slade,

Good quibble.

I think the Gaussian assumption was made simply because it made the analysis easier. Kinda like assuming linearity, even though we know the world is non-linear.

The problem in these situations is that “hard” data tends to dominate “soft” judgments. It wouldn’t surprise me if an S&P analyst said that there was a huge risk to the system. But my guess is that the underwriter of a new issuance (a different guy in a different department on a different floor) still fed the same stuff into the same model and it spit out a AA- rating, and anyone who was involved in the review of this rating still believed that the model was just fine, thank you very much. After all, it had worked so far – like Taleb’s turkey.

Where things also went haywire was when the rating was then fed to a bond insurer who simply priced his AAA premium based upon the existing spread – probably using a 5-cell Excel worksheet that his MBA summer intern had built 10 years ago.

At the end of the day, though, the market is self-correcting. All of the bond insurers and most of the investment banks that drove this process are toast. Bond holders are taking huge write-downs. Equity is vaporizing, and the turkeys are coming home to roost (at least those who kept their heads).

I guess my fundamental point was that if you believe that fraud is the problem, you go after the people (e.g. Madoff). If you believe concentration is the problem, you go after the system. People make a much more convenient (for politicians) and dramatic (for the press) target, but if you simply put new people in an old system you’ll get the same results.

But systemic change is much trickier, mainly because no one understands the whole thing well enough to avoid unintended consequences. Better to have a few simple rules – transparency, liquidity requirements, leverage limitations, concentration limitations, etc. – than to add layer upon layer of rules that simply add to complexity and don’t really prevent the next meltdown since they only address the last one.

Cheers.

L3