The Recognition Research Of Chinese Listed Companies' Financial Fraud

Listed company's financial statements fully reflects the com- pany's operation, and supplies information to company stakeholders in order to help them make the right decisions in writing.However, The financial fraud problem of listed company is repeated that seriously disturbes the normal operation of Chinese securities market, prevents healthy development of Chinese securities market, has a huge impact on national economy.Now, the issue has become the focus that was concerned investors, creditors, national policy authorities. There- fore, It is a practical significance to establish a reasonable,high recognition rate of recognition model of financial fraud. Nowadays most studies are based on the research of methods.The recognition model of financial fraud recognition models are rarely in different industries.For the study of financial fraud has been very mature, mainly through the following approaches:logistic regression, principal component analysis, discrimination analysis, etc. Filter, model and forecast the financial data of company. This article elaborated on the main principle and basic idea of five modes:logistic regression, univariate analysis, neural network model, bayesian discriminant analysis and decision tree. On these basis we choose 104 companies of financial fraud as the fraud sample between 2009 and 2014, and 208 companies without financial fraud with the matching. Fist identified the established model of financial indicators using the mann-whitney U-test and correlation analysis, then set up a corresponding identi- fication model using logistic regression, bayesian discriminant analysis and decision tree. We analyzed the industry-classified model using the bayesian discriminant analysis and decision tree with better recognition results of these three models. Finally, we analyzed the identification effect of all kinds of models, proposed the limitation of research and outlook on the future.The research resultes show that the bayes discriminant analysis and decision tree can build higher recognition rate of financial fraud identification model. At the same time, the recognition effect of identification model of the different industries is significantly better than the same industries.Financial fraud identification model can be more effective to help the company's stakeholders to recognize the financial fraud company.