On January 27, 2009, Nassim Taleb and Daniel Kahneman shared a stage for an hour or so. We finally got around to watching the video, which is embedded below.
It is a bit surprising to learn that the speakers seem to regard each other’s work positively. Here’s an oversimplification of their key theses:
- Kahneman: Prospect Theory says that given some symmetric variation in gains and losses, humans tend to weight losses more heavily. We are risk averse toward unlikely losses that would have a large impact, and risk seeking toward likely losses that would have a small impact. A good example of this is the notion of someone buying a cheap lottery ticket (happily seeking the risk of a very likely small loss) and also buying insurance (happily avoiding the risk of a very unlikely large loss).
- Taleb: the Black Swan metaphor says that people are risk seeking (or even wilfully ignorant) toward low probability, high impact losses, and risk averse toward high probability, low impact losses. Banks sold credit default swaps willy-nilly, LTCM squeezed every penny from its available leverage, etc., all without regard for the unlikely possibility of some fundamental assumption proving to be wrong. Taleb’s solution has reportedly been to pursue a strategy that buys cheap deep out of the money puts, taking frequent small losses over many years in order to profit from the eventual market crash.
Regardless of the merits of the views themselves, doesn’t it seem that these are directly opposed theories on what human behavior is like? Maybe it’s possible to massage the statements above so that one can consistently agree with both – maybe Taleb’s is a description of some smaller set of human behavior, say among certain professionals but not humans as such. But in any case, their reactions to one another’s work don’t make sense when the context is the financial crisis and institutional/professional behavior. Taleb should be criticizing Kahneman’s work as failing to describe what took place, not praising it relentlessly; and vice versa, mostly certainly vice versa. Readers, are we missing something?
UPDATE: Reader Aaron B. writes in with some insightful comments:
“My off-the-cuff read of Taleb/Kahneman (I haven’t seen the video yet): Their work is consistent in so much as we’re talking about subjective probability estimates. Translated to mechanism, Taleb’s thesis implies that the machinery for estimating subjective probability is incapable of modeling events in the far tails of high-kurtosis reality. Kahneman et al, meanwhile, have a series of experiments which show that people underestimate the probability of low-frequency events and overestimate the probability of high-frequency ones . Loss aversion is a separate effect that describes decisions made among options with known, “objective” probabilities.. There has been a great deal of attention paid these last few years to “decisions from experience,”  with a number of studies claiming that loss aversion does not hold up in situations where probabilities must be estimated from sequential experience (such as drawing cards from a deck), and in fact might even show the reverse effect . The jury is still out. Regardless, I think that the claim of dissonance between our stated intended behavior when presented with paper facts and actual operation under sequential uncertainty is one both Taleb and Kahneman should feel comfortable agreeing on.”
 google keywords: “subjective probability weighting function”