A comparative study between logistic model and Z-score model : empirical evidence from US markets

Abstract

Academics in the fields of accounting and finance have actively studied bankruptcy since the revolutionary work of Beaver (1966, 1968) and Altman (1 968). The majority of these studies focused on the statistical methodology and expanding the set of explanatory variables with the objective of improving prediction accuracy. Along the same vein, this study relied on accounting based measures as the predictor variables. In particular, we attempt to adopt the statistical approach as proposed by Westgaard Wijst (2000) “Default Probabilities In A Corporate Bank Portfolio : A Logistic Model Approach” to calibrate a bankruptcy prediction model and using it to investigate bankruptcies events in the United States from 1999 to 2001. Eivind (2001)’s “A Model Of Bankruptcy Prediction” also applied a similar methodology on the Norwegian market. Of the multivariate approaches, the multivariate discriminant analysis (MDA) approach is frequently cited and notable results are provided by Altman (2001 ) “Predicting Financial Distress Of Companies”. The effectiveness and reliability of prediction results from the logistic regression model is studied and then compared to those obtained by ZScores.