The Expirical Sduty On Credit Risk Evaluation Of Business Loan In Chinese Commerical Banks

The main content of this thesis focuses on how to quantify credit risk in loan appraisal. The quickly economic development and the complexity of modern finance market have raised the credit risk confronted by our commercial banks. It is necessary to establish an effective credit risk evaluation and control system for our commercial banks.Firstly, this thesis review the important literature on credit risk evaluation from domestic and abroad papers, and introduces three important evaluation methods, namely classical specialist's score method, credit score model and modern credit risk models. After compared each method's advantages, disadvantages and their adaptability in presented China, we found classical specialist's score method is not objective enough, only suitable to our consumer credit risk appraisal; modern credit evaluation methods is out of condition to apply in China; credit score model have practicability and objective, is competent for evaluation of credit risk in business loan. Then the thesis discusses some typical modeling techniques for credit score model, such as discriminate analysis, Logit model and PNN etc. Lastly the Logit model is applied to measurement of credit risk for listed companies in China.In the empirical study, we use the default probability to quantify the credit risk in enterprise loan, and choose the listed companies which are"ST"from 2001 to 2006 as the"default"sample and the component of SHANGHAI SHENZHEN 300 INDEX as the"good"sample. The thesis firstly uses the t-test to select 19 significance variables, and then use principal component analysis to extract 4 communal components, which respectively stand for the ability of redemption, profit, operation and capital structure. We use the 4 components as the final input variables in Logit model. We use the estimate sample to build the credit risk predicting Logit model, and use the test sample to inspect the accuracy of the model. Experimental results show the model has good prediction ability and can classify the"default"and"good"companies well one year before.In conclusion, it is feasible to use Logit model and choose financial data as its input variables to evaluate credit risk in business loan. In the credit risk evaluation and control system of our commercial banks, we should combine the quantification analysis with the qualitative analysis, put credit score model first and take classical specialist's score method as the subsidiary, which can develop accuracy of the loan appraisal and reduce commercial banks'credit risk exposure at greatest degree.