Nomura finds limitations in Monte Carlo

Nomura Securities released a study this morning on the unquantifiable correlation risk particularly associated with assets that are not modeled using actuarial methods. In the piece, headlined What a Coincidence?, Nomura urges investors to tread cautiously into securitizations backed by these assets, such as CDOs, aircraft ABS and franchise loan ABS, all of which have suffered severed ratings volatility over the past few years.

Logically, CDOs backed by these types of deals have been hit hardest, as they feature two layers of this “model risk,” as Nomura coins it.

“Correlation was not explicitly an issue in the early years of the ABS markets,” Nomura writes. “The situation became tougher as structured finance expanded to include deals backed by small numbers of corporate or commercial credits. In such cases, it was impractical or impossible to use an actuarial approach based on the historical performance of similar ‘pools.’ Structured finance professionals had to develop new techniques for estimating the credit risk of such deals. In certain cases, they turned to mathematical models and Monte Carlo simulations. Unfortunately for some investors, those techniques appear not to have fully captured important credit risk correlations. In addition, those techniques almost never addressed the issue of time-varying correlations.”