

affects tax avoidance. As discussed in the previous section, Desai and Dharmapala (2006)

develop and test a model that links managers’ equity-based compensation to aggressive tax

avoidance. They conjecture the existence of complementarities between tax-sheltering and rent
 
1
A related set of studies attempts to identify and validate measures of tax avoidance. For example, Frank et al.
(2009), Wilson (2009), Lisowsky (2010), and Dyreng et al. (2008) develop alternative measures of tax avoidance.
Since the passage of the Financial Accounting Standards Board (FASB)’s Interpretation No. 48 (Accounting for
Uncertainty in Income Taxes) (hereafter FIN 48), several recent papers measure firms’ tax aggressiveness using the
magnitude of their uncertain tax benefits (e.g., Rego and Wilson, 2012; Lisowsky et al., 2013). As we discuss in
more detail below, we carefully consider the specific attribute of tax avoidance that is called for by our research
question (Hanlon and Heitzman, 2010).

 


extraction. The essence of their argument is that managers of well-governed firms will have

greater incentives for tax avoidance because the presence of other governance mechanisms will

prevent these managers from extracting the rents that are generated from their tax avoidance

activities. In contrast, managers of poorly-governed firms will not have incentives for aggressive

tax avoidance because the lack of monitoring and oversight would otherwise allow these

managers to extract the rents that are generated from their aggressive tax planning.

Recent work questions several of the fundamental assumptions in Desai and

Dharmapala’s model. For example, their model assumes that managers can extract rents

generated by tax avoidance because operational complexity (and the accompanying information

asymmetry) results in a more opaque information environment and therefore lowers the cost and

expands the scope for rent extraction. However, Gallemore and Labro (2014) find that tax

avoidance is associated with higher quality (internal) information environments and the precise

channels through which managers extract (or personally benefit from) the rents that are

generated from tax avoidance are not clear. Moreover, there is limited empirical evidence that

managers do, in fact, extract rents that are generated by tax avoidance.2 Desai and Dharmapala

(2006) also implicitly assume that equity-based compensation does not provide a direct,

mechanical tax benefit in the form of a tax shield. Seidman and Stomberg (2011) challenge this

assumption and find that firms with higher levels of equity-based compensation are less likely to

benefit from additional tax avoidance. Seidman and Stomberg (2011) suggest that Desai and

 
2
Desai et al. (2007) present evidence that Russian oligarchs appear to extract meaningful rents from firms that avoid
more taxes. However, the authors do not find evidence that this is the case for Russian firms that operate in
regulated (e.g., U.S.) markets. Blaylock (2011) also fails to find evidence that managers of U.S. firms extract
economically meaningful rents through tax avoidance.





Dharmapala’s association between equity compensation and tax avoidance can be explained by

“tax exhaustion.”3

In a related concurrent study, Robinson et al. (2012) also examine the association

between incentives and governance for both general and “risky” forms of tax avoidance.4

Robinson et al. report that the proportion of accounting experts on the board is associated with

more general and less “risky” tax planning. To the extent their measure of “risky” tax planning

explains variation in relatively extreme levels of tax avoidance (i.e., the right tail of the tax

avoidance distribution), our findings regarding the relation between board financial expertise and

tax avoidance are consistent with those of Robinson et al. (2012). However, one important

difference between our study and Robinson et al. is that we also examine the effect of

governance and incentives on tax avoidance across the entire tax avoidance distribution, which

provides better insight into the link between governance and tax avoidance. For example, firms

in the left tail of the (conditional) tax avoidance distribution are those that engage in less tax

avoidance than expected (i.e., “under-shelter”), and constitute an important and unresolved

puzzle in the tax literature.5 Our findings also shed light on the relation between various

governance mechanisms and tax avoidance of these firms.

In summary, although extant literature provides some insight into the role of incentives

on tax avoidance, inferences are still limited regarding whether (and how) corporate governance
 
3
Graham et al. (2006) suggest that firms require fewer alternative tax shelters if they utilize more stock option
grants. This “tax exhaustion” occurs because option exercises provide the corporation with tax deductions, which
reduces the need to engage in alternative tax avoidance strategies. The same reasoning can also be found in the
equity incentives literature. For example, Core and Guay (1999, 159) argue that “when future corporate tax rates are
expected to be higher, the future tax deduction from deferred compensation becomes more favorable relative to the
immediate tax deduction received from cash compensation.”
4
Robinson et al. (2012) measure risky tax planning using estimates of tax shelter likelihood derived from the level
of activity in tax havens (Dyreng and Lindsey, 2009; Balakrishnan et al., 2012) and the shelter estimation models
described in Wilson (2009) and Lisowsky (2010). They also use the predicted uncertain tax benefits as an additional
proxy for firms’ involvement in risky tax planning following Cazier et al. (2009) and Rego and Wilson (2012).
5
The lack of evidence of significant costs of tax avoidance is frequently referred to in the literature as the “under-
sheltering puzzle” (e.g., Desai and Dharmapala, 2006; Weisbach, 2002; Hanlon and Heitzman, 2010; and Gallemore
et al., 2012).




influences firms’ tax avoidance. Moreover, the evidence that does exist is confined to explaining

the conditional mean of the tax avoidance distribution, but does not describe the relation in other,

potentially more interesting and informative parts of the distribution—most notably in the tails.

2.2. Research hypotheses

2.2.1. Managerial incentives

The majority of most CEOs’ monetary incentives stem from changes in the value of their

equity holdings. Prior studies discuss how the sensitivity of a CEO’s equity portfolio to changes

in stock price, or delta, has two opposing effects on their risky project selection decisions. On

one hand, since delta captures the sensitivity of a CEO’s wealth to stock price, it encourages

CEOs to take risks that are expected to generate a sufficient increase in stock price. On the other

hand, Armstrong et al. (2013) note that delta also “amplifies the effect of equity risk on the total

riskiness of the manager’s portfolio,” which discourages risk-averse managers from pursuing

risky projects. As a result of these two opposing effects, the net incentive effect of delta is

theoretically ambiguous for risky projects. If CEOs believe that the increase in stock price will

more than offset the accompanying increase in equity risk associated with aggressive tax

positions, then delta should encourage tax avoidance. Alternatively, if CEOs believe that the risk

to their equity wealth exceeds any increase in stock price from adopting aggressive tax positions,

then delta should discourage tax avoidance. Because of these opposing effects, it is not clear

whether CEOs’ equity portfolio delta will be positively or negatively associated with tax

avoidance across the distribution.

Unlike delta, the sensitivity of a CEO’s equity portfolio value to changes in stock return

volatility, or vega, provides them with an unambiguous incentive to take risk.6 Therefore, if

 
6
Armstrong and Vashishtha (2012) develop a numerical example that shows that although equity portfolio vega
provides risk-averse CEOs with incentives to take both systematic and idiosyncratic risk, the incentive to take




CEOs believe that more aggressive tax avoidance increases stock price volatility, we expect vega

to have a positive relation with tax avoidance. Moreover, the magnitude of this relation should be

higher (lower) in the right (left) tail of the tax avoidance distribution where there is thought to be

more (less) risk associated with the tax avoidance. This prediction is similar to that in Rego and

Wilson (2012), except that we also predict a stronger relation at higher levels—and, conversely,

a weaker relation at lower levels—of the tax avoidance distribution.

2.2.2. Board characteristics

Similar to other activities that entail both costs and benefits, the effect of tax avoidance

on firm value should be concave with an interior optimum. In particular, there are likely to be

positive net benefits (e.g., cash savings) from engaging in tax avoidance up to some firm-specific

optimal level. Beyond this point, the marginal cost of additional tax avoidance (e.g., costs related

to structuring complicated tax transactions, an inability to repatriate and invest foreign earnings,

and potential political, regulatory, or reputational costs that are detrimental to future operations)

exceeds the marginal benefit.7 If shareholders and managers have different preferences for tax

avoidance then governance mechanisms will be used to influence managers’ tax avoidance

decisions. For example, certain governance mechanisms may prevent (or mitigate) over- and

under-investment in tax avoidance. Although a firm’s governance structure is comprised of a

complex set of contracts, relationships, and institutional features, we focus on governance

mechanisms that we believe are closely related to tax decisions. Specifically, we focus on the

    

systematic risk is stronger than the incentive to take idiosyncratic risk. Our analysis does not distinguish between the
nature of the risk (i.e., systematic versus idiosyncratic) associated with tax avoidance and does not rely on the
differential strength of risk-taking incentives provided by vega. Our analysis instead relies on vega providing CEOs
with incentives to take both types of risk.
7
For example, some studies suggest that firms incur potential tax penalties and reputational costs (e.g., Chen et al.,
2010) or face significant costs defending aggressive tax positions (e.g., Rego and Wilson, 2012). Although there is
little direct evidence supporting the existence of explicit reputational costs (e.g., Gallemore et al., 2012; Hanlon and
Slemrod, 2009), there is clearly some underlying cost of extreme tax avoidance because not all firms have extremely
low effective tax rates.




financial sophistication and independence of the board. We expect that more financially

sophisticated boards will be better able to monitor their firm’s tax positions. This prediction is

consistent with recent guidance that recommends that tax issues, including implementing and

monitoring tax planning, should be placed on the audit committee’s agenda and, more generally,

advocates “greater awareness in the Boardroom of the importance of tax issues.”8 Accordingly,

we predict a positive (negative) relation between the financial sophistication of boards and tax

avoidance in the left (right) tail of the tax avoidance distribution. The extent to which

independent directors affect a firm’s tax policy is not ex ante clear. Independent directors may

not have sufficient firm-specific expertise to affect firm tax policy. Alternatively, they may be

able to draw on outside experience with other firms’ tax positions and therefore be more likely to

influence tax planning activities. Independent directors should recognize that there are potential

costs associated with extreme tax positions and, consequently, should attempt to mitigate

extreme tax avoidance. If independent directors monitor managers’ tax positions or are otherwise

sensitive to extreme tax avoidance, then we expect to find a positive (negative) relation between

board independence in the lower (upper) tail of the tax avoidance distribution.

3. Sample Selection

Our sample selection starts with all firms listed on Compustat for the 2007-2011 fiscal

years for which we have data to compute at least one of the tax attributes that we define below.9

We eliminate foreign registrants and firms designated as real estate investment trusts (REITs)

because these firms are subject to different tax rules. We also eliminate firms that have an annual

 
8
See Jeffrey Owens, Good Corporate Governance: the Tax Dimension—OECD Forum on Tax Administration,
September 2006 and Deloitte Hot Topics: Taxes: What the audit committee should know, October 2011.
9
As recommended by Lisowsky et al. (2013), we delete observations with missing Compustat data for
ENDFIN48BAL. We also truncate ETR observations to be between zero and one. Because we use three-year ETRs,
very few observations are truncated.




average stock price of less than $1.00 per share and firms with average total assets of less than

$10,000. These requirements yield 12,275 firm-year observations. We then retain firm-year

observations for which we have data available for our control variables (defined below). This

yields 7,231 firm-year observations. Finally, we retain firm-years for which we have data

available for our governance and incentives variables.10 This yields a final sample of between

3,137 and 4,128 firm-year observations depending on the measure of tax avoidance.

Table 1 Panel A provides descriptive statistics for all of the variables used in our analysis

including measures of tax avoidance, governance, incentives, and the control variables. Table 1

Panel B compares characteristics of our sample to those of the Compustat population for fiscal

year 2009, which is the year that has the largest representation in our sample. Panel B shows that

the firms in our sample are represented in each of the Barth et al. (1998) industry groups and, on

average, are larger and more profitable than those in the broader Compustat population.

4. Research Design

Since our hypotheses relate to the extreme tails of the tax avoidance distribution, our

primary statistical tests and inferences are based on quantile regression estimates. Quantile

regression allows us to draw more complete inferences beyond those that can be drawn from

traditional ordinary least squares (OLS) regressions, which only describe the relation between

independent variables and the conditional mean of the dependent variable of interest. Quantile

regression is more general and describes the relation between the independent variables and any

specified percentile of the conditional distribution of the dependent variable. In describing the

 
10
We obtain governance and incentives data from Equilar, which is similar to the ExecuComp database in that it
provides executive compensation and equity holdings data collected from annual proxy filings (Form DEF 14A)
with the SEC. We use Equilar data because it provides more than twice as many annual observations as
ExecuComp.




advantages of quantile regression, Hao and Naiman (2007) note that “the focus on the central

location has long distracted researchers from using appropriate and relevant techniques to

address research questions regarding noncentral locations on the response distribution. Using

conditional-mean models (e.g., OLS regression) to address these questions may be inefficient or

even miss the point of the research altogether. …A set of equally spaced conditional quantiles

can characterize the shape of the conditional distribution in addition to its central location.”

The essential features of quantile regression can be illustrated through the following

example. Suppose that a researcher is interested in determining whether education affects pay to

help inform public policy regarding subsidized education. Further suppose that estimates from an

OLS regression of pay on the level of education (e.g., years of schooling) indicate a small

positive association between education and pay, leading the researcher to infer that the average

effect of education on pay is modest. The policy recommendation based on inferences from OLS

might be that further education investment generates only modest pay benefits. Quantile

regression estimates at the conditional 5th percentile of the pay distribution might instead reveal a

much larger association between education and pay, suggesting that the returns to education are

much larger for low-paid workers, who are more likely to receive educational subsidies.

In a related application, Eide and Showalter (1998) use quantile regression to determine

whether the relation between school quality and standardized test performance differs across the

conditional distribution of test performance. Their results indicate that per pupil expenditures “is

a variable that is generally found to be insignificant in most regressions which focus on the mean

effect. In the quantile regressions, the coefficient is … insignificant at the 0.25 quantile and

higher, but is relatively large and significant for the bottom tail of the distribution suggesting that

these expenditures may increase math scores for the lower part of the conditional distribution.”




Quantile regression has also been applied in finance and economics to assess the effects of

401(K) participation on wealth (Chernozhukov and Hansen, 2004); the determinants of house

prices (Zietz et al., 2008); the determinants of gender wage differences (Garcia et al., 2001); the

effect of education on women’s labor market value (Buchinsky, 2002); and to evaluate value-at-

risk models (Gaglianone et al., 2011).

In our research setting, quantile regression allows us to determine whether the relation

between various governance characteristics and tax avoidance varies across the tax avoidance

distribution. As discussed in Section 2, we expect that more financially knowledgeable and more

independent boards will more actively engage managers about their tax choices when the level of

tax avoidance departs from the average. Thus, any relation between managerial incentives or

corporate governance and tax avoidance should be more pronounced in the tails of the tax

avoidance distribution. From a methodological perspective, this prediction implies that the

impact of managerial incentives and related governance mechanisms will alter the shape of the

entire tax avoidance distribution, rather than simply shift its central location as described by the

mean or median. Consistent with some of the results from prior studies, it is possible that the

impact of governance is slight or negligible at the mean or median, yet strong at other points of

the tax avoidance distribution. Since traditional OLS regression estimation methods do not detect

anything other than a shift in central location (i.e., the conditional mean), we use quantile

regression as our primary method of estimation.

The basic intuition for quantile regression in the context of our hypotheses is illustrated in

Figure 1. Conditioning on a particular governance mechanism may “rotate” the conditional tax

avoidance cumulative distribution function (CDF) around the median in a counterclockwise

manner. Such an effect would manifest as a negative (positive) coefficient on governance in the




 right (left) tail of the tax avoidance distribution. However, in this example, the coefficient on

governance at the median (i.e., the central location) would be zero. Since traditional OLS and

median regression only estimates the relationship at the “center” of the distribution (i.e., the

mean and median, respectively), these techniques cannot detect shifts elsewhere in the

distribution of interest and would not detect the shifts illustrated in Figure 1. In contrast, quantile

regression is more general and describes changes in both the location and shape of the

distribution of interest.11

4.1. Quantile regression specification

To test our research hypotheses, we estimate the following specification using quantile

regression.12

TaxPositioni,t = β0i,t + β1LogNumFinExpi,t-1 + β2PctIndepi,t-1 +

β3LogNumDirsi,t-1 + β4LogCEOPortDeltai,t-1 +
β5 LogCEOPortVegai,t-1 + β6CFOpsi,t +
β7LogMVEi,t + β8LogForAssetsi,t +β9GeoCompi,t + İi,t, (1)

TaxPosition is one of two proxies that measure a firm’s level of tax avoidance for a given year.

Our first proxy for tax avoidance is EndFin48Bal, which we measure as the ending balance of

the firm’s uncertain tax benefit (“UTB”) account (Compustat item TXTUBEND) scaled by total

assets. The UTB represents management’s estimate of the amount of tax savings generated by

 
11
To better understand the technical aspects of quantile regression, recall that OLS regression minimizes the
following loss function:
  


  
 . Quantile regression is similar in that it minimizes an objective
function with different weights for different quantiles. To illustrate, the special case of regression at the 50th
percentile (i.e., median regression) is described by the following objective:
 
 
  
. More
generally, to estimate the association at any other specified percentile of the Y distribution, quantile regression
minimizes the following objective with asymmetric weights that are determined by the parameter c:

 
  
     where is the quantile (i.e., percentile) of interest, u is the error estimate,
 ! "  #$! % &' ( #  #$! ) &'! "   #$! ) &'!*and 1[*] is the “indicator function” that equals one
if the bracketed condition regarding u is true and zero otherwise. (See Imbens, Guido W. and Jeffrey M.
Wooldridge, 2007, Lecture Notes 14, Summer ’07, available at http://www.nber.org/WNE/lect_14_quantile.pdf, for
a further discussion.)
12
Following Koenker and Hallock (2011), we both tabulate and graph the coefficient estimates at decile intervals to
show the relationship across the entire support of the tax avoidance distribution. We also report OLS coefficient
estimates of the conditional mean for comparative purposes.




tax planning that is potentially payable to the tax authorities upon audit. Because the UTB

represents an aggregation of multiple tax positions, it is highly visible to the board. In addition,

directors can easily compare a firm’s UTB to that of its peers to gauge whether management is

over- or under-investing in tax planning.

Our second proxy for tax avoidance is TAETR, which is the difference between the firm’s

three-year average GAAP effective tax rate (hereafter, ETR, computed as the firm’s total tax

expense scaled by pre-tax income) and the three-year average GAAP ETR of the firm's size and

industry peers (i.e., those in the same quintile of total assets in the same Fama-French 48

industry). This measure of tax avoidance captures cross-sectional variation in firms’ total tax

planning (including both timing and permanent differences), and benchmarks a given firm’s tax

aggressiveness relative to that of similar-sized firms in the same industry (see Balakrishnan et al.,

2012). Similar to the UTB, a firm’s ETR is a highly visible summary measure of tax planning

that the board can monitor and evaluate. By comparing a firm’s ETR to that of its peers, we are

effectively performing a cross-sectional comparison that the board could make to determine

whether the firm is investing too much or too little in tax planning activities.13

LogNumFinExp is the natural logarithm of one plus the number of financial experts on

the board in the previous year (as indicated by RiskMetrics). We use the number rather than the

proportion of financial experts on the board because we believe that the former better captures

how the tax expertise of the board as a whole is determined by a subset of specialists.14 For

example, we believe that a board with two financial experts out of ten total members is more

 
13
One potential concern is that both of our measures of tax avoidance include accounting accruals. Accordingly, an
alternative explanation of our results is that the boards do not monitor tax planning, but rather managers’ accounting
decisions. Although this explanation is consistent with our results, it is also requires that boards have detailed
knowledge of their firm’s tax positions to ascertain whether management’s accounting choices are conservative (in
the right tail) or aggressive (left tail).
14
We observe similar results (untabulated) when we use the proportion rather than the number of financial experts
on the board.

 


likely to have more tax expertise than a board with one financial expert out of five total

members. The proportion of financial experts, which is a common measure of boards’ financial

expertise, would treat these two scenarios as equivalent. However, we believe that the board has

a better pool of expertise in the former scenario.

PctIndep is the proportion of independent directors in the previous year (as indicated by

Equilar). LogNumDirs is the natural logarithm of one plus the total number of directors on the

board in the previous year (as indicated by Equilar). LogCEOPortDelta is the natural logarithm

of the (risk-neutral) dollar change in the CEO’s equity portfolio value for a 1% increase in stock

price (Core and Guay, 2002) during the previous year. As discussed earlier, portfolio delta

measures the change in the CEO’s wealth that relates to changes in stock price.

LogCEOPortVega is the natural logarithm of one plus the (risk-neutral) dollar change in CEO

equity portfolio value for a 0.01 increase in annual stock return volatility (Core and Guay, 2002)

during the previous year.15 Portfolio vega measures the change in the CEO’s wealth that relates

to changes in stock price risk (i.e., return volatility).

We also control for economic determinants of firms’ tax positions by including variables

that are common in the tax literature that capture costs, benefits, and opportunities to engage in

tax avoidance. CFOps is cash flow from operations divided by average total assets in the

previous year; LogMVE is the natural logarithm of the firm’s market value of equity in the

previous year; LogForAssets is the natural logarithm of total foreign assets (Oler et al., 2007) in

the previous year; GeoComp is a revenue-based Hirfindahl-Hirschman index that captures

 
15
The value of a CEO’s stock and restricted stock is assumed to change dollar-for-dollar with changes in the price
of the underlying stock. The value of a CEO’s stock options is assumed to change according to the option’s delta
(vega), which is the derivative of its Black-Scholes value with respect to price (return volatility). Annualized return
volatility is calculated using continuously compounded monthly returns during the previous 36 months, with a
minimum of 12 months of returns. The risk-free rate is calculated using the interpolated interest rate on a Treasury
Note that is closest in maturity to the remaining life of the option multiplied by 0.70 to account for the prevalence of
early option exercise. Dividend yield is calculated as the dividends paid during the previous twelve months scaled
by the stock price at the beginning of the month.





within-firm geographic segment complexity (Bushman et al., 2004) in the previous year; and i

and t index the firm and fiscal year, respectively.

Panel C of Table 1 reports the mean values of CEO equity incentives (i.e., CEOPortDelta

and CEOPortVega) for each decile of the unconditional tax avoidance distribution (where tax

avoidance is measured as EndFin48Bal). Table 1 Panel C indicates that CEO equity incentives

tend to increase with the level of tax avoidance. Table 1 Panel D reports Pearson correlations

between the primary variables. Note that tax avoidance is associated with cash flow from

operations, the amount of foreign assets, and the degree of geographic complexity, which

underscores the importance of controlling for these factors in our multivariate specifications.

5. Results

Table 2 and Figure 2 present our primary results regarding the relation between corporate

governance—including CEO equity incentives—and tax avoidance. We first consider the

relation between CEO equity incentives and tax avoidance. OLS estimates of the conditional

mean of tax avoidance presented in Panel A of Table 2 provide no evidence of a relation between

TaxPosition and LogCEOPortDelta. In particular, the estimated coefficient is 0.0001 (t-stat of

0.46) when tax avoidance is measured with EndFin48Bal and is -0.0035 (t-stat of -1.56) when

tax avoidance is measured as TAETR. Consistent with Rego and Wilson (2012), OLS estimates

provide evidence of a positive relation between TaxPosition and LogCEOPortVega, particularly

when tax avoidance is measured with EndFin48Bal (coefficient of 0.0004 and t-stat of 4.44).

Figure 2 Panel A plots the quantile regression coefficient estimates from Table 2 Panel A.

In general, the patterns indicate that the relation between CEOs’ equity incentives and tax

avoidance is generally positive and increasing in magnitude in the right tail of the distribution.




This pattern is particularly pronounced for LogCEOPortVega, which measures CEOs’ equity

risk-taking incentives. Tests of differences in coefficients across the quantiles when tax

avoidance is measured as EndFin48Bal indicate that the coefficient at the 90th percentile is

significantly larger than the coefficient at the 50th percentile (p-value of 0.002), which, in turn, is

significantly larger than the coefficient at the 10th percentile (p-value of 0.000). This observed

positive relation is consistent with equity incentives being an important determinant of

TaxPosition. Moreover, depending on whether observed equity incentives are “optimal,” the

large differences at the extremes of the tax aggressiveness distribution are consistent with

unresolved agency problems with respect to tax avoidance.

Figure 2 Panel B and Table 2 Panel B both present results for the relation between tax

avoidance and board expertise and independence. We focus first on LogNumFinExp, which is

our proxy for the financial sophistication on the board. We expect that more sophisticated boards

are better able to assess when the firm might be over- or under-investing in tax avoidance, and

will therefore have a greater effect on the firm’s tax avoidance at extreme levels of tax

avoidance. Consequently, we expect to observe a positive relation between LogNumFinExp and

TaxPosition in the left tail and a negative relation in the right tail of the tax avoidance

distribution.

The OLS estimates in Panel B of Table 2 provide no evidence of a statistical relation

between TaxPosition and LogNumFinExp for either tax avoidance measure. The OLS coefficient

is -0.0009 (t-stat of -0.96) when EndFin48Bal is the measure of tax avoidance and is -0.0089 (t-

stat of -1.08) when TAETR is the measure of tax avoidance. In contrast, Figure 2 shows that the

relation between the financial expertise of the board and both measures of tax avoidance differs

across their respective distributions. Specifically, the relation between financial expertise and tax




avoidance is positive in the left tail but negative in the right tail. Tests of coefficient differences

across quantiles indicate that the coefficient at the 90th percentile (t-stat of -1.94) is significantly

more negative than the coefficient at both the 10th (t-stat of 5.58, difference p-value of 0.014) and

the 50th percentiles (t-stat of 1.33, difference p-value of 0.011) when EndFin48Bal is the measure

of tax avoidance. This result is consistent with our hypothesis and indicates that boards’ financial

sophistication does not have a uniform relation with tax avoidance, but that the relation differs

according to the level of tax avoidance.16 Moreover, the quantile regression estimates indicate

that OLS estimates of the conditional mean are not representative of the relation at other points

of the tax avoidance distribution. Therefore, generalizing based on OLS estimates of the

conditional mean provides a misleading picture of the relation between this important

governance mechanism and tax avoidance.

Next we consider PctIndep, which measures board independence and has been used as a

proxy for the amount of board monitoring and oversight of management. Desai and Dharmapala

(2006) suggest that agency problems may arise with respect to tax avoidance which would occur

if shareholders and managers evaluate the costs and benefits of tax avoidance differently. If more

independent boards can better identify and mitigate these agency problems, then we expect to

observe a positive (negative) relation in the left (right) tail of the tax avoidance distribution.

The OLS estimates presented in Panel B of Table 2 provide no evidence of a relation

between TaxPosition and PctIndep. The coefficient is -0.0014 (t-stat of -0.46) when TaxPosition

is measured with EndFin48Bal and 0.0103 (t-stat of 0.37) when TaxPosition is measured with

TAETR. However, similar to our results for LogNumFinExp, the quantile regression coefficient

estimates in Table 2 and the corresponding graph in Figure 2 provide some evidence that the
 
16
This result is consistent with the hypothesis that sophisticated board members affect tax policy decisions at
extreme levels of tax avoidance. An alternative inference is that sophisticated directors do not affect tax policy
decisions, but instead affect the tax accruals that are reported by managers.




relation between board independence and tax avoidance differs across the tax avoidance

distribution. Tests of coefficient differences across quantiles indicate that the coefficient at the

90th percentile (t-stat of -2.28) is significantly more negative than the coefficient at the 50th

percentile (t-stat of 2.01, difference p-value of 0.005) when TAETR is the TaxPosition proxy.

This pattern is consistent with our hypothesis that more independent boards alleviate over- and

under-investment in tax avoidance, which is one potential symptom of unresolved agency

problems with respect to tax planning.

Although the results in Table 2 and Figure 2 provide insight into the relation between

firms’ governance attributes and tax avoidance, we acknowledge that our inferences are limited

because variation in firms’ governance structures may not be exogenous with respect to their

level of tax avoidance. Therefore, we cannot rule out the possibility that our inferences are

confounded by reverse causality or more general concerns related to correlated omitted variables.

We considered several ways to address these concerns. First we considered quantile regression

with instrumental variables (e.g., Frolich and Melly, 2012; Kaplan and Sun, 2013; Powell, 2013).

However, there is no obvious instrument that satisfies the necessary conditions of being both

“relevant” (i.e., induces sufficient variation in firms’ governance structures) and “valid” (i.e.,

satisfies the exclusion restriction that requires the instrument to affect firms’ tax avoidance only

through its effect on their governance structures). Similarly, we are not aware of a suitable

exogenous shock or “natural experiment” during our sample period that might allow for

identification of causal effects.

Regarding the potential for reverse causality, it could be the case that firms’ governance

structures are designed primarily to influence tax avoidance activity. We believe that tax

planning is one of many factors that is considered when selecting the design and composition of




the board and is therefore, at most, a second-order concern.17 Moreover, board composition and

structure is typically quite stable over time and would therefore be unlikely to change—or at

least change quickly—in response to changes in the firm’s tax position. Nevertheless, it is

possible that firms with more aggressive tax positions select more financially sophisticated

directors specifically to oversee these tax positions. We believe that this scenario is unlikely

because it predicts not only a positive relation between LogNumFinExp and TaxPosition at both

high and low levels of tax avoidance, but also a larger positive relation for high levels of tax

avoidance where financially sophisticated directors would be most beneficial. Our results are

inconsistent with this scenario.

We also believe that our research design helps to establish the nature of causality because

we are not aware of an obvious candidate for an omitted correlated variable that would induce a

positive relation between LogNumFinExp and TaxPosition for low levels of tax avoidance, but a

negative relation for high levels of tax avoidance. Nevertheless, as with every observational

study, our inferences are inherently descriptive.

6. Supplemental Analysis

6.1. Sensitivity analysis: alternative measures of investment

As previously discussed, tax avoidance can be view as one of many possible alternative

risky investment opportunities. If this characterization is accurate, other types of investment

(e.g., capital expenditures and research and development) should exhibit relations that are similar

to the one that we document between governance and tax avoidance. Specifically, the relation

between board financial sophistication and other types of investment should differ across the

 
17
This belief is supported by the recent guidance cited above (see supra note 8) that advocates “greater awareness in
the Boardroom of the importance of tax issues.”




investment distribution. Consistent with this conjecture, the estimated coefficients from quantile

regressions (untabulated) of CAPEX and R&D exhibit patterns that are similar to those reported

for Tax Avoidance in Figure 2.18 This finding suggests that board financial sophistication and

independence may also ameliorate agency problems that are associated with extreme levels of

these other risky investment decisions.

6.2. Sensitivity analysis: evidence of mean reversion in extreme tax avoidance

If financially sophisticated boards detect and mitigate over- and under-investment in tax

avoidance, we expect to observe subsequent changes in tax avoidance. To test this conjecture, we

sort EndFin48Balt-1 into quintiles and examine whether there is a predictable change in the next

period’s ending FIN 48 balance (∆EndFin48Balt). In other words, we examine whether firms

with relatively high (low) levels of tax avoidance subsequently reduce (increase) tax avoidance.

We then investigate whether this effect—if it exists—is more pronounced for firms that have

more financial experts and a greater percentage of independent directors on the board.

In untabulated analysis, we find that firms in the lowest quintile of EndFin48Balt-1

experience an increase of 0.0015 in the following year when they have at least three financial

experts on the board. We also find that firms with two or fewer financial experts experience a

smaller—albeit not statistically different—increase of 0.0008 in the following year. Conversely,

we find that firms in the highest quintile of EndFin48Bal t-1 experience a decrease of -0.0054 in

the following year when they have at least three financial experts on the board, and a

significantly smaller decrease of -0.0011 when there are two or fewer financial experts on the

board.

 
18
We do not observe a similar pattern when ROA is the dependent variable, which suggests that we are not simply
documenting quantile regression coefficient patterns that are spurious or mechanically determined by a particular
research design choice.




 In the lowest quintile of EndFin48Bal t-1, we also observe a change of 0.0010 (0.0009) in

the following year when the percentage of independent directors on the board is greater (less)

than the sample median. These changes are not statistically different from each other. In the

highest quintile, we observe a -0.0028 (-0.0017) change in the following year when the

percentage of independent directors on the board is greater (less) than the sample median. The

difference between these changes is marginally significant. Collectively, these results are

consistent with our conjecture that more financially sophisticated and more independent boards

are associated with attenuated changes in relatively extreme levels of tax avoidance.19

6.3. Re-examining Desai and Dharmapala (2006)

Desai and Dharmapala (2006) present evidence that their measures of governance and

equity-based compensation have an interactive effect on the average level of tax aggression.

Specifically, they provide evidence (in their Table 4) that there is no statistical relation between

tax aggressiveness and equity incentives—measured as the annual value of stock option grants

scaled by total compensation of the firm’s top five executives—in “well governed” firms (i.e.,

firms with low scores on a governance index).20 However, they report a significant negative

relation between tax aggressiveness and equity-based compensation in “poorly governed” firms.

Desai and Dharmapala (2006) discuss why it is ambiguous whether equity-based

compensation provides incentives for managers to engage in more or less aggressive tax

strategies. They also discuss how it is reasonable to expect that the incentives for tax

aggressiveness provided by equity-based compensation depend on the firm’s other governance

 
19
We also attempt to assess the relation between changes in the number of financial experts on the board and
subsequent changes in EndFin48Bal. However, there is little variation in the former measure, and the resulting test
lacked sufficient power.
20
Desai and Dharmapala (2006) measure tax aggressiveness as the residual from a regression of the difference
between book and taxable income on total accruals (see their discussion, pp. 159-160). They also rely on the
Gompers et al. (2003) “G-index”, which is primarily a measure of shareholder rights with respect to takeovers, as
their measure of governance quality. We believe that there are more relevant measures of governance quality
available, particularly as it pertains to firms’ tax avoidance.




mechanisms, which can provide either reinforcing or countervailing incentives in the case of

complements and substitutes, respectively. However, they neither model these complex

interactions in their research design nor consider whether the relation might differ across the tax

avoidance distribution. We estimate Desai and Dharmapala’s (2006) primary empirical

specification (i.e., their Table 4 Column (4)) using quantile regression to determine whether the

relation that they document for the conditional mean is representative of the relation elsewhere in

the distribution or whether the relation instead differs across the distribution, as our earlier

findings suggest.

Figure 3 and Table 3 provide OLS and quantile regression estimates of Desai and

Dharmapala’s specification for our sample. Our OLS estimates are generally similar to those

reported in Desai and Dharmapala. Importantly, we are able to replicate their result of a

(marginally significant) negative coefficient on their measure of equity-based compensation and

a positive (but insignificant) coefficient when this variable is interacted with an indicator for

whether the firm is “well governed.” Quantile regression estimates, however, show that the

interaction between equity-based compensation and “good governance” differs—and,

importantly, leads to different inferences—across the tax avoidance distribution. Specifically, we

find that the relation is negative in the right tail of the tax avoidance distribution, which suggests

that both equity incentives and certain governance mechanisms may, in fact, attenuate aggressive

tax positions. These results and the corresponding inferences are inconsistent with Desai and

Dharmapala’s conclusion, but are in line with our results reported in Figure 2 and Table 2.

Moreover, these results further illustrate the importance of examining how governance relates to

tax avoidance across its entire distribution to gain a more complete understanding of the nature

of their relationship.




7. Conclusion

We examine the relation between corporate tax avoidance and corporate governance

characteristics and managers’ equity incentives. We expand the scope of prior research by

estimating the relation not only at the conditional mean, but also across the entire tax avoidance

distribution. Our specific research question relates to the tails of the tax avoidance distribution,

which represent relatively extreme levels of tax avoidance. The ability to document shifts in the

relation is particularly important in our research setting because certain governance mechanisms

are likely to exhibit different relations with tax avoidance at different points in the distribution—

especially if the net benefits of tax avoidance differ at different levels of tax avoidance.

Consistent with the hypothesis that managers expect greater personal benefits from

increased tax avoidance (e.g., Rego and Wilson, 2012), we find evidence that risk-taking equity

incentives are positively related to tax avoidance and that this relation is stronger in the right tail

of the tax avoidance distribution. This result is consistent with the notion that relatively high

levels of risk-taking equity incentives have the potential to motivate managers to invest in risky

tax avoidance beyond the level that is desired by shareholders. We also find evidence that board

financial sophistication and independence exhibit a positive (negative) relation with tax

avoidance in the left (right) tail of the tax avoidance distribution. This finding is consistent with

the hypothesis that more financially sophisticated and more independent boards mitigate agency

problems related to relatively extreme levels of tax aggressiveness.

Collectively, our results provide yield a more nuanced understanding of how managerial

incentives and certain corporate governance mechanisms are related to the level of corporate tax

avoidance. Our results are also consistent with the notion that, similar to other investments that

 


entail a risk-return tradeoff (i.e., costs and benefits), the optimal level of tax avoidance is

unlikely to be at the “corner.” Instead, the optimal level of tax avoidance is more likely to occur

at an interior point at which the marginal costs and benefits (to managers) equate.





References
Angrist, J. D. and J. S. Pischke, 2009. Mostly harmless econometrics. Princeton University Press.
Armstrong, C., Blouin, J., D. Larcker, 2012. The incentives for tax planning. Journal of Accounting and
Economics 53(1): 391-411.
Armstrong, C., Jagolinzer, A. D., and D. F. Larcker, 2010. Chief Executive Officer Equity Incentives and
Accounting Irregularities. Journal of Accounting Research 48: 225-271.
Armstrong, C., Guay, W. R., and J. P. Weber, 2011. The role of information and financial reporting in
corporate governance and debt contracting. Journal of Accounting and Economics 50: 179-234.
Armstrong, C., Larcker, D. F., Ormazabal, G., and D. J. Taylor, 2013. The relation between equity
incentives and misreporting: the role of risk-taking incentives. Journal of Financial Economics 109:
327 – 350.
Armstrong, C., and R. Vashishtha, 2012. Executive stock options, differential risk-taking incentives, and
firm value. Journal of Financial Economics 104: 70-88.
Balakrishnan, K., Blouin, J., and W. Guay, 2012. Does tax aggressiveness reduce corporate transparency?
Working paper, University of Pennsylvania.
Barth, M., Beaver, W., and W. Landsman, 1998. Relative valuation roles of equity book value and net
income as a function of financial health. Journal of Accounting and Economics 25: 1-34.
Bebchuk, L., and J. Fried, 2006. Pay without performance. The unfulfilled promise of executive
compensation. Harvard University Press, Cambridge.
Blaylock, B., 2011. Do managers extract economically significant rents through tax aggressive
transactions? Working paper, Oklahoma State University.
Buchinsky, M., 2002. Quantile regression with sample selection: Estimating women’s return to education
in the U.S. Economic Applications of Quantile Regression, Physica-Verlag HD.
Bushman, R., Chen, Q., Engel, E., and A. Smith, 2004. Financial accounting information, organizational
complexity and corporate governance systems. Journal of Accounting and Economics 37: 167-201.
Cazier, R., Rego, S., Tian, X., and R. Wilson, 2009. Did FIN 48 limit the use of tax reserves as a tool for
earnings management? Working paper, Texas Christian University.
Chen, K. P. and C. Y. C. Chu, 2005. Internal control vs. external manipulation: a model of corporate
income tax evasion. RAND Journal of Economics 36: 151-164.
Chen, S., Chen, X., Cheng, Q., and T. Shevlin, 2010. Are family firms more tax aggressive than non-
family firms? Journal of Financial Economics 95: 41-61.
Chernozhukov, V., and C. Hansen, 2004. The effects of 401(K) participation on the wealth distribution:
An instrumental quantile regression analysis. The Review of Economics and Statistics 86: 735-751.
Cloyd, C. B., Pratt, J., and T. Stock, 1996. The use of financial accounting choice to support aggressive
tax positions: public and private firms. Journal of Accounting Research 34: 23-43.
Core, J.E., and W. R. Guay, 1999. The use of equity grants to manage optimal equity incentive levels.
Journal of Accounting and Economics 28: 151-184.
Core, J. E., and W. R. Guay, 2002. Estimating the value of employee stock option portfolios and their
sensitivities to price and volatility. Journal of Accounting Research 40: 613-630.
Crocker, K. and J. Slemrod, 2005. Corporate tax evasion with agency costs. Journal of Public Economics
89: 1593-1610.
Demsetz, H. and K. Lehn, 1985. The structure of corporate ownership: causes and consequences.
Journal of Political Economy 93: 1155-1177.
Desai M. A., and D. Dharmapala, 2006. Corporate tax avoidance and high-powered incentives. Journal of
Financial Economics 79: 145-179.




Desai M. A., and D. Dharmapala, 2008. Taxation and corporate governance: An economic approach. Max
Planck Institute Studies on Intellectual Property, Competition and Tax Law, (ed. Wolfgang Schon),
Volume 3, 13-30.
Desai M. A., and D. Dharmapala, 2009. Corporate tax avoidance and firm value. Review of Economics
and Statistics 91(3), 537-546.
Desai, M., and A. Dyck, and L. Zingales. 2007. Theft and taxes. Journal of Financial Economics 84, 591-
623.
De Waegenaere, A., R. Sansing, and J. Wielhouwer. 2010. Financial accounting measures of tax reporting
aggressiveness. Working paper, Tilburg University.
Dyreng, S., M. Hanlon and E. Maydew, 2008. Long-run corporate tax avoidance. The Accounting Review
83: 61-82.
Dyreng, S., and B. Lindsey, 2009. Using financial accounting data to examine the effect of foreign
operations located in tax havens and other countries on U.S. multinational firms’ tax rates. Journal of
Accounting Research 47: 1283-1316.
Dyreng, S., M. Hanlon and E. Maydew, 2010. The effects of managers on corporate tax avoidance. The
Accounting Review 85: 1163-1189.
Eide, E. and M. Showalter, 1998. The effect of school quality on student performance: A quantile
regression approach. Economics Letters 58: 345-350.
Fich, E. M. and A. Shivdasani, 2006. Are busy boards effective monitors? Journal of Finance 61: 689-
724.
Frank, M. M., Lynch, L., and S. Rego, 2009. Tax reporting aggressiveness and its relation to aggressive
financial reporting. The Accounting Review 84: 467-496.
Frolich, M., and B. Melly, 2012. Unconditional quantile treatment effects under endogeneity. Working
paper, Universitat Mannheim.
Gaglianone, W., Lima, L., Linton, O., and D. Smith, 2011. Evaluating value-at-risk models via quantile
regression. Journal of Business & Economic Statistics 29: 150-160.
Gallemore, J., and E. Labro, 2015. The importance of the internal information environment for tax
avoidance. Journal of Accounting and Economics, forthcoming.
Gallemore, J., Maydew, E. L., and J. R. Thornock, 2012. The reputational costs of tax avoidance and the
under-sheltering puzzle. Working paper, the University of North Carolina.
Garcia, J., Hernandez, P., and A. Lopez-Nicolas, 2001. How wide is the gap? An investigation of gender
wage differences using quantile regression. Empirical Economics 26: 149-167.
Gompers, P., J. Ishii, and A. Metrick, 2003. Corporate governance and equity prices. The Quarterly
Journal of Economics February: 107-155.
Graham, J., and A. Tucker, 2006. Tax shelters and corporate debt policy. Journal of Financial Economics
81: 563-594.
Gupta, S. and K. Newberry, 1997. Determinants of the variability in corporate effective tax rates:
evidence from longitudinal data. Journal of Accounting and Public Policy 16: 1-34.
Hanlon, M. and J. Slemrod, 2009. What does tax aggressiveness signal? Evidence from stock price
reactions to news about tax shelter improvement. Journal of Public Economics 93: 126-141.
Hanlon, M. and S. Heitzman, 2010. A review of tax research. Journal of Accounting and Economics 50:
127-178.
Hanlon, M., Mills, L., and J. Slemrod, 2007. An empirical examination of corporate tax noncompliance.
Taxing Corporate Income in the 21st Century, Cambridge: Cambridge University Press.
Hao, L., and D. Q. Naiman, 2007. Quantile regression. Sage publishing.




Harris, M. and A. Raviv, 2008. A theory of board control and size. Review of Financial Studies 21: 1797-
1832.
Ho, D. E., Imai, K., King, G., and E. A. Stuart, 2007. Matching as nonparametric preprocessing for
reducing model dependence in parametric causal inference. Political Analysis 15: 199-236.
Kaplan, D., and Y. Sun, 2013. Smoothed Estimating Equations for Instrumental Variables Quantile
Regression. Working paper, University of Missouri.
Koenker, R., and K. F. Hallock, 2011. Quantile regression. Journal of Economic Perspectives 15: 143-
156.
Larcker, D., and T. Rusticus, 2010. On the use of instrumental variables in accounting research. Journal
of Accounting and Economics 49: 186-205.
Lisowsky, P., 2010. Seeking shelter: Empirically modeling tax shelters using financial statement
information. The Accounting Review 85: 1693-1720.
Lisowky, P., L. Robinson and A. Schmidt, 2013. Do publicly disclosed tax reserves tell us about privately
disclosed tax shelter activity? Journal of Accounting Research 51: 583-629.
Mills, L. and K. Newberry, 2001. The influence of tax and nontax costs on book-tax reporting
differences: public and private firms. Journal of the American Taxation Association 23: 1-19.
Minnick, K. and T. Noga, 2010. Do corporate governance characteristics influence tax management?
Journal of Corporate Finance 16, 703-718.
Oler, M., Shevlin, T., and R. Wilson, 2007. Examining investor expectations concerning tax savings on
the repatriations of foreign earnings under the American Jobs Creation Act of 2004. Journal of the
American Taxation Association 29: 25-55.
Phillips, J. D., 2003. Corporate tax planning effectiveness: the role of compensation-based incentives. The
Accounting Review 78: 847-874.
Powell, D., 2013. A new framework for estimation of quantile treatment effects: nonseparable disturbance
in the presence of covariates. Working paper, Rand Corporation.
Rego, S. O., and R. Wilson, 2012. Equity risk incentives and corporate tax aggressiveness. Journal of
Accounting Research 50: 775-810.
Robinson, J. R., Xue, Y., and M. H. Zhang, 2012. Tax planning and financial expertise in the audit
committee. Working paper, University of Texas at Austin.
Scholes, M. S., Wolfson, M. A., Erickson, M. M., Maydew, E. L., and T. J. Shevlin, 2009. Taxes and
business strategy: a planning approach, Prentice Hall.
Seidman, J. and B. Stomberg, 2011. Why are option compensation and tax sheltering negatively related?
Working paper, University of Texas.
Shevlin, T., 2007. The future of tax research: From an accounting professor’s perspective. Journal of the
American Taxation Association 29: 87-93.
Shevlin, T. and D. A. Shackelford, 2001. Empirical tax research in accounting. Journal of Accounting and
Economics 31: 321-387.
Slemrod, J., 2004. The economics of corporate tax selfishness. National Tax Journal 57:877-899.
Weisbach, D., 2002. Ten truths about tax shelters. Tax Law Review 55: 215-253.
Wilson, R. 2009. An examination of corporate tax shelter participants. The Accounting Review 84(3),
969-999.
Zietz, J., Zietz, E., and G. Sirmans, 2008. Determinants of house prices: A quantile regression approach.
Journal of Real Estate Finance and Economics 37: 317-333.




 Figure 1

Example of the Effect of Governance on the Tax Avoidance Distribution

This figure illustrates the basic intuition for quantile regression in the context of the relation between Governance
and Tax Avoidance. The treatment effect of a particular governance mechanism may “rotate” the conditional tax
avoidance cumulative distribution function (CDF) around the median in a counterclockwise manner. Such an effect
would manifest as a negative (positive) coefficient on governance in the right (left) tail of the tax avoidance
distribution. However, the coefficient on governance at the median (i.e., the central location) would be zero. Since
traditional OLS or median regression only estimates the relationship at the “center” of the distribution (i.e., the mean
and median, respectively), these techniques cannot detect shifts that occur in other locations of the distribution of
interest. In contrast, quantile regression is much more general and is designed to characterize changes in both the
location and shape of the distribution of interest.




 Figure 2
Coefficient Estimates at Various Quantiles of the Tax Avoidance Distribution

Panel A: CEO Incentives

TaxPosition = EndFin48Bal X = LogCEOPortDelta TaxPosition = TAETR X = LogCEOPortDelta
  

 


 


 
 
 
 
 
 


 
 
 
 
 

 
 
 
 

 
 
 
 
 
 
 


 
 

 

 

 
 

TaxPosition = EndFin48Bal X = LogCEOPortVega TaxPosition = TAETR X = LogCEOPortVega

    

 


 

 


 
 
 
 

   
 
 
 
 
 
 


 
 
 

 
 
 
 
 
 
 


 
 

 

 
 




 Figure 2 (cont’d)

Panel B: Board Expertise and Independence

TaxPosition = EndFin48Bal X = LogNumFinExp TaxPosition = TAETR X = LogNumFinExp
   

 


 


 
   
 
 
 
 
 
 


 
 

 
 

 

 
 
 
 
 
 


 
 

 
 

 

 
 

 
 

TaxPosition = EndFin48Bal X = PctIndep TaxPosition = TAETR X = PctIndep

    

 


 


 


 
 
 
 
 
 


 
 
 

 
 
 
 
 
 
 
 


 
 

 
 

 
 

This figure plots quantile regression coefficient estimates at various quantiles of the tax avoidance distribution.
Panels A and B correspond to estimates reported in Panels A and B of Table 2. EndFin48Bal is measured as the
firm’s ending balance of the uncertain tax benefit account (Compustat TXTUBEND), scaled by the firm’s average
assets over the period. TAETR is computed as the mean three-year GAAP ETR (computed as the firm’s total tax
expense scaled by pre-tax income)of the firm's size and industry peers minus the firm's three-year GAAP ETR. Size
peers are firms within the same quintile of total assets and industry peers are firms within the same Fama-French 48
industry portfolios. PctIndep is the percentage of independent directors to total directors sitting on the board in the
previous year (as indicated by Equilar). LogNumFinExp is the natural logarithm of one plus the number of financial
experts designated on the board of directors in the year preceding the fiscal year (as indicated by RiskMetrics).
CEOPortDelta (CEOPortVega) is the (risk-neutral) dollar change in the firm CEO’s equity portfolio value for a 1%
change in the value (volatility) of the firm’s stock price (Core and Guay, 2002). LogCEOPortDelta
(LogCEOPortVega) is the natural logarithm of CEOPortDelta (CEOPortVega).




 Figure 3

Desai and Dharmapala (2006) Quantile Regression

Table 4 (All Firms)
StkMixGrant * WellGov
  
 
 

 


 
 
 
 
 
 
 
 
 
 
 
 
 
 




 
 
 
 

 

 

  

 

 

 

 

Coefficient plot for StkMixGrant * WellGov from quantile regression estimation of Desai and Dharmapala (2006)
Table 4, Column (4). The following equation is estimated (Desai and Dharmapala, 2006, adjusted to include a mean
effect for WellGov):
TSi,t = β0 + β1STKMIXGRANTi,t + β2(WELLGOVt * STKMIXGRANTi,t) + Other Interaction Terms
+ Firm Fixed Effects + Year Fixed Effects + Controls + vi,t. (17)
TS is the residual estimated from regressing the difference between book and tax income on total accruals (see Desai
and Dharmapala, 2006 discussion, p. 159-160). StkMixGrant is the ratio of stock option grant value to total
compensation for the firm’s top five executives. WellGov is a dichotomous variable that equals one for firms that
have a low governance index score (G-score less than or equal to 7) and equals zero otherwise. Estimation includes
controlling for the level of deferred taxes and year, firm, and firm-year size fixed effects.




 Table 1
Panel A: Descriptive Statistics
Standard 25th 75th
n Deviation percentile Mean Median percentile
Tax Position
EndFin48Bal 4,128 0.025 0.003 0.015 0.008 0.018
TAETR 3,137 0.198 -0.104 -0.028 -0.031 0.094
Governance and Incentives
NumFinExp 4,128 1.273 1.000 1.711 1.000 3.000
PctIndep 4,128 0.131 0.700 0.773 0.800 0.880
NumDirs 4,128 2.182 8.000 9.082 9.000 10.000
CEOPortDelta ($) 4,128 606,123 39,818 330,331 129,767 355,567
CEOPortVega ($) 4,128 51,095 0.000 16,532 320 106,832
Controls
CFOps 4,128 0.096 0.006 0.104 0.103 0.151
MVE ($millions) 4,128 25,979 532 8,418 1,647 5,419
ForAssets ($millions) 4,128 57,447 6 5,101 330 1,701
GeoComp 4,128 1.852 0.324 0.628 0.541 0.985
 
  




 Table 1 (cont’d)
Panel B: Descriptive Statistics

Sample % Compustat %
Industry
Mining & Construction 2.09 4.09
Food 2.55 2.06
Textiles, printing & publishing 5.25 2.86
Chemicals 3.94 2.33
Pharmaceuticals 5.18 6.69
Extractive Industries 3.71 4.70
Durable Manufacturers 26.67 16.71
Computers 18.62 12.75
Transportation 4.64 6.24
Utilities 3.94 4.69
Retail 10.90 7.16
Financial Institutions 2.70 14.77
Insurance & real estate 1.00 5.37
Services 8.42 7.62
Other 0.39 1.94

Firm Characteristic Sample Median Compustat Median

ROA 0.029 0.005***
Book to Mkt 0.509 0.567***
LT Debt/Assets 0.156 0.078***
Sales/Assets 0.826 0.590***
MVE 1,084.391 230.125***


Industries are classified in accordance with Barth et al. (1998). Sample and Compustat observations are selected for fiscal year 2009,
which provides the largest contribution of observations (1,294) to the total estimation sample. *** denotes statistically significant
median differences at the 1% levels using a Kolmogorov-Smirnov non-parametric test.

 


 Table 1
Panel C: Mean Incentives by Tax Avoidance Decile
EndFin48Bal
Decile 1 2 3 4 5 6 7 8 9 10
Incentives
CEOPortDelta ($) 135,550 210,512 245,666 271,839 374,635 357,027 389,706 381,002 440,935 496,950
CEOPortVega ($) 10,791 9,112 13,002 15,121 16,504 20,934 14,654 17,420 21,761 26,040





 Table 1 (cont’d)

Panel D: Pearson Correlations

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
1.0000 0.1264 0.0063 -0.0953 -0.0178 0.0170 0.0626 -0.0015 -0.0491 0.1031 -0.0309
EndFin48Bal (1)
<.0001 0.6866 <.0001 0.2532 0.2762 <.0001 0.9239 0.0016 <.0001 0.0474
0.1264 1.0000 0.0146 -0.0110 -0.0274 -0.0428 0.0467 -0.0243 -0.1133 0.0352 -0.0171
TAETR (2)
<.0001 0.4201 0.5432 0.1302 0.0183 0.0100 0.1796 <.0001 0.0524 0.3452
0.0063 0.0146 1.0000 0.1795 0.1859 0.1092 0.0898 0.1998 0.0117 0.2012 -0.0332
PctIndep (3)
0.6866 0.4201 <.0001 <.0001 <.0001 <.0001 <.0001 0.4507 <.0001 0.0328
-0.0953 -0.0110 0.1795 1.0000 0.2799 0.2833 -0.0437 0.5540 0.0410 0.2954 -0.0202
LogNumDirs (4)
<.0001 0.5432 <.0001 <.0001 <.0001 0.0050 <.0001 0.0085 <.0001 0.1953
-0.0178 -0.0274 0.1859 0.2799 1.0000 0.3265 -0.0884 0.4189 0.1238 0.2295 -0.0559
LogNumFinExp (5)
0.2532 0.1302 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.0003
0.0170 -0.0428 0.1092 0.2833 0.3265 1.0000 -0.1571 0.6656 0.2635 0.3648 -0.0192
LogCEOPortDelta (6)
0.2762 0.0183 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.2168
0.0626 0.0467 0.0898 -0.0437 -0.0884 -0.1571 1.0000 -0.2284 -0.1210 -0.0630 0.0384
LogCEOPortVega (7)
<.0001 0.0100 <.0001 0.0050 <.0001 <.0001 <.0001 <.0001 <.0001 0.0136
-0.0015 -0.0243 0.1998 0.5540 0.4189 0.6656 -0.2284 1.0000 0.2818 0.4908 -0.0419
LogMVE (8)
0.9239 0.1796 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 0.0070
-0.0491 -0.1133 0.0117 0.0410 0.1238 0.2635 -0.1210 0.2818 1.0000 0.0722 -0.0150
CFOps (9)
0.0016 <.0001 0.4507 0.0085 <.0001 <.0001 <.0001 <.0001 <.0001 0.3370
0.1031 0.0352 0.2012 0.2954 0.2295 0.3648 -0.0630 0.4908 0.0722 1.0000 -0.1047
LogForAssets (10)
<.0001 0.0524 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001 <.0001
-0.0309 -0.0171 -0.0332 -0.0202 -0.0559 -0.0192 0.0384 -0.0419 -0.0150 -0.1047 1.0000
GeoComp (11)
0.0474 0.3452 0.0328 0.1953 0.0003 0.2168 0.0136 0.0070 0.3370 <.0001




 EndFin48Bal is measured as the firm’s ending balance of the uncertain tax benefit account (Compustat TXTUBEND), scaled by the firm’s average
assets over the period. TAETR is computed as the mean three-year GAAP ETR (computed as the firm’s total tax expense scaled by pre-tax income)of
the firm's size and industry peers minus the firm's three-year GAAP ETR. Size peers are firms within the same quintile of total assets and industry peers
are firms within the same Fama-French 48 industry portfolios. PctIndep is the percentage of independent directors to total directors sitting on the board
in the previous year (as indicated by Equilar). LogNumDirs is the natural logarithm of one plus the number of total directors sitting on the board in the
previous year (as indicated by Equilar). LogNumFinExp is the natural logarithm of one plus the number of financial experts designated on the board of
directors in the year preceding the fiscal year (as indicated by RiskMetrics). CEOPortDelta (CEOPortVega) is the (risk-neutral) dollar change in the
firm CEO’s equity portfolio value for a 1% change in the value (volatility) of the firm’s stock price (Core and Guay, 2002). LogCEOPortDelta
(LogCEOPortVega) is the natural logarithm of CEOPortDelta (CEOPortVega). LogMVE is the natural logarithm of market value of equity computed
for the fiscal year. CFOps is cash flow from operations divided by average total assets. LogForAssets is the natural logarithm of total foreign assets
computed for the fiscal year. GeoComp is a revenue-based Hirfindahl-Hirschman index that captures within-firm geographic segment complexity
(Bushman et al., 2004) computed for the fiscal year.




 Table 2
Panel A: CEO Incentives
TaxPosition = EndFin48Bal TAETR EndFin48Bal TAETR
n= 4,128 3,137 4,128 3,137
X= LogCEOPortDelta LogCEOPortDelta LogCEOPortVega LogCEOPortVega
Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat
OLS 0.0001 0.46 -0.0035 -1.56 0.0004 4.44 0.0014 1.69
Quantile
0.10 0.0001 3.00 -0.0051 -2.30 0.0000 4.02 -0.0031 -2.66
0.20 0.0002 3.33 -0.0031 -1.68 0.0001 4.81 -0.0006 -0.87
0.30 0.0002 2.35 -0.0014 -0.88 0.0001 5.59 0.0010 1.84
0.40 0.0001 1.03 -0.0011 -1.06 0.0001 3.35 0.0014 2.35
0.50 0.0002 1.37 -0.0014 -1.12 0.0002 3.69 0.0022 3.85
0.60 0.0003 1.36 -0.0019 -1.22 0.0002 5.40 0.0033 5.04
0.70 0.0001 0.21 -0.0024 -1.30 0.0004 4.70 0.0045 5.65
0.80 0.0002 0.64 -0.0004 -0.15 0.0006 4.70 0.0049 4.85
0.90 0.0008 1.94 -0.0013 -0.44 0.0009 5.02 0.0030 3.21
Avg Pseudo R2 0.039 0.046 0.039 0.000
Q(0.80) = Q(0.20) 0.989 0.309 0.000 0.000
Q(0.90) = Q(0.10) 0.097 0.213 0.000 0.000
Q(0.90) = Q(0.50) 0.144 0.975 0.002 0.301
Q(0.10) = Q(0.50) 0.374 0.044 0.000 0.018

Panel B: Board Expertise and Independence

TaxPosition = EndFin48Bal TAETR EndFin48Bal TAETR
n= 4,128 3,137 4,128 3,137
X= LogNumFinExp LogNumFinExp PctIndep PctIndep
Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat
OLS -0.0009 -0.96 -0.0089 -1.08 -0.0014 -0.46 0.0103 0.37
Quantile
0.10 0.0005 5.58 0.0152 1.08 0.0009 2.50 -0.0274 -0.84
0.20 0.0008 6.57 0.0056 0.73 0.0005 0.88 0.0236 0.97
0.30 0.0007 3.59 0.0011 0.18 0.0015 2.13 0.0275 1.55
0.40 0.0009 2.30 -0.0044 -0.67 0.0021 2.04 0.0384 2.30
0.50 0.0006 1.33 -0.0120 -1.85 0.0025 2.14 0.0426 2.01
0.60 0.0004 0.61 -0.0144 -1.91 0.0014 0.61 0.0388 1.68
0.70 0.0004 0.55 -0.0167 -2.41 0.0023 0.95 0.0153 0.48
0.80 -0.0008 -0.69 -0.0100 -0.99 0.0043 1.35 0.0021 0.05
0.90 -0.0040 -1.94 -0.0027 -0.32 -0.0009 -0.15 -0.0707 -2.28
Avg Pseudo R2 0.039 0.046 0.039 0.046
Q(0.80) = Q(0.20) 0.173 0.254 0.215 0.425
Q(0.90) = Q(0.10) 0.014 0.394 0.174 0.604
Q(0.90) = Q(0.50) 0.011 0.314 0.568 0.005
Q(0.10) = Q(0.50) 0.772 0.125 0.094 0.016




 Table 2 (cont’d)
Panel C: Control Variables
TaxPosition = EndFin48Bal EndFin48Bal EndFin48Bal EndFin48Bal EndFin48Bal
n= 4,128 4,128 4,128 4,128 4,128
X= LogNumDirs LogMVE CFOps LogForAssets GeoComp
Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat
OLS -0.0159 -8.26 0.0008 2.07 -0.0155 -3.68 0.0010 7.29 -0.0003 -1.47
Quantile
0.10 0.0002 1.04 0.0000 0.37 0.0006 1.76 0.0002 6.50 -0.0001 -0.48
0.20 -0.0004 -1.21 0.0002 3.00 0.0007 1.57 0.0003 7.31 -0.0002 -0.49
0.30 -0.0010 -2.57 0.0004 3.72 0.0008 1.05 0.0005 8.65 -0.0000 -0.07
0.40 -0.0020 -3.05 0.0005 2.63 0.0006 0.34 0.0006 7.87 -0.0001 -0.06
0.50 -0.0050 -4.36 0.0008 4.31 -0.0012 -0.57 0.0007 5.32 -0.0001 -0.08
0.60 -0.0094 -7.68 0.0013 4.62 -0.0025 -0.95 0.0008 4.67 -0.0002 -0.09
0.70 -0.0136 -7.92 0.0016 4.02 -0.0041 -0.75 0.0009 4.06 -0.0002 -0.09
0.80 -0.0177 -6.22 0.0018 3.35 -0.0118 -1.67 0.0013 3.65 -0.0002 -0.04
0.90 -0.0349 -8.28 0.0024 2.66 -0.0291 -2.54 0.0015 2.93 -0.0004 -0.07

TaxPosition = TAETR TAETR TAETR TAETR TAETR

n= 3,137 3,137 3,137 3,137 3,137
X= LogNumDirs LogMVE CFOps LogForAssets GeoComp
Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat Coef. t-stat
OLS -0.0175 -0.98 0.0036 1.05 -0.2108 -5.77 0.0033 2.71 -0.0015 -0.90
Quantile
0.10 -0.0367 -1.43 0.0324 7.86 -0.0298 -0.51 -0.0051 -3.63 -0.0027 -0.34
0.20 0.0023 0.18 0.0167 6.39 -0.1879 -4.48 0.0001 0.11 -0.0048 -0.56
0.30 0.0103 0.81 0.0100 4.86 -0.2257 -7.74 0.0024 2.49 -0.0059 -0.59
0.40 0.0028 0.22 0.0077 3.61 -0.2800 -12.12 0.0045 3.97 -0.0066 -0.45
0.50 0.0043 0.41 0.0003 0.14 -0.2975 -9.11 0.0064 6.53 -0.0073 -0.65
0.60 -0.0163 -1.06 -0.0072 -2.95 -0.3026 -9.23 0.0090 9.17 -0.0043 -0.51
0.70 -0.0182 -0.89 -0.0137 -5.27 -0.2797 -8.58 0.0108 11.14 0.0024 0.38
0.80 -0.0053 -0.21 -0.0228 -6.14 -0.2592 -4.18 0.0115 8.23 0.0015 0.25
0.90 0.0310 1.37 -0.0220 -5.54 -0.1709 -2.95 0.0039 2.23 0.0006 0.08

EndFin48Bal is measured as the firm’s ending balance of the uncertain tax benefit account (Compustat
TXTUBEND), scaled by the assets. TAETR is computed as the mean three-year GAAP ETR (computed as the firm’s
total tax expense scaled by pre-tax income)of the firm's size and industry peers minus the firm's three-year GAAP
ETR. Size peers are firms within the same quintile of total assets and industry peers are firms within the same Fama-
French 48 industry portfolios. PctIndep is the percentage of independent directors to total directors sitting on the
board in the previous year (as indicated by Equilar). LogNumFinExp is the natural logarithm of one plus the number
of financial experts designated on the board of directors in the year preceding the fiscal year (as indicated by
RiskMetrics). CEOPortDelta (CEOPortVega) is the (risk-neutral) dollar change in the firm CEO’s equity portfolio
value for a 1% change in the value (volatility) of the firm’s stock price (Core and Guay, 2002). LogCEOPortDelta
(LogCEOPortVega) is the natural logarithm of one plus CEOPortDelta (CEOPortVega). Two-sided p-values are
reported for the tests of coefficient differences between quintiles 0.80 vs. 0.20 and 0.90 vs. 0.10.




 Table 3
Desai and Dharmapala (2006) Quantile Regression
Table 4 (All Firms)
StkMixGrant WellGov StkMixGrant * WellGov
Coef. t-stat Coef. t-stat Coef. t-stat
OLS -0.0086 -1.60 0.0323 0.57 0.0016 0.18
Quantile
0.05 0.0025 1.50 0.0081 0.01 0.0046 1.11
0.10 0.0025 0.86 0.0081 0.01 0.0046 0.76
0.15 0.0029 0.82 0.0102 0.01 0.0012 0.15
0.20 0.0022 0.56 0.0081 0.01 0.0009 0.11
0.25 0.0020 0.50 0.0090 0.03 -0.0007 -0.09
0.30 0.0012 0.34 0.0175 0.13 -0.0023 -0.27
0.35 0.0006 0.16 0.0186 0.29 -0.0035 -0.50
0.40 0.0000 0.01 0.0064 0.12 -0.0039 -0.47
0.45 -0.0002 -0.07 0.0044 0.09 -0.0038 -0.51
0.50 -0.0004 -0.11 0.0016 0.04 -0.0032 -0.45
0.55 0.0007 0.24 -0.0004 -0.01 -0.0041 -0.60
0.60 0.0011 0.36 0.0079 0.18 -0.0028 -0.41
0.65 0.0014 0.35 0.0086 0.15 -0.0040 -0.52
0.70 0.0015 0.39 0.0048 0.04 -0.0044 -0.63
0.75 0.0023 0.53 0.0065 0.02 -0.0065 -0.84
0.80 0.0022 0.53 0.0059 0.01 -0.0077 -0.80
0.85 0.0014 0.46 0.0211 0.02 -0.0109 -1.65
0.90 -0.0006 -0.24 0.0253 0.02 -0.0080 -1.52
0.95 -0.0006 -0.47 0.0262 0.02 -0.0079 -2.65
Replication of Desai and Dharmapala (2006) Table 4, Column (4) using quantile regression estimation. The
following equation is estimated (equation 17 from Desai and Dharmapala, 2006, adjusted to include a mean effect
for WellGov which was mistakenly excluded from their specification):
TSi,t = β0 + β1 STKMIXGRANTi,t + β2 (WELLGOVi * STKMIXGRANTi,t)
+ Other Interaction Terms + Firm Fixed Effects + Year Fixed Effects + Controls + υi,t

TS is the residual estimated from regressing the difference between book and tax income on total accruals (see Desai
and Dharmapala, 2006, 159-160). StkMixGrant is the ratio of stock option grant value to total compensation for the
firm’s top five executives. WellGov is a dichotomous variable that equals one for firms that have a low governance
index score (G-score less than or equal to 7) and equals zero otherwise. Estimation includes controlling for the level
of deferred taxes and year, firm, and firm-year size fixed effects.