The M-Score is the output of a logistic regression model estimated on 74 manipulator firms and 1,708 non-manipulators from SEC enforcement actions. The threshold of −2.22 was chosen by Beneish to minimize total misclassification cost, assuming the cost of missing a manipulator (false negative) is roughly twice the cost of a false alarm (false positive). At this cutoff, the model correctly classified 76% of actual manipulators with a 17.5% false-positive rate. The alternative −1.78 threshold reduces false negatives at the cost of more false alarms — appropriate for short-sellers where missing a fraud is more costly than a wasted investigation. The M-Score is a continuous variable: a score of −0.5 is far more alarming than −2.18 even though both technically exceed the threshold.

TATA = (Net Income − Operating Cash Flow) / Total Assets is the most direct measure of accrual-based earnings manipulation. When a company reports income far above its cash generation, those earnings consist of accruals — revenue recognized but not yet collected, or expenses deferred into the future. Honest, healthy companies typically have TATA near zero or slightly negative (cash earnings match or slightly exceed reported income). The Sloan (1996) accrual anomaly — which predates Beneish — showed that high-accrual companies systematically underperform over the following year as accruals reverse. During an earnings manipulation, accruals are the mechanism: premature revenue recognition creates accounts receivable without cash; deferred expense recognition creates liabilities without cash payment. TATA captures both. Critically, TATA must be calculated using GAAP operating cash flow from the formal cash flow statement — not an estimated or adjusted figure — otherwise it loses most of its predictive power.

These models answer fundamentally different questions. The Altman Z-Score predicts bankruptcy risk: “Will this company become insolvent within 2 years?” The Beneish M-Score detects earnings manipulation risk: “Has management distorted the financial statements to misrepresent performance?” A financially healthy company can have a high M-Score (manipulating from a solid foundation). A distressed company can have a low M-Score (honestly reporting its poor results). They are complementary: a company with a high M-Score and a deteriorating Z-Score is a particularly serious warning — the manipulation may be concealing a worsening fundamental position that will eventually force insolvency. Professional forensic analysts and activist short-sellers typically run both models as the first two filters in an earnings quality framework before conducting primary research.

The model was developed on US public companies (GAAP/SEC reporting). For IFRS-reporting non-US companies: the key inputs (receivables, sales, PP&E, net income, operating cash flow) map consistently between GAAP and IFRS on most line items. The model can be applied directionally, but the −2.22 threshold has not been independently validated for IFRS reporters — use it as a relative ranking tool rather than a binary classifier. For private companies: the model requires two consecutive years of accrual-basis financial statements with a formal cash flow statement. Without a cash flow statement, TATA (the most predictive variable, weight +4.679) cannot be properly computed, and the model loses most of its predictive power. In this case, use the remaining 7 indices as a partial screen and supplement by manually inspecting accounts receivable aging, management representation letters, and auditor independence disclosures.