Research and improvement of fraud identification model of Chinese A-share listed companies based on M-score

Purpose - The purpose of this paper is to start with the background of the construction of the M-score model, find the variables that can represent the fraud characteristics of Chinese companies, and use the data of Chinese A-share listed companies to modify the M-score model. Design/methodology/approach - In this paper, the fraud behavior of Chinese enterprises that M-score cannot detect is summarized as the basis of adding variables. Then, based on the data of Chinese listed companies, a modified M-score model including nine variables is constructed by the logistic regression method based on Wald. Findings - Based on the original 8 variables of M-score, this paper adds 10 new variables that can represent the fraud characteristics of Chinese listed companies, and finally, constructs a modified M-score model with 9 variables. Results indicated that indexes such as gross profit margin, fixed assets depreciation rate, equity concentration and audit opinion can characterize the financial fraud of Chinese listed companies. Practical implications - The modified M-score model based on the characteristics of Chinese enterprises’ fraud is more suitable for Chinese market, which can help investors avoid fraud risks, protect their own rights and interests and reduce losses. Originality/value - Starting from the background of the model, this paper looks for variables that can characterize the characteristics of fraud in Chinese listed companies. Then, subdivides the research samples into specific fiscal years in which fraud occurs, so that the modified M-score model can be more suitable for the Chinese market.