A Semiparametric Method for Predicting Bankruptcy

(Submitted to Journal of Finance, 2005)

 by

Ruey-Ching Hwang, K.F. Cheng, and Jack C. Lee

Abstract

A bankruptcy prediction method based on a semiparametric logit model is proposed for either the prospective (simple random) or the case-control (choice-based) data. The unknown quantities in the model are estimated by the local likelihood approach, and the resulting estimates are analyzed by their asymptotic biases and variances. The relationship on bankruptcy prediction methods using the prospective data and the case-control data is established. It is shown that the prediction methods using these two types of data are essentially equivalent. Empirical studies demonstrate that our prediction method is competitive with alternatives, in the sense of yielding smaller “leave-one-out” misclassification rate.