Sunday, December 28, 2008

How I Misrepresented Nassim Nicholas Taleb On Barbell Strategies

[long, wonkish, and personal]

If one googles "barbell strategy," one mostly finds discussions of a bond-trading strategy (and a related definition) that involves putting half of funds in long term bonds and half in short term bonds, with no money in intermediate term bonds. The money at both ends makes this a "barbell." Whether or not such a strategy makes more money than the "bullet strategy" of buying only intermediate term bonds (or all time horizon bonds) depends on the shape of the term structure of interest rates and how it changes over time. Another variation is for a takeover raider to buy both very large firms and very small ones, but not mid-sized ones. A more general definition of a barbell strategy is one that combines very safe investments with very risky ones, while not buying ones of intermediate riskiness. In his book, The Black Swan (TBS), Nassim Nicholas Taleb recommends (pp. 205-26) a variation on this in which one puts 85-90% of one's assets in "extremely safe assets, like Treasury bills" and the rest in "extremely speculative bets, as leveraged as possible (like options), preferably venture-style portfolio," with a footnote suggesting that this remaining 10-15% involve "as many of these small bets as possible."

In various blogs, including here, I have described his barbell strategy accurately with regard to the first part, but misrepresented the second part as involving specifically buying puts on major crashes. I then criticized this as a strategy that would lose money in most years (although obviously it would make money this year if properly done) based on a paper by Oleg Bonderenko, "Why are Put Options So Expensive," available at http://tigger.uic.edu/~olegb/research.htm. In a link on his website, http://fooledbyrandomness.com/fake.htm, Taleb has sharply criticized me for this (others dumped on there are Tyler Cowen, Robin Hanson, Alan Greenspan, Robert C. Merton, Kenneth Rogoff, and Paul Seabright, for various alleged sins, some more serious in my eyes than others, some not sins at all). In any case, I have now gone back to his book and see that my frequently repeated description of his barbell strategy was a misrepresentation. For this, I apologize to him publicly.



Needless to say, there is more to this than I have said so far, for those of you who are interested, and I will say that quite a bit of what Taleb says on his website involves serious misrepresentations. I would strongly suggest that he alter the inaccurate parts and publicly apologize, as I altered parts of an original posting back on maxspeak over a year ago at his request that was at the origin of all this. So, for those of you who are curious, here is the rest of the story, at least as I see it.

So, in the summer of 2007 I happened to read TBS. I found it mostly very interesting and stimulating, and in general I was then and remain in agreement with most of the ideas in it. I enjoyed (and still do) many of the tales and neologisms he came up with there, although on my second reading I find much of it more superficial and self-contradictory than I did the first time around, although this may reflect a more critical approach given our bad relations since. Anyway, after reading it, and becoming aware that he had a number of technical papers floating around on substantive aspects of it that were unpublished, I sent him a friendly email, praising his book and inviting him to submit any papers he might wish to the journal I edit, the Journal of Economic Behavior and Organization (JEBO). He has posted part of that email on his site.

What he does not note on his site is that his reply to me, somewhat delayed, was a form email telling me that he was too busy to reply to my email. I was not all that surprised by this as this was the time when TBS was on the bestseller list, and I could understand that indeed he might be very busy. He said nothing about JEBO and did not say anything about my invitation.

Now, there is at this point a crucial event that he did not realize. He argues that what followed was due to my "feeling rebuffed" that he had "refused [my] invitation to submit in his technical journal." Well, he had not refused. He had simply said he was too busy to reply, and I was in fact still awaiting a reply. However, I became aware of this paper by Bondarenko and saw some discussions on some blogs by some other people whom he does not attack and whom I shall not bring up who argued that Bondarenko's results showed that Taleb's barbell strategy was a money loser in most years, even if the losses would not be all that great, which is his ultimate defense of the strategy ("My barbell strategy has nothing to do with making money - although it does OK - but with being robust to model error"). It was after seeing this, and also hearing of some bizarre conduct of his regarding another blogger, whom I shall not mention, that I then put up a post on the old maxspeak, the predecessor of this blog.

In that post I spent most of my time praising TBS, something that Taleb has somehow never noticed or commented on, I guess assuming that everybody should praise his book without any questions or criticisms. I then, as had Tyler Cowen in a review in Slate (the reason Tyler is on his bad list), said that the main weakness of the book was that his barbell strategy would lose money most of the time. I went too far and said that he would probably not be able to make much money personally with it and would do better by writing and selling his successful books.

Well, now within 20 minutes of putting this up, I got an email from this guy who was "too busy" to reply to my friendly earlier email. It warned me in pretty strong language that I did not know how he made his money and that I should be careful what I said, pointing out specific wording mentioned above. So, I altered that wording to make the post not say what he definitely would do or not do in terms of making money, but still criticizing him. He then sent me another email in which he essentially threatened to sue me, which he has done to others who criticize his financial strategies (whatever else he is, he is ridiculously thin-skinned). I then put up his emails in the comments section of the post and ripped him harshly for his hypocrisy.

I now realize that part of what happened here was a miscommunication. He did not realize that I had changed my views somewhat on his book because of something I read after I sent my original email. That was unfortunate, because, as I said to him both on the blog and in an email, he and I are very close in views and admire and respect many of the same people, some of whom we share as friends. We actually ought to get along, but fell into a very unpleasant contretemps. In any case, I have continued to blast him here and there until now on all this.

A few substantive remarks beyond all this soap opera, and I shall let this go. One is that whether or not any barbell strategy either makes money or even succeeds in insuring one against big losses while preserving that option of making big money depends on the details of the strategy. He sells it as insurance against black swans (unforeseen events), but its ability to insure depends on having "safe assets" that will be safe even in the event of a black swan. In the case of his specific recommendation of US T-bills as the safe asset, while all the world has run to them in the last few months, it is now very far from clear how safe they are, at their near zero rates, and with the possibility of a major crash of the dollar sitting out there. Great grandma's old "cash in the cookie jar" may well be better, maybe along with some euros and yen and a couple of other currencies. A crash of US government securities could well be the next Black Swan.

The strategy again only will work if not too many people are doing it. If too many people are doing it, then the generalization of Bondarenko's critique will hold: those endpoint assets will be overpriced and money will be made by buying the intermediately risky ones. Also, if everybody does it, the economy will belly up with no financing of those intermediately risky activities that constitute the majority of the economy.

Furthermore, while it is very popular now to attack "economists" in general, Taleb rather makes a hash of things. On the broader part of his website he says somewhere that the only two economists he respects are dead: Hayek and Shackle. However, in TBS we find him praising Keynes, Knight, Minsky, and Kindleberger, all of whom I also admire. He also cites many living economists at least not critically, and some positively, with Robert Shiller probably most frequently. Yes, Merton and Scholes (and Samuelson) all look bad, and yes, the textbooks should stop pushing models based on Gaussian distributions, but by now most of us, and certainly the contributors to and readers of this blog know better.

Finally, he does remain in contradiction with himself in TBS, although he knows this and sort of acts like it is all very cute and philosophical (p. 296: "Half the time I am intellectual, the other half I am a no-nonsense practitioner... Half the time I am shallow, the other half I want to avoid shallowness," etc.). But he does have major contradictions, which make me less inclined to be so charitable about his slams on so many people over pretty trivial stuff (although not all of his slams are over trivial matters). So, he spends most of the book denouncing Gaussian distributions and those who push them (not very many these days, kind of a straw man). He seems to contrast that "Mediocristan" with "Black Swan" world of "Extremistan." But then, he spends time talking about econophysics and his papers with Mandelbrot on multi-fractal distributions and so forth, which he admits are not really either; they are "grey swans." Yes, he does cover his behind, but after all his rhetoric and carrying on, it looks pretty hypocritical.

A final btw. On his site he states that he did not wish to submit to JEBO because of his "no-nonsense orientation [that] clashes with academic resume building." Hmmm. Well, that might be fine, but I do see him publishing in American Statistician, Quantitative Finance (edited by our mutual friend, Jean-Philippe Bouchaud), and currently in the process of co-editing a special issue of the International Journal of Forecasting. Well, I wish him good luck with all that academic resume building (Oh, and I still think that he is a popularizer, although that is not necessarily a bad thing, but no, he will not get a Nobel for his ideas, sorry).

12 comments:

Ken Houghton said...

Good summary. (Only quibble: the link to Bonderenko should have ~olegb, not just ~oleg, in its penultimate position. (Corrected is this.)

rosserjb@jmu.edu said...

Thanks, Ken.

I have sent an email to Nassim telling him this is here, but he has yet to see fit to reply. I am not holding my breath that he will correct any of the outright errors on his website, not all of which I have listed here.

Motasim said...

Well all this maybe very valid. But, the fact still remains that the world has lost trillions through the hands of risk managers and courtesy their models based on gaussian distribution and other quantitative techniques.

As far as grey swans are concerned at least he should get credit for accepting and disclosing its limitations as it is impossible to have a financial model that is free from all the flaws. Remember three body problem...........

Barkley Rosser said...

Motasim,

I would say that Taleb has performed a useful publicizing function, hence my labeling of him as a "popularizer," which really ticks him off (see his website), although I have said that there is nothing wrong with that. The problem is that he wants to be all things to all people, the "no-nonsense trader," the popularizer who writes bestselling books that get him on TV, and the deep philosopher and mathematician coming up with new and innovative ideas of possible Nobel worthiness. He wants to be Paul Krugman, who sells lots of books, gets on TV, and also got a Nobel (although regular readers here will know that I have some problems with PK's haveing gotten the Nobel all by himself at this time, especially without citing people who generated the ideas he got his Nobel for).

Also, Taleb wildly overstates how much the recent techniques and models have relied on Gaussian distributions. Most of the more sophisticated modelers got out of that after the 1987 crash, with student t, Weibull, power laws of various sorts (which is what Taleb is pushing with Mandelbrot) being used by lots of actual traders and practitioners. Nobody serious has used the Black-Scholes formula one finds in a textbook for at least two decades. So, has done some wild exaggerating with all his ranting and denouncing. It has been a battle among a bunch of grey swans for some time now, and while he proposed a nice variation on the barbell one, it is not the end all or be all.

BTW, I cannot resist reporting that I did receive an email reply from Taleb, another form message, this one telling me he is "offline" and "without a blackberry," so I shall just have to wait to hear from him "dear correspondent." Been there, and waited before. Duh.

Nassim, if you are actually checking this out. Here is the deal. While I have apologized for certain things, I still think you are plenty guilty of plenty of garbage, including lies on your website. Communicate with me somewhere in a reasonable fashion, and I am prepared to stop going after you in the econoblogosphere. But, if I do not hear from you, or get some nonsensical b.s. from you as I did before, half-baked threats or your thug relatives from the Mionstrtry of the Interior in Lebanon showing up at my door, I shall continue to wipe your ass up all over the blogosphere. So, there it is. I am offering a truce, if not peace, but sitting there sending me form emails is not going to cut it. Been there, done that. Phooey.

Oh, and Motasim, I was writing about the implications of the three body problem for financial markets long before Taleb every did. Been at chaos theory for a very long time. Go check my vita for the stuff that is too old to be up on my website. Again, Taleb makes a nice popularizer of this stuff, but that is it.

Barkley Rosser said...

Since I am on my high horse here, I shall note that regarding the three body problem and chaotic dynamics in financial markets, I shall note that I am the person who first applied the term "chaotic bubbles" to them, back in 1991 in my book, From Catastrophe to Chaos: A General Theory of Economic Discontinuities, which was rejected by 13 publishers before Kluwer took it on. The term originated in 1987 in the fluid dynamics literature, where it continues to have its greatest use.

BTW, I do not really mind if Taleb keeps me on his "phonies" list, although I would prefer if he changed some of the outrigt lies there. Heck, while I think that Greenspan and Merton probably belong in jail, I agree with the stuff that Cowen, Seabright, and Hanson said that has him so ticked off. And Ken Rogoff's sin is apparently "wearing a coat and tie."

Now, I realize that Taleb likes to denounce "suits," and apparently never wears one. But I would point out that his friend Benoit Mandelbrot is almost never seen in public not wearing a suit, albeit one that is pretty rumpled.

Actually, while I understand that Taleb is really going after overpaid spokespeople for drivel on Wall Street when he invokes this, I have begun to find it amusing that claiming that wearing a suit shows that somebody is stupid has become very popular among econophysicists (physicists in general rarely wear suits). At an AEA session on econophysics, I saw a physicist, who will remain unnamed, who was wearing some very fancy designer label jeans and neckties, making this argument about why the econophysicists were better, or at least smarter, than the economists, with Mandelbrot sitting next to him in a suit. Oh well.

Barkley Rosser said...

Still have not heard from dear old Nassim. He is not raising his respect level in my eyes with this (non) performance.

Blissex said...

«the world has lost trillions»

The world has made and then lost trillions of paper gains and then losses.

«through the hands of risk managers and courtesy their models based on gaussian distribution and other quantitative techniques.»

Courtesy of their using whatever mathematical technique looked likely to underestimate risks and lead to higher papers profits and cash bonuses by requiring less than prudential reserves.

Barkley Rosser said...

As of late on New Year's Day, I have still not heard from Taleb via email. Just too busy.

Barkley Rosser said...

I understand from the papers that Taleb is now in a big fight with Merton. I agree with him on that.

Anonymous said...

looks like you are not on
http://fooledbyrandomness.com/fake.htm anymore.

Peter said...

I find it astonishing that anyone takes Taleb seriously, especially in regard to statements about optimization. I'm prompted to post a "refutation" of the bar-bell at this link, as if one were needed

http://www.quantapology.com/

Unknown said...

How can we judge the performance of a barbell without knowing the specifics of its construction?