Earnings are an opinion; cash is a fact. The gap between them is where a company can flatter its results — booking revenue early, stretching out depreciation, letting receivables balloon. In 1999, an accounting professor named Messod Beneish asked a sharp question: if a company is quietly doing this, does it leave a fingerprint in its own numbers? He found that it usually does — and turned the fingerprint into a single score. It's called the Beneish M-Score, and this is how to read it without either ignoring it or over-trusting it.
What the score actually is
The M-Score is one number that answers one question: how much does this company's year look like the companies that were later caught manipulating earnings? It's built from eight ratios, each comparing a line of the financials this year to last year. Add them up with Beneish's weights and you get the score. The line that matters is -1.78: above it, the model classifies a company as a likely manipulator; below it, it doesn't.
That threshold isn't arbitrary. Beneish tuned it on a sample of companies known to have manipulated earnings, and above the line his model caught roughly 76% of them. The catch — and the whole reason this guide exists — is that it also flags honest companies. A score above -1.78 is a reason to look, never a reason to conclude.
The eight ingredients, in plain terms
You don't need to compute these by hand, but knowing what they watch tells you why a company scored the way it did:
- Receivables vs. sales (DSRI) — are customers taking much longer to pay, or is revenue being booked before cash is real?
- Gross margin (GMI) — is the margin deteriorating, giving management a motive to dress up the numbers?
- Asset quality (AQI) — is a growing share of assets the soft kind (capitalized costs, goodwill) rather than hard assets?
- Sales growth (SGI) — growth itself isn't manipulation, but fast growers face the most pressure to keep it up.
- Depreciation (DEPI) — is the company slowing depreciation to nudge profit higher?
- SG&A (SGAI) — are overheads rising faster than sales, a sign of trouble being papered over?
- Leverage (LVGI) — is debt climbing in a way that raises the incentive to hit covenants by any means?
- Total accruals (TATA) — the heavyweight. How much of this year's earnings is accounting entries rather than cash? This is the term that most often separates real manipulation from noise.
Why the score cries wolf — and who it cries wolf about
Beneish built the model on manufacturers. That origin story matters, because it means the score misreads two kinds of company on a good day:
Banks and insurers. The receivables, margin, and inventory terms are close to meaningless for a balance sheet made of loans and reserves. Running the M-Score on a bank isn't a red flag; it's using the wrong tool. The right tools are provisioning and reserve-adequacy checks.
Fast-growing companies. This is the trap that fools most people using the raw score. When a company's sales jump from a small base, its sales-growth index (SGI) and receivables index (DSRI) can explode — and because SGI carries real weight in the formula, the score sails past -1.78. But that's growth, not fraud. The tell: in genuine manipulation, the accruals (TATA) and gross-margin (GMI) signals move too; in a healthy growth story, they sit in normal ranges while only the growth terms spike.
So the single most useful habit: when a score crosses the line, don't stop at the headline number — look at which components pushed it over. Accruals and margin doing the pushing is a real signal. Sales growth doing all the pushing is usually an expansion artifact.
How to read it responsibly
Treat the M-Score the way a doctor treats an elevated lab result: a prompt to investigate, not a diagnosis. A reading above -1.78 sends you to three places in the 10-K — the receivables note (is revenue outrunning cash?), the gross-margin trend, and the cash-flow statement, where you compare net income to cash from operations. If earnings are drifting well above cash and the accruals term is doing the work, you have a real question to ask management. If the company is simply growing fast and cash tracks earnings, the flag was noise. Either way, the score did its only job: pointing you at the right paragraph faster.
How AnalystBook computes it — and where it disagrees with the raw score
We compute the M-Score for every company we cover, straight from the machine-readable numbers in their SEC filings — no estimates, and the same filing always yields the same score. Across the companies we've scored, about one in seven trips the -1.78 line. If we stopped there, we'd be handing you hundreds of false alarms.
So we don't stop there. Rather than show a bare flag, AnalystBook classifies the reading: it marks the model not applicable for financial institutions, calls out a crossing that's growth-driven when only the expansion terms spike, and surfaces a possible-manipulation verdict only when the accrual and gross-margin components — the ones that actually evidence manipulation — agree. And when too many inputs were missing from a filing to trust the result, it says so instead of pretending the company is clean. The number is the same one Beneish published; the judgment around it is what keeps a growth company from being branded a fraud. Every reading links back to the exact financial lines it came from, so you can check it yourself in one click.
That's the honest version of a forensic score: not a verdict machine, but a very fast way to find the one company in a hundred whose earnings deserve a second look — and to know which paragraph to read first. For research purposes only; not investment advice.