Estimating the proportion of misstated records in an audit data set using Benford’s law

Abstract

Auditors are required to provide high levels of assurance that financial statements are free of material misstatements. This paper contributes to the literature on the field of audit sampling, by proposing a procedure to estimate the proportion of misstated records in a numerical audit data set based on stratified sampling, which can also be of assistance in financial fraud detection. Stratification rules based on the expected profile of misstated records and on Benford's law are evaluated and compared through an empirical experiment. The results show that: 1) the examined stratification rules perform significantly better than a simple random sampling approach; 2) when using Benford’s law, combining it with other methods does not seem to improve the performance of the estimation. The proposed procedure can be embedded in an audit software and contribute to enhance the effectiveness of audits and fraud detection.