How leverage turned Nobel-level certainty into system risk

1

Salomon’s half-true lesson

Meriwether’s desk proved leverage plus precision could mint fortunes—a lesson that turned fatal once the same culture operated without institutional brakes.

Lowenstein opens with Meriwether building Salomon’s fixed-income arbitrage group by hiring mathematicians and economists rather than conventional traders. The unit hunted tiny pricing gaps between related bonds and used heavy borrowing to turn thin edges into outsized profit, becoming the firm’s most profitable and most secretive desk.

A Treasury auction scandal on another desk, driven by false bids, engulfed the firm. Meriwether was not accused of wrongdoing, yet he had known of irregularities and failed to report them promptly; he resigned under pressure with the chairman. The pattern flagged here is a firm willing to look away from risk while returns kept arriving.

He left with his genius reputation intact, deep trading-desk relationships, and a clear aim: a vehicle free of public-bank disclosure and bureaucracy. Those years supplied both the intellectual toolkit LTCM would scale and the self-belief that made outside scrutiny feel unnecessary to its founders.

In practice: Paul Mozer’s false Treasury bids triggered the scandal that pushed Meriwether out even though the arbitrage desk was not the accused unit.

2

Pedigree replaces disclosure

Founders assumed elite academic theory plus trading instinct could erase ordinary uncertainty—an assumption the fund’s history would disprove.

In 1993 and 1994 Meriwether recruited Scholes and Merton, whose options work had reshaped derivatives, along with Larry Hilibrand, Eric Rosenfeld, and former Fed vice chairman David Mullins. The new firm opened with roughly 1.25 billion dollars from outside investors.

Terms were unusually stiff: a ten-million-dollar minimum, a three-year lockup, and fees of two percent of assets plus twenty-five percent of profits. Investors received almost no position-level information and were told to trust partners and models.

Banks competed to lend and clear, often waiving normal collateral because of who was asking. Lowenstein treats this as the structural setup for disaster: insulation from scrutiny, eager bank financing, and leaders convinced rigorous theory had driven risk near zero.

In practice: Two future Nobel laureates and a former Fed official made a leveraged convergence book feel like applied science to both clients and lenders.

3

Eventually is not a plan

Relative-value trades only pay if gaps close before leverage demands cash; calm liquidity was the unstated common factor.

Lowenstein details the method: spot two securities that should trade nearly alike, buy the cheap, short the rich, profit as the gap closes. The classic pair was newly issued on-the-run Treasuries versus slightly older off-the-run bonds of similar maturity—nearly identical risk, different prices from liquidity preference.

Gaps were fractions of a percent, so the fund borrowed through repurchase agreements and used derivatives to control positions many times its capital. The approach spread into swap spreads, merger arbitrage, emerging markets, and mortgages. Models scored each trade as low-risk and the portfolio as diversified.

The flaw, invisible in quiet markets, was dependence on one condition: markets staying calm and liquid long enough for convergence on schedule. If that broke, trades could move against the fund together, and leverage left no spare time to wait.

In practice: On-the-run versus off-the-run Treasury spreads were tiny liquidity gaps that heavy financing turned into the signature profit engine.

4

Success that raised gearing

Partners answered shrinking edges by returning outside capital and holding positions steady, spiking leverage instead of caution.

Early results were extraordinary: over 40 percent after fees in 1995 and again in 1996. Imitators crowded in and profitable spreads narrowed.

Rather than scale back, the partners pushed into Russian government debt, corporate merger deals, and directional equity-volatility bets that drifted from the original low-risk brief. At the end of 1997 they returned about 2.7 billion dollars of investors’ capital while keeping the trading book roughly the same size.

That concentrated the same risk on a thinner equity base. By early 1998 leverage stood above 25 to 1, with derivatives notionals exceeding a trillion dollars. Lowenstein calls the choice self-inflicted fragility—an edge eroding with size while the firm refused to accept it.

In practice: Handing capital back while leaving positions intact was framed as efficient management, not as a deliberate leverage increase.

5

History mistaken for law

Value-at-risk treated a rapid capital wipeout as roughly once-per-universe by assuming past correlations and liquidity would hold.

Risk systems assumed daily moves followed something near a normal distribution drawn from years of data. Under that lens, losing a large fraction of capital quickly looked so remote it could be ignored.

Two premises did the heavy lifting: future correlations would match the past, and markets would stay liquid enough to adjust. Neither accounted for the fund now being large enough to move the markets it traded, distorting the history the models trusted.

The 1997 Nobel Prize awarded to Scholes and Merton draped scientific authority over the methods and discouraged skepticism, including among lending banks. Lowenstein’s point is not simple model error about yesterday; it is treating a pattern as a law that could break when enough capital chased the same bets into a real shock.

In practice: Internal risk language cast a severe short-period loss as an event expected once in the life of the universe.

6

Asia read as proof

Surviving 1997’s emerging-market turmoil confirmed the framework for partners and bankers instead of warning that shocks couple markets.

Currency and debt crises swept Thailand, Indonesia, South Korea, and neighbors, sending a global flight into safer assets. LTCM took losses in some emerging-market and volatility sleeves yet recovered as markets steadied, finishing the year still solidly positive though below earlier peaks.

Inside the firm the outcome looked like validation: models had allowed a bad but survivable stretch, and the fund endured. Partners read this as evidence the risk frame captured plausible shocks, not as a signal that a larger version of the same coupling could overwhelm them.

Wall Street banks watching the navigation without apparent damage kept extending favorable financing with little independent scrutiny of true exposures. Lowenstein marks the episode as the destroyer in miniature—supposedly unrelated markets moving together—answered with more confidence, not less gearing.

In practice: Counterparties continued soft terms after Asia because survival looked like control rather than a near-miss on correlation.

7

Diversification’s hidden bet

Russia’s default revealed that dozens of trades were one wager on orderly liquidity; when calm vanished, so did the hedge.

On August 17, 1998, Russia defaulted on domestic debt and devalued the ruble. Investors scrambled away from risk toward the safest liquid assets, chiefly U.S. Treasuries.

LTCM was built for risk premiums to narrow back toward normal relationships. Instead spreads widened sharply and simultaneously across fixed income, equity volatility, and merger books the models had treated as largely independent. Panic itself became the correlating force no historical matrix had encoded.

Leverage meant the fund could not wait for theoretical convergence; further marks would exhaust equity first. This is Lowenstein’s turning point: correlation models collapse exactly when crisis pushes all correlations toward one.

In practice: Sleeves that looked like separate relative-value ideas all required the same orderly market and lost together when it ended.

8

The squeeze dynamic

Leverage that multiplied gains on the way up multiplied panic on the way down once rivals could trade against a known forced seller.

As September losses mounted, firms that had lent to or cleared for LTCM could infer much of the book. Some traded in ways that anticipated forced sales, pushing prices further against the fund and profiting from distress—a common pattern around a large leveraged player in visible trouble.

Capital fell to under 1 billion dollars by late September even as the derivatives book stayed enormous. The fund could not simply liquidate for cash: unwinding that size into illiquid markets would itself crash prices and worsen losses for LTCM and for banks on the other side.

The narrative shows the paradox cleanly: too large to fail quietly, too illiquid to fail slowly. Any attempt to sell down risk threatened to become the shock that transmitted crisis into the broader banking system.

In practice: Rivals needed no conspiracy—only the expectation that a desperate seller would appear—to accelerate the mark-to-market spiral.

9

Plumbing, not partners

The Fed organized a private rescue to prevent credit-market seizure, establishing a template for treating interconnected firms as too big to fail.

By late September, regulators at the Federal Reserve Bank of New York under William McDonough judged that a disorderly failure could chain through dealers holding offsetting trades and freeze credit amid wider fragility. No taxpayer funds went into the fund.

McDonough convened the major exposed banks and brokerages and pressed them toward a private solution. An earlier Buffett-led approach with AIG and Goldman Sachs would have wiped out existing partners entirely, but a tight deadline and structural conditions let that bid lapse.

A consortium then took dominant ownership so positions could be unwound gradually rather than dumped. Markets stabilized within weeks. Lowenstein’s claim is that the intervention, though private, set the pattern: when collapse threatens financial plumbing, authorities act to prevent chaos—a precedent tested far larger a decade later.

In practice: Control shifted to the banks so the book could be wound down slowly instead of thrown into already stressed markets.

10

The lesson left unlearned

After the orderly wind-down, Meriwether and much of Wall Street rebuilt similar leveraged, model-driven habits; humility did not stick.

Under consortium control, positions came off through 1999 and into 2000 without fresh crisis; the rescuing banks eventually recovered their money and the fund closed.

Meriwether started JWM Associates in 1999 with strategies similar in spirit, though at somewhat lower leverage. That fund later took severe losses during the 2008 crisis and shut down, underscoring how beliefs about convergence and risk measurement survived their first failure.

More broadly, banks and hedge funds adopted leverage-heavy approaches often using the same value-at-risk techniques that missed 1998. Regulators studied concentrated counterparty exposure, yet structural changes to leverage limits or hedge-fund transparency stayed modest. The book frames LTCM as preview rather than freak: models, reputations, and bank balance sheets can hide shared risk until one shock exposes it.

In practice: JWM Associates’ collapse in 2008 showed the same philosophical approach surviving direct contact with its earlier ruin.