16 November 2008

Comments on Taleb

Filed under: Commentary — András Salamon @ 23:50

Nassim Nicholas Taleb makes provocative statements with several interesting ideas underpinning them.  To try to get past the over-the-top rhetoric of his interviews and two widely known books Fooled by Randomness and The Black Swan, I spent some time reading the papers at Taleb’s site.

Simplifying vastly, Taleb currently seems to have three main themes.

  1. The convergence to the central limit theorem is too slow to be of any use in processes based on social interaction.  This means that economics (and other social sciences) often cannot use a central limit theorem and work with the limiting stable distributions, but has to grapple with the unknown and often very weird distribution such data has.  In particular, using Gaussian distributions as any kind of approximation is often a very bad idea, even if variance is finite.
  2. The second point is that many social phenomena have self-similar distributions.  Barabási made this idea popular, and it has been widely observed in many kinds of network structures over the last decade.  Taleb argues that most economic time series behave this way also.  This means that their distributions are heavy-tailed, and that it is very difficult to estimate their parameters.
  3. Finally, Taleb notes that many social and economic phenomena are dominated by the effect of single events.  When regulators rejig the rules of a market, or war breaks out, or a meteor hits the earth, the effect is so large that it completely dominates the many years of “normal” behaviour.

This adds up to a pretty pessimistic point of view, on the surface.  However, Taleb appears to be formulating prescriptions for living with this state of affairs.  It seems he believes it is important to bound losses, through mechanisms such as reinsurance or options.  If dragons lurk in the heavy tails, then one can at least reduce their effect by donning some armour.  Taleb hasn’t yet gone far beyond the ranting stage in his published writing, as far as I can see.  It would seem likely that he does have specific prescriptions, since he is one of the principals at Universa Investments, which aims to achieve capital preservation regardless of market conditions.

One way to address the first issue would be to move further along the way towards continuous markets.  Based on the convergence of various distributions, one could estimate the number of events that would have to be observed for some summary event to be close enough to a desired limiting distribution.  There is already a strong move towards markets that are ever less discrete, and doing this analysis would tell us how far we have to go.  Academics in quantitative finance dealing with high frequency data and rough path theory are addressing these kinds of issues.

I’m not sure how to argue with the second issue, unless one can find new types of interaction underlying processes that do not generate self-similar laws.  This might be feasible, but would perhaps require new markets and ways of interacting.  All markets are to some extent designed, so this may be a long term approach, harnessing the work of many people, from different disciplines, on network science.

The issue of single events is probably unavoidable.  If one plans for the meteor strike, then in the interval until it happens life is pretty miserable (what with living underground, keeping a large stock of tinned food, and keeping up a gloomy outlook).  It seems unlikely that human nature could be changed to prepare for the worst, when that could be years or decades away.  The engineering solution (of doubling or trebling tolerances, because we don’t actually know the distributions), or the insurance solution (of keeping a reserve for a rainy 40 days), are difficult to implement when the outliers really are likely to be off the scale.  But again, people are working on this.  (An example was the recent Global Catastrophic Risks Conference.)

In conclusion: Taleb makes some good points, but his claim that everyone is ignoring these issues appears somewhat hollow.


1 Comment »

  1. Sometimes to reverse the paradigm is to shift the emphasis. What Taleb does, to my mind, is just to call attention, with good examples, of those difficult spots of prediction, where the qualitative makes the quantitative a little arbitrary and subjective. You are right, some are not ignoring these issues, but according to my experience, you would be surprised how many people in science are only barely aware of the implications.

    So, we are not buying Taleb’s points, we are being furnished with examples to contrast a purely quantifying view of things, which may impair comprehension of the phenomenon at large.



    PS. you have a rather neat collection of references in your blog, thank you for making a review of ECAI available!

    Comment by Mario Negrello — 22 February 2009 @ 0:49 | Reply

RSS feed for comments on this post. TrackBack URI

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

Blog at

%d bloggers like this: