By Grant Mercer
There is a substantial risk of loss in trading commodity futures and options products. Losses in excess of your initial investment may occur. Past performance is not necessarily indicative of future results.
In late 1993, one of Germany's great industrial conglomerates nearly destroyed itself with a hedge. Metallgesellschaft AG — a 112-year-old metals and engineering giant — had built, through its American subsidiary, a large and apparently sensible program: sell long-term fixed-price petroleum contracts to customers, and hedge the price exposure with oil futures. It was the kind of transaction taught in textbooks. The hedge worked exactly as designed.
And it nearly killed the company anyway — not because the hedge was wrong about prices, but because it was wrong about cash. This article covers the construction of the strategy: what the subsidiary built, why the design looked sound, and the structural feature that would later prove fatal. (The margin calls, the liquidation, and the losses are covered in the companion piece on Metallgesellschaft's undoing.)
What happened: the setup
The strategy was conceived at MG Refining & Marketing (MGRM), the U.S. petroleum marketing subsidiary of Metallgesellschaft. Beginning around 1991, MGRM offered its customers — independent gas station operators, heating-oil distributors, and other petroleum jobbers — something unusual: long-term supply contracts, up to ten years, at fixed prices. For customers, this was attractive insurance against rising oil prices. For MGRM, it created an enormous exposure: the company was now short oil for a decade, obligated to deliver petroleum at fixed prices no matter what the market did.
To hedge this exposure, MGRM bought oil futures — but there was a structural problem. No exchange listed crude oil futures ten years out. The longest-dated liquid contracts extended only a year or two. So MGRM did what hedgers in this situation do: it built a "stack" — a large long position in near-term futures contracts, equivalent in barrel terms to the long-term delivery obligation (reported at roughly 160 million barrels across futures and related derivatives) — and planned to "roll" the stack forward month by month, selling expiring contracts and buying the next month out, for the life of the customer contracts.
The strategy had an additional elegance. In the early 1990s, the oil market was in backwardation — near-term prices above distant prices. Rolling a long stack forward in a backwardated market earns a positive "roll yield": each month, you sell the expiring contract high and buy the next month low, pocketing the difference. The hedge was not just protection; in backwardation, it was designed to be profitable in its own right.
By 1993, MGRM's program was enormous — one of the largest oil derivatives positions in the world. And then two things changed. The price of oil fell, from the high teens toward the low teens per barrel. And the forward curve flipped: backwardation gave way to contango, with near-term prices below distant prices. Both changes struck at the heart of the stack-and-roll design.
The mechanism: why a stack and roll?
A "stack and roll" hedge exists because the futures market's calendar is shorter than the hedger's obligation. If you owe a customer oil deliveries at fixed prices for ten years, the textbook hedge is a ten-year long futures position — but that contract doesn't trade, or trades too thinly to use. So you stack your entire hedge in the near months, where liquidity is deep, and roll it forward as time passes. Each month, you close the expiring contracts and open the next month's, keeping the total barrel-equivalent exposure roughly constant against the remaining delivery obligation.
Economically, the logic is coherent. The customer contracts make MGRM short oil for a decade. The futures stack makes MGRM long oil. If oil prices rise, the futures gain offsets the rising cost of supplying fixed-price oil. If oil prices fall, the futures lose, but the customer contracts become more valuable — MGRM is locked into selling oil at above-market prices for years to come. On a mark-to-market basis, the combined position is roughly balanced. That is the textbook case, and MGRM's economists understood it well.
The roll yield was the design's second pillar. In backwardation — when near-term futures trade above later-dated ones — rolling a long position forward month after month generates a gain at each roll: you sell the expiring contract at the higher near-term price and buy the next month at the lower price. In the early 1990s, oil markets were backwardated often enough that the strategy's architects treated this roll yield as a reliable feature of the design. The hedge wasn't merely defensive; it was structured to pay its own way.
But both pillars contained hidden assumptions. The first assumed that a long-dated economic hedge could be replicated by a sequence of short-dated futures positions without friction — that rolling would be cheap and that margin financing would always be available. The second assumed that the forward curve's shape was a durable feature of the market rather than a variable that could flip. When oil prices fell through 1993 and the curve moved into contango — near-term prices below distant prices — the roll yield reversed sign: each monthly roll now crystallized a loss, selling the expiring contract low and buying the next month high.
The warning signs embedded in the design
Viewed as a funding structure rather than a price view, MGRM's program displayed its fragility well before the crisis:
The margin mechanism was the position's true size. The relevant measure of MGRM's exposure was never the notional barrel count — it was the maximum margin call the position could generate in an adverse move, and whether the company could fund it. A 160-million-barrel-equivalent long funded month-to-month is a different instrument than the same exposure funded for a decade. The futures stack was marked to market daily, and every dollar the oil price fell was a dollar per barrel that had to be wired to the clearinghouse immediately — regardless of the decade-long horizon of the customer contracts. The position's danger was denominated in dollars of potential margin call, not barrels.
The roll yield was assumed, not guaranteed. The strategy's economics depended on backwardation persisting. But the shape of the forward curve is itself a market variable — driven by inventories, interest rates, and expectations — and it had flipped before. A hedge whose profitability requires a particular curve shape is a hedge with a hidden view on the curve. When the view proved wrong and contango arrived, the monthly roll began bleeding cash in addition to the margin calls.
Accounting obscured cash. The financial statements showed the economics; they did not show the weekly cash demands. As oil prices fell, the customer contracts — the long-term fixed-price obligations — were simultaneously becoming more valuable, an unrealized gain that, on paper, offset the futures losses. But that gain existed only on paper: the customer contracts paid off over a decade, while the margin calls were due that week. Any risk report that presents a hedged position as "balanced" without showing the timing of the cash flows is answering the wrong question. The right question is always: what do we have to pay, when, and from what source?
The board didn't understand the book. Perhaps the most human warning sign: the supervisory board that would ultimately liquidate the position reportedly did not fully grasp the strategy it was unwinding. An organization running one of the world's largest derivatives positions, overseen by directors who could not evaluate it under stress, was an accident waiting for a trigger. Governance is part of the position.
The lessons in the construction itself
The Metallgesellschaft affair remains the most intellectually contested disaster in the history of derivatives — the subject of a famous academic debate between economists Merton Miller and Christopher Culp, who argued the stack-and-roll strategy was fundamentally sound, and Antonio Mello and John Parsons, who held that it was reckless regardless of its theoretical elegance. The debate has never been fully resolved, which is precisely why the episode remains so instructive. Both sides agree on the central fact: the strategy's viability depended on funding, not just on price views.
For the student of risk, the construction phase teaches four durable lessons:
Hedges transform risk; they do not eliminate it. MGRM's program exchanged price risk for funding liquidity risk and calendar basis risk. That exchange may have been worthwhile — but it was an exchange, not an elimination, and it needed to be managed as one. Every hedge should be analyzed not just for what it protects against but for what new exposures it creates, and when those exposures demand cash.
Cash flow is the binding constraint. A position that is balanced on a mark-to-market basis and fatal on a cash basis is a fatal position. Risk measurement that stops at P&L — at values, at Greeks, at scenario valuations — is incomplete. Liquidity risk is not a separate topic from market risk; it is market risk's cash register.
Design for the margin call, not the thesis. Every margined strategy should be stress-tested against its own funding: what is the largest cash demand this position can generate, and where does that cash come from? If the answer is "we'll figure it out," the strategy is incomplete. The thesis can be brilliant and the funding absent — and the market settles in cash, on schedule, without interest in your thesis.
Governance must understand the book. The board that later liquidated MGRM's hedge was deciding under conditions of incomprehension. An institution running large derivatives positions needs overseers who understand them well enough to distinguish a funding squeeze from a failed strategy — otherwise the governance meant to protect the firm becomes the mechanism of its destruction.
In December 1993, the board did intervene: it dismissed the responsible executives, liquidated the entire futures stack at reported losses of roughly $1.3 billion, and unwound the customer contracts. How margin calls forced that decision, and why liquidation crystallized the losses, is the subject of the companion article — and of the debate that still divides the economists who study it.
There is a substantial risk of loss in trading commodity futures and options products. Losses in excess of your initial investment may occur. Past performance is not necessarily indicative of future results.