When Genius Failed

On September 23, 1998, William J. McDonough, president of the Federal Reserve Bank of New York, summoned the heads of every major Wall Street bank to the Fed's boardroom. For the first time, the chief executives of Bankers Trust, Bear Stearns, Chase Manhattan, Goldman Sachs, J. P. Morgan, Lehman Brothers, Merrill Lynch, Morgan Stanley Dean Witter, and Salomon Smith Barney gathered to consider rescuing a single hedge fund: Long-Term Capital Management (LTCM), a private bond-trading partnership based in Greenwich, Connecticut. LTCM had amassed roughly $100 billion in assets, virtually all borrowed, and entered into derivative contracts, essentially side bets on market prices, covering more than $1 trillion in exposure. If the fund defaulted, every bank would be left holding one side of a contract for which the other side no longer existed, potentially triggering a panic that could freeze the entire financial system. When Bear Stearns CEO James Cayne flatly refused to invest a nickel, the room exploded.

LTCM's origins traced to John W. Meriwether, who grew up in an Irish Catholic neighborhood on Chicago's South Side, excelled at math, and earned a golf scholarship to Northwestern University. After teaching high school math and earning an M.B.A. from the University of Chicago, Meriwether joined Salomon Brothers in 1974 just as deregulation, inflation, and computers were transforming bond markets. In 1979, he took over the losing position of a failing securities dealer in Treasury bill futures. Despite weeks of further losses, the prices converged and Salomon profited handsomely, teaching Meriwether a formative lesson: if a spread trade is properly conceived, ride the losses until they turn into gains.

In 1977, Meriwether had formed the Domestic Fixed Income Arbitrage Group within Salomon and recruited academics from top universities, including Eric Rosenfeld from Harvard, Victor Haghani from the London School of Economics, and Lawrence Hilibrand from MIT. The academics downloaded historical bond prices into computers, identified when market prices deviated from predicted patterns, and exploited the discrepancies. Meriwether created an insular, fiercely loyal culture around the group. As profits swelled, so did confidence; Hilibrand in particular doubled down on losing bets, believing markets always revert. In 1991, a trader named Paul Mozer confessed to Meriwether that he had submitted false bids to the U.S. Treasury. When the scandal erupted, CEO John Gutfreund was forced to resign, and Warren Buffett took over as interim CEO. Under pressure, Meriwether himself resigned, settled a civil complaint with the Securities and Exchange Commission (SEC), and began planning an independent fund.

By the early 1990s, hedge funds, private and largely unregulated investment pools open only to wealthy individuals and institutions, had proliferated to roughly 3,000 in the United States. Meriwether envisioned LTCM as a "relative value" bond fund that bought undervalued bonds and shorted overvalued ones, profiting from tiny spreads between pairs of securities. Because these spreads were small, the fund planned to leverage its capital—borrowing money to multiply its bets—twenty to thirty times to amplify returns. Meriwether enlisted Merrill Lynch to raise capital and recruited two brilliant academics: Robert C. Merton, a Harvard scholar who pioneered "continuous time finance," and Myron Scholes, co-inventor of the Black-Scholes option-pricing formula, a widely used equation for pricing stock options. He also recruited David W. Mullins, vice chairman of the Federal Reserve, giving LTCM unparalleled access to institutional investors worldwide. The fund opened in late February 1994 with $1.25 billion, the largest start-up ever.

LTCM's timing proved ideal. In February 1994, the Fed raised short-term interest rates, triggering a bond sell-off that widened spreads and created the opportunities LTCM's strategy was designed to exploit. From the start, the fund refused to pay the standard "haircut," the extra margin collateral banks normally require, and virtually every major bank acquiesced. Haghani, running the London office, made Italy the fund's biggest single-country bet, exploiting the gap between Italian government bond yields and swap rates, essentially wagering that Italy's credit risk would improve. An insider estimated that an Italian default could have cost half the fund's capital, a risk never disclosed to investors. LTCM earned 28 percent in its first year and a stunning 59 percent before fees in 1995.

The fund's risk framework rested on historical volatility: by plugging past bond prices into computers, the partners derived expected future fluctuations and calibrated leverage accordingly. This approach assumed prices followed a bell-curve distribution with stable volatility. Yet Eugene Fama, Scholes's thesis adviser, had documented that markets exhibit far more extreme movements than the bell curve predicts, with "fat tails" that make them inherently riskier. Despite such warnings, virtually every investment bank adopted similar models.

By spring 1996, LTCM had $140 billion in assets, thirty times its underlying capital. The fund earned $2.1 billion in profits in 1996, more than McDonald's or Merrill Lynch. But competition intensified as imitators piled in, and LTCM expanded aggressively into unfamiliar territory: Haghani bet $2.3 billion on the spread between Royal Dutch Petroleum and Shell Transport; Hilibrand entered merger arbitrage despite fierce internal opposition from Scholes and Merton. In September 1997, Meriwether forced outside investors to take back $2.7 billion, causing leverage to jump from 18 to 1 to 28 to 1. That October, Merton and Scholes won the Nobel Memorial Prize in Economic Science, though both were privately uneasy about the fund's direction.

In early 1998, LTCM began selling enormous quantities of equity volatility, shorting options on the S&P 500 and major European indices. Morgan Stanley nicknamed the fund the Central Bank of Volatility. Credit spreads, the yield difference between risky bonds and safe government bonds, narrowed to historic lows by April, the fund's high-water mark. Then Salomon Smith Barney closed its bond arbitrage desk, and because Salomon held positions overlapping LTCM's, the liquidation pushed spreads wider and triggered the fund's first sustained losses.

On August 17, Russia declared a debt moratorium, shattering the assumption that nuclear powers do not default. Markets worldwide buckled. Swap spreads, the premium of swap yields over government bond yields and a gauge of credit risk, surged. LTCM lost $553 million in a single day, a loss its models had deemed virtually impossible. By month's end, the fund had lost $1.9 billion in August alone. With $2.28 billion in equity against $125 billion in assets, leverage exceeded 55 to 1. The partners frantically sought capital, but George Soros's conditional offer expired, Buffett declined, and no rescue materialized.

In September, losses accelerated as rivals and banks, aware of LTCM's positions, sold in advance of an anticipated liquidation. By September 21, another single-day loss of $553 million left the fund with under $1 billion against over $100 billion in assets. Peter Fisher of the New York Fed visited LTCM's offices and reviewed its exposures, realizing the fund had made the same spread trade in virtually every market worldwide. He estimated a disorderly collapse could cost the fund's biggest counterparties $3 billion to $5 billion.

On Wednesday, September 23, Berkshire Hathaway, AIG, and Goldman Sachs faxed an offer to buy the fund for $250 million, inject $3.75 billion, and fire the partners, with a deadline of 12:30 P.M. LTCM's general counsel James Rickards identified structural problems with the bid, and Buffett was unreachable in Alaska. The bid was withdrawn. That afternoon, Herbert Allison of Merrill Lynch drafted a consortium plan calling for each bank to invest $300 million. Bear Stearns's Cayne refused to participate, provoking fury, while Goldman Sachs CEO Jon Corzine, under pressure from unhappy partners, nearly withdrew but ultimately committed.

Over the following weekend, tense negotiations unfolded at the law firm Skadden, Arps, where 140 lawyers hammered out the terms while the partners waited in a separate conference room. Goldman threatened to quit unless Chase Manhattan waived its right to repayment of a $500 million syndicate loan; after an explosive confrontation, Chase conceded. On Monday, September 28, with markets still falling, Hilibrand, hopelessly in debt from personal borrowing, read the contract through tears and initially refused to sign. Meriwether persuaded him the group needed him. Hilibrand signed, and fourteen banks took over the fund, investing $3.65 billion in exchange for 90 percent of the equity.

The partners personally lost $1.9 billion; most lost 90 percent or more of their wealth. The portfolio continued to plummet before Fed interest rate cuts stabilized markets. LTCM earned 10 percent in the year after the bailout before the consortium redeemed its capital. Lowenstein argues that LTCM's failure exposes a flaw at the heart of modern finance: the belief that tomorrow's risks can be reliably inferred from yesterday's prices. Diversification proved illusory when every trade amounted to the same bet; leverage magnified the consequences; and no independent risk oversight existed to check the dominant traders. In December 1999, fifteen months after losing $4.5 billion, Meriwether raised $250 million for a new fund, much of it from former LTCM investors, and was off and running again.