Random thinning model with a truncated credit quality vulnerability factor: Application to top-down-type credit risk assessment

Abstract

In the top-down approach of intensity-based credit risk modeling, a procedure called “random thinning” is needed to obtain credit event intensities for sub-portfolios. This paper presents a random thinning model incorporating a risk factor called the credit quality vulnerability factor (CQVF) to capture time-series variation in credit event occurrence in a target sub-portfolio. In particular, we propose a type of CQVF that follows truncated normal distributions specified by macroeconomic variables. Using credit event samples of Japanese firms, our empirical analysis aims to clarify the applicability and effectiveness of the proposed model to practical credit risk management. Since macroeconomic variables are included in our model, it is applicable to the macro-stress testing of portfolio credit risk management within a top-down-type framework.