Application of Beneish and Dechow Linear Models in Identifying and Predicting Financial Reporting Distortions: Evidence from Companies Listed on the Tehran Stock Exchange

Abstract

The study aims to assess the effectiveness of the Beneish and Dechow linear models in identifying and predicting financial reporting distortions among Tehran Stock Exchange-listed firms. This applied, quantitative, and ex post facto study examined all listed firms on the Tehran Stock Exchange during 2018–2022. Samples were selected through systematic elimination, classifying firms with qualified audit opinions indicating misstatements, tax disputes, or major restatements as distortion cases. Data were collected through documentary review of financial statements and processed using EViews software. Logistic regression was employed to test hypotheses and compare the predictive ability of the Beneish (M-Score) and Dechow (F-Score) models. The Beneish model achieved an overall accuracy rate of 84.26%, while the Dechow model reached 68.88%. Comparative analysis confirmed the superior performance of the Beneish model. Key predictors such as sales growth, asset quality, and total accruals significantly contributed to detecting manipulation. The findings indicate that the Beneish model, emphasizing internal financial ratios, performs better in volatile economic environments like Iran. Both Beneish and Dechow models effectively detect financial reporting distortions in Iran’s market context; however, the Beneish model demonstrates higher predictive power and practical applicability for regulators, auditors, and investors. These findings underscore the importance of localizing fraud detection models for emerging markets.