Upload
others
View
1
Download
0
Embed Size (px)
Citation preview
Tax Risk and the Cost of Equity Capital
Michelle Hutchens Indiana University
Sonja Rego* Deloitte Foundation Accounting Faculty Fellow
Indiana University
September 2013
Abstract A primary benefit of corporate tax avoidance is greater after-tax cash flows and therefore, increased shareholder value. However, in the accounting literature, some measures of aggressive tax avoidance have also been utilized as proxies for the level of a firm’s tax risk, since aggressive tax strategies involve uncertain future outcomes and can impose significant costs on the firm. This study evaluates the extent to which proxies for aggressive tax avoidance capture a firm’s tax risk, as measured by a positive association with the implied cost of equity capital. We find that the level of a firm’s reserve for income taxes is significantly, positively associated with the cost of equity capital, consistent with tax reserves capturing uncertainty surrounding a firm’s future after-tax cash flows. We also find that several other proxies for tax risk are not significantly associated with the cost of equity capital, including cash effective tax rates. We conclude that these tax metrics do not capture uncertainty surrounding a firm’s future after-tax cash flows. Keywords: Tax risk, cost of capital, tax reserves, uncertain tax positions, tax avoidance. * Corresponding Author: Indiana University; Kelley School of Business; 1309 E. 10th St., Bloomington, IN 47405. Office: (812) 855-6356; Email: [email protected]. We are grateful for helpful comments from T.J. Atwood, Daniel Beneish, Joe Fisher, Leslie Hodder, Pat Hopkins, Nathan Marshall, Brian Miller, John Phillips, Mort Pincus, Terry Shevlin, Logan Steele, Siew Hong Teoh, Jim Wahlen, Dave Weber, Michael Willenborg and workshop participants at Florida State University, Indiana University, University of California – Irvine, and the University of Connecticut. Professor Rego appreciates research funding from the Deloitte Foundation and the Kelley School of Business.
1
1. Introduction
Since the Sarbanes-Oxley Act of 2002, many tax professionals have shifted their focus
from traditional tax compliance and planning to “tax risk management.” As a result, public
accounting firms regularly publish “tax risk management” surveys and market “tax risk
management” services. Consistent with this shift, a 2004 Ernst & Young survey indicates that
68 percent of tax directors view tax risk management a “critical factor” in corporate governance,
while 91 percent claim they receive “active direction” from their CEO and/or CFO on tax risk
matters (Ernst & Young 2004). More recently, increased scrutiny by global tax authorities,
changes in tax reserve disclosure requirements for both financial and tax reporting purposes,1
combined with substantial economic uncertainty since the start of the financial crisis have only
further heightened tax practitioners’ focus on tax risk management (Ernst & Young 2011).
Tax practitioners often define “tax risk” as involving transactional risk, operational risk,
compliance risk, and financial reporting risk.2 For purposes of this study, we define tax risk as
all tax-related risks and uncertainties associated with a firm’s operating, investing, and financing
decisions, including uncertainty in the application of tax law to company facts, the risk of audit,
including assessments of additional tax, interest, and penalties, and uncertainty in the financial
accounting for income taxes. Taken together, these tax-related risks and uncertainties can
impose substantial costs on a firm, both in current and future time periods. However, it is
1 FASB Interpretation No. 48 “Accounting for Uncertainty in Income Taxes,” codified in ASC 740 and commonly referred to as “FIN 48,” now requires firms to disclose details regarding uncertain tax positions in their financial statements. In addition, Schedule UTP requires businesses that disclose uncertain tax positions in their financial statements to provide additional information in their federal income tax returns regarding uncertain tax positions. 2 A PriceWaterhouseCoopers guide to tax risk management lays out four primary components of tax risk including transactional risk (uncertainty in a specific transaction in either law or fact, or risk through extreme complexity), operational risk (uncertainty in applying tax laws to regular operations), compliance risk (reliance on professionals and accounting systems in gathering data for tax return preparation), and financial accounting risk (uncertainty in estimates made in the tax accrual and tax related financial statement disclosures) (PriceWaterhouseCoopers 2004). Grant Thornton also defines tax risk as including the same four components, plus personal tax risk (http://www.grantthornton.com.au/Services/Tax/TaxRiskMgmt.asp).
2
unclear the extent to which investors measure and evaluate a firm’s “tax risk.” To that end, in
this study we investigate the extent to which tax risk is associated with a firm’s implied cost of
equity capital.
We examine the association between tax risk and the implied cost of equity capital for
several reasons. First, Rego and Wilson (2012) provide evidence that empirical proxies for tax
risk are positively associated with equity risk incentives, stock return volatility, and the standard
deviation of pretax income. Their results suggest a link between tax risk and firm risk. Second,
we primarily focus on the cost of equity capital rather than other measures of firm risk because
all else equal, the benefits of tax risk (i.e., lower tax liabilities) accrue to shareholders.
Moreover, greater tax risk (as defined in this paper) increases uncertainty regarding a firm’s
future after-tax cash flows, and thus should impact a firm’s implied cost of equity capital.
Given the difficulty in measuring a firm’s tax risk and cost of equity capital, we utilize
several proxies for each underlying construct. Because our study utilizes a broad definition of
tax risk (see above), we require a tax risk metric that captures the tax consequences for a broad
set of transactions that involve greater levels of uncertainty with regard to future after-tax cash
flows. Recent accounting studies utilize discretionary permanent book-tax differences and tax
shelter prediction scores as proxies for “aggressive” tax avoidance, which involves a high degree
of uncertainty with respect to future tax payments (e.g., Frank, Lynch, and Rego 2009; Wilson
2009; Lisowsky 2010; Rego and Wilson 2012). In addition, Lisowsky et al. (2013) provide
evidence that the contingent liability for income taxes (aka tax reserve) is a superior predictor of
tax shelter activity relative to other measures of aggressive tax avoidance.3 We utilize all three
3 The contingent liability for income taxes (aka tax cushion, tax reserves, and/or unrecognized tax benefits as disclosed under FIN 48) represents the amount of income taxes that the firm may be required to pay to tax authorities related to uncertain tax positions. For example, if a firm deducts an expense that is more likely than not
3
measures (i.e., tax reserves, discretionary book-tax differences, and a tax shelter prediction
score) as proxies for a firm’s exposure to tax risk. Consistent with Rego and Wilson (2012), we
assert that aggressive tax positions increase the uncertainty surrounding future after-tax cash
flows.4 As a result, increased tax aggressiveness leads to increased tax risk and higher cost of
equity capital.
Contrary to some recent studies, we do not consider the cash effective tax rate (ETR) a
proxy for aggressive tax positions. The cash ETR compares cash taxes paid to adjusted pretax
income, often over an extended period of time. Because tax strategies that reduce cash tax
payments (but not adjusted pretax income) reduce a firm’s cash ETR, this tax avoidance measure
captures tax planning that involves both certain and uncertain outcomes. Thus, we view the cash
ETR as a weak proxy for aggressive tax positions but also the most direct measure of the cash
tax savings from tax avoidance strategies, which should increase shareholder value. Koester
(2011) hypothesizes that investors positively value uncertain tax avoidance because they expect
firms to retain most of their unrecognized tax benefits (i.e., the tax savings from tax avoidance
strategies) and because uncertain tax avoidance signals that managers are good stewards of
shareholder wealth. Indeed, Koester finds a positive association between tax reserves and stock
price. However, this positive association does not address the extent to which tax risk (as
defined in this study) is reflected in a firm’s cost of equity capital. Because our proxies for tax
risk potentially reflect the tax savings from tax avoidance strategies and uncertainty surrounding
to be rejected by the Internal Revenue Service, then the firm must increase the tax reserves reported in its financial statements by the tax benefit associated with the expense deduction. See section 2 for more detailed discussion. 4 Our views of tax aggressiveness and tax risk are similar to those in Guenther, Matsunaga, and Williams (2013). Those authors define tax aggressiveness as the extent to which a firm takes tax positions that are unlikely to survive a challenge by the IRS (as measured by a firm’s tax reserves) and tax risk as uncertainty regarding the firm’s future tax payments (as measured by the firm’s volatility of cash effective tax rates).
4
future after-tax cash flows, we control for the cash tax savings generated by tax avoidance
strategies (as measured by cash ETRs) in our multivariate tests.
Similar to tax risk, the cost of equity capital is not directly observable. However,
accounting research has developed a variety of empirical proxies for the rate of return that
investors require from their equity investments. Consistent with several recent studies (Dhaliwal
et al. 2006; Hail and Leuz 2006; Daske et al. 2008; Callahan et al. 2012), we calculate a firm’s
implied cost of equity capital as the average of four measures developed in prior research.5 A
firm’s cost of equity capital is comprised of the risk-free rate of return and a risk premium, which
has been empirically linked to firm-specific risk factors including firm size, book-to-market
ratio, beta, leverage, and accrual and internal control quality (e.g., Francis et al. 2004; Dhaliwal
et al. 2006; Hail and Leuz 2006; Ashbaigh-Skaife et al. 2009; Callahan et a. 2012). Additionally,
Lambert et al. (2007) provide evidence that higher quality accounting information and higher
quality firm disclosures reduce a firm’s cost of equity capital. They demonstrate that the quality
of accounting information influences investors’ assessments of uncertainty surrounding a firm’s
future cash flows, which has a direct effect on the assessed covariances with other firms’ cash
flows and thus impacts the firm’s cost of equity capital. If our proxies for tax risk influence
investors’ assessments of the distribution of future after-tax cash flows, then they should be
associated with the cost of equity capital.
We provide consistent evidence that the total amount of tax reserves disclosed under FIN
48 is positively associated with the implied cost of equity capital. In contrast, the one-year
change in tax reserves, discretionary permanent book-tax differences, and a tax shelter prediction
score are not significantly associated with the cost of equity capital. These results suggest that
5 Specifically, we use the implied cost of capital measures developed in Claus and Thomas (2001), Gebhardt, Lee, and Swaminathan (2001), Ohlson and Juettner-Nauroth (2005), and Easton (2004).
5
larger reserves for income taxes increase investors’ assessments of uncertainty surrounding a
firm’s future after-tax cash flows, while the other proxies for tax risk do not. We also find that
the cash tax savings from income tax avoidance (as measured by cash ETRs) are not
significantly associated with the implied cost of equity capital.6 We note that our primary
analyses control for numerous firm characteristics that influence expected future cash flows,
including characteristics related to a firm’s operating, investing, and financing activities. Given
findings in Guenther, Matsunaga, and Williams (2013), we re-estimate our cost of capital
regressions to include an alternative proxy for tax risk – the volatility of cash ETRs – in addition
to our other proxies for tax risk. We continue to find that tax reserves are positively associated
with the implied cost of capital, while other tax metrics, including cash ETR volatility, are not.
Taken together, we infer that investors perceive larger tax reserves as requiring greater equity
risk premiums, while other measures of tax risk and tax avoidance are not associated with the
implied cost of equity capital.
In supplemental analyses, we also examine whether tax reserves are associated with other
measures of firm risk. Based on evidence in Rego and Wilson (2012) that tax risk is positively
associated with stock return volatility, we utilize current year stock return volatility as one
alternative measure of firm risk. We also utilize operating cash flow volatility and market beta
as additional measures of firm risk. While the stock return and cash flow volatility measures
reflect dispersion in expected and actual payoffs, respectively, market beta captures systematic
(i.e., non-diversifiable) firm risk. In analyses that regress these alternative measures of firm risk
on tax risk and control variables we find that tax reserves are positively associated with both
6 This result is contrary to findings in Goh, Lee, Lim, and Shevlin (2013), which finds that greater tax avoidance – as proxied by lower cash ETRs – is associated with significantly lower implied cost of equity capital. Given the different sample periods between the two studies, and the absence of a proxy for tax risk in the Goh et al. (2013) regression model, it is difficult to reconcile our contrasting findings.
6
stock return and cash flow volatility, but negatively related to market beta. Similar to our
findings for the cost of equity capital, the results for tests based on stock return volatility as our
measure of firm risk suggest that investors perceive larger tax reserves as increasing the
uncertainty surrounding expected future after-tax cash flows.
Our study provides initial insights into whether and how investors evaluate tax risk. Prior
research develops a variety of tax measures and acknowledges that each of these proxies captures
different aspects of aggressive tax planning and tax risk (Hanlon and Heitzman 2010). FIN 48
specifically claims to improve financial reporting by providing “more information about the
uncertainty in income tax assets and liabilities.” Consistent with that claim and with findings in
Lisowsky et al. (2013), our study highlights the uniqueness of tax reserves as a measure of tax
risk. We find that discretionary book-tax differences, a tax shelter prediction score, and cash
effective tax rates are not associated with the implied cost of equity capital and therefore, likely
do not capture greater investor uncertainty regarding future after-tax cash flows. In contrast, tax
reserves are associated with the implied cost of capital and therefore, capture some aspects of a
firm’s overall tax risk. Our findings also contribute to the growing literature on uncertain tax
positions as disclosed under FIN 48. Our results combined with those in Koester (2011) suggest
investors understand that a firm’s tax reserves reflect both cash tax savings from tax avoidance
strategies and uncertainty surrounding future after-tax cash flows (i.e., tax risk). Overall, our
results are consistent with transparency in the reporting of uncertain tax positions under FIN 48
providing investors with value-relevant information regarding a firm’s exposure to tax risk.
7
2. Background and Hypothesis Development
2.1 Aggressive Tax Avoidance and Tax Risk
Recent research has evaluated the costs and benefits of “aggressive” tax avoidance,
which is often defined as tax positions that involve greater uncertainty with respect to outcomes
with tax authorities. In his evaluation of tax shelters, Weisbach (2001) asks why firms do not
avoid more income taxes through tax shelter transactions, especially given the availability of tax
shelters, the relatively low costs of implementation, and the low probability of challenge. A
primary benefit of aggressive tax planning is greater after-tax cash flows and therefore, increased
shareholder value. Frischmann et al. (2008) provide evidence that a firm’s initial disclosure of
tax reserves related to “permanent,” uncertain tax positions is positively associated with
abnormal stock returns in the 3-day window around the initial disclosure. Their result suggests
that investors view uncertain tax avoidance as value enhancing. It is also consistent with
uncertain tax positions possessing reputational benefits, as shareholders infer that management is
a good steward of company resources (Koester 2011).
However, tax strategies that involve greater uncertainty with respect to future outcomes
are inherently risky and often involve significant costs. Firms with more aggressive tax
strategies incur both internal and external costs to reduce their overall tax burden. Uncertain tax
avoidance that involves unique transactions can be costly to implement, given complexities in
the application of tax law and in understanding company facts (e.g., costs associated with
internal tax staff, external tax service providers, and/or coordination with other functional units
within the firm). Uncertain tax strategies also increase financial reporting risk, as the firm must
decide whether each and every tax position requires a tax reserve under FIN 48 and if so, how
8
large the tax reserve must be.7 Tax risk can result from a firm’s tax positions coming under audit
by the IRS or other tax authorities, in which case the firm can experience significant costs in
complying with the audit and paying unpaid taxes, penalties, and/or interest. In addition,
Balakrishnan, Blouin, and Guay (2012) provide evidence that tax aggressiveness reduces
corporate transparency, as measured by larger analyst forecast errors and dispersion and greater
information asymmetry. Their results are consistent with assertions in Desai and Dharmapala
(2006, 2009) that aggressive tax avoidance obscures financial reporting and thus increases
agency costs. In sum, increasing tax risk can impose significant costs on a firm.
Since aggressive tax avoidance increases both tax risk and after-tax cash flows, it is
unclear to what extent empirical proxies for uncertain tax avoidance capture tax risk, as defined
in this paper. Tax strategies can have highly certain outcomes (e.g., tax-exempt interest income
earned on municipal bonds) or highly uncertain outcomes (e.g., transfer pricing schemes
designed to shift profits from high tax to low tax locations) and so the magnitude of tax risk can
vary substantially across firms with seemingly similar rates of tax avoidance. For example, two
firms can have identical cash ETRs but different levels of tax risk because one firm engaged in a
highly uncertain tax shelter transaction while the other firm took advantage of bonus
depreciation and tax-exempt interest income to reduce its cash ETR. Yet, it is unclear whether
investors recognize and evaluate the differing levels of tax risk across these two firms.
2.2 Background on Tax Reserves (aka Unrecognized Tax Benefits)
FASB Interpretation No. 48 Accounting for Uncertainty in Income Taxes, codified in
ASC 740 (commonly referred to as “FIN 48”), requires firms to evaluate and disclose contingent
7 Since 2010, firms have also been required to provide detailed information in their U.S. federal income tax returns (on Schedule UTP) regarding the uncertain tax positions for which the taxpayer has recorded a reserve in its financial statements. Thus, tax reserve disclosures in a firm’s financial statements now subject the firm to even greater tax risk, since the IRS intends to use Schedule UTP to refine its audit process and procedures.
9
income tax liabilities. A firm’s contingent liability for income taxes, which we refer to as tax
reserves, informs financial statement users of tax positions that have a relatively high level of
uncertainty based on tax laws and therefore, tax positions that are inherently more risky. In
2006, FIN 48 required all publicly traded firms to record tax reserves on the balance sheet and
also to disclose in the footnotes of the financial statements tabular details regarding tax reserves.
A firm is required to record a tax reserve for the full benefit of any tax position that, based on tax
laws and regulations, has a fifty percent chance or less of being successfully upheld. When
evaluating the recognition of each tax position, FIN 48 requires a firm to assume that each
position will be examined by the relevant tax authority, which has full knowledge of all relevant
information.8 In addition, for any tax position deemed to have a greater than fifty percent chance
of success based on the technical merits of the position, the firm must record a tax reserve for the
difference between the total benefit of the tax position and the amount that has a fifty percent
likelihood of being sustained. In sum, FIN 48 requires firms to accrue and disclose tax reserves
for tax positions that involve highly uncertain outcomes, and thus involve greater tax risk.
Rego and Wilson (2012) utilize estimated tax reserves as a proxy for aggressive tax
positions and find that equity risk incentives are associated with firms having larger estimated
tax reserves, consistent with greater corporate risk-taking. In addition, Lisowsky et al. (2013)
provide evidence that tax reserves are superior predictors of tax shelter activity relative to other
measures of corporate tax avoidance. Given the stated purpose and rules embedded in FIN 48, it
would seem that tax reserves are an appropriate proxy for tax risk. However, in a study
examining whether investors view tax reserves as value-increasing or decreasing, Koester (2011)
finds that tax reserves are positively associated with stock price. Nonetheless, this positive
association between tax reserves and stock price does not address the question of whether greater 8 ASC 740-10-25-6
10
tax risk increases the rate of return that investors require from their equity investments in a firm
(i.e., the firm’s cost of equity capital).
Given her findings that investors positively value tax reserves, Koster (2011) suggests
that uncertain tax avoidance in the current year is an indicator of future uncertain tax avoidance
that will generate future tax savings. Based on this evidence, Koester posits that the expected
future tax savings associated with uncertain tax avoidance are larger than the expected costs
generated by such avoidance. However, when examining the association between a given
parameter and a firm’s stock price it is difficult to separate the impact on a firm’s cost of equity
capital and the impact on forecasted cash flows (Botosan and Plumlee 2005). By utilizing the
cost of equity capital as our dependent variable, while controlling for current year cash ETRs, we
attempt to carefully evaluate the extent to which tax risk is reflected in tax reserves. However,
while tax reserves are designed to quantify uncertain tax positions, they are influenced by
managerial discretion and judgment and are subject to manipulation by opportunistic managers.
In fact, Cazier et al. (2012) provide evidence that tax reserves are frequently used to achieve
earnings targets, even in the post-FIN 48 time period. Thus, it is an empirical question whether
tax reserves accurately capture a firm’s tax risk.
2.3 Tax Risk and the Cost of Equity Capital
Lambert et al. (2007) provide evidence that despite the ability of investors to diversify
risk, higher quality accounting information and higher quality firm disclosures reduce a firm’s
cost of equity capital. They demonstrate that the quality of accounting information influences
investors’ assessments of uncertainty surrounding a firm’s future cash flows, which affects the
assessed covariances with other firms’ cash flows and thus impacts the firm’s cost of capital.
Lambert et al. (2007) also show that accounting system quality has an indirect effect on a firm’s
11
cost of capital, since accounting system quality affects firms’ real decisions and real decisions
influence expected net cash flows to investors.
Expanding on Lambert et al.’s (2007) analysis, Ashbaugh-Skaife et al. (2009) provide
evidence that firms with weak internal controls have higher implied costs of equity capital. They
theorize that when a firm reports internal control deficiencies, the quality of the firm’s
accounting signals is impaired, limiting an investor’s ability to assess the firm’s cash flows
relative to those of the market. To better understand a firm’s implied cost of equity capital,
accounting researchers have also evaluated the association between the implied cost of capital
and disclosure level (Botosan 1997; Botosan and Plumlee 2002), accruals quality (Francis et al.
2004), and financial reporting under FIN 46 (Callahan et al. 2012). In addition, Dhaliwal et al.
(2006) provide evidence that the positive association between the cost of equity capital and
leverage is decreasing in a firm’s tax benefit from debt. In their research setting, income taxation
generates cost savings (i.e., tax deductions for interest expense) for highly levered firms, which
moderates the association between the implied cost of capital and leverage. However, to our
knowledge, prior research has not evaluated whether a firm’s tax risk is reflected in its cost of
equity capital.
In this study, we assert that a firm’s exposure to tax risk should be reflected in its implied
cost of equity capital, where greater tax risk involves greater uncertainty surrounding future
after-tax cash flows. Income taxes consume a large proportion of a firm’s pretax profits, and
thus constitute a material, recurring expense that significantly impacts a firms’ after-tax cash
flows. But income taxes are not only material, they are also highly complex. Today’s global
businesses must contend with tax laws and tax authorities that not only cross state borders, but
for U.S. multinational companies, they also cross international borders. Numerous strategies for
12
reducing global income tax liabilities exist, but given increased scrutiny by global tax authorities,
substantial economic uncertainty since the onset of the financial crisis in 2008, and changes in
tax reserve disclosure requirements for both financial and tax reporting purposes, these tax
reduction strategies can expose firms to substantial tax risk.9 Building on Lambert et al. (2007),
we argue that if tax risk influences investors’ assessments of the distribution of a firm’s future
after-tax cash flows relative to those for the market, then tax risk should be associated with the
cost of equity capital. Our first hypothesis, stated in the alternative:
H1: Tax risk is positively associated with the cost of equity capital. Measuring tax risk (i.e., tax strategies that involve highly uncertain outcomes) has proven
a difficult task for accounting researchers. Rego and Wilson (2012) utilize four different proxies
for corporate tax avoidance, which they assert should reflect tax risk to varying degrees, although
they acknowledge that all four proxies contain measurement error. For example, discretionary
permanent book-tax differences (DTAX) do not capture uncertain tax avoidance that generates
temporary (rather than permanent) book-tax differences. The tax shelter prediction score
(SHELTER) provides insight into firm characteristics that are associated with aggressive tax
sheltering; however, this proxy also captures many aspects of a firm’s business model and does
not directly measure uncertain tax avoidance. Rego and Wilson (2012) also note that firms with
low cash ETRs are as likely to employ low-risk tax reduction strategies as high-risk tax reduction
strategies. In our research setting, tax reserves are the most direct measure of uncertain tax
avoidance that is publicly available. Recall that FIN 48 requires firms to provide tax reserves for
tax positions that are less than highly certain under the current tax law. As a result, we expect
tax reserves to contain less measurement error than other proxies for a firm’s tax risk, which
9 The financial crisis increased tax risk at many firms because their pre-tax operating profits were subject to so much uncertainty they found it difficult to anticipate tax planning needs and strategies in a timely manner.
13
should translate into tax reserves having greater explanatory power for the cost of equity capital,
relative to other proxies for tax risk. Thus, our second hypothesis (stated in the alternative) is:
H2: Tax reserves are more highly associated with the cost of equity capital than other proxies for a firm’s tax risk.
3. Sample Selection Procedures & Research Design
3.1 Sample Selection
The FASB required public companies to adopt the provisions of FIN 48 for their financial
reporting year beginning after December 15, 2006.10 The sample for this study therefore
includes all public firms with fiscal years ending between December 15, 2007 and December 31,
2011, for which we are able to gather all necessary data. To perform the empirical tests outlined
below, we require: (1) annual financial statement data from the Compustat North America
Fundamentals Annual database, (2) monthly stock return data from the CRSP Monthly Stock
File, (3) daily stock price data from the CRSP Daily Stock File, and (4) analysts’ forecasts of
earnings per share, dividends per share, book value per share, and long-term growth from the
I/B/E/S Summary Statistics file.
For a firm-year observation to be included in our final sample, the firm must have
reported non-zero tax reserves in their financial statements and have the data necessary to
compute other proxies for corporate tax avoidance.11 To compute DTAX, we require each
industry-year combination to have at least 15 observations (Frank et al. 2009). To compute
SHELTER, we require each industry-year combination to have at least 5 observations for the
10 FASB Interpretation No. 48. 11 We do not include in our sample firms that report zero tax reserves because we are concerned that some firms either chose or were not able to comply with FIN 48 (at least in the first several years after its implementation), in which case a zero tax reserve could indicate either highly certain tax positions or non-compliance. For future drafts of this study we intend to perform robustness tests that include in our sample firms with zero tax reserves to determine if our results are sensitive to their inclusion.
14
computation of discretionary accruals. To compute cash ETRs, we require a firm to have
positive cumulative pretax income (adjusted for special items) for the five year period ending in
the observation year and positive cumulative cash taxes paid over the same five year period.
This restriction also focuses our analyses on firms that are more likely to engage in risky tax
avoidance, as firms with cumulative losses likely have less incentive to tax plan (Rego and
Wilson 2012). Each firm-year observation must also have the necessary data to compute the
implied cost of equity capital and each control variable. We exclude real estate investment
trusts, financial institutions, and utilities, as regulation of these industries likely affects both a
firm’s cost of equity capital and tax risk tolerance.
As reported in Table 1, we obtain 11,147 firm-year observations (2,989 unique firms) for
which Compustat reports non-zero tax reserve data during our sample period. After applying all
of the data restrictions described above, including the elimination of observations with
cumulative pre-tax losses , our final sample includes 3,263 firm-year observations for 1,075
unique firms. We winsorize all continuous variables at the 1st and 99th percentiles.
[Insert Table 1 here]
3.2 Research Design
To examine the association between tax risk and the cost of equity capital (H1) we
estimate the following regression model:
AVG_RATEt = α1TAX_RISKt + α2CASH_ETRt + α3CAP_EXPt + α4R&Dt + α5SG&At + α6FOR_OPERt + α7LEVt + α8ROAt + α9DISCR_ACCRt + α10FC_BIASt + α11EARN_VOLt + α12MKTt + α13SMBt + α14HMLt (1),
where tax risk represents our proxies for tax risk, including tax reserves, discretionary permanent
book-tax differences (DTAX), and the tax shelter prediction score (SHELTER). We measure tax
reserves in several ways. Our primary calculation is simply the total tax reserve reported at
15
fiscal year-end (TAX_RES). This amount captures all tax reserves on a firm’s balance sheet. We
also calculate the change in total tax reserves from year t-1 to year t (TAX_RES), since
investors may differentially evaluate a firm’s tax risk based on whether the firm increases or
decreases its tax reserves during the current fiscal year. The dependent variable (AVG_RATE) is
the average of four commonly-used implied cost of capital measures (discussed in greater detail
below), less the median yield on a 10-year treasury bond.
We include numerous variables to control for factors that are likely associated with a
firm’s cost of equity capital. First, we control for firm characteristics that are known to be
associated with corporate tax avoidance, but may also be associated with a firm’s implied cost of
capital, including: capital and research and development expenditures (CAP_EXP and R&D,
respectively), selling, general, and administrative costs (SG&A), the presence of foreign
operations (FOR_OPER), leverage (LEV), profitability (ROA), and discretionary accruals
(DISCR_ACCR). Second, we control for the cash tax savings from corporate tax avoidance
(CASH_ETR), which should allow our proxies for tax risk to capture tax positions with greater
uncertainty with respect to future outcomes. Third, given evidence in Hail and Leuz (2006),
Daske et al. (2008), and McInnis (2010), we control for the potential influence of both analyst
forecast errors (FC_BIAS) and earnings volatility (EARN_VOL) on the implied cost of capital.
We also include the three Fama and French (1993) risk factors (MKT, SMB, and HML) in our
implied cost of capital model, consistent with Dhaliwal et al. (2006).12 Lastly, we include firm
and industry fixed effects. See Appendix A for complete descriptions of all variables.
We next empirically evaluate which of our tax risk proxies has greater explanatory power
for our model of the implied cost of capital (H2). Given that a firm’s reported tax reserves 12 Other studies include some or all of the firm characteristics that underlie the Fama and French (1993) risk factors (e.g., beta, size, and book-to-market ratio) directly in their cost of capital models (instead of the three Fama and French (1993) factors), including Hail and Leuz (2006), Ashbaugh-Skaife et al. (2009), and Callahan et al. (2012).
16
capture tax positions with a relatively high level of uncertainty, we expect total tax reserves to
have greater explanatory power for our cost of capital model. However, tax reserves are subject
to managerial discretion and therefore, it is possible that one of the other tax risk proxies has
greater explanatory power for our cost of capital model. To empirically evaluate the explanatory
power of each tax risk proxy, we utilize the Vuong (1989) likelihood ratio test for non-nested
models13 and the Clarke (2003) paired sign test for non-nested models.14 Both tests compare the
relative explanatory power of two separate estimations of equation (1), where each estimation is
based on a different proxy for tax risk. The null hypothesis of the Vuong and Clarke tests is that
both regression estimations are equally able to predict a firm’s cost of equity capital. The
alternative hypothesis is that one model has greater explanatory power than the other.
3.3 Computing of the Implied Cost of Equity Capital
Prior research utilizes a variety of methods for computing a firm’s implied cost of equity
capital (e.g., Botosan and Plumlee 2005 review five common methods). Consistent with more
recent research (Dhaliwal et al. 2006; Daske et al. 2008, Hail and Leuz 2009, Callahan et al.
2012), our primary proxy (AVG_RATE) is the average of four measures of the implied cost of
capital, less the median rate on a 10-year Treasury bond for the year immediately preceding the
date of the cost of capital computation.15 The four cost of capital measures are derived from
accounting valuation models and estimate the implied cost of equity capital based on stock price
and analysts’ forecasts of dividends and earnings. First, we calculate the implied cost of capital
13 The Vuong (1989) test compares the average log-likelihood ratio of each model to zero. One of the benefits and drawbacks of the Vuong test is that the test does not require that one of the models be the true model and simply tests which model is closer to the true model. The drawback is that because the test is relative it does not tell us if both models are a poor fit and far from the true model (Clarke 2001). 14 The Clarke (2003) paired sign test is similar to the Vuong (1989) test but it evaluates the median log-likelihood ratio of each model. 15 Hail and Leuz (2009) perform sensitivity analyses aggregating each of the four measures in different ways and utilizing different weights and find consistent results. In light of their findings, both Daske et al. (2008) and Callahan et al. (2012) utilize a simple average of the four cost of capital measures.
17
based on the models developed by Claus and Thomas (2001) and Gebhardt, Lee, and
Swaminathan (2001). Each of these models is based on the residual income valuation model;
however, the Claus and Thomas (2001) model assumes that long-term residual income grows at
a rate equal to inflation and the Gebhardt, Lee, and Swaminathan (2001) model assumes that
long-term earnings revert to an industry median return. We also calculate the implied cost of
equity capital based on an abnormal earnings growth valuation model developed by Ohlson and
Juettner-Nauroth (2005). Lastly, we compute the implied cost of capital based on the modified
price earnings growth (PEG) model developed by Easton (2004). For each of these calculations
we utilize the mean values of analysts’ forecasts of earnings, dividends, and book value per
share, the mean analyst long-term growth forecast (gathered from I/B/E/S), and the CRSP
reported market price as of the last day of the sixth month of the fiscal year. See Appendix B for
more detailed discussion of the formulas and model specifications for each cost of capital
measure included in AVG_RATE.
All four implied cost of capital measures utilize analyst forecasts of earnings per share,
dividends per share, book value per share, and/or forecasted long-term growth and therefore are
not without measurement error and bias. Consistent with Daske et al. (2008) and McInnis
(2010), we attempt to control for bias in analyst forecasts by including in equation (1) the prior
year analyst forecast error (FC_BIAS). The implied cost of capital models also require
assumptions regarding the forecast horizon and long-term growth, and as observed by Hail and
Leuz (2006), these models are based on earnings and therefore could include bias related to
accounting conservatism. These limitations are prevalent in the implied cost of capital literature
and highlight the difficulties in accurately calculating a firm’s cost of equity capital. Given these
limitations, our analyses are based on the average of the four implied cost of capital measures.
18
In supplemental tests, we evaluate the robustness of our results by estimating equation (1)
separately for each individual cost of capital measure (consistent with Callahan et al. 2012).
4. Results
4.1 Descriptive Statistics and Correlations
Table 2, Panel A presents descriptive statistics for the sample utilized to test H1 and H2.
The mean (median) firm-year observation reports total tax reserves (TAX_RES) that are 1.25
(0.74) percent of a firm’s total assets and similar to other recent studies of corporate tax
avoidance, has near zero discretionary permanent book-tax differences (DTAX). By design, the
statistics for SHELTER range from 0 to 1, since we rank all firm-year observations based on raw
SHELTER scores (as calculated in Wilson 2009) and then rank them by decile (i.e., 0 to 9) and
scale by 9. Table 2 also indicates that the mean (median) five-year CASH_ETR is 27.2 (25.8)
percent, while the mean (median) volatility of cash ETRs over the five prior years
((CASH_ETR)) is 0.422 (0.108). During our sample period the mean (median) average implied
cost of capital, adjusted for the risk free rate of return (AVG_RATE) is 8.2 (6.1) percent. These
statistics are similar to those in Callahan et al. (2012). We also note that the average firm has
operations in foreign countries (FOR_OPER indicator variable mean = 0.715); reports relatively
high R&D expenditures (mean = 3.4 percent of lagged total assets) and leverage (mean = 0.487);
but exhibits little analyst forecast bias (FC_BIAS). In addition, because most observations have
positive ROA, most sample firms have incentives to tax plan and reduce their corporate income
taxes.
[Insert Table 2 here]
19
Panel B presents the distribution of sample observations across the 30 Fama and French
industry classifications, available on Ken French’s website.16 The industries with the largest
proportions of observations are Personal and Business Services (line 22) and Business
Equipment (line 23), followed by Healthcare, Medical Equipment, and Pharmaceutical Products
(line 8) and Retail (line 27).
Table 3 provides the Pearson and Spearman correlation coefficients amongst the four
implied cost of capital measures, on which AVG_RATE is based. The results indicate that the
Gebhardt, Lee, and Swaminathan (2001) measure is highly correlated with all three of the other
implied cost of capital measures, while the Easton (2004) measure exhibits the smallest
correlations with the other measures. Nonetheless, all of the implied cost of capital measures are
significantly correlated with each other (correlations ranging from 0.142 to 0.698), consistent
with the measures capturing similar aspects of the implied cost of equity capital.
[Insert Table 3 here]
Table 4 provides the Pearson and Spearman correlation coefficients amongst the proxies
for tax risk and measures of firm risk, including AVG_RATE. Most of the correlations between
AVG_RATE and the proxies for TAX_RISK (in column and row 7) are not as predicted. For
example, the Spearman correlation between AVG_RATE and TAX_RES is -0.044 and the Pearson
correlation between AVG_RATE and SHELTER is -0.161. We note, however, that cash ETR
volatility ((CASH_ETR)) is positively, significantly correlated with AVG_RATE, consistent
with predictions in Guenther et al. (2013). Nonetheless, most of the correlations amongst the
proxies for TAX_RISK are as expected (e.g., TAX_RES is positively correlated with TAX_RES
and SHELTER, while DTAX is positively correlated with SHELTER). Consistent with prior
16 http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/f-f_factors.html.
20
research, CASH_ETR is negatively correlated with all four proxies of TAX_RISK, consistent with
high tax risk firms reporting lower cash ETRs. Amongst the proxies for firm risk, AVG_RATE is
significantly, positively correlated with all three measures: stock return volatility ((RET)), cash
flow volatility ((CFO)), and market beta (MKT), but all of the correlations are relatively small
in magnitude, consistent with these alternative proxies for firm risk capture different aspects of
firm risk.
[Insert Table 4 here]
4.2 Multivariate Results for the Association between Tax Risk and the Cost of Equity Capital
H1 predicts that tax risk is positively associated with the implied cost of equity capital.
We present our primary results for tests of this hypothesis in Table 5. Each column contains
results for estimations of equation (1) based on a different proxy for tax risk. In column 1, we
find a positive and significant coefficient on TAX_RISK (where TAX_RISK = TAX_RES),
consistent with larger tax reserves being associated with a higher cost of equity capital. This
result supports H1. However, none of the other three proxies for tax risk (i.e., TAX_RES in
column 2, DTAX in column 3, and SHELTER in column 4) are significantly associated with
AVG_RATE in Table 5. Contrary to concurrent research, we find no evidence that CASH_ETR is
significantly associated with AVG_RATE, after controlling for a firm’s exposure to tax risk.17
Amongst the other control variables, the results in Table 5 indicate that capital expenditures,
foreign operations, profitability, discretionary accruals, analyst forecast errors, market beta, and
the book-to-market risk factor (HML) are all consistently associated with the implied cost of
equity capital, in addition to a firm’s ending tax reserve balance (TAX_RISK in column 1).
17 In a concurrent working paper, Goh et al. (2013) find that the implied cost of equity capital is decreasing in corporate tax avoidance, as proxied by CASH_ETR. Given the different sample periods and the lack of controls for tax risk in the Goh et al. (2013) study, it is difficult to reconcile the contrasting results for CASH_ETR between the two studies.
21
Overall, the results in Table 5 suggest that only a firm’s ending tax reserve balance captures
greater uncertainty with respect to future outcomes, as measured by its positive association with
the implied cost of capital.
[Insert Table 5 here]
The lower portion of Table 5 presents the results of the Vuong (1989) and Clarke (2003)
log-likelihood ratio tests. The Clarke statistics presented at the bottom of column 1 indicate that
TAX_RES has greater explanatory power for the cost of equity capital than TAX_RES (column
2), DTAX (column 3), and SHELTER (column 4). These results are consistent with H2, which
predicts that tax reserves are more highly associated with the cost of equity capital than other
proxies for a firm’s tax risk. Nonetheless, we acknowledge that the adjusted R-squareds exhibit
little variation across the four columns (range from 63.45 percent in column 3 to 63.68 percent in
column 1). Thus, variation in the explanatory power of the four TAX_RISK proxies is
economically small, despite the significance of the Clarke statistics at the bottom of column 1.
Guenther et al. (2013) utilize CASH_ETR as a proxy for tax avoidance, TAX_RES as a
proxy for tax aggressiveness, and (CASH_ETR) as a proxy for tax risk. Those authors assert
that (CASH_ETR) best captures uncertainty regarding a firm’s future tax payments (i.e., their
definition of tax risk), while TAX_RES best captures the extent to which a firm takes tax
positions that are unlikely to survive challenge by the IRS (i.e., their definition of tax
aggressiveness). However, Guenther et al. (2013) acknowledge that aggressive tax policies
could increase firm risk if there is a high degree of uncertainty with regard to future tax
payments. We build on their analyses, which test whether CASH_ETR, TAX_RES, and/or
(CASH_ETR) are associated with future stock return volatility (whereas we focus on the
implied cost of capital), and add cash ETR volatility to equation (1) as an alternative proxy for
22
tax risk. Consistent with Guenther et al. (2013), we calculate (CASH_ETR) as the standard
deviation of one-year cash effective tax rates over years t-4 through year t.
Table 6 presents the results for these alternative estimations of equation (1), where our
proxy for TAX_RISK alternates between TAX_RES, TAX_RES, DTAX, and SHELTER in
columns 1-4, but we include CASH_ETR and (CASH_ETR) in all four regressions models. We
find that the coefficients on TAX_RISK in Table 6 are substantially similar to those in Table 5
(the coefficient on TAX_RISK = TAX_RES is similar in both size and significance level). In
addition, none of the coefficients on either CASH_ETR or (CASH_ETR) are significant. These
results suggest that only the ending balance of tax reserves captures uncertainty regarding future
after-tax cash flows, as reflected in the implied cost of equity capital. We acknowledge,
however, that we cannot reconcile the inferences from our results in Table 6 to inferences from
the results in Guenther et al. (2013), given the substantially different sample periods and research
designs.18
[Insert Table 6 here]
4.3 Results of Supplemental Analyses and Robustness Tests
We perform a variety of supplemental analyses to evaluate the strength of our results.
First, we consider alternative measures of firm risk, including stock return volatility, the
volatility of cash flow from operations, and market model beta. We calculate stock return
volatility ((RET)) based on monthly stock returns starting with the fourth month of the current
fiscal year through the third month of the following year. We calculate stock return volatility
over this time period because stock returns during this one-year window should reflect investor
18 The sample period in Guenther et al. (2013) is 1987-2011 and their primary multivariate test regresses future stock return volatility on CASH_ETR, (CASH_ETR), the level and the volatility of cash flow from operations, and other control variables.
23
expectations based on information available during the current fiscal year and until the Form 10-
K is filed with the SEC (for most firms). That is, we view this calculation are consistent with our
calculation of the implied cost of equity capital. In contrast, we calculate cash flow volatility
((CFO)) based on annual data starting with year t-4 through year t. We choose this time period
primarily for practical reasons, as tax reserve data is only available for fiscal years 2007 and
thereafter, which severely limits our ability to calculate cash flow volatility based on future cash
flow data. Lastly, we also utilize market beta as an alternative measure of systematic (i.e., non-
diversifiable) firm risk, which is calculated as described in Appendix B. We estimate equation
(1) based on these three alternative measures of firm risk, but exclude EARN_VOL from the
regression where (CFO) is the dependent variable (due to their extremely high correlation
coefficients), and for obvious reasons we also exclude market beta from the regression where
market beta is the dependent variable.
Table 7 presents results for regressions that are based on the three alternative proxies for
firm risk. In these regressions we include TAX_RES as our only proxy for tax risk, since only the
coefficients on TAX_RES are significant in our cost of capital regressions in Tables 5 and 6. The
results in Table 7 indicate that the ending balance of tax reserves is significantly and positively
associated with both the stock return ((RET)) and cash flow volatility ((CFO)) measures. In
contrast, the coefficient on TAX_RES is negative and significant in column 3, where market beta
(MKT) is our proxy for firm risk. We also note that firms with lower cash ETRs tend to have
lower stock return and cash flow volatility, consistent with predictions in Goh et al. (2013).
From the (RET) regression results in column 1, we infer that investors perceive tax reserves as
capturing greater uncertainty with respect to future after-tax cash flows. We also conclude
(based on the (CFO) results in column 2) that tax reserves are associated with recent cash flow
24
volatility, consistent with tax reserves capturing more volatile and uncertain tax positions that
inconsistently affect after-tax cash flow from operations. Lastly, we are somewhat puzzled by
the significant and negative coefficient on tax reserves in the market beta (MKT) regression, as
this result suggests that firms with more uncertain tax positions actually experience lower
systematic risk.
[Insert Table 7 here]
Given the difficulty in computing a firm’s cost of equity capital, the regressions in Tables
5 through 7 are based on AVG_RATE as the dependent variable. AVG_RATE is the average of
four commonly used cost of capital measures developed in Claus and Thomas (2001), Gebhardt,
Lee, and Swaminathan (2001), Ohlson and Juettner-Nauroth (2005), and Easton (2004). To
evaluate the consistency of our primary results across these four separate measures of the cost of
equity capital, we re-estimate equation (1) based on each separate measure. Moreover, given our
consistent findings that TAX_RES is the only proxy for tax risk significantly associated with
AVG_RATE, we only include TAX_RES as our proxy for tax risk in these robustness tests. We
report the results for these separate regressions in Table 8. In three of four regressions, the
coefficient on TAX_RES is positive and significant. Only the coefficient on TAX_RES in column
1 is not significant, where the implied cost of capital is based on the measure in Claus and
Thomas (2001). We infer that our results are fairly robust across the four separate implied cost
of capital measures.
[Insert Table 8 here]
Lastly, to be included in our primary analyses (Table 5), an observation need only have
data for one (or more) cost of capital measures, on which we base AVG_RATE. In untabulated
robustness tests, we also require observations to have data for all four implied cost of capital
25
measures to evaluate whether inconsistent data requirements in calculating AVG_RATE influence
our results. Based on this reduced sample of 933 firm-year observations, we continue to find a
significant and positive coefficient on TAX_RES (coefficient = 0.800; t-statistic = 2.04), but none
of the coefficients on the other tax risk proxies are significant. Thus, inconsistency in the
computation of AVG_RATE does not account for our primary findings.
5. Conclusions
The objective of this study is to evaluate the extent to which financial statement-based
proxies for tax risk (i.e., tax positions that increase uncertainty with regard to future outcomes)
are associated with a firm’s implied cost of equity capital. Theory would suggest that each
measure captures, to varying degrees, both tax risk and higher after-tax cash flows. By analyzing
the association between several measures of aggressive tax avoidance and the cost of equity
capital, we provide evidence that the level of a firm’s tax reserves (as reported under FIN 48)
best captures tax positions with uncertain future outcomes, while the change in tax reserves,
discretionary permanent book-tax differences, and a tax shelter prediction score likely do not.
We also examine whether tax risk is associated with other measures of firm risk,
including stock return volatility, cash flow volatility, and market model beta. Consistent with
our implied cost of capital results, we also find that the level of a firm’s tax reserves are
significantly associated with both stock return and cash flow volatility measures. Overall, we
infer that the reserve for income taxes best captures tax positions that involve uncertain future
outcomes, as measured by market expectations and recent cash flow volatility measures. Our
results are robust to numerous controls for factors known to be associated with measures of firm
risk, and also alternative measures of tax risk, including volatility of cash effective tax rates.
26
Our study expands our understanding of how investors measure and evaluate corporate
tax avoidance and tax risk. The findings support Hanlon and Heitzman’s (2010) call to be
cautious when selecting a measure of aggressive tax avoidance, since each measure likely
captures elements of both tax risk and increased after-tax cash flows. While DTAX, SHELTER,
and CASH_ETR have been recently used to measure aggressive tax planning, they do not appear
to significantly capture tax risk. These variables may nonetheless capture the benefits of
corporate tax avoidance, namely increased after-tax cash flows. Overall, our results suggest that
tax reserves are a superior proxy for tax positions that increase uncertainty with regard to future
outcomes.
Our study is subject to several limitations. First, like other studies on the cost of equity
capital, we measure the implied cost of capital with error. We attempt to reduce the impact of
measurement error by controlling for numerous factors that are likely associated with firm risk
and/or tax risk in our multivariate analyses. Nonetheless, to the extent tax reserves are correlated
with the error in our implied cost of capital estimates, our findings may be spurious. Second,
because FIN 48 has only been in effect for fiscal years 2007 and thereafter, our analyses are
based on a limited time series of data. Consequently, results based on a longer time series of
data may differ due to increased power (and in fact our short time series may contribute to
differences between our results and those in Guenther et al. (2013) and Goh et al. (2013)).
Lastly, our current study does not consider the impact of corporate governance on the association
between tax risk and the cost of equity capital. Future research should investigate whether
corporate governance strength – and which governance mechanisms – impact tax risk and also
the extent to which corporate governance influences how investors and analysts perceive tax risk.
27
REFERENCES
Ashbaugh-Skaife, H., D. W. Collins, and R. LaFond. 2009. The effect of SOX internal control deficiencies on firm risk and cost of equity. Journal of Accounting Research 47 (1): 1-43.
Balakrishnan, K., J. Blouin, and W. Guay. 2012. Does tax aggressiveness reduce corporate
transparency? Working paper. Botosan, C. A. 1997. Disclosure level and the cost of equity capital. Accounting Review: 323-
349. Botosan, C. A., and M. A. Plumlee. 2002. A re‐examination of disclosure level and the expected
cost of equity capital. Journal of Accounting Research 40 (1): 21-40. ———. 2005. Assessing alternative proxies for the expected risk premium. The Accounting
Review 80 (1): 21-53. Callahan, C. M., R. E. Smith, and A. W. Spencer. 2012. An examination of the cost of capital
implications of FIN 46. The Accounting Review 87 (4): 1105-1134. Cazier, R., S. Rego, X. Tian, and R. Wilson. 2012. Did increased disclosure requirements and the
standarization of accounting practices reduce earnings management through the reserve for income taxes? Working paper.
Clarke, K. A. 2001. Testing nonnested models of international relations: Reevaluating realism.
American Journal of Political Science: 724-744. ———. 2003. Nonparametric model discrimination in international relations. Journal of Conflict
Resolution 47 (1): 72-93. Claus, J., and J. Thomas. 2001. Equity premia as low as three percent? Evidence from analysts'
earnings forecasts for domestic and international stock markets. The Journal of Finance 56 (5): 1629-1666.
Daske, H., L. Hail, C. Leuz, and R. Verdi. 2008. Mandatory IFRS reporting around the world:
Early evidence on the economic consequences. Journal of Accounting Research 46 (5): 1085-1142.
Desai, M. A., and D. Dharmapala. 2006. Corporate tax avoidance and high-powered incentives.
Journal of Financial Economics 79 (1): 145-179. Desai, M. A., and D. Dharmapala. 2009. Corporate tax avoidance and firm value. The Review of
Economics and Statistics 91 (3): 537-546. Dhaliwal, D., S. Heitzman, and O. Zhen Li. 2006. Taxes, leverage, and the cost of equity capital.
Journal of Accounting Research 44 (4): 691-723.
28
Easton, P. D. 2004. PE ratios, PEG ratios, and estimating the implied expected rate of return on
equity capital. The Accounting Review 79 (1): 73-95. Fama, E. F., and K. R. French. 1993. Common risk factors in the returns on stocks and bonds.
Journal of financial economics 33 (1): 3-56. Financial Accounting Standards Board (FASB). Accounting for Uncertainty in Income Taxes:
An Interpretation of FASB Statement No. 109. FASB Interpretation No. 48. Norwalk, CT: FASB. 2006.
Financial Accounting Standards Board (FASB). Accounting Standards Codification: 740
Income Taxes. Norwalk, CT: FASB. 2009. Francis, J., R. LaFond, P. M. Olsson, and K. Schipper. 2004. Costs of equity and earnings
attributes. The Accounting Review 79 (4): 967-1010. Frank, M. M., L. J. Lynch, and S. O. Rego. 2009. Tax reporting aggressiveness and its relation to
aggressive financial reporting. The Accounting Review 84 (2): 467-496. Frischmann, P. J., T. Shevlin, and R. Wilson. 2008. Economic consequences of increasing the
conformity in accounting for uncertain tax benefits. Journal of Accounting and Economics 46 (2): 261-278.
Gebhardt, W. R., C. Lee, and B. Swaminathan. 2001. Toward an implied cost of capital. Journal
of Accounting Research 39 (1): 135-176. Goh, B.W., J. Lee, C.Y. Lim, and T. Shevlin. 2013. The effect of corporate tax avoidance on the
cost of equity. Working paper. Guenther, D.A., S.R. Matsunaga, and B.M. Williams. 2013. Tax avoidance, tax aggressiveness,
tax risk and firm risk. Working Paper. Hail, L., and C. Leuz. 2006. International differences in the cost of equity capital: Do legal
institutions and securities regulation matter? Journal of Accounting Research 44 (3): 485-531.
———. 2009. Cost of capital effects and changes in growth expectations around US cross-
listings. Journal of Financial Economics 93 (3): 428-454. Hanlon, M., and S. Heitzman. 2010. A review of tax research. Journal of Accounting and
Economics 50 (2): 127-178. Koester, A. 2011. Investor valuation of tax avoidance through uncertain tax positions. Working
Paper.
29
Lambert, R., C. Leuz, and R. E. Verrecchia. 2007. Accounting information, disclosure, and the cost of capital. Journal of Accounting Research 45 (2): 385-420.
Lisowsky, P. 2010. Seeking shelter: Empirically modeling tax shelters using financial statement
information. The Accounting Review 85 (5): 1693-1720. Lisowsky, P., L. Robinson, and A. Schmidt. 2013. Do Publicly Disclosed Tax Reserves Tell Us
About Privately Disclosed Tax Shelter Activity? Journal of Accounting Research. McInnis, J. 2010. Earnings smoothness, average returns, and implied cost of equity capital. The
Accounting Review 85 (1): 315-341. Ohlson, J. A., and B. E. Juettner-Nauroth. 2005. Expected EPS and EPS growth as determinants
of value. Review of Accounting Studies 10 (2-3): 349-365. PRICEWATERHOUSECOOPERS LLP. 2004. “Tax Risk Management.” (April 2004). Rego, S. O., and R. Wilson. 2012. Equity risk incentives and corporate tax aggressiveness.
Journal of Accounting Research 50 (3): 775-810. Vuong, Q. H. 1989. Likelihood ratio tests for model selection and non-nested hypotheses.
Econometrica: Journal of the Econometric Society: 307-333. Weisbach, D. A. 2001. Ten truths about tax shelters. Tax L. Rev. 55:215. Wilson, R. J. 2009. An examination of corporate tax shelter participants. The Accounting Review
84 (3): 969-999.
30
APPENDIX A Variable Definitions
Proxies for Tax Risk: TAX_RES = Total tax reserves at fiscal year-end (TXTUBEND), scaled by total assets (AT) at
fiscal year-end. TAX_RES = The change in total tax reserves from year t-1 (TXTUBBEGIN) to year t
(TXTUBEND), scaled by total assets at the beginning of the fiscal year. DTAX = Discretionary permanent differences, as calculated in Frank et al. 2009. Equals the
residual of the following equation: PERMDIFF = α0 + α1INTANG + α2UNCON + α3MI + α4CSTE + α5ΔNOL +
α6LAGPERM + ε (1) Where all variables are scaled by beginning of year total assets. (2) PERMDIFF is total book-tax differences [pre-tax book income (PI) less
current federal expense (TXFED) and current foreign expense (TXFO) divided by the statutory tax rate] less temporary book-tax differences [deferred tax expense (TXDI) divided by the statutory tax rate of 35 percent]
(3) INTANG is the sum of goodwill (GDWL) and other intangibles (INTANO) (4) UNCON is income reported under the equity method (ESUB) (5) MI is minority interest (MII) (6) CSTE is the current state income tax expense (TXS) (7) ΔNOL is the change in net operating loss carryforwards (TLCF) (8) And LAGPERM is the one year lagged PERMDIFF.
SHELTER = The rank value of the tax shelter prediction score in Wilson (2009), i.e., the predicted
value from the following equation: TSPS = -4.86 + 5.2*BTD + 4.08*DAP – 1.41*LEV + 0.76*SIZE + 3.51*ROA +
1.72*FINC + 2.43*R&D (1) Where BTD, book-tax difference, is equal to pre-tax book income (PI) less
taxable income [federal tax expense (TXFED) plus foreign tax expense (TXFO) divided by the statutory tax rate of 35%] less the change in NOL carryforwards (TLCF).
(2) DAP, discretionary accruals, is calculated using the cross-sectional modified Jones model w/ lagged return on assets.
(3) LEV, leverage ratio, long-term debt (DLTT) divided by total assets (AT). (4) SIZE is the natural log of total assets (AT). (5) ROA, return on assets, is equal to pre-tax income (PI) divided by total assets
(AT). (6) FINC, is an indicator variable for foreign operations, and equals 1 if there is
non-zero foreign income (PIFO), and 0 otherwise. (7) R&D is research and development (XRD) scaled by total assets.
Proxies for Firm Risk: AVG_RATE = Average implied cost of equity capital less the median yield on a 10-year treasury
bond. See Appendix B. (RET) = The annual standard deviation of monthly stock returns from CRSP, calculated
starting with the fourth month of year t through the third month of year t+1.
31
(CFO) = The standard deviation of annual operating cash flows (OANCF) for years t-4 to t,
scaled by total assets at the beginning of year t. Control Variables for Equations (1) – (3): CASH_ETR = The cash effective tax rate (Dyreng et. al 2008), which is the sum of cash taxes paid
(TXPD) for years t-4 through year t, divided by the sum of adjusted pretax income (PI - SPI) for years t-4 through year t.
(CASH_ETR) = The standard deviation of the annual cash effective tax rate for year t-4 through year
t, where the annual cash effective tax rate is cash taxes paid (TXPD) divided by adjusted pretax income (PI-SPI), consistent with Guenther et al 2012.
MKT SMB HML
= The Fama and French (1993) risk factors are computed by regressing a firm’s monthly stock returns (for the period starting sixty-six months prior to fiscal year-end and ending six months prior to fiscal year-end, i.e., the date we calculate the cost of equity capital) on the monthly Fama and French (1993) factors, available at http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/f-f_factors.html.
LEV = Financial leverage, calculated as the ratio of total liabilities (LT) to total assets (AT). FC_BIAS = An estimate of the forecast bias in analysts’ forecasts. Calculated as the prior year’s
one-year ahead earnings per share forecast from I/B/E/S minus this year’s actual earnings per share (NI divided by SHOUT), scaled by total assets (AT).
CAP_EXP = Total capital expenditures for the fiscal year (CAPX), scaled by total assets at the
beginning of the year. R&D = Total research and development expenditures for the fiscal year (XRD), scaled by
total assets at the beginning of the year. SG&A = Total general and administrative expense for the fiscal year (XSGA), scaled by total
assets at the beginning of the year. FOR_OPER = Indicator variable for if the firm has foreign operations. DISCR_ACCR = Discretionary accruals calculated using the cross-sectional modified Jones model
with lagged return on assets. EARN_VOL = The standard deviation of adjusted income (PI-SPI) for the period from t-4 to t,
scaled by total assets at the beginning of year t. ROA = Return on assets, calculated as income before extraordinary items (IB) divided by
lagged total assets. Notes: Where applicable, Compustat variable names are provided in parentheses. Variables gathered from CRSP, I/B/E/S, and Risk Metrics are noted accordingly.
32
APPENDIX B Cost of Capital Models Utilized in the Calculation of AVG_RATE
Following prior literature (Dhaliwal et al. 2006, Daske et al. 2008, Hail and Leuz 2009, Callahan
et al. 2012), AVG_RATE is the average of four commonly-used implied cost of capital measures
developed in Claus and Thomas (2001), Gebhardt, Lee, and Swaminathan (2001), Ohlson and
Juettner-Nauroth (2005) and Easton (2004), less the median rate on the 10-year treasury note.
Claus and Thomas (2001):
0∗
1∗ 11
The Claus and Thomas (2001) model is based on the residual income valuation model. A key
assumption of this model is that for year five and beyond residual income grows at a rate equal to
inflation. Because the model does not have a closed form solution, an iterative process is utilized
to solve for the cost of equity capital (CTR).
− P0 is the price as of the last day of the 6th month before the fiscal year-end.
− bv is the book value per share as of the last day of the 6th month before the fiscal
year-end.
− eps is the earnings per share forecasted as of the last day of the 6th month before
the fiscal year-end.
− f is the expected inflation rate, which we set at 3 percent. Note that the inflation
rate sets the lower bound for the computed cost of capital.
Gebhardt, Lee, and Swaminathan (2001):
∗1
∗1
The Gebhardt, Lee, and Swaminathan (2001) model is also based on the residual income
valuation model; however, this model assumes that in years 4 through 12 residual income reverts
33
to an industry median and in years 13 through 25 residual income remains constant. We
compute the industry median based on the forecasted income for years 1 through 3 for each two
digit SIC code. Similar to the Claus and Thomas (2001) model, an iterative process is utilized to
solve for the cost of equity capital (GLSR) because the model does not have a closed form
solution,
− P0, bv, and eps are previously defined.
Ohlson and Juettner-Nauroth (2005):
∗ ∗ /
The Ohlson and Juettner-Nauroth (2005) model is based on an abnormal earnings growth
valuation model and has a closed form solution for the cost of capital (OJR), provided the firm
has a positive change in forecasted earnings.
− gst is the short-term growth rate, which is estimated as the average of the change
in earnings from year 1 to year 2 and the long-term growth forecast provided by
I/B/E/S.
− glt is the long-term growth forecast provided by I/B/E/S.
− dps is dividends per share as of the last day of the 6th month before the fiscal year-
end.
− P0 and eps are previously defined.
Easton (2004):
Easton (2004) computes a firm’s cost of capital based on the price earnings growth ratio. The
model also provides a closed form solution for the cost of capital (ER), provided the firm has
forecasted earnings growth. Consistent with Callahan et al. (2012), we assume that dividends
per share are zero.
34
− P0 and eps are previously defined.
Additional Specifications:
(1) We require the long-term growth rate to be positive (consistent with Daske et al. 2008)
and constrain the long-term growth rate to 50 percent or less (following Callahan et al.
2012).
(2) We require that forecasted earnings per share is positive (consistent with Daske et al.
2008) and winsorize forecasted earnings per share at the 99th percentile to reduce the
influence of unattainable forecasts (consistent with Callahan et al. 2012).
(3) For the Ohlson and Juettner-Nauroth (2005) model we constrain the dividend payout ratio
(dps1 / eps1) to between 0 and 1 (following Callahan et al. 2012).
(4) For firm-year observations where a long-term growth forecast is available and an
earnings per share forecast is provided for one and two years ahead, but not for
subsequent years, we compute forecasts for three, four, and five years ahead based on the
following equation: epst = epst-1*(1-glt), (consistent with Daske et al. 2008).
(5) For firm-year observations where forecasts of book value per share are not available for
two, three, and four years ahead, we compute the forecasts based on the following
equation: bpst = bpst-1 + epst*(1-DPS1/EPS1), (consistent with Daske et al. 2008).
35
TABLE 1 Sample Selection Procedures
Firm-Years Firms
Number of observations with non-zero total UTBs for fiscal years ending between 12/15/07 and 12/31/11
11,147 2,989
Less: Observations with a cumulative loss for the five year period ending in the observation year
(3,131) (700)
Less: Observations with a negative cumulative tax expense for the five year period ending in the observation year
(523) (114)
Less: Observations with insufficient data to compute DTAX (2,164) (517)
Less: Observations with insufficient data to compute SCORE (198) (50)
Less: Observations with insufficient information to compute AVG_RATE (insufficient data to compute all individual cost of capital calculations)
(1,560) (426)
Less: Real Estate Investment Trusts (SIC 6798) 0 0
Less: Financial Institutions (SIC 60**, 61**, and 62**) (88) (37)
Less: Utilities (SIC 49**) (48) (15)
Less: Observations with insufficient data to compute control variables
(172) (55)
Total Sample 3,263 1,075
Notes: This table presents an overview of the sample selection procedure for the sample utilized in the tests of the association between the measures of tax risk and the cost of equity capital. The table begins with all firms for which total UTBs were reporting in Compustat, with fiscal years ending between December 15, 2007 and December 31, 2011.
36
TABLE 2 Descriptive Statistics
Panel A: Descriptive Statistics Variable N Mean Std Dev 5th 25th Median 75th 95th
TAX_RES 3,263 0.0125 0.0145 0.0007 0.0032 0.0074 0.0162 0.0435
TAXRES 3,263 0.0002 0.0049 -0.0073 -0.0011 0.0001 0.0016 0.0076 DTAX 3,263 0.0032 0.3430 -0.3691 -0.0281 0.0093 0.0739 0.2361 SHELTER 3,263 0.5517 0.2867 0.1000 0.3000 0.6000 0.8000 1.0000 CASH_ETR 3,263 0.2724 0.1763 0.0496 0.1723 0.2580 0.3321 0.5104
(CASH_ETR) 3,263 0.4223 2.3762 0.0275 0.0604 0.1081 0.2216 0.9140 AVG_RATE 3,263 0.0819 0.0773 0.0078 0.0342 0.0609 0.1045 0.2295
(RET) 3,263 0.1018 0.0550 0.0430 0.0659 0.0900 0.1234 0.1923
(CFO) 3,263 0.0460 0.0347 0.0127 0.0236 0.0373 0.0572 0.1089
MKT 3,263 0.0114 0.0061 0.0028 0.0073 0.0107 0.0148 0.0221
CAP_EXP 3,263 0.0486 0.0477 0.0072 0.0181 0.0326 0.0615 0.1454 R&D 3,263 0.0343 0.0577 0.0000 0.0000 0.0047 0.0467 0.1499 SG&A 3,263 0.2850 0.2290 0.0410 0.1331 0.2334 0.3767 0.6665 FOR_OPER 3,263 0.7150 0.4515 0.0000 0.0000 1.0000 1.0000 1.0000 LEV 3,263 0.4866 0.2167 0.1561 0.3267 0.4805 0.6234 0.8458 ROA 3,263 0.0622 0.0982 -0.0697 0.0267 0.0607 0.1016 0.3414 DISCR_ACCR 3,263 -0.0040 0.0395 -0.0670 -0.0247 -0.0022 0.0150 0.0614 FC_BIAS 3,263 0.0004 0.0024 -0.0016 -0.0002 0.0000 0.0005 0.0043 EARN_VOL 3,263 0.0426 0.0378 0.0089 0.0195 0.0320 0.0523 0.1132
SMB 3,263 0.0076 0.0090 -0.0056 0.0017 0.0068 0.0127 0.0222
HML 3,263 0.0000 0.0094 -0.0147 -0.0055 -0.0002 0.0051 0.0157 Notes: This table presents descriptive statistics for all variables in the sample utilized in the tests of the association between tax risk and the cost of equity capital. The sample includes firms with fiscal years ending between December 15, 2007 and December 31, 2011 with non-missing values for all variables. All continuous variables were winsorized at the 1st and 99th percentiles. All variables are defined in Appendix A.
37
TABLE 2 - Continued
Panel B: Industry Membership Proportion of:
Industry Description Our
Sample Compustat Population
1 Food Products 2.12% 2.08% 2 Beer & Liquor 0.61% 0.30% 3 Smoke 0.00% 0.11% 4 Games & Recreation 2.18% 1.86% 5 Books, Printing & Publishing 1.44% 0.63% 6 Household Consumer Goods 2.15% 1.06% 7 Clothing/Apparel 2.42% 0.94% 8 Healthcare, Medical Equipment, Pharmaceutical Products 9.68% 11.27% 9 Chemicals 3.25% 1.85% 10 Textiles 0.00% 0.19% 11 Construction and Construction Materials 2.88% 2.04% 12 Steel Works 1.47% 1.14% 13 Fabricated Products and Machinery 5.03% 2.53% 14 Electrical Equipment 1.26% 1.42% 15 Automobiles and Trucks 1.50% 1.32% 16 Aircraft, ships, and railroad equipment 1.66% 0.56% 17 Precious Metals, Non-Metallic, and Industrial Metal Mining 0.46% 3.21% 18 Coal 0.28% 0.34% 19 Petroleum and Natural Gas 3.43% 5.03% 20 Utilities 0.00% 4.19% 21 Telecommunications 3.52% 3.44% 22 Personal and Business Services 16.06% 11.30% 23 Business Equipment 16.55% 9.90% 24 Business Supplies and Shipping Containers 2.52% 1.07% 25 Transportation 1.53% 2.89% 26 Wholesale 5.27% 2.74% 27 Retail 6.99% 3.57%
28 Restaurants, Hotels, Motels 2.36% 1.33%
29 Insurance, Real Estate, Trading 1.53% 21.03%
30 Other 0.00% 0.68%
Missing 1.87% 0.00%
Total Firm-Year Observations 3,263 40,261 Notes: This table presents industry classifications for the sample utilized in the tests of the association between tax risk and cost of equity capital. Industry classifications are based on the Fama and French 30-industry model using four-digit SIC codes. Classification specifications are available on the website of Ken French: http://mba.tuck.dartmouth.edu/pages/faculty/ken.french/Data_Library/det_30_ind_port.html
38
TABLE 3 Spearman (Pearson) Correlations on Upper (Lower) Diagonal for Individual Cost of
Capital Measures Included in AVG_RATE
(1)
Easton
(2) Ohlson and
Juettner
(3) Claus and Thomas
(4) Gebhardt, Lee, and Swamin.
(1) 0.61167 0.26483 0.46299
(2) 0.29505 0.31193 0.57385
(3) 0.14233 0.31275 0.68302
(4) 0.20323 0.40706 0.69811 Notes: This table presents Spearman (Pearson) correlation coefficients in the upper (lower) diagonal. Bolded correlation coefficients are significant at the 10% level or better, based on two-sided tests. Sample size is 857 observations. All variables are as defined in Appendix B.
39
TABLE 4 Spearman (Pearson) Correlations on Upper (Lower) Diagonal for Tax and Firm Risk Measures
(1)
TAX_RES (2)
TAXRES (3)
DTAX (4)
SHELTER (5)
CASH_ETR (6)
(CASH_ETR) (7)
AVG_RATE (8)
(RET) (9)
(CFO) (10) MKT
(1) 0.256858 0.004173 0.240204 -0.147150 -0.049245 -0.044261 0.033373 -0.042420 0.052857
(2) 0.249647 0.001687 0.055848 -0.123350 -0.078951 -0.059334 0.104963 0.036273 0.047827
(3) -0.033389 -0.024185 0.119209 -0.113020 -0.048233 -0.084531 -0.069844 -0.060659 0.027283
(4) 0.170027 0.034410 0.053331 -0.110650 -0.284264 -0.173593 -0.146519 -0.370496 0.029506
(5) -0.082760 -0.070110 -0.074300 -0.141560 0.281260 0.090140 0.032088 0.024117 -0.133870
(6) 0.006253 -0.020705 0.018634 -0.074447 0.149930 0.187016 0.211728 0.298067 0.159423
(7) -0.014162 -0.031861 -0.022748 -0.161566 0.092630 0.031145 0.064132 0.237860 0.162604
(8) 0.119686 0.116383 -0.092454 -0.105581 0.057826 0.030531 0.056985 0.290529 0.123815
(9) -0.015186 0.032176 -0.029724 -0.330124 0.085581 0.076455 0.306694 0.210569 0.239014
(10) 0.049287 0.034174 0.019759 0.028831 -0.053230 0.048071 0.167517 0.121126 0.233077
Notes: This table presents Spearman (Pearson) correlation coefficients in the upper (lower) diagonal. Bolded correlation coefficients are significant at the 10% level or better, based on two-sided tests. Sample size is 3,263 observations. All variables are as defined in Appendix A.
40
TABLE 5 Results for OLS Regressions of the Implied Cost of Equity Capital (AVG_RATE) on Proxies for Tax Risk and Control Variables
(1)
TAX_RISK = TAX_RES (2)
TAX_RISK = TAX_RES (3)
TAX_RISK = DTAX (4)
TAX_RISK = SHELTER Coeff T-Stat Coeff T-Stat Coeff T-Stat Coeff T-Stat TAX_RISK 0.412 1.97** 0.114 0.43 -0.001 -0.35 0.002 0.21
CASH_ETR -0.006 -0.43 -0.006 -0.42 -0.005 -0.39 -0.006 -0.42
CAP_EXP -0.440 -8.56*** -0.443 -8.62*** -0.444 -8.71*** -0.443 -8.63***
R&D -0.079 -1.11 -0.074 -1.05 -0.074 -1.04 -0.074 -1.04
SG&A -0.042 -1.52 -0.040 -1.45 -0.038 -1.38 -0.039 -1.45
FOR_OPER 0.019 1.97** 0.019 1.96** 0.019 1.94* 0.019 1.81*
LEV -0.000 -0.01 0.003 0.16 0.004 0.20 0.003 0.18
ROA -0.167 -6.87*** -0.171 -7.08*** -0.169 -6.93*** -0.173 -6.78***
DISCR_ACCR -0.096 -2.66*** -0.098 -2.70*** -0.101 -2.8*** -0.099 -2.74***
FC_BIAS -2.905 -3.57*** -2.945 -3.61*** -2.921 -3.59*** -2.949 -3.61***
EARN_VOL 0.128 1.71* 0.134 1.79* 0.130 1.74* 0.137 1.83*
MKT 0.998 2.95*** 0.961 2.85*** 1.025 3.06*** 0.960 2.84***
SMB 0.300 1.22 0.300 1.21 0.330 1.34 0.300 1.21
HML 0.451 2.21** 0.448 2.19** 0.471 2.32** 0.447 2.18**
Industry FE? Y Y N Y Firm FE? Y Y Y Y # of Observations 3,263 3,263 3,263 3,263 Adjusted R2 63.68% 63.61% 63.45% 63.61%
Relative Explanatory Power: Vuong / Clarke P-Values Vuong / Clarke P-Values Vuong / Clarke P-Values Vuong / Clarke P-Values
Different from TAX_RES? .2899/<.0001 .1100/<.0001 .2678/<.0001
Different from TAX_RES? .2899/<.0001 Mixed Results Mixed Results
Different from DTAX? .1100/<.0001 Mixed Results Mixed Results
Different from SHELTER? .2678/<.0001 Mixed Results Mixed Results
41
Notes: This table presents the results for regressions of the cost of equity capital on proxies for tax risk and control variables. All regressions include both industry and firm fixed effects, except where noted. *, **, *** indicate statistical significance at the 10, 5, and 1 percent levels, respectively, based on two-sided t-tests. Variables are defined in Appendix A. The variable DTAX is calculated by year and 2-digit SIC code combinations. Thus, the regression that includes DTAX does not include industry fixed effects. The p-values for the Voung and Clarke tests are presented in that order in the lower portion of Table 4. “Mixed Results” indicates that results from the Voung and Clarke tests do not agree with respect to which model has greater explanatory power for the cost of capital model.
42
TABLE 6 Results for OLS Regressions of the Implied Cost of Equity Capital (AVG_RATE) on Proxies for Tax Risk, including (CASH_ETR)
(1)
TAX_RISK = TAX_RES (2)
TAX_RISK = TAX_RES (3)
TAX_RISK = DTAX (4)
TAX_RISK = SHELTER Coeff T-Stat Coeff T-Stat Coeff T-Stat Coeff T-Stat
TAX_RISK 0.405 1.93* 0.116 0.44 -0.001 -0.37 0.002 0.23
CASH_ETR -0.007 -0.48 -0.007 -0.48 -0.006 -0.47 -0.007 -0.48
(CASH_ETR) 0.001 1.03 0.001 1.10 0.001 1.09 0.001 1.10
CAP_EXP -0.439 -8.56*** -0.442 -8.61*** -0.444 -8.70*** -0.443 -8.62***
R&D -0.077 -1.08 -0.073 -1.02 -0.072 -1.02 -0.072 -1.01
SG&A -0.043 -1.57 -0.041 -1.51 -0.039 -1.43 -0.041 -1.50
FOR_OPER 0.019 1.96** 0.019 1.96* 0.019 1.94* 0.018 1.80*
LEV -0.000 -0.02 0.003 0.14 0.004 0.20 0.003 0.17
ROA -0.166 -6.83*** -0.170 -7.03*** -0.168 -6.88*** -0.172 -6.74***
DISCR_ACCR -0.094 -2.61*** -0.096 -2.65*** -0.099 -2.75*** -0.097 -2.69***
FC_BIAS -2.866 -3.51*** -2.902 -3.56*** -2.882 -3.54*** -2.908 -3.56***
EARN_VOL 0.125 1.67* 0.131 1.75* 0.127 1.70* 0.133 1.78*
MKT 0.995 2.94*** 0.958 2.84*** 1.024 3.05*** 0.957 2.83***
SMB 0.295 1.19 0.294 1.19 0.325 1.32 0.294 1.19
HML 0.444 2.17** 0.441 2.15** 0.464 2.28** 0.439 2.15**
Industry FE? Y Y N Y Firm FE? Y Y Y Y # of Observations 3,263 3,263 3,263 3,263 Adjusted R2 63.70% 63.63% 63.47% 63.63% Notes: This table presents the results for regressions of the cost of equity capital on proxies for tax risk, σ(CASH_ETR), and control variables. All regressions include both industry and firm fixed effects, except where noted. *, **, *** indicate statistical significance at the 10, 5, and 1 percent levels, respectively, based on two-sided t-tests. Variables are defined in Appendix A. The variable DTAX is calculated by year and 2-digit SIC code combinations. Thus, the regression that includes DTAX does not include industry fixed effects.
43
TABLE 7 Results for OLS Regressions of Alternative Measures of Firm Risk on Tax Risk and Control Variables
(1)
FIRM_RISK = (RET) (2)
FIRM_RISK = (CFO) (3)
FIRM_RISK = MKT Coeff T-Stat Coeff T-Stat Coeff T-Stat
TAX_RES 0.431 2.72*** 0.266 4.16*** -0.039 -2.91***
CASH_ETR 0.017 1.65* 0.012 2.99*** -0.002 -2.53***
(CASH_ETR) 0.000 0.18 0.001 2.63*** 0.000 0.50
CAP_EXP -0.029 -0.74 0.023 1.46 -0.017 -5.36***
R&D 0.043 0.80 0.068 3.14*** -0.013 -2.80***
SG&A -0.014 -0.69 0.068 8.35*** 0.002 1.10
FOR_OPER -0.008 -1.08 0.003 1.05 0.000 0.46
LEV 0.112 7.75*** 0.022 3.76*** 0.000 0.01
ROA -0.006 -0.34 0.072 10.33*** 0.003 1.61
DISCR_ACCR -0.006 -0.20 -0.059 -5.34*** -0.003 -1.24
FC_BIAS 2.848 4.62*** 0.629 2.56** -0.002 -0.04
EARN_VOL 0.258 4.56*** 0.014 2.90***
MKT 1.273 4.98*** 0.229 2.23**
SMB -0.153 -0.82 0.075 1.00 -0.245 -16.59***
HML 0.745 4.82*** 0.107 1.71* 0.032 2.48**
Industry FE? Y Y Y
Firm FE? Y Y Y
# of Observations 3,263 3,263 3,263
Adjusted R2 59.06% 83.31% 76.60% Notes: This table presents the results for regressions of various measures of firm risk on proxies for tax risk and control variables. All regressions include both industry and firm fixed effects. *, **, *** indicate statistical significance at the 10, 5, and 1 percent levels, respectively, based on two-sided t-tests. Variables are defined in Appendix A.
44
TABLE 8 Results for OLS Regressions of Individual Implied Cost of Equity Capital Measures on TAX_RES and Control Variables
(1)
Cost of Equity Capital = CT_R
(2) Cost of Equity Capital =
GLS_R
(3) Cost of Equity Capital =
OJ_R
(4) Cost of Equity Capital =
EASTON_R
Coeff T-Stat Coeff T-Stat Coeff T-Stat Coeff T-Stat
TAX_RES -0.175 -0.70 0.311 2.30** 0.914 2.12** 1.452 3.01***
CASH_ETR -0.011 -0.50 0.002 0.25 -0.007 -0.17 0.063 1.96*
CAP_EXP -0.145 -2.21** -0.106 -3.19*** -0.328 -3.05*** -0.951 -8.40***
R&D -0.046 -0.58 -0.038 -0.84 0.253 0.74 0.136 0.72
SG&A -0.005 -0.14 -0.072 -4.03*** -0.052 -0.76 -0.092 -1.38
FOR_OPER 0.031 2.48** 0.016 2.47** -0.003 -0.13 0.004 0.17
LEV 0.004 0.16 0.054 4.42*** 0.027 0.65 -0.098 -2.09***
ROA -0.120 -3.60*** -0.053 -3.28*** -0.177 -2.85*** -0.101 -1.80*
DISCR_ACCR 0.024 0.50 -0.037 -1.57 -0.139 -1.66* -0.142 -1.63
FC_BIAS -7.269 -5.41*** -0.194 -0.34 -5.142 -2.31** -2.146 -1.26
EARN_VOL -0.114 -1.13 0.074 1.52 0.817 4.94*** 0.463 2.63***
MKT 1.142 2.49** 0.789 3.63*** 1.875 2.45** 2.332 2.97***
SMB 0.480 1.44 -0.152 -0.92 0.494 0.85 1.052 1.91*
HML 0.472 1.80* 0.265 2.00** 0.470 1.02 1.479 2.93***
Industry FE? Y Y Y Y
Firm FE? Y Y Y Y
# of Observations
2,117 2,966 1,292 1,958
Adjusted R2 82.14% 73.30% 73.01% 65.79%
45
Notes: This table presents the results of the regression of the cost of equity capital on TAX_RES, utilizing each of the four measures of cost of equity capital as dependent variables. All regressions include both industry and firm fixed effects. *, **, *** indicate statistical significance at the 10, 5, and 1 percent levels, respectively, based on two-sided t-tests. Variables are defined in Appendix A and Appendix B.