A one-factor copula-based model for credit portfolios

Marek Kolman

Abstract

ABSTRACT

The aim of this paper is to develop an alternative model for portfolio credit risk to those widely used: CreditRiskC and CreditMetrics. The model aspires to patch the usual weak spots of portfolio credit models and is easy to implement. The general engine relies on a one-factor copula model and the concept of generating functions. The proposed model also uses only elementary inputs, which make it quick to deploy in a banking environment. The output is the usual credit value-at-risk. We also illustrate various methods related to the implementation and incorporation of credit wrong-way risk. As an example of this implementation, the double t copula is chosen in combination with a stochastic loss given default feature. Finally, we show the impact of various models on capital requirement by testing them on a sample portfolio.