Analysis Of Loan Risks Of Commercial Banks In China

From the deep research on the formation of commercial banks'credit and specified Chinese commercial credit management and risk classification rules, the dissertation builds up a relation structure model of the credit risk factors as well as a risk evaluation model.lt also states the realization process of the three parts of the credit risk analysis system including the setup of standards system, the factor function simulation of credit risk evaluation model and risk early warning model.The dissertation not only studies the rules of choosing risk analysis index and related issues, but also sets up pre-selected indices of credit risk analysis, of which the meaning, algorithm and functions are stated in the dissertation. After the study of the main-factor analysis principal theory, algorythm and its defects, the author establishes the reform direction and sets up KLMS method applicable to the choice of quantitative index.Starting from the study of the characteristics of artificial neural network NN, the research focuses on a three-layer neural network with single output BP-based fuzzy evaluation model of credit assurance risk and a three-layer neural network with five-node output BP-based credit risk classification. It also couplings a double NN and constructs double additive BP neural netwok with vertical distribution, which leads to a scientific evaluation of commercial credit risk.Based on the discussion of the meaning and property of commercial credit risk early-warning, the dissertation points out its limits and redefines the meaning of commercial credit risk early warning. Meanwhile, this dissertation makes a deep research into the refinement of grey method(GM) analysis principles and GM(1,1) mechanism and then establishes SAGM model and SAGM-based wanrning method of commercial credit risk.The dissertation conducts a deep research on the structure, function and interface design of risk evaluating indices establishing module, neural network training module and five-level classification module of credit risk management, as well as credit risk pre-alarming module of the integrated commercial credit risk management system. Using Visual Basic in Windows98 environment, the author designs original integrated decision model for the credit risk management and conducts the relative deep empirical study based on substantial real data. The empitical research shows that this system is of much more flexibility, more precision of identifying risk, and higher level of automation and intelligence. Moreover, the integration of alarming module endows the system with more pre-alarming capability.