How The Game Was Played And Why Long Term Lost

Its computer models didn't foresee how bad bad could be

The rocket scientists at Long-Term Capital Management wrote the book on how to use derivatives. Now, they can write the sequel on how to misuse them.

Long-Term didn't make money the old-fashioned way, by searching out assets whose inherent value was not reflected in the market. Instead, it bet on the interior dynamics of the markets themselves: exploiting anomalies in the price difference between, say, two bonds with different credit ratings or between shares of merging companies. Instead of using derivatives to hedge against risk, Long-Term mainly used them as enhancers--amplifying its gambles and ultimately amplifying its losses when bets proved wrong.

As it pursued this strategy, the hedge fund in Greenwich, Conn., was able to make bets on the markets entirely out of proportion with its financial resources. And now the bailout of the fund is being hampered by the difficulty of unwinding these complicated investment plays that consist, mainly, of bundle upon bundle of these investment products known as derivatives--the squishy value of which is based on the worth of underlying securities.

What's shocking about the screwup is thAt no one should have understood derivatives better than LTCM. Partners Myron S. Scholes and Robert C. Merton shared a 1997 Nobel Prize in economics for options-pricing theory, the basis of the derivatives business.

To be sure, the exact positions that Long-Term Capital holds remain secret. (Anyone who knows what the new owners plan to liquidate could profit by betting the other way.) Still, the general outlines of Long-Term's strategy are known. Here, in a nutshell, is how it played the game--and how it was able to pile up such huge stakes that Wall Street and the Federal Reserve ultimateLy decided it was too big to fail.

Long-Term's favorite bet was on the difference in the yield, or rate of return, between two debt instruments. Let's say that five-year junk bonds rated BB3 usually yield about two percentage points more than five-year Treasuries because of their higher risk. Long-Term's strategists might bet that if the yield spread ever got much wider than two percentage points, it was likely to narrow again soon.

Computer models would show that's a smart bet. But with the global economy in turmoil, investors lost their appetite for risk. Starting in July, they fled junk bonds and poured money into Treasuries. That pushed yields higher on junk bonds and lower on Treasuries. By late September, the yield spread had widened to a punishing 2.8 percentage points (chart). Instead of converging, the yields diverged.

NO ONE KNOWS. Long-Term Capital made matters worse by using derivatives to increase the size of its bets without borrowing. For instance, it could have made a bet such as the one above without putting up a penny. Instead of spending $100 million for a portfolio of junk bonds and shorting a like-size T-bond portfolio, Long-Term may have used a popular derivative known as a swap. In a swap, no money changes hands initially because typically the two parties start out by paying each other equal amounts. Long-Term would agree to pay the current yield on $100 million of junk bonds and would receive in turn the current yield on $100 million of Treasuries (plus some fixed percentage markup to make it a fair trade). The $100 million is called a "notional amount" since neither party owns the underlying securities.

How much did Long-Term Capital risk? By some estimates, it had bets with "notional amounts" of $1 trillion or more. But that overstates risks, since many of its positions essentially offset each other. While its balance sheet showed about $100 billion in assets on a capital base of less than $4 billion, its true risk came from derivative deals that don't appear on the balance sheet at all. The firms that took it over are still sorting out what they've inherited.

The beauty of a swap is that if the spread moves in your favor, you can make enormous profits without tying up any of your own money--in essence, your leverage is infinite. If the spread moves against you, though, you have to come up with correspondingly large amounts of collateral. Prudent firms won't enter huge swap deals unless they have capital adequate to withstand severe moves against them. Long-Term Capital's partners and its lenders apparently ignored that principle--or seriously underestimated how bad bad could be. Risk-assessment computer models "tend to ignore low-probability catastrophic events," says Henry T.C. Hu, a law professor and derivatives expert at the University of Texas.

Still, how could one small firm's bad bets threaten the entire U.S. financial system, to the point where the Federal Reserve Bank of New York brought together 14 major financial institutions to bail it out? Simple: Not knowing how much aggregate risk Long-Term Capital was taking on, these same firms extended enormous amounts of credit because Long-Term had always paid off in the past. "It's a bit like a junkie with crack. You gotta have more of this stuff," says a banking industry consultant. By mid-September, Long-Term owed so much money that its failure might have set off a worldwide chain reaction of creditors cutting off credit. In other words, a liquidity crisis.

Keeping Long-Term Capital on life support became even more important, since most derivatives aren't readily salable. One bet might employ half a dozen derivatives to hedge out various unwanted risks. Like the twigs in a bird's nest, each derivative is worth less alone than as part of a carefully built whole.


SHOT IN THE ARM. The injection of $3.6 billion in fresh capital into LTCM will allow it to meet its margin calls, easing fears of a ripple effect. And Long-Term Capital's new owners will be able to keep many of the firm's core bets in place with hopes that convergence will ultimately make the bets pay off.

On the other hand, some important spreads have actually widened in recent weeks. The Federal Reserve's Sept. 29 rate cut should help. But if new margin calls and money-losing asset sales eat away too much of its current rescue package, Long-Term Capital might need to be bailed out all over again.

Before it's here, it's on the Bloomberg Terminal.