Negative WTI and floating storage

A price falls through zero

On Monday 20 April 2020, the May-2020 West Texas Intermediate (WTI) crude oil futures contract on the New York Mercantile Exchange (NYMEX) opened at $17.73 a barrel, an unremarkable price for a rough year. By early afternoon it had turned negative, and it kept falling. It touched an intraday low of −$40.32 at 2:29pm New York time, then closed, officially, at −$37.63 a barrel. That was the first time in the contract's 37-year history that the price had gone below zero. Sellers were, in effect, paying buyers to take barrels of oil off their hands.

A contract with nowhere to go

The contract expired the next day, and whoever held it at expiry owed physical delivery of 1,000 barrels of crude at Cushing, Oklahoma, the pipeline hub that is the contract's sole delivery point. Normally that is not a problem: almost every trader closes out or rolls into the next month long before expiry, and only a small number of contracts stand for delivery. April 2020 was not normal. Covid-19 lockdowns had gutted demand for petrol and jet fuel just as a Saudi-Russian price war added extra barrels to the market. Cushing's storage tanks, which can hold around 76 million barrels, were reported at about 76% full by the Friday before the crash, and most of the remaining space was already leased or promised to someone else.

The only way out was to pay someone to take it

That left a small number of traders still holding the May contract in an impossible position. They could not take delivery, because there was nowhere to put the oil. They could not simply let the contract lapse either, because a futures contract is a binding obligation, not an option. Their only way out was to sell, at whatever price a buyer with spare storage would accept. Whoever could supply that storage, for instance by chartering an oil tanker to hold crude at sea rather than move it, was, in effect, being paid to solve someone else's problem. The very next day, the June contract, now the front month, traded at a completely ordinary +$20.43. The gap between the two, $58.06 a barrel, was the price the market put on one month of storage space.

Storage, not the barrel, was the scarce asset

The shipping data tell the same story. Floating storage on tankers hit record levels within a week: about 60 very large crude carriers, each able to hold roughly 2 million barrels, were chartered to sit at sea and wait. The May contract itself expired the following day at +$10.01, a perfectly normal price, once the traders who had nowhere to put physical oil were gone from the market.

WTI is the US benchmark crude grade, priced and delivered at Cushing, Oklahoma, a pipeline and storage hub. Unlike a financial future on a stock index, the NYMEX WTI contract is physically deliverable: a trader still holding the contract at expiry must make or take delivery of 1,000 barrels a lot. That is the detail that drives everything else in this case. Most participants, hedgers and speculators alike, close their position or roll it into a later month long before that point, precisely to avoid the delivery obligation. In April 2020, Covid-19 travel and lockdown restrictions had collapsed demand for road and air fuel just as Cushing's tanks were filling from the Saudi-Russia price war. With storage scarce and the contract about to expire, the small remaining pool of long positions had far fewer ways out than usual, and far less room to manoeuvre.

Crude oil is a consumption asset, not an investment asset, and that distinction does all the work in this case. For an investment asset, both sides of the no-arbitrage argument work, and together they pin the forward price down to an equation: \(F_0 = S_0 e^{rT}\) (extended for income, dividend yield, or storage costs). For a consumption asset, only one side survives. If the future trades too high, anyone holding physical inventory can buy spot, store it, and sell the future, and that trade caps the price. The mirror trade would require selling spot inventory short and buying it back later, but refiners who hold physical oil will not do this, because it means halting production, and they value having the barrel on hand, a value known as the convenience yield. That asymmetry leaves an inequality, not an equation:

$$F_0 \le S_0\, e^{(r+u)T}$$

April 2020 is what happens when even the ceiling-side trade stops working. That ceiling assumes a trader can always buy spot, store it, and sell the future. By 17 April, Cushing was reported at about 76% of working capacity, with the rest largely spoken for. A trader who did not want, or could not arrange, storage had no way to convert the expiring contract into a stored barrel. The only way out was to sell the contract outright, to anyone who could supply the missing storage. Once that arm of the arbitrage was gone, nothing capped the price any more, not even zero.