Greenspan: Don't Blame The Nobel-Winning Models

I caught about 30 seconds of Greenspan's grilling on the Hill today, but it was a priceless half-minute.  He was responding to some politician's question about risk and derivatives on Wall St. and his answer was an apologetic version of Taleb's "black swan" thesis.  My earlier pieces here and in print explain that argument, but here it is in a nutshell: we can't reasonably measure risk in financial markets with models based on standard deviation.

Here's my paraphrase of the man who once said "If you think you understand what I said, I probably misspoke..."

The models that we used to evaluate securities and risk in financial markets were Nobel Prize-winning models.  It's just that we input the wrong data... since we were relying on market information from the past ten years only.

Sounds self-evident enough to anyone who passed freshman algebra.  Garbage in, garbage out... yeah, yeah, we get that.  What he means, of course, is that the models in question all use some form of that work-horse of statistics, standard deviation, to calculate returns and variance.  Modern Portfolio Theory (MPT), the Capital Asset Pricing Model (CAPM), and the Black-Scholes option pricing model are all built with components that require standard deviation as an input.

If you are new to the world of options, standard deviation in the option pricing model is simply volatility.

So, Greenspan is saying that Wall St. risk managers and derivatives dealers simply had the wrong volatility input in their models.  Oh, is that all.

Even he misses the point of the black swan.  The models don't work because nobody knows the right volatility for complex, long-term securities.  If you rely on ten or twenty years of data-even 100 years of data-the model still won't account for the outliers.

Why?  Because standard deviation by its very nature ignores its own fat tails.  Extreme events get washed out in the averaging.  Lots of people have lost a lot of money using "standard" deviation in the stock market now too, after the credit market explosions and implosions.  That's why the VIX has recently seen unprecedented volatility levels-investors bid wildly for puts so they would not be burned again.

What's the correct volatility input?  For options traders who either define their risk in advance or dynamically hedge, it doesn't matter much.  They can adapt in the short-term, and control the risk until expiration.

But for complex, long-term investments like mortgage-backed securities, collateralized debt obligations, and credit default swaps, the models are about as good as a Ouija  board.  Alan doesn't look like the type who sees dead people, but he's definitely still chasing modern finance's favorite ghosts-mathematical models that presume to define, reflect, and predict reality.  Oh, the spooky things quants and bankers can do with your money.