1998 LTCM Crisis
1998 LTCM Crisis
In 1998, Long-Term Capital Management, a hedge fund run by Nobel Prize winners and the best quants on Wall Street, lost $4.6 billion in four months as its mathematical models failed to account for the correlations that emerged when Russia defaulted, forcing the Federal Reserve to orchestrate a $3.6 billion government-backed rescue to prevent global financial collapse.
Key takeaways
- LTCM used leverage ratios exceeding 25-to-1 to execute relative-value trades that exploited small price differences between bonds.
- Russia's default in August 1998 caused yield spreads to widen far beyond LTCM's models' predicted bounds, creating losses that overwhelmed the fund's capital.
- Correlation, which LTCM's risk models assumed would remain stable, broke down catastrophically; all markets sold off together.
- The Fed coordinated a $3.6 billion rescue funded by 14 major banks to prevent LTCM's bankruptcy from spreading systemic damage.
- The crisis revealed that models based on historical correlations can fail when the world changes, and that even the smartest strategies can fail under leverage.
The Fund That Was Too Smart to Fail
Long-Term Capital Management was not a typical hedge fund. Its founders included John Meriwether, a legendary Salomon Brothers bond trader; Robert Merton and Myron Scholes, who had won the Nobel Prize in Economics for their Black-Scholes options-pricing model; and James Relax, another academic giant. The fund's pitch was seductive: use cutting-edge mathematics to find mispricings in fixed income and derivatives that the market had overlooked.
The strategy was not stock-picking or macroeconomic forecasting. It was relative-value trading. LTCM identified pairs of bonds—say, US Treasuries and Treasury strips (zero-coupon bonds derived from Treasuries)—that historical analysis showed should trade at a predictable relationship. If that relationship temporarily widened beyond normal, LTCM would buy the cheap one and short the expensive one, betting the relationship would snap back.
The beauty of this strategy was that it appeared market-neutral. LTCM was not betting on rates going up or down; it was betting on technical mispricings disappearing. The risks, on paper, were small. And the returns—driven by leverage and the compounding of small, consistent profits—were stunning. In 1995, LTCM returned 19.9%. In 1996, 40.8%. In 1997, 17.1%. Investors believed they had found a perpetual motion machine.
Leverage as a Leveling Instrument
By early 1998, LTCM had approximately $5 billion of investor capital. To scale the relative-value strategy, the fund had borrowed heavily. The balance sheet showed approximately $125 billion in notional positions. The leverage ratio—total assets divided by equity—exceeded 25-to-1. This was not unusual for a fixed-income trading operation, but it concentrated risk in ways that were invisible to LTCM's partners at the time.
The strategy relied on one critical assumption: that correlations between market segments would remain relatively stable. A widening in, say, the spread between General Electric bonds and Treasury bonds would not occur simultaneously with widening in swap spreads, on-the-run versus off-the-run Treasury spreads, and credit spreads. The models said these moves were independent or negatively correlated. History supported this view.
But correlation is a two-way street. When correlations rise—when, instead of moving independently, asset classes begin moving together—diversification stops working. And leverage, which is magnificent during calm periods, becomes a destroyer when correlations spike.
Russia Breaks the System
In August 1998, Russia defaulted on its domestic debt. The immediate impact on US Treasuries was zero—Russia owed money to its own currency holders, not the US government. But Russia's default triggered a global re-risk event. Investors worldwide, suddenly worried about credit risk and emerging-market exposure, began selling any asset that wasn't ultra-safe. And ultra-safe meant US Treasuries.
This should have been LTCM's trade in reverse. But it wasn't. Instead, everything sold off together. Emerging-market bonds plunged. High-yield credit spreads blew out. Even Treasury strips—supposedly safe—widened relative to current-coupon Treasuries. The "carry trade" unraveled: investors who had borrowed cheap yen or dollars to buy higher-yielding assets began unwinding simultaneously.
On-the-run Treasury spreads, which LTCM was long, widened to levels the models said were impossible. The fund was losing money on its supposedly independent trades simultaneously. A $5 billion fund with 25-to-1 leverage was now facing losses of $100+ million per day.
Within four months, LTCM had lost $4.6 billion of its $5 billion in capital. The fund was near insolvency. Because LTCM had counterparty relationships with virtually every major bank on Wall Street—via the swap dealers, repo desks, and derivatives dealers who had financed the strategy—the threat was not just to the fund. It was to the global financial system.
If LTCM filed for bankruptcy, its counterparties would have to liquidate $125 billion of collateral into a market that was already dislocated. The cascade of forced selling would have spread losses across the banking system. Several large banks had direct exposure to LTCM; others would be hurt by the liquidation flows.
The Federal Reserve Steps In
On September 23, 1998, Federal Reserve Chair Alan Greenspan, Treasury Secretary Robert Rubin, and the presidents of the Federal Reserve Banks convened in New York. The decision was clear: LTCM could not be allowed to fail. But the Fed itself could not legally inject capital into a private firm.
Instead, the Fed coordinated a rescue: 14 major banks (including Merrill Lynch, Goldman Sachs, Lehman Brothers, JP Morgan, and others) would inject $3.6 billion of capital into LTCM in exchange for a 90% ownership stake. The banks would then unwind the fund's positions in an orderly fashion over weeks and months, rather than dumping them onto the market in a panic.
The rescue was not a gift. The consortium of banks took losses—on the order of hundreds of millions—but those losses were far smaller than the systemic losses would have been. The Fed also cut interest rates 50 basis points to inject liquidity into the system.
The key point: in 1998, the Fed determined that the failure of a private hedge fund was a systemic risk to the financial system. This decision would be controversial in the years following but was vindicated in 2008 when systemic risk became the dominant concern.
The Model Failure
LTCM's catastrophe was not a failure of mathematics per se; it was a failure of assumptions. The Black-Scholes model, which Merton and Scholes had won the Nobel Prize for, was elegant and theoretically sound. But it assumed that market prices followed a normal distribution with finite variance. In reality, on days like August 1998, markets experienced "tail events"—outcomes multiple standard deviations away from the historical mean.
The models also assumed that historical correlations would persist. They did not. When fear spread, correlations moved toward 1 (all assets moved together). The diversification that looked real in calm markets was illusory in stress.
A second issue was liquidity. LTCM's models had assumed that positions could be liquidated at prices close to the last quote. But in August-September 1998, spreads widened dramatically. The bid-ask spread on mortgage-backed securities, normally 1/32, blew out to 1/8. The illiquidity cost of exiting positions was an order of magnitude higher than the models had predicted.
Lessons for Bond Investors
The LTCM crisis had several implications for bond market participants. First, leverage amplifies both gains and losses. A 25-to-1 leverage ratio meant that a 4% loss became a 100% loss to equity. Second, models should be stress-tested against tail events and should never assume that correlations are stable. A strategy that works in calm markets can fail catastrophically in stress.
Third, counterparty risk became visible. LTCM had relationships with dozens of banks; a failure would have spread directly. This lesson would resurface in 2008, when the interconnectedness of the banking system became the central regulatory concern.
For individual bond investors, the LTCM crisis reinforced that even Nobel Prize winners and brilliant mathematicians could misjudge risk. A bond fund that returned 40% one year (as LTCM did in 1996) was taking risk that was not fully captured in traditional metrics. Backtesting and historical analysis can be misleading if they do not account for rare but catastrophic events.
The Relative-Value Trade Lives On
Ironically, LTCM-style relative-value trading did not disappear after 1998. The strategy was sound; the execution was too leveraged. In the years following, quant hedge funds, bank proprietary desks, and mutual funds continued to exploit small mispricings in bonds, but with more conservative leverage and better stress testing.
The regulatory response, however, was limited. The 1998 crisis led to some heightened disclosure requirements for hedge funds and a focus on counterparty risk, but large-scale regulatory overhaul did not occur. The banks absorbed their losses and continued their prop trading. The Federal Reserve's moral-hazard concern—that bailing out LTCM would encourage future risk-taking—was noted but not acted upon vigorously.
This would be a problem a decade later, when leverage had once again accumulated in the financial system, this time in mortgage-backed securities and the shadow banking sector.
Decision tree
Next
The bond market would enjoy relative calm for most of the 1990s and 2000s, until 2008, when mortgage-backed securities—the most fundamental asset class in fixed income—would unravel and threaten the solvency of the banking system itself.