Benchmark-driven earnings management: Empirical analysis of earnings management to avoid losses and earnings decreases

This thesis examines benchmark-driven earnings management from two distinct aspects. Firstly, the author examines whether companies included in the study have engaged in earnings management to avoid reporting losses and earnings decreases. Secondly, the author examines whether the incentives to avoid reporting losses and earnings decreases are increasing with the length of the preceding string of positive earnings and earnings increases. To examine the four distinct hypotheses formulated in accordance with the research objectives of this thesis, the author studied visually the scaled distributions of earnings and earnings change figures to determine whether a discontinuity in the underlying distributions could be found. Moreover, the visual inspection was complemented with the Burgstahler and Dichev (1997) test static to provide statistical evidence for the hypotheses of this study. The sample of this thesis included all earnings observations from the Compustat North America database for years 1991 to 2014. The results of this study showed both visual and statistical evidence to support the acceptance of three out of four hypotheses tested in this study. The results suggest that earnings have been managed both to avoid losses and earnings decreases. Moreover, the author was able to document that the incentives to avoid earnings decreases are increasing with the string of consecutive years of earnings increases. The empirical findings offered, however, no evidence that the incentives to avoid losses are increasing with the string of consecutive years of profits