Stress Testing with a Credit Risk Model [extrait BMI 113]

Version numérique PDF

The aim of this paper is to build and estimate a macroeconomic model of credit risk for the French manufacturing sector. This model is based on Wilson?s CreditPortfolioView model (1997a, 1997b); it enables us to simulate loss distributions for a credit portfolio for several macroeconomic scenarios. We implement two simulation procedures based on two assumptions relative to probabilities of default (PDs): in the first procedure, firms are assumed to have identical default probabilities; in the second, individual risk is taken into account. The empirical results indicate that these simulation procedures lead to quite different loss distributions. For instance, a negative one standard deviation shock on output leads to a maximum loss of 3.07% of the financial debt of the French manufacturing sector, with a probability of 99%, under the identical default probability hypothesis versus 2.61% with individual default probabilities.

Auteurs :Avouyi-Dovi Sanvi , Jardet Caroline ??0000000000000000000 , Kendaoui Ludovic ??0000000000000000000 , Moquet Jérémy ??0000000000000000000 , Bardos Mireille ??0000000000000000000
Extrait de la revue BMI 113