Social Sharing
Extended Viewer
Good, Bad or Useful?
Jun 1, 2008
Fashionable as it is to bludgeon quantitative models, it is a stance we have avoided in the past, noting that models were generally more in need of rational defense than of emotional indictment. But portfolio credit derivatives are an area in need of some subtle bludgeoning, if such a thing exists. Importantly, though, if we are going to cast out old standards, it should be for good reasons, relating specifically to how the model is in fact used. We propose here a few empirical tests, show that the standard Gaussian copula model barely outperforms the most simple benchmark and propose that these tests be used when we consider new candidates for the title of standard.
Download