Sunday, December 07, 2008

Taleb Blames VAR, Merton, Scholes for Crimes

You can't blame Taleb for arguing the same point over and over--he's a bestselling author and much wanted speaker, so supply is meeting demand. But like Naomi Klein, I think he plays into straw man caricatures that are not merely simplifications, but profoundly wrong (when not totally incoherent). In Taleb's case, his latest broadside is courtesy of the Financial Times and Co-author Pablo Triana:
A crime has been committed. Yes, we insist, a crime...Yet a method heavily grounded on those same quantitative and theoretical principles, called Value at Risk, continued to be widely used. It was this that was to blame for the crisis....Few people would agree that the illusion of knowledge is a good thing. Almost everyone would accept that the failure in 1998 of Long Term Capital Management discredited the quantitative methods of the Nobel economists involved with it (Robert Merton and Myron Scholes) and their school of thought called “modern finance”...Ask for the Nobel prize in economics to be withdrawn from the authors of these theories, as the Nobel’s credibility can be extremely harmful. Boycott professional associations that give certificates in financial analysis that promoted these methods. Remove Value-at-Risk books from the shelves – quickly.

Making the crew at RiskMetrics do a frogwalk is a funny thought (What are you in for, bud? Estimating a VAR for mortgages based on daily data during a period where housing prices were appreciating. Bastard!) Nobel Prize revocation? As they haven't even revoked the Nobel prize to the guy who invented Lobotomies as a cure for epilepsy (Egaz Moniz, 1949) or Arafat's Nobel Peace Prize (a guy who lectured the UN wearing a holster), I doubt Black-Scholes-Merton meets whatever standard is currently being applied.

The idea that LTCM was relevant to Merton and Scholes's seminal model is simply absurd. Long Term Capital Management (LTCM) went down because they became too big in simple trades that were about as related to option pricing as they were to the dividend discount model or the Fisher equation, and eventually the market had them by the balls, because they knew they had to get out at any price, and like Niederhoffer getting busted in 1997, the market will punish you for becoming too big. Thus, the main trades that took down LTCM were a Treasury-Eurodollar trade, a volatility trade (trade on volatility via over-the-counter derivatives), and directional bets in emerging markets. Scholes and Merton were primarily marketing props, not strategizers, and they were punished as investors lost something like 95% of their capital. Thus, perhaps Merton and Scholes let themselves be used by people who took too much risk, but the validity of the Black-Scholes/Merton option model was independent of LTCM's business model. The recent downturn also, is pretty independent of the Black-Scholes/Merton model.

Value at Risk is a broad concept, that basically is a framework that has as its goal estimating a 99% or 95% worst-case-scenario, over an arbitrary period of time (day, week, year). Now, clearly many people underestimated the risk in mortgage-backed paper of all sorts, but this error was not particular to Value at Risk, rather, the assumption that housing prices would not decline, which then lead to errors in your stress test, or whatever you call your 'worst case scenario'. This error was made at many levels, by investors, issuers, rating agencies, banks who warehoused these loans, regulators, etc. Again, Value at Risk is pretty independent of this error, though the error would be reflected in any VAR that had such erroneous assumptions (garbage in, garbage out). Further, Value at risk is pretty agnostic as to method, whether you use data with really fat tails, 99.99% worst case scenarios, whatever you want. Blaming Value at Risk for the latest crisis is like blaming soup, rather than sanitation, for Typhoid Mary.

My argument still stands: he's basically someone who makes the perfect the enemy of the good. Like a communist reveling in the Great Depression, Taleb may take heart in recent events vindicating his worldview, but as Taleb knows, there are always 'lucky fools' who merely were fortunate. I would like to see a risk management report from Taleb, perhaps a page always saying "risk=inf", because of course, we could lose everything in WW3, a new virus, all the banks fail, etc. Every assumption isn't true, so make no assumptions. Be prepared for anything. Go long volatility (especially extreme tails). I don't think that's very good advice, though clearly it worked great this year (and funds he's affiliated have done very well, being long tail volatility).

He is now emboldened by his popular success, and not amending his critique, rather thinking that he has financial academy on the ropes by the weight of his best-seller The Black Swan. Indeed, he notes "So how can we displace a fraud? Not by preaching nor by rational argument (believe us, we tried). Not by evidence." So much for reasoned debate. But regardless of popularity, there needs to be something more specific than his strident calls to stop using Value-at-Risk, option models, and portfolio theory from business schools. What sort of principles are we going to teach? Clearly reality is more complicated than these models, but there are some good ideas in there, and they do have many applications, though one need common sense in applying these tools fruitfully. Perhaps, MBA's should be taught common sense? That is, in some sense, the case-study approach of the Harvard Business School, which walks through various scenarios, with all their idiosyncrasies, and while I think this is very useful as part of a curriculum, it needs to be balanced with boring tools, like learning about covariance and how that affects portfolio risk.

I remember seeing Taleb's syllabus from the course he taught at UMass, and it seemed to be a bunch of papers (half by him) about various behavioral finance items. While any thread this big will have some good pieces, I don't think it could be the basis for learning about finance, or risk management, or pricing. As Eugene Fama says, half of behavioral finance is over-reaction, half under-reaction, about what you expect in an efficient market, but in any case, a risk manager should still understand options, value-at-risk and portfolio theory.


The Rioja Kid said...

Merton was pretty much the marketing department for LTCM, but Scholes did provide one fairly important innovation - he invented the Swapco system which LTCM use to manage the hypothecation and rehypothecation of its collateral. Since he didn't bother to simultaneously invent sensible standardised terms or back office procedures, he was actually responsible in large part for the demise of LTCM because the millefeuille of economically offsetting but legally distinct swap contracts created meant that it was nearly impossible for an external party to verify the solvency of the funds.

Plamen said...

A professor of Economics in a course I took said she had some "insider information" that the biggest bet that threatened the solvency of LTCM was a bet on the spread between Russian and Finnish interest rates, which did not shrink, as LTCM had expected, but actually broadened. Of course, about three months later the rates converged, and the bet was ultimately profitable, but it was a bit too late - LTCM had had to take the bailout money.

Agreed on Taleb, he hit the lottery, and now is running a lottery-investing niche industry, but hey, someone has to provide liquidity for deep out-of-the-money options too, let the guy pat his shoulder and blow kisses to his reflection in the mirror, and make money off him [grin].

Anonymous said...

Taleb is right. Wall St'ers aren't stupid but they can't see beyond what their wallet tells them. There will be another 1 in a 1000 years event soon enough and often enough. Position accordingly.