Beneish M-Score: Eight Ratios That Flag Possible Earnings Manipulation

A company can report a lovely jump in profit and still be telling you a half-truth. The Beneish M-score is one of the few tools built specifically to catch that. It takes eight ratios from two years of financial statements, weights them, and gives a single number that says whether the accounts look like those of companies that were later found to have manipulated earnings.

It doesn’t prove fraud. Nothing on a spreadsheet does. What it does is tell you where to start digging, and that is often the hardest part when a stock everyone loves is reporting results that look a little too smooth.

Who built the Beneish M-score, and why

The model comes from Professor M. Daniel Beneish of the Kelley School of Business at Indiana University. As Investopedia’s explainer on the Beneish model notes, he had worked on it for years before his paper, “The Detection of Earnings Manipulation”, was published in 1999, and he has written follow-up studies since.

The thinking behind it is fairly intuitive. Companies are more tempted to dress up profits when the business underneath is getting tougher: gross margins slipping, leverage creeping up, and sales growing fast enough that everyone expects the growth to continue. Pressure creates motive. The M-score tries to measure the symptoms of that pressure, plus the accounting footprints that manipulation tends to leave behind.

The eight ratios, one by one

Every variable compares this year (t) with last year (t-1). Most are built so that a value above 1 means “things moved in a direction that deserves a closer look”. The formulas below follow the Wikipedia entry on the Beneish M-score.

1. DSRI: Days’ sales in receivables index

(Receivables t / Sales t) divided by (Receivables t-1 / Sales t-1). If receivables are growing much faster than sales, either customers are paying slower or some of those “sales” may have been booked a bit early. This is the classic channel-stuffing tell.

2. GMI: Gross margin index

Last year’s gross margin divided by this year’s. Note the direction: it’s old over new, so a value above 1 means margins have shrunk. A business with worsening economics has a reason to look elsewhere for profit.

3. AQI: Asset quality index

This looks at the share of total assets that is not current assets, net plant and equipment or securities, and compares it year on year. That leftover bucket is where capitalised costs and other soft, hard-to-verify assets sit. If it’s swelling, the company may be pushing expenses onto the balance sheet instead of the profit and loss account.

4. SGI: Sales growth index

Simply sales this year divided by sales last year. Growth isn’t wrongdoing. But fast-growing companies face the most pressure to keep the streak going, and their financing often depends on it.

5. DEPI: Depreciation index

Last year’s depreciation rate (depreciation over net PP&E plus depreciation) divided by this year’s. A value above 1 means the company is depreciating its assets more slowly, perhaps by stretching useful lives. Slower depreciation quietly lifts reported profit.

6. SGAI: SG&A expense index

Selling, general and administrative expenses as a share of sales, this year over last. Rising overheads relative to revenue signal weaker efficiency.

7. LVGI: Leverage index

(Current liabilities plus long-term debt) over total assets, this year over last. Higher leverage brings debt covenants, and covenants give management a reason to keep ratios looking healthy.

8. TATA: Total accruals to total assets

(Income from continuing operations minus cash flow from operations) divided by total assets. This is the one we’d watch most closely. When profit races ahead of the cash actually coming in, the gap is made of accruals, which rest on management’s estimates.

Profit is an opinion built on estimates. Operating cash flow is much harder to argue with, and TATA measures the gap between the two.

The Beneish M-score formula and the -1.78 cutoff

The eight-variable version combines the ratios like this:

M = −4.84 + 0.92 × DSRI + 0.528 × GMI + 0.404 × AQI + 0.892 × SGI + 0.115 × DEPI − 0.172 × SGAI + 4.679 × TATA − 0.327 × LVGI

For this version, the threshold is −1.78. Both Investopedia and Wikipedia put it the same way: a score below −1.78 suggests the company is unlikely to be a manipulator, and a score above it suggests it is likely to be one. Because the scores are negative, this trips people up. −2.5 is the comfortable side. −1.5 is the worrying side.

Look at the weights for a second. TATA carries 4.679, DSRI 0.92 and SGI 0.892. Accruals, receivables and sales growth do most of the heavy lifting. SGAI and LVGI actually carry negative coefficients in the fitted model, so an increase in them pulls the score down slightly. That’s counter-intuitive given the story above, and it’s a good reminder that these are statistical weights fitted to a sample, not a moral ranking of red flags.

A worked example in rupee crore

Let’s run the numbers on a made-up mid-sized company, call it Company X. Every figure here is hypothetical and exists only to show the arithmetic.

On the surface this is a growth story. Sales up 40%, profit of ₹150 crore. Now the ratios.

• DSRI: receivables were 12% of sales last year (120/1,000) and about 18.6% this year (260/1,400). 0.1857 / 0.12 = 1.548.
• GMI: gross margin fell from 40% to about 35.7%. 0.40 / 0.357 = 1.120.
• AQI: the soft-asset share went from about 9.5% of total assets to 10%. 0.100 / 0.0952 = 1.050.
• SGI: 1,400 / 1,000 = 1.400.
• DEPI: the depreciation rate slipped from about 9.1% (50/550) to about 7.9% (48/608). 0.0909 / 0.0789 = 1.152.
• SGAI: SG&A fell from 15% of sales to 13%. 0.13 / 0.15 = 0.867.
• LVGI: leverage rose from about 42.9% of assets (450/1,050) to about 48.6% (680/1,400), giving 1.133.
• TATA: (150 − 40) / 1,400 = 0.079.

Plug them in:

−1.17 is well above −1.78. By the model’s rule, Company X’s accounts look like a manipulator’s.

What actually drove the number

Break the score into pieces and the story is obvious. The receivables jump alone contributes about 1.42. Sales growth adds another 1.25. Margin compression and the thinner depreciation rate pile on. Meanwhile, of ₹150 crore of reported profit, only ₹40 crore showed up as operating cash.

Now play with it. Suppose operating cash flow had been ₹140 crore instead of ₹40 crore. TATA falls to about 0.007 and the M-score improves to roughly −1.51. Still above the cutoff. Now also suppose receivables had grown in line with sales, to ₹168 crore, so DSRI is exactly 1. The score drops to about −2.01, comfortably on the safe side.

That little exercise tells you more than the final number does. In this example, receivables racing ahead of sales were the single biggest reason for the red flag. If you were analysing Company X for real, your next job would be to open the notes to accounts and the ageing schedule for trade receivables.

The final M-score is a headline. The breakdown by variable is the story, and it usually points straight at the note in the annual report you need to read.

The Enron story, and why people still quote it

The model’s most famous moment came in 1998, when a group of Cornell University business students used it and concluded that Enron was manipulating its earnings. According to Investopedia, Enron stock was trading at about $48 at the time, roughly half the $90 it later reached before collapsing into bankruptcy in 2001. Nobody on Wall Street paid much attention to the students’ warning, and Wikipedia notes that analysts were still recommending the stock then.

It’s a great anecdote, and it’s also a single case. A model that flags one famous fraud early is not the same as a model that is right most of the time. Keep both thoughts in your head.

Where the M-score falls short

• It’s probabilistic. The Wikipedia entry is explicit that it cannot detect manipulators with 100% accuracy. Plenty of honest high-growth companies will score above −1.78, and a careful fraud can slip under it.
• Not for banks and NBFC-style lenders. Financial institutions were excluded from Beneish’s sample because they earn money differently, and sales and receivables, two core inputs, don’t mean much for a lender. Don’t run it on a bank’s balance sheet.
• Growth companies get penalised. SGI carries a large weight. A company that grows 40% in a year starts with a handicap, even if that growth is real.
• Accounting mapping is messy. The formula was built on US-style statements. Mapping Ind AS line items, such as contract assets, right-of-use assets or capital work in progress, into the “current assets”, “PP&E” and “securities” buckets involves judgement calls that can move AQI and DEPI. Write down your choices and stay consistent across years and companies.
• One-off events distort it. A large acquisition can blow up receivables, assets and sales in one go and make the ratios meaningless for that year.

Interestingly, the idea has also been used at the level of the whole economy. Wikipedia cites a 2023 research paper that aggregated scores across many companies to try to predict US recessions, and found the aggregate score in early 2023 was the highest in some 40 years. That’s an academic finding about the US, not a trading signal, but it shows how far the concept has travelled.

How we’d use it on Indian results

1. Use annual audited numbers. The model compares full years. Quarterly figures are too noisy and many balance-sheet items are only published half-yearly.
2. Screen, don’t conclude. Run it across a watchlist and treat anything above −1.78 as “read the notes”, not “sell”.
3. Go straight to the biggest contributor. If DSRI is driving the score, look at receivables ageing. If TATA is, compare profit with operating cash flow over several years. Our guide on operating cash flow vs net profit walks through how to read that gap.
4. Cross-check with other scores. The Altman Z-score looks at distress rather than manipulation, and the Piotroski F-score checks whether fundamentals are improving. A company that looks bad on all three deserves real scepticism.
5. Read the governance disclosures too. Auditor changes, qualified opinions and heavy related-party dealings often travel alongside aggressive accounting. If you’re building the habit from scratch, start with how to analyse quarterly results of a company.

The short version

The Beneish M-score takes eight ratios covering receivables, margins, asset quality, growth, depreciation, overheads, leverage and accruals, and combines them into one number with −1.78 as the line. Above it, the accounts resemble those of past manipulators. Below it, they don’t. It is quick, uses public data and has a famous early call on Enron to its name. It also misfires on fast growers, doesn’t work for lenders and can’t replace reading the annual report. Treat it as a flashlight. It shows you which drawer to open.

Disclaimer: This article is for educational purposes only and is not investment advice. Company X and all figures in the worked example are hypothetical. Investing involves risk of loss; consult a SEBI-registered investment adviser before making investment decisions.