The day US oil went below zero

Too much oil, nowhere to put it. A Saudi–Russian price war in March had added supply just as lockdowns spread. On 12 April OPEC+ agreed to cut output by a record 9.7m barrels a day — but only from 1 May, and April’s demand loss was three times as big. Unwanted barrels poured into tanks, and stocks at Cushing, the delivery point for WTI futures, rose at a record pace.

A physical contract with financial holders. Each WTI contract is for 1,000 barrels, delivered at Cushing. Anyone holding May contracts purely for price exposure had no tanks and no way to take delivery, so they had to sell before trading stopped on 21 April.

Most had already rolled into later months: open interest, the number of contracts still outstanding, in May fell from more than 600,000 contracts at the start of April to about 108,600 by Friday 17 April. But whoever was left had to sell — and the only natural buyers, firms with spare storage, could demand to be paid.

Liquidity vanished at the worst moment. With most trading already in June, the May contract thinned out. In the 20 minutes before settlement, its price fell about $40 with almost no one buying. The warning was already flashing on Friday, when May settled $6.76 below June — an extreme contango showing that the market would pay heavily to push delivery back by a month.

‘Oil’ isn’t one price. The headline number was true only for oil delivered at Cushing in May. A little contango is normal — later delivery has to cover the cost of storage — but a $6.76 gap between neighbouring months, days before expiry, was screaming that storage was scarce. Brent never went negative: its futures are cash-settled against an index, and seaborne crude can be shipped anywhere or stored on tankers, so no holder had to find a tank in landlocked Oklahoma.

Roll risk is real risk. Products that roll on a fixed timetable are predictable, and predictable flows get run over. Bank of China’s ‘Crude Oil Treasure’, sold to retail investors, still held May exposure on 20 April and settled clients’ positions at -$37.63.

About 60,000 investors were hit, with losses reported at around $1bn. China’s banking regulator later fined the bank 50.5m yuan for failures in stress-testing the product, managing its risks and disclosing them.

‘It can’t go below zero’ is an assumption, not a law. On 15 April CME told its members that NYMEX energy futures could trade at zero or negative prices, and offered a test environment for their systems. Anything built on the assumption of a zero floor had to be fixed in a hurry: CME itself switched its oil options to a pricing model that can handle negative prices from 22 April.

The squeeze was violent but brief. The May contract traded below zero for part of the next day, then expired on 21 April at $10.01. But the fear rolled forward: June, now the front month, fell 43% to $11.57 that day, as the market worried its own expiry would bring a repeat.

USO, the largest oil exchange-traded fund, had rolled out of May before the crash, but it held June contracts. Within days it spread its holdings across the June-to-September contracts, after CME capped how many contracts it could hold in each month.

In November 2020 a CFTC staff report listed an oversupplied market, an unprecedented collapse in demand and worries about storage among the fundamental factors that may have influenced the negative price. The lesson for every desk: know what your contract obliges you to do at expiry, and don’t be the last one holding it.