Abstract
This study reassesses the usefulness of the Beneish Model as a financial statement fraud-risk screening tool in an emerging-market setting using evidence from 872 companies listed on the Stock Exchange of Thailand during 2013–2022. The study applies the original eight-ratio Beneish M-Score and a Thai-sample logistic fraud-probability model to evaluate the discriminatory relevance of the underlying indicators. Robustness is assessed through classification accuracy, 10-fold cross-validation, sensitivity analysis, comparison with the Dechow F-Score, a TFRS 15 regime-change analysis, and additional specifications controlling for temporal and industry heterogeneity. The results show strong and balanced screening performance, correctly identifying 91.14% of fraud-accused companies and 90.90% of normally operating companies. Most Beneish indicators provide significant discriminatory information, whereas SGAI is not statistically significant, highlighting the context-dependent behavior of individual financial-statement signals. The overall M-Score does not differ significantly between the pre- and post-TFRS 15 periods, and the main relationships remain broadly stable after accounting for reporting-regime, temporal, and industry-specific conditions. These findings demonstrate that the Beneish Model remains a transparent and decision-useful screening mechanism, but its indicators should not be interpreted as mechanically transferable across institutional settings. The study contributes to international accounting and fraud-detection literature by showing how the effectiveness and interpretation of an established fraud-screening model depend on institutional and reporting contexts. Practically, the model provides auditors, regulators, investors, and other external users with a low-cost preliminary screening mechanism for identifying firms that warrant further investigation when access to internal evidence is limited.