John L. Campbell, Jenny Xinjiao Guan, Oliver Zhen Li, and Zhen Zheng

Recent research finds that stock-based compensation contracts encourage CEOs to take more risks and, thus, engage in more tax planning (Rego and Wilson 2012; Armstrong, Blouin, Jagolinzer, and Larcker 2015). However, stock options only provide managers with upside gain potential, and fail to protect managers from downside risk (Cadman, Campbell, and Klasa 2016). Downside risk protection is important in the tax setting because tax positions have been shown to draw scrutiny from outside parties (i.e., the IRS, the SEC, analysts, and the media), and thus have the potential to attract unwanted attention on CEOs and their firms (Bozanic, Hoopes, Thornock, and Williams 2017; Kubick, Lynch, Mayberry, and Omer 2016; Ehinger, Lee, Stomberg, and Towery 2017; Chen, Powers, and Stomberg 2016). This attention can even potentially lead to CEO job termination (Chyz and Gaertner 2018).

Prior research provides evidence that severance pay reduces managers' career concerns and encourages otherwise risk-averse managers to take higher levels of risk (Ju, Leland, and Senbet 2014; Chen, Cheng, Lo, and Wang 2015; Cadman et al. 2016). In this study, we examine whether the increased risk-taking behavior induced by severance pay extends to the tax setting. As in these other risk-taking settings, it is not necessary to assume that a CEO believes that s/he will be fired solely for engaging in incremental tax planning. Rather, as long as tax planning increases the attention paid to a CEO in a way that would increase her/his career concerns, and this expected increase in career concerns prevents the CEO from taking these tax positions, severance pay will mitigate this tendency and the CEO will take more tax risks than s/he would absent the severance pay.

We find that CEO severance pay is positively associated with corporate tax planning, consistent with CEO severance pay providing contractual protection against managers' career concerns and thereby inducing otherwise risk-averse managers to engage in incremental levels of tax planning. This result holds under an instrumental variable approach and propensity score matching, and survives alternative measures of CEO severance pay and corporate tax planning. Finally, we find that severance pay provides stronger tax planning incentives in situations where managers are expected to face greater career concerns—when they are less experienced, when they face stronger shareholder monitoring, and when they manage firms with higher idiosyncratic volatility. Overall, our results suggest that CEO severance pay represents a form of efficient contracting with otherwise risk-averse managers.

We add to the literature on compensation contracts and corporate tax planning. Prior studies focus primarily on the risk incentives provided by stock options. Importantly, these studies examine compensation contracts that only provide incentives when firm performance is positive. In contrast, we examine a compensation agreement, CEO severance pay, that completes the spectrum of compensation incentives provided to risk-averse CEOs by mitigating their career concerns (i.e., providing protection for downside risk). Our study also extends the literature on CEO severance pay. Prior research finds that severance pay encourages managers to take risks that increase firm value. We extend this empirical work to the tax setting. Finally, our study has practical implications. As a prevalent compensation practice, severance pay has received a lot of attention from practitioners and regulators. Our results, taken together with prior studies, suggest that ex ante severance pay, on average, provides incentives for otherwise risk-averse managers to take value-increasing risks.

Jimmy F. Downes, Mollie E. Mathis, and Lisa Kutcher

The accounting for multinational firms' foreign earnings is complex. Prior to the Tax Cuts and Jobs Act (TCJA) of 2017, the U.S. had a worldwide tax system that taxed firms' domestic earnings immediately but deferred taxation of foreign earnings until repatriation to the U.S. parent or when the subsidiary was sold or liquidated. U.S. Generally Accepted Accounting Principles (GAAP) include the current earnings of foreign subsidiaries in firms' net income. The timing difference between when foreign profits were earned and when they were repatriated resulted in the recognition of a deferred income tax expense and a deferred tax liability equal to the future taxes payable on these earnings. Although several repatriation methods existed, the most common method was to pay a dividend from the foreign subsidiary to the U.S. parent. In calculating U.S. taxes, repatriated earnings were “grossed up” to earnings before foreign taxes. Applying the U.S. statutory tax rate to the grossed-up foreign earnings determined the U.S. tax liability. The foreign tax credit reduced this liability for foreign taxes paid.

We examine whether the repatriation of foreign earnings is sensitive to firm-specific currency exposure and whether investors price these repatriations differently than repatriations that are not subject to currency exposure. When multinational corporations repatriate, the repatriated earnings are converted into U.S. dollars (USD) at the exchange rate in effect during the repatriation period (i.e., the spot rate), but the foreign tax credit remains anchored at the exchange rates in effect when the corporation paid foreign taxes (i.e., the blended rate). Because the period in which foreign taxes are paid often differs from the repatriation period, as the USD strengthens relative to firm-specific foreign currencies, firms face a lower effective tax cost of repatriation. This lower tax cost comes at the potential expense of a lower after-tax dividend. We find that repatriation likelihood increases during periods of firm-specific USD strength and that investors place a premium on repatriation costs when the USD strengthens against a firm-specific basket of currencies for repatriating firms. This premium implies that investors value the benefit of a lower effective tax cost of repatriation more than the potential cost of a lower after-tax dividend available to the U.S. parent.

Our findings are relevant given the deemed repatriation in the TCJA, which requires a one-time deemed repatriation of post-1986 unremitted foreign earnings with a tax rate of 15.5 percent for cash and cash equivalents, and 8 percent for all other earnings. Revenues derived from this repatriation are dependent on the strength of the USD relative to the anchored foreign tax credit at the time of the deemed repatriation. This study also contributes to the literature on foreign currency exposure. We are the first to develop a firm-specific measure of foreign currency exposure that does not rely on a measure of stock returns. Additionally, our study contributes to the literature on the determinants of repatriation by providing evidence that firm-specific USD strength increases repatriation likelihood, and financial reporting incentives may motivate this decision. Finally, this study answers the call for research on users of tax-based information in the financial statements (Graham, Raedy, and Shackelford 2012).

Mark (Shuai) Ma and Wayne B. Thomas

Many governments and organizations around the world are attempting to deter corporate tax avoidance. Proposals include penalties imposed directly on corporate officers as well as penalties imposed on the corporation (i.e., shareholders). For example, penalties on shareholders have doubled since 2005 for large Australian multinational firms that do not have a reasonable position for their taxable income. Similar proposals for penalties on shareholders are also considered by the U.K. treasury and the U.S. government (e.g., the Stop Tax Haven Abuse Act). However, the former U.K. Liberal Democrat chief secretary, Danny Alexander, and others suggest that the government should instead employ more penalties directly on corporate officers.

Our study examines how a jurisdiction's legal environment—penalties on corporate officers versus penalties on the corporation—affects corporate tax avoidance. We use a sample of U.S. firms that operate in multiple states. For penalties on corporate officers, we use the maximum penalty in each state in terms of length of prison sentence. The maximum prison sentence is intended to reflect not only the perceived threat of penalty, but also the state's attitude toward imposing penalties directly on the officer. Penalties on the corporation are defined in terms of the maximum monetary fine as a percentage of underpayment.

We find that the extent of penalties on officers deters state tax avoidance, while the extent of penalties on the corporation/shareholders does not. These findings are consistent with theory. A penalty on an officer has a direct impact on the officer's expected utility and therefore is expected to reduce the officer's willingness to make tax avoidance decisions. In contrast, a penalty on the corporation and its shareholders may not directly impact the officer. We conclude that the threat of penalties on officers may offer governments a lower-cost deterrent effect that persists over a longer period of time, increasing net tax revenues.

Mollie E. Mathis

Prior to the Tax Cuts and Jobs Act of 2017, U.S. GAAP allowed firms to avoid accruing for U.S. taxes on foreign income while still including the earnings in net income when the firm stated the earnings were permanently reinvested earnings (PRE). Krull (2004) finds evidence that firms manage earnings up through PRE, using a modified version of the “backing-out” methodology. However, the backing-out methodology is potentially problematic due to mechanical associations between variables. Prior literature shows that earnings management studies using the backing-out methodology are subject to Type I errors because of this mechanical association (Lim and Lustgarten 2002; Elgers, Pfeiffer, and Porter 2003). Specifically, rejection of the null hypothesis in these studies is consistent with firms managing earnings through (1) the suspect earnings management tool, (2) an alternative earnings management tool, or (3) firms not managing earnings (Lim and Lustgarten 2002), making the results difficult to interpret.

In this paper, I reexamine earnings management through PRE using two methods to measure pre-managed earnings. Gupta, Laux, and Lynch (2016) provide a measure (the Gupta measure) of pre-managed earnings by using analysts' forecasts of pretax and after-tax income to estimate pre-managed earnings, which are earnings before any earnings management through the income tax accrual. The Gupta measure provides a measure that is independent of the dependent variable and, thus, is not subject to the specification issues of the Krull measure. In the cross-section, I find evidence consistent with upward and downward earnings management through PRE using the Krull measure, but not using the Gupta measure. I follow the Lim and Lustgarten (2002) methodology to test the pervasiveness of Type I errors in the Krull and Gupta measures separately. I find that the Krull measure is susceptible to Type I errors and the Gupta measure is not. I repeat my main analyses on a subsample of firms with the greatest ability to manage earnings up through PRE. I limit my sample to firms with less than 90 percent of unremitted earnings designated as PRE. Consistent with my primary analysis, I continue to find support for upward earnings management through PRE using the Krull measure, and I also find support for upward earnings management through PRE using the Gupta measure within this subsample.

My research contributes to the literature on specification issues within the earnings management literature. Lim and Lustgarten (2002) demonstrate the Type I error rate in earnings management studies that use the backing-out method for specific earnings management tools. Using their methodology, I demonstrate that these issues persist in the tax literature. I also contribute to the literature on earnings management through income tax expense. I provide a methodological improvement by using a measure of pre-managed earnings free from Type I errors. Using this measure, my results suggest that only a subset of firms with the greatest ability manage earnings upward through PRE, indicating that earnings management through PRE is not as pervasive as prior research indicated.

Terry Shevlin, Oktay Urcan, and Florin P. Vasvari

Prior evidence indicates that bank loan spreads and public bond offering yields are increasing in corporate tax avoidance, with the economic effects being substantially larger in bonds than bank loans (Hasan et al. 2014). However, the literature has not investigated the channels through which borrowers' tax avoidance activities influence the spreads of their newly issued debt. We examine the importance of the paths through which corporate tax avoidance affects bond yields and bank loan spreads and provide additional cross-sectional analyses on firm characteristics that likely affect the relationship between corporate tax avoidance on bond yields.

We consider two paths through which tax avoidance could be positively associated with bond yields and loan spreads. First, corporate tax avoidance activities could induce higher uncertainty about the magnitude and volatility of the firms' future cash flow, which is viewed negatively by lenders. Second, corporate tax avoidance activities contribute to a decrease in the quality and transparency of the financial statements (Desai and Dharmapala 2006; Desai et al. 2007; Balakrishnan et al. 2019).

Our empirical tests use a sample of 7,945 individual bonds issued by 1,035 U.S. domiciled nonfinancial firms over the period 1990–2007 and 6,015 bank loans obtained by 1,297 firms over the same period. We document that tax avoidance predicts lower future pre-tax cash flow levels, higher future pre-tax cash flow volatility, and poorer future information quality, and that tax avoidance influences bond yields and bank loan spreads through these mediating variables. The standardized coefficients and significance levels are higher in the bond sample, consistent with the interpretation that these mediating variables play a larger role for bonds than for bank loans. We find that approximately one half of the total effect of tax avoidance on bond yields is explained through the negative effect of tax avoidance on future pre-tax cash flow levels and volatility and, to a lesser extent, poorer information quality.

In additional cross-sectional analyses, we further investigate the role of two sets of firm characteristics that potentially affect the association between corporate tax avoidance and debt spreads: (1) lender wealth expropriation incentives, which reflect conflicts of interest between equity holders and lenders, and (2) the probability of an IRS audit, which reflects an alternative monitoring source that reduces lender risk but also increases tax risk. We capture lenders' wealth expropriation incentives using four empirical constructs: the presence of strong antitakeover provisions, the presence of large blockholders, high managerial equity incentives, and high credit risk. We find that corporate tax avoidance increases the cost of bond financing, but not bank financing, when managers have incentives to expropriate lenders' wealth and firms face a high probability of an IRS audit.

Overall, our results suggest that, relative to bond investors, banks are better able to reduce information asymmetry problems and effectively monitor borrowers' activities due to banks' better access to borrowers' private information.

Beth Y. Vermeer, Brian C. Spilker, and Anthony P. Curatola

This study examines whether and to what extent tax professionals' economic and social relationships with clients separately and jointly affect tax professionals' propensity to recommend aggressive tax positions to clients facing ambiguous tax issues. Because such relationships are independent of the information necessary to resolve tax issues, economic and social relationships normatively should not affect recommendations. To the extent economic and social relationships with clients cause professionals to recommend more aggressive positions than they would otherwise, economic and social relationships can expose both the professionals and the client to more risk than is intended.

In general, we expect that the economic importance of a client provides incentives for tax professionals to recommend aggressive tax positions. However, the potential exposure to increased risks, especially as client fees become more significant, may result in an inflection point where the potential costs of aggressive reporting outweigh the potential rewards. Specifically, as clients become more economically important, tax professionals must balance the positive economic incentives of maintaining fees and enhancing future revenue streams with the increasing risks associated with aggressive tax reporting. While benefits are expected to increase consistently with client fees, perceived costs and risks are expected to rise acutely at high levels of economic importance. Thus, we predict an inverted U-shaped response to economic importance such that tax professionals will make more aggressive recommendations to clients of moderate economic importance than to clients of low or high economic importance.

We draw on social identity theory (SIT) to predict that strong interpersonal relationships with tax clients lead to higher levels of client identification and that, the more strongly an individual identifies with another, the more that individual's attitudes and behaviors will be favorably biased toward the other person such that professionals will provide more aggressive recommendations for high identification clients than for low identification clients when client economic importance is low or moderate. However, when economic importance is high, we expect professionals to be more concerned about their own risks than about the social relationship with the client. Thus, we predict that client identification (social relationships) will not affect recommendations when economic importance is high.

We test our hypotheses in an experiment with 133 experienced tax professionals. Consistent with our hypotheses, we find that tax professionals more strongly recommend the aggressive position when the economic importance of the client is moderate than when it is either low or high. This pattern of results holds separately for both the low and the high client identification conditions. We also find that, when the economic importance of the client is either low or moderate, tax professionals provide more aggressive recommendations when client identification is high compared to when it is low. However, when the economic importance of the client is high, there is no significant difference in recommendations between the high and low client identification conditions.

This study provides new insights about how client economic importance affects the aggressiveness of tax professionals' recommendations and that client economic importance moderates the effect of client identification on tax professionals' recommendations. Thus, this study also contributes to the literature by identifying a boundary condition on client identification that has not been considered in prior accounting research.

2020