"We believed," he said, "there was a possibility of institutional failure in the banking and financial system so we did all sorts of exercises, simulation exercises. But what we didn't see, and nobody saw, was the possibility of complete market failure, that markets seized up across the world."
That sounds suspiciously like one of these idiotic Monte Carlo exercises, where you assume everything in the world is governed by an uncorrelated random walk, and assume that one in five hundred year events will never happen.
Of course, the real world is not governed by random walks, and when the smelly stuff starts hitting the fan, everything gets very correlated, very fast. One in five hundred year events suddenly become one in ten year events, or worse.
This is the source of the idiotic approach to credit judgments that got us into this mess. I hadn't realised that HMT was doing the same sort of thing at a systemic level, though I suppose I shouldn't be surprised.