WHEN GENIUS FAILED: The Rise and Fall of Long-Term Capital Management. - Review - book review

Washington Monthly, Sept, 2000 by James K. Galbraith

WHEN GENIUS FAILED: The Rise and Fall of Long-Term Capital Management

By Roger Lowenstein Random House, $26.95

Capital Mismanagement

Nobel prizes and billions of dollars don't always equal success

ONE DAY IN THE EARLY FALL OF I975 I SAT in the office of Rep. Henry S. Reuss (D-Wis.), Chairman of the House Banking Committee, for whom I then worked, along with Governor Hugh Carey of New York and Felix Rohatyn, Peter Goldmark, and David Burke of Carey's emergency team. Our purpose was to nail down a legislative plan to prevent the impending bankruptcy of the city of New York.

Toward the end of the meeting, Reuss asked me whether I had anything to add. "What about the windfall problem?" I asked. New York City bonds were wading at 30 cents on the dollar (or so I recall). When the bill went through, they would jump and anyone who bought them as we spoke would make a fortune. Carey turned to Rohatyn, "What about that, Felix?" Rohatyn shrugged, "Well, he's right."

Little did I know. The speculator existed, and his name, as we learn from Roger Lowenstein's genial account of Long-Term Capital Management, When Genius Failed, was John Meriwether. The New York crisis marked the takeoff of Meriwether's career, which went from bailout to bailout over 23 very exciting years.

When Genius Failed is a very good account of LTCM: clear, entertaining, informative, and judicious. The narrative has drama; the fate of the financial world was briefly at stake. It has complexity; there are serious issues of mathematics, statistics, and social science behind this story. Most of all, there is here a haunting portrait of our financial culture, where very ordinary people control extraordinary amounts of money.

Meriwether, pathologically self-effacing, emerges not unsympathetically in Lowenstein's portrait, except of course that he was crazily unsuited to running a financial firm. It is the other partners who fascinate, above all the professors Robert Merton and Myron Scholes, the fund's "philosophical fathers." The professors, we learn here, were peripheral to trading operations (Scholes, a "lesser partner" was "forever angling for more money"), but they were central to the marketing campaign. In this respect, their shared 1997 Nobel Memorial Prize didn't hurt at all.

Long-Term's basic strategy was to bet on the eventual convergence between the prices of extremely similar assets (the archetypal case being 30-year Treasury bonds issued today, "on the run," and the same bonds issued six months ago, "off the run"). By buying the cheap and selling the dear, convergence would bring a profit no matter the common movement of bond prices. In principle, it was a low-risk strategy, with tiny returns on each trade. And Long-Term made up the difference by leverage--by borrowing 10, 20, 50 times capital.

Leverage multiplies risk, but Long-Term believed that the underlying risk was so low that multiplication by leverage would not matter. This belief stemmed from a critical assumption: that the spreads between two essentially identical assets would be randomly distributed under a normal (bell) curve. Thus, the probability of a spread between highly similar assets moving too far in the wrong direction--of the cheap asset falling in price while the dear one rose, would be very slight and could be precisely calculated from the historical record.

Lowenstein presents the statistical issue competently, given that his own understanding is not deep and he expects his readers to have none at all. But he underplays a key issue. Economists have known for decades that the assumption of normality in the distribution of future events is false, particularly in financial markets. In 1937, John Maynard Keynes made the distinction between risk and incalculable uncertainty the foundation stone of his theoretical revolution. But Merton, Scholes, and the others came from a branch of economics that had rejected Keynes. That was their first mistake.

Worse, the fact that the distribution of asset returns is not normal was well known even in Chicago's financial-theory circles in the 1960s. Scholes' teacher Eugene Fama knew it; the case was first made to Fama by the mathematician Benoit Mandelbrot, father of fractals. Asset returns, Mandelbrot argued, follow a Pareto-Levy distribution. They have (as Lowenstein puts it) "fat tails." Infinite variance. This means that catastrophic events are likely to happen all the time.

Fama and Mandelbrot should have rung down the curtain on the "efficient-markets hypothesis" 30 years ago. But economists would not give it up. Dumping the bell curve would have meant abandoning the use of econometrics in finance, and hence nullifying a vast body of published work. Careers would have been compromised, academic empires diminished. And then along came Merton and Scholes, who forgot the difference between an academic game and real life.

They also did not notice a key fact about their own operation. By buying cheap and selling dear, they were always betting on convergence. In this arcane way, all of their trades--no matter what they were superficially about, whether Italian bonds or common stocks and options--shorted the asset perceived to be safe and bought the asset perceived to be risky. As Lowenstein puts it: "It apparently did not occur to Rosenfeld that since Long-Term tended to buy the less liquid security in every market, its assets were not entirely independent of one another." And so they exposed themselves to the classically Keynesian event: a flight to quality. In 1998, the event occurred. Spreads all went the wrong way. LTCM lost all its capital in about five weeks.