Credit risk evaluation: a review and the application using Backpropagation Neural Networks

The dramatic increase in the corporate failures over the last decade has led to the increasing interests in failure prediction model. This paper reviews various quantitative methods and adopts one neural network approach, Backpropagation Neural Networks (BPNN), to identify the credit risk. BPNN gives convincing 54.55% bankruptcy and 100% non-bankruptcy out-of-sample prediction accuracies. The promising results validate that the neural network approach is an excellent supplement to the traditional prediction techniques and it provides management tremendous benefits on credit approval, loan securitization and loan portfolio management.