Every famous accounting fraud was visible in the filed numbers before it was visible in the headlines. Not proven — visible. Enron’s revenue grew 750% in four years while its gross margin fell from 21% to 6%; its return on assets sank to 0.4% while ExxonMobil earned 10.7%; its return on capital sat below its cost of capital; it never once generated positive free cash flow. All of that was in the 10-K. This guide is the checklist that turns those patterns into questions you can ask of any stock, dividend or not, and it is what powers the Accounting Anomalies card on the Financial Health tab of the Stock Modal.
Five elements, ten items
Fraud researchers used to describe the conditions for fraud as a triangle — pressure, opportunity, rationalization — later a diamond with capability added. The Polytope Fraud Theory (2024) adds a fifth element, accounting anomalies: the idea that whatever the motive, fabricated earnings leave a systematic fingerprint in the statements, because a fake dollar of profit has to be parked somewhere on the balance sheet and never arrives as cash. It then lists ten things to check.
| Item | What to look for | In DiviDrip |
|---|---|---|
| 1 · Income statement bubble | Revenue booming while gross and net margins fall; receivables outrunning sales (DSRI > 1.2); book income that never shows up in cash taxes; earnings propped up by asset sales. | Computed |
| 2 · Balance sheet bubble | Fake profit has to be parked somewhere: assets swelling faster than revenue, ROA low and falling, goodwill and intangibles a large share of assets, debt mostly short-term. | Computed |
| 3 · Value destruction | Return on invested capital below the cost of capital year after year. Chanos’s original Enron short thesis. | Computed (WACC approximated) |
| 4 · Financial distress | Current ratio under 1, liabilities above 75% of assets, negative free cash flow funded by borrowing or share issuance. | Computed |
| 5 · Related-party dealing | Large transactions with entities insiders control. Enron’s SPEs were run by its own CFO. | Manual — 10-K notes |
| 6 · Aggressive remuneration | Bonuses keyed to revenue size or stock price; option grants with no holding period; hostility to questions. | Manual — DEF 14A |
| 7 · Unreasonable M&A | Goodwill jumping 25%+ with impairments that never arrive. Roughly 83% of mergers fail to add value. | Computed |
| 8 · Special industry / weak jurisdiction | Foreign filers where auditors cannot inspect (PCAOB access), or newly deregulated sectors. | Computed (static flag) |
| 9 · Insiders & auditors | Auditor changes (8-K Item 4.01), waves of executive departures (Item 5.02), heavy insider selling. | Computed from 8-Ks |
| 10 · Complexity & opacity | Hundreds of subsidiaries with no operating reason, inconsistent accounting policies, refusal to explain. | Manual — Exhibit 21 |
The paper’s rule of thumb: more than three items triggered, be cautious; more than seven, walk away. DiviDrip computes seven of the ten and links you to the SEC filing for the other three, so the card’s scale is 0 clean, 1–3 look closer, 4–5 multiple red flags, 6–7 avoid until explained.
The three scores underneath
The checklist is qualitative; three academic scores make it quantitative, and the card shows all three because their agreement is the signal. The weighs eight ratios — receivables growth, margin change, asset quality, sales growth, depreciation rate, SG&A, leverage and total accruals — into one number; above −1.78 is the alarm. Montier’s ignores the weights and simply counts how many of six manipulation levers are moving at once; above 2 is the alarm. The is the bluntest: the share of this year’s profit that did not arrive as operating cash. And minus the cost of capital is item 3 in one line — the number Jim Chanos used to short Enron in November 2000, a year before the collapse.
Enron, year by year
| 1996 | 1997 | 1998 | 1999 | 2000 | |
|---|---|---|---|---|---|
| Revenue ($B) | 13.3 | 20.3 | 31.3 | 40.1 | 100.8 |
| Gross margin | 21.2% | 14.6% | 15.6% | 13.3% | 6.2% |
| Net margin | 4.4% | 0.5% | 2.2% | 2.2% | 1.0% |
| Return on assets | 3.6% | 0.5% | 2.4% | 2.7% | 1.5% |
| Liabilities / assets | 76.9% | 75.1% | 76.0% | 71.3% | 82.5% |
| ROIC − WACC | −1.6% | −7.2% | +6.7% | −4.2% | +0.9% |
Read across, not down. Revenue up 7×, margins down 3×, returns on assets a third of the peer group, leverage at a five-year high, value destroyed in most years — and the consensus analyst rating in October 2001 was still Buy. The year-over-year table inside the Accounting Anomalies card is exactly this layout for the stock you are looking at, going back as far as 2010.
Reading the card without fooling yourself
- Every flag has a benign case. Tap it. Coca-Cola’s accruals jumped in 2024 because it deposited $6 billion with the IRS in a tax dispute — the check fires, the explanation is a footnote.
- Persistence beats magnitude. One bad year of DSRI is a customer paying late. Three years is a business booking sales nobody pays for.
- Sector matters. Banks have no gross margin; REITs are judged on FFO, not net income; the card marks those checks not applicable instead of guessing.
- Agreement is the signal. M-Score, C-Score and Sloan all red at once, plus two or more checklist items, is when you read the 10-K before you buy — or before you keep holding.
- The three manual items are the ones that convicted Enron. Related parties, pay design and subsidiary sprawl are text, not numbers; the card puts the EDGAR link one tap away.
None of this is a verdict. It is the list of questions a short-seller asks first, applied to your own holdings before someone else does. Pair it with the Piotroski F-Score for financial strength and the analyst coverage guide for why the sell-side is usually the last to notice.
FAQ
- What is an accounting anomaly?
- A number in the filed statements that moves in a way honest operations rarely produce: receivables growing faster than sales, profit growing while operating cash flow shrinks, assets swelling faster than revenue, margins collapsing during a revenue boom. One anomaly is a question. Several at once, for several years, was Enron.
- Does a flagged item mean the company is committing fraud?
- No. Every flag in DiviDrip’s Accounting Anomalies card has a benign explanation listed next to it — a tax deposit, an acquisition year, a deliberate low-margin expansion. The checklist tells you where to read the 10-K, not what to conclude. The paper it comes from advises caution above three flagged items and avoidance above seven of ten.
- What is the Polytope Fraud Theory?
- A 2024 framework that extends the classic Fraud Triangle (pressure, opportunity, rationalization) and Fraud Diamond (+ capability) with a fifth element, Accounting Anomalies, and turns it into a ten-item checklist an outside investor can run from public filings. Items 1–4 and 7–9 are computable from the statements; 5, 6 and 10 need the 10-K text and the proxy statement.
- How would the checklist have caught Enron?
- Revenue rose 750% in four years while gross margin fell from 21% to 6% and net margin from 4.4% to 1.0% (item 1). ROA fell to 0.4% against peers at 4–10% while assets ballooned (item 2). Return on invested capital sat below its cost of capital in four of five years (item 3). Debt hit 82.5% of assets with negative free cash flow every year (item 4). Hundreds of related-party SPEs (5), pay keyed to revenue size (6), goodwill write-offs (7), 68 executive departures and $1B of insider sales (9), and a structure analysts called a black box (10).
- What do Beneish M-Score, C-Score and Sloan accruals each measure?
- Beneish M weighs eight ratios into one manipulation probability (alarm above −1.78). Montier’s C-Score simply counts how many of six manipulation levers are moving (alarm above 2). Sloan’s accruals ratio is the share of profit that has not turned into cash (alarm beyond 10% of assets). They overlap on purpose — agreement between them is the signal.
- Why does DiviDrip skip some checks for banks, insurers, BDCs and REITs?
- Because the metrics mean something else there. A bank has no gross margin and its “receivables” are its loan book; a REIT’s net income is depressed by depreciation on appreciating buildings, so ROA is misleading and FFO is the right lens. The card marks those checks not applicable rather than producing false alarms.
- How many years of data does the card use?
- Up to sixteen annual statements (2010 onward for most large caps). Trend checks such as DSRI, the C-Score and accrual drift need at least three years; a recently listed company with one or two years gets only the point-in-time checks and says so.
- Where do I check the three items the card cannot compute?
- The card links straight to the company’s SEC EDGAR filing list. Related-party dealings are in the 10-K notes (search “related party”); executive pay design is in the DEF 14A proxy; the subsidiary count and offshore jurisdictions are in the 10-K’s Exhibit 21.