The Ooghe-Joos-De Vos failure prediction models : a cross-industry validation

Abstract

This study tests the validity of the Belgian Ooghe-Joos-De Vos (1991) failure prediction models (1 and 3 years prior to failure) across 18 different industries and different size classes. Firstly, the performance results and the trade-off functions reveal a wide range of performances for the different industries. However, we notice that the OJD models perform best for the classical manufacturing industries and financial services, while they show the worst performance results for the service industries and the no-industry category. Furthermore, when using new, industry specific cut-off points, the error rates of the models are significantly reduced. Secondly, the OJD model 1 year prior to failure seems to perform best for large companies and companies with complete form annual accounts. Finally, the performance differences between the various subgroups with respect to industry, size class and form of annual account of the model 3 years prior to failure the are much smaller than those of the model 1 year prior to failure.