The 1994 Bond Massacre: Seven Fed Hikes and $1.5 Trillion Lost

At twenty past two on the afternoon of Friday 4 February 1994, the Federal Reserve did something it had never done before. It issued a press release saying what it had just decided. Four sentences went out over the wires, the first of them reading: "Chairman Alan Greenspan announced today that the Federal Open Market Committee decided to increase slightly the degree of pressure on reserve positions."

Translated, that meant the federal funds target was going from 3 per cent to 3.25 per cent — a quarter of a point, the smallest move the Committee had available. Bond dealers who had spent seventeen months borrowing overnight and buying long paper understood it as something larger. By the close the Dow Jones Industrial Average had lost 96 points, about 2.4 per cent, and the thirty-year Treasury had begun a decline that would run for nine months and take roughly a fifth off the price of the long bond.

Over the following year the Fed raised rates six more times, tripling the funds rate to 6 per cent. Fortune, surveying the damage in October, put the loss in global bond-market value at $1.5 trillion and gave the episode the name that stuck: the great bond massacre (Ehrbar, 1994).

Three Per Cent for Seventeen Months

Understanding why a quarter-point did so much damage requires going back to September 1992, when the Fed cut the funds rate to 3 per cent and left it there. Nominal 3 per cent against consumer price inflation running near 3 per cent meant a real short rate of approximately zero, held for a year and a half — a posture the Committee had adopted to repair banks whose balance sheets had been wrecked by the commercial property bust and the savings and loan collapse.

Cheap funding plus a steep yield curve produces a single obvious trade. Borrow at 3 per cent in the repo market, buy a ten-year note yielding 6, and pocket the carry; lever the position eight or ten times, and an unspectacular spread becomes a handsome return on equity. Banks did it. Hedge funds did it with more leverage. Municipal treasurers, insurance companies and Japanese life insurers did versions of it. Between them they had assembled a position whose only real risk was that short rates would move.

Treasury yields obliged the carry trade all the way down. On 15 October 1993 the thirty-year bond closed at 5.78 per cent and the ten-year note at 5.17 per cent, the lowest readings either had recorded in the quarter-century the modern series had existed. Corporate treasurers refinanced, homeowners refinanced, and mortgage originations set records. Anyone holding a bond had spent three years being paid to do nothing.

Washington had noticed the market's new authority. James Carville, fresh from running Bill Clinton's campaign and watching the new administration's fiscal plans get marked to market by the Treasury curve, gave the Wall Street Journal the line that defined the era in February 1993: "I used to think that if there was reincarnation, I wanted to come back as the president or the pope or as a .400 baseball hitter. But now I would like to come back as the bond market. You can intimidate everybody."

Why Greenspan Told Everyone

Nothing obliged the Fed to publish. For decades the Committee had changed the funds rate and left the open market desk to signal the move through its daily operations, so that dealers inferred policy from the size and type of repo the New York Fed conducted that morning. Guessing wrong was part of the job, and the ambiguity suited an institution that preferred not to be pinned down.

Greenspan broke with it deliberately. A first tightening after five years of easing was, he judged, too consequential to communicate by implication; leaving the market to work it out invited a misreading in either direction. The Committee also wanted the move understood as pre-emptive. Inflation was not rising — it would stay below 3 per cent through 1994 and 1995 — and the case for tightening rested entirely on where inflation would be eighteen months later if a real short rate of zero persisted into an expansion already three years old.

That reasoning has worn well in the academic literature, where 1994 is usually treated as the clearest demonstration of forward-looking policy in the Greenspan years and the episode that established the Committee's anti-inflation credibility for the rest of the decade (Blinder and Reis, 2005). It went down badly at the time. A market that had positioned for a zero real rate to persist did not much care that the Fed was acting against a forecast rather than against the data.

Seven Moves in Twelve Months

What followed was not a gradual campaign. The Committee began with quarter-point steps, then accelerated as the bond market failed to settle, delivering half-point moves in May and August and a three-quarter-point move on 15 November that remains the largest single increase of the modern era. The discount rate was pulled up alongside it.

Long yields rose further and faster than the funds rate that was supposedly driving them. By early November 1994 the thirty-year bond yielded about 8.16 per cent and the ten-year note had pushed just past 8 per cent, roughly 240 and 285 basis points above their October 1993 lows. A curve that begins the year steep and ends it flat has handed a loss to every institution funding long assets with short money, and it does so without any borrower missing a payment.

Mortgage securities made the sell-off self-reinforcing. Homeowners stop refinancing when rates rise, so the expected life of a mortgage bond extends exactly when its holder least wants more duration — negative convexity, in the trade's vocabulary. Portfolios that had been happily short duration in 1993 found themselves involuntarily long in 1994, and hedging the extension meant selling Treasuries into a market already falling. Each leg down produced mechanical selling that produced the next.

Where the Bodies Were

Four years of stable short rates had also produced an industry devoted to manufacturing yield for investors who were not permitted to take credit risk. Structured notes did the job by embedding leveraged interest-rate bets in instruments that looked, to a trustee reading a prospectus, like government paper. An inverse floater paid a coupon that fell as short rates rose, often at a multiple; a cost-of-funds note took the bet on the shape of the curve. Frank Partnoy's account of selling such paper out of Morgan Stanley's derivatives desk describes clients who understood the coupon and not the embedded option (Partnoy, 1997).

David Askin ran the first fund to break. His Granite Partners, Granite Corporation and Quartz Hedge Fund held collateralised mortgage obligation residuals — interest-only and principal-only strips whose values moved violently with prepayment speeds — financed on repo from Kidder Peabody, Bear Stearns and Donaldson, Lufkin & Jenrette. When the April margin calls came, Askin had been marking the positions from his own models rather than dealer quotes, and the dealer quotes that arrived were far lower. Roughly $600 million disappeared, and the funds filed for bankruptcy protection on 7 April 1994.

Robert Citron lasted longer because his losses were unrealised. The Orange County treasurer had levered a $7.5 billion pool to about $20 billion through reverse repurchase agreements and stuffed it with inverse floaters, a bet that short rates would stay low that had made him the highest-returning public investment officer in California. Asked during his 1994 re-election campaign how he knew rates would not rise, he replied: "I am one of the largest investors in America. I know these things." The pool lost about $1.64 billion, and the county filed the largest municipal bankruptcy then on record on 6 December 1994, a sequence examined in the Orange County bankruptcy of 1994. Philippe Jorion's study of the portfolio found the strategy required only that rates remain flat to succeed and was fatal at almost any speed of increase (Jorion, 1995).

Corporate treasuries were not spared. Procter & Gamble had entered two leveraged swaps with Bankers Trust in late 1993 whose payoffs were geared to the spread between five-year and thirty-year yields; the position lost money at a rate its own treasury staff could not calculate, and on 12 April 1994 the company took a pre-tax charge of $157 million. Chairman Edwin Artzt was blunt about it: "Derivatives like these are dangerous and we were badly burned. We won't let this happen again." Gibson Greetings lost close to $20 million on similar trades, Air Products and Chemicals about $113 million, and litigation discovery produced internal Bankers Trust tapes in which salesmen discussed the "rip-off factor" built into client pricing. Regulators fined the bank $10 million in December 1994, and its franchise never recovered.

Michael Steinhardt's loss was the largest in dollars and the least discussed. He had extended a celebrated equity record into levered positions in European government bonds, a market that followed Treasuries down through the spring, and finished 1994 roughly 30 per cent lower with about $1.3 billion gone. Piper Jaffray's institutional government income fund, sold to Minnesota municipalities as a conservative alternative to a money fund, fell about 28 per cent. In September the Community Bankers US Government Money Market Fund in Denver liquidated below a dollar a share, the first money market fund in American history to break the buck, undone by adjustable-rate derivative notes bought for yield.

Outside the United States the carry trade unwound in the same direction. German, Italian, Spanish and Swedish yields rose in sympathy with no domestic tightening to justify them, which is what happens when the marginal holder of every bond market is funded in the same currency. Mexico's tesobono programme — short-dollar-indexed paper rolled to finance a widening current account — became unrollable as dollar rates climbed, and the peso broke on 20 December, twelve days after Orange County filed, opening the Mexican peso crisis and the tequila effect.

The Bill

Measured by index, 1994 was the worst calendar year the American bond market had recorded since data began. The Lehman Brothers Aggregate returned minus 2.92 per cent, a number that sounds mild until set against an asset class whose entire purpose is not to lose money; it stood as the index's worst annual result from its 1976 inception until 2022. Long-dated Treasuries returned about minus 8 per cent. Fortune's $1.5 trillion estimate covered the global stock of bonds and was, if anything, conservative about instruments with no observable price.

What makes the year unusual among market dislocations is the absence of a credit event. No sovereign defaulted before December, no large bank failed, no fraud of consequence came to light beyond Joseph Jett's $350 million of phantom profits at Kidder Peabody, which cost General Electric a $210 million charge and ended with Kidder sold to PaineWebber in October. Every loss in the table above came from the price of money changing by three percentage points over twelve months. That was enough, because the positions had been built on the assumption that it would not.

Washington responded with disclosure rather than prohibition. The General Accounting Office had already delivered a report in May 1994 warning that derivatives exposure was concentrated in a handful of dealers and that no regulator saw the whole picture (GAO, 1994). The Financial Accounting Standards Board issued Statement 119 that October, requiring companies to disclose derivative positions and the purposes for which they were held. Municipal investment pools acquired mark-to-market reporting and investment-policy statements of the kind Orange County had never possessed. The Chicago end of the business drew a parallel lesson from the Metallgesellschaft oil hedge of 1993–1994: a hedge that requires funding is a financing decision before it is a risk decision.

The Soft Landing Nobody Enjoyed

Greenspan got what he had tightened for. Growth ran near 4 per cent in 1994, unemployment fell through the year, inflation did not accelerate, and the expansion continued without interruption until 2001 — the outcome the Committee had been aiming at and the one most of its critics had said was unattainable. The Fed began cutting in July 1995 and had returned the funds rate to 5.25 per cent by the end of January 1996. Economists have used the episode ever since as the reference case for a central bank tightening into a forecast rather than into an inflation print.

The market drew a different conclusion, and kept it. Four years later, when Long-Term Capital Management assembled convergence positions at leverage of more than twenty-five times equity, the lesson of 1994 available to its principals was that a funding shock moves every correlated position at once; the sequence that followed is traced in the collapse of Long-Term Capital Management. The instruments differed. The mechanism — cheap short money financing long assets until the price of short money changes — did not.

Nothing about 1994 was hidden. The Fed published every move on the day it made it, precisely so that no one would be surprised. What the year established is that transparency about policy does nothing for portfolios whose leverage was calibrated to the previous policy, and that a quarter-point announced at twenty past two on a Friday afternoon is still a quarter-point.