Forensic Analytics, 2nd Edition

CHAPTER 3Benford's Law: The Basic Tests

BENFORD'S LAW (BENFORD) IS an interesting and counterintuitive statistical regularity that is a part of the main set of forensic accounting tests. Benford-based tests are effective at detecting abnormal duplications in our data. Benford gives us the expected frequencies of the digits in tabulated accounting data. Actual frequencies that deviate significantly from the norm point us towards a set of abnormal duplications. Bolton and Hand (2002) state that the statistical tools for fraud detection usually compare a single numeric value, or a set of observed values, to a set of expected values. This digit-based test is our fifth high-level overview test and is essentially a continuation of Chapter 2. Nonconformity to Benford, under certain circumstances, is an indicator of an increased risk of fraud or error. However, nonconformity to Benford does not signal fraud or error with certainty. More review work is always needed.

Our Benford-based tests start with the high-level first-two digits test. After that we drill deeper into our data searching for abnormal duplications of specific numbers and exact or near-duplicate records. Other related tests include the last-two digits test and the summation test. The digit-based cycle of tests is designed to point forensic accountants to (a) finding fraud (where the fraudster has excessively duplicated their actions), (b) finding errors (where the error is systematic), (c) finding outliers, (d) finding ...