Marathon Oil, Manipulation, Misplaced Concreteness and the Cheesy Strategy

Michael Giberson

The Houston Chronicle reports, "Marathon Oil Corp. says regulators have warned the company that it may face legal action for allegedly attempting to manipulate crude oil prices in 2003."

Marathon allegedly traded aggressively one day in 2003 in an after-hours market to drive down the price of a commonly-used benchmark, presumably to benefit on holdings of contracts tied to the benchmark price.

There are a number of reasons a company might want to sell oil below market prices, said Craig Pirrong, energy markets director for the Global Energy Management Institute at the University of Houston.

A company with plans to buy 1 million barrels of oil, for example, could decide to first sell a smaller amount it already owns, say 100,000 barrels, at a lower price. That lower-priced sale could help drive down the price of oil, making the 1 million barrel purchase less expensive for the company, Pirrong said.

If it works, the company would save more from buying the lower-priced oil than it lost from the below-market sale.

Pirrong said it's a tactic that a well-known cheese company was accused of following in the 1980s and 1990s, when it regularly sold large quantities in an auction at low prices because its contracts to buy cheese were tied to the auction price.

Part of the problem comes from a kind of misplaced concreteness - acting as if the benchmark price is an objective fact, when in fact it is the product of an inter-subjective social process.

Wouldn't these multi-million dollar contracts be less subject to both outright manipulation and various random events if they were tied to a small population of benchmarks? While any set of benchmark prices relevant to a transaction would naturally be closely related - say the price of oil in London, the price in Amsterdam and the price in New York - relying on the median of the three seems superior than relying upon any single one of the three. Kind of like judging figure skaters - throw the top and bottom scores out, and you naturally reduce the bias.