Reassessing Benford’s Law, Sarbanes-Oxley & Dodd-Frank: A Non-Parametric Exploration of Financial Statement Numbers 1970-2013

• In Event: Forecasting, Accruals, and Benford's Law

Abstract

This trifold investigation contributes forensic accounting substance and statistical methodology in applying Benford’s Law to forty years of public company financial statement numbers. First, it uses non-parametric, generalized additive modelling to produce evidence that the Benford’s Law conformity of public company financial statement numbers has not noticeably increased since the implementation of the Sarbanes Oxley Act (2003) or the Dodd Frank Act (2010). This finding suggests possible failure of these regulatory initiatives to meet some of their intended objectives. Second, the paper presents evidence that Mean Absolute Deviation (MAD), a leading metric for testing conformity with Benford’s Law, is negatively correlated with sample size, raising questions regarding the optimal use of MAD as a screen for Benford’s conformity. Finally, addressing these questions, the paper uses Monte Carlo simulation to probe the Benford’s Law-MAD-sample-size relationship and proposes a MAD transformation, styled “Excess MAD,” as a more accurate and stable measure of conformity to Benford’s Law.

Authors

• Bradley J. Barney, Kennesaw State University

• Kurt S. Schulzke, Kennesaw State University