A New Methodology for Estimating Internal Credit Risk and Bankruptcy Prediction under Basel II Regime

Abstract

Credit estimation and bankruptcy prediction methods have utilized Altman's Z-score method for the last several years. It is reported in many studies that Z-score is sensitive to changes in accounting figures. Researchers have proposed different variations to conventional Z-score that can improve the prediction accuracy. In this paper, we develop a new multivariate nonlinear model for computing the Z-score. In addition, we develop a new credit risk index by fitting a Pearson type 3 distribution to the transformed financial ratios. The results of our study have shown that the new Z-score can predict the bankruptcy with an accuracy of 98.6% as compared to 93.5% by Altman's Z-score. Also, the discriminate analysis revealed that the new transformed financial ratios could predict the bankruptcy probability with an accuracy of 93.0% as compared to 87.4% using the weights of Altman's Z-score.

Predictive Potential and Risks of Selected Bankruptcy Prediction Models in the Slovak Business Environment

“…They recommend to expand studies to other methods such as conditional tress, KNN (k-nearest neighbours), SVM (support vector machines), as well as to improve the prediction accuracy mainly by using balanced sampling of training sets and Cut-point selection. Interesting results from recent studies can be found by researchers Kumar and Rao (2015) who developed a "new multivariate nonlinear model for computing the Z-score" as well as "a new credit risk index by fitting a Pearson type 3 distribution to the transformed financial ratios". By their application they showed that the new Z-score predicts bankruptcy with an accuracy of 98.6%, whereby original Z score reached 93.5%.…”