Metallgesellschaft's rolling-hedge collapse

A decade of fixed prices, sold to American retailers

In 1991, MG Refining and Marketing (MGRM), the American oil-marketing subsidiary of the German industrial group Metallgesellschaft AG, hired Arthur Benson from Louis Dreyfus Energy and began building an oil-marketing business in the United States. By 1992, MGRM was running an aggressive sales programme, offering US retailers long-term fixed-price contracts for petrol, heating oil and diesel, with terms of five or ten years. The appeal to customers was straightforward: a fixed price meant certainty for a decade, at a modest premium over the spot price of oil at the time the contract was signed, usually $3 to $5 a barrel.

The risk this created sat with MGRM, not the customer. Selling a fixed price years into the future is a real commercial risk. If oil prices rose, MGRM would have to buy fuel at the new, higher market price to honour contracts it had sold years earlier at a lower one.

A stack-and-roll hedge built on one assumption

MGRM hedged the exposure. It bought a large stack of short-dated NYMEX crude oil and gasoline futures, together with OTC energy swaps, matching the barrels of its long-dated fixed-price commitments roughly one for one. The constraint that shaped the whole design was liquidity: exchange-traded oil futures do not trade in real size much beyond about eighteen months, so MGRM could not simply buy contracts that matured on the same dates as its ten-year delivery obligations. Instead it ran a stack-and-roll strategy, holding near-month futures and rolling the position forward every month as the contracts it held approached expiry.

The scale grew quickly. By September 1993, MGRM had committed to deliver 102 million barrels under "firm-fixed" contracts, 94% of them with a ten-year tenor, plus a further 47.5 million barrels under ten-year "firm-flexible" contracts and 10.5 million barrels under five-year firm-flexible contracts. By mid-to-late 1993 its cumulative short position and its matching stack of futures and swaps had both reached roughly 160 million barrels, hedged barrel-for-barrel.

The strategy rested on one assumption: that the oil futures market would stay in backwardation, the state where near-dated futures trade above more distant ones, so that each monthly roll would, on average, earn a small profit as the expiring contract was replaced by a cheaper one. For most of 1993, that assumption failed. Oil prices, already low by recent historical standards, kept falling, and the futures curve shifted from backwardation into contango. From that point, every monthly roll cost money instead of paying a benefit.

A profit on paper, a cash crisis in real time

The falling price turned from a commercial inconvenience into a cash problem through the mechanics of daily settlement. NYMEX futures are marked to market daily: when the price of oil fell, the value of MGRM's long futures position fell with it, and the loss had to be funded in cash, that day. The offsetting gain, the value of being able to buy oil cheap and deliver it against contracts signed at higher fixed prices, was real, but it would only be realised over the following five or ten years as the delivery contracts were fulfilled. MGRM was profitable on paper over the life of the whole position, yet it was haemorrhaging cash in the short run. That distinction is the whole case in one sentence.

The board dismisses Schimmelbusch and unwinds the hedge

By early December 1993, rumours of MGRM's difficulties were circulating. NYMEX raised MGRM's margin requirements, and nervous OTC counterparties began terminating contracts or demanding extra collateral, tightening the cash squeeze further. On 17 December 1993, with WTI trading at $13.91 a barrel, the crisis became public. On 20 December 1993, Metallgesellschaft's supervisory board dismissed executive chairman Heinz Schimmelbusch and other senior managers, citing lax operational control, and installed new management with orders to liquidate MGRM's hedge and renegotiate or cancel its long-term customer contracts. NYMEX withdrew MGRM's hedging exemption once the wind-down was announced, forcing a further reduction in its remaining futures positions. Over December 1993 and January 1994, MG's board negotiated a $1.9 billion rescue package with a consortium of 120 creditor banks to keep the group solvent. Oil prices began recovering not long after the hedge was unwound, and that timing is exactly why the case still fuels debate over whether liquidation was necessary at all.

Counting the cost

MG reported operating losses of DM 1.8 billion for the fiscal year ended 30 September 1993, on top of roughly DM 1.5 billion (about $1 billion at the exchange rates of the time) that auditors attributed specifically to the oil hedging programme. Later press accounts settled on a figure of about $1.3 billion as the loss attributable to MGRM's marketing and hedging programme, and that is the figure this page uses throughout. A rounder $1.5 billion also appears in some accounts: that was an early December 1993 press estimate, made before the figure was refined down to $1.3 billion, not a separate loss. Two larger figures turn up later in the record too, $1.75 billion in November 1994 and a further $2.2 billion in February 1995, but these belong to the wider Metallgesellschaft group's continuing troubles in 1994 and 1995, not the original hedge unwind. In July 1995, the US Commodity Futures Trading Commission brought administrative proceedings against MGRM and MG Futures Inc. for material inadequacies in internal control and for selling illegal off-exchange futures contracts. Both subsidiaries settled, without admitting or denying the charges, paying $2.5 million and agreeing to internal-control reforms.

A debate that has never quite settled

The case split economists into two camps, and they still disagree today. Christopher Culp and Merton Miller argued MGRM was a prudent hedger ruined by a board that panicked: at maturity, the delivery contracts and the rolled hedge would broadly have offset, and the real failure was a lack of funding and patience, not a flawed strategy. Antonio Mello and John Parsons, and separately Stephen Craig Pirrong, argued the mismatch itself was the flaw: hedging a ten-year exposure with one-month contracts, barrel-for-barrel, is not really a clean hedge. It is a long-dated bet on the shape of the futures curve, dressed up as a hedge, plus a recurring cost every time the stack rolls in contango.

MGRM was short a very large book of long-dated, fixed-price forward delivery obligations. That is the underlying commercial exposure this course covers in Lecture 3: a promise to sell at a fixed price for years to come, which loses value if the market price rises above what the contract locked in.

To offset this, MGRM went long a large stack of short-dated NYMEX crude oil and gasoline futures and OTC swaps, rolling the stack forward roughly every month. This is exactly the rolling-hedge problem this course studies: a rolled short-dated position converts one long-dated basis exposure into a sequence of repeated roll decisions, and each roll costs money or earns money depending on whether the futures curve sits in backwardation or contango. MGRM's programme was built on the assumption that backwardation would persist. When the market moved into contango for most of 1993, the roll became a recurring toll rather than a recurring benefit.

What turned a large but arguably sound position into a genuine cash crisis was the margin mechanism covered in Lecture 2. NYMEX futures are marked to market and settled in cash daily. When oil prices fell, MGRM's long futures position lost value immediately and had to be funded with real cash straight away. The offsetting gain on the fixed-price delivery book would only be realised as those contracts were fulfilled or terminated, years in the future. A position that could be economically sound over its full life still failed for want of cash in the short run. The same timing mismatch recurs later in the course, in the 2022 UK gilt-LDI crisis and in several Chinese corporate hedging cases: margin turns a market-risk position into a liquidity-risk position.

The size of the mismatch is also a basis-risk story. MGRM hedged barrel-for-barrel, buying one barrel of near-month futures or swaps for every barrel of long-dated exposure, regardless of whether delivery was six months or ten years away. Academic re-analysis finds that a variance-minimising hedge would have been considerably smaller, on the order of 38% to 50% of the exposure at the shorter end of the book. MGRM's 1:1 ratio meant it was carrying roughly two to two-and-a-half times more near-term price risk than a hedge designed purely to minimise variance would require. Whether that extra risk was a reasonable bet on persistent backwardation, as Culp and Miller argue, or speculation dressed as hedging, as Mello and Parsons argue, is still an open question in the literature.