Pricing Synthetic CDOs Using a Three Regime Random-Factor-Loading Model

Synthetic Collateralized Debt Obligations (CDOs) were among the driving forces of the rapid growth of the market for credit derivatives in recent years. Possibly the most popular model beside the Gaussian copula for pricing CDO tranches is the Random-Factor-Loading-Model of Andersen and Sidenius (2005). We extend this model by allowing more than two regimes of default correlations. The model is calibrated to market spreads at times of financial distress and during calm periods. For both points in time the model correlation skews are similar to the steep skews observed in the market and lead to an improvement to the standard Random-Factor-Loading-Model.