Is Tax Volatility Priced by Lenders in the Syndicated Loan Market?

Abstract

In this study, I consider the effects of tax risk from tax volatility on the pricing of syndicated debt. Tax volatility is an interesting feature in that managers have some discretion over the risks they take with their tax strategies, which, however, are often harder to monitor for outsiders than risks related to other business activities. Framing my predictions based on the theoretical model developed by Merton [1974], I hypothesize and find that tax volatility is incrementally informative to other priced risks suggesting that tax risks per se are relevant to lenders. Moreover, I find that the results are stronger when the loan contract does not include performance pricing provisions or other restrictions, such as capital expenditure covenants, that protect lenders. This evidence adds to knowledge about the real effects of tax risk.

Keywords:

• Tax volatility
• Debt contracting
• Tax risk
• Borrowing costs

Acknowledgement

I thank Martin Jacob (the editor) and two anonymous referees for their detailed suggestions, which greatly improved the paper. Furthermore, I appreciate the helpful comments from David Aboody, Brian Akins, Paul Asquith, Joshua Anderson, Kathleen Andries, Judson Caskey, John Core, Scott Dyreng, Josh Filzen, Henry Friedman, David Guenther, Michelle Hanlon, Bradford Hepfer, Jeff Hoopes, Jack Hughes, Thomas Kubick, Zawadi Lemayian, Rebecca Lester, Lilian Mills, Thomas Omer, Reining Petacchi, Morton Pincus, Leslie Robinson, Antoinette Schoar, Terry Shevlin, Nemit Shroff, Eric So, Erin Towery, Brett Trueman, Rodrigo Verdi, Joseph Weber, Ben Yost, and seminar participants at the AAA Annual Meeting, AAA Southeast Meeting, AAA Western Meeting, the Massachusetts Institute of Technology, the University of North Carolina Tax Symposium, and the University of Iowa Tax Readings Group.

Supplemental Data and Research Materials

Supplemental data for this article can be accessed accessed https://doi.org/10.1080/09638180.2018.1520641.

Notes

1 A conceptually related term used in the literature is ‘tax uncertainty’ (e.g., Hanlon et al., 2017; Jacob et al., 2017). For parsimony, I use tax risk throughout the paper.

2 Lenders generally view cash flows as the primary repayment source. To estimate cash flows they rely on past financial performance. (See e.g., Wells Fargo website: https://www.wellsfargo.com/biz/help/faqs/lending-decisions/.)

3 The McGuire et al. (2013) measure is calculated slightly differently as the standard deviation of annual cash ETRs for the five-year period scaled by the mean of annual cash ETRs over the same five-year period. Given the similarity of both measures, and for parsimony, I use the tax volatility measure as suggested by Guenther et al. (2017).

4 I use a five-year period because this matches the measurement period of Tax Volatility.

5 For example, my results are qualitatively unchanged if I require that Annual Cash ETR be at least 50% or, alternatively, 70%.

6 In the paper, I set missing special items equal to zero. My results are unchanged if I drop observations with missing special items.

7 The bias is conservative given that lenders are likely to be more sensitive to risks posed by future tax obligations for a sample of firms closer to financial distress (e.g., firms that have losses). For how tax decisions of loss firms affect debt contracting terms, see Saavedra and Hughes (2017).

8 I stop in 2012, because this is the last year for which WRDS provides data to link Compustat to Dealscan. Further, in order to avoid the confounding effects of the FAS 109 adoption, I also used a sample that didn’t begin until 1995. Using this smaller sample, I find similar results to the ones presented in the paper.

9 Previous research suggests that there is a link between financial reporting aggressiveness and tax reporting aggressiveness (Frank, Lynch, & Rego, 2009). As a result, it is possible that tax volatility and aggressive financial reporting behavior are associated, in which case financial misreporting might be a correlated omitted variable in my tests. To address this concern, I drop all firms that experienced a restatement. In untabulated results, I continue to find that lenders penalize firms with higher tax volatilities.

10 In the online appendix, I provide robustness tests using alternative denominators such as total assets or market value of assets.

11 Consistent with the suggestion in Angrist and Pischke (2009), I use a linear probability model as opposed to a non-linear limited dependent variable model. This allows for the easy interpretation of the coefficients as well as the use of fixed effects in the model.

12 For instance, Fahlenbrach et al. (2012) provide evidence that past bank performance is predictive of future bank performance. Banks that performed poorly in the 1998 crisis were more likely to perform poorly in the 2008 crisis. This result is also consistent with concurrent findings in Drake et al. (2017) that past tax volatility predicts future tax volatility.

13 Taking the logarithm of Loan Spread does not affect my main results.

14 For loans not based on LIBOR, Dealscan converts the spread into LIBOR terms by adding or subtracting a differential that is adjusted periodically.

15 In robustness tests, I use unrecognized tax benefits (FIN 48) as an alternative measure of tax risk. The results are presented in the online appendix.

16 Results are similar using alternative specifications such as dummies for low ETR firms or annual cash ETRs. Moreover, in the online appendix, I present results when controlling for book-tax differences.

17 In the online appendix, I also include earnings volatility as an additional control. My results are unchanged. I do not use cash flow volatility and earnings volatility together in the paper because they are highly collinear. I keep cash flow volatility because it is the more common control in debt contracting papers.

18 I use a modified Altman (1968) Z-score as in Graham et al. (2008). In particular, Z-score=1.2 (Working Capital/ Total Assets) + 1.4 (Retained Earnings/Total Assets) + 3.3 (EBIT/Total Assets) + (Sales/Total Assets). Firms with lower Z-scores have a higher probability of default.

19 To ensure that I use only accounting information that is publicly available at the time of a loan, I employ the following procedure (see, e.g., Bharath, Dahiya, Saunders, & Srinivasan, 2007): for those loans made in calendar year t, if the loan activation date is four months or later than the fiscal year ending month in calendar year t, I use the data from that fiscal year. If the loan activation date is less than four months after the fiscal year ending month, I use the data from the fiscal year ending in calendar year t−1.

20 For instance, in a recent credit agreement between lenders and Amazon, the loan contract stipulates that Amazon warrants that

The Company has filed all Federal income tax returns, and the Company and its Subsidiaries have filed all other material tax returns and reports required to be filed, and have paid all Federal and other material Taxes levied or imposed upon them or their properties, income or assets otherwise due and payable. (Amazon 10Q, June 2016)

21 In a third test (untabulated), I find that large tax payments or tax volatility associated with tax settlements disclosed in financial statements lead to particularly high interest rates in the syndicated loan market. In contrast, I find little evidence that lenders penalize large tax payments due to tax deferral strategies or repatriation of foreign earnings.

22 In this context, it is worth mentioning that (as shown in Table 2), few low ETR firms have a prior large tax payment, which might reduce the power of this test.

Additional information

Funding

I gratefully acknowledge the financial support of the MIT Sloan School of Management, the UCLA Anderson School of Management and the Deloitte Foundation.