Measurement of the Probability of Insolvency with Mixture-of-Experts Networks

Abstract

The information how probable it is that a given company becomesinsolvent is important for owners, creditors and other financiers ofthis company. Especially investors are in need of this information tocalculate and control the risk they take with all investment decision.We show in this paper how the probability of corporate failure can bemeasured with artificial neural networks (ANN), namelymixture-of-experts networks. With the help of 8,660 financial statementsof 3,125 industrial companies we developed a mixture-of-experts networkthat is able to classify 90% of all companies which became insolventwithin the next three years correctly; the correspondingmisclassification rate of actually solvent firms is only 29%(Jerschensky (1998)).