A Chaos Approach To Bankruptcy Prediction

Abstract

Chaotic systems, although deterministic and predictable over short horizons, appear to be random. This study applied chaos theory to bankruptcy prediction using a pair-matched sample of bankrupt firms. Given that healthy systems exhibit more chaos than unhealthy ones, it was hypothesized that the returns of firms nearing bankruptcy would exhibit significantly less chaos, measured with Lyapunov exponents, than at an earlier period. Successful univariate and multivariate bankruptcy prediction models were then constructed using the obtained Lyapunov exponents.