Where it comes from
Messod Beneish, an accounting professor, published the model in 1999. He studied companies later found to have manipulated their earnings and compared their statements, in the years before discovery, with those of other companies.
The eight ratios
Each ratio compares this year with last year, so the model needs two consecutive annual reports.
- Days' sales in receivables index: are customers taking longer to pay, relative to sales?
- Gross margin index: has the gross margin deteriorated?
- Asset quality index: has more of the asset base moved into long-term assets that are harder to value?
- Sales growth index: how fast did revenue grow? Rapid growth creates pressure to keep it going.
- Depreciation index: has the rate of depreciation slowed, which flatters profit?
- SG&A index: have selling, general and administrative costs risen relative to sales?
- Leverage index: has debt risen relative to assets?
- Total accruals to total assets: how much of profit is not backed by cash?
The formula and the threshold
M = −4.84 + 0.920 × receivables index + 0.528 × gross margin index + 0.404 × asset quality index + 0.892 × sales growth index + 0.115 × depreciation index − 0.172 × SG&A index + 4.679 × accruals − 0.327 × leverage index.
WylthIQ flags a score above −1.78, the threshold commonly used with the eight-variable model. Below it, nothing unusual is flagged.
What it does not tell you
False positives are common, and every flag needs context.
- Plenty of companies with straightforward accounts trip it. Fast growth, acquisitions and a changing mix of products all move these ratios for ordinary reasons.
- It was built on a particular set of cases, from a particular period.
- It does not suit banks and insurers, whose statements are structured differently, so WylthIQ does not compute it for them.
- A flag says where to look — receivables, accruals, capitalised costs — not what will be found there.