Data snooping and the global accrual anomaly

Abstract

Naïvely testing for accruals mispricing in 26 equity markets – one market at a time – we find statistical evidence of anomalous returns in some countries. However, some of these findings might well be spurious because of data snooping biases that arise when simultaneously testing several hypotheses. While the accrual anomaly is not deemed to be robust in some countries when properly accounting for multiple testing, we find the international momentum effect to by and large pass the battery of multiple testing procedures. Moreover, we find the few robust accrual anomalies vanishing in recent times, indicating that investors have been exploiting the mispricing.

Keywords:

• accrual anomaly
• market efficiency
• multiple hypotheses testing
• momentum effect

JEL Classification:

• C12
• G14
• M41

Acknowledgements

We are grateful to Rajiv Banker, Andrea Buraschi, Jean-Marie Dufour, Florian Esterer, Andreas Gintschel, Christopher Hennessy, Bendik Höhn, Soeren Hvidkjaer, Robert Kosowski, Marcel Marekwica, Thorsten Neumann, Bernd Scherer, Clemens Sialm, Sandra Sizer, Richard Taffler, Rudolf Volkart, Michael Wolf, Liuren Wu, Josef Zechner, Guofu Zhou, and the seminar participants at the 2007 European Finance Association Meeting in Ljubljana, the 2007 Meeting of the Econometric Society in Budapest, and the 2007 FMA European Conference in Barcelona for helpful comments and suggestions. Note that this article expresses the authors' views that do not have to coincide with those of Deka Investment. Part of this research has been carried out within the Swiss Finance Institute and the National Centre of Competence in Research ‘Financial Valuation and Risk Management’ (NCCR FINRISK). The NCCR FINRISK is a research program supported by the Swiss National Science Foundation. Markus Leippold gratefully acknowledges the financial support by Bank Vontobel.

Notes

¹ Note that we do not investigate whether potential abnormal returns of this strategy are more likely due to market misweighting of operating cash flows as compared to misweighting of accruals. Hence, any result that we are attributing to the accrual anomaly might reflect two anomalies, i.e. accrual and operating cash flow anomaly. An examination of this possibility is beyond the scope of this study. Also, distinguishing between both anomalies has typically been accomplished by means of the Mishkin Test which has been recently demonstrated to be flawed (Kraft et al., 2007).

² These results can be obtained upon request.

³ To obtain these correlation figures, we have recomputed our study's αs using the respective sample periods of Lafond (2005) and Pincus et al. (2007). Thus, the remaining differences are most likely due to differences in sample design and employed risk controls.

⁴ Since we do not have access to the datatype ΔOTH _cf , this item is not included in our computation of ACC _cf .

⁵ ILLIQ is the absolute daily return divided by the associated dollar volume, thus capturing price impact. To obtain an aggregate monthly value of ILLIQ, we simply compute its mean over the corresponding daily values.

⁶ That is, estimates the SD of w _T,s .

⁷Actually, controlling the FWE corresponds to controlling the FDP with the extreme choice γ = 0.

⁸ Using the stationary bootstrap with an average block 6 of six months leaves results virtually unchanged.

⁹ For the US markets see, e.g. Jegadeesh and Titman (1993, 2001), and for a study for the UK market, see, e.g. Siganos (2007). For international studies see, e.g., Rouwenhorst (1998), Griffin et al. (2003, 2005) and Leippold and Lohre (2011). Also, Schwert (2003) and Fama and French (2008) argue that momentum is among the few robust anomalies. However, for a conflicting conclusion, see Pan and Hsueh (2007).

¹⁰ See, e.g. Alford et al. (1993) and Ali and Hwang (2000).