Social Sharing
Extended Viewer
On the White Board - June 2008
Jun 15, 2008
Since it was first widely adopted, in the early part of this decade, the one factor Gaussian copula (OFGC) model for credit tranches has been scrutinized. The scrutiny, though, has been for the most part reserved for whether the model could explain, that is, calibrate to, tranche prices observed in the market. After being developed in 2005 as a response to its predecessor's failure to calibrate, the base correlation framework is now itself under scrutiny, as depending on the specifics of the implementation, it can either not calibrate or can only calibrate with unrealistic parameter values. But calibration is not an end in itself. Rather, what we should be concerned with are the actual uses for the model. Two uses - hedging and risk aggregation - can be tested empirically, and it is the performance in these tests that is the real measure of the value of the model. We have tested (see Research Monthly) the model on these two fronts, and compared the results with those of a very simple regression approach. The results are not encouraging, and indeed for some series, the regression approach produces better hedges than the much more sophisiticated models. All signs point to the need for new models, particularly ones that explain more market dynamics than the OFGC approach.