Manipulation screen

What it measures¶

The Beneish M-score combines eight comparisons of this year's accounts with last year's into one number. Each comparison is an index: this year's ratio divided by last year's, so 1.0 means no change.

• money owed by customers, against sales;
• the gross margin (last year's over this year's, so a fading margin raises it);
• the share of assets that are neither current assets nor property and equipment;
• sales growth itself;
• the rate of depreciation;
• selling, general and administrative costs, against sales;
• total accruals (profit not backed by cash), against assets;
• leverage (liabilities against assets).

Each index is multiplied by a weight Messod Beneish estimated, the products are added to a constant, and the result is the M-score — a more negative score looks less like the companies in his study that manipulated their accounts. The check asks: is M below −2.22?

Why it is on the page¶

It is the best-known published screen for earnings manipulation, famous for flagging Enron's accounts before the collapse. A later study found it still the only one of the common models that gave investors a net benefit after the cost of its false alarms. Because the largest weight sits on accruals, its reading is read beside the accruals check rather than as a separate opinion on the same fact.

Thresholds, their source and the population they were fitted on¶

Beside a flagged score the evidence panel shows how much of it came from the sales-growth index, with a note: fast growers trip this screen often for innocent reasons, because growth itself raises the score.

• Source: Beneish (1999), "The detection of earnings manipulation"; −1.78 is his published line, −2.22 the more cautious line widely used since.
• Fitted on: 74 companies charged with manipulation by the US securities regulator, 1982–1992, against a matched sample. At −1.78 it missed 26% of the manipulators and wrongly flagged 13.8% of the others; out of its sample it flagged about 17% of company-years.
• Evidence grade: peer-reviewed and replicated.
• Basis: absolute.

A worked example¶

Apple, fiscal year 2025 (statements stored on 25 October 2026): M = −2.32, just below −2.22 — passed.

Honeywell International, the same year: no net property, plant and equipment line is stored for fiscal 2025, which the asset-quality index needs, so the score was not evaluable — net property, plant and equipment FY2025.

What “No verdict” means here¶

The score needs two consecutive years of eight groups of lines — receivables, gross profit, property and equipment, depreciation, selling costs, debt and more. When any is missing it is not evaluable, the first missing input is named, and the check leaves the area's count. A partial score is never computed: it would be a different number wearing the same name.

What the verdict is and is not¶

A flag is a reason to look closer, not a finding of manipulation: in Beneish's own study about one company in seven that had done nothing wrong scored above −1.78. A pass is not a certificate that the accounts are sound. It is not a forecast and not an instruction to trade the shares. The eight indices are shown on the Beneish M-Score row of the Fundamentals page.