Modeling stochastic recovery rates and dependence between default rates and recovery rates within a generalized credit portfolio framework. (2nd April 2016)

• Record Type: Journal Article
• Authors: Fischer, Matthias; Köstler, Christoph; Jakob, Kevin
• Is Part Of: Journal of statistical theory and practice. Volume 10:Number 2(2016)
• Page Start: 342
• Page End: 356
• Publication Date: 2016-04-02
• Subjects: Credit portfolio model -- CreditRisk+ -- sector correlation -- stochastic LGDs -- saddlepoint approximation -- Monte Carlo -- PD-LGD correlation; 62P05; 519.505
• DOI: 10.1080/15598608.2016.1141733
• Languages: English
• ISSNs: 1559-8608

Abstract

ABSTRACT: Without any doubt, the CreditRisk + model that was launched by Credit Suisse Financial Products in 1997 is one of the most popular credit portfolio models in the banking industry. In order to accommodate more flexible dependence structures, Fischer and Dietz in 2012 introduced a generalized CreditRisk + framework. Focusing on the extension of Fischer and Dietz, the contribution of this article is twofold: First, we derive an analytic framework that allows for stochastic recovery rates, and for which the corresponding risk figures can be obtained via saddlepoint approximation. Second, we propose a straightforward approach for how to take dependencies between recovery rates and default rates into account. The corresponding loss distribution has to be derived using Monte Carlo simulations. We illustrate the effects of both stochastic recovery rates and dependence between recovery rates and default rates on the level of risk figures for a specific benchmark portfolio.