Friday, February 13, 2009

Taleb and Kahneman in Germany

see video here.

A couple quick comments. Taleb noted that he would not speak to the US based Chartered Financial Analyst meeting unless they stopped teaching Modern Portfolio Theory. They did not, so he withheld his eminence. Now, first, this is an absurd request, and I'm no fan of the CAPM. Second, Taleb argues MPT, including the CAPM, is worse than useless. Yet Peter Bernstein wrote a nice blurb on his book! Bernstein wrote a couple hagiographies of the founders of the Modern Portfolio Theory (Against the Gods, and Capital Ideas). I guess most people think inconsistency--MPT is the greatest intellectual achievement of the past 60 years, and worse that worthless--is no big deal.

As you get older, it is interesting to find people you really know are totally clueless as eminent authorities. Like when you are 24 or so, and see talking heads on TV, and think they know a lot, until they start talking about something you really know, like ethanol or health care, and then realize they don't know much about that. Perhaps they don't know anything about Somalian relief efforts, or fiscal policy, as well?

Taleb notes the Kahneman is the only Nobelist to be able to predict, via his 'prospect theory'. This theory was developed to explain why people gamble and buy insurance. It 'predicts' only what it was designed to explain. One could call this a successful prediction (people will buy insurance and lottery tickets next year too) only in the way that one considers superstring's prediction of gravity amazing. I find his prospect theory useless, because Markowitz, and Friedman and Savage introduced doubly inflected utility curves back in the 1950's, but the profession abandoned this approach because it explained too muc. A theory that people can be risk averse or risk loving, depending on the situation, is not useful. It's like George Soros's book, where in The Alchemy of Finance, he notes that stock prices are usually biased--either too high or too low. Well, that's one interpretation. A better one is that the expected return is unbiased, and stock prices will fluctuate. Behavioral finance supposes that economists have never tried to understand humans, which is simply false.

Unlike what Taleb's colleague and friend Paul Wilmott states, my criticisms of Taleb are not driven by envy, any more than I might envy Suze Orman or Naomi Klien, both highly popular in economics (I think Suze is generally correct if banal, Naomi strident and plain wrong). To the extent Taleb makes a consistent, novel point--which is rare--he is wrong, pure and simple. Further, he is truly crazy.

I agree with Taleb on some things, including the limitations of Value at Risk. But his inference than VaR is worse than useless, or the cause of the recent crisis, is simply stupid, and highlights that he does not understand how strategic decisions are made in large organizations. VaR had nothing to do with the subprime crisis, it has a very limited area of application, and strategic decisions are not one of them. Just as 'risk managers' do not okay acquisitions, risk managers do not okay strategy, they merely make sure the rules are being followed, in part by using a Value at Risk. That VaR might have been used to validate investing in mortgages (UBS had some poor Value-at-Risk numbers underlying their mortgage exposures), does not mean it was a critical driver. The decision, really the assumption, was widely held that default risk on mortgages was a trivial risk because the US had not had an aggregate decline in housing since WW2. Everything after that is incidental.

Further, the idea that 'shit happens' should be a center to one's thinking is totally perverse. It does not explain, it does not instruct, but rather just gives people an excuse to be lazy, overpay for insurance, and go on wild goose chases. They can always blame a 'Black Swan' for their failure to anticipate, or their lack of results.


Dan Stowell said...

Picking on Taleb or Krugman may be entertaining (it's one of the reasons I read this blog), but it's not too instructive. Once you accept the premise that most people don't know that much even if they speak out loudly and often, it's kinda like watching a cartoon character slip on a banana peel over and over. Finding people who can explain the nuances in their thinking might be harder, but more helpful in the long run for your audience.

Eric Falkenstein said...

It is easier to criticize that generate good ideas, ideas that are new, true and important. So I criticize more than I present new observations, mainly because I'm a realist about my good idea generation rate. If they are in the news, I figure it's topical.

But I'm editing my book now, which is chock-full of new, true, and important information. Yet, I want to save most of those ideas for the release date (though the theme, risk not related to return, is an old one for me).

A Concerned Taxpayer said...

As a CFA and Finance grad, I never found MPT or CAPM to be useful in any manner due to the limiting assumptions. Anyone building their investment strategy around this deserves to lose their money. Taleb is right. The normal distribution curve assumption in finance is dead and probably should never have been alive.

Anonymous said...

Well if you don't use the normal curve, what do you use?

If you don't assume that risk is priced by beta (which everyone already knows is at most partially true) what do you assume?

If you don't assume anything how do you make decisions? I feel like I'm teaching kindergarten here, but these things are just tools. The responsibility rests with the decision maker. If the building falls down do we blame Home Depot where the construction tools were purchased?