Oil has been knocked back this week, but gasoline has taken a real beating. Since Independence Day, WTI crude is down 5.8 percent, but Nymex gasoline has slumped more than 8 percent. Tuesday marked its biggest one-day drop since February's oil market massacre, according to Bloomberg data.
Bad as that is, though, the market may be even weaker than it looks.
The U.S. oil refining year breaks down roughly into two halves, with gasoline dominating spring and summer as Americans take to the roads and heavier fuels like diesel and heating oil picking up the baton for the rest of the year. That is apparent in this chart comparing refining margins for gasoline and heating oil over the past decade or so:
What isn't obvious from that chart is that something a little weird is happening this year. From 2006 through 2015 (excluding that crazy 2008) gasoline's premium to heating oil usually disappeared sometime between late July and late August, with the latter month being the most common. In 2016, it may have already vanished, and the first week of July isn't even over yet. Gasoline's crack spread is now about $2 a barrel lower than that for heating oil.
This matters because strong U.S. gasoline demand has been an article of faith in this year's oil rally. The International Energy Agency projects global oil demand growth this year of 1.33 million barrels a day, with the U.S. accounting for 210,000 barrels, or almost one-sixth, of that. The vast majority of that extra U.S. demand is gasoline.
Yet, as I wrote last week, it's odd that even with American gasoline consumption so strong, inventories now cover more days of demand than they have in years. My colleague Julian Lee pointed out that the problem may be with the demand data themselves, with a large revision to the Energy Information Administration's April numbers presenting a big red flag. The revision took gasoline demand for April down by 260,000 barrels a day compared with the weekly numbers the EIA publishes. That implies demand through the first four months of the year was running about 2.5 percent higher than during the period a year earlier, significantly lower than the figure of almost 4 percent implied by the weekly estimates.
Now a new report from energy economist Phil Verleger suggests even 2.5 percent may be overstating the case. The weekly demand numbers from the EIA aren't actually demand -- rather, they're a mathematical plug to explain the difference between oil production, imports, exports and changes in stocks. This leaves them particularly vulnerable to revision if data on exports, which are always at least six weeks out of date and also have to be estimated -- are off. And they certainly have been.
Through April, exports were overstated in the weekly numbers by about 177,000 barrels a day, compared with an underestimation of about 50,000 barrels a day in the period a year earlier. The corollary, according to Verleger, is that gasoline demand was slightly underestimated last year but significantly overestimated in 2016. He calculates that demand is actually up by just 129,000 barrels a day so far this year, or just 1.4 percent.
Weaker demand would certainly fit with the amount of inventory sloshing around in storage tanks and margins' early migration south this summer. It also would make sense given what has happened with pump prices. While gasoline is cheaper this year, the pace of decline has slowed since last year's steep drop really revved drivers' pulses.
If even part of U.S. gasoline demand growth proves illusory, that's a setback for a global oil market that is already facing other macroeconomic risks like the fallout from the Brexit vote.
Refiners are at the sharp end of this. With gasoline margins weakening early, margins on distillate will have to take up the slack. If inventories of that fuel were low, this could help ease the burden as buyers would take barrels early to help build stocks ahead of winter.
But about that:
This column does not necessarily reflect the opinion of Bloomberg LP and its owners.
To contact the author of this story:
Liam Denning in San Francisco at firstname.lastname@example.org
To contact the editor responsible for this story:
Daniel Niemi at email@example.com