Metallgesellschaft, 1993

A hedge that was economically sound and financially fatal. The textbook case for the difference between being right and being able to stay solvent while you wait.

What happened

  • Early 1990s — the US subsidiary of Metallgesellschaft signs long-term contracts to supply heating oil and gasoline to customers at fixed prices, for up to ten years. Roughly 160 million barrels of committed delivery.
  • The exposure — if oil prices rise, the firm must buy at market and sell at the old fixed price. A genuine and very large short position in oil, created by the sales contracts.
  • The hedge — buy oil futures. But no liquid market existed ten years out, so the firm stacked the entire hedge in short-dated contracts and rolled them forward month after month: the "stack-and-roll".
  • 1993 — oil prices fall sharply, from around $20 to roughly $15 a barrel.
  • The consequence — the futures hedge loses money immediately and in cash. The offsetting gain on the supply contracts is real but sits in the future, unrealised and uncollectable.
  • December 1993 — after roughly $1.3bn of margin outflows, the supervisory board removes management and liquidates the hedge near the bottom, converting a funding problem into a realised loss.
  • 1994 onwards — oil prices recover. Academics have argued ever since about whether the strategy was sound and the unwind was the error.

The mechanism

The hedge worked as designed: as spot fell, the long-dated contracts gained exactly what the futures lost. But the futures losses were paid in cash every day, and the contract gains were not collectable for years. The vertical line is where the funding gap ended the position.
Hedge unwoundSpot priceCumulative margin paid on the hedgeValue of the long-dated contractsTime (monthly rolls)Price / cumulative cash
  • Maturity mismatch. A ten-year exposure hedged with one-month instruments. The price risk was covered; the timing of the cash flows was not, and the two are different risks.
  • Roll risk turned against them. The strategy quietly relied on backwardation — short-dated contracts priced above longer ones, so each roll earned money. When the curve flipped to contango, every roll cost money instead. Run the storage arbitrage tool to see what determines which regime holds.
  • Asymmetric cash flows. Exchange-traded futures demand variation margin daily. Bilateral supply contracts pay nothing until delivery. The hedge was marked to market; the thing it hedged was not.
  • The hedge ratio was also wrong. Stacking a full ten years of exposure into the front month over-hedges: a barrel due in year eight is far less sensitive to today's spot price than a barrel due next month. The position was larger than the exposure it protected.

What it teaches

  • A hedge that is right on paper can still bankrupt you. Solvency and liquidity are separate constraints, and only one of them can be satisfied later. This is the same lesson as LTCM, reached from the opposite direction — a hedger rather than a speculator.
  • Model the cash flows, not just the payoff. Before any hedge, ask what it costs in cash under an adverse move, and where that cash comes from. The margin-call simulator exists for this question.
  • Basis risk is where hedges actually fail. Different maturity, different grade, different location, different settlement timing — each opens a gap between the hedge and the exposure. A hedge is never a mirror; it is an approximation with named residuals.
  • Unwinding under duress is a decision too. Liquidating at the worst moment converted a paper problem into a permanent one. Whether that was prudence or panic is genuinely still debated — which is itself the lesson: the decision was forced by governance and funding, not by analysis.
  • Size the hedge to the sensitivity, not to the notional. Matching barrels rather than duration-weighted exposure was an independent error, and a common one — the same trap the beta hedge calculator is built to avoid.