Applying Credit Score Models to Multiple States of Nature

Abstract

We conduct a comparative analysis of there popular methodologies; the Z-score, the O-score, and a Cash Flow Based Model-score, in different scenarios constructed from various business conditions and diverse rating classes. We detect dissimilarities in the predictive power of the different models, demonstrate the economic importance, and analyze the sources of this assortment. We find that profitability measurements affect default odds distinctively among different rating categories and within diverse business segments, but cash flow quantities do not. We further deploy a canonical discriminant analysis and propose unique score for the examined states. These alternative scores often challenge the traditional approaches.