quiet. He welcomed frank debates among the partners, but he usually chimed in only at the very end or not at all.
The firm’s headquarters were the ground floor of a glassy four-story office complex, on a street that ran from the shop-lined center of affluent Greenwich past a parade of Victorian homes on Long Island Sound. Several dozen of Long-Term’s growing cadre of traders and strategists worked on the trading floor, where partners and nonpartners sat elbow to elbow, cramped around a sleek, semicircular desk loaded with computers and market screens. The office had an elaborate kitchen that had been put in by a previous tenant, but the partners lunched at their desks. Food meant little to them.
J.M., Merton, and Scholes (the latter two because they didn’t trade) had private offices, but J.M. was usually on the trading floor, a mahogany-paneled room that looked out through a full-length picture window to the water, resplendent and often speckled with sailboats. Aside from the natty Mullins, the partners dressed casually, in Top-Siders and chinos. The room hummed with trader talk, but it was a controlled hum, not like the chaos on the cavernous New York trading floors. Only occasionally did the partners revert to their past life for a few rounds of liar’s poker.
Besides the Tuesday risk meetings, which were for partners only, Long-Term had research seminars on Wednesday mornings that were open to associates and usually another meeting on Thursday afternoons, when partners would focus on a specific trade. Merton, usually in Cambridge, would join in by telephone. The shared close quarters fostered a firm togetherness, but the associates and even some of the partners knew they could never be part of the inner circle. One junior trader perpetually worried that his trades would be found out by the press, which he feared could cost him his job. Associates in Greenwich, even senior traders, were kept so much in the dark that some resorted to calling their London counterparts to find out what the firm was buying and selling. Associates were never invited back to the partners’ homes—there seemed to be an unwritten rule against partners and staff fraternizing. Leahy, a college hockey player, exchanged the normal office banter with the employees, but most of the partners treated the staff with cool formality. They were polite but interested only in one another and their work. The analysts and legal and accounting staffs were second-class citizens, shunted to a room in the back, where the pool table was.
Like everyone else on Wall Street, Long-Term’s employees made good money. The top staffers could make $1 million to $2 million a year. There was subtle pressure on the staff to invest their bonuses in the fund, but most of them were eager to do so anyway—ironically, it was considered a major perk of working at Long-Term. And so, the staff confidently reinvested most of their pay.
Just as predicted, Long-Term’s on-the-run and off-the-run bonds snapped back quickly. Long-Term made a magical $15 million—magical because it hadn’t used any capital. As Scholes had promised, Long-Term had scooped up a nickel and, with leverage, turned it into more. True, many other firms had done the same kind of trade. “But we could finance better,” an employee of Long-Term noted. “LTCM was really a financing house.”
Long-Term preferred to reap a sure nickel than to gamble on making an uncertain dollar, because it could leverage its tiny margins like a high-volume grocer, sucking up nickel after nickel and multiplying the process thousands of times. Of course, not even a nickel bet was absolutely sure. And as Steinhardt, the fund manager, had recently been reminded, the penalty for being wrong is infinitely greater when you are leveraged. But in 1994, Long-Term was almost never wrong. In fact, nearly every trade it touched turned to gold.
Long-Term dubbed its safest bets convergence trades, because the instruments matured at a specific date, meaning that convergence appeared to be a sure thing. Others were known as relative value trades, in which convergence was expected but not guaranteed.16
The bond market turmoil of 1994 seeded the ground for a huge relative value trade in home mortgage securities. Mortgage securities are pieces of paper backed by the cash flow on pools of mortgages. They sound boring, but they aren’t. Some $1 trillion of mortgage securities is outstanding at any given time. What makes them exciting is that clever investment bankers have separated the payments made by homeowners into two distinct pools: one for interest payments, the other for principal payments. If you think about it—and Long-Term did, quite a bit—the value of each pool (relative to the other) varies according to the rate at which homeowners pay off their loans ahead of schedule. If you refinance your mortgage, you pay it off in one lump sum—that is, in one giant payment of principal. Therefore, no further cash goes to the interest-only pool. But if you stand pat and keep writing those monthly checks, you keep making interest payments for up to thirty years. Therefore, if more people refinance, the interest-only securities, known as “IOs,” will fall; if fewer people prepay, they will rise. The converse is true for principal-only securities, or POs. And since the rate of refinancings can change quickly, betting on IOs or POs can make or lose you a good deal of money.
In 1993, when Long-Term was raising capital, America was experiencing a surge of refinancings. With mortgage rates dropping below 7 percent for the first time since the Vietnam War, baby boomers who had never given their mortgages a second thought were suddenly delirious with the prospect of cutting their payments by hundreds of dollars a month. Getting the lowest rate became a point of pride; roughly two in five Americans refinanced in that one year—in fact, some folks did it twice. Naturally, the prices of IOs plummeted. Actually, they fell too much. Unless you assumed that the entire country was going to refinance tomorrow, the price of IOs was simply too low. Indeed, Meriwether, Hilibrand, Rosenfeld, Haghani, and Hawkins bought buckets of IOs for their personal accounts.
In 1994, as Long-Term was beginning to trade, IOs remained cheap for fear there would be another wave of refinancings. Bill Krasker had designed a model to predict prepayments, and Hawkins—an outgoing, curly-haired mortgage trader with backwoods charm—continually checked the model against the record of actual prepayments. The IO price seemed so out of whack that Hawkins wondered, “Is there something wrong with the model, or is this just a good opportunity?” The methodical Krasker carefully retooled the model, and it all but screamed, “Buy!” So Long-Term—once again with massive leverage—started buying IOs by the truckload. It acquired a huge stake, estimated at $2 billion worth.
Now, when interest rates rise, people aren’t even going to think about refinancing. But when rates fall, they run to the mortgage broker. That means that IOs rise and fall in sync with interest rates—so betting on IOs is like betting on interest rates. But the partners didn’t want to forecast rates; such outright speculation made them jittery, even though they did it on occasion. Because interest rates depend on so many variables, they are essentially unpredictable. The partners’ forte was making highly specific relative bets that did not depend on broad unknowns.
In short, the partners merely felt that, given the present level of interest rates, IOs were cheap. So the partners shrewdly hedged their bet by purchasing Treasurys, the prices of which move in the opposite direction from interest rates. The net effect was to remove any element of rate forecasting. The partners excelled at identifying particular mispriced risks and hedging out all of the other risks. If Haifa oranges were cheap relative to Fuji apples, they would find a series of trades to isolate that particular arbitrage; they didn’t simply buy every orange and sell every apple.
In the spring, when interest rates soared, Long-Term’s IOs also soared, although its Treasury bonds, of course, fell. Thus, it was ahead on one leg of its trade and behind on the other. Then, in 1995, when interest rates receded, Long-Term’s Treasurys rose in price. But this time people did not rush to refinance, so Long-Term’s IOs held on to much of their former gains. Presumably, people who had gotten new mortgages in 1993 were not so eager to do it again. Long-Term made several hundred million dollars. It was off to a sizzling start.
Despite appearances, finding these “nickels” was anything but easy. Long-Term was searching for pairs of trades—or
