Assessment of Portfolio Credit Risk under Dynamic Default Correlation

Abstract

This paper describes a methodology to address topical issues in credit risk assessment under variable default correlation. The methodology solves the black box problem, typical of Markov regime switching, autoregressive conditional heteroscedasticity and other known models, accommodates heterogeneous loan portfolios, in contrast to the analytical approach to Value-at-Risk calculation, and provides for a variety of stress scenarios pursuant to the Bank of Russia’s requirements. This methodology may be applied in such areas as internal capital adequacy assessment, supervisory stress testing, financial stability recovery plans, and internal ratings-based models for credit risk assessment. The approaches based on the presented methodology have been integrated into the risk management system of one of the top Russian banks by assets.