This paper is intended to develop some conditional credit risk models through a cursory approach in which any quality deteriorations in banks’ cash credit portfolios, measured as unfavourable changes in the ratio of delinquent credits to total credits, are considered to be a signal for an increase in overall credit risk and the weights of credit segments in entire portfolio are used as predictors. In modelling, two separate studies with consolidated and non- consolidated financial statement data covering the time period between March 2003 and March 2009 have been carried out. Our models based on Neural Networks and Multivariate Adaptive Regression Splines provide significant evidence that dynamic structure of credit portfolios are among the important determinants of credit risk. Furthermore, there exist some findings supporting the active role of macroeconomic conditions and our network models yield sound proofs suggesting that corporate governance concerns are influential on credit risk and quality.
Note: There is a file embedded within this post, please visit this post to download the file.↧