Discriminant Analysis and Logistic Regression Applying To Credit Risk Management

Abstract

The financial crisis that is currently shaking the world, particularly the successive failures of the major banks have brought the issue of banking risks, including credit risk, back to the forefront. This risk must now be managed by more sophisticated methods. In this paper we present two methods that allow us to establish two functions, namely Fisher discriminant analysis and logistic regression; these two functions allow us to evaluate the risk of non-repayment incurred by a bank in the light of our data. It emerges that Fisher discriminant analysis is more effective or efficient than logistic regression for the evaluation of the risk of non-repayment of credit. Discriminant analysis and logistic regression are two methods of credit risk management here the problem we are trying to solve is how to help banks choose the most efficient method between the latter two.