Tails of the unexpected

Credit models have been heavily criticised following their breakdown in the face of recent market moves. Could a new breed of models incorporating unknown factors driving defaults be the answer? By Mark Pengelly

Credit models have come unstuck due to the crisis triggered by souring US subprime mortgage loans. While the base correlation approach using Gaussian copula models has long been criticised as a grossly oversimplified way of pricing bespoke collateralised debt obligation (CDO) tranches, the torrid markets of the past 10 months have raised the volume of such criticisms. Banks have reported billions of dollars in losses on CDO holdings - in particular, on super-senior tranches of CDOs referencing