An Affine Intensity Model for Large Credit Portfolios

Beatrice Acciaio and Stefano Herzel

Abstract

The paper proposes a reduced form model for credit risk in a multivariate setting. The default intensities are obtained as a linear combination of three independent factors. The factors are affine jump-diffusion processes that can be interpreted as the intensities of general, sectoral and idiosyncratic credit events. The model is very flexible and can be efficiently calibrated to the term structures of obligors’ default probabilities and to their correlations. We formulate an algorithm for the exact simulation of the default times that does not rely on time discretization.

[PDF]