Economic Consequences of Increasing the Conformity in Accounting for Uncertain Tax Benefits

Transcription

1 Economic Consequences of Increasing the Conformity in Accounting for Uncertain Tax Benefits Peter J. Frischmann Professor of Accounting Idaho State University College of Business Pocatello, Idaho Terry Shevlin Paul Pigott/PACCAR Professor of Business Administration Foster School of Business University of Washington Box Seattle, WA Ryan Wilson Tippie School of Business University of Iowa W280 Pappajohn Business Building Iowa City, IA August 8, 2008 Abstract: Commentary during the development of FASB Interpretation No. 48 suggests the interpretation could be costly for firms because new disclosure requirements could be used by the IRS to more effectively challenge uncertain tax positions. Stock returns around FIN 48 pronouncements suggest investors were not concerned about an increase in tax costs, and investors responded favorably to initial disclosures required under FIN 48. However, we document a significant negative market reaction to subsequent news of a Senate inquiry into these disclosures consistent with investors revising their beliefs over the potential for additional tax costs. We thank Cristi Gleason, Linda Krull, Lillian Mills, Tom Omer, Kathy Petroni, Sonja Olhoft Rego, John Robinson, Casey Schwab, Stephanie Sikes, an anonymous referee, and workshop participants at Michigan State University, Westfälische Wilhelms-Universität Münster, and the 2008 ATA mid-year meeting for helpful comments. We appreciate valuable research assistance from Jessica Tanner and Xiaoli Tian. Shevlin acknowledges financial support from his Paul Pigott/PACCAR Professorship, Frischmann from Idaho State University, and Wilson from the Iowa Old Gold Summer Fellowship. 1

2 Economic Consequences of Increasing the Conformity in Accounting for Uncertain Tax Benefits I. Introduction Tax law is often subject to varied interpretation, and many firms take positions on their income tax returns that might not be fully sustained under audit. Pursuant to SFAS No. 5 (FASB 1975) firms must accrue a loss contingency for uncertain tax positions if the loss is probable and reasonably estimable. Accordingly, under SFAS No. 5 managers have to exercise discretion in assessing both the materiality and the probability of a loss associated with these contingencies. The application of this discretion led the Financial Accounting Standards Board (FASB) to become concerned diverse accounting practices had developed resulting in inconsistency in the criteria used to recognize, derecognize, and measure benefits related to income taxes (FIN 48). For this reason, on July 13, 2006 the FASB issued Interpretation No. 48, Accounting for Uncertainty in Income Taxes: An Interpretation of FASB Statement No.109 (FASB 2006a, FIN 48). The purpose of FIN 48 is to improve the quality of financial reporting by reducing the significant diversity in practice associated with recognition and measurement in accounting for income taxes. FIN 48 requires firms to evaluate a tax position using a two-step process. First, a firm should recognize the financial statement benefit of a tax position only after determining it is morelikely-than-not the relevant tax authority will sustain the position following an audit. 1 Second, the amount recognized should be the largest benefit that has a greater than 50 percent likelihood of being realized upon ultimate settlement with the relevant tax authority. The amount not recognized is referred to as the unrecognized tax benefit (UTB) and is recorded as a tax 1 Recognizing means recording a financial statement benefit and could involve, among other items, recording a smaller current tax expense or reducing an existing tax related liability. 2

3 contingent liability. FIN 48 also contains significant disclosure requirements including a tabular reconciliation of the beginning and ending balances of unrecognized tax benefits. On July 14, 2005 the FASB issued an exposure draft (ED) on accounting for uncertain tax benefits that preceded the adoption of FIN 48. Critics expressed concern the new disclosure requirements would provide the IRS with a roadmap to better identify and examine firms most controversial tax positions. Robert Willens, a tax analyst at Lehman Brothers, noted firms will virtually be holding an arrow, pointing the IRS toward controversial tax benefits it might audit (Yoon 2006). Willens speculated the new rule would cost U.S. firms billions of dollars in charges to earnings and additional taxes. Uncertain tax positions represent an economically significant contingency for many firms. For example, prior to the adoption of FIN 48 Microsoft discloses in their 2006 Annual Report to Shareholders a long-term liability of $4.194 billion for tax contingencies. This amount is nearly as large as their 2006 net deferred tax assets of $4.551 billion and their 2006 tax expense of $5.663 billion. Despite the dire warnings of some commentators about the potential tax costs associated with FIN 48, FASB board members did not believe the new disclosures would provide too much information to tax authorities. At a May 16, 2006 board meeting, FASB board member Katherine Schipper stated the IRS has a far more detailed and effective roadmap in its schedule M-3 than it would be provided by any disclosure in the final interpretation (FASB 2006b). If the FASB board assertions are correct then FIN 48 would not be expected to result in additional tax costs for firms. Further, it is also possible FIN 48 could actually benefit shareholders. Prior research suggests managers use discretion in accounting for uncertain tax benefits to meet key earnings targets or smooth earnings (Dhaliwal et al. 2004; Gupta and Laux 2007; Blouin and 3

4 Tuna 2007). To the extent the passage of FIN 48 circumscribed this type of opportunistic behavior on the part of managers then we might expect to see a positive reaction from investors. To investigate investor perceptions of the costs and benefits associated with the adoption of FIN 48 we examine stock prices on the dates of important pronouncements from the FASB associated with the development and implementation of this new interpretation. If investors believe the tax burden of firms will increase as a result of FIN 48 we expect to find a negative market reaction to key pronouncements associated with FIN 48. For our complete sample of firms we do not observe a significant market reaction to any of the key pronouncement dates leading up to the implementation of FIN 48. Further, we also examine whether the market reaction to any of the key pronouncement dates varies cross-sectionally with two alternative measures of tax aggressiveness (the five-year cash effective tax rate and book-tax differences). We find little evidence the market reaction to key FIN 48 pronouncements varied as a function of our measures of tax aggressiveness. In total, we interpret these results as consistent with investors finding only limited credibility in claims FIN 48 would cost firms billions of dollars in additional taxes. Next, we examine the market reaction to the initial disclosure of unrecognized tax benefits required by FIN 48 in the first quarter of The purpose of these tests is twofold. First, the tests provide insight into whether these disclosures provide new information to investors. It is not clear prior to the adoption of FIN 48 whether investors were able to assess the magnitude of firms uncertain tax benefits. Second, to the extent these disclosures provide new information to investors, our tests show whether investors reacted negatively (as predicted by the increased tax and political scrutiny argument above) or positively. A positive reaction is consistent with 4

5 investors responding positively to news of tax aggressiveness 2 and/or revising downward their estimates of firms effective tax rates and consequently its after-tax net income upwards). We hand collect data on unrecognized tax benefits from the 2007 first quarter reports for a sample of 361 firms in the S&P 500 with calendar year-ends. Controlling for the level of unexpected earnings, we find the component of unrecognized tax benefits managers indicate would impact the firm s effective tax rate if recognized is positively associated with returns in the three-day window surrounding the release of the firms first quarter 2007 reports. 3 This finding is consistent with investors responding positively to the realization these initial disclosures would be of only limited use in helping tax authorities more effectively audit firms. 4 This finding is also consistent with investors responding positively to news of aggressive tax reporting and/or revising downward their estimates of firms effective tax rates (with a consequent upward revision in after-tax net income). On August 23, 2007 the U.S. Senate Permanent Subcommittee on Investigations sent letters to at least 30 companies seeking to obtain additional details about the underlying transactions associated with their FIN 48 disclosures (Drucker 2007). The subcommittee, chaired by Senator Carl Levin, had previously held numerous hearings on tax shelters and tax avoidance. 2 While aggressive tax planning/reporting strategies can help maximize shareholder value through reducing the tax burden of the firm, recent research suggests aggressive tax planning might also be associated with opportunistic behavior on the part of managers (Desai and Dharmapala 2006). As a result, it is not clear whether investors will react positively to news that firms are engaged in aggressive tax planning. 3 The total contingency for uncertain tax benefits (that is, unrecognized tax benefits, UTB) can be composed of two types of tax positions. Uncertainty can arise as to the timing of when to recognize income or take a deduction. Uncertainty can also arise when the ultimate deductibility is at issue. For tax positions associated with timing issues, adjustments to the UTB for those tax positions would impact the firms deferred taxes, but would not change the firm s total tax expense or effective tax rate (and therefore reported income). The component of UTB management indicates will impact the effective tax rate is associated with tax positions where the ultimate deductibility of a position is at issue. 4 This finding is consistent with the statements made by the IRS in their LMSB Field Examiners Guide on the implications of FIN 48. On the issue of FIN 48 providing a roadmap for the IRS the guide states The disclosures required under FIN 48 should give the Service a somewhat better view of a taxpayers uncertain tax positions; however, the disclosures do not have the specificity that would allow a perfect view of the issues and amounts at risk. 5

6 We predict this event represents a realization of the fears expressed by critics leading up to the adoption of FIN 48. That is, political pressure resulting from the size of the contingencies for uncertain tax benefits disclosed by firms under FIN 48 would lead to pressure on the IRS to abandon its policy of restraint with regard to obtaining tax accrual workpapers. 5 The Senate subcommittee request could be viewed as a strong signal by Congress to the IRS to abandon its policy of restraint. The request also raises the possibility Congress could enact new tax laws targeting the specific transactions underlying the reported UTBs. Consistent with these concerns, we find a significant negative market reaction on the date the Wall Street Journal first reported news of the subcommittees probe for our sample of 361 firms in the S&P 500 with a calendar year-ends. In supplemental analysis we also evaluate the association between the new disclosures for unrecognized tax benefits and two traditional measures of tax aggressiveness (the cash effective tax rate and book-tax differences researchers used before the availability of the disclosed FIN 48 unrecognized tax benefits). Unrecognized tax benefits provide an excellent measure of a firm s tax aggressiveness because they represent management s beliefs about the tax positions most likely to be challenged. We find no association between total unrecognized tax benefits and these traditional measures of tax aggressiveness. However, we do find both measures are significantly associated with the portion of the contingency for uncertain tax benefits management indicates will impact the effective tax rate. These findings provide some support for studies relying on variations of these measures to assess tax aggressiveness (e.g., Desai and Dharmapala 2006; Dyreng, Hanlon, and Maydew 2008; Wilson 2008). 5 In 1984, the U.S. Supreme Court ruled in the United States v. Arthur Young & Co. the IRS could subpoena auditor workpapers related to the tax reserve. However, until very recently, the IRS chose not to pursue this right and instead followed a policy of restraint in requesting auditor workpapers. Gleason and Mills (2007) report the IRS has recently begun to change this policy to examine audit workpapers associated with listed transactions (tax shelters). 6

7 Our paper contributes to the role of mandated disclosures in the presence of voluntary disclosures. The theoretical literature (Verrecchia 2001) suggests firms will voluntarily disclose information if the benefits exceeds the costs the benefits include informing external stakeholders if you have good news, while the costs include proprietary costs (often suggested to be informative to competitors). One reason firms might not voluntarily disclose tax contingencies is concern the disclosures provides useful information to the IRS increasing tax audit costs and tax settlements. Our results for the initial passage of FIN 48 (little negative market reaction) and first disclosures of firm-specific unrecognized tax benefits (positively associated with announcement period returns), suggests investors did not view these costs as large. Further, our results are consistent with investors responding favorably to the initial disclosure of the component of firms contingency for uncertain tax benefit management indicates will reduce their effective tax rate if the tax benefit is recognized. However, the stock market reaction to the subsequent Senate request for more information suggests investors might have revised their initial beliefs FIN 48 could impose additional tax costs on firms and firms initial concerns about increased tax costs were not ill founded. The remainder of the paper is organized as follows. The next section presents our hypotheses on the capital market effects of key pronouncements and the initial firm disclosures. Section III describes our sample selection procedure, tests and results of the market reaction during key event dates associated with the passage of FIN 48. Section IV describes our sample selection procedure, tests and results of the market reaction to the initial disclosures of firms unrecognized tax benefits and to the disclosure of the Senate inquiry. Section V examines the association between unrecognized tax benefits and traditional measures of tax aggressiveness and the final section concludes. 7

8 II. Summary of FIN 48 and Hypothesis Development Summary of FIN 48 Financial Accounting Standards No. 109 (FAS 109) ushered in an asset and liability approach for financial accounting and reporting for income taxes (FASB 1992). However, these principles do not adequately address the complexities of uncertain tax positions. Firms instead referred to SFAS 5, Accounting for Contingencies, to provide the primary guidance concerning the recognition and disclosure of uncertain tax benefits. SFAS No. 5 requires firms to accrue a loss contingency for uncertain tax positions if the loss is probable and reasonably estimable. SFAS No. 5 does not specifically allow firms to consider the probability of audit detection when assessing the probability of a loss, however Blouin et al. (2007) note many firms considered the probability of audit and detection in determining the likelihood and amount of potential losses. In contrast to SFAS No. 5, FIN 48 explicitly prohibits the consideration of audit detection risk in determining whether to recognize uncertain tax benefits. Providing explicit guidance on the consideration of audit risk in assessing the probability of loss is consistent with the FASB s goal of reducing the diversity of practice in accounting for uncertain tax benefits. In developing FIN 48, FASB considered the qualitative characteristics discussed in Concept Statement No. 2 emphasizing providing users with comparable information allows them to better detect similarities and differences between two economic events. FIN 48 provides a recognition threshold and a measurement attribute for the financial statement recognition and measurement of a tax position. The recognition threshold requires it is more-likely-than-not a tax position will be sustained upon audit. The threshold is based solely on the position s technical merits assuming the tax authority has all relevant information and ignores the probability of audit. The measurement attribute is the largest amount of benefit that has a greater than 50 percent 8

9 likelihood of being realized upon settlement. The resulting unrecognized tax benefit is recorded as a contingent liability. FIN 48 only addresses uncertain tax positions and does not change other FAS 109 deferred tax accounting and the use of the valuation allowance account for deferred tax assets. New disclosures include a tabular roll-forward of the total amounts of unrecognized tax benefits from the beginning to the end of the period and the total amount of unrecognized tax benefits, if recognized, would affect the effective tax rate. This disclosure also includes the total amount of interest and penalties recognized in the income statement and balance sheet and, for positions where it is reasonably possible unrecognized benefits will significantly change within 12 months, the nature of the uncertainty, the event that would cause the change, and an estimate of the range of the change, if possible. The pronouncement is effective for fiscal years beginning after December 15, Hypotheses development for events surrounding passage of FIN 48 Primarily, FIN 48 has two key objectives. First, it is intended to reduce the diversity in practice associated with accounting for uncertain tax positions. Second, it is intended to provide investors with additional information as to changes in and the amount of uncertain tax positions taken on existing tax returns. The adoption of FIN 48 could have an income statement and balance sheet effect; however the direction of the effect depends on the pre-fin 48 accounting practices of the firm. Because FIN 48 has an income effect, where the direction of the effect will differ by firm, we look to the disclosure characteristics of the pronouncement for directional predictions during its passage. To the extent investors believe increased disclosure will lead to an increased tax burden for firms we expect events increasing the probability of passage and events requiring disclosure of additional information to governmental authorities, will elicit a negative market reaction. We 9

10 classify the events leading up to the final issuance of FIN 48 as either positive or negative depending on this criterion (i.e., assuming here expected increased tax costs). In order to identify all possible event dates where FIN 48 was discussed by the FASB or the financial press we conducted a search of the Wall Street Journal, New York Times, and the FASB website. Table 1 provides a listing of each pronouncement associated with the passage of FIN 48 we examine in our study, and our ex ante prediction regarding whether the event will elicit either a positive or negative market reaction. Appendix A provides a complete discussion of each key pronouncement date. The predicted signs in both Table 1 and Appendix A are based on the assumption of increased tax costs resulting from FIN 48. The three dates receiving the most extensive press coverage were July 14, 2005 when the ED was issued, July 13, 2006 when the final interpretation was issued, and January 17, 2007 when the FASB rejected calls for a delay in implementation. The press coverage surrounding these events suggested the new requirements would result in more restrictive accounting for uncertain tax benefits for most firms and could provide the IRS with additional information to more effectively identify and audit uncertain tax positions. Our first hypothesis is: H1a: There will be a negative market reaction to pronouncements that increase the probability of passage and/or the extent of disclosure requirements in accounting for uncertain tax positions. An alternative prediction is increased disclosures under FIN 48 would lead to a reduction in managerial opportunism, benefiting shareholders and resulting in a positive stock market reaction. Desai and Dharmapala (2006) suggest in some contexts managers use tax shelters opportunistically in an effort to extract rents from the firm. Prior research also suggests managers might take advantage of the lack of transparency in accounting for income taxes to meet financial reporting objectives. Dhaliwal et al. (2004) and Gupta and Laux (2007) provide 10

11 evidence consistent with firms using discretion in accounting for these tax contingencies to meet analysts forecasts. Blouin and Tuna (2006) find evidence firms use their tax contingencies to smooth earnings. To the extent investors believed the new disclosures under FIN 48 would mitigate this type of managerial opportunism we would expect to see a positive stock market reaction around key pronouncements. This argument leads to an alternative prediction: H1b: There will be a positive market reaction to pronouncements that increase the probability of passage and/or the extent of disclosure requirements in accounting for uncertain tax positions. Cross-sectional predictions for events surrounding passage of FIN 48 Mills, Robinson, and Sansing (2007) develop a model of strategic tax compliance with tax law ambiguity in a setting where the government is provided a signal (FIN 48 disclosure) about the taxpayer s level of uncertainty regarding its tax filing position. Mills et al. (2007) find in cases where the taxpayer has strong facts to support their filing position, providing a signal to the government concerning the firm s level of uncertainty concerning the tax position (disclosure of a small UTB) actually increases the expected payoff to the position. However, in their model they show in cases where the taxpayer has weak facts resulting in disclosure of large a UTB, FIN 48 leads to a decrease in the taxpayers expected payoff. The reduction in expected payoff results from these firms either choosing to pursue less aggressive tax reporting strategies following the adoption of FIN 48, and/or these firms being audited more frequently after FIN In some targeted tax rule changes some firms benefit, but other firms might be losers as the beneficiaries bid up prices of the now to them tax-favored assets. For an example, see Berger s (1993) analysis of the corporate R&D tax credit. As modeled by Mills et al. (2007) some firms might win under FIN 48 (those classified as tax aggressiveness using the traditional proxies available pre FIN 48 but who have strong facts as evidenced by low post FIN 48 UTBs). However, investors cannot observe the UTBs until post FIN 48 so investors (and we as researchers) cannot observe the winners and losers prior to the new disclosures. 11

12 Based on the Mills et al (2007) analysis we believe firms engaged in aggressive tax reporting are likely to incur the most significant costs associated with FIN 48. For purposes of our analysis we define aggressive tax reporting as engaging in significant tax positions with relatively weak supporting facts. If the concerns the IRS will utilize the additional disclosures required by FIN 48 to more effectively audit firms or Congress will act to increase the tax burdens of the most aggressive firms (by curtailing tax laws that allow the tax aggressive firms to save taxes) are valid, then this will have the greatest impact on the most aggressive corporate taxpayers. 7 As a result, we predict firm-specific variation in aggressive tax reporting behavior will lead to crosssectional differences in the market reaction to events leading up to FIN 48. This prediction leads to our second hypothesis: H2a: The stock price reaction to pronouncements that increase disclosure requirements for uncertain tax positions will be more negative for firms investors believe are more tax aggressive. However, the same arguments used to predict a possible positive market reaction to the passage of FIN 48 (reduced managerial opportunism relating to both rent extraction and earnings management via the tax cushion) apply to firms considered more tax aggressive. This argument leads to the alternative prediction: H2b: The stock price reaction to pronouncements that increase disclosure requirements for uncertain tax positions will be more positive for firms investors believe are more tax aggressive. Hypotheses development for initial FIN 48 disclosures In our second set of tests we examine the stock market reaction to firms first disclosures of the unrecognized tax benefits (UTB). In the unlikely event investors are somehow able to 7 This political cost argument is similar to that surrounding the passage of the Tax Reform Act of 1986 that, although it reduced corporate tax rates, it widened the tax base (by for example restricting deductions) in response to a series of reports by the Citizens of Tax Justice identifying many large profitable U.S. corporations not paying any U.S. taxes. See Murray and Birnbaum (1987) for a discussion of the political background to TRA Shevlin and Porter (1992) examine the effect of TRA 86 on corporate tax burdens. 12

13 estimate the extent of UTBs prior to the adoption of FIN 48 and the filing date for the first quarter of Q then the disclosure will not present new information to investors. In contrast, to the extent the UTB disclosures provide new information to the market about the magnitude of firms UTBs then we expect a market reaction. Building on the increased tax (and political) costs argument above, firms reporting large UTBs are most likely to receive the closest scrutiny by the IRS and Congress. If investors believe the disclosures provide information allowing tax authorities to more effectively identify and challenge firms uncertain tax positions then we expect the market reaction to be negatively associated with the disclosed contingency for uncertain tax benefits. Another argument supporting a negative association between returns and the initial UTB disclosures arises from recent literature arguing tax aggressiveness can be used to obscure or mask managerial opportunism. Desai and Dharmapala (2006) examine the issue of aggressive tax reporting within the context of the agency problems faced by public corporations. They argue because tax shelters are designed to obscure the underlying purpose of a transaction, shelters represent a potential vehicle for rent extraction on the part of management. To the extent investors perceive this to be the case, and the UTB disclosures cause them to reassess their beliefs about the level of firms tax aggressiveness, then we would expect a negative association between UTBs and abnormal returns. Hanlon and Slemrod (2007) also investigate the market reaction to news of tax aggressiveness by examining the stock price reaction to news a firm participated in a tax shelter. They find, on average, a company s stock price declines on news of tax shelter participation, but the price decline is smaller than the reaction to major accounting mishaps (such as restatements). While these results suggest the market reacts negatively to tax sheltering it is not clear whether 13

14 this reaction is driven by news a firm is more tax aggressive than previously thought or whether it is a reaction to news a firm s aggressive tax planning is subject to IRS and public scrutiny that could result in additional penalties and legal costs. Examining the initial UTB disclosures under FIN 48 allows us to investigate the market reaction to news of tax aggressiveness without having to disentangle the effects of news of additional penalties and legal costs. 8 The above arguments lead us to predict a negative reaction to firm-specific UTB disclosures. H3a: The stock market reaction will be negatively associated with firm s first disclosures of their unrecognized tax benefits. On the other hand, it is possible investors will react positively to the first disclosures. If investors view tax aggressiveness as a firm value maximizing activity, and if large UTB s reflect tax aggressiveness, then investors could react positively to UTB disclosures. Examining a sample of identified tax shelter participants, Wilson (2008) finds the tax shelter firms with strong corporate governance exhibit positive abnormal returns during the period of active tax sheltering. This finding is consistent with tax sheltering being a tool for wealth creation within wellgoverned firms. Using confidential IRS data, Lisowsky (2008) documents a link between estimates of the contingency for uncertain tax benefits reported under FIN 48 and listed and reportable transactions from the IRS office of tax shelter analysis. As such, to the extent large contingencies for uncertain tax benefits are a function of tax shelter activity prior research indicates can be a tool for wealth creation within the firm then we might expect investors to react positively to firms disclosing large contingencies for uncertain tax benefits. 8 If investors believe managers have not sufficiently reserved for uncertain tax positions and firms with significant UTB disclosures actually have even greater uncertain tax positions that could be reversed under audit, then we would expect a negative reaction by investors. Anecdotal evidence suggests this is unlikely to be the case because firms rarely appear to be under-reserved at the time disputes concerning uncertain tax positions are resolved with the relevant tax authorities. In fact, recent research suggests firms have an incentive to be over-reserved for uncertain tax positions, because firms use discretion in accounting for these reserves to meet analyst forecasts (Dhaliwal et al and Gupta and Laux 2007). 14

15 It is also possible investors will react positively to the disclosure of large contingencies for uncertain tax benefits because they believe these contingencies are overstated. In assessing whether a tax position is more-likely-than-not to be sustained under audit, FIN 48 requires managers to assume the IRS has access to all relevant information and to ignore the probability of audit. In a comment letter to FASB regarding FIN 48 the Tax Executives Institute noted these standards would lead to systematic overstatements of tax liabilities compared with an enterprise s best, reasonable estimate of the likely outcome of its tax positions (TEI, 2005). The letter goes on to argue firms should be able to rely on their prior settlement experience and the settlement experience of others in evaluating whether a tax position is sustainable. Consistent with this reasoning, it is possible investors will initially assess firms contingencies for uncertain tax benefits as being overstated. These arguments lead us to predict a positive reaction to firmspecific UTB disclosures H3b: The stock market reaction will be positively associated with firm s first disclosures of their unrecognized tax benefits. Hypothesis development for Congressional Inquiry On September 11, 2007 the Wall Street Journal reported the U.S Senate Permanent Subcommittee on Investigations had sent letters to at least 30 firms requesting additional information associated with their FIN 48 disclosures (Drucker 2007). Among the questions asked in the letter was a request to describe any tax position or group of similar tax positions representing 5 percent or more of the firm s total unrecognized tax benefit. News of the Senate probe raised concern Congress would push the IRS to abandon its policy of restraint regarding tax accrual workpapers. Bill Smith, a tax attorney and tax director of CBIZ, Inc. noted if the Senators believe that corporations are pushing the tax envelope too hard, they will get their 15

16 hackles up and push the IRS to change its policy (Leone 2007). It is reasonable to assume the tax accrual workpapers provide more detailed information concerning firms most aggressive tax transactions and would be more useful to the IRS than the disclosures required under FIN 48. Investors might also believe Congress would enact new tax laws targeting the transactions underlying the disclosed UTBs increasing the future tax burdens of firms. Further, some critics noted the Congressional inquiry could have a chilling effect on legitimate transactions with unclear tax consequences. These arguments lead us to predict a negative reaction to news of the Senate investigation. H4: There will be a negative market reaction to news of the Congressional inquiry into firms FIN 48 disclosures. III. Tests of Market Reaction to Key Event Dates Surrounding Passage of FIN 48 We conduct our analysis of H1 and H2 on a sample of firms meeting the following criteria: (1) they are included on the CRSP daily returns and the Compustat Annual Industrial files; (2) they have daily stock return data available from January 1, 2004 through March 31, 2007; (3) they have the necessary Compustat data available to calculate our two (pre-fin 48) aggressive tax reporting proxies described below ending in the 2004 fiscal year; (4) they have positive cumulative pretax income over the five-year period ending with the 2004 fiscal year; and (5) they have a cash ETR between plus and minus 1. Criteria (1) through (3) are employed to ensure we have data available to test hypotheses 1 and 2. Criteria (4) is included to ensure our tests focus on firms expected to be significantly impacted by FIN 48. Firms that have not been profitable over a sustained period of five years are unlikely to have a significant tax burden and therefore are unlikely to face increased costs as a 16

17 result of FIN 48 (unless their losses are due to aggressive tax planning). Criteria (5) is adopted to eliminate extreme and unusual observations as it is not clear what these extreme Cash ETRs signal to investors about the level of a firm s tax aggressiveness. Application of these selection criteria resulted in a final sample of 1,381 firms. In order for the market reaction to vary as a function of a firm s tax aggressiveness investors must be able to identify the firms that are the most aggressive taxpayers. Because the characteristics investors rely on to identify tax aggressive firms prior to FIN 48 are not clear we perform our analysis using two alternative measures proposed in recent research as proxies for tax aggressiveness. Specifically, we examine whether the market reaction to key FIN 48 pronouncements varies as a function of the long-run cash effective tax rate developed by Dyreng et al. (2008), and as a function of aggregate book-tax differences. 9 The cash effective tax rate Our first tax aggressiveness measure is the long-run cash effective tax rate (Cash ETR) introduced by Dyreng et al. (2008) calculated as the ratio of the five-year sum of cash taxes paid to the five year sum of pretax financial accounting income. As discussed in Dyreng et al. (2008), this measure of tax aggressiveness has several advantages over the traditional ETR measure. First, the Cash ETR measure is not affected by changes in the firm s tax contingency. Regardless of whether a firm recognizes the benefit associated with an aggressive tax position in earnings, the reduced cash tax payments resulting from that aggressive position will be reflected in a lower Cash ETR for the firm. Second, to the extent firms are aggressively accelerating expenses or 9 To the extent the assumption investors believe firms that are the most tax aggressive will incur the greatest additional costs associated with FIN 48, these tests provide insight into the question of how investors identify aggressive corporate taxpayers. That is, observing cross-sectional variation in the market reaction to key pronouncements as a function of these two tax aggressiveness measures provides some evidence investors rely on these measures (or other firm characteristics correlated with these measures) in identifying aggressive corporate taxpayers. 17

18 deferring income for tax purposes this will be reflected in a lower Cash ETR provided these timing differences do not reverse within the five-year period Cash ETR is measured. Finally, the Cash ETR measure will be reduced by the tax benefit associated with employee stock options and therefore provides a better measure of the firm s true tax burden than the traditional ETR measure. Of course, if Cash ETR is low due mostly to ESO tax benefits, this lowers the power of our tests because our proxy might be classifying firms as tax aggressive when, in fact, they are not. Following Dyreng et al. (2008) we calculate Cash ETR as the five-year sum of taxes paid (data 317) divided by the corresponding five-year sum of pre-tax book income (data 170) minus special items (data 17). For firms with missing data on cash taxes paid, cash taxes paid are estimated by current tax expense. Although not without problems, this measure overcomes many of the shortcomings associated with ETR. 10 It is unaffected by valuation and contingency estimates, and takes into account the tax benefits of employee stock options. Book-tax differences Our second proxy for tax aggressiveness is an estimate of the difference between pretax book income and taxable income (BTDs). Some previous research suggests BTDs could be an effective signal of tax aggressiveness. Desai (2003) posits the growing difference between book and taxable income during the 1990 s was caused by increased levels of tax sheltering. Wilson (2008) finds BTDs are positively associated with actual incidence of tax sheltering. Further, Mills (1998) finds proposed IRS audit adjustments are positively related to large BTDs. Despite evidence large positive BTDs are associated with aggressive tax reporting this measure also has 10 Cash taxes paid represents the actual taxes paid by the firm during a given year, and as a result could include estimated tax payments associated with the prior year s income. However, because we utilize a five-year measure of Cash ETR we expect the affect of the timing of estimated tax payments will add only a limited amount of noise to our measure of the Cash ETR. Use of the 5-year average of PTBI reduces the effect of earnings management on the ratio leaving the ratio as a measure of tax aggressiveness. 18

19 limitations. Hanlon (2003) and Manzon and Plesko (2002) identify firm specific characteristics that are determinants of BTDs, but not necessarily reflective of aggressive tax reporting. For example, firms with large capital expenditures could have significant book-tax differences associated with depreciation, but this would not be reflective of an aggressive tax planning/reporting strategy. To the extent earnings management activities and innate firm characteristics unrelated to aggressive tax reporting are the primary determinants of BTDs, this could limit the usefulness of BTDs as a proxy for tax aggressiveness. BTDs are computed as pre-tax book income less an estimate of taxable income. Taxable income is calculated by grossing up the sum of the current federal tax expense (data 63) and the current foreign tax expense (data 64) and subtracting the change in NOL Carryforward (data 52). If the current federal tax expense is missing, total current tax expense is calculated by subtracting deferred taxes (data 50), state income taxes (data 173) and other income taxes (data 211) from total income taxes (data 16). Book income is calculated as pre-tax book income (data 170) less minority interest (data 49). If our estimate of BTDs is associated with the type of firm investors identify as tax aggressive then under H2a we expect the stock price reaction to key pronouncement associated with FIN 48 to be greater for firms with large BTDs. Descriptive statistics for the samples and variables of interest appear in Panel A of Table Column (1) of Table 2 indicates the median sample firm has a cash effective tax rate of 25% and positive book-tax differences. Table 2 also reports descriptive statistics for the firms categorized into the most aggressive tax quintiles according to both measures of tax aggressiveness. The firms in the lowest Cash ETR quintiles are larger than the median firm from the entire sample and report larger book-tax differences. Firms in the highest book-tax difference quintile are 11 To ensure that our results are not driven by outliers, each of the tax aggressiveness variables are winsorized at the 1 and 99 percent level. 19

20 smaller than the median firm from the full sample and report lower cash effective tax rates than the entire sample of firms. Panel B of Table 2 presents the Pearson and Spearman correlation coefficients between the variables of interest. As expected, we observe a significant negative Pearson and Spearman correlation between the Cash ETR measure of tax aggressiveness and BTDs, consistent with these proxies both capturing to some extent the same underlying construct of tax aggressiveness. The industry distribution of the sample firms is presented in Panel C of Table 2. Of the 1,381 sample firms, the largest concentration of firms is 631 in the manufacturing industry followed by 185 in the wholesale industry. Market response to key FIN 48 pronouncements Our first hypothesis examines the impact of 11 events associated with FIN 48 on stock prices. We examine hypothesis 1 using a Multivariate Regression Model (MVRM) proposed by Schipper and Thompson (1983). The MVRM model conditions the return generating process on the occurrence or nonoccurrence of an event by adding a dummy variable to the market model for each event (Espahbodi et al. 2002). Each dummy variable is set equal to 1 for key event dates and zero otherwise. The coefficient on each dummy variable measures the impact of each respective event on stock returns. Because the exact timing of the information release associated with each pronouncement is not clear we use a three-day window corresponding to the days t = - 1, 0, +1 associated with each announcement date described in Table 1. We estimate the stock market reaction using the following regression model: pt p p mt pk kt pt, k = 1 K R = α + β R + g D + e (1) R = the return on portfolio p on day t (t = 1,2, T). T is the total number of daily return pt observations from the beginning of 2004 through March of

21 R = the return on the CRSP value-weighted portfolio on day t; mt α = intercept coefficient for portfolio p; p β p = risk coefficient for portfolio p; g pk = the effect of event k (k = 1,2,.K) on portfolio p s return. That is, g is an estimate of the abnormal return on the portfolio on event date k. K is the total number of events examined: 11 in this study; D kt = dummy variable for the kth event set equal to 1 during the 3-day period (t = -1, +1 relative to the announcement date), zero otherwise; and e = random disturbance assumed to be both normal and independent of the explanatory pt variables. We estimate the regression for three portfolios: all sample firms, and two portfolios comprising the firms categorized in the top quintile of aggressive tax reporting firms using each of the two measures of tax aggressiveness. Within each portfolio, firm returns are equallyweighted. Equation (1) estimates the impact of the key pronouncements associated with FIN 48 on the sample firms in each portfolio. The estimation of the model in equation (1) assumes the residuals are independently and identically distributed. Because returns are assumed to be serially independent and the unit of analysis is portfolio returns we are not concerned with either cross-sectional heteroscedasticity or time-series dependence. However, to correct for possible time-series heteroscedasticity we use a procedure developed by White (1980) that allows the variance-covariance matrix of the residuals to vary across observations. Table 3 reports the portfolio abnormal returns and the t-statistics for the full sample as well as the two portfolios containing the top quintile of each tax aggressiveness proxy for the 3-day period around each of the 11 key pronouncement dates. The portfolio abnormal returns reported in Table 3 are the coefficient estimates on the dummy variables in equation (1). We consider coefficient estimates significant if p <.05 in a two-tailed test reflecting the competing predictions in H1a and H1b. The results in Column (1) of Table 3 indicate there was not a 21

22 significant negative market reaction for the full sample of firms around any of the 11 key event dates. The remaining columns in Table 3 present the abnormal returns for the two portfolios of aggressive tax reporting firms. Note, these columns do not present a direct test of whether the returns are associated with tax aggressiveness characteristics (those tests are presented in Table 4). Rather, these columns simply present the equally-weighted abnormal returns for these two alternative sub-samples of aggressive tax reporting firms. The only significant market reaction to any of the key event dates for either sub-sample of firms is for the low Cash ETR firms around the issuance of the initial exposure draft on July 14, Table 3 also reports an F-statistic for each sub-sample of firms testing whether the cumulative total returns across the key dates are significantly different than zero. The results indicate the cumulative market response across the pronouncement dates is not significant at the 0.05 level or better for the complete sample of firms and for both sub-samples of tax aggressive firms. Taken together, the results in Table 3 provide only limited evidence investors believed firms would incur significant additional costs as a result of FIN 48. We do not interpret these results as consistent with investors believing firms would incur billions of dollars in additional tax costs as suggested by some commentators. Examining cross-sectional variation in the market response to key pronouncements Hypothesis 2 examines whether the impact of key FIN 48 pronouncements is a function of firm characteristics associated with tax aggressiveness. To test the effects of these firm characteristics on the stock price reaction to our key event dates we use the portfolio weighting procedure developed by Sefcik and Thompson (1986). The Sefcik and Thompson (1986) method 12 In supplemental tests (not tabulated) we also examined the market reaction to key pronouncements for portfolios of the least tax aggressive quintiles of firms (lowest BTD and highest Cash ETR). We found no significant stock market reactions for these portfolios of firms to any of the key pronouncement dates. We also examined (not tabulated) the market reaction for the 93 firms in both the low Cash ETR and high BTD quintile portfolios. We do not observe a significant market reaction to the key pronouncement dates for this subset of 93 firms. 22

23 is used to account for cross-sectional heteroscedasticity and cross-correlation of the residuals likely to occur in the presence of common event dates. Under this estimation approach, the information pertaining to the full covariance matrix of the residuals is included by creating separate portfolios for P types of tax aggressiveness proxies. Each portfolio is then used to reestimate equation (1) above. The estimates of g pk reflect the effect of the pth tax aggressiveness proxy on the stock-price reaction to the kth event. Consistent with Espahbodi et al. (2002), we implement the Sefcik and Thompson (1986) procedure using the following approach. The first step in this procedure is the formation of an N x 2 matrix X having a column of ones and a column containing each firm s tax aggressiveness characteristic, namely either Cash ETRs or BTDs. The next step is to create portfolio weights 13 (W) and compute the portfolio returns ( R pt ) as follows: (X X) -1 X = W R pt = W p R it, p = 1,2 t = 1,2,.,T, i = 1,2,,N, (2) where: W = P N matrix of portfolio weights (P = 2 and N = 1,446 firms). Column 1 contains a column of 1 s and column 2 contains the tax aggressiveness characteristic for each firm; R pt = the portfolio return on day t; and R it = N 1 vector of individual firms security returns on day t. Using the weighted portfolio returns (where the weight is based on the firm s tax aggressiveness measure), we re-estimate equation 1. We form weighted portfolio returns separately for each measure of tax aggressiveness because we are interested in testing the association between 13 The Sefcik and Thompson (1986) approach is similar to a cross-sectional regression of abnormal returns on firms tax aggressiveness characteristic at each event date: AR it = α 0 + α 1 Tax Agg it + ε. In this equation the coefficient α 1 is an estimate of the tax aggressiveness measure on the market reaction to a key FIN 48 pronouncement. This estimate, α 1, is identical to the estimate, g pk, produced by the Sefcik and Thompson (1986) weighted return approach. However, the Sefcik and Thompson (1986) approach accounts for cross-sectional heteroscedasticity and crosscorrelation of the residuals providing a correct estimate of the standard error. 23

24 returns and the construct of tax aggressiveness rather than the incremental effect of each measure of tax aggressiveness. Table 4 presents the results of our tests of hypothesis 2. In interpreting the results presented in Table 4 it is important to note all of the firms in our sample were profitable during the fiveyear period leading up to these key pronouncements. It is therefore likely most of the firms in our sample would have an incentive to take some aggressive tax positions and would all be impacted to some degree by the implementation of FIN 48. This fact could limit our ability to detect crosssectional differences in the market response to key events based on measures of tax aggressiveness. Because H2a and H2b provide competing predictions about the sign of the association between stock returns and tax aggressiveness, we again conduct two-tailed significance tests. The results in Columns (1) and (2) of Table 4 indicate significance in the predicted direction in only one of the 11 event dates: July 14, 2005, (event 5) the date the initial exposure draft was released. The positive association in column (1) indicates the lower the Cash ETR (the more tax aggressive the firm) the more negative the stock return. The negative association in column (2) indicates the higher the BTD (the more tax aggressive the firm) the more negative the stock return. We also observe a positive association between BTDs and the market reaction to pronouncements on November 17, However, the FASB pronouncements on this date received limited press coverage and we are uncertain as to what is driving this association. Taken together, Tables 3 and 4 provide at best limited or weak evidence investors expected FIN 48 to impose increased tax costs on firms. An alternative explanation for the results in Table 4 is we have measurement error in our proxies for tax aggressiveness. However, we shed some light on this in our supplemental analysis when we regress the disclosed UTB amounts on our tax 24