The Cost of Fraud Prediction Errors

ABSTRACT We compare seven fraud prediction models with a cost-based measure that nets the benefits of correctly anticipating instances against costs borne by incorrectly flagging non-fraud firms. find even best trade off false to true positives at rates exceeding 100:1. Indeed, high number makes all considered too costly for auditors implement, in subsamples where misreporting is more likely. For investors, M-Score and, higher cut-offs, F-Score, are only providing net benefit. regulators, several economically viable as positive limited investigations regulators can initiate, and relatively low market value loss “falsely accused” firm would bear denials requests under Freedom Information Act (FOIA). Our results similar whether we consider or two alternative restatement samples. Data Availability: available from public sources cited text. JEL Classifications: G31; G32; G34; M40.