The probability of default in internal ratings bades (IRB) models in Basel II: an application of the rough sets methodology

Abstract

The new Capital Accord of June 2004 (Basel II) opens the way for and encourages credit entities to implement their own models for measuring financial risks. In the paper presented, we focus on the use of internal rating based (IRB) models for the assessment of credit risk and specifically on the approach to one of their components: probability of default (PD). In our study we apply the rough sets methodology to a database composed of 106 companies, applicants for credit, with the object of obtaining those ratios that discriminate best between healthy and bankrupt companies, together with a series of decision rules that will help to detect the operations potentially in default, as a first step in modelling the probability of default. Lastly, we compare the results obtained against those obtained using classic discriminant análisis. We conclude that the rough sets methodology presents better risk classification results.