Analysing Bankruptcy Data with Multiple Methods

Neural networks have proved in many ways and in a number of publications to be real challengers to statistical methods especially to logit analysis in predicting failures. However, most of the studies have used a rather small data set, very often close to only one hundred observations. Therefore, it has been difficult to say whether there are any significant differences between the methods tested. In this study, we extend a previous study and compare rule-based learning with neural networks and logit analysis using a larger data set consisting of 570 companies. We investigate the effects of the prediction capabilities of the methods using different sample sizes and different time periods for estimation. Our study shows that in this domain neural networks and rule-based learning perform better than logit analysis, but there is substantial variation in the results depending on the sample size and