An exploration of two accounting-based models for earnings misstatements and their implications for stock returns

Abstract

Using two popular accounting-based models for earnings manipulation (i.e., the Beneish M-Score and the Dechow F-Score) and the financial data of public companies from 2004 to 2012, 1 find that the M-Score (F-Score) predicts less (more) earnings overstatements during the recent financial crisis in 2007-2008 than other sample years. However, a detailed investigation at the industry level reveals that this does not hold in all industries. I further show that the potential misstating firms flagged by the M-Score tend to under-perform the market both at the aggregate and the industry level, and some of those flagged by the F-Score under-perform at the industry level. Finally, by running Fama-French three-factor regressions at the aggregate level, I provide evidence that the firms flagged by the MScore generally yield negative risk-adjusted stock returns. The evidence suggests public availability of financial statements alone does not ensure that all the elements of financial statements are fully integrated into prices in a timely manner. Overall, this study provides substantial support for the use of quantitative accounting analysis in equity trading.