Predicting Corporate Failure and Payment Behaviour : Addressing Some Pertinent Issues for Practitioners

This study aims to address a number of the issues which are pertinent to practitioners trying to use published models in day to day risk assessment. Firstly we look at an approach to handling missing data, then at the predictivity of payment behaviour information, which has become available on credit reports in the UK in recent years but has not so far been considered by researchers. Finally we consider the extent to which payment behaviour can be predicted, an issue of particular interest to trade credit managers and again an issue which is not the subject of published research. The sample used in this study is the first to be as large as those used by commercial model developers and is very heterogeneous, covering firms from sole traders to public limited companies. The models presented therefore provide a “generic model” benchmark for future work. We find that the implementation of corporate failure models in commercial environments, where missing information is common, could be facilitated by the estimation of models using a dummy variable structure to represent the missing data. We find that payment behaviour data can add incrementally to the predictivity of corporate failure models and can be used to successfully predict future payment behaviour in a trade credit context.