Neural network models and the prediction of bank bankruptcy

A neural network approach to bank failures prediction is presented and empirical results show that among alternative models, neural networks is a competitive instrument for evaluating the financial condition of a bank.

Abstract

The number of failed banks has reached a high unparalleled since the great Depression. Research in developing predictive models for bank failures is therefore warranted and desirable in this turbulent period. In this paper, we present a neural network approach to bank failures prediction and compare its performance with existing models. Empirical results show that among alternative models, neural networks is a competitive instrument for evaluating the financial condition of a bank. The study concludes with a discussion on the potential and limitations of neural networks as a general modelling tool for financial applications.