Bank modelling methodologies: a comparative non-parametric analysis of efficiency in the Japanese banking sector

Following triggered corporate bankruptcies, an increasing number of prediction models have emerged since 1960s. This study provides a critical analysis of methodologies and empirical findings of applications of these models across 10 different countries. The study’s empirical exercise finds that predictive accuracies of different corporate bankruptcy prediction models are, generally, comparable. Artificially Intelligent Expert System (AIES) models perform marginally better than statistical and theoretical models. Overall, use of Multiple Discriminant Analysis (MDA) dominates the research followed by logit models. Study deduces useful observations and recommendations for future research in this field.