A Question for the Rocket Scientists

A QUESTION FOR THE ROCKET SCIENTISTS….Every couple of weeks I take a whack at trying to understand what Collateralized Debt Obligations and Structured Investment Vehicles really are. My latest take is that they’re basically ways to game the credit rating agencies (a game the agencies were none too eager to stop, as it turns out). Is that about right?

UPDATE: OK, here’s what I mean by this. Warning: abysmal ignorance about to be displayed.

Basically, the whole selling point of a CDO is that its risk-adjusted return is higher than the risk-adjusted return of its underlying parts — i.e., that you’ve modeled the CDO’s diversification or correlation or something better than Moody’s has, and it will thus yield a bit more than you’d normally expect based on the CDO’s rating.

Since the rated tranches are pretty straightforward, this mostly depends on convincing the world (and the rating agencies) that the small unrated tranche doesn’t affect the riskiness of the rated tranches by very much. Not enough to affect their stated ratings, anyway.

But in the long run, this can’t work. If you’re right, then eventually everyone else’s models will catch up and the yield of the securities in the unrated tranche will go down, thus eliminating the pooling advantage claimed by the CDO in the first place. If you’re wrong, then the CDO goes bust. The game only continues if you can consistently value the unrated tranche more accurately than anyone else, including the rating agencies. But how many people can actually do that?