How internal tax and legal expertise affect corporate income taxes. Lisa De Simone Stanford Graduate School of Business

Transcription

1 How internal tax and legal expertise affect corporate income taxes Lisa De Simone Stanford Graduate School of Business Bridget Stomberg University of Georgia November 2015 PRELIMINARY PLEASE DO NOT QUOTE OR CITE WITHOUT PERMISSION Acknowledgements: The authors appreciate helpful discussions with Peter Barnes, Charles Boes, and Lawrence De Simone; comments from Carlos Jimenez Angueira (discussant), workshop participants at the University of Santa Clara, and AAA reviewers; and invaluable programming advice and assistance from Jordan Nickerson. The authors also gratefully acknowledge funding from the Stanford Graduate School of Business (De Simone) and the University of Georgia (Stomberg).

2 How internal tax and legal expertise affect corporate income taxes ABSTRACT: Because corporate income taxes are governed by codified law with inherent uncertainty, we expect taxes to be an area where individuals with prior tax or legal experience influence corporate policy. We therefore examine the effect of executives with tax backgrounds, and independent directors with tax or legal backgrounds, on corporate income tax avoidance and financial reporting quality. Focusing on several measures of tax avoidance and financial reporting quality, we find that although firms with tax executives engage in higher levels of tax avoidance as measured using long-run GAAP effective tax rates (ETR), we find no difference in cash ETRs. Additionally, these firms exhibit smaller reserves for tax uncertainty that map more closely into future settlements with tax authorities. These results suggest that firms with tax executives engage in the same level of tax avoidance as other firms but exhibit different financial reporting choices. In contrast, we find limited evidence that independent directors with tax or legal backgrounds influence corporate tax avoidance or financial reporting for taxes. Our study extends our understanding of the effect of the characteristics and qualifications of internal governance on firm behavior and financial reporting quality. Keywords: Corporate governance, tax background, legal background, financial reporting quality, uncertain tax positions, tax avoidance JEL Classifications: H25, M41, M48

3 1. Introduction We examine how prior tax experience at the CEO or CFO level, and prior legal or tax experience at the director level, affects corporate tax avoidance and financial reporting for income taxes. We consider income tax avoidance to be any action a company takes to minimize its explicit tax burden. Although an extensive literature explores how firm-level characteristics such as size (Armstrong, Blouin, and Larcker 2012, Gupta and Newberry 1997, Manzon and Plesko 2002, Rego 2003, Zimmerman 1983) or multinationality (Collins, Kemsley, and Lang 1998, De Simone, Mills and Stomberg 2015, Klassen, Lang, and Wolfson 1993, Mills and Newberry 2004, Rego 2003) influence the level of corporate income tax avoidance, few studies examine how individuals inside the firm shape tax policy. Indeed, Huang and Kisgen (2013) note that much prior finance literature ignores the influence a specific manager has on decision making (p. 822). We contribute to a growing stream of research that examines how manager-specific attributes affect corporate tax policy by investigating whether corporations with CEOs, CFOs or independent directors who have prior professional experience in a tax or legal role engage in different levels of tax avoidance and exhibit different levels of quality in accounting for income taxes. Understanding the role that individuals inside the firm play in shaping their corporations tax policy is important for at least two reasons. First, prior literature documents an apparent under-sheltering puzzle whereby not all firms appear to optimize their level of tax avoidance given firm-level characteristics (Desai and Dharmapala 2006, Hanlon and Heitzman 2010, Weisbach 2002). For example, anecdotal evidence suggests many businesses, especially small and medium-sized firms, often leave tax benefits on 1

4 the table because they do not think they qualify for various incentives (Zerbe 2014). Having individuals inside the firms with an awareness of tax incentives and the knowledge to implement tax avoidance strategies can bring firms closer to an expected level of tax avoidance based on existing (and often hard-to-change) firm-level characteristics. Second, individuals inside the firm with prior tax and legal experience might also be effective in financial reporting for income taxes. Accounting for income taxes is a particularly complex area of financial accounting and a significant component of overall financial reporting quality. Income taxes continue to be a significant cause of restatements and the incidence of tax-related comment letters continues to rise (PwC 2014). Financial restatements and responding to SEC comment letters are costly for firms. Regulators, boards of directors and shareholders should therefore be interested in identifying managerial characteristics associated with more transparent and/or accurate income tax amounts and disclosures. We expect firms with CEOs or CFOs who have prior tax experience to engage in greater levels of tax avoidance, all else equal. Tax avoidance requires timely knowledge of tax planning opportunities compatible with business operations. We propose that CEOs and CFOs with tax backgrounds are more familiar with a broad range of tax avoidance strategies and can work with internal or external tax service providers to implement strategies appropriate given the nature of the firms structure and activities. Although external tax consultants have an extensive knowledge of tax planning strategies, these individuals are not as well positioned to identify matches between strategies and firm operations. Firms relying solely on outside advice may therefore miss 2

5 tax avoidance opportunities because they do not become aware of important transactions inside the firm in a timely manner. Additionally, even if a firm engages an external consultant to implement a strategy, it is the firm s legal obligation to the tax authority and the firm s name in the press associated with suspected wrongdoing. Thus, without someone inside the firm willing to take on the risk of reputational, tax authority audit or legal costs, strategies proposed by outside consultants may not be implemented. Tax executives could also positively influence tax financial reporting quality by providing more accurate assessments of tax uncertainty and thereby increasing the correlation between tax expense and future cash flows. These individuals could have a better understanding of the rules governing income tax accounting broadly and a greater awareness of hot-button issues driving tax restatements or comment letters. CEOs and CFOs with tax backgrounds could therefore potentially avoid these negative financial reporting outcomes. However, a CEO or CFO with tax knowledge could use that knowledge to obscure tax avoidance or use tax accounts to manage earnings. It is therefore unclear ex ante what effect executives with prior tax experience have on financial reporting for income taxes. We also examine the monitoring role that independent directors with tax or legal backgrounds play in determining the level of corporate tax avoidance. Given the materiality of income taxes and heightened awareness of tax uncertainty following a 2006 change in financial statement disclosure regulations, we expect the board of directors to be involved in reviewing a firm s tax avoidance activities. These individuals could influence the level of tax avoidance inside the firm by (a) identifying knowledge gaps within the tax function that could lead to inappropriate levels of investment in tax 3

6 planning, (b) being committed to allocating a greater level of resources to the tax function to facilitate proper support and documentation of positions undertaken, and/or (c) being a source of additional information about tax planning opportunities through their networking connections with other board members (Brown 2011). We also acknowledge the possibility, however, that directors on average do not influence the level of tax avoidance because they are not actively involved in the development or identification of tax avoidance strategies. Finally, we explore the effect independent directors could have on accounting for income taxes. Directors with legal backgrounds have been shown to improve accrual quality, likely due to their heightened awareness of litigation risks and costs that can stem from poor financial reporting quality (Krishnan, Wen, and Zhao 2011). In a tax context, these results suggest that firms with tax and legal directors may report more conservatively for uncertainty in the tax accounts to adequately alert shareholders to the magnitude of tax benefits potentially subject to repayment. Alternatively, directors with tax and legal backgrounds may be better prepared to substantiate and more willing to litigate their uncertain tax positions because of their greater understanding of their fiduciary duties of due care. This latter notion suggests that these firms may report less tax uncertainty to shareholders. 1 Similarly, directors with a tax background may have a better understanding of the audit process and be better able to estimate outcomes in tax accruals. We use Boardex data from to identify firms with CEOs and CFOs with tax backgrounds as well as with independent directors who have tax or legal 1 The complex reporting rules covering reserves for unrecognized tax benefits (UTBs) arising from tax uncertainty require firms to assume a 100 percent probability of tax authority audit under full disclosure of the position s details, unless the firm is willing to litigate for the benefits claimed. 4

7 backgrounds. Both univariate and multivariate analyses suggest that firms with tax executives report lower five-year forward looking GAAP effective tax rates (ETRs). However, we find no evidence of differential cash ETRs. These results suggest that firms with tax executives engage in the same level of tax avoidance as other firms but either focus on strategies that are more permanent in nature and therefore reduce reported GAAP ETRs, or record smaller tax reserves for uncertain tax avoidance. Consistent with the latter explanation, we also find weak evidence that firms with tax executives accrue smaller reserves for tax benefits, but that these smaller tax reserves better map into future tax cash flows. Finally, we document that tax executives are associated with a reduced likelihood of tax-related comment letters and restatements. With respect to independent directors, we find little evidence that tax or legal directors influence the level of tax avoidance or financial reporting for income taxes. Our results concerning financial reporting quality are in stark contrast to those in Krishnan et al. (2011), who document that firms with audit committee members with legal backgrounds exhibit higher short-term accrual quality and fewer discretionary accruals. This discrepancy is striking because we examine a specific legal contingency account (the UTB) that is susceptible to managerial discretion (Cazier, Rego, Tian and Wilson 2015; De Simone, Robinson, and Stomberg 2014), and therefore an account where we expect to see the documented positive monitoring effects of legal directors. These results suggest that tax and legal directors may not have sufficient influence over tax reporting decisions to provide the monitoring benefits documented in Krishnan et al. (2011). This study contributes to the literature in several areas. First, we extend the growing literature in accounting that examines the effect of specific individual traits on 5

8 corporate tax avoidance (Chyz, Gaertner, Kausar and Watson 2014, Olsen and Steckelberg 2015) and the broader literature in finance that links managerial characteristics to corporate decision making. We also contribute to the related literature that examines the effect of executives on corporate tax avoidance (e.g., Dyreng, Hanlon and Maydew 2010). Whereas much of this research focuses on the influence of executive incentives (Armstrong et al. 2012, Gaertner 2014, Rego and Wilson 2012, Powers, Robinson and Stomberg 2015), we exploit a new link between the qualifications executives and the level of corporate tax planning. Our results also contribute to the literature examining the effect of corporate governance on tax avoidance. We find limited evidence that independent directors influence the level of corporate tax avoidance, which could suggest they are not effective monitors of management s actions with respect to tax planning. Further, we provide evidence on the effect of internal monitors on the quality of income tax accounting, finding no evidence that tax or legal directors are associated with improved accounting for income taxes. Shareholders wishing to maintain strong internal monitoring of tax accounts may therefore have to use other mechanisms to achieve this goal. 2. Related Literature and Hypothesis Development The role of tax executives on tax avoidance A majority of research exploring the determinants of corporate tax avoidance focuses on firm-level characteristics such as size, profitability, the extent of foreign operations, and R&D. These studies document several firm attributes that significantly drive the level of corporate tax avoidance. Recent studies in finance and accounting have begun to explore the very likely possibility that individuals inside the firm influence 6

9 outcomes based on their own personal characteristics. As noted by Graham, Harvey and Puri (2013), although traditional economic theory suggests companies should pursue positive NPV projects to maximize wealth even within the U.S., firms in the same industry, of similar size and facing similar investment opportunities behave differently (p. 103). A growing line of research examines how executives influence corporate tax avoidance. Much of this recent research was spurred by Dyreng et al. (2010), which finds evidence of a manager fixed effect on corporate tax avoidance. These results are incremental to controlling for firm characteristics and suggest that given the same opportunities for tax avoidance, different executives make different decision. However, Dyreng et al. (2010) do not identify any single managerial characteristic that explains these individuals effects on tax avoidance. Since that time, studies have linked corporate tax avoidance with executive incentives (Armstrong et al. 2012, Gaertner 2014, Powers et al. 2015, Rego and Wilson 2012), CEO narcissism (Olsen and Steckelberg 2015) and CEO overconfidence (Chyz et al. 2014). We extend this literature by testing whether prior job experience in a tax role is associated with CEO or CFO influence over corporate tax avoidance. The link between an executive s prior tax knowledge and corporate tax planning is straightforward. Implementing a tax avoidance strategy requires not only knowledge that a particular strategy exists, but also technical know-how and/or ability to implement the strategy. Alstadsaeter and Jacob (2015) find that incentives, access and awareness are all important factors in explaining cross-sectional differences in the extent to which individuals avail themselves of legal and observable tax avoidance opportunities (p. 1). 7

10 This finding is consistent with Bonner, Davis and Jackson (1992), who find that integrated tax and transactional knowledge may be important to high-quality tax planning (p. 25). CEOs and CFOs with tax backgrounds therefore likely have a greater awareness of tax avoidance strategies as well as the ability to work with internal or external tax service providers to implement strategies appropriate for the firm given the nature of the firm s structure and business operations. The prevalence of outside tax consultants could, however, mitigate the incremental benefit of executives with prior tax experience. For example, McGuire, Omer and Wang (2012) provide evidence that firms purchasing tax services from their industryexpert audit firms engage in higher levels of tax avoidance, all else equal. These results could suggest that external consultants provide sufficient tax technical knowledge to make internal tax awareness irrelevant. We do not believe this is the case for two reasons. First, implementing a tax avoidance strategy to maximize cash flow benefits requires a timely awareness of both the strategy and the opportunity to implement the strategy. For example, an external consultant cannot provide tax planning advice about a transaction of which he or she is unaware. Therefore, we believe that having internal tax awareness increases the likelihood that tax planning opportunities are identified and either implemented with internal resources or with the help of outside consultants. Second, managers deciding whether to implement tax avoidance strategies must be willing to tolerate potential IRS audits, legal actions and media scrutiny. Graham, Hanlon and Shevlin (2014) provide survey evidence that 60 to 70 percent of executives have not implemented a proposed strategy due to concerns that the IRS would detect and challenge the position. Therefore, even if an external consultant proposes a strategy, 8

11 executives within the firm must be comfortable bearing any potential uncertainty associated with the strategy before agreeing to the proposal. We believe greater familiarity with tax authority audit procedures, audit rates, detection rates, and the technical knowledge that come with prior tax experience translate into a higher tolerance for tax uncertainty. We therefore predict in our first hypothesis that firms with executives that have a tax background engage in more tax avoidance, all else equal. H1: Executives with tax backgrounds are associated with increased corporate income tax avoidance. The role of tax executives on accounting for income taxes Income tax law contains sufficient ambiguity to create uncertainty about what amount of a tax benefit claimed in given year will be ultimately sustained. Because income tax returns are largely confidential, shareholders rely almost exclusively on financial statement disclosures to assess tax uncertainty. Given growing attention to tax uncertainty and limited disclosures of tax uncertainty, the Financial Accounting Standards Board enacted Accounting Standards Codification (ASC) Topic , also known as Financial Interpretation No. 48 (FIN 48), in 2006 to improve the transparency and comparability of accounting for income taxes. 2 Effective in 2007, FIN 48 allows managers to recognize only those tax benefits that are more likely than not to be sustained upon an audit by the tax authority. Further, FIN 48 requires the firm to recognize only the largest amount of tax benefit that is greater than 50 percent likely of being realized upon ultimate settlement, thereby requiring firms to establish a reserve for the remaining portion of the benefit claimed on the tax return that does not meet the 2 Tax uncertainty arises due to the possibility of losing a tax benefit whether due to tax audit, a change in facts and circumstances or business operations, or a change in regulations that reduce the support for or availability of a tax position. 9

12 50 percent sustainability hurdle. Firms are required to disclose a reconciliation of the beginning and ending balance of the reserve each year. Proper financial reporting for tax uncertainty under FIN 48 requires an understanding of the strength of the facts and circumstances surrounding a tax position. Tax executives could positively influence tax financial reporting quality by providing more accurate assessments of tax uncertainty and increasing the correlation between the reserve for uncertain tax benefits and future cash flows. Magro and Nutter (2012) find that experiment participants with more tax experience (tax managers) have different assessments of the relevance and strength of tax authorities than less experienced participants (tax graduate students), suggesting accounting judgments by decision makers could be significantly impacted by the individual s level of tax experience. In our setting, this finding suggests that managers with greater tax experience are better able to accurately assess the strength of income tax positions, leading to a more accurate mapping between future cash flows and the reserve for tax uncertainty. Aside from the reserve for tax uncertainty, accounting for income taxes is a particularly complex area of accounting. Tax expense related issues account for over 10 percent of all financial restatements from 2007 to 2013 (PwC 2014). Additionally, we identify over 2,000 SEC comment letters related to taxes from Audit Analytics over the same time period. Executives with tax backgrounds could have greater technical knowledge of the rules generally governing accounting for income taxes and be more aware of areas of SEC concern. These individuals might therefore improve the firm s overall tax reporting quality by providing more accurate and transparent disclosures. 10

13 However, some studies suggest that executives use tax avoidance for rent extraction and therefore have incentives to obfuscate financial disclosures (e.g., Desai and Dharmapala 2006). In this case, tax executives could impair tax reporting quality to mask suboptimal levels of tax avoidance. Prior work also suggests that income taxes are subject to earnings management, particularly through the tax reserve (Dhaliwal, Gleason and Mills 2004). Even though FIN 48 attempted to curb this opportunism, the assessment of a more-likely-than-not probability under FIN 48 is subject to managerial discretion (De Simone et al. 2014), and it remains unclear how successfully FIN 48 limited earnings management through the tax reserve (Cazier et al. 2015, Gupta, Laux and Lynch 2015). Executives with greater knowledge of the tax accounts could use that knowledge to respond to financial reporting incentives, like managing earnings to meet or beat analyst forecasts, or to otherwise obfuscate the uncertainty associated with the firm s tax avoidance. Because it is unclear ex ante what effect executives with prior tax experience have on financial reporting for income taxes, we state our second hypothesis in the null form below. H2: Executives with tax backgrounds are not associated with financial reporting for taxes. The role of independent directors on tax avoidance A growing area of research examines the effect of the board of directors as internal monitors of the firm (e.g., Agrawal and Chadha 2005; Efendi, Srivastava, and Swanson 2007; Dechow, Sloan, and Sweeney 1996; Beasley 1996; Farber 2005). This stream of research examines a wide variety of characteristics, including independence (Carcello, Neal, Palmrose, and Scholz 2011; Klein 2002), financial expertise and status (Badolato, Donalson, and Ege 2014), legal expertise (Krishnan et al. 2011) and incentive 11

14 compensation (Larcker, Richardson, and Tuna 2007). Krishnan et al. (2011) propose that because legal experts on the audit committee have a greater understanding of litigation risks and associated costs, these board members play a greater monitoring role. We believe independent directors are also likely to provide a monitoring role over taxes. Practitioners encourage active involvement of the board in tax-related issues, advising that the responsibility of corporate tax policy is shared by the board of directors (Deloitte 2014). IRS Director Doug Shulman similarly encouraged boards of directors to communicate regularly with management about taxes (Deloitte 2011). Together with management, boards of directors must decide whether that policy entails (a) lower levels of tax avoidance that involve only those positions unlikely to be challenged by tax authorities, (b) higher levels of tax avoidance that exploit all possible tax savings opportunities or (c) something in between those two extremes (Deloitte 2014). Directors with tax and legal background could influence corporate tax avoidance in the following ways. First, they can identify knowledge gaps within the tax function that could lead to forgone tax planning opportunities. Second, they are likely to be more aware of the resource requirements of the tax function and be committed to ensuring an appropriate level of tax investment. Third, directors with tax and legal backgrounds may also have a greater awareness of the types of tax avoidance strategies available to a firm and of firm-specific opportunities to exploit those strategies. We acknowledge the possibility that directors on average do not influence the underlying level of tax avoidance at the firm. These directors may not be involved with day-to-day tax planning decisions and given the infrequency of board meetings might not have sufficient interaction with the tax function to influence tax avoidance. Alternatively, 12

15 these directors could assert their influence by ensuring appropriate managers are hired to oversee the tax function, leaving little room for an incremental effect of their presence on the board. Nonetheless, we expect that independent directors with a tax or legal background positively influence a firm s level of tax avoidance, as stated our third hypothesis below. H3: Independent officers with tax or legal backgrounds are associated with more corporate income tax avoidance. The role of independent directors on accounting for income taxes Our final hypothesis addresses how directors with tax or legal backgrounds affect financial reporting for income taxes. Because tax accruals and related disclosures are a large component of financial reporting quality, the tax accounts are a plausible setting for directors to influence financial reporting. Several studies document an effect of directors on financial reporting quality, many concluding that board members with legal or financial expertise are associated with improved financial reporting quality (Carcello et al. 2011, Klein 2002, Badolato et al. 2014, Larcker et al. 2007). For example, Agrawal and Chadha (2005) find that the probability of restatement is lower in companies whose boards have an independent director with financial expertise. Krishnan et al. (2011) report that firms with legal experts on their audit committees exhibit improved accrual quality. If these findings apply to accounting for income taxes as well, we would expect directors with tax and legal backgrounds to be associated with improved financial reporting for income taxes. However, there are reasons to believe that tax and legal directors may not exercise similar influence over the tax accounts. Krishnan et al. (2011) allude to the fact that legal directors influence accrual quality because they are more aware of litigation risks and 13

16 feel the urgency to correct any wrongdoings (p. 2100) before legal problems arise. Because the rules governing accounting for income taxes are complex, however, an acute awareness of litigation risk may be insufficient to improve tax accruals and disclosures. For example, a Deloitte analysis of tax restatements in 2009 revealed that, in many cases, the restatement arose from poor integration between the tax department and other areas of the company and not any intentional management wrongdoing. If directors with legal expertise are focused on litigation rather than specific accounting issues or on the integration of tax and other departments, they may not benefit tax accounting. Additionally, legal directors may not focus on routine tax issues but instead become involved in taxes only when the company is involved in disputes with tax authorities. Directors with tax knowledge potentially have a deeper understanding of accounting for income taxes and feel a greater responsibility to be involved in routine tax matters. However, as noted in the Deloitte analysis referenced above, another leading cause of tax restatements is carelessness often stemming from top-side adjustments or last minute entries that can be endemic to accounting for income taxes. Because it is not the responsibility of directors to audit financial statements, even material errors in complicated tax calculations may go unnoticed. Regarding accounting for income tax uncertainty, specifically, it is also unclear what affect tax and legal directors will have. On one hand, if directors with tax and legal experience have a heightened awareness of shareholder litigation risks and costs, they could influence managers to report more conservatively for tax uncertainty, suggesting a positive association between directors with tax and legal experience and the magnitude of firms tax reserve accruals. On the other hand, if directors with tax and legal experience 14

17 are more sensitive to the potential costs of signaling uncertain tax avoidance strategies to tax authorities and/or are better prepared to substantiate uncertain tax positions claimed, these directors may be confident that the level of uncertainty associated with each position is lower and therefore that smaller reserves are required. Due to competing predictions, we present our final hypothesis in the null form below. H4: Independent directors with tax or legal backgrounds are not associated with financial reporting for taxes. 3. Research Design Data and Sample Boardex Data We use the Boardex 2013 database containing biographical data on executives and directors to identify prior tax and legal experience. We develop a Python script to search the prior work experience, education, and certifications of directors and officers for text indicating prior tax or legal experience. We then manually review a random subset of results to verify that these searches correctly identify executives and directors with prior tax or legal experience. We create indicator variables to summarize by year whether a firm has a CEO or CFO with a tax background, or an independent director with a tax or legal background. We focus on independent directors as they likely have a greater monitoring role (Boone, Field, Karpoff and Raheja 2007; Cai, Lui and Qian 2009; Coles, Daniel and Naveen 2007; Ferreira, Ferreira, and Raposo 2011; Klein 2002; Petra 2004). 15

18 Sample We match Boardex firms to the Compustat database using tickers and CIK codes. In untabulated analysis, we find that firms covered by Boardex are larger (both in terms of assets and market capitalization) than other firms in Compustat. We also compare the industry distributions of Boardex firms with the rest of Compustat and find they are largely similar. Using Compustat annual financial statement data from 1999 to 2013, we limit the sample to firms with data available to calculate variables required for analysis. The nature of some variables of interest requires firms to have five years of non-missing data. We also exclude real estate investment trusts and other flow-through entities. The final sample yields 31,084 firm-year observations from 4,193 firms. 38 firms (114 firmyear observations) have an executive with a tax background during at least one year in the sample. 247 firms (1,447 firm-year observations) have an independent director with a tax background during at least one year in the sample. 3,640 firms (16,735 firm-year observations) have an independent director with a legal background during at least one year in the sample. < Table 1 here > Panel A of Table 1 presents the number of sample firms by year. We further break out the sample into firm-years with a CEO or CFO with a tax background (TaxExec =1), with an independent director with a tax background (TaxIndDir =1), and with an independent director with a legal background (LegalIndDir =1). The full sample is split fairly evenly across all years. The instance of CEOs or CFOs with a tax background is rare, with at most 11 firms having an executive with a tax background in any given year. The frequency of tax background at the independent director level is also small, peaking 16

19 at 130 observations in 2005 and shrinking each year since. In contrast, the instance of LegalIndDir has increased dramatically over time from less than 40 percent in 1999 to almost 60 percent every year since the financial crisis. Panel B provides the industry classification of sample firms. In our full sample, roughly 36 percent of observations come from SIC code 3, Rubber, Metal and Machines, 18 percent from Food, Textile and Chemicals (SIC code 2) and 15 percent from Health and Other Services (SIC code 7). The majority of observations with tax executives come from firms in SIC 3. The frequency of tax and legal directors is also very high in SIC3 as well as SIC 2. We find no tax executives or tax independent directors in finance, insurance and real estate firms. Table 2 provides descriptive statistics for the full sample of firms. On average (at the median), sample firms have $3B ($0.4B) in assets. Mean (median) pre-tax return on assets is 2.4 (6.9) percent. Mean (median) foreign sales as a share of total worldwide sales are 24.0 (13.3) percent. R&D expenditures are 5.1 (0.4) percent of lagged total assets on average (at the median). We calculate one- and five-year cash and GAAP ETRs. The mean (median) cash ETR for the full sample (CETR) is 23.1 (22.7) percent, the mean (median) GAAP ETR (ETR) is 27.2 (32.0) percent, and the average long run, five year cash (GAAP) ETR is 27.1 (29.1) percent. These figures suggest that most firms in our sample engage in significant tax avoidance activities. The mean (median) UTB balance as a percent of total assets is 1.2 (0.6) percent, while the mean (median) current year addition to the UTB scaled by assets is 0.1 (0.0) percent. < Table 2 here > 17

20 Multivariate Tests To evaluate the effect of individuals on a firm s chosen level of tax avoidance (H1 and H3), we model one- and five-year measures of the cash and GAAP ETR as a function of executive or director characteristics and controls as in Equation (1) below. TaxAvoidance = β 0 + β 1 *Individual + β 2 *PT_ROA + β 3 *Size + β 4 *PctForeign + β 5 *R&D + β 6 *Leverage + β 7 *DiscAcc + β 8 *SGA + β 9 *MTB + β 10 *Growth + IndustryFE + YearFE + ε (1) For tests of H1, Individual is an indicator variable for the presence of an executive with a tax background (TaxExec) in that firm-year. For tests of H3, Individual is one of two indicator variables for the presence of either an independent director with a tax background (TaxIndDir) in that firm-year, or an independent director with a legal background (LegalIndDir) in that firm-year. Our measures of tax avoidance are the level of a firm s effective tax rate over a single- or multi-year period. Cash ETR (CETR) is total taxes paid divided by pre-tax income (TXPD/PI). GAAP ETR (ETR) is total tax expense divided by pre-tax income (TXT/PI). We also calculate five-year forward-looking GAAP and cash ETR (ETR5 and CETR5) from year t through year t+4 using the methodology from Dyreng, Hanlon and Maydew (2008). Consistent with much prior literature, we truncate values outside of [0, 1] to mitigate the influence of outliers. We use forward-looking measures to better detect the effect of a particular individual in year t on future contemporaneous and future tax avoidance. Using lagged measures would potentially confound inferences by attributing firm-level characteristics in prior years to the individuals we examine in year t. These measures provide evidence of the level of a firm s tax avoidance activities, with lower ETRs suggesting more avoidance. 18

21 We obtain controls from Rego and Wilson (2012). We include pre-tax return on assets (PI/LagAT) and the natural log of total assets (AT) because effective tax rates are correlated with firm performance and size (Armstrong et al. 2012, Gupta and Newberry 1997, Manzon and Plesko 2002, Rego 2003, Zimmerman 1983). We control for the foreign sales scaled by worldwide sales because foreign operations (and foreign tax rates) can affect worldwide consolidated ETRs. For R&D, we scale research and development expenditures by lagged total assets and set missing values of R&D to zero. Leverage is total debt scaled by lagged total assets and is included to control for debt tax shields that reduce the marginal tax benefits of aggressive tax avoidance. DiscAcc is performanceadjusted discretionary accruals using the Modified Jones model. Both Lisowsky (2010) and Wilson (2009) find a positive association between discretionary accruals and tax shelter incidence. SGA is selling, general and administrative expenses, scaled by lagged assets. MTB is the market to book ratio (PRCC_F*CSHO/AT). Growth is one-year percentage sales growth. We estimate Equation (1) using OLS with two-digit SIC industry and year fixed effects and cluster standard errors by firm. To evaluate the effect of individuals on a firm s financial reporting for taxes (H2 and H4), we estimate Equation (2) below. TaxReporting = β 0 + β 1 *Individual + β 2 *PT_ROA + β 3 *Size + β 4 *PctForeign + β 5 *R&D + β 6 *Leverage + β 7 *DiscAcc + β 8 *SGA + β 9 *MTB + β 10 *Growth + IndustryFE + YearFE + ε (1) As before, Individual is TaxExec for tests of H2 and TaxIndDir or LegalIndDir for tests of H4. TaxReporting takes one of five values. The cumulative reserve for uncertain tax benefits (UTB_Bal) is the UTB ending balance scaled by assets (UTB/AT). This variable captures reserves related to all uncertain positions that remained unresolved as of the end 19

22 of year t. As with our ETR measures, we also consider forward-looking financial statement measures to more accurately capture the effects of Individual in year t. The current year addition to the UTB (UTB_Add) is the increase in the reserve for uncertain tax benefits related to positions claimed in year t. Settle%3 is the percent of the UTB that is settled with tax authorities in t through t+2. 3 Following Robinson, Stomberg, and Towery (2015), we include additions to the reserve made in the current year but associated with prior year positions in the denominator of these settlement measures. This mitigates differences in timing as to when reserves are established. This measure reflects the accuracy of the UTB, with higher percentages corresponding to more accurate accruals relative to amounts actually paid to tax authorities. 4 Finally, we define indicator variables equal to 1 if firm received a tax-related comment letter in that year from the SEC (Letter) and, separately, if the firm had a taxrelated financial restatement in that firm-year (Restate). Information related to both of these financial reporting variables is available from the Audit Analytics website. We estimate Equation (2) using OLS with two-digit SIC industry and year fixed effects and cluster standard errors by firm. Following Hanlon and Hoopes (2015), we use a linear probability model to estimate these likelihoods to allow for easier interpretation of the coefficients as well as the use of fixed effects in the model. As noted by Ai and Norton (2003), fixed effects can create statistical issues for non-linear limited dependent variable 3 The statute of limitations for the IRS to assess a deficiency is three years. Statutes vary in other jurisdictions. We provide average settlements over a three-year period but acknowledge that settlements can take longer to finalize. However, requiring four or five-year ahead settlement data severely restricts our sample and reduces power. 4 The FIN 48 rollforward shows how reserves change from year to year. Firms are required to report decreases in reserves due to settlements with tax authorities. It is not clear whether this settlements line of the rollforward represents cash payments to tax authorities or the entire amount of the reserve accrued for the settled position. However, practitioner guidance suggests that cash payments should be reported as settlements and any differences between accrued reserves and cash payments should be reported as increases or decreases in the reserve associated with positions taken in prior years (Deloitte 2008). 20

23 models. Greene (2004) notes that fixed effects in non-linear limited dependent variable models can produce biased and inconsistent coefficients when group sizes are small, which is the case here, but not when using a linear probability model. Other control variables are defined above. 4. Results Tests of H1 and H2 The effect of tax executives To address H1 and H2, we examine differences in the level of tax avoidance and financial reporting for taxes between firms with a CEO or CFO with a tax background (TaxExec = 1) and other firms (TaxExec = 0). Table 3, Panel A reports these univariate results. Results relating to tax avoidance are generally weak and inconsistent. Using twotailed confidence intervals, we find significant differences only related to Settle%3 and Restate with tax executive firms reporting higher three-year settlement percentages and lower probabilities of tax-related restatements. Using one-tailed confidence intervals, we find mean and median one-year cash ETRs of tax executive firm-years are higher but median five-year GAAP ETRs are lower. We report significant differences across many firm characteristics, however, which highlight the importance of a multivariate analysis. [Insert Table 3 here.] We report multivariate results of H1 in Table 3 Panel B. Columns (1) and (2) use the one- and five-year cash ETR, respectively, as the dependent variable and Columns (3) and (4) use the one- and five-year GAAP ETR, respectively. Across all four measures, we only find an association between the presence of a CEO or CFO with a tax background and the firm s level of tax avoidance that is significant at conventional levels when we use the long-run GAAP ETR. We estimate a coefficient on TaxExec of

24 (p-value = ), suggesting that firm-years with a tax executive report average ETRs on their income statements over the following five years that are approximately 4.2 percent lower than those of other firms. Although insignificant at conventional levels, the signs on the estimated coefficients on TaxExec are negative when using the long-run cash ETR and the one-year GAAP ETR. Further, the negative relation between TaxExec and the one-year GAAP ETR is significant one-tailed (p-value = ). These results suggest that firms with tax executives engage in the same level of tax avoidance as other firms but either focus on strategies that are more permanent in nature and therefore reduce reported GAAP ETRs, or record smaller tax reserves for uncertain tax avoidance. This pattern of result is similar to Powers et al. (2015) who find that CEOs compensated on after-tax earnings measures report lower GAAP ETRs but similar cash ETRs as other firms. It also corroborates survey evidence in Graham et al. (2014) that some executives tend to focus on those tax planning strategies that provide a financial statement benefit. < Table 3 here > We next estimate the effect of tax CEO and CFOs on reported UTB ending balances, current-year additions, and settlement percentages, controlling for other determinants of tax reporting. We also test the likelihood of receiving an SEC comment letter or having a tax restatement in this multivariate setting. We report these results in Table 3 Panel C. Column (1) uses UTB_Bal as the measure of financial reporting for taxes, Column (2) uses UTB_Add, and Column (3) uses Settle%3, Column (4) uses Letter and Column (5) uses Restate. Consistent with univariate results, we estimate a positive and significant association between the presence of an executive with a tax background and the three-year forward-looking UTB settlement percentage. We estimate that firm- 22

25 years with tax executives settle 17.5 percent more of their UTB balance over the threeyear period relative to other firms. Because settlement percentages of 100 percent suggest more accurate accruals for tax uncertainty, together with Columns (1) and (2) we take these results as consistent with tax executive firms engaging in comparable levels of uncertain tax avoidance as other firms but generating higher quality accruals for uncertain tax positions. We also find in Column (5) that tax executive firm-years are less likely to have tax restatements, suggesting higher financial reporting quality over taxes for these firms. Similarly, the negative coefficient on TaxExec in Column (4) is weakly significant (two-tailed p-value = ) suggesting a beneficial effect on the instances of SEC comment letters. Tests of H3 and H4 The effect of tax and legal independent directors Panels A and B of Table 4 provides univariate tests of H3 and H4. We examine differences in ETRs between firm-years with an independent director that has a tax background (TaxIndDir = 1) and other firm-years (TaxIndDir = 0) in Panel A. We find that firm-years with an independent director with a tax background report average oneyear cash (GAAP) ETRs that are 1.7 (3.2) percent higher than other firms, and average five-year cash (GAAP) ETRs that are 1.2 (3.6) percent higher than other firms. All median ETR measures are also higher for firm-years with independent directors with a tax background. These results are statistically significant at conventional levels. Firms with independent tax directors also report higher three-year settlement ratios but are more likely to receive tax-related comments letters. We also test differences in tax avoidance measures between firms with an independent director with a legal background (LegalIndDir=1) and other firms 23

26 (LegalIndDir=0) in Panel B. Similarly, we find that all four ETR measures of tax avoidance are higher in firm-years characterized by having an independent director with a legal background. Firms with legal directors report higher reserves for uncertain tax positions at the median, larger values of Settle%3 and are also more likely to receive SEC comment letters. As with tax executives, however, we also report significant differences in most firm-level characteristics across these groups, revealing the need to perform multivariate analyses before drawing conclusions about the effect of these individuals on firms taxes. Panel C of Table 4 provides results of multivariate tests of H3. Columns (1) and (2) report results using the one-year cash ETR, Columns (3) and (4) use the five-year cash ETR, Columns (5) and (6) use the one-year GAAP ETR, and columns (7) and (8) use the five-year GAAP ETR. For each pair of results, we separately test the effect of an independent director with a tax background and the effect of an independent director with a legal background. We find that once we control for other determinants, different independent director backgrounds have a limited impact on the level of tax avoidance. We only estimate a statistically significant effect of the presence of an independent director with a tax background when using the long-run GAAP ETR measure as the dependent variable. In this case, we estimate that in the following five years, firms with tax directors report GAAP ETRs that are 1.8 percentage points higher than those of other firms. Similarly, we only estimate a statistically significant effect of legal backgrounds on the one-year cash ETR, finding that firm-years with a legal independent director report cash ETRs that are less than 1 percentage point higher than other firms. < Table 4 here > 24

27 We report multivariate results of H4 in Panel D of Table 4. Inconsistent with predictions and prior work, we find no effect of independent director backgrounds on financial reporting for tax uncertainty. 5. Additional Tests Matched Sample Design Because firms with executives and independent directors who have tax or legal backgrounds differ in key ways from other firms, we also re-estimate equations (1) and (2) using a matched sample design. Specifically, we match each treatment firm by year to every control firm in the same size (asset) decile and industry using the 30 industry classification from Fama and French to produce a one-to-many match. Results are generally consistent with our main tests. Re-examining H1, we find that tax executives are associated with five-year GAAP ETRs that are 3.1 percentage points lower than other firms. These results are consistent with those reported in Table 3 Panel B, although statistical significance falls just above the 10 percent threshold (p-value =.1004). In re-examining H2, we find that tax executives are associated with a lower probability of financial restatement (estimated coefficient = , p-value =0.0881). Consistent with our results for H3, we find that tax independent directors are associated with five-year GAAP ETRs that are 1.9 percentage points higher (p-value < 0.01) than other firms. This magnitude is marginally higher than that reported in Panel C of Table 4. Consistent with results for H4, we find no relation between independent director characteristics of interest and financial reporting for taxes. Note we do not conduct a matched analysis for LegalIndDir because the 25

28 number of observations where LegalIndDir=1 is greater than 50 percent of our sample and we therefore do not obtain sufficient matches. Propensity Score Matching Because differences in firm-level characteristics between our treatment group and our control group remained after using the matched sample approach described above, we use propensity scores to match each firm-year observation in our treatment sample to one firm-year observation in the control sample. Specifically, we estimate the following logistic regression to obtain propensity scores: Pr(Individual=1) = F(+ β 2 *PT_ROA + β 3 *Size + β 4 *PctForeign + β 5 *R&D + β 6 *Leverage + β 7 *DiscAcc + β 8 *SGA + β 9 *MTB + β 10 *Growth + β 11 *IndustryComplexity+ β12*geographiccomplexity + β 13 *Segments + β 14 *CapEx + β 15 *Intangibles + IndustryFE + YearFE + ε (3) Most control variables are as defined above. We include complexity measures based on the sum of squared ratios of sales by industry (IndustryComplexity) or geographic (GeographicComplexity) segment to total sales as reported in the Compustat Segments database. We also include the total number of business segments reported (Segments). Finally, we include CapEx (CAPEX/LagAT) and Intangibles (INTAN/LagAT) to control for different asset mixes. In untabulated analysis, we estimate that larger firms, less geographically complex firms and firms with higher levels of SGA are more likely to have executives with tax backgrounds. Larger firms, less capital intensive firms and less R&D intensive firms are more likely to have independent directors with tax backgrounds. This approach yields treatment groups that are statistically similar to the control groups along the vast majority of variables reported in Table 2. However, this approach drastically reduces our sample sizes (to at most 228 total observations for tests of 26