Lesson 9 of 9
The Beneish M-Score, used carefully
Compute the eight index values behind the score, and see why fast real growth alone can push an honest company above the cutoff.
By Javier Sanz
Founder of ValueScreener. Built Ninety Nine, a securities services provider, and sold it to a tier-one global bank; ran operations and internationalization at Alpaca, the brokerage API.
Who writes thisEditorial processLinkedIn (opens in a new tab)X (opens in a new tab)
In one minute
- The M-Score combines eight ratios, seven of them this year against last, into one number, with weights fitted on one sample of known manipulators and industry-matched companies.
- Sales growth raises one of the eight, and receivables that outgrow sales raise another, with or without manipulation.
- A score above the published cutoff, minus 1.78, is a reason to open the filing, not a finding on its own.
The Beneish M-Score is eight financial ratios, seven of them compared with the year before, combined with published weights into one number. A score above minus 1.78 is the model's flag for a possible manipulator. Two of the ratios can also rise in an honest company: one with fast sales growth, one when receivables grow faster than sales.
The seven indices behind one company's score
- DSRI, receivables against sales · receivables doubled as a share of sales2.00
- SGI, sales growth · sales grew 50 percent1.50
- GMI, gross margin1.00
- AQI, asset quality1.00
- DEPI, depreciation rate1.00
- SGAI, SGA expense1.00
- LVGI, leverage1.00
No change from last year: 1.0
What the score adds up
Beneish published the M-Score in 1999, fitted on a sample of companies later found to have manipulated earnings, matched against companies that had not. Seven ratios compare this year with last. The eighth, TATA, is this year's accruals as a share of total assets. A weight fixed by the original fitting multiplies each ratio, and an intercept of minus 4.84 is added to the sum.
| Input | What it compares | Rises when |
|---|---|---|
| DSRI | Receivables ÷ sales, this year against last | Receivables outgrow sales |
| SGI | Sales, this year ÷ last year | Sales grow |
| GMI | Gross margin, last year ÷ this year | Gross margin shrinks |
| AQI | Share of assets that are neither current assets nor plant, this year against last | More cost sits in soft assets |
| DEPI | Depreciation rate, last year ÷ this year | Depreciation slows |
| SGAI | Selling and administrative cost ÷ sales, this year against last | Overhead outgrows sales |
| LVGI | Current liabilities plus long-term debt ÷ total assets, this year against last | Leverage rises |
| TATA | Working-capital accruals after depreciation ÷ total assets, this year only | Profit runs ahead of cash |
The published cutoff is minus 1.78. A score above it flags a possible manipulator, and a score below it does not. The cutoff was chosen to keep the expected cost of errors low, not to mark a likely case. It came out of the same fitted sample as the weights, so it carries the restriction the earlier lesson in this course describes: it travels only as far as that sample does.
Two ratios growth can move
DSRI compares receivables as a share of sales, this year against last, so it rises only when receivables grow faster than sales. Honest causes exist. A company may give longer credit terms to win customers, or book more of its sales late in the year. An acquisition near year end adds the target's receivables but only weeks of its sales, and its goodwill lifts AQI too. SGI is sales this year divided by sales last year, and its weight, 0.892, is one of the largest in the model. A company growing sales fast scores higher on SGI for the plain reason that it is growing fast.
Beneish's own sample found growth mattered: rising receivables and rising sales were among the strongest predictors of a later manipulation finding. A later study of 2,190 SEC enforcement releases, covering 677 firms, also found growth in receivables and revenues mattered. It warned that its own model flags many false positives, because misstatement is rare. A fast-growing, entirely honest company sits close to a fast-growing dishonest one on these two ratios alone.
Reading the score, not the zone
Run the eight ratios and read the number as a rank against a group you chose, not as a fixed zone. A score fitted on one sample carries that sample's mix of industries and outcomes, and a fast grower crossing minus 1.78 is one number, not a finding. Treat a high M-Score as a reason to open the filing and check what actually grew, not as a verdict already reached.
How many US companies cross minus 1.78
Beneish's cutoff, minus 1.78: between the 75th percentile (−2.21) and the 90th percentile (−1.07).As a floor it admits between a tenth and a quarter of the cohort.
- 10th percentile: −3.29
- 25th percentile: −2.82
- median: −2.54
- 75th percentile: −2.21
- 90th percentile: −1.07
Between a tenth and a quarter of US companies crossed it. Caught misstatement is far rarer, so most flags are false alarms.
Check yourself
A company's receivables grew faster than its sales. Which index does this move most directly?
- DSRI
- LVGI
- DEPI
Show the answer
Answer: A.
DSRI compares receivables as a share of sales, this year against last, and it rises only when receivables grow faster than sales.A company's M-Score is minus 1.20, above the published cutoff of minus 1.78. What does that number mean on its own?
- A finding that earnings were manipulated
- A reason to open the filing and check which ratios moved
- A guarantee that the company will restate its results
Show the answer
Answer: B.
The cutoff came from a fitted sample and flags a pattern, not a conclusion. It tells you where to look, not what you will find.Two companies have identical inputs except two: Company A's sales grew 50 percent and its receivables, as a share of sales, doubled. Company B had no change in either. Which scores higher?
- Company A
- Company B
- They are equal, since neither company's accruals changed
Show the answer
Answer: A.
SGI and DSRI both carry positive weights, so faster sales and a larger receivables share push the score up regardless of the other six inputs.Sources
- Beneish, Financial Analysts Journal, 1999, "The Detection of Earnings Manipulation": the eight indices, the published weights and the cutoff of minus 1.78.
- Dechow, Ge, Larson and Sloan, Contemporary Accounting Research, 2011, "Predicting Material Accounting Misstatements": growth in receivables and revenues as predictors, from 2,190 SEC enforcement releases covering 677 firms, and the false positives a model of a rare event produces.
What this leaves out
- The figure and the worked example use round invented numbers for one hypothetical company. They are not any company's filed figures.
- The eight weights and the minus 1.78 cutoff come from Beneish's own fitted sample and were not re-estimated on today's market. ValueScreener withholds the score for banks, capital markets firms, equity REITs, insurers and mortgage REITs.
- A score above the cutoff is not a finding, and this lesson draws none from the illustrative company above.
- Nothing here is advice about any security.