The Information Content of Deferred Taxes Under IFRS

Abstract

As prior evidence on deferred taxes under IFRS is scarce and inconclusive, this study examines whether deferred taxes reported in a German conservative accounting environment have information content. To improve informativeness, I propose a new split into four deferred tax categories. Applying this split, I replicate prior literature’s value relevance tests. Next, I examine whether the (lack of) value relevance can be explained by predictive ability. My findings indicate that aggregate deferred taxes are value relevant and that the new split informs about the drivers of this value relevance. It further helps to disentangle value relevance and predictive ability which do not always go hand in hand. Taken together, my results suggest that applying the new categorization improves the information content of deferred taxes.

Keywords:

• Deferred taxes
• IAS 12
• Predictive ability
• Value relevance

JEL Classification:

• M41
• M48

Acknowledgements

I sincerely thank Martin Jacob (associate editor) and two anonymous reviewers for their guidance and constructive comments. This paper has also benefitted from comments of Bianca Beyer, Jeffrey Hoopes, Urška Kosi, Thomas Kourouxous, Edward Maydew, Jens Müller, Regina Ortmann, Sönke Sievers, seminar participants at the Kenan-Flagler Business School at the University of North Carolina, conference participants at the 2017 European Accounting Association conference as well as participants of the TAF Brown Bag and TAF Research Workshop at the University of Paderborn. I also thank Regina Müller and Louisa Frischmuth for their excellent research assistance. Part of the research was conducted when Vanessa Flagmeier was affiliated with the University of Paderborn and while she was visiting the Kenan-Flagler Business School at the University of North Carolina.

Supplemental Data and Research Materials

Supplemental data for this article can be accessed on the Taylor & Francis website, doi:10.1080/09638180.2020.1826338.

Online Appendix 1. Deferred Tax Category Illustration.

Online Appendix 2. Reliability Test Details.

Online Appendix 3. Deferred Taxes under IFRS and US GAAP.

Online Appendix 4. Correlation Matrix.

Notes

1 The International Accounting Standards Board (IASB) decided not to implement any fundamental revisions of International Accounting Standard (IAS) 12 but frequently issues smaller amendments, for example an Exposure Draft on deferred taxes related to assets and liabilities arising from a single transaction (IASB, 2019).

2 Laux (2013) uses the categories ‘tax first’ and ‘GAAP first’. As this is an IFRS study, I replace the term ‘GAAP first’ by ‘financial statement (FS) first’.

3 Laux (2013) codes deferred tax assets (liabilities) as positive (negative) numbers and examines the net amount in the categories ‘FS first’ and ‘tax first’.

4 Conceptually, the recognition of deferred taxes under US GAAP and IFRS is similar. The main difference is the recognition of a valuation allowance under US GAAP. See Online Appendix 3 for details on the differences.

5 Online Appendix 3 provides details, particularly on the US-GAAP valuation allowance which supplements deferred taxes with valuable information (Dhaliwal et al., 2013) and does not exist under IFRS.

6 These deferred taxes are subsumed in the category OTHER. See Section 5.1 for details.

7 Due to the US-GAAP setting, ‘FS first’ is labeled ‘GAAP first’ in Laux (2013).

8 Cheung et al. (1997) use aggregate deferred taxes which represent net balance sheet numbers, making it hard to draw inferences about the relation between deferred taxes and future tax payments (and cash flows) from their findings. Furthermore, the study is criticized for having a number of shortcomings (Chludek, 2011a, p. 6).

9 Two recent studies focus on other tax items but include deferred taxes as control variables in their future income tax cash outflow models. Ciconte et al. (2016) analyze the relation between unrecognized tax benefits and future tax payments in their main tests while Hanlon et al. (2017) test the association between tax uncertainty and future taxes in a supplemental test. Both studies include the control variables net deferred taxes and net operating loss carryforwards in their tests and generally find negative and significant (insignificant) coefficients for the net operating loss carryforwards (net deferred taxes). The insignificant deferred tax coefficients could possibly be explained by not considering the distinction between ‘FS first’ and ‘tax first’ suggested by Laux (2013).

10 Dreher et al. (2017) analyze deferred taxes in an IFRS setting. However, they focus on the recognized and unrecognized deferred taxes for TLC and examine whether these items provide information about future earnings and cash flows.

11 My sample (short period and unbalanced sample) does not have enough within-firm variation to include firm fixed effects. Generally, Petersen (2009) recommends using standard errors clustered by firm and year for panel data. However, for small samples with few clusters in one of the dimensions, standard errors clustered by both firm and year can be biased (Petersen, 2009; Thompson, 2011). As my data has only six sample years, I do not use two-way clustered standard errors in my main tests. Repeating the main tests with standard errors clustered by firm and year results in only minor changes regarding the significance of coefficients and does not affect inferences.

12 Following Chludek (2011b), I estimate expected normal operating earnings as 12 percent of the book value of equity less net financial assets.

13 Examples for the categories can be found in Online Appendix 1.

14 As a specification check, I repeat my tests by using the sum of TXC (TXPD) for the period t+1 and t+2 as dependent variables. Results are unchanged regarding sign and significance while the magnitude of the coefficients roughly doubles.

15 Prior literature indicates that it is likely that deferred taxes are informative in less conservative settings such as the UK and Australia (Badenhorst & Ferreira, 2016).

16 See https://www.bundesanzeiger.de.

17 While I compare my value relevance analysis to Chludek (2011b), who uses a different sample period (2005 to 2008), I intentionally choose this period to test whether her findings of non-value relevance also hold in a period that is not biased by financial crisis effects.

18 The negative coefficient of TAXFIRSTDTL and the positive coefficient of TAXFIRST are consistent because TAXFIRST is a net amount, calculated as TAXFIRSTDTA minus TAXFIRSTDTL. Hence, TAXFIRSTDTL has generally a positive sign but is signed negatively as part of TAXFIRST.

19 This finding is in line with recent evidence by Dreher et al. (2017), documenting that deferred tax assets for TLC do not improve performance predictions although they are correlated with future earnings and cash flows.