Comparative analysis of corporate failure prediction methods: Evidence from Chinese firms

Abstract

This paper examines four most popular alternative methods that have been applied in financial failure prediction: linear discriminant analysis, logit analysis, neural networks and support vector machine. The main purpose is to make comparisons of the prediction abilities among different methods. It was implemented by using the Chinese firms data one, two and three years prior to failure in the empirical analysis. The results indicate that the classification accuracy of support vector machine is the highest. © 2009 Springer Berlin Heidelberg.