Confessions of an Econometrician

From FIN Alternatives (bold emphasis added)

A co-founder of Long-Term Capital Management said that the legendary collapsed hedge fund’s leverage ensured its fate.

LTCM collapsed in 1998—a year after Myron Scholes won the Nobel Prize in economics—forcing the Federal Reserve to arrange a bailout. At the time, it was the largest-ever hedge fund failure.

Scholes, who went on to found hedge fund Platinum Grove Asset Management, now says that the firm’s leverage doomed it in the wake of Russia’s sovereign debt default.

“LTCM ran leveraged positions at too-high risk levels,” Scholes told Risk. “It was not a sustainable business in the longer run if you have to reduce leverage and seek extra capital at a time when risk transfer costs are high.”

And this risks inherent in LTCM’s portfolio were higher than anyone realized at the time.

“It was a much higher-probability event than people thought, because it told people they were going to make 40% a year at 20% volatility—a high risk level,” Scholes said of LTCM’s demise. “The problem comes because, as a hedge fund, you don’t really have deep pockets, so it’s hard to run at a high risk for a long time.”

Scholes also blamed an over-reliance on classic portfolio theory.

“Capital models should give levels that are required to sustain the business at times of shock, and this is different for leveraged hedge funds because they can’t call for additional capital from investors,” he told Risk. “I believe capital models should not rely on portfolio theory, because the correlation structure is just not constant—in a crisis, you have intermediaries reducing risk simultaneously, so things that appeared to be independent clusters in the past become correlated, and diversification against those clusters does not provide staying power.”

In short, taking too much risk via leverage based on the assumption of the infallibility of econometric models.

As the great Ludwig von Mises warned (bold emphasis mine)

But it is not permissible to argue in an analogous way with regard to the quantities we observe in the field of human action. These quantities are manifestly variable. Changes occurring in them plainly affect the result of our actions. Every quantity that we can observe is a historical event, a fact which cannot be fully described without specifying the time and geographical point.

The econometrician is unable to disprove this fact, which cuts the ground from under his reasoning. He cannot help admitting that there are no "behavior constants."