How did two Nobel laureates lose $4.6 billion?

Barings was a control failure. Long-Term Capital Management was the opposite: the most sophisticated risk management in the world at the time, applied with rigour, and destroyed anyway. It is the more instructive case for exactly that reason.

Who and what

LTCM was founded in 1994 by John Meriwether, formerly head of bond arbitrage at Salomon Brothers. Its board included Myron Scholes and Robert Merton, who in 1997 received the Nobel Memorial Prize in Economic Sciences for the option-pricing work that underpins the modern derivatives market.

The strategy was relative value: identify two closely related instruments whose prices had drifted slightly apart, buy the cheap one, sell the rich one, and wait for the gap to close. A representative trade was buying a slightly older, less liquid US Treasury bond and shorting the newly issued one, capturing the small premium paid for liquidity as it decayed.

These trades are genuinely low-risk per unit. They are also barely profitable per unit — a fraction of a percent. There is only one way to turn a fraction of a percent into a hedge-fund return.

The leverage

At the start of 1998, LTCM had roughly $4.7 billion of equity supporting about $125 billion of balance-sheet assets — 26.6:1 — plus derivative positions whose notional was frequently quoted above $1 trillion.

Apply the formula from Unit 3. At L = 26.6, equity change = 26.6 × asset change, and the ruin point is 1 ÷ 26.6 ≈ 3.8%. A move of under 4% across the asset book erases the fund. Every model said such a move was effectively impossible, because the positions were hedged against direction and diversified across dozens of markets and countries.

August 1998

On 17 August 1998 Russia defaulted on its domestic debt and devalued the rouble. What followed was a global flight to quality: investors everywhere simultaneously sold anything less liquid and bought the safest, most liquid instruments available.

Every one of LTCM's spreads was, in economic substance, a bet that the premium for illiquidity would narrow. All of them widened at once. The diversification across countries, asset classes and instruments turned out to be diversification across one shared factor — liquidity — that had simply never moved in the calibration sample.

Losses came to approximately $4.6 billion in under four months, wiping out substantially all of the fund's capital — against the $4.7 billion it began the year with, more than 97%. The positions were also too large to exit: attempting to sell moved prices further against the remaining book. Size, which had been the source of the returns, became the reason the exit was closed.

On 23 September 1998 the Federal Reserve Bank of New York convened a consortium of 14 financial institutions, which injected about $3.6 billion for 90% of the fund, allowing an orderly wind-down. No public money was used; the concern was that a forced liquidation of that book would damage the market itself. That concern — that a private fund's positions could be systemically relevant — was new, and it has shaped regulation ever since.

What actually caused it

1. Leverage converted a normal event into insolvency. A 3.8% adverse move was fatal. Spreads widened by considerably more than that in relative terms. The trades were not wrong — most eventually converged, and the consortium wound the book down at a profit. LTCM simply could not survive the interval.
2. Models calibrated on a period without the event. Value-at-Risk built on five years of data cannot see a liquidity crisis that did not occur in those five years. The model was not miscalculated; it was answering a question about a world that was about to change.
3. Correlation assumptions failed precisely when needed. Positions diversified in normal conditions became a single position in stressed ones. This is the most repeated lesson in financial history and the least internalised.
4. Size removed the exit. Liquidity is not a fixed property of a market; it is a property of a market and the size trying to move through it.

Note the pattern with Barings: different failure mode, same two ingredients — leverage, and a risk nobody could see in the reported numbers.

Try it now

1. Compute the ruin point 1 ÷ L for leverage of 10, 25 and 50. Note how small the fatal move becomes, and how little the leverage number has to rise to halve it.
2. The table below follows two US Treasury yields, the 3-month bill and the 30-year bond, through four days of the second half of 1998. Work out how far each moved from July to 8 October, and what happened to the gap between them.

3. Write one sentence explaining how a portfolio can be "diversified" and still lose everything at once. The word you need is correlation, and the condition you need is stress.