Research On Two-class Logistic Model Based On Modified Benford Law

The financial problems of listed companies have always been the focus of the capital market,and also the focus of the work of audit institutions and government regulatory agencies.China’s government departments have continuously improved relevant laws and regulations,and increased supervision and punishment of listed companies’ financial affairs.However,some listed companies will still carry out black-box operations in order to survive.For example,ST and ST*listed companies have a great motivation to whitewash the report data in order to regain the normal listing.For investors and credit-related stakeholders,a two-category model is needed to classify whether a listed company is in a normal listing state,correctly assess investment risks,and make rational investment.For managers,it is necessary to establish available models to evaluate the current status of listed companies,to correctly evaluate the financial situation of listed companies,to adjust the operating principles of listed companies,to make correct decisions,and to ensure the healthy development of listed company.In view of this,it is very valuable to establish a feasible two-category model to identify the financial status of listed companies.Based on the existing research on the two-class model,this paper uses the financial data of the annual financial statements of listed companies in China from 2010 to 2017 as an example of empirical analysis to verify the advantage of the two-class Logistic model based on the modified Benford distribution law.This paper selects six financial indicators that can represent the financial level of listed companies as the financial indicators of listed companies.ST and ST*listed companies are used as "ST" samples,and normal listed companies are used as control samples.Based on the modified Benford law,the revised Benford factor is constructed.The modified Benford factor is added to the traditional two-class logistic regression model as a new explanatory variable,and a two-class logistic model based on the modified Benford distribution law is constructed.The data analysis results show that compared with the traditional logistic regression model,the two-class Logistic model based on the modified Benford distribution law improves the classification accuracy of the model.The binary classification logistic regression model with modified Benford factor has higher classification accuracy rate for Chinese listed companies,and the model prediction effect is better.The two-class Logistic model based on the modified Benford distribution law also provides a new method for the study of other two classification problems.