China's Commercial Banks, Credit Risk Assessment Model Study

Credit risk is one of the most ancient risks in the financial market. In China, as the opening of financial market is more and more formal and standard, domestic commercial banks are gradually facing drastic competition with foreign capital banks. But the status quo of credit risk management of domestic commercial banks is not optimistic. For example, system of managing credit risk is from being complete, and model of analyzing credit risk is outdated. This makes it urgent to support ideas for researches of analyzing credit risk for China commercial banks. Thus, this paper gives a method based on Principal Component Analysis and Neural Networks and hopes to probe the evaluation of credit risk that baffle commercial banks.Based on the traditional and modern management theory of credit risk, this paper has studied the evaluation model of credit risk in commercial banks. There are four sectors in this paper.Firstly, we define the meanings and characteristics of credit, credit risk and credit risk of commercial banks. Secondly, we analyze the reasons of credit risk in domestic commercial banks from macroscopical and microscopical aspects, and probe the problems in the practice of credit risk management. Thirdly, with the guidance of principle in evaluation system design, we analyze the finance factors that affect credit risk in commercial banks. Thus, we have established more satisfying evaluation system of credit risk. The last part is the most important part of this paper. Two aspects are there in this part: on the one hand, based on the theory of Principal Component Analysis and Neural Networks, we analyze the feasibility of evaluation model which is the combination of this two methods. On the other hand, we establish new evaluation system of credit risk in commercial banks, using the method of Principal Component Analysis. Then, we put forward the evaluation model of credit risk in commercial banks, using the Neural Networks, then make specific analysis and check on the result of neural networks model according to practical cases.