← All posts
Published 2026-07-24 · 6 min read · Quant Models

The Beneish M-Score, Explained: Why a High Earnings-Quality Score Usually Isn't Fraud

A high Beneish M-Score sounds alarming, but for a fast-growing company it is often just growth showing up in the math. Here is what the model's eight ratios measure, where the threshold sits, and why an elevated score is a prompt to dig deeper rather than a verdict.

What the Beneish M-Score actually measures

The M-Score was published in 1999 by Messod Beneish, an accounting professor at Indiana University. It is a statistical model, built by comparing a sample of companies later found to have overstated earnings against a control group that did not, and it is designed to estimate the probability that a company has manipulated its reported earnings. It does not audit the books, prove intent, or measure how much a figure was inflated. It simply asks whether a set of financial statements carries the statistical fingerprints that, historically, appeared more often in manipulators.

There are two published versions: an eight-variable model and a stripped-down five-variable model that drops the inputs relying on the cash-flow statement and a few harder-to-source ratios. The eight-variable version is the one most tools compute, and it is the one worth understanding line by line.

The eight variables, one by one

Every input is an index — this year's ratio divided by last year's — so the model is really measuring change. A value near 1.0 means "no change from last year." Values that drift well above 1.0 are what move the score.

RatioWhat it compares (this year vs last)What an elevated reading can hint at
DSRI — Days' Sales in ReceivablesReceivables as a share of salesSales growing faster than cash collection — possibly looser credit or premature revenue
GMI — Gross Margin IndexLast year's gross margin over this year'sDeteriorating margins, which Beneish linked to greater pressure to manage earnings
AQI — Asset Quality IndexShare of "soft" assets (not current, not PP&E)More costs being capitalized or parked in intangibles rather than expensed
SGI — Sales Growth IndexThis year's sales over last year'sRapid growth — not manipulation itself, but a setting where pressure runs high
DEPI — Depreciation IndexLast year's depreciation rate over this year'sSlowing depreciation — possibly longer assumed asset lives that lift income
SGAI — SG&A IndexSG&A as a share of salesOverhead rising faster than revenue, a sign of eroding efficiency
LVGI — Leverage IndexTotal debt as a share of assetsRising leverage and the covenant pressures that can accompany it
TATA — Total Accruals to Total AssetsAccrual-based earnings against total assetsEarnings not backed by cash flow — the single most heavily weighted input

The formula and the threshold

The eight indices are combined with fixed weights that Beneish estimated from his sample:

M = −4.84 + 0.920·DSRI + 0.528·GMI + 0.404·AQI + 0.892·SGI + 0.115·DEPI − 0.172·SGAI + 4.679·TATA − 0.327·LVGI

Two things stand out. First, TATA carries by far the largest coefficient (4.679), so how well earnings are backed by actual cash flow tends to dominate the result. Second, the output is a single number mapped to a probability. In the eight-variable model, a score above roughly −1.78 falls in the range Beneish associated with a higher probability of manipulation; below it, the probability is lower. (The five-variable variant is often cited with a cutoff near −2.22.) Because the scale is negative, "higher" scores are the less-negative ones — −1.5 is a higher score than −3.0.

The threshold is a statistical cut-point chosen to balance false alarms against missed cases. It is not a bright line where one side is "clean" and the other "guilty."

Why an elevated score usually isn't fraud

Here is the part that gets lost in headlines: the same signals that flag manipulation also describe a perfectly healthy, fast-growing business. Consider a company doubling its revenue:

Stack those together and a company that is simply growing quickly — or has just made an acquisition, changed its business mix, or shifted credit terms — can post an "elevated" M-Score without a single questionable entry. This is a well-known and common false positive: an elevated reading is growth-driven far more often than it is manipulation-driven. It is a prompt to investigate, never evidence of wrongdoing, and it should never be read as an accusation against any specific company.

Reading an elevated score responsibly

Treated as a screening flag rather than a verdict, the M-Score is genuinely useful. When a score lands in the elevated range, investors often examine the surrounding context before drawing any conclusion:

  1. Is the company growing fast? If SGI is the main driver, the flag is likely a growth artifact.
  2. Which variable is doing the work? A score pushed by TATA (weak cash backing) or DSRI (receivables outrunning sales) tends to warrant a closer read than one pushed by SGI alone.
  3. Does cash flow track earnings? Comparing net income to operating cash flow over several years is a simple cross-check on the accrual signal.
  4. What do the filings say? Revenue-recognition policy, receivables aging, and auditor commentary in public filings (SEDAR+ / EDGAR) add context a single number cannot.

The M-Score is one input among many, and it is most useful when paired with other earnings-quality and distress measures rather than read in isolation.

What the model can't do

The model earned its reputation partly because, in a now-famous graduate-student study, it flagged a large energy company's statements as unusual well before that company collapsed in an accounting scandal. That case is a reminder of what the tool is for — surfacing statements worth a second look — and not a template for labeling companies from a single number.

How Quintarthai helps

Quintarthai computes the Beneish M-Score across its US and Canadian coverage from public filings (SEDAR+ / EDGAR), and surfaces it in two places: as a sortable column in the screener and inside each company's deep-dive. Because an elevated score is so often growth-driven, it is shown in a neutral amber state — never a red "fraud" flag — next to its growth context (the sales-growth and accrual inputs) so the number reads as a prompt to investigate rather than a conclusion.

See the Beneish M-Score computed with its full growth context on the free Core dashboard — try it with a fast-grower like SHOP at quintarthai.com/app.
This article is for educational purposes only and is not investment, tax, or financial advice. Quintessentia Network Inc. (operating as Quintarthai) is not a registered investment adviser, broker-dealer, or securities exchange. Consult a qualified professional before making decisions. See Disclosures and AI Transparency.
Share this article
🐦 X / Twitter in LinkedIn
You might also like
Quant Models
Altman Z-Score, Explained: Reading Bankruptcy Risk With the Exact Formula
The Altman Z-Score explained in plain English: the exact 1968 five-ratio formula, the safe/grey/distress zones, the Z' and Z'' variants, and its real
Stock Screening
How to Screen Canadian Stocks (TSX & TSX-V): A Practical Guide
A repeatable method for screening TSX and TSX-V stocks: define your universe, filter junior distress, dedupe dual-listings, and apply academic risk mo

📚 Related research

📈 Markets
All Daily Market Wraps
Auto-published daily synthesis on movers, sectors, cross-border arb watch
💱 Investing
Cross-Listed Arbitrage Explained
FX-adjusted TSX/NYSE spread math — when a 2% gap is and isn\'t tradeable
⚠️ Risk
TSX-V Delisting — 9 Red Flags Worth Knowing
Float concentration, NI 51-102 timeliness, auditor downgrades

📰 Trending external news

🏦 Finance
BNN Bloomberg — Canadian Markets Financial Post — Investing Reuters — Finance
📊 Investing
Seeking Alpha — Market News Morningstar Canada Stockchase — Canadian Picks
₿ Crypto
CoinDesk — Markets Decrypt — News The Block