Metallgesellschaft: The Oil Hedge That Cost $1.3 Billion

A heating-oil distributor in rural Pennsylvania who signed with MG Refining and Marketing in 1992 walked away with something no refiner in the country would sell him: a fixed price for every gallon he would take delivery of for the next ten years, set on the day he signed, with no escalator and no review clause. If crude collapsed he was stuck paying above the market. If crude tripled he was protected, and he could tear the contract up and take cash instead. Thousands of independent retailers, jobbers and small industrial users signed versions of that deal, and by September 1993 the American subsidiary of a Frankfurt metals conglomerate had promised to deliver about 160 million barrels of gasoline, heating oil and diesel at prices fixed years in advance.

Fifteen months later Metallgesellschaft AG was reporting losses of roughly $1.3 billion on the position built to cover those promises, its chairman had been dismissed, and a syndicate of some 120 banks was assembling a DM 3.4 billion rescue to keep a 113-year-old industrial group out of insolvency. Nobody had stolen anything. No trader had hidden a position. The firm had simply discovered that a hedge which is sound over ten years can bankrupt you in ten months.

A Metals House That Became a Trading House

Metallgesellschaft was founded in Frankfurt in 1881 as a metals dealer and by the 1990s had become one of Germany's larger industrial groups: mining and smelting, engineering, plant construction, chemicals, trade finance, roughly 20,000 employees and annual revenues around DM 26 billion. Its shareholder register read like a directory of German capitalism — Deutsche Bank, Dresdner Bank, Allianz and Daimler-Benz all held stakes, alongside the Kuwait Investment Authority with about 20 per cent.

Heinz Schimmelbusch, who became chairman of the management board in 1989 at the age of 44, pushed the group hard toward trading and financial services. He built up commodity operations in the United States through MG Corp. in New York, and within it a subsidiary called MG Refining and Marketing, run by an American oil trader named W. Arthur Benson whom the group had recruited from Louis Dreyfus Energy in 1989. MGRM took a stake of roughly half in Castle Energy Corporation and signed long-term offtake agreements with Castle's refineries, giving it a physical supply base of about 100,000 barrels a day.

Benson's commercial insight was plain enough. Independent distributors competing against integrated majors had no way to lock in their input costs beyond a few months, because the liquid end of the futures market runs out somewhere past eighteen months and nobody was writing ten-year retail fuel contracts. MGRM would write them. The price MGRM quoted sat a few dollars above the prevailing spot price, and that spread — six to eight dollars a barrel on some contracts — was the business.

Synthetic Storage

Selling a decade of fixed-price fuel leaves you short 160 million barrels. The orthodox way to cover it is to buy matching long-dated forwards, but no such market existed at the far end of the curve. MGRM instead used what Christopher Culp and Merton Miller would later call synthetic storage: it bought near-month NYMEX futures in unleaded gasoline and heating oil, plus over-the-counter energy swaps arranged with banks, in a quantity roughly equal to its entire remaining delivery obligation — about 55 million barrels of exposure at its peak — and rolled the whole stack forward every month.

Stacking the far-dated obligation into the front month is not a neutral choice. Rolling a long position is profitable when the market is in backwardation, where the near contract trades above the more distant one, so each month you sell the expiring contract high and buy the next one cheaper. Oil had been backwardated in a large majority of months since the mid-1980s, and that roll yield was an explicit part of MGRM's expected return, not an accident of it.

Two other features mattered. Contracts in the firm-fixed programme let a customer walk away if the near-month futures price rose above the contract price, collecting half the difference on the volume not yet taken; the firm-flexible programme paid the full difference. Culp and Miller (1995) argued that this cash-out option was precisely why a barrel-for-barrel front-month stack made sense, since an early termination would be triggered by exactly the price spike that made the front-month hedge profitable, and the two would settle against each other.

The Curve Flips

Through 1993 the oil market did the two things MGRM's structure could least tolerate at once. Prices fell, and the shape of the forward curve inverted from backwardation into contango.

OPEC's meeting in Vienna in late November 1993 broke up without an agreement to cut quotas, and West Texas Intermediate finished December averaging around $14.50 a barrel against roughly $20 in the spring. Every dollar of that decline showed up immediately as a variation-margin payment on 55 million barrels of long futures. Against it sat an offsetting gain on the fixed-price forward book, which was worth more with every dollar crude fell — but that gain would arrive in dribs over ten years, as customers took delivery, while the margin calls arrived by wire the next morning.

Contango made it worse in a second, quieter way. With the near contract now trading below the next one, each monthly roll meant selling low and buying high, and MGRM was paying that penalty on the entire stack, month after month. What had been a positive carry became a recurring cash cost.

By December 1993 the cash drain on the futures and swap positions was reported at something approaching $900 million. NYMEX, watching a single account hold the largest long position on the exchange in several contracts, raised margin requirements and withdrew MGRM's hedger exemption, which had allowed it to exceed speculative position limits. Removing the exemption forced the position smaller at the worst possible moment and told every other trader in the pit which direction MG would have to go.

What Frankfurt Saw on the Books

German accounting turned a funding problem into a solvency problem. Under the imparity principle of the Handelsgesetzbuch, an unrealised loss must be recognised as soon as it is foreseeable while an unrealised gain may not be booked until it is realised. The futures losses therefore hit the income statement in full. The offsetting appreciation of the ten-year forward book, which was the whole economic point of holding the futures, did not appear at all. A position that a US filer might have presented as a broadly matched hedge appeared in Frankfurt as a one-sided hole.

Culp and Miller attacked this treatment directly in an article in Risk under the title "Auditing the Auditors" (Culp and Miller, 1995), arguing that MG's supervisory board had been shown an accounting artefact and had responded to it as though it were an economic fact.

The supervisory board, chaired by Ronaldo Schmitz of Deutsche Bank, met in the middle of December 1993 and reached the opposite conclusion. On 17 December it dismissed Schimmelbusch and the chief financial officer, Meinhard Forster. Kajo Neukirchen, a restructuring specialist, was brought in as chairman in January. The board ordered the American positions closed.

Liquidation

Unwinding was the decisive act, and it was done into a market that knew what was coming. MGRM sold its futures and swaps and, critically, also cancelled a large part of the fixed-price customer contracts the futures had been bought to cover — which meant abandoning the asset side of the hedge while crystallising the liability side at the bottom of the price cycle.

Deutsche Bank and Dresdner Bank led the rescue, a package of roughly DM 3.4 billion — about $1.9 billion — combining a rights issue with standby credit lines from the group's lending syndicate. Neukirchen sold businesses and cut some 7,500 jobs over the following two years. Benson sued Metallgesellschaft for $1 billion, alleging he had been made the scapegoat for a strategy the parent had approved; the group countersued, and the dispute was settled out of court.

Crude bottomed at a monthly average of $14.68 in March 1994, eight weeks after MG finished selling, and was above $19 by June. Had the position been carried rather than liquidated, the front-month leg would have recovered a large part of its loss inside two quarters.

The Argument Economists Are Still Having

Few corporate losses have generated a literature this contentious, because the facts admit of two readings that are both internally consistent.

Culp and Miller held that MGRM's programme was a coherent synthetic-storage strategy destroyed by a board that mistook a margin call for a loss, and that the liquidation, not the position, cost the shareholders their money. Antonio Mello and John Parsons answered with "Maturity Structure of a Hedge Matters" (Mello and Parsons, 1995), making the technical objection that a one-for-one barrel stack over-hedges a long-dated obligation, because the far end of the forward curve moves far less than the front end; on their arithmetic the minimum-variance hedge ratio was well below one, and the excess was a speculative bet on the shape of the curve dressed as risk management. Franklin Edwards and Michael Canter posed the question in their title — unhedgeable risks, poor hedging strategy, or just bad luck? — and concluded that funding risk was a genuine, unhedgeable feature of the structure rather than a failure of execution (Edwards and Canter, 1995). Craig Pirrong's later reconstruction found the position too large to be explained as a pure hedge and identified a deliberate speculative component in it (Pirrong, 1997).

What none of them dispute is the mechanism. MGRM had matched the price risk of its forward book and left the timing risk wide open. A hedge that is correct in net present value can still fail if the losing leg pays cash daily and the winning leg pays over a decade, and no amount of being right about the level of oil prices fixes a mismatch in when the money moves.

What It Changed

Metallgesellschaft arrived at the front of a run of disasters that taught the same lesson in different currencies. Orange County's leveraged repo book failed in December 1994 on collateral calls rather than on credit losses. Barings' collapse in February 1995 began with margin wired to Singapore against positions the London board could not see. Long-Term Capital Management failed in 1998 holding convergence trades that were mostly right about relative value and fatally wrong about how long its funding would last. Sumitomo's copper book unravelled the same summer. Each involved a firm whose economic exposure looked manageable and whose cash requirements did not.

Risk management absorbed the point slowly and then formally. Liquidity-adjusted measures, cash-flow-at-risk alongside value-at-risk, and stress tests run on margin requirements rather than on mark-to-market values all became standard in the second half of the decade. Accounting caught up too: the Financial Accounting Standards Board issued FAS 133 in 1998, imposing symmetrical treatment of derivatives and the items they hedge, exactly the asymmetry that had made MGRM's book look like a catastrophe in Frankfurt while it looked balanced in New York.

Metallgesellschaft survived, merged with Gesellschaft für Elektrometallurgie in 1993 and eventually with Chemie in 2000 to become mg technologies, and dropped out of metals entirely. Schimmelbusch went on to build a specialty-metals group of his own. The Pennsylvania heating-oil dealers who had signed ten-year contracts mostly found their agreements bought out in the winter of 1994, at a price good for them and set on the day the seller most needed to stop paying.