Beneish M-Score: eight things that moved out of line
The Beneish M-Score (Beneish, 1999) combines eight warning signs found in the statements of companies caught overstating earnings into a single score. Most of its inputs compare this year with last year, so it points to where the numbers suddenly moved out of line.
Formula
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
The original cut-off is -1.78: above it the profile resembles the caught companies, below it the ordinary ones. An index above 1 generally means a move for the worse versus last year (TATA is a ratio in its own right).
The eight indices, with a worked example
Values for this guide series' example company (calculated by the Caveat engine).
| Index | Calculation | Coefficient | Example |
|---|---|---|---|
| DSRI Days sales in receivables index | (receivables ÷ sales) this year ÷ last year | 0.92 | 1.04 |
| GMI Gross margin index | last year's gross margin ÷ this year's | 0.528 | 0.90 |
| AQI Asset quality index | (1 − (current assets + PP&E) ÷ total assets) this year ÷ last year | 0.404 | 1.21 |
| SGI Sales growth index | this year's sales ÷ last year's | 0.892 | 1.11 |
| DEPI Depreciation index | last year's depreciation rate ÷ this year's (rate = depreciation ÷ (depreciation + PP&E)) | 0.115 | 0.98 |
| SGAI SG&A index | (SG&A ÷ sales) this year ÷ last year | -0.172 | 0.98 |
| LVGI Leverage index | (total liabilities ÷ total assets) this year ÷ last year | -0.327 | 0.98 |
| TATA Total accruals to total assets | (net income − operating cash flow) ÷ total assets | 4.679 | -0.022 |
| M-Score | -2.41 |
M of -2.41 is below the cut-off, in the ordinary range. Sales grew (SGI 1.11) but cash came in ahead of profit (TATA -0.022), which kept the score low.
How to read it
- The drivers, not just the score: when M is high, find which index contributed most (coefficient × value). For DSRI check collections; for TATA check the gap between profit and cash.
- The growth trap: a fast-growing company scores higher through SGI alone. The original paper treats growth as a source of pressure to manipulate, but healthy growth produces the same score.
- With the Dechow F-Score: Caveat's reliability cross-check asks whether the two models agree. Both flagging is the first thing to check; one flagging means checking just what that model looks at. The scores use different units and are never averaged. Receivables, accruals and asset structure feed both models, so both flagging there is the same number seen twice, not two independent signals.
Both sides: crossing the cut-off does not prove manipulation; acquisitions, heavy investment or restructuring move the indices too. Below the cut-off, a single index that stands out, such as DSRI, still deserves its own check.
Look at these too
- Dechow F-Score: a different model asking the same question
- Days sales outstanding: when DSRI is high
- Profitable but short of cash: when TATA (accruals) is large
Limitations
- The model was estimated on US listed companies in the 1980s and 1990s; its fit for other countries or private SMEs has not been validated.
- Caveat does not separate long-term investments in AQI and uses total liabilities in LVGI, slightly different from the original definitions.
- Operating cash flow is estimated from the statements.
- For reference only; it does not replace credit, investment or tax decisions.
See it in your own statements
Upload a balance sheet and income statement and Caveat calculates these for your company.
- Do both yardsticks point the same way? →
- Does anything look out of step with last year? →
- The eight checks against last year →
- Whether the margin on each sale shrank from last year →
- Whether assets that are neither cash nor equipment grew →
- How much sales grew on last year →
- Whether equipment is being written down more slowly than last year →
- Whether selling and admin costs grew faster than sales →
- Whether debt grew against total assets since last year →
- The share of profit no cash has arrived for →
FAQ
- Does an M-Score above the cut-off mean manipulation?
- No. It is a statistical resemblance to companies caught in the past. Acquisitions or fast growth can raise it, so check the indices that drove the score.
- How does it differ from the Dechow F-Score?
- Beneish looks at eight year-on-year ratio changes; Dechow estimates a misstatement probability from accruals, changes in receivables and inventory, and financing. The inputs and units differ, so Caveat checks whether they agree rather than combining them.
- Why do sources quote different cut-offs?
- Some sources use other cut-offs such as -2.22. Caveat uses -1.78, the cut-off for the eight-variable model in the original 1999 paper.
- Which figures are needed?
- Two years of revenue, cost of sales, SG&A, receivables, current assets, PP&E, total assets, total liabilities, depreciation and net income, plus inventory and payables to estimate operating cash flow.