An excellent – straightforward – paper giving a sound overview of the development of structured finance has been published by Joshua D. Coval, Jakub Jurek, and Erik Stafford at Harvard Business School.
There’s not a mathematical lemma in sight, but the paper doesn’t skimp on detail. Well worth a read. Of note in particular is Table 3, which shows just how crucially important default correlation assumptions were to the under-estimation of CDO risks. Misunderstanding correlation (through the misapplication of the David Li 2001 Gaussian Copula function) was probably the single most egregious mathematical modelling mistake the banks made.
Note how disproportionately changes to the default correlation parameter affect senior tranches – the ones the banks were holding onto themselves. A mezz tranche (BBB-) moving from a 20 per cent default correlation assumption to an 80 per cent one loses only four notches. A senior tranche (AAA) making the same transition loses eleven notches.
This entry was posted by Sam Jones on . Tagged with Basel II, Capital Markets, CDO, Rating Agencies.