Percent accruals and the accrual anomaly: evidence from the UK

ABSTRACT

Using the percent accrual measure proposed by Hafzalla, N., Lundholm, R., & Van Winkle, M. (2011. Percent accruals. The Accounting Review, 86, 209–236), we empirically evaluate the predictions of earnings fixation hypothesis and limits to arbitrage hypothesis on the accrual anomaly in the U.K. stock market. We show a strong negative relation of percent accruals with future profitability and stock returns. We find that the effect of percent accruals on future earnings and stock price performance is stronger across loss firms relative to profit firms. Furthermore, we show that the effect of percent accruals on stock returns is more pronounced for micro stocks relative to small stocks, while it doesn’t occur across big stocks. Overall, we conclude that earnings fixation is a key factor in explaining the occurrence of the percent accrual anomaly, while limits to arbitrage are of great significance in explaining the persistence of the anomaly.

KEYWORDS:

• Percent accruals
• earnings
• stock returns
• earnings fixation
• limits to arbitrage

JEL Descriptors:

• G1
• M4

Acknowledgments

The author appreciates helpful comments from the seminar participants at the 39th European Accounting Association Annual Congress (2016), at the 42nd Annual Conference of the European International Business Academy (2016) at the University of Bern (2017) and at the University of Koc (2018). The author thanks Gikas Hardouvelis, Alexis Kunz and Dimitrios Thomakos and two anonymous reviewers for insightful comments and suggestions. The publication of this paper has been partly supported by the University of Piraeus Research Centre. The usual disclaimer applies.

Disclosure statement

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

Notes

1 By focusing on UK data, enables also us to make comparisons to the frequently examined US market (see also Wisniewski & Yekini, 2015).

2 According to the earnings fixation hypothesis, investors presumably do not appropriately adjust for the different mean-reversion tendencies of accruals and cash flows.

3 Several other studies report differences on return predictability attributable to working capital accruals and total accruals. In particular, Richardson et al. (2005), Dechow et al. (2008) and Hafzalla et al. (2011) among others, show larger accruals mispricing for U.S. firms, based on total accruals relative to working capital accruals.

4 As argued by IASB, managerial discretion following the adoption of IFRS should be used to increase the value relevance of accounting figures. However, managerial discretion under IFRS can be misused (e.g. negative earnings management, Fargher & Zhang, 2014). For instance, firm executives, may use discretion in timely recognition of losses (e.g., Siekkinen, 2016).

5 The return index recorded by Datastream for measurement of stock returns is slightly different before and after 1988.

6 Konstantinidi et al. (2016) consider proxies for losses based on the signs of cash flows, the change in cash flows and industry-adjusted cash flows, and the sign of abnormal returns. Our results remain qualitatively similar, if we use the proxies suggested by Konstantinidi et al. (2016) to separate profit from loss firms.

7 We use a broader measure of accruals that is equal to the difference between net income and free cash flows. After scaling this measure with absolute value of net income, we get percent total accruals. Hafzalla et al. (2011) consider in their study both percent operating accruals (i.e., the difference between difference between operating income and operating cash flows scaled by absolute value of net income) and percent total accruals and find qualitatively similar results for both measures. Kim et al. (2015) study the accrual anomaly in Korea by considering percent operating accruals, since prior research on the accrual anomaly in Korea focuses on operating accruals. Our results remain robust, if we instead consider percent operating accruals.

8 Since earnings can be near zero, percent accruals will sometimes have extreme values. Thus, in our regression analysis we consider scaled decile ranks of percent accruals. Further, in our portfolio analysis the extremity of percent accruals is irrelevant and will have no impact on the results.

9 The size of the sample, when we consider the combined measure of firms’ life cycle stage is reduced to 20,125 firm-year observations. Thus, after taking into account only the three LCS quintiles, we get a final sample that consists of 12,075 firm-year observations, or 4,025 at each life-cycle stage.

10 Our findings regarding the accrual effect on stock returns for profit and loss firms are in line with those reported by Patatoukas (2016), but in contrary with those reported by Konstantinidi et al. (2016). The differences may arise from the accrual measures used in the empirical tests. The empirical evidence reported in Konstantinidi et al. (2016) is based on working capital accruals, in Patatoukas (2016) on total accruals, while in our study on percent accruals. Working capital accruals are a narrow accrual measure relative to total accruals and percent accruals, due to the fact that they omit long-term accruals (see Dechow et al., 2008; Richardson et al., 2005). We need to stress here that the results reported by both Patatoukas (2016) and Konstantinidi et al. (2016) are applicable to UK firms.