To Shift or Not To Shift? Intertemporal Income Shifting as a Response to the Risk Capital Allowance Introduction in Belgium

Abstract

This study examines how firms shift profits from one period to another in response to the introduction of an allowance for corporate equity. We focus on the introduction of the risk capital allowance in Belgium by the law of 22 June 2005. We predict and find that firms with relatively low (but positive) earnings in 2006 have incentives to defer profits in 2005 and that this effect is stronger for firms with higher equity ratios. Conversely, we predict that firms that are highly profitable in 2006 and have large net operating loss carryforwards in 2005 have incentives to accelerate profits. We find that only subsets of firms react to the latter incentive, in particular firms with higher equity ratios. Our findings show that tax-related benefits motivate firms to engage in conforming tax avoidance and provide evidence of cross-sectional variations in their reaction to these incentives. These insights contribute to the literature outlining the costs and benefits of changes in tax regimes and documenting earnings management in response to tax incentives in a high book-tax-conform environment.

JEL Classifications:

• M41 – Accounting
• M48 – Government Policy and Regulation
• H21 – Efficiency & Optimal Taxation
• H25 – Business Taxes and Subsidies
• H32 – Fiscal Policies and Behavior of Economic Agents – Firm

View correction statement:

Erratum

Acknowledgements

We thank the editor Richard Sansing, an anonymous referee, Kay Blaufus (discussant), Dhammika Dharmapala, Martin Jacob, Giorgia Maffini (discussant), Ed Maydew, Jim Seida, Robert Ullman, Alexandra Van den Abbeele, Christoph Watrin, Kelly Wentland, Marleen Willekens, seminar participants at Erasmus University Rotterdam, IESEG Business School, KU Leuven and WHU – Otto Beisheim School of Management, and attendees at the 4th EIASM Workshop on Current Research in Taxation, the 2015 ATA Midyear Meeting and the 1st Berlin-Vallendar Tax Conference for helpful comments and suggestions.

Supplemental Data and Research Materials

Supplemental data for this article can be accessed on the Taylor & Francis website, doi:10.1080/09638180.2016.1202854.

Notes

1 The RCA regulation is an ACE-type system. A detailed discussion of the RCA regulation in Belgium is provided in Section 2.1.

2 The RCA is applicable to: Belgian companies, Belgian branches of foreign companies, not-for-profit organizations subject to Belgian corporate income tax and foreign companies that own real estate located in Belgium. It excludes companies that already benefit from certain other beneficial tax rules (see Table 3 for more details).

3 The first year in which the RCA deduction was allowed is 2006. The allowed rate is based on the average interest rate on 10-year linear bonds issued by the Belgian government, for the year preceding the fiscal year (see Table 2 for more details).

4 The corrected equity base is calculated by reducing shareholder's equity at the beginning of the fiscal year with a number of items (see Table 2 for more details). Originally, firms were able to carry this surplus forward for up to 7 years if the RCA deduction exceeded the value of taxable profits. This carryforward clause has been removed in 2012.

5 The first country to implement an ACE system was Croatia in 1994, but Croatia abolished the regime in 2000 after deciding to implement a lower STR (Keen & King, 2002; Klemm, 2007). Italy introduced a partial ACE system that was in place from 1997 to 2003 and has introduced a new ACE-based system in 2011 (Klemm, 2007; Panteghini, Parisi, & Pighetti, 2012). Austria introduced an ACE system that was restricted to new equity in 2000 which was abolished in 2004 (Klemm, 2007). Outside Europe, only Brazil has applied an ACE system, which is still in place. This regulation was introduced in 1996 and is conditioned on payouts to shareholders (Klemm, 2007).

6 We acknowledge that it is impossible for firms to estimate their exact profitability of 2006 while finalizing the financial statements of 2005. However, as the final date allowed to file the financial statements with the NBB is 7 months after the book-year-end of 2005, we argue that most firms will have had a decent idea about the profitability of the first two quarters. This is even more so if they are experiencing extreme results in these quarters.

7 Because 31 December 2005 is the first possible year end date that is relevant for the calculation of the corrected equity and consequently, the RCA deduction in the following year, most (but not all) of the book years of interest in the pre-RCA year coincide with the 2005 calendar year. Henceforth, we refer to the pre-RCA year as 2005 and the first RCA year as 2006 to improve readability.

8 It is this incentive that has spurred prior intertemporal profit shifting research as deferring (accelerating) income for one period will lead to a considerable benefit in terms of the NPV of the tax expense equal to MTR_t− MTR_t+1/(1 + r) for each EUR deferred (accelerated) in the case of a STR decrease (increase) (Lin et al., 2013; Maydew, 1997).

9 In Belgium, the STR is a flat rate of 33.99%. However, a progressive rate (varying from 24.98% to 35.54%) is sustained for companies with a taxable income lower than EUR 322,500 that meet certain specific criteria (in terms of ownership, dividend payout, executive compensation, etc.). We use the flat rate throughout the sample for simplicity and expect similar inferences when using the progressive rate.

10 This discount rate is not subject to taxation, as we are using it to measure the NPV of the tax expense itself.

11 Mathematical proof of these predictions is outlined in an online appendix, which can be accessed as an online supplement at the journal's Taylor and Francis website.

12 Throughout the example, we assume that all earnings are retained. Note that 65% of our sample reports a payout ratio of 0, and nearly half of the firms that do pay dividends have a payout ratio below 50%. Hence, our assumption is quite realistic.

13 For the purpose of this study, we deviate from the traditional ROE definition and instead use corrected equity of the prior period (CE_t) (instead of common equity of the current or prior period) in the denominator.

14 Please note that our expectations would also be applicable to any other tax regulation change that introduces book-tax differences which lead to variations that affect the marginal tax rates of firms with differing characteristics. Our predictions mainly result from the possibility to have positive book income while having negative or zero taxable income.

15 Reports by privately held companies in Belgium do not include a cash flow statement. Consequently, the use of more direct accruals measures is unfeasible.

16 When estimating Equation (4) for the computation of the parameter coefficients of our size-based peer groups and our industry-based peer groups, we use the broadest yearly sample possible, composed of the firms included in the potential sample (Table 3). This implies that the sample used for the estimation of parameter coefficients α₁ and α₂ is larger than the final sample in our panel regressions.

17 Firms that fall below certain size requirements are not required to mandatorily report the extended financial reporting format that incorporates revenues. Because the estimation of discretionary accruals requires this item, the smallest firms are omitted from our sample.

18 While the equivalent figures for DCACC_SIC and DCACC_10Y show similar developments over time, we opt to include the figures that are representative of the highest number of observations.

19 Throughout the results section we always use the specification of HR_i,t+1 that is based on return on corrected equity (ROE_ADJ_i,t+1) because this measure is arguably the most accurate indication of the relevant level of profitability. In the robustness section, we deviate from this practice.

20 The significance of the different effects on discretionary accruals is tested by using a seemingly unrelated estimations technique that controls for covariances and allows the testing of differences in coefficients across nested samples.

21 Prior research has indicated that tax-related decisions are made based on the salience of corporate tax rates (Amberger, Eberhartinger, & Kasper, 2016; Graham, Hanlon, Shevlin, & Shroff, 2016).

22 Firms that filed within 6 months after book-year-end had a maximum of 5 months to prepare their financial statements. A filing date between 180 and 245 days after book-year-end indicates that this subsample of firms had considerably more time to optimize their profits in response to the RCA introduction while not being subject to retributions because of late filing (8 months after book-year-end). We refrain from incorporating firms that have a filing lag that is longer than 245 days because this could indicate that there were some irregularities: these are the firms that would have been subject to retribution for filing too late (even though untabulated results show that our results for H₂ improve when including firm-year observations with a filing lag of 245 days or more).

23 Ideally, we compare each firm to a historical record of observations from the same firm. We aim to do this by using a time-series estimation over 10 years. However, as this introduces a survivorship bias, we use well-accepted cross-sectional estimation procedures that aim to provide the most comparable benchmark for each estimation. Nevertheless, there will be differences in the comparability, but it is very unlikely that they are biasing the measures in favor of our hypotheses.