Research On Credit Risk Evaluation Models For Commercial Banks

The globalization and fluctuation in financial world have brought big challenge to bankers and investors all around the world. Many new methods are introduced into the credit risk evaluation area and a lot of quantitative techniques have put into the market. Comparing with famous international commercial banks, banks in China have a long way to go. By summarizing and analyzing the main approaches and models in the area of evaluating borrowers for commercial banks, this paper used three dominant methods (i.e. Multivariate Discriminant Analysis, Logit Analysis, and Neural Network) to select the characteristic financial indexes of default risk for enterprise borrowers. With these indexes, we applied the accelerated genetic programming (GP), which is up rising at present, for simulation and optimization purposes. The expressions evaluated from the genetic-programming models are then applied in judging the degree and establishing the limit of credit of the borrowers. We demonstrated that the accelerated GP is successful in that it evolves solutions with greater generalization and forecast capacity than traditional methods.