Predicting Bankruptcy Using Neural Networks in the Current Financial Crisis: A Study of U.S. Commercial Banks

Abstract

We develop a model of neural networks to study the bankruptcy of U.S. banks. We provide a new model to predict bank defaults some time before the bankruptcy occurs, taking into account the specific features of the current financial crisis. Based on data from the Federal Deposit Insurance Corporation, our results corroborate that distressed banks undertook higher credit risks and were more heavily concentrated on real estate. Interestingly, the distressed banks do not show lower cost efficiency than their wealthy counterparts, suggesting that bank failures are a consequence of careless bank strategies rather than low cost efficiency. After drawing the profile of distressed banks, we use our model to predict future bankruptcies and test the performance of the model by comparing our predictions with the actual bankruptcies between January-June 2010. Our model shows a high discriminant power and is able to differentiate correctly wealthy and distressed banks. Specifically, our model would have been able to predict in December 2009 around 60% of failures that occurred in the first six months of 2010.