GSH Financial mark GSH Financial
(431) 305-0352 Book a Consultation
GSH FinancialGSH Financial

Home / Financial Tools / Benford's Law Forensic Digit-Distribution Analyzer

Graduate-Level Modeling · Forensic Accounting

Benford's Law Forensic Digit-Distribution Analyzer

Real, unmanipulated transaction populations follow a predictable, logarithmic leading-digit pattern. This tool tests whether yours actually does, and flags exactly which digits and how badly it deviates.

How To Use This Model

Reading This Tool

Paste a column of transaction amounts, or load one of the two sample ledgers, then choose whether to test the first digit or the second digit.

The tool strips every number down to its leading digit(s), builds the observed frequency distribution, and compares it against the frequency Benford's Law predicts for naturally occurring numbers, scoring the fit with Nigrini's Mean Absolute Deviation (MAD) conformity bands.

Dataset

Only the leading digit(s) of each number's magnitude matter, decimal points, currency symbols and commas are ignored automatically. Numbers with a magnitude under 1 are excluded, since Benford's Law is defined on the significant digits of a number, not on its scale.

Benford Conformity

-

Observations Tested

0

Mean Absolute Deviation

-

Chi-Square Statistic

-

Observed vs. Benford-Expected Digit Frequency

ObservedBenford Expected

-

-

What The MAD Score Is Actually Telling You

-

Why MAD, Not Just Chi-Square

Chi-square grows mechanically with sample size, a large enough ledger will eventually flag statistically significant deviations even from genuinely clean data, simply because the test has more power to detect trivial differences. Nigrini's MAD bands are sample-size independent, which is why forensic accountants lean on MAD for the conformity call and treat chi-square as a secondary check.

What This Test Can't Tell You

Nonconformity is a reason to look closer, not a finding of fraud. Legitimate datasets with a narrow range (rents in one building, salaries in one pay band) routinely fail Benford tests because they lack the multi-order-of-magnitude spread the law assumes. Conversely, a sophisticated manipulator who understands Benford's Law can fabricate numbers that pass it. Treat this as a triage screen that tells you where to sample next, not a verdict.

MAD Conformity Meter (Nigrini Bands)

Per-Digit Significance Test

DigitObservedExpectedz-StatisticFlag

A digit is flagged when |z| exceeds 1.96, the standard 5% two-tailed threshold, meaning its observed frequency is unlikely to have arisen by chance if the true distribution really were Benford's Law.

Forensic Accounting & Fraud Detection

The Core Formulas

P(d1=d) = log10(1 + 1/d), d = 1..9
P(d2=d) = Σk=1..9 log10(1 + 1/(10k+d)), d = 0..9
MAD = (1/K) Σ |Observedi − Expectedi|
χ² = Σ (Observedi − Expectedi)² / Expectedi

Benford's Law predicts that in many naturally occurring collections of numbers, the leading digit is not uniformly distributed, small leading digits occur far more often than large ones, because the digits emerge from numbers spanning multiple orders of magnitude on a logarithmic scale.

Nigrini's MAD Conformity Bands (First-Digit Test)

  • 0.000 – 0.006: Close conformity.
  • 0.006 – 0.012: Acceptable conformity.
  • 0.012 – 0.015: Marginally acceptable conformity, worth a closer look.
  • Above 0.015: Nonconformity, the strongest signal to sample this population further.

When To Actually Use This Test

  • Screening a general ledger, disbursements file, or expense report population before selecting an audit sample.
  • Teaching the digit-frequency logic behind forensic accounting and fraud risk analytics courses.
  • Flagging populations for follow-up procedures like duplicate-payment testing or vendor master file review.
  • Spotting threshold-avoidance patterns, transactions clustered just under an approval or reporting limit.

Key Assumptions & Limitations

  • Works best on populations spanning several orders of magnitude, at least a few hundred observations, without an artificial minimum or maximum.
  • Assigned or sequential numbers (invoice numbers, cheque numbers, zip codes) are not expected to follow Benford's Law and should never be tested with it.
  • A nonconforming result is evidence for further sampling, not proof of manipulation, and a conforming result does not clear a population of risk.

Foundational References

Benford, F. (1938). The Law of Anomalous Numbers. Proceedings of the American Philosophical Society, 78(4), 551-572.

Nigrini, M. J. (2012). Benford's Law: Applications for Forensic Accounting, Auditing, and Fraud Detection. John Wiley & Sons.

Want an earnings-manipulation screen too?

Open The Beneish M-Score Tool →