What the M-score measures
Messod Beneish published the model in 1999. He compared companies identified through SEC enforcement actions and press reports as having manipulated their earnings with other companies, and found eight measures that had moved differently in the year the manipulation began. The M-score combines them, each with its own weight.
Most indices divide this year’s ratio by last year’s, so a value near 1 means little changed. The score needs two years of figures.
The eight indices
- DSRI, days sales in receivables. Receivables relative to sales, against last year. A sharp rise can mean sales recorded before customers are really committed.
- GMI, gross margin. Last year’s gross margin divided by this year’s, so it rises when margins fall. Falling margins can put pressure on managers.
- AQI, asset quality. The share of assets other than current assets, property and equipment, and securities: mostly intangibles and deferred costs. A rise can mean more costs are being capitalized.
- SGI, sales growth. Growth is not manipulation, but fast-growing companies face more pressure to keep growing.
- DEPI, depreciation. Last year’s depreciation rate divided by this year’s; above 1 when assets are being depreciated more slowly.
- SGAI, selling, general and administrative expenses. SG&A relative to sales, against last year. It carries a negative weight in the model.
- TATA, total accruals to total assets. Income from continuing operations minus cash flow from operations, relative to total assets: earnings not yet backed by cash.
- LVGI, leverage. Debt relative to assets, against last year. It also carries a negative weight.
Reading the score
Beneish treated scores above −1.78 as a sign of possible manipulation. A score above that cut-off means the statements look more like those of the manipulators in his sample; it does not show that anything improper happened. Fast growth, an acquisition or a change in the business can push several indices up at once.
Some sources quote a cut-off of −2.22. The −1.78 figure is the one Beneish’s 1999 paper gives for this eight-variable model, and the one the calculator uses.
What the M-score does not show
- It is a screen, not evidence. Many companies above the cut-off have done nothing wrong, and some manipulators score below it.
- It was built on US public companies, and it is not meant for banks and insurers.
- It needs two years of comparable statements; a restatement or a large acquisition between them distorts most indices.
- Each index is a ratio of ratios, so a small figure in the prior year can produce a very large index.
Common mistakes
- Using net income including discontinued operations for TATA, instead of income from continuing operations.
- Mixing gross and net receivables between the two years.
- Reading the score as a finding of fraud rather than a reason to read the notes and the cash flow statement more closely.
Related measures
Two indices start from ratios with their own pages: days sales outstanding and gross profit margin. The Piotroski F-score includes a similar check on earnings against cash flow, and the Altman Z-score looks at the risk of financial distress.
To keep the calculation, download the Beneish M-score template for Excel or Google Sheets.