Value-Trap Radar — Stocks Flagged by Beneish M-Score

What you’re looking at

In 1999, Messod Beneish published a paper in Financial Analysts Journal that’s quietly become the most-cited number in forensic accounting. He built a probit model on eight financial-statement ratios designed to spot the kind of aggressive accounting that precedes a restatement. He trained it on a sample that included some of the most famous earnings manipulators of the 1980s and 1990s, validated out-of-sample, and showed the score did a better job than chance at flagging companies that would later be caught.

A year later, a class of MBA students at Cornell ran Beneish’s model on the financials of a company most of their professors had given top marks. The score came back red. That company was Enron. Beneish’s M-Score has been earning its keep ever since.

This list is the S&P 500 companies whose most recent M-Score is above -1.78 — the threshold Beneish himself defined as the manipulator zone. The higher the number, the more aggressive the financial-statement pattern looks. A score above 0 means the model is very suspicious. The threshold doesn’t mean fraud; it means the company’s reported financials display the same structural patterns that historical manipulators have. Sometimes the explanation is benign — a one-time accounting change, a real growth inflection, a working-capital swing from a big new customer. Sometimes it isn’t.

The eight ratios, plain English

Beneish’s model is an interaction of eight year-over-year ratios. You don’t need to compute them by hand to use this list, but it helps to understand what the model is reacting to when it flags a name.

DSRI — Days Sales in Receivables Index. Did the receivables grow faster than sales? If yes, the company may be recognizing revenue it hasn’t actually collected — the textbook channel-stuff pattern.

GMI — Gross Margin Index. Did gross margin deteriorate? Companies under margin pressure sometimes get more creative on the revenue side to keep reported profits looking steady.

AQI — Asset Quality Index. Did the share of “soft” assets (everything other than current assets and PP&E — think intangibles, deferred costs, goodwill) go up? Companies that capitalize what they should expense will show this pattern.

SGI — Sales Growth Index. Is sales growth itself accelerating? Beneish found rapid growth correlates with manipulation pressure — high-growth companies face stronger incentives to keep the streak going.

DEPI — Depreciation Index. Did the depreciation rate slow down? Stretching useful lives boosts current-period earnings, which is one of the easiest manipulations to spot if you know to look.

SGAI — Sales, General & Administrative Index. Did SG&A as a % of sales jump? Disproportionate overhead growth can signal margin compression management hasn’t acknowledged yet.

LVGI — Leverage Index. Is leverage rising? Companies approaching debt covenants have a strong reason to manage reported earnings upward.

TATA — Total Accruals to Total Assets. Are accruals (the gap between accrual-basis earnings and cash earnings) growing? This is the single most powerful component in the model — a high TATA almost always shows up in manipulators.

Beneish weights each, sums them, and outputs one number. A simple, auditable score. Like Piotroski, the appeal is that you can verify it yourself with one fiscal year and the prior comparison year.

How to read this list

Names above -1.78 are in the suspicious zone. The classic interpretation:

• -1.78 to -1.0: Mild warning. Worth a paragraph of explanation in the MD&A. Often nothing — a one-time working-capital swing or an accounting change disclosed in the notes. Read those notes.
• -1.0 to 0: Notable warning. Filing.fyi flags these in the forensic panel on every ticker hub. We’d want to see a clear, documented reason in the company’s own filings before treating the underlying earnings number as clean.
• Above 0: Strong warning. Historically, names crossing 0 had elevated rates of subsequent restatements. The model isn’t always right — false positives are real — but at this level we’d treat the reported earnings as needing independent verification before relying on them for valuation.

Critically: the M-Score is a forensic accounting score. It doesn’t say anything about whether a stock is overvalued or undervalued in the market. A clean M-Score doesn’t make a name a good investment. A flagged M-Score doesn’t make it a short. It just means the reported numbers may not be what they appear.

Why this list won’t catch every manipulation

The model is trained on US GAAP financials with at least two full fiscal years of comparison data. It struggles with:

• Newly-public companies with less than two complete fiscal years on file. Some IPOs that are demonstrably aggressive in the rearview were “clean” on M-Score at the time because the model didn’t have enough history.
• Financial-sector companies (banks, insurers, broker-dealers). The ratios assume a normal industrial balance sheet. M-Score outputs for financial-services firms are technically valid but interpretively weak.
• Companies with unusual one-time events: a big acquisition, divestiture, or recapitalization can produce M-Score readings that look like manipulation but are accounting-level reflections of a structural change. Read the company’s own segment disclosures before drawing conclusions.

The model also doesn’t catch all manipulators. Companies that manipulate via non-financial-statement channels — like fake customers, round-trip transactions with off-balance-sheet vehicles, or revenue recognized in stub periods — can score clean on M because the manipulation doesn’t show up in the trailing fiscal-year ratios. Wirecard, for instance, was clean on M-Score before it collapsed; the manipulation was in fictitious cash balances that had no obvious financial-statement footprint.

How we use this list internally

Filing.fyi treats a name on this list with elevated scrutiny. Specifically: any time a ticker shows up here, we want to see (a) a Piotroski F-Score on the other side that confirms operational health, (b) cash flow from operations clearly exceeding net income for at least two of the past three years, (c) a clean accounts-receivable trend in the MD&A. If two of those three are missing, the M-Score signal stands.

We won’t short a name just because it sits on this list — Beneish himself has been clear the model is for risk screening, not direction. But we will demand much stronger evidence of management quality and competitive position before allocating to a name above -1.0. The number of long-term great businesses with structurally high M-Scores is small.

Cross-references

• going-concern-radar — companies whose Altman Z-Score puts them in the distress zone. Overlap with this list is the highest-risk cohort on Filing.fyi.
• highest-piotroski — the inverse. Companies on the high-strength side. Names that appear on both lists deserve a full read of the underlying 10-K.

Common questions

A company I respect is on this list. Should I sell? No — at least not because of this list. The M-Score is a risk screen. It says “the financial-statement pattern looks like the pattern past manipulators have shown.” It doesn’t say “this is fraud.” Read the filing notes. Check whether the score moved sharply year-over-year (a structural shift) or stayed roughly stable (likely a normal feature of the business). The hub page on each ticker shows the trend.

Why are growth companies disproportionately on this list? Beneish’s SGI (Sales Growth Index) explicitly penalizes accelerating sales growth, because the historical sample showed manipulation pressure rises with growth. That’s a feature, not a bug — but it does mean clean high-growth companies appear here too. Don’t read the list as a verdict; read it as a starting point.

Is this updated automatically? Yes. The underlying scores recompute every time a constituent’s new 10-K hits EDGAR. The ranking on this page reflects the most-recent 10-K processed for each name in our universe.

Where do I see the per-component breakdown? On each ticker’s hub page. The forensic panel shows each of the eight Beneish ratios with the year-over-year delta — so you can see which test the company is failing, not just the aggregate score.