Optimization Of M-score Model In Financial Fraud Identification

Financial frauds of enterprises,which belong to economic illegal behaviors with adverse natures,have seriously infringed rights and interests of investors and disrupted correct orders of the capital market.Therefore,people from academic circles and practice fields have always focuses on disclosing and restraining financial frauds of enterprises.In order to identify financial reporting frauds of enterprises,Professor Beneish created the MScore Model,a financial warning model featured by simplicity and efficiency which is very popular in academic circles and practice fields of America.Since the beginning of this century,as short selling institutions of America sold short many China Concept Stocks,MScore Model gradually attracted the attention of Chinese scholars and the masses.However,not many research literatures of Chinese academic circles research M-Score model.Most of the researches used this model to analyze financial fraud cases of A-share listed companies,but few literatures empirically tested the rationality and applicability of M-Score model according to the background of A-share market.Although some researches reveal that Mscore model has defects,few literatures conducted optimization study of the M-score model according to frauds of listed companies in A-share market.In this case,using this model as a research tool may lead to misjudgment of the research object.Nowadays,many Internet investment forum and financial we-media have recommended M-Score model to investors as a stock selection tool without theoretical basis,which is highly likely to mislead investors in their making right judgment and to damage investors' rights and interests.This may greatly affect lead to corporate images of the target company.Therefore,this paper holds that it is necessary to conduct an in-depth research on M-score model based on the background of Ashare market in China.This paper uses theoretical analysis and empirical test to test the rationality and applicability of the model and optimized the model according to the fraud triangle theory.Besides,this paper also selects and pairs 83 enterprises which can meet the definition of corporate financial frauds from the data of Chinese A-share listed companies from 2010 to 2019.Besides,binary Logistic model is used to obtain the optimization model with higher recognition efficiency.This paper is divided into seven chapters.Chapter 1 is the introduction,which explains the research background and significance of this paper,summarizes research ideas,methods and contents and summarizes the innovation points.Chapter 2 is a review of domestic and foreign literatures on fraud triangulation theory and M-score model.Chapter 3 defines the concept of financial fraud and introduces the principal-agent theory and fraud triangulation theory.Chapter 4 evaluates M-score model and conducts theoretical analysis and application effect of M-score model from the two aspects of rationality and applicability.Chapter 5 is about the optimization design of M-Score model and its contents include model design and index selection.Chapter 5 also puts forward relevant hypotheses.Chapter 6 is about the inspection and analysis of M-Score optimization model,which includes six parts: paired sample test,multicollinearity analysis,binary logistic regression analysis,evaluation of optimization model,description of hypothesis test and robustness analysis.The M-score model is optimized by empirical research method.Chapter 7 describes the research conclusions,suggestions and prospects,summarizes researches of this paper and puts forward some specific suggestions.Main research conclusions of this paper are as follows:1.M-score model has limitations in identifying financial frauds of A-share listed companies(1)M-score model is not reasonably set.This paper evaluates the rationality of the Mscore model factor setting on the basis of the fraud triangle theory and draws the following conclusions: M-Score model can not cover all fraud factors and the design of some index formulas can easily lead to extreme values and invalid values.Therefore,the model has a high probability of distortion.(2)Under the background of A-share market,M-score model is applicable.The statistics and tests of experimental sample data show that there is no significant difference between the experimental group and the control group for most indicators and that the correct rate of the threshold discriminant method is only 59.03%,which shows poor practical application effect.In the test of binary Logistic regression model,M-score model does not have an obliviously overall significance,the indicators are indistinctive and the fficiency of model recognition is only 60.24%.All these further demonstrate that M-score model lacks applicability.2.The optimized M-Score model can effectively identify financial fraud behaviors of A-share listed companies.(1)The optimized model is highly reasonable.This paper optimizes measurement methods of the retained original model indicators and selects and designs several supplementary indicators according to the fraud triangulation theory.The 10 variables retained after multiple screening comprehensively cover the three aspects of pressure,opportunity and excuse and contain multiple classification variables,which significantly solve the problem of model distortion caused by abnormal or extreme calculations of indicators.(2)The optimized model is highly applicable.The direct recognition efficiency of the optimized model is much higher than that before the optimization and the accuracy rate is76.51%.After the discriminant threshold is reduced to 0.41,the recognition efficiency of the optimized model can be further improved.According to interval probability distribution situations of the model,this paper finds that the closer the discrimination probability is to the extreme case,the higher the recognition efficiency of the model will be and that they are not uniformly distributed.Therefore,this paper holds that we can judge the tendency of fraud or innocence of target enterprises according to model discrimination results.Therefore,this paper believes that the optimized model,which is highly applicable,can effectively detect financial frauds of A-share listed enterprises and make basic judgments on financial fraud risks of enterprises.Based on the research conclusions of this paper,this paper believes that: this study provides an efficient model for internal and external users of enterprise information to test the financial fraud behavior of A-share listed enterprises.The proper application of this model can effectively improve the rationality of investors' decision-making and the working efficiency of supervisors,and also help enterprises to improve their corporate governance ability.