Does Divergence from Benford's Distribution Signal Earnings Management?

Abstract

In this paper, we have analyzed the validity of the Financial Statement Divergence (FSD) score—a measure of the deviation of financial statement numbers from Benford’s Distribution that has gained popularity as a measure of financial information reliability. Our theoretical discussion addresses the fact that FSD scores are not monotonically related to the number or materiality of accounting manipulations, as confirmed by our empirical results using artificial manipulations. Moreover, even after we induced manipulations of extreme materiality, only a small proportion of observations diverged significantly from Benford’s Distribution, showing the extremely low power of the FSD score for detecting accounting manipulations. We also document contradicting results of the FSD score when the firm-year results are compared to the line-item results. Taken together, our results cast serious doubt on the validity of the FSD score for assessing financial information reliability.

Keywords: Financial information reliability, Benford’s Law, earnings management

JEL Classification: M41, M42