The Effects of Increased Financial Statement Disclosure Quality on Tax Avoidance: An Examination of SEC Comment Letters

Thomas R. Kubick Assistant Professor

University of Kansas tkubick@ku.edu

Daniel P. Lynch* Assistant Professor

University of Wisconsin –Madison dlynch@bus.wisc.edu

Michael A. Mayberry Assistant Professor

University of Florida michael.mayberry@warrington.ufl.edu

Thomas C. Omer Professor of Accounting

University of Nebraska-Lincoln thomas.omer@unl.edu

August 2014

Abstract:

This study examines the tax avoidance behavior of firms prior to the issuance, and following the resolution, of tax-related SEC comment letters. We find that firms engaging in higher levels of tax avoidance are more likely to receive a tax-related SEC comment letter. In order to resolve SEC comment letter issues, firms must address the comments and increase disclosure quality to satisfy the SEC’s demands. If the costs of the new disclosures exceed the benefits of the higher tax avoidance, we predict firms will decrease tax avoidance following a tax-related comment letter. Supporting this conjecture we find firms subsequently reduce their level of tax avoidance following the resolution of the SEC comment letter. This is consistent with increased disclosure quality increasing the expected costs of tax avoidance ultimately leading to a decrease in tax avoidance. We also present evidence that investors positively value within-firm increases in tax avoidance but assign lower valuations after receipt of a tax-related comment letter, consistent with investors anticipating additional tax-related costs from regulatory scrutiny. Overall, these results illustrate the interplay between increased financial statement disclosure quality and tax avoidance.

*Corresponding Author We gratefully acknowledge helpful comments from Karla Johnstone and Terry Warfield.

 

2  

The Effects of Increased Financial Statement Disclosure Quality on Tax Avoidance: An Examination of SEC Comment Letters

I. Introduction

The disclosure of tax-related information in financial statements has undergone dramatic

changes in recent years. From the enactment of FIN 48 (ASC 740) in 2007, requiring all public

firms to make financial statement disclosures of aggregated reserves for uncertain tax positions,

to recent calls for corporations to publicly disclose tax returns and report information on a

country by country basis, firms are under pressure from regulators and standard-setters to

increase disclosures of tax-related information in their financial statements.1 Although prior

research has examined the impact of changes in mandatory disclosures, the literature provides

little evidence on the impact of firm-specific changes in disclosure on firms’ tax avoidance

behavior and investors’ reaction to these new disclosures.2 This study contributes to the ongoing

public discussion by examining the tax outcomes of firms before the issuance of and following

the resolution of tax-related Securities & Exchange Commission (SEC) comment letters to

investigate the relation between increased tax-related financial statement disclosure quality and

tax avoidance, as well as, investors’ responses to an increase in disclosure quality.

The SEC examines publicly traded companies’ 10-K filings at least once every three

years (SOX: Section 408) and comment letters are issued for deficient disclosures. Firms are

required to respond to comment letters within ten days.3 Firms’ responses include additional

information, disclosures, and a commitment to adjust future filings, or restating previously issued

 1 http://www.bloomberg.com/news/2011-10-06/on-corporate-taxes-put-the-public-in-publicly-traded-view.html 2 This is in contrast to studies that consider the effects of mandated, general disclosure standards of tax information. See Donohoe and McGill (2012), Abernathy et al. (2013), Ciconte et al. (2014), De Simone et al. (2014), and Robinson and Schmidt (2013) for information regarding the disclosure of tax information through new standards. 3 The SEC examines filings other than the 10-K (e.g. 8-Ks), but concentrates its resources on these critical disclosures. Thus, in this study we focus on SEC comment letters on 10-K filings.

 

3  

filings (Cassell et al. 2013). When the matter is resolved, SEC comment letters and firms’

responses to them are posted on the SEC’s web site, revealing new firm-specific information.

To proxy for firm-specific, tax-related disclosure quality and increases in disclosure

quality, we use tax-related SEC comment letters and their ultimate resolutions. The primary

reason for tax-related SEC comment letters is a lack of disclosure and clarity in the income tax

footnote related to items such as undistributed foreign earnings, uncertain tax positions and

valuation allowances (Deloitte 2012; PWC 2013). Thus, we first examine whether firms with

higher levels of tax avoidance, and presumably higher costs related to disclosure, are more likely

to receive a tax-related SEC comment letter. We then investigate if, subsequent to increases in

disclosure quality (i.e., the resolution of disclosure issues cited in SEC comment letters), firms

engage in lower levels of tax avoidance. We next examine how investors change their

expectations about the probability of losing tax avoidance strategies by considering the valuation

of tax avoidance preceding and subsequent to the disclosure of tax-related SEC comment letters.

Prior research documents inconsistent results with respect to the association between tax

avoidance and mandatory disclosure quality. Consistent with a proprietary cost explanation,

some prior research documents an association between higher levels of tax avoidance and

decreased disclosure quality (Hope, Ma, and Thomas 2013; Robinson and Schmidt 2013; Dyreng

et al. 2014). However, Towery (2012) documents that firms complying with new tax return

disclosure requirements do not reduce tax avoidance in response to these increased private

disclosures to the IRS but instead alter their financial reporting of potentially sensitive tax

positions. Additionally, concurrent research suggests higher tax avoidance is associated with

greater financial complexity, leading to reduced corporate transparency (Balakrishnan et al.

2012, Neuman et al. 2013).This reduction in disclosure quality results in firms attempting to

 

4  

increase the volume of their disclosures in order to offset the reduction in transparency but

without a corresponding increase in the quality of their tax disclosures (Balakrishnan et al. 2012,

Neuman et al. 2013). Given the mixed results concerning the complex relation between tax

avoidance and disclosure quality, we focus on tax-related SEC comment letters as an exogenous

shock to firms’ tax disclosures that allows us to use the firm as its own control, potentially ruling

out alternative explanations and providing a stronger test of the impact of increased tax-related

financial statement disclosure quality on tax avoidance.

There are costs to disclosing tax-related information in the financial statements. Indeed,

strong corporate opposition to new disclosures is consistent with managers’ belief that disclosing

tax information in financial statements is informative to both the IRS and competitors (Graham

et al. 2012). Thus, increasing the disclosure quality of tax information can increase detection

risk, thereby increasing the expected costs of tax avoidance. Tax authorities could use improved

financial statement disclosures to target companies for audit and to target particular tax planning

strategies for examination during audit. We argue that increases in disclosure quality potentially

reveal sensitive tax information, thereby increasing the expected cost of tax avoidance by

increasing detection risk. All else equal, higher expected costs of tax avoidance related to

improved disclosure quality should result in a new, lower level of tax avoidance.

Using a sample of firm-years from 2004-2011, we find a positive association between

higher levels of tax avoidance and the propensity to receive a tax-related SEC comment letter.4

This is consistent with firms’ tax-related disclosures obfuscating information which may result in

additional costs from revealing their tax planning activities to tax authorities. In our primary

analysis, we find that following the resolution of the disclosure issues cited in the SEC comment

 4 We use textual and manual analyses to examine SEC comment letters for the references to disclosure issues related to tax matters.

 

5  

letters, firms engage in lower levels of tax avoidance. Specifically, our estimates suggest that

firms report higher GAAP effective tax rates and smaller total/discretionary permanent book-tax

differences following comment letter resolution. This is consistent with increased financial

statement disclosure quality of tax-related information leading to a decrease in tax avoidance.

Finally, we find that investors appear sensitive to the receipt of tax-related comment letters. Prior

to the resolution of a comment letter, tax avoidance is positively related to stock returns.

However, we document a significant attenuation of this positive relation following a comment

letter, suggesting SEC scrutiny is related to heighten concern over the loss of value-enhancing

tax planning.

We also perform a number of supplemental tests. First, we confirm using a firm fixed

effects specification, which controls for unobservable time-invariant firm characteristics and help

rule out alternative explanations. Second, we employ a difference-in-difference design using

propensity-score matching where we match each firm receiving a tax-related comment letter to a

control firm based on company and industry characteristics including prior period tax avoidance

and other variables that could impact the likelihood of receiving a tax-related comment letter.

Inferences generally remain unchanged or become stronger. Finally, we document that tax

avoidance is positively related to the length of correspondence between the firm and the SEC,

suggesting that tax avoidance is also related to the severity of the comment letter.

The impact of tax-related financial statement disclosures on tax avoidance is important

for several reasons. First, tax authorities potentially use financial statement disclosures (or lack

of disclosures) in determining which firms to target for examination and the scope of subsequent

audits. Second, although the SEC and IRS are both governmental entities, in the past minimal

 

6  

coordination and information sharing has occurred between the two agencies.5 This study

documents a material “spill-over” effect from improving financial reporting quality that favors

additional revenue collection. Thus, we inform legislators and regulators on the potential benefits

of better cross agency allocation of resources. Third, by investigating investors’ responses to

increased disclosure for firms engaging in greater levels of tax avoidance, we provide evidence

on the market’s perception of increased disclosures and the value of firms’ tax avoidance.

This study contributes to several lines of literature. First, we contribute to the literature

investigating the determinants (Cassell et al. 2013; Ettredge et al. 2011) and consequences

(Johnston and Petacchi 2012; Robinson, Xue, and Yong 2011) of SEC comment letters. We

expand the scope of the literature to include tax-related topics, which represent a unique tension

between the interests of regulators that favor full disclosure of tax information and investors that

may prefer less disclosure to facilitate an increase in after-tax cash flows through tax avoidance.

Second, we extend studies examining the effect of disclosure on tax avoidance (Robinson and

Schmidt 2013; Gupta et al. 2014; Donohoe and McGill 2011; Abernathy et al. 2012) by focusing

on firm-specific disclosures rather than newly mandated standards. This approach allows us to

investigate the variability of individual firm disclosures rather than attempting to assess the

impact based on an assumed complete or uniform compliance among firms. Third, we contribute

to the literature on the consequences of regulatory scrutiny (Hanlon et al. 2014; Hoopes et al.

2012). We document a ‘spill-over’ effect where the actions of the SEC benefit the IRS and

potentially increase tax collections.

The remainder of this paper proceeds as follows: Section II provides background on the

SEC comment letter process along with a discussion of prior research and develops the research

 5See https://www.sec.gov/info/municipal/sec-irs-mou030210.pdf for an exception where the SEC and IRS worked together regarding tax-exempt bonds and municipal securities.

 

7  

hypotheses. Section III discusses the methodology and empirical procedures, Section IV

discusses the results, Section V provides a discussion of additional analyses performed, and

Section VI concludes.

II. Related literature and hypotheses development

SEC Comment Letters and Disclosure

Under Section 408 of the Sarbanes-Oxley Act (SOX), the SEC is required to review each

registrant’s 10-K at least every three years (Cassell et al. 2013). Comment letters are issued when

a registrant’s filing is materially deficient or if the filing requires additional clarification (SOX

Section 408). Registrants are required to respond to comment letters within ten days and

responses typically include supplemental information addressing the deficiency or providing

additional clarification.6 In some cases, firms are required to restate the reviewed filing, but

typically the firm commits to providing the requested information in future disclosures.7

Prior literature has modeled the determinants of SEC comment letters. Cassell et al.

(2013) examine SEC comment letters related to 10-K disclosures during 2004-2009 and find that

lower profitability, higher complexity, weaker corporate governance, and retaining a non-Big 4

auditors are all positively associated with the likelihood of receiving a SEC comment letter.

Boone et al. (2013) find that firms subject to rule-based standards and standards that require

subjective accounting estimates are positively associated with the propensity to receive SEC

comment letters. Additionally, Johnston and Petacchi (2012) find that firms with a history of

amended filings and restatements are more likely to receive comment letters.

 6 Although the SEC comment letter process is advisory in nature, research shows the firms modify disclosures in response to the receipt of SEC comment letters (Bozanic et al. 2013; Brown et al. 2013). 7 Johnston and Petacchi (2012) report that more than 17% of firms amend filings to resolve SEC comment letters during the 2004-2006 time period.

 

8  

Given the determinants and consequences of comment letters, prior research has used

SEC comment letters as a measure of disclosure quality to examine the consequences of the SEC

review process. Ettredge et al. (2011) use SEC comment letters to identify firms failing to

disclose bad news (i.e., circumstances of auditor changes) and find that identified firms had

weaker corporate governance and relied less on external creditors. Robinson, Xue and Yong

(2011) use SEC comments on mandatory compensation disclosures to document a positive

association between defective disclosures and excess CEO compensation as well as prior media

criticism of CEO compensation, suggesting excessive compensation incentivized managers to

withhold information for private benefit. Further, Johnston and Petacchi (2012) document an

increase in earnings response coefficients (ERCs) and a reduction in return volatility and trading

volume around earnings announcements following the resolution of comment letters suggesting

that investors are cognizant of comment letters and that the additional disclosure from the

resolution of the comment letter process results in a decrease in information asymmetry.

Appendix B provides two examples of tax-related comment letters. In the first example,

General Motors Inc. (GM) receives a tax-related comment letter related to its 2010 10-K filing in

which the SEC asks for clarification regarding GM’s assumptions related to permanently

reinvested earnings (PRE). In response, GM claims that the liquidity needs of the firm can be

met by using foreign cash through intercompany loans and pledges to make this disclosure in

future filings. While the SEC deemed GM’s response to be satisfactory to comply with tax

footnote disclosures, the particular tax information disclosed in this exchange potentially

increases the probability of IRS scrutiny. The use of intercompany loans can be treated as a

deemed dividend under section 956 of the Internal Revenue Code. Thus, the disclosure necessary

to satisfy the SEC potentially exposed GM to costs in the form of higher expected tax costs.

 

9  

In a second example, Apple, Inc. (Apple) received a comment letter regarding its foreign

tax structure in which the SEC asks for additional risk factor disclosures and clarification on the

location of PRE. In response, Apple refers the SEC to existing disclosures and further discloses

that the majority of PRE are located in Ireland. Notably, Apple withheld a pledge to materially

change future disclosures to include Ireland. The SEC did not deem this private disclosure to be

acceptable and then issued a second comment letter asking for a change in disclosures which

specifically references Ireland and the risks associated with changes in tax rates. Apple requested

confidential treatment of this second comment letter, suggesting that even revealing such

mundane details as the location of PRE is proprietary information that can hamper tax avoidance

activities. The SEC ultimately denied Apple the ability to withhold information regarding Ireland

and tax risks from their disclosures, releasing this information to the public. Notably, in each of

these examples it appears that the firm reveals new information related to its tax avoidance

activities.

Disclosure Quality and Tax Avoidance

Because of the separation of ownership and control in most large U.S. corporations,

external shareholders are exposed to loss through self-interested managers diverting resources

for their own private benefit (Jensen and Meckling 1976; Healy and Palepu 2001). Disclosure of

private information is one potential, yet complicated, avenue to alleviate the agency costs

associated with the separation of ownership and control. However, the means of obtaining the

optimal level of firm-level disclosure is not straightforward.

Market forces can motivate the disclosure of information. Grossman’s (1981) analysis of

disclosure suggests an ‘unraveling’ phenomenon where investors punish managers for failure to

disclose information by discounting share prices. All but the poorest quality firm discloses

 

10  

information to avoid being characterized by shareholders as the worst. This disclosure pattern

allows investors to infer that a lack of disclosure represents bad news and effectively reveals

private information to the market. In addition to Grossman’s (1981) work, the management

forecast (Bamber and Cheon 1998; Hirst, Koonce, Venkataraman 2008) and voluntary disclosure

(Verrecchia 1990; Leuz and Verrecchia 2000) literature suggest that managers reveal private

information without being compelled through regulation.

In contrast, another means of improving disclosure is through regulation. Proponents of

disclosure regulation view disclosure as a public good in which firms do not reap sufficient

firms-specific benefits from disclosure and thus underinvest in disclosure from a social-welfare

maximizing perspective (Beaver 1998). Because of the socially suboptimal level of disclosure

resulting from this free-rider problem, regulators must mandate higher levels of disclosure rather

than rely on market forces.

Proponents of regulation also point to a number of issues complicating the unraveling

result of Grossman (1981). First, in addition to complications with respect to disclosing bad

news, proprietary costs can also hinder the disclosure of good news (Verrecchia 1983;

Verrecchia and Weber 2006). If the disclosure of good news erodes a firms’ competitive position

by allowing competitors to gain vital information that can be used to alter competitive behaviors,

managers might optimally select non-disclosure and withhold information from the market. For

example, firms in more competitive industries are more prone to request that the SEC allow them

to withhold information regarding material contracts, presumably to keep competitors from

learning sensitive price and quantity information (Thompson 2013). Second, the potential

existence of proprietary costs provides managers a credible explanation for withholding

information that exposes agency costs (Bens et al. 2011; Berger and Hann 2007; Hope and

 

11  

Thomas 2008). Rather than disclose information demanded by investors that would reveal

managerial self-dealing, managers can assert disclosure would harm investors by revealing

sensitive information to competitors and potentially avoid discounts on firm value associated

with agency costs (Berger and Hann 2007). For example, empirical evidence suggests managers

withheld the disclosure of segment information primarily when segment performance was low

and would reveal unresolved agency costs (Berger and Hann 2007).

Third, Dye (1985) explains private information might be uncertain, meaning managers

with an incentive to fully disclose information may not possess the information to disclose to the

market, further complicating the unraveling result of Grossman (1981). If this is the case, the

failure to disclose information does not imply withholding of information and investors are

therefore unsure of how to assess management actions. Investors are therefore unable to fully

punish firms that withhold information through discounting share prices to accommodate agency

costs. In summary, a lack of disclosure may represent (1) the optimal withholding of information

to protect firm value from proprietary costs, (2) agency costs of managerial self-dealing, or (3)

the lack of actual information.

Prior literature investigating the relation between disclosure quality and tax avoidance

focuses on mandated disclosures and offers mixed evidence. One line of literature is consistent

with tax avoidance leading to a decrease the level of disclosure quality. Hope et al. (2013) find

that firms engaged in higher levels of tax avoidance exploited discretion in segment reporting to

hide foreign segments, presumably to withhold information from taxing authorities. Robinson

and Schmidt (2013) find that FIN 48 disclosure quality is inversely related to tax avoidance and

that investors rewarded firms with poor disclosure quality. Dyreng et al. (2014) find that firms

that fail to comply with mandatory subsidiary disclosures in the United Kingdom (U.K.) had

 

12  

higher levels of tax avoidance and that firms subsequently increase ETRs after pressure from

social activists to disclose a complete list of subsidiaries, consistent with firms withholding tax

information to prevent consumer backlash.8

In contrast to the literature suggesting that tax avoidance incentivizes managers to

withhold information, some literature suggests that managers do not withhold information

regarding tax avoidance. Balakrishnan et al. (2012) find that managers of firms engaging in

higher levels of tax avoidance attempt to offset the obfuscation resulting from higher tax

avoidance by increasing their disclosure quantity. Lennox, Lisowsky, and Pittman (2013) find

that firms engaged in higher levels of tax avoidance are less likely to receive an Accounting and

Auditing Enforcement Release, suggesting tax avoidance does not incentivize managers to

distort information on mandated financial statements and potentially implying a lack of incentive

to withhold tax-related information from disclosure in general.

Given the results of some prior literature supporting a positive association between tax

avoidance and poor disclosure quality and the nature of the issues cited in tax-related SEC

comment letters such as uncertain tax positions, foreign tax planning, etc. we contend that higher

levels of tax avoidance result in lower disclosure quality. We therefore hypothesize:

H1: The receipt of tax-related comment letters is positively associated with prior period tax avoidance.

Increases in Disclosure and Tax Avoidance

If the disclosure of new information reveals sensitive tax information, the relative costs

and benefits of tax avoidance should change. All else equal, increased regulatory scrutiny should

 8 Both Dyreng et al. (2014) and Robinson and Schmidt (2013) focus on the presence of a specific, required disclosure. In contrast, the use of SEC comment letters allows us to examine both the presence and precision of specific disclosures. Additionally, we contribute to the literature by examining the impact of regulatory scrutiny of disclosures rather than pressure from a non-governmental entity (Dyreng et al. 2014)

 

13  

increase the expected cost of detection and thus the expected cost of tax avoidance. Evidence on

mandated disclosure regulation is consistent with this conjecture. Gupta et al. (2014) find a

reduction in multistate tax avoidance following the implementation of FIN 48. Similarly, Blouin

et al. (2010) find that firms reduced tax reserves prior to the implementation of FIN 48,

consistent with a change in the costs of financial reporting of uncertain tax positions eliciting a

firm-level response. Moreover, our previously-discussed examples identified in Appendix B

provide anecdotal evidence consistent with this intuition. Subsequent to the disclosure of

intercompany loans, GM’s GAAP ETR increased from 9% in 2010 to 27% in 2012.9 Following

their comment letter regarding the tax risks of having material PRE located in Ireland, Apple’s

GAAP ETR increased from 13.8% to 18.2% and cash ETR increased from 25.2% to 26.2%.10

However, we should not observe a change in tax avoidance behavior to the extent that

comment letters do not require the revelation of new, sensitive tax information. Indeed, in a non-

tax context, extant research finds that not all SEC comment letters are equally important and

some fail to produce stock price reactions, consistent with at least some comment letters

revealing no new information (Ryans 2014). Similarly, in a non-tax setting, Robinson, Xue, and

Yu (2011) fail to find changes in executive compensation following SEC comment letters on the

quality of compensation disclosures, suggesting that even for comment letters which might

reveal sensitive information, the newly incurred costs from disclosure may not exceed the benefit

of current corporate policies. In a tax context, previous studies examining newly mandated tax

disclosures do not consistently find evidence of disclosures producing novel information. Dunbar

et al. (2009) finds only limited evidence that FIN 48 reduced Federal tax avoidance activities,

 9 GM had negative tax expense in 2011. 10 In the Apple example a potential confounding factor is the Senate subcommittee hearings on May 21st, 2013 where Apple testified on its foreign tax planning strategies.

 

14  

consistent with a mandated disclosure of tax information not eliciting a change in taxpayer

behavior. Hasegawa, Hoopes, Ishida, and Slemrod (2013) fail to detect changes in tax avoidance

for Japanese firms following new, public disclosure guidelines for corporate tax returns. Similar

to mandated disclosures, SEC tax-related comment letters may represent mundane or boilerplate

issues that provide no new information to investors.

While there are counter-arguments, eliciting scrutiny from the SEC appears to be a

material event. First, tax-related comment letters are targeted to specific firms and therefore have

a greater likelihood of producing additional information compared to generic, one-size-fits-all

disclosure standards. Second, empirical evidence confirms that firms responded to comment

letters concerning non-tax subjects by materially altering disclosures (Bozanic et al. 2013;

Brown et al. 2013). Therefore, we hypothesize in the alternative form:

H2: The resolution of tax-related comment letters is negatively associated with future tax avoidance.

Stock Market Response to SEC Comment Letters

We examine investors’ responses to changes in disclosure quality to consider whether

investors appear sensitive to changes in the expected cost of tax avoidance due to the receipt of

tax-related comment letters. If managers are hesitant to disclose tax-related information because

of costs relating to increased detection by taxing authorities, then the scrutiny from the SEC

comment letter should decrease shareholders’ valuation of tax avoidance. Donohoe and McGill

(2011) and Abernathy et al. (2012) find negative abnormal stock returns surrounding key

Schedule M-3 and Schedule UTP dates, respectively. Similarly, Frischmann et al. (2008) find

negative price reactions to Congressional inquiries about specific firm’s FIN 48 disclosures.

These negative stock returns suggests shareholders expected a potential loss of tax benefits

 

15  

resulting from the increases in the mandated disclosure of tax information and is consistent with

an increased cost explanation.

However, Frischmann et al. (2008) also find no evidence of abnormal returns for tax

aggressive firms around key FIN 48 dates, suggesting that disclosure standards either do not

produce useful information or that regulators do not use all tax-related information contained in

disclosures. Moreover, Frischmann et al. (2008) find investors responded favorably to the initial

disclosures of FIN 48 for certain types of firms. Given prior research’s lack of consensus, we

hypothesize in the null form:

H3: The resolution of a tax-related comment letter does not change the valuation of tax avoidance.

III. Methodology

Sample

We use the intersection of Compustat Annual files and Audit Analytics databases to

construct an initial sample comprised of 16,061 firm-year observations for the period 2004-2011.

We begin our sample in 2004 because this is when the SEC began publicly disclosing comment

letters and end in 2011 because of data limitations.11 Consistent with prior research, we remove

firms in the financial and utilities industries because of their unique regulatory and institutional

structures. We also exclude firms with missing total tax expense (Compustat TXT), missing cash

taxes paid (Compustat TXPD), or negative pretax book income before special items (Compustat

PI - Compustat SPI) because these firms are in a different tax planning position compared to

firms traditionally examined in this line of research (Dyreng et al. 2008). We require each

observation to have each of our four tax avoidance proxies discussed in the next section. Table 1

 11 Prior 2005, comment letters were only available via Freedom of Information Act requests (Johnston and Pettachi 2012).

 

16  

presents the sample breakdown by fiscal year and industry. We note a slight decrease in the

annual distribution of firms in Panel A that is likely due to the omission of loss firms in our

sample during years in which the Financial Crisis occurred.

To identify firms that receive a tax-related comment letter, we begin with the Audit

Analytics Comment Letter database. Panel B reports the annual distribution of comment letters,

including tax-related comment letters, and depicts a fairly even distribution across time. From

here, we differentiate tax from non-tax-related comments by searching the text of the letter for

the words “Tax”, “FAS 109”, “FIN 48” and “ASC 740.” Our search methodology captures

variations of these keywords, including the usage of lowercases and the usage of “SFAS” in lieu

of “FAS.” Panel C reports the annual distribution of tax-related comment letters. Similar to the

general comment letter distribution in Panel B, tax-related comment letters are fairly evenly

distributed across time. Finally, Panel D reports the industry distribution (one-digit SIC) of our

tax-related comment letter sample.

Measures of tax avoidance

Following Hanlon and Heitzman (2010) and related tax research, we define tax avoidance

as the reduction in a firm’s explicit tax liability. Accordingly, we employ four commonly used

proxies to measure tax avoidance: two based on effective tax rates and two based on book-tax

differences. First, we use annual GAAP and cash ETR measures. The GAAP ETR (ETR) is

computed as the ratio of total income tax expense (TXT) to pretax income (PI) and reflects tax

avoidance activities that generate permanent book tax differences and thus impact accounting

earnings. In contrast, the cash ETR (CETR) is the ratio of cash taxes paid (TXPD) to pretax

income (PI) and represents the effects of both permanent and temporary tax planning strategies.

Following prior literature, we winsorize our effective tax rate measures to lie between zero and

 

17  

one to ensure a valid economic interpretation. Both effective tax rate measures (ETR and CETR)

are decreasing in tax avoidance.

Second, we also use two measures of book-tax differences. Permanent book-tax

differences (BTD) are the differences between book and taxable income that do not reverse over

time, scaled by lagged total assets (AT). Book income equals the difference between pretax

income (PI) and minority interest in earnings (MII). Taxable income equals the sum of federal

(TXFED) and foreign taxes (TXFO) less deferred taxes (TXDI) divided by 35%. Finally, as a

proxy for potentially more aggressive forms of tax avoidance, we measure discretionary

permanent BTDs (DTAX) following Frank et al. (2009).12 By controlling for common

determinants of permanent BTDs and using the residual, we hope to capture more aggressive tax

planning activities (McGuire, Omer, and Wang 2012; Armstrong, Blouin, and Larker 2012).

Both book-tax difference measures (BTD and DTAX) are increasing in tax avoidance.

The association between tax avoidance and disclosure (H1)

To examine our first hypothesis (H1) regarding the relationship between tax avoidance

and disclosure quality, we estimate a probit model to predict the issuance of a tax-related

comment letter:

TAXCOMMLETTi,t = 0 + TAXi,t + 2SIZEi,t + 3ROAi,t + 4ACCi,t + 5LEVi,t + 6FOREIGNi,t

+ 7RESTRUCTUREi,t + 8MERGERi,t + 9PRIORLITi,t + 10Gi,t + 11RESTATEi,t-1

+ 12TAXICWi,t + 13NONTAXICWi,t + 14TAXHAVENi,t + 15NUM_CL_INDi,t + t +i

+ i,t (1)

 12 We specifically follow Frank et al. (2009) and measure DTAX as the residual from regressing (by two-digit SIC and fiscal year) permanent book-tax differences on intangibles, unconsolidated earnings, non-controlling interest in earnings, state income tax expense, change in NOL, and lagged permanent book-tax differences. Following Frank et al. (2009), we require at least 15 observations for each industry-year group in order to estimate DTAX.

 

18  

Our model is in the spirit of Cassell et al. (2013) whom estimate the factors that influence

the issuance of a comment letter. However, we augment this model to incorporate necessary tax-

related factors that influence the issuance of a tax-related comment letter.13 The dependent

variable (TAXCOMMLETT) is an indicator variable equal to one if the firm receives a tax-related

SEC comment letter for the 10-K filing for the fiscal year ended and zero for firms that do not

received a tax-related comment letter. Our primary variable of interest, TAX, is one of the four

previously defined tax avoidance proxies. A negative coefficient on TAX for ETR and CETR as

well as a positive coefficient on BTD and DTAX would suggest that firms engaging in higher

levels of tax avoidance are more likely to receive a tax-related comment letter, consistent with

H1.

We also include a number of factors that we believe have predictive ability in the

issuance of a tax-related comment letter. We include the natural log of the lagged market value

of equity (PRCC_F*CSHO) to control for size (SIZEt-1). We include the ratio of pretax income

(PI) to lagged total assets (AT) to control for profitability (ROA). Discretionary accruals (ACC)

are performance-matched discretionary accruals calculated as in Frank et al. (2009).14 Leverage

(LEV) is the ratio of long term debt (DLTT) to assets (DLTT/AT).

We include a number of variables that represent the complexity of disclosure. FOREIGN

equals one if firms report a foreign currency translation adjustment (FCA). RESTRUCTURE

equals one if firms report non-zero restructuring costs (RCP). MERGER equals one if firms

 13 In supplemental analyses, we use this model to form matched-pairs using propensity score matching in order to estimate our difference-in-difference regressions to test H2. We set TAXCOMMLETT to missing if the firm has ever received a tax-related comment letter in a different year. This safeguards the integrity of our difference-in-difference estimation in that we are ensuring our control firms have never received a tax-related comment letter. 14 We specifically follow Frank et al. (2009) and require at least 10 observations for each industry-year group in order to estimate pretax discretionary accruals (ACC).

 

19  

engage in any merger or acquisition activity during the fiscal year (AQC). PRIORLIT equals one

if the firms’ 10-Ks indicate any litigation within the last five years.15

Prior literature finds that monitoring and internal accounting systems are related to

receiving comment letters. Accordingly, we include an indicator variable (BIG4) that equals one

when firms retain a Big 4 auditor and zero otherwise. RESTATE is an indicator equaling one if

firms experienced a restatement in prior years and zero otherwise. We include indicator variables

for whether firms reported a tax or non-tax-related internal control weakness (TAXICW and

NONTAXICW, respectively). We separate tax from non-tax internal controls because prior

literature finds differential effects for tax internal control weaknesses versus non-tax weaknesses

on tax avoidance (Bauer 2014; Lynch 2014). We also include an indicator variable that tracks

whether firms disclose material operations in a known tax haven (TAXHAVEN).16 Finally, we

control for the number of tax-related comment letters issued for a given industry-year in order to

capture the possibility that the SEC may be scrutinizing tax-related disclosures among certain

industries (NUMBER_CL_INDUSTRY).17 Industry (t ) and year fixed effects (I ) are also

included and robust standard errors are clustered at the firm level.

The impact of disclosure improvements on tax avoidance (H2)

To examine our second hypothesis (H2), regarding the effects of resolving a tax-related

comment letter, we estimate the following ordinary least squares (OLS) regression, in the spirit

of Chen et al. (2010):

TAXi,t = 0 + TAXCLFIRMi + 2POSTi,t + k Controlsk + t +i + i,t (2)

 15 To capture prior litigation (PRIORLIT), we use the Stanford University lawsuit database. 16 Data procedures used to identify tax haven operations are described in Dyreng and Lindsey (2009). We thank Professor Scott Dyreng for generously sharing this data on his faculty website. 17 All variables are defined in Appendix A.

 

20  

The dependent variable is one of our four tax avoidance measures, defined previously.

TAXCLFIRM is an indicator variable equaling one if a firm received a tax-related comment letter

at any point during our sample period and zero otherwise. POST is an indicator variable equaling

one for periods after the resolution of a tax-related comment letter, only among firms that have

ever received a tax-related comment letter. Given that we estimate Model (2) over a pooled

sample, observations that have a zero value for TAXCLFIRM must have a zero value for POST.

Thus, POST serves as an implicit interaction and represents the effect that receiving a tax-related

comment letter has on firms’ tax policies, controlling for the level of tax avoidance before

receiving the comment letter. We expect a positive (negative) and significant interaction (POST)

when using ETR and CETR (BTD and DTAX), consistent with firms decreasing their tax

avoidance following the resolution of a tax-related comment letter because of the costs of

additional disclosure.

We include a vector of controls (CONTROLS) previously identified in related tax

research as determining firm-level tax avoidance. EQINC equals the ratio of the equity method

income in earnings (ESUB) to lagged total assets (AT) to control for book-tax differences related

to income received from unconsolidated entities (Dyreng, Hanlon, and Maydew 2010). INTAN

equals intangible assets (INTAN) scaled by lagged total assets (AT) and controls for differing

treatments of intangibles for book and tax purposes. We control for depreciation tax shields

(PPE) by including the ratio of property, plant, and equipment (PPENT) to lagged total assets

(AT) (Stickney and McGee 1982; Gupta and Newberry 1997). We also control for net operating

losses by including NOL, an indicator variable equal to one if firms report a positive tax loss

carryforward (TLCF) during the year and the change in NOL scaled by lagged total assets

(NOL).

 

21  

We include the lagged market-to-book ratio (MTB), calculated as the ratio of the market

value of equity (PRCC_F*CSHO) to the book value of equity (CEQ) to control for differences in

growth opportunities. In addition, we control for leverage, LEV (DLTT/AT), and research and

development, R&D (XRD/AT), to control for the effect of debt usage and R&D on tax outcomes,

respectively. Finally, we control for free cash flow (FCF), defined as operating cash flows

(OANCF) net of capital expenditures (CAPX), because prior research finds firms with greater

levels of cash holdings engage in greater tax avoidance (Dhaliwal, Huang, Moser, and Pereira

2011). We include year and industry fixed effects as well as adjust standard errors for clustering

by firm and by year.

The stock market response to SEC comment letters (H3)

We investigate our third hypothesis (H3) relating to changes in the valuation of tax

avoidance by using the following firm-fixed effects model:

RETi,t = 0 + POSTi,t + 2TAXi,t + 3TAX*POSTi,t + 4 TAX*TAXCLFIRM,t +

5ROAi,t + 6SIZEi,t-1 + 7MTBi,t-1 + 8LEVi,t + µi + t + i,t (3)

The dependent variable (RET) is the firm’s buy and hold annual return, inclusive of

dividends. POST equals one for years after receipt of a tax-related comment letter. TAX equals

one of four measures of tax avoidance described earlier. Each specification is estimated using

firm fixed effects, allowing for a test of investors’ reaction to within firm changes in tax

avoidance before and after receipt of a tax-related comment letter and controlling for

unobservable time-invariant firm characteristics. This firm fixed effect specification provides a

test of investors’ valuation of within-firm changes in tax avoidance both before and after receipt

of a tax-related comment letter relative to firms that have never received a tax-related comment

letter.

 

22  

The coefficient on TAX captures investors’ average valuation of within-firm changes in

tax avoidance, regardless of whether the firm has a received a tax-related comment letter or not.

We control for any incremental differences in valuation that exist before the receipt of a tax-

related comment letter by including the interaction TAX*TAXCLFIRM, which captures the

average valuation effect of tax avoidance for firms that will receive a tax-related comment letter

at some point in our sample. Note that we do not include the main effect TAXCLFIRM in our

model as this is collinear with the firm fixed effects.

Our primary variable of interest is the interaction between TAX and POST. This

interaction captures investors’ valuation of changes in tax avoidance after receipt of a tax-related

comment letter. A negative coefficient is consistent with an additional cost of disclosure

explanation and suggests shareholders assign an incrementally less positive value to tax planning

following a tax-related comment letter consistent with a concern regarding the potential loss of

value-enhancing tax benefits.18 The remaining variables are previously defined. We include ROA

in the model in order to distinguish between investors’ reaction of changes in tax avoidance from

changes in profitability. We control for SIZE, MTB, and LEV as these are common controls from

prior asset pricing research (Fama and French 1992, 1993).

 18 Model (3) also allows us to test whether or not investors perceive the withholding of tax-related information as indicative of agency costs. Desai and Dharmapala (2006) assert that tax avoidance necessarily requires a level of obfuscation in order to be successful. Depending upon the nature of incentive contracting and corporate governance, self-interested managers may exploit this obfuscation for their own private benefit. To the extent Desai and Dharmapala’s (2006) claims hold in our context, we would expect to find a positive and significant interaction between TAX and POST in Model (3), consistent with shareholders expecting lower agency costs associated with tax avoidance due to greater regulatory scrutiny. However, given that Desai and Dharmapala (2006) do not assert such an effect to hold generally across all firms and recent empirical evidence questioning its validity (Blaylock 2011; Seidman and Stomberg 2012), we omit such arguments from our paper as they are out of the scope of our research question. Subsequent results presented in Table 5 are inconsistent with an agency based explanation of our findings as we find tax avoidance is, on average, positively valued and that decreases in tax avoidance following a tax-related comment letter are less positively valued.

 

23  

IV. Results

Descriptive Statistics

Table 2 presents our descriptive statistics. Consistent with tax accruals, on average,

increasing reported tax rates, the average ETR is 0.302 and is higher than the average CETR of

0.252. Permanent BTDs average 3.7% of assets, of which 3% remain unexplained and appear in

the average value of DTAX. Approximately 44% of our samples receive a SEC comment letter

regarding any topic. However, only 7.6% of our sample received a tax-related SEC comment

letter. Overall, our descriptive statistics are broadly consistent with prior literature (Chen et al.

2010; Kubick et al. 2015).

Univariate Tests

Figure 1 contains a graphical representation of our tax avoidance measures before and

after the resolution of a tax-related comment letter. The first graph pictures the change in

effective tax rates while the second illustrates the change in book-tax difference-based measures.

Consistent with H2, we find that ETR increases from approximately 29% to approximately 30%

following the resolution of a comment letter (p< 0.05). Similarly, we find significant decreases

for BTD (from 0.045 to 0.035, p < 0.05) and DTAX (0.03 to 0.015, p < 0.05). We fail to find a

significant increase for CETR, though the direction is consistent with H2. Collectively, these

patterns provide preliminary evidence consistent with firms decreasing tax avoidance following

the resolution of a tax-related comment letter and supporting H2.

Multivariate results – H1

We present the results of estimating Model 1 in Table 3. In each specification, the area

under the ROC curve is over 0.70 which suggests that our probit models have reasonable

 

24  

discriminate power (Hosmer and Lemeshow 2000).19 We find that receipt of a tax-related

comment letter is negatively related to book effective tax rates (ETR, Column (1)), and positively

related to book-tax differences (BTD and DTAX, Columns (3) and (4), respectively). These

results support our first hypothesis (H1) that tax avoidance is positively associated with receipt

of a tax-related comment letter. In contrast, we do not find a relation between the cash effective

tax rate (CETR, Column (2)) and receipt of a tax-related comment letter, suggesting that the SEC

is less concerned with disclosures related to tax issues that do not affect reported earnings.

Overall, tax avoidance appears to impair tax-related disclosure quality and thus positively

predicts tax-related comment letters, consistent with an additional cost of disclosure explanation.

Multivariate results – H2

Table 4 contains the results of estimating Model (2) for the pooled sample. The negative

and significant coefficient on TAXCLFIRM in the ETR specification (p < 0.05) indicates that

firms receiving a tax-related comment at some point in the sample period engaged in a higher

level of tax avoidance and the positive and significant coefficients on TAXCLFIRM in the BTD

and DTAX specifications (p < 0.01) support this result. Moreover, consistent with H2, we find

that firms decrease their level of tax avoidance following the resolution of a tax-related comment

letter. In particular, we find a positive and significant coefficient on POST in the ETR

specification (p < 0.05). We also find negative and significant coefficients on POST in the BTD

and DTAX specifications (p < 0.01). We fail to find significant results in the CETR specification.

Multivariate results – H3

Our third hypothesis (H3) considers whether the receipt of a tax-related comment letter

affects investors’ valuation of changes in tax avoidance. Table 6 reports the results from testing

 19 We note a smaller sample (N = 11,770) in these specifications because zero values for the dependent variable (TAXCOMMLETT) only captures firms that have never received a tax-related comment letter.

 

25  

H3 with the specification in Model (3). Consistent with tax avoidance increasing after-tax cash

flows and increasing firm value, we find stock returns are decreasing in ETR and CETR (p <

0.01) and increasing in BTD and DTAX (p < 0.01). Moreover, we find no evidence that investors

differentially value tax avoidance of firms before receiving comment letters because the

interaction between TAX and TAXCLFIRM is insignificant in all four specifications. However,

consistent with investors lowering the value of firms’ tax avoidance following tax-related

comment letters, we find negative interactions between TAX and POST in the CETR specification

and positive interactions in the BTD and DTAX specifications (p < 0.05). Accordingly, we reject

hypothesis H3 in favor of a decreased valuation of tax avoidance. We fail to find a significant

interaction between TAX and POST in the ETR specification. Overall, results suggest investors

discount the tax avoidance of firms subsequent to an increase in tax-related disclosure quality

(i.e., the resolution of a tax-related comment letter).

V. Additional analyses

Firm fixed effects

To mitigate the possibility that our results are sensitive to unobservable firm

characteristics, such as firm culture or tax technologies, we re-estimate Model (2) using a firm

fixed effects specification. We continue to find support for our hypothesis H2 under this

methodologically restrictive design. Specifically, we continue to find firms engaged in

significantly higher levels of tax avoidance before the receipt of a tax-related comment letter (p <

0.05) and that, following the disclosure of tax information through the resolution of a tax-related

comment letter, firms significantly decrease their level of tax avoidance.

 

26  

Difference-in-difference regressions using propensity score matching

In addition to Model (2), that uses a pooled sample, we also estimate a refined model

using propensity score matching. We first generate the expected probability of receiving a tax

comment letter using our probit specification in Model (1). We then match, on a one-to-one

basis, firms receiving a tax-related comment letter (treatment firms) to firms never receiving a

tax-related comment letter during the sample period with the closest propensity score (control

firms). Then, using this matched-pairs sample, we modify Model (2) to accommodate the

following research design:

TAXi,t = 0 + TAXCLFIRM,t + 2POSTi,t + 3TAXCLFIRM*POSTi,t +

k Controlsk + t +i + i,t (4)

The dependent variable of Model (4) continues to be one of our four tax avoidance

measures, defined previously. Given that this sample is generated through propensity score

matching, control firms have POST equal to one in the same year in which their matched

treatment firm receives a tax-related comment letter. An interaction between TAXCLFIRM and

POST is therefore necessary to determine the effect of tax-related comment letters and serves as

our difference-in-difference estimator. Accordingly, our variable of interest, and only

modification from Model (2), is the interaction between TAXCLFIRM and POST

(TAXCLFIRM*POST). We expect a positive (negative) and significant interaction when using

ETR and CETR (BTD and DTAX), consistent with firms decreasing their tax avoidance following

the receipt of a tax-related comment letter.20

 20 We require firms to have valid pre and post data in order to be considered for Model (1). Furthermore, firms that have received a tax-related comment letter in a different year are not considered as a potential control for the current fiscal year. These additional constraints reduce our sample size for Model (4), relative to the pooled OLS model in Model (2), and thus serves as a more powerful test of H2.

 

27  

Table 6 contains the results of estimating Model (4) over our propensity-score matched

sample. Panel A reports the probit regression of Model (1). Results confirm H1, namely that

firms with lower book effective tax rates (ETR) are more likely to receive a tax-related comment

letter. Panel B reports univariate differences in control variables between treatment and control

firms to ensure that our matching procedures are sound. Panel C contains our estimation of

Model (4). We find even stronger results in support of hypothesis H2. The coefficients on the

interaction of TAXCLFIRM and POST are positive and significant in both the ETR and CETR

specifications (p < 0.05) and negative and significant in the BTD and DTAX specifications (p <

0.05).21 These findings are consistent with firms decreasing their level of their tax avoidance

following the resolution of a tax-related comment letter and supports our notion that the

disclosure of additional tax information in the comment letter process changes the expected costs

and benefits of tax avoidance such that tax avoidance is relatively more costly.

Severity of Comment Letter Process

Prior literature acknowledges variation in the severity of comment letters and commonly

employs the number of rounds of communication between the SEC and the registrant firm as a

proxy for severity (Cassell et al. 2013). As additional analysis, we consider the extent to which

tax avoidance affects the severity of comment letters. We modify Model (1) such that the

dependent variable equals the natural log of the number of rounds in which taxes are referenced

in a comment letter thread.

We present our results on the severity of the comment letters in Table 7. We find

evidence consistent with tax avoidance being positively related to the severity of the comment

letter. In particular, we find negative and significant coefficients on ETR and CETR (p < 0.05)

 21 Results are robust to matching on any one of the four tax avoidance proxies used in the primary analyses.

 

28  

and positive and significant coefficients on BTD and DTAX (p < 0.01). Not only does tax

avoidance appear to impair the tax-related disclosure quality of firms, higher levels of tax

avoidance also appear to produce greater scrutiny by the SEC.

Uncertain Tax Benefits

We also examine changes in uncertain tax benefits (UTBs) surrounding tax-related

comment letters to help triangulate our overall findings. If managers believe that the comment

letter reveals sensitive tax information and elicits regulatory scrutiny, managers should increase

the cumulative likelihood of sustaining these positions under audit and accordingly lower the

maximum tax benefit that is more likely than not to be realized. We replace the dependent

variable of Model (2) with UTB, equaling the ratio of uncertain tax benefits (TXTUBEND) to

total assets (AT).22 Our sample decreases to 6,223 observations. In untabulated results, before

receipt of a tax-related comment letter we find an insignificant coefficient on TAXCLFIRM (p =

.564) suggests no difference in UTB balances. However, consistent with our expectations, we

find a positive coefficient on POST, suggesting firms increase their total disclosed UTBs

following the resolution of a tax-related comment letter (p < 0.05).

VI. Conclusion

We examine the relation between increases in firm-specific disclosure quality and tax

avoidance using SEC comment letters. Consistent with firms attempting to limit the disclosure of

information which would incur proprietary costs, we find the probability of receiving a tax-

related SEC comment is increasing in tax avoidance. We further document that the level of tax

avoidance decreases following the resolution of the SEC tax-related comment letter. This finding

is consistent with the release of tax-related information increasing the expected costs of tax

 22 Consistent with Lisowsky, Robinson, and Schmidt (2013), we omit firms with missing UTB balances due to Compustat’s poor coverage of UTB variables.

 

29  

avoidance and firms responding by lowering their average level of tax avoidance. Moreover, we

find that investors assign an incrementally less-positive value to tax avoidance following the

resolution on a tax-related comment letter, consistent with investors increasing the expected

costs associated with tax avoidance following regulatory scrutiny. We perform a battery of tests

to verify the robustness of our finding, including a firm fixed effect design, using propensity-

score matching to control for the underlying determinants of receiving a tax-related comment

letter, and examining the severity of the comment letters.

This study expands our knowledge regarding the costs of tax avoidance and our

understanding of why shareholders may not demand full disclosure of information in all settings.

Our findings are important to legislators as we document evidence of a spillover of benefits from

a securities regulator to the tax regulator. Moreover, our findings inform legislators that the

comment letter process mandated by the Sarbanes-Oxley Act appears to be impacting firm

behavior in a tax context. Our conclusions also inform tax regulators, suggesting that targeting

firms with poor disclosure quality in their tax footnote for audit should aide in the proper

identification of firms engaged in potentially aggressive tax avoidance behavior. Finally, our

evidence on investor responses to comment letters suggests that shareholders could prefer a

higher level of tax avoidance rather than complete compliance with current disclosure standards.

 

30  

APPENDIX A Variable Definitions

Tax Avoidance Measures (TAX)

ETR

Book effective tax rate: computed as total tax expense (Compustat TXT) divided by pretax book income less special items (Compustat PI - SPI) constrained between the [0,1] interval.

CETR Cash effective tax rate: Computed as cash taxes paid (Compustat TXPD) divided by pretax book income less special items (Compustat PI - SPI) constrained between the [0,1] interval.

BTD Permanent book-tax differences: Computed as pretax income, less minority interest in earnings (Compustat PIi,t – MIIi,t), minus the sum of federal and foreign tax expense less deferred taxes (Compustat TXFEDi,t + TXFOi,t – TXDIi,t), scaled by 35%, all scaled by lagged total assets (Compustat ATi,t-1).

DTAX Discretionary permanent book-tax differences: constructed following Frank et al. (2009).

SEC Comment Letter Variables

TAXCOMMLETT Equal to 1 if the firm has resolved an SEC comment letter which upon manual inspection discloses tax-related matters as an issue during the current fiscal year, zero otherwise.

TAXCLFIRM Equal to 1 if the firm has resolved an SEC comment letter which upon manual inspection does not disclose tax-related matters as an issue during the current fiscal year, zero otherwise.

Determinants of Tax-Related Comment Letters

SIZE Firm size: Natural log of the lagged market value of equity (Compustat PRCC_F*CSHO).

ROA Return on assets: computed as pretax book income (Compustat PIi,t) divided by lagged total assets (Compustat ATi,t-1).

ACC Performance-matched pretax discretionary accruals: computed following Frank et al. (2009).

LEV Leverage: computed as total long term debt (Compustat DLTT) divided by lagged total assets (Compustat AT). We set missing observations of DLTT equal to zero.

FOREIGN Foreign operations indicator: equals one if the firm reports a foreign currency translation adjustment (Compustat FCA), zero otherwise.

RESTRUCTURE Restructuring charges indicator: equals one if the firm reports non-zero restructuring costs (Compustat RCP), zero otherwise.

MERGER Merger activity indicator: equals one if the firm engages in any merger or acquisition activity during the fiscal year (AQC), zero otherwise.

PRIORLIT Prior litigation indicator: equals one if the firm’s 10-K is involved in any litigation within the last five years (Source: Stanford University lawsuit database), zero otherwise.

BIG4 Big 4 auditor indicator: equals one if a firm retains a Big 4 auditor, zero otherwise

RESTATE Restatement indicator: equal to one if the firm experienced a restatement in prior years, zero otherwise

 

31  

TAXICW Tax-related internal control weakness: equal to one if the firm discloses a tax-related 404(b) material weakness in internal control in the current year, zero otherwise.

NONTAXICW Non-tax internal control weakness: equal to one if the firm discloses a non-tax related 404(b) material weakness in internal control in the current year, zero otherwise.

TAXHAVEN Tax haven indicator: equal to one if the firm discloses a subsidiary in a listed tax haven (Source: Dyreng and Lindsey 2009), zero otherwise.

NUM_CL_IND Number of comment letters in industry-year: Number of tax-related comment letters issued in a firm’s industry year.

Additional Tax Avoidance Control Variables

EQINC Equity method earnings: computed as equity in earnings (Compustat ESUB) divided by lagged total assets (Compustat AT). We set missing observations of ESUB equal to zero.

INTAN Intangibles: computed as reported intangibles (Compustat INTAN) divided by lagged total assets (Compustat AT). We set missing observations of INTANi,t equal to zero.

PPE Property, plant & equipment: computed as net property, plant and equipment (Compustat PPENT) divided by lagged total assets (Compustat AT). We set missing observations of PPENT equal to zero.

NOL Presence of NOL: Indicator variable equal to one if the firm reports a positive tax loss carryforward during the year (Compustat TLCF).

ΔNOL Change in NOL: computed as the change in firm i's NOL during the year scaled by lagged total assets (Compustat AT).

MTB Market-to-book ratio: computed as the ratio of lagged market value of equity (Compustat PRCC_F*CSHO) to lagged book value of equity (Compustat CEQ).

R&D Research & development: Computed as R&D expense (Compustat XRD) scaled by lagged total assets (Compustat AT). We set missing observations of XRD equal to zero.

FCF Free cash flow: computed as operating cash flows less capital expenditures (Compustat OANCF - CAPX) scaled by lagged total assets (Compustat AT). We set missing observations of OANCF or CAPX equal to zero.

 

 

32  

APPENDIX B Examples of SEC Comment Letters and Resolutions

1. General Motors 12/31/2010 10-K Filing SEC Comment Letter and Response

SEC Comment (4/7/2011): We note from your disclosure in Note 23 that you have material assets that you consider permanently reinvested overseas. In this regard, it appears there may be amounts recorded in your financial statements, and included in your discussion of liquidity, for which there are material tax-driven restrictions on the free flow of funds from foreign subsidiaries. If so, please add a discussion of such assets. This discussion should include the potential charges that may be incurred if such amounts were repatriated.

Management Response (4/19/2011): The Company respectfully advises the Staff that $4.7 billion of the $6.9 billion of permanently reinvested amounts is not recorded as liquid assets on the balance sheet as of December 31, 2010, as it is derived from the Company’s investments in non-consolidated joint ventures which were adjusted to their fair value as a result of fresh-start reporting. As such, the cash, marketable securities and lines of credit of these joint ventures are not included in the Company’s total reported liquidity. The remaining $2.2 billion is primarily composed of earnings subject to excessive additional taxes on distributions, such as earnings generated in Mexico. Although this liquidity is permanently reinvested, the Company has taken and will continue to take steps to utilize this liquidity globally, as needed, on a less than permanent basis via inter-company loans and other cash pooling initiatives. The Company continues to believe that these amounts are appropriately included in total available liquidity. Because these earnings are effectively available for use by the Company, we do not believe disclosure of the potential charges that would be incurred if such amounts were repatriated is meaningful to our liquidity discussion. In response to the Staff’s comment, the Company proposes to revise disclosures in future filings which contain a discussion of Liquidity and Capital Resources by adding the following: “A portion of our total liquidity includes amounts deemed permanently reinvested in our foreign subsidiaries. Such amounts are available for global operations through methods such as inter-company loans.”

 

33  

2. Apple, Inc. 09/29/2012 10-K Filing SEC Comment Letters and Responses (6/24/13 & 7/22/2013

1st SEC Comment Letter (6/13/2013): 1) You state that because you are subject to taxes in the United States and numerous foreign jurisdictions, you could be subject to changes in tax rates, the adoption of new United States or international tax legislation, or exposure to additional tax liabilities. These appear to be risks that could apply to any registrant with international operations. Please tell us what consideration you have given to including a more tailored discussion of any specific risks associated with your current tax structure, including any agreements or arrangements that provide material tax benefits. See Item 503(c) of Regulation S-K. 2) We note the caption in your effective income tax rate reconciliation table “Indefinitely invested earnings of foreign subsidiaries.” Please explain how such amounts were determined in each period and identify the significant components of this line item for each period presented. Management Response (6/24/2013): 1) The Company acknowledges the Staff’s comment and notes that Item 503(c) of Regulation S-K states, among other things, that the Company is to “Explain how the risk affects the issuer . . . .” Accordingly, the Company’s existing risk factor disclosure specifically explains that its future effective tax rates are subject to “changes in the mix of earnings in countries with differing statutory tax rates, changes in the valuation of deferred tax assets and liabilities, or changes in tax laws or their interpretation.” 2) In response to the Staff’s comment, the Company notes that the tax benefits disclosed in the effective tax rate reconciliation table under the heading “Indefinitely Invested Earnings of Foreign Subsidiaries,” solely represent the difference between the U.S. statutory rate of 35% and the local tax rate on those indefinitely reinvested undistributed earnings, substantially all of which were generated by subsidiaries organized in Ireland. As disclosed in the effective tax rate reconciliation table, the Company had tax benefits from indefinitely reinvested earnings of foreign subsidiaries totaling $5,895 million, $3,898 million, and $2,125 million for 2012, 2011, and 2010, respectively. 2nd SEC Comment Letter (7/9/2013): We note your response to prior comment 1. The revised language still refers to “foreign” jurisdictions and appears to discuss risks that could apply to any registrant with international operations. Your responses state that your Irish subsidiaries generated substantially all of your $40.4 billion in undistributed international earnings, creating a tax benefit of approximately $5.9 billion in 2012. Thus, it appears that you should specifically reference the potential risks associated with any changes in Irish tax laws. Management Response (7/22/2013): In response to the Staff’s comment, the Company intends to include additional disclosure in the risk factor in future filings substantially similar to the underlined language set forth below (which also reflects modified additional language proposed by the Company in its previous response letter to the Staff, dated June 24, 2013):

 

34  

“The Company could be subject to changes in its tax rates, the adoption of new U.S. or international tax legislation or exposure to additional tax liabilities.” The Company is subject to taxes in the U.S. and numerous foreign jurisdictions, including Ireland, where a number of the Company’s subsidiaries are organized. Due to economic and political conditions, tax rates in various jurisdictions may be subject to significant change. The Company’s future effective tax rates could be affected by changes in the mix of earnings in countries with differing statutory tax rates, changes in the valuation of deferred tax assets and liabilities, or changes in tax laws or their interpretation, including in the U.S. and Ireland. The Company is also subject to the examination of its tax returns and other tax matters by the Internal Revenue Service and other tax authorities and governmental bodies. The Company regularly assesses the likelihood of an adverse outcome resulting from these examinations to determine the adequacy of its provision for taxes. There can be no assurance as to the outcome of these examinations. If the Company’s effective tax rates were to increase, particularly in the U.S. or Ireland, or if the ultimate determination of the Company’s taxes owed is for an amount in excess of amounts previously accrued, the Company’s operating results, cash flows, and financial condition could be adversely affected.”

 

35  

References Abernathy, J., Davenport, and E. Rapley. 2013. Schedule UTP: Stock Price Reaction and Economic

Consequences. The Journal of the American Taxation Association 35(1): 25–48. Balakrishnan, K., Blouin, J., and W. Guay, 2012. Does tax aggressiveness reduce financial reporting

transparency? Working Paper, University of Pennsylvania. Bamber, L. and Y. Cheon. 1998. Discretionary Management Earnings Forecast Disclosures:

Antecedents and Outcomes Associated with Forecast Venue and Forecast Specificity Choices. Journal of Accounting Research 36(2): 167-190

Bauer, A. 2014. Internal Control Quality as an Explanatory Factor of Tax Avoidance. Working Paper, University of Illinois

Beaver, W. 1998. Financial Reporting: an accounting revolution. Prentice-Hall, Englewood Cliffs, NJ.

Bens, D., P. Berger, S. Monahan. 2011. Discretionary Disclosure in Financial Reporting: An Examination Comparing Internal Firm Data to Externally Reported Segment Data. Accounting Review 86(2): 417-449

Berger, P. and R. Hann. 2007. Segment Profitability and the Proprietary and Agency Costs of Disclosure. Accounting Review 82(4):869-906

Blaylock, B. 2011. Do Managers Extra Economically Significant Rents Through Tax Aggressive Transactions? Working Paper, Oklahoma State University

Blouin, J., C. Gleason, L. Mills, and S. Sikes. 2010. Pre-Empting Disclosure? Firms’ Decisions Prior to FIN No. 48. Accounting Review 85(3): 791-815

Boone, J., C. Linthicum, and A. Poe. 2013. Characteristics of accounting standards and SEC review comments. Accounting Horizons 27(4): 711-736.

Cassell, C., L. Dreher, and L. Myers. 2013. Reviewing the SEC’s Review Process: 10-K Comment Letters and the Cost of Remediation. The Accounting Review 88(6): 1875–1908.

Ciconte, W., M. Donohoe, P. Lisowksy, and M. Mayberry. 2014. Predictable Uncertainty: The Relation between Unrecognized Tax Benefits and Future Income Tax Cash Outflows. Working Paper, University of Florida

De Simone, L., J. Robinson, and B. Stomberg. 2014. Distilling the Reserve for Uncertain Tax Positions: The Revealing Case of Black Liquor. Review of Accounting Studies 19(1): 456–72.

Deloitte. 2012. SEC Comment Letter Examples: Income Taxes. Accessible at: http://www.deloitte.com/assets/DcomUnitedStates/Local%20Assets/Documents/Tax/us_tax_SEC_Comment_Examples_Income_Taxes_092112.pdf

Desai, M. Dharmapala, D., 2006. Corporate tax avoidance and high-powered incentives. Journal of Financial Economics 79, 145-179.

Dhaliwal, D., S. Huang, W. Moser, and R. Pereira. 2011. Corporate tax avoidance and the level and valuation of firm cash holdings. Working Paper. University of Arizona, University of Arkansas, University of Missouri at Columbia.

Donohoe, M., and G. McGill. 2011. The Effects of Increased Book-Tax Difference Tax Return Disclosures on Firm Valuation and Behavior. Journal of the American Taxation Association 33(2): 35–65.

Dunbar, A., J. Phillips, and G. Plesko. 2009. The Effect of FIN 48 on Firms’ Tax-Reporting Behavior. Working Paper, University of Connecticut.

Dye, R. 1985. Disclosure of Nonproprietary Information. Journal of Accounting Research 23(1): 123-145

Dyreng, S., M. Hanlon, E. Maydew. 2008. Long-run corporate tax avoidance. The Accounting Review 83, 61-81.

 

36  

Dyreng, S., M. Hanlon, E. Maydew. 2010. The effects of executives on corporate tax avoidance. The Accounting Review 85, 1163-1189.

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., J. Hoopes, and J. Wilde. 2014. Real Costs of Subsidiary Disclosure: Evidence from Corporate Tax Behavior. Working Paper, University of Iowa

Ettredge, M., K. Johnstone, M. Stone, and Q. Wang. The effects of firm size, corporate governance quality, and bad news on disclosure compliance. Review of Accounting Studies 16 (4) 866-889.

Fama, E., French, K., 1992. The cross-section of expected stock returns. Journal of Finance 47, 427-465.

Fama, E., French, K., 1993. Common risk factors in the returns on stocks and bonds. Journal of Financial Economics 33, 3-56.

Frank, M., L. Lynch, and S. Rego. 2009. Tax reporting aggressiveness and its relation to aggressive financial reporting. The Accounting Review 64, 467-496.

Frischmann, P., T. Shevlin, R. Wilson. 2008. Economic Consequences of Increasing the Conformity in Accounting for Uncertain Tax Benefits. Journal of Accounting & Economics 46(2): 261-278

Graham, J., J. Raedy, and D. Shackelford. 2012. Research in Accounting for Income Taxes. Journal of Accounting and Economics 53 (1): 412–34.

Grossman, S. 1981. The Informational Role of Warranties and Private Disclosure about Product Quality. Journal of Law & Economics 24(3):461-483

Gupta, S., L. Mills, and E. Towery. 2014. The Effect of Mandatory Financial Statement Disclosures of Tax Uncertainty on Tax Reporting and Collections: The Case of FIN 48 and Multistate Tax Avoidance. Journal of the American Taxation Association, Forthcoming

Gupta, S., Newberry, K., 1997. Determinants of the variability in corporate effective tax rates: Evidence from longitudinal data. Journal of Accounting and Public Policy, 16: 1-34. Hanlon, M., S. Heitzman. 2010. A review of tax research. Journal of Accounting and Economics 50,

127-178. Hanlon, M., J. Hoopes, N. Shroff. 2014. The Effect of Tax Authority Monitoring and Enforcement

on Financial Reporting Quality. Journal of the American Taxation Association. Forthcoming Hasegawa, M., J. Hoopes, R. Ishida, and J. Slemrod. 2013. The Effect of Public Disclosure on

Reported Taxable Income: Evidence from Individuals and Corporations in Japan. National Tax Journal 66(3): 571-607

Healy, P., and K. Palepu, 2001. Information asymmetry, corporate disclosure and the capital markets: A review of the empirical disclosure literature. Journal of Accounting and Economics.

Hirst, D., L. Koonce, S. Venkataraman. 2008. Management Earnings Forecasts: A Review and Framework. Accounting Horizons 22(3):315-338

Hope, O., M. Ma, and W. Thomas. 2013. Tax Avoidance and Geographic Earnings Disclosure. Journal of Accounting and Economics 56 (2):170–89.

Hope, O., and W. Thomas. 2008. Managerial Empire Building and Firm Disclosure. Journal of Accounting Research 46(3): 591-626

Hoopes, J., D. Mescall, and J. Pittman. 2012. Do IRS Audits Deter Corporate Tax Avoidance? Accounting Review 87(5):1603-1639

Jensen, M. and W. Meckling. 1976. Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Strucutre. Journal of Financial Economics. 1976(3):305-360

Johnston, R. and R. Petacchi. 2012. Regulatory Oversight of Financial Reporting: Securities and Exchange Commission Comment Letters. Working Paper, Purdue University

 

37  

Kubick, T. D. Lynch., M. Mayberry, and T. Omer. 2015. Product Market Power and Tax Avoidance: Market Leaders, Mimicking Strategies, and Stock Returns. The Accounting Review. Forthcoming

Lennox, C., P. Lisowksy, J. Pittman. 2013. Tax Aggressiveness and Accounting Fraud. Journal of Accounting Research 51(4): 739-778

Leuz, C. and R. Verrecchia. 2006. The Economic Consequences of Increased Disclosure. Journal of Accounting Research 38(Supp): 91-124

Lisowsky, P., L. Robinson, A. Schmidt. 2013. What Do Publicly Disclosed Tax Reserves Tell Us About Privately Disclosed Tax Shelter Activity? Journal of Accounting Research, 51 (3): 583-629, 2013

Lynch, D. 2014. Investing in the corporate tax function: The effects of remediating material weaknesses in internal control on tax avoidance. Working Paper, University of Wisconsin-Madison

Neuman, S., T. Omer, M. Shelley. 2013. Corporate Transparency, Sustainable Tax Strategies, and Uncertain Tax Activities. Working Paper, University of Nebraska-Lincoln

PWC. 2013. Stay informed Automotive Sector: 2013 SEC comment letter trends. www.pwc.com. Robinson, J., Y. Xue, Y. Yu. 2011. Determinants of Disclosure Noncompliance and the Effect of the

SEC Review: Evidence from the 2006 Mandated Compensation Disclosure Regulations. The Accounting Review 86 (4): 1415–44.

Robinson, L., and A. Schmidt. 2013. Firm and Investor Responses to Uncertain Tax Benefit Disclosure Requirements. The Journal of the American Taxation Association 35 (2): 85–120.

Ryans, J. 2014. Textual Classification of SEC Comment Letters. Working Paper, University of California-Berkley

Seidman, J. and B. Stomberg. 2014. Equity Compensation and Tax Avoidance: Incentive Effect of Tax Benefits? Working Paper, University of Georgia.

Stickney, C., and V. McGee. 1982. Effective corporate tax rates: The effect of size, capital intensity, leverage, and other factors. Journal of Accounting and Public Policy 1 (2): 125–152.

Towery, E., 2012. How Do Disclosures of Tax Aggressiveness to Tax Authorities Affect Reporting Decisions? Evidence from Schedule UTP. Working Paper, University of Georgia.

Thompson, A. 2013. SEC Confidential Treatment Orders: Determinants and Consequences of Disclosure Exemption Decisions. Working Paper, University of Illinois

Verrecchia, R. Discretionary disclosure. Journal of Accounting & Economics 5(3): 179-194 Verrecchia, R., Information Quality and Discretionary Disclosure. Journal of Accounting &

Economics 12(4):365-380 Verrecchia, R. and J. Weber. 2006. Redacted Disclosure. Journal of Accounting Research 44(4):

791-814

 

38  

Figure 1 Mean tax avoidance before (Pre) and after (Post) receipt of a tax-related SEC comment letter.

t-test of differences in mean: ETR: p-value <0.05; CETR > 0.10.

t-test of differences in mean: BTD: p-value <0.05; DTAX: p-value <0.05

0.230

0.240

0.250

0.260

0.270

0.280

0.290

0.300

0.310

Pre Post

Effective tax rates

ETR CETR

0.000

0.005

0.010

0.015

0.020

0.025

0.030

0.035

0.040

0.045

0.050

Pre Post

Book-tax differences

BTD DTAX

 

39  

Table 1 Sample composition. Panel A: Time distribution.

Fiscal year Frequency % Cumulative Freq. Cumulative % 2004 2,290 14.26 2290 14.26 2005 2,259 14.07 4,549 28.32 2006 2,208 13.75 6,757 42.07 2007 2,066 12.86 8,823 54.93 2008 1,769 11.01 10,592 65.95 2009 1,677 10.44 12,269 76.39 2010 1,917 11.94 14,186 88.33 2011 1,875 11.67 16,061 100.00

Panel B: Time distribution of comment letters. Fiscal year Frequency % Cumulative Freq. Cumulative %

2004 763 10.8 763 10.8 2005 904 12.80 1,667 23.61 2006 891 12.62 2,558 36.22 2007 1,001 14.17 3,559 50.40 2008 954 13.51 4,513 63.91 2009 908 12.86 5,421 76.76 2010 832 11.78 6,253 88.54 2011 809 11.46 7,062 100.00

Panel C: Time distribution of tax-related comment letters.

Fiscal year Frequency % Cumulative Freq. Cumulative % 2004 139 11.34 139 11.34 2005 151 12.32 290 23.65 2006 113 9.22 403 32.87 2007 135 11.01 538 43.88 2008 156 12.72 694 56.61 2009 119 9.71 813 66.31 2010 239 19.49 1,052 85.81 2011 174 14.19 1,226 100.00

Panel D: Industry distribution.

Industry (1-digit SIC) Frequency % Cumulative

Freq Cumulative

% 0-1 (Agriculture, mining, oil, construction) 988 6.15% 988 6.15% 2 (Food, tobacco, textiles, paper, chemicals) 2,769 17.24% 3,757 23.39% 3 (Manufacturing, machinery, electronics) 5,560 34.62% 9,317 58.01% 4 (Transportation and communications) 913 5.68% 10,230 63.69% 5 (Wholesale and retail) 2,329 14.50% 12,559 78.20% 7 (Services) 2,483 15.46% 15,042 93.66% 8-9 (Health, legal, educational services) 1,019 6.34% 16,061 100.00%

 

40  

Table 2 Descriptive statistics. This table reports summary statistics for the variables used in our primary analyses. The sample is comprised of 16,061 observations spanning fiscal years 2004 through 2011. Effective tax rates (ETR, CETR) are constrained to lie on the [0,1] interval to ensure a valid economic interpretation. Utilities and financial firms are excluded from the sample. All continuous tax and control variables are winsorized at the 1% and 99% level to mitigate the influence of outliers.

Variable N Mean Std dev 10th Pctl 50th Pctl 90th Pctl ETRt 16,061 0.302 0.175 0.011 0.335 0.422 CETRt 16,061 0.252 0.221 0.006 0.230 0.488 BTDt 16,061 0.037 0.088 -0.009 0.013 0.095 DTAXt 16,061 0.030 0.126 -0.065 0.008 0.139 COMMLETTt 16,061 0.440 0.496 0.000 0.000 1.000 TAXCOMMLETTt 16,061 0.076 0.266 0.000 0.000 0.000 NOTAXCOMMLETTt 16,061 0.363 0.481 0.000 0.000 1.000 ROAt 16,061 0.127 0.108 0.024 0.100 0.259 ACCt 16,061 0.011 0.089 -0.074 0.000 0.102 SIZEt-1 16,061 6.365 2.074 3.511 6.449 9.011 FIt 16,061 0.022 0.044 0.000 0.000 0.075 EQINCt 16,061 0.001 0.004 0.000 0.000 0.002 INTANt 16,061 0.213 0.248 0.000 0.125 0.552 PPEt 16,061 0.276 0.264 0.039 0.187 0.664 NOLt 16,061 0.444 0.497 0.000 0.000 1.000 ΔNOLt 16,061 -0.002 0.084 -0.036 0.000 0.026 MTBt-1 16,061 3.240 3.403 1.017 2.281 6.022 LEVt 16,061 0.165 0.198 0.000 0.105 0.429 FCFt 16,061 0.067 0.109 -0.045 0.066 0.186 R&Dt 16,061 0.032 0.056 0.000 0.000 0.112

 

41  

Table 3 Determinants of tax-related comment letters, probit regressions. This table reports results from a probit regressions in which TAXCOMMLETT is the dependent variable. TAXCOMMLETT equals zero if the firm has never received a tax-related comment letter. Each specification in Columns (1) through (4) contain a different tax avoidance measure for TAX (ETR, CETR, BTD, DTAX). For brevity, time and industry dummies are not tabulated. Standard errors are clustered by firm. Variables of interest are bolded and italicized. Robust standard errors are employed and clustered at the firm level. All p-values are two-tailed.

(1) TAX = ETRt (2) TAX = CETRt (3) TAX = BTDt (4) TAX = DTAXt

Variable Estimate p-

value Estimate p-

value Estimate p-

value Estimate p-

value Intercept -1.393 0.001 -1.489 0.000 -1.582 0.000 -1.551 0.000 TAXt -0.478 0.000 0.051 0.546 1.178 0.000 0.439 0.002 SIZEt-1 0.138 0.000 0.137 0.000 0.144 0.000 0.142 0.000 ROAt -0.306 0.154 -0.266 0.217 -0.685 0.003 -0.330 0.125 ACCt 0.303 0.183 0.424 0.063 0.287 0.204 0.308 0.175 LEVt 0.314 0.004 0.310 0.004 0.304 0.005 0.297 0.006 FOREIGNt 0.178 0.000 0.179 0.000 0.175 0.001 0.179 0.000 RESTRUCTUREt 0.171 0.000 0.173 0.000 0.165 0.000 0.167 0.000 MERGERt -0.059 0.602 -0.059 0.597 -0.063 0.575 -0.060 0.593 PRIORLITt 0.064 0.183 0.070 0.148 0.063 0.191 0.065 0.175 BIG4t -0.214 0.000 -0.222 0.000 -0.220 0.000 -0.220 0.000 RESTATEMENTt 0.044 0.497 0.041 0.530 0.049 0.452 0.044 0.497 TAXICWt 0.661 0.000 0.651 0.000 0.637 0.000 0.653 0.000 NONTAXICWt 0.141 0.189 0.128 0.235 0.136 0.206 0.136 0.206 TAXHAVENt 0.112 0.024 0.112 0.024 0.110 0.028 0.111 0.026 NUMBER_CL_INDUSTRYt 0.038 0.000 0.037 0.000 0.037 0.000 0.037 0.000

Year dummies? Yes Yes Yes Yes Industry dummies? Yes Yes Yes Yes Pseudo R2 0.11 0.10 0.11 0.11 Area under the ROC curve 0.73 0.73 0.73 0.73 N 11,770 11,770 11,770 11,770

 

42  

Table 4 Pooled sample. Levels of tax avoidance following tax-related comment letter resolution. OLS. This table reports results from OLS regressions in which ETR, CETR, BTD, and DTAX are the dependent variables. For brevity, time and industry dummies are not tabulated. Standard errors are clustered by firm and year. Variables of interest are bolded and italicized. Robust standard errors are employed and clustered at the firm and year level. All p-values are two-tailed.

(1) ETRt (2) CETRt (3) BTDt (4) DTAXt

Variable Estimate p-

value Estimate p-

value Estimate p-

value Estimate p-

value Intercept 0.262 0.000 0.298 0.000 0.091 0.046 0.125 0.000 TAXCLFIRMt -0.014 0.021 0.005 0.348 0.010 0.000 0.007 0.008 POSTt 0.012 0.061 0.001 0.894 -0.009 0.000 -0.011 0.000 ROAt 0.060 0.043 -0.065 0.139 0.204 0.000 -0.020 0.531 ACCt -0.297 0.000 -0.505 0.000 0.134 0.000 0.361 0.000 SIZEt-1 0.003 0.004 0.001 0.632 -0.005 0.000 -0.011 0.000 FIt -0.231 0.000 0.022 0.773 0.111 0.000 0.156 0.000 EQINCt -1.117 0.051 -0.256 0.698 0.285 0.322 -1.124 0.000 INTANt 0.014 0.110 0.011 0.388 0.001 0.764 0.073 0.000 PPEt -0.022 0.060 -0.089 0.000 0.005 0.318 0.020 0.001 NOLt -0.019 0.000 -0.038 0.000 0.010 0.000 0.013 0.000 ΔNOLt 0.094 0.000 0.147 0.000 -0.092 0.000 0.112 0.002 MTBt-1 -0.002 0.017 0.000 0.759 0.002 0.000 0.003 0.000 LEVt 0.002 0.892 -0.049 0.002 0.006 0.387 -0.011 0.144 FCFt -0.138 0.000 -0.327 0.000 0.079 0.000 0.161 0.000 R&Dt -0.449 0.000 -0.452 0.000 0.274 0.000 0.433 0.000

Year dummies? Yes Yes Yes Yes Industry dummies? Yes Yes Yes Yes R2 0.10 0.12 0.22 0.16 N 16,061 16,061 16,061 16,061

 

43  

Table 5 The association between annual stock returns and tax avoidance after receipt of a tax-related SEC comment letter. OLS. This table reports results from OLS regressions where the dependent variable is the firm’s buy and hold annual stock return inclusive of dividends (RET) (Source: CRSP). All other variables are defined previously. Variables of interest are bolded and italicized. Robust standard errors are employed. All p-values are two-tailed.

(1) TAX = ETRt (2) TAX = CETRt (3) TAX = BTDt (4) TAX = DTAXt

Variable Estimate p-

value Estimate p-

value Estimate p-

value Estimate p-

value

Intercept -1.749 0.000 -1.786 0.000 -1.903 0.000 -1.924 0.000 POSTt -0.067 0.094 -0.099 0.002 -0.007 0.774 -0.031 0.168 TAXt -0.383 0.000 -0.377 0.000 1.129 0.000 0.831 0.000 TAX*POSTt 0.064 0.566 0.188 0.031 -0.916 0.000 -0.420 0.016 TAX*TAXCLFIRMt 0.028 0.758 0.063 0.365 0.133 0.449 0.152 0.214 ROAt 1.541 0.000 1.419 0.000 1.266 0.000 1.456 0.000 SIZEt-1 0.370 0.000 0.371 0.000 0.376 0.000 0.379 0.000 MTBt-1 -0.069 0.000 -0.067 0.000 -0.070 0.000 -0.070 0.000 LEVt 0.377 0.000 0.377 0.000 0.368 0.000 0.330 0.000

Year dummies? Yes Yes Yes Yes Firm fixed effects? Yes Yes Yes Yes R2 0.47 0.48 0.48 0.48 N 16,052 16,052 16,052 16,052

 

44  

Table 6 Difference-in-difference regressions, propensity score matched sample. Panel C reports results from OLS regressions in which ETR, CETR, BTD, and DTAX are the dependent variables. The sample is comprised of treatment and control firms formed using the probit model displayed in Panel A. For brevity, time and industry dummies are not tabulated. Standard errors are clustered by firm and year. Variables of interest are bolded and italicized. Robust standard errors are employed and clustered at the firm level. All p-values are two-tailed. Panel A: Probit model

(1) TAXCOMMLETTt Estimate Estimate p-value ETRt -0.788 0.000 SIZEt-1 0.154 0.000 ROAt -0.360 0.268 ACCt 0.756 0.051 LEVt 0.459 0.003 FOREIGNt 0.211 0.003 RESTRUCTUREt 0.198 0.003 MERGERt 0.006 0.970 PRIORLITt 0.108 0.136 BIG4t -0.343 0.000 RESTATEMENTt 0.080 0.384 TAXICWt 0.738 0.000 NONTAXICWt 0.414 0.010 TAXHAVENt 0.056 0.415 NUMBER_CL_INDUSTRYt 0.036 0.000

Year dummies? Yes Industry dummies? Yes Pseudo R2 0.14 Area under the ROC curve 0.75 N 5,316

Panel B: Univariate differences in means across treatment and control groups based on the probit in Panel A.

(1) Treatment group (2) Control group Estimate N Mean N Mean p-value ETRt 443 0.308 443 0.292 0.125 SIZEt-1 443 7.012 443 7.121 0.411 ROAt 443 0.134 443 0.140 0.407 ACCt 443 0.003 443 0.007 0.526 LEVt 443 0.188 443 0.167 0.120 FOREIGNt 443 0.357 443 0.330 0.396 RESTRUCTUREt 443 0.327 443 0.325 0.943 MERGERt 443 0.950 443 0.957 0.632 PRIORLITt 443 0.704 443 0.713 0.768 BIG4t 443 0.777 443 0.792 0.568 RESTATEMENTt 443 0.074 443 0.074 1.000 TAXICWt 443 0.023 443 0.020 0.817 NONTAXICWt 443 0.038 443 0.025 0.250 TAXHAVENt 443 0.467 443 0.456 0.737 NUMBER_CL_INDUSTRYt 443 11.124 443 11.014 0.882

 

45  

Panel C: Regressions using propensity-score matched sample. (1) ETRt (2) CETRt (3) BTDt (4) DTAXt

Variable Estimate p-

value Estimate p-

value Estimate p-

value Estimate p-

value Intercept 0.368 0.000 0.484 0.000 0.002 0.796 0.063 0.033 TAXCLFIRMt -0.013 0.106 -0.014 0.072 0.009 0.001 0.006 0.104 POSTt -0.016 0.012 -0.009 0.347 0.006 0.117 0.004 0.329 TAXCLFIRM*POSTt 0.022 0.040 0.025 0.012 -0.009 0.004 -0.008 0.040 ROAt 0.096 0.010 -0.043 0.459 0.141 0.000 -0.095 0.037 ACCt -0.268 0.000 -0.589 0.000 0.098 0.000 0.320 0.000 SIZEt-1 0.001 0.547 -0.005 0.047 -0.003 0.000 -0.006 0.000 FIt -0.390 0.000 -0.097 0.395 0.185 0.003 0.213 0.000 EQINCt -1.219 0.100 -0.122 0.876 0.286 0.391 -1.051 0.000 INTANt 0.011 0.426 0.005 0.798 0.002 0.806 0.057 0.003 PPEt -0.049 0.004 -0.088 0.001 0.026 0.016 0.047 0.000 NOLt -0.010 0.029 -0.024 0.001 0.007 0.015 0.005 0.101 ΔNOLt 0.155 0.001 0.214 0.000 -0.129 0.000 0.036 0.375 MTBt-1 0.000 0.682 0.002 0.070 0.001 0.292 0.000 0.995 LEVt 0.005 0.772 -0.042 0.037 0.005 0.624 -0.017 0.250 FCFt -0.142 0.011 -0.423 0.000 0.098 0.000 0.207 0.000 R&Dt -0.410 0.000 -0.485 0.000 0.227 0.000 0.334 0.000

Year dummies? Yes Yes Yes Yes Industry dummies? Yes Yes Yes Yes R2 0.11 0.15 0.21 0.12 N 5,030 5,030 5,030 5,030

 

46  

Table 7 The association between the number of rounds in which taxes were referenced in the comment letter and tax avoidance. The dependent variable in this table is the natural logarithm of the number of rounds in which taxes were referenced in the comment letter (Rounds). All other variables are defined previously. All p-values are one-tailed. Standard errors are clustered by firm and year.

(1)

Rounds (2)

Rounds (3)

Rounds (4)

Rounds

Estimate Estimate p-

value Estimate p-

value Estimate p-

value Estimate p-

value Intercept 0.693 0.008 0.653 0.006 0.604 0.007 0.608 0.008 ETRt -0.202 0.004 CETRt -0.067 0.036 BTDt 0.411 0.001 DTAXt 0.241 0.002 SIZEt-1 0.020 0.128 0.020 0.123 0.021 0.107 0.022 0.080 ROAt -0.048 0.760 -0.055 0.741 -0.186 0.305 -0.064 0.674 ACCt 0.039 0.849 0.048 0.803 0.024 0.910 0.037 0.855 LEVt -0.076 0.415 -0.087 0.361 -0.079 0.399 -0.087 0.345 FOREIGNt 0.022 0.361 0.022 0.370 0.019 0.435 0.022 0.352 RESTRUCTUREt 0.006 0.784 0.006 0.796 0.001 0.942 0.002 0.918 MERGERt 0.077 0.105 0.078 0.113 0.072 0.157 0.077 0.120 PRIORLITt -0.002 0.965 0.002 0.962 0.000 0.989 0.002 0.964 BIG4t -0.059 0.240 -0.064 0.221 -0.066 0.231 -0.066 0.208 RESTATEMENTt -0.056 0.357 -0.057 0.349 -0.054 0.383 -0.057 0.360 TAXICWt 0.005 0.878 0.007 0.861 -0.005 0.895 0.001 0.987 NONTAXICWt -0.089 0.273 -0.092 0.276 -0.089 0.271 -0.095 0.249 TAXHAVENt -0.007 0.828 -0.008 0.824 -0.006 0.865 -0.008 0.807 NUMBER_CL_INDUSTRYt -0.001 0.507 -0.002 0.417 -0.002 0.354 -0.002 0.356

Year dummies? Yes Yes Yes Yes Industry dummies? Yes Yes Yes Yes R2 0.11 0.11 0.12 0.11 N 1,226 1,226 1,226 1,226