Commercial Mortgage Default: A Comparison of Logit with Radial Basis Function Networks.

This article explores the use of artificial neural networks in the modeling of foreclosure of commercial mortgages. The study employs a large set of individual loan histories previously used in the literature of proportional hazard models on loan default. Radial basis function networks are trained (estimated) using the same input variables as those used in the logistic. The objective is to demonstrate the use of networks in forecasting mortgage default and to compare their performance with that of the logistic benchmark in terms of prediction accuracy. Neural networks are shown to be superior to the logistic in terms of discriminating between "good" and "bad" loans. The study performs sensitivity analysis on the average loan and offers suggestions on further improving prediction of defaulting loans. Copyright 1998 by Kluwer Academic Publishers