LTCM’s story was told in detail not long after the fact in an excellent book by Roger Lowenstein, When Genius Failed. The reason to revisit the story now is to show how the 2008 panic was foretold by 1998’s events. Circumstances were different in 1998 and 2008, yet the dynamics were the same. Disturbingly, the next panic is now foretold by both 1998 and 2008. No lessons were learned. Elites simply expanded the bailout each time. Except next time the panic will be too large, and the bailout too small to stop it.
Ingredients like overleverage, derivatives, and reliance on obsolete risk models were identical in 1998 and 2008. The 2008 collapse could have been avoided if the lessons of LTCM were learned and applied. Instead, Wall Street and Washington turned a blind eye to what occurred in 1998. Policymakers including Fed chair Alan Greenspan and Treasury Secretary Larry Summers persisted with their belief in flawed risk models. Rather than learn lessons, Greenspan and Summers doubled down by supporting Glass-Steagall repeal and derivatives deregulation that made the 2008 collapse inevitable.
Today, we see lessons being ignored again. Wall Street is back to business as usual, relying on misleading models such as value at risk. The next catastrophe will be exponentially larger than the last two. The next time the world will not bounce back.
The Experts
I joined LTCM in February 1994 and reported to the firm’s founder, legendary bond trader John Meriwether, known as “JM.” I came on board before the fund opened for business, and remained through the collapse, rescue, and unwind until August 1999. The founding partners included two future Nobelists and other fathers of modern finance. There was never as great a collection of financial talent in one place, including universities and think tanks, as there was at LTCM.
LTCM’s birth resulted from the near death of Salomon Brothers in 1991. The Salomon name is obscure today, yet in the 1980s, Salomon was synonymous with huge bond bets and complex trading strategies, many invented by Meriwether. In August 1990 and May 1991, a Meriwether subordinate, Paul Mozer, illegally cornered the market in two-year Treasury notes, and lied to the Federal Reserve about his trades. Mozer confessed his crimes to Meriwether, who immediately reported them to the CEO, John Gutfreund, and the president and general counsel of Salomon. Those three officials bungled the internal investigation and never reported the crimes to the government in a timely way.
The scandal reached a critical state on August 18, 1991, when the Treasury Department banned Salomon from bidding in Treasury securities auctions. Salomon’s biggest investor, Warren Buffett, knew this was a death sentence and that Salomon would have to file for bankruptcy, wiping out his investment. Yet Buffett was worried about more than his investment. Coming so soon after the 1990 bankruptcy of Wall Street giant Drexel Burnham, Buffett believed another bond dealer default might destabilize the global financial system. A Drexel-Salomon one-two punch might be more than markets could bear.
Buffett called Treasury Secretary Nicholas F. Brady and got Treasury to partially rescind the auction bidding ban four hours later. In exchange, Buffett agreed to clean house, put up new money, and assume operating control of Salomon until the firm stabilized.
Gutfreund resigned under pressure along with president Tom Strauss and general counsel Don Feuerstein. Meriwether’s case was more difficult because he reported the wrongdoing internally. Still, he was vice chairman of Salomon, and in the hue and cry the Fed felt all top management needed to leave. Meriwether was allowed to resign; his career on Wall Street was over. JM found himself “on the beach,” in Wall Street parlance.
Meriwether set out to build a new firm, a hedge fund not regulated by the Fed or SEC. This allowed him to pursue his complex trading strategies in secrecy, without scrutiny by the government, media, or banks. He systematically recruited former colleagues at Salomon and new faces from academia. The firm was called Long-Term Capital Management. A coming-out announcement for LTCM appeared in The New York Times on September 5, 1993, under the headline “JOHN MERIWETHER RIDES AGAIN.”
LTCM was based in Greenwich, Connecticut. In addition to Meriwether, the LTCM partners included two future Nobel Prize winners, Myron Scholes and Robert C. Merton, and a former vice chairman of the Federal Reserve Board, David Mullins Jr. Yet the talent went far beyond headline names like Scholes, Merton, and Mullins. Less well known, yet equally accomplished, was Alberto Giovannini, the Italian economist who led the technical design team that created the euro. Another key figure was Greg “The Hawk” Hawkins, from the finance faculty at Berkeley, where he was a colleague of Janet Yellen’s. Among the younger talents was Matt Zames, now next in line to be CEO of JPMorgan on Jamie Dimon’s departure.
The LTCM spider web went past the partners to include the fund’s financial backers. One major investor was the Italian Treasury. This connection was crucial because LTCM was the world’s largest trader in Italian government debt. Another investor was the Kuomintang, Chiang Kai-shek’s Nationalist Chinese Army, later a political party that dominated Taiwan for decades.
Some of the world’s largest banks including Japan’s Sumitomo Bank, Germany’s Deutsche Bank, and Switzerland’s UBS made large investments euphemistically called “strategic relationships.” This meant a two-way information flow between LTCM’s traders and the banks’ top management would be maintained. In the global financial web, LTCM was now near the center.
What united this talent was a rock-solid belief in the tenets of modern finance: efficient markets, mean reversion, rational expectations, and normal distribution of risk. In practical terms, this meant that if two instruments had substantially the same credit risk and cash flow present values, they should trade at similar prices. Markets were scrutinized using sophisticated modeling and massive computing power to spot situations where a price relationship was misaligned.
For example, a U.S. Treasury five-year note issued three years ago has two years left to maturity. Treasury also issues new two-year notes. An old five-year note with two years to maturity and a new two-year note with the same maturity should trade at prices that produce nearly identical yields to maturity for the two notes. There was no material difference between the notes because they were issued by the same government, and matured at the same time.
At times, this yield equivalence did not hold true. Two Treasury notes might trade at different yields for reasons unrelated to credit or cash flow. The reasons for divergent prices included institutional liquidity preferences. Some investors wanted only new or so-called on-the-run notes, and avoided old off-the-run notes. They would implement this preference at government debt auctions by selling their off-the-run notes, using proceeds to buy on-the-run notes. Investor liquidity preference temporarily depressed the price of old notes and placed a premium on new ones.
To devotees of efficient markets, liquidity preference made no sense because the two notes were identical from a credit and cash flow perspective. LTCM considered a pricing difference in this situation an anomaly, and exploited it by buying the old note and selling the new note short. In effect, LTCM took the other side of the trade from the investor. By buying the “cheap” note and selling the “rich” one, LTCM owned the spread between the two notes.
In time, markets would normalize. The new note became “old” in the marketplace. The spread between the two notes converged. LTCM would unwind the trade by selling its long position, covering its short position, and pocketing the spread as profit. Because the notes represented the same risk, and because LTCM had offsetting long and short positions, this strategy was regarded as practically risk free. Profit was just a matter of exploiting investors’ irrational liquidity preference.
There were infinite variations on this risk-free arbitrage. Price differences arose in situations that did not involve auctions of new debt. Securities might be taxed differently, and an arbitrage might arise from the tax accounting. Differences might arise between bonds in two different currencies. Pricing adjustments would be made for the currency
risk, which could be hedged in a different trade. Yet whenever some difference arose, LTCM’s computers were waiting to buy the cheap bond and short the expensive bond. Then LTCM could sit back, wait until the spread converged, and pocket a risk-free profit. LTCM would be rational whenever markets were irrational.
One problem with this strategy was that profits on each trade were steady, yet small. Market forces tended to keep spreads from getting too far out of line. LTCM solved this problem with leverage. Profit on one trade might be small, say 2 percent on an annualized basis. Yet if the trade was leveraged 20-to-1, the 2 percent return became a 40 percent return.
LTCM was not a bank, although it acted like one. How could a hedge fund borrow enough money to achieve 20–1 leverage? Money was borrowed through repurchase agreements, or “repo.” In a repo, a bond purchased from one dealer is pledged to another dealer to raise cash to buy more bonds. The result is an inverted pyramid of bonds, pledges, and loans poised on a thin sliver of cash.
Another leverage technique was to hide trades off-balance-sheet using swaps. Swaps are contracts that have the economics of bond trades without actual bonds. Swap counterparties specify a fixed rate, maturity, and currency that synthetically replicate desired cash flows without buying bonds. Swap leverage was higher than with repo because swaps were off-balance-sheet from the bank counterparty perspective. Capital requirements for the banks were lower with swaps than repos. If the off-balance-sheet swaps were accounted for the same as repos, the real leverage at LTCM was not 20-to-1, but 300-to-1. LTCM’s swap agreements eventually exceeded $1 trillion.
These arbitrage and leverage strategies worked. Returns to investors were 20 percent in 1994, 43 percent in 1995, 41 percent in 1996, and 17 percent in 1997. In four years, LTCM had almost tripled investors’ money. This occurred by doing supposedly risk-free trades. From the outside, it looked as if LTCM had invented a perpetual motion money machine. High profits, and high fees charged by LTCM, meant the LTCM partners personally made hundreds of millions of dollars. Outsiders did not know that by 1997, the biggest investors in LTCM were the fund’s partners themselves.
This was a heady time. Professors and policymakers visited to see what the mystery was about. These office tours were awkward because there wasn’t much to see. The panoramic Greenwich Harbor views were beautiful, but strangely quiet. In contrast to hectic shouting in Wall Street dealing rooms, LTCM was mostly silent; the computers did the work. LTCM’s trading style meant that once a trade was on the books, it might remain there for months or years as spreads slowly converged and risk-free profits rolled in. The partners debated strategies in meetings that were more like academic seminars than the knife fights that go on in some banks. When the weather was nice, Meriwether and his partners were as likely to be on the links at nearby Winged Foot Country Club as on the trading floor. There was no reason not to golf; the computers were in control.
JM’s public persona was the bold, brash, Wall Street “master of the universe,” as portrayed in two iconic books, Tom Wolfe’s 1987 novel, Bonfire of the Vanities, and Michael Lewis’s 1989 comic memoir, Liar’s Poker. Wolfe told the story of larger-than-life bond trader Sherman McCoy, who held a position similar to Meriwether’s at Salomon Brothers. The comparison was impossible to miss. Lewis told a real-life anecdote about a million-dollar bet made on the Salomon trading floor between Meriwether and Gutfreund on a single liar’s poker hand, a legend that dogged JM.
In fact, JM was soft-spoken and somewhat shy. He was gregarious with friends on a golf course or racetrack, yet shunned the media and the social circuit that are the usual accompaniments of success in the hedge fund world. JM was a Thoroughbred racing enthusiast and owned several horses. Among his few outside activities was membership on the board of the New York Racing Association, operator of the three largest racetracks in New York including Belmont Park, home of the Belmont Stakes, last leg of the Triple Crown. A day at the races was a typical team-building excursion for LTCM staff.
Thoroughbreds, golf courses, and Greenwich calm were at odds with a perception that LTCM must be a frantic beehive of testosterone-laced shouting and sharp elbows. It wasn’t. Models and computers were producing the profits. Partners were pilots of an aircraft with automatic navigation. You might intervene now and then to deal with unexpected bad weather, but otherwise you let the autopilot guide the plane to its final destination. Provided the autopilot was properly programmed, all was well.
Every big bank in the world wanted a piece of the action. Foreign banks lined up to become investors. They wanted the same strategic relationship that early bank investors enjoyed. Others wanted to be swap counterparties. Banks could book LTCM trades, then lay off the risk in the marketplace, making their own risk-free profits. Often this risk was laid off with banks that had other swaps with LTCM. It was a merry-go-round of risk passed around and around, ending up in the same place—the banks. It seemed the carousel music would never stop.
LTCM’s financial technology was not limited to the fixed income arbitrage Meriwether invented in the 1980s. New structures were discovered. LTCM coinvented the sovereign credit default swap market in 1994 around the same time as a better-known initiative by JPMorgan bankers on a lost weekend in Miami, as recounted in Gillian Tett’s brilliant book Fool’s Gold.
LTCM was the largest holder of Italian government debt, part of a complex arbitrage involving Italian interest rates, different debt classes, and Italian withholding taxes on interest paid to foreign investors. LTCM could hedge interest rates, foreign exchange, and tax risk. One risk the fund could not hedge was default by the Italian government. That risk was tiny, yet not zero. The bond position was so large that even a trivial risk produced a huge expected loss when analyzed statistically. LTCM needed to make Italian default risk disappear. We needed insurance that did not exist at the time, so we created it.
An LTCM trader, Arjun Krishnamachar, and I teamed up to invent this new insurance. We found a willing counterparty in the Milan branch of Japanese banking giant Sumitomo. The branch had assets in Italy and was willing to take on Italian liabilities for a price. Arjun’s job was to devise formulas to price the insurance. My job as chief counsel was to write a contract that defined events of default. This was long before the industry standardized default swap terms. We started with a blank sheet of paper.
Obviously if the government didn’t pay you, that was an event of default. That was an easy one. But there are numerous ways for governments to renege on obligations to bondholders, including capital controls, withholding taxes, asset freezes, and hyperinflation. We had to think of them all; otherwise the insurance might not pay off when needed. That was like buying hurricane insurance that covered wind damage, but not floods. We wanted to make sure we were covered against floods. The sovereign credit default swap was our first innovation. It was not the last.
Avarice
After initial success and billions of dollars in profits, greed came to call. The search was on for new ways to make money using leverage and derivatives.
The partners expanded into stock market arbitrage on mergers and takeovers. So-called merger arb involved the spread between the price at which one company offered to buy another, and the current price of the target company stock.
If company A offered to buy company B for $25 per share, payable in company A stock, and company B’s stock traded at $21 per share, it was a simple matter to short company A and buy company B. In that case you captured a $4 per share spread. You unwound the trade by delivering your B stock at closing, receiving the A stock in exchange, then covering the short position in A shares, pocketing a $4 per share profit.
The risk in such trades is that the deal falls through, and company B stock drops back to a lower level. The LTCM partners understood this. They reasoned that statistically most deals go through, and profits on the winners offset losses on occasional failed deals. The key was to make huge profits on winners, and the key to that was of
f-balance-sheet leverage.
LTCM did not trade actual takeover stocks. LTCM partners knew little about the stock market; for them it was all about the math. Buying and selling shares in takeover deals is expensive because of commissions and margin interest on short positions. LTCM used an equity basket swap arranged by its prime broker, Bear Stearns. In an equity basket swap, a limit is placed on the basket’s size. In the case of the LTCM–Bear Stearns equity swap, the basket held $15 billion in equities. LTCM put items in the basket or took them out with a phone call to the Bear Stearns swap desk. The swap gave LTCM the same profit or loss as owning actual shares without the expense or capital requirements of ownership.
Old-school arbitrageurs were mystified by LTCM’s trading in their market. They had spent decades developing analytic models for assessing if a deal would go through or not. They bought and sold actual stocks and paid high financing costs to do so. Busted deals were their worst nightmare. LTCM knew almost nothing about stocks and was indifferent to the occasional busted deal. LTCM’s edge was high leverage and statistical odds, just another mathematical game.
The partners pursued this strategy in the era’s biggest takeover deals, including Lockheed-Boeing, MCI-WorldCom, and Citicorp-Travelers. LTCM put long and short stock picks in the swap basket, then Bear Stearns did real stock trades to cover its basket exposure. LTCM and Bear Stearns were both hedged. Bear Stearns got cheap financing because it was a dealer. LTCM got cheap financing because it used an off-balance-sheet swap. Everyone was a winner, everyone was hedged. So it seemed.
In 1996, near the height of LTCM’s profits and praise, JPMorgan offered to buy a 50 percent interest in LTCM for $5 billion, a reasonable price considering the management company was making more than $300 million per year in management fees alone. JPMorgan also calculated its ownership position would give it preferred status as an investor, and the opportunity to reap proprietary trading profits.
The Road to Ruin Page 14