A PRACTICAL APPROACH TO DEVELOPMENT AND VALIDATION OF CREDIT RISK MODELS BASED ON DATA ANALYSIS

Abstract

The main objective of this research is to define how the performance of the models used by commercial banks in granting loans and for calculating the ECL according to IFRS9 is developed and validated. The development and validation of a high-performance model are two of the fundamental processes that can avoid the risk of non-payment in case of granting loans by a banking institution. Once a model has been developed, it is validated to assess the predictive power of risk estimators and rating models. Thus, in this research, several techniques for validating the performance of credit risk models will be presented and, with the help of the Python programming language, we will test these techniques on a data set consisting of observations regarding the clients of a credit portfolio. The case study illustrates how a credit risk model has been developed for default probability and how its performance has been validated in terms of power of discrimination, stability and accuracy.