A Commercial Bank Customer Credit Rating And Empirical Research

Commercial banks operating process faces many risks,the credit risk is one of the major risks for Chinese commercial banks, so the credit risk management is especially important,The Basel II suggest commercial banks should have a complete credit rating system to improve credit risk management capabilities. Customer credit rating is one of the important parts in credit rating system, it can quantify the credit risk of loan customers by calculating the probability of customer default and do credit rating on the customers. Therefore, the im-portance of customer credit rating in credit risk management is obvious. This paper introduces a customer credit rating methods as follows:1.Calculating the weight of evidence (WOE), information value (IV)of finan-cial indicators; preliminarily choosing indicators for the logistic regression.2. Predicting the model of the default probability by logistic regression.3. Using Kolmogorov-Smirnov to test it (K-S Testing).ROC curve test the dis-tinction ability and accuracy of the predicted model; then determine customer credit rating model.Unlike using tradtional principal component analysis and factor analysis in selecting the indicators, this approach directly selects those indicators closely related to default and whose risk trend is linear by taking account of IV from information theory and the goodness fit of WOE. When calculating default probability which is the core element of credit rating by logistic regrssion, we employ a variable-ordering technique that substitutes the value of WOE for original index value to improve the toughness of the logistic regression and to reduce the impact of abnormalities. With data of the IT industry obtained from a large commercial bank, an empirical study is carried out to form a customer credit rating model; hence the feasibility of the approach is justified.