Predicting business failure: an application of multicriteria decision aid techniques in the case of small UK manufacturing firms

Abstract

Small firms play an important role in most modern economies. Nevertheless, they are not without their problems. Therefore, the development of business failure prediction models for small firms is quite important. In the present study we employ three multicriteria decision aid (MCDA) techniques for the development of failure prediction models for small UK firms. The MCDA models are developed using a ten-fold cross validation and benchmarked against models developed with discriminant analysis and logit analysis. All models achieve satisfactory classification accuracies while including only a small number of input variables, hence avoiding the commonly encountered problem of data availability when dealing with small firms. Also we examine two non-financial variables – the age of the firm and auditors' opinion – but the classification results are only marginally affected.