The Chaos Based Bankruptcy Model -- Current Status

Abstract

This study reexamines whether chaos theory can be used to develop a bankruptcy prediction model. In a 1996 paper, we applied chaos theory to bankruptcy prediction using a pair-matched sample of bankrupt firms and demonstrated that returns of firms nearing bankruptcy would exhibit significantly less chaos, as measured by Lyapunov exponents. The current study extends that work but differs from the earlier study by utilizing a binary logistic regression and bootstrapping to test a recent sample of bankrupt firms and their pair-matches. The current results show that the phenomena observed in 1996 persist.