An Approximation for credit portfolio losses

This paper discusses a new approximation for the loss distribution in mixture models of portfolio credit risk, based on a normal approximation to the conditional loss distribution and the Berry-Esseen inequality. Applications to the risk management for credit portfolios and to the pricing of multi-name credit derivatives in factor copula models are discussed. A numerical case study shows that the method provides substantially more accurate results than the standard Vasicek approximation, while being at the same time computationally less expensive than standard Monte Carlo algorithms.