Journal Entry Testing with a Benford Threshold
An audit senior runs a Benford's Law test over 180,000 manual journal entries for a manufacturing client. The first-digit distribution shows a significant deviation (chi-square p < 0.01), driven mainly by an excess of entries beginning with '9' and a deficit beginning with '1'. The senior drafts a finding that manual journals show 'evidence of manipulation'. The audit manager notes that: (a) the client posts a standard month-end accrual reversal batch of about 12,000 entries, all between $9,000 and $9,999, generated by the ERP; (b) the population includes credit notes and reversals as negative amounts, which the test took the absolute value of; (c) 60% of entries are below $500 and are auto-generated by a workflow tool. The engagement partner asks whether the finding should go to the audit committee.
1. Assess whether the Benford result supports the draft finding. What should be done before any conclusion is drawn?
Read the worked answer
The result as drafted does not support the finding. Benford's Law applies to naturally occurring, unconstrained, same-scale numeric populations. This population violates that in at least three ways:
- The $9,000-$9,999 batch is a designed, bounded range. It will mechanically inflate the '9' digit. It is not manipulation; it is a control design choice. These entries should be excluded or tested separately.
- Absolute values of credits and reversals destroy the sign structure. A reversal is by definition the mirror of an original; mixing them distorts digit frequencies and can create spurious excesses. Sign should be preserved and debits/credits tested separately.
- Auto-generated sub-$500 entries are system output with constrained formats, not human-entered amounts. They dilute the population and can mask or manufacture deviations.
Before concluding, the senior should: define the population precisely (manual, human-entered, positive, above a materiality-linked floor), remove the known structured batches, re-run the test, and - critically - treat Benford as a screening tool, not evidence. A deviation identifies a population to investigate; it never proves manipulation. The next step is to drill into the excess '9' entries that remain after exclusions and test them against expectations.
2. The partner asks whether the finding should go to the audit committee. Advise, and describe what the communication should contain if anything is reported.
Read the worked answer
Not as drafted. Reporting 'evidence of manipulation' to the audit committee on an unadjusted Benford result would be a serious professional error: it asserts a conclusion the evidence does not support, could damage the client relationship unfairly, and may mislead those charged with governance.
What should happen first: re-run the test on a properly defined population. If, after removing the structured batch and separating signs, a deviation persists and drill-down identifies specific unexplained entries, then there is a matter to communicate.
If a matter remains, the communication should contain:
- The population tested and the exclusions, with reasons.
- The method (Benford first-digit, chi-square) and its limitations - it is a screening technique, not proof.
- The specific entries or patterns identified on drill-down, with amounts and account impact.
- Management's explanation and the auditor's assessment of it.
- The effect on the audit approach (extended testing, adjusted risk assessment).
If the re-run is clean, no audit committee report is needed; document the work and the exclusions in the file so a reviewer can see why the initial result was not pursued. Either way, the senior's draft should be rewritten to describe a screening outcome, not a finding of manipulation.