I recently reread ‘The Black Swan’ by Nassim Nicholas Taleb. Much of what he says in three hundred pages could’ve been said as well in a hundred. He is exasperatingly ignorant and downright condescending about the insights of modern science, game theory and extreme value theory. He ties himself in contradictory knots about epistemology and empiricism. Which explains why when I read it the first time, I felt it to be a waste of my time. But his main point about our ignorance of high-impact, low probability events is more or less true. We overestimate our ability to predict and serially assign lower than warranted probabilities to extreme events.
Right up until 2007, every single big bank assigned a single number to their best estimate of the probability of U.S. house prices falling: zero. PhDs from Harvard and MIT who worked at Goldman Sachs, Lehman Brothers and JP Morgan, literally picked this numerical value to plug into their risk models. During the week of the Lehman collapse, bank analysts claimed that they had seen 3 six sigma events (each with probability lower than 1% of 1%)!
The Japanese Nuclear Safety Commission (NSC) insists on a safety level that would keep the probability of a disaster lower than that of a one-in-a-million-year event. Let me quote from their website:
“The mean value of acute fatality risk by radiation exposure resultant from an accident of a nuclear installation to individuals of the public, who live in the vicinity of the site boundary of the nuclear installation, should not exceed the probability of about 1×10-6 per year.”
The current disaster was only 66 years after 1945.
Now consider the assessment of the stability of the Arab Petro-dictatorships in the Middle East. The assumed effective probability of events that have already transpired was close to zero. But even now policy makers in Washington and traders in energy markets are implicitly assigning extremely low probabilities to game changers like regime change in Saudi Arabia. That is why smart hedge fund managers are betting massively on tail events like Brent crude hitting $200 a barrel this June. Just to be clear, the claim is not that it is likely. Surely, the fall of the Saudi regime is unlikely even now. The claim is that our assignment of probabilities is subject to huge errors. And because it is such a game changer, we ought to be really, really conservative. With Brent crude spot price at $113 a barrel today, it is certainly arguable that the instability premium is too low.
We just don’t seem to be good at assigning probabilities to extreme events.
[Update: The March 26th issue of the Economist has an article called Fat-Tailed attraction. The recent spate of black swans (Banking collapse, Middle East revolt, Japan quake) has investors freaking out. PIMCO is overseeing $400 billion in tail risk products and our very own Taleb’s Tail fund has grown from $200m in 2007 to $6 billion now. Another article on the world economy reports that volatility is up sharply across the board.]