Tax Avoidance and Financial Statement Readability

Abstract

This paper examines whether managers of firms that engage in high levels of tax avoidance (TA) strategically reduce their financial statement readability (FSR) to mitigate the risk of exposing their TA strategies. On average, results are inconclusive, but mainly hold in the sample of firms with low effective tax rates (i.e., high TA levels). Specifically, focused on firms with above-industry-median TA, the panel regression results show a negative relation between TA and FSR, and the difference-in-differences analysis, based on the ‘Check-the-Box’ regulation in 1997 that exogenously increases tax planning opportunities, suggests the negative impact of TA on FSR is likely causal. The relationship is stronger for firms faced with a greater likelihood of tax audit ex ante, firms in industries with a higher tax risk, and firms with a larger institutional ownership. The evidence adds to our understanding of the influence of corporate TA on financial disclosures.

Keywords:

• Tax avoidance
• Financial statement readability
• Tax audit risk

JEL codes:

• G18
• G21
• H26
• M41

Acknowledgements

I greatly acknowledge helpful comments and suggestions from Martin Jacob (the Editor), the two anonymous reviewers, Henk Berkman, Steven Cahan, Iftekhar Hasan, Paul Healy, Ben Marshall, Jake Thornock, Terry Shevlin, Nam Tran, Brian Williams, Ryan Wilson, and conference participants at the 2019 American Accounting Association (AAA) Annual Meeting, and seminar participants at The University of Auckland, Massey University, and Victoria University of Wellington. I thank Brian Miller for sharing financial statement readability (the Bog index) data, Jeffrey Hoopes for sharing IRS audit probability and IRS attention data, Bill McDonald for sharing financial statement length (10-K file size and number of words) data, Feng Li for sharing Fog index data, Kerry Inger for sharing tax footnote readability data, and Daniel Taylor for share managers’ and analysts’ linguistic complexity data. An earlier version of the paper was circulated under the title ‘Does It Pay to Not Write Well? The Case of Tax Avoiders.’ All errors remain mine.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Notes

1 Firms may have incentives to obfuscate information about their TA activities provided to other stakeholders as well. For example, firms do not want investors or customers to view them as irresponsible corporate citizens that engage in noncompliant TA activities (Hanlon & Slemrod, 2009).

2 Prior studies find several potential costs of having less readable financial statements, including poorer information environment (Loughran & McDonald, 2014), higher cost of debts (Bonsall & Miller, 2017; Ertugrul et al., 2017), higher crash risk (Ertugrul et al., 2017), less efficient investments (Biddle et al., 2009), and lower firm value (Hwang & Kim, 2017).

3 In a concurrent study, Beuselinck et al. (2018) also find a negative relation between aggressive TA and FSR.

4 Please refer to Bonsall, Leone, et al. (2017) for more details on the Bog index construction. The Bog Index data can be found at: https://kelley.iu.edu/bpm/activities/bogindex.html. I thank Brian P. Miller for making the data available.

5 The data on financial statement length can be found at: https://sraf.nd.edu/textual-analysis/resources/. The data on Fog index can be found at: http://webuser.bus.umich.edu/feng/. I thank Bill McDonald and Feng Li for making the data available.

6 I thank Kerry Inger for providing me with the tax footnote readability data.

7 The results on TAXFOG become positive and statistically significant when I replace firm with industry fixed effects, which is consistent with Inger et al. (2018). My results suggest the positive relation between TA and TAXFOG may not hold within firm.

8 Please refer to Hoopes et al. (2012) for more details of the construction of the IRS audit probability. The related data can be found at: http://www.jeffreyhoopes.com/data.html. I thank Jeffrey L. Hoopes for making the data available.

9 In an additional analysis, I further classify institutional investors into dedicated and transient ones (Bushee, 1998, 2001) and find that the amplifying effect of institutional ownership on the negative association between TA and FSR among high-TA firms is even stronger for dedicated ones, consistent with the notion that these dedicated institutional investors are more long-term focused and hence more sensitive to tax audit risk.

10 I thank Daniel J. Taylor for providing me with the managers’ and analysts’ Fog index data.

11 The data on tax havens can be found at: https://sites.google.com/site/scottdyreng/Home/data-and-code/EX21-Dataset. I thank Scott D. Dyreng for making the data available.

12 In untabulated tests, I follow Frank et al. (2009) to adopt a measure of discretionary permanent differences, and follow Edwards et al. (2016) to employ a measure of deferral-based tax planning strategies, as alternative proxies for tax aggressiveness. The robustness check results are consistent with the main ones.

13 In an untabulated test, I redo the sensitivity check using TA quintiles and the results show statistically significant and positive coefficients for subgroups of the 4th and 5th quintiles while it is (weakly) negative for the subgroup of 1st quintile.

14 I find consistent results if I define treated firms as those with positive foreign income (pifo>0) as in Dyreng, Jacob, et al. (2019).

15 Note that the data on tax-footnote readability is not available for the DiD estimation period 1994–2000.