Overcooking US gasoline demand
Has the country been burning more gasoline than ever before? The data paint a confusing picture
Herculean US gasoline demand has been propping up cracking margins on both sides of the Atlantic since the beginning of last year. In response, refiners have pumped out as much of the road fuel-and other products besides-as they can, chasing maximum profit.
But are American drivers really hitting the roads with such abandon? To find out, the oil industry relies on Energy Information Administration (EIA) data, the best and most transparent source of consumption numbers available.
Those data have been bullish, showing that as the annual driving season kicked in US gasoline demand soared to a record 9.82m barrels a day in the week ending 17 June. That was 160,000 b/d more than a year earlier.
In its latest Short Term Energy Outlook, a rolling publication giving a forward-looking snapshot of market balances, the EIA said average gasoline demand would rise by 130,000 b/d, or 1.5%, this year, to 9.29m b/d. It would set another annual record.
More people in work in the US, combined with cheap pump prices and a 2.5% rise in highway traffic are behind the demand surge. The EIA expects consumption in 2017 to be close to this year's level.
But the organisation publishes more than one useful set of data. The Petroleum Status Report comes out weekly. It also releases a Petroleum Supply Monthly report, which includes revised demand figures for major oil products. These data are usually two months in arrears. That's where discrepancies start to show up.
The EIA uses an average of 4.3 weeks per month as a model to calculate an average figure for weekly data, which can then be compared with the monthly data.
It calls this its "monthly from weekly" (MFW) calculation. According to the MFW data, between January 2015 and April this year it overstated US gasoline demand in nine out of the 16 months in that period.
Between January and April this year, it overstated demand in every month except March, culminating in a combined 443,000 b/d overestimation for that three-month period. That's an average of almost 148,000 b/d for each month.
The biggest example was in April, when the department's MFW figure overstated gasoline demand by 273,000 b/d (or around 2.3%). The actual demand figure for that month, which was confirmed later, was 9.213m b/d, not the 9.21m b/d the market was originally told.
To put that into context, ExxonMobil's Joliet refinery processes 250,000 b/d of crude, from which it produces 9m gallons of gasoline and diesel for drivers in the US midwest. When five-week months are taken into account the data are, predictably, even more difficult to pin down.
The EIA is also prone to underestimating demand. In September the department's MFW figure was 231,000 b/d below the 9.275m b/d actually consumed.
The anomalies extend to crude data too. Average monthly demand estimates for total US liquids consumption in May were almost 1m b/d higher than the actual 19.4m b/d it later transpired had been consumed. That's an overestimate of over 5% for just one month.
Oil markets react immediately to the weekly EIA data release, and an apparent increase in demand or stocks can move prices on both sides of the Atlantic. This can affect the entire value chain, from cracking margins to arbitrage economics, and right down to pump prices.
Rob Merriam, manager of the EIA's Petroleum Supply Statistics department, says the weekly data should be treated as a rough estimate only. "A lot of analysts say we grossly overestimate weekly volumes but normally we're within 1-2%," Merriam says. "A lot of prognosticators out there would love to be that accurate. People expect a high level of precision on our weekly numbers and that's just not reality."
To compile the monthly figures, the EIA sends out surveys to over 3,000 companies. They have 20 days to provide information ranging from field production to refinery throughput to storage levels.
Then a team of 20 EIA analysts scrutinise around 100,000 data points, investigating anomalies and then finally publishing the figures. This process can take up to eight weeks to complete.
That just isn't possible for the weekly numbers. So instead the EIA surveys just the largest 1,200 companies, which provide around 90% of the volume data, and then estimate the remaining 10%.
Since January, Brent crude prices surged by around $17 a barrel by mid-August, partly because of the perception that demand is increasing-with the US and its thirsty drivers a strong factor in the price climb.
That's why refiners on both sides of the Atlantic have spent months maximising gasoline production, leading to bulging stocks of the road fuel. US inventories of the fuel reached 241.5m barrels for the week ending 22 July, the second highest level on record. Gasoline stocks in Europe-which exports the bulk to the US-stood at around 11.7m barrels at the end of July, according to BNP Paribas, a bank that compiles such data. That's almost a quarter above year-earlier levels.
Other products such as distillates have also been affected. At the end of July US distillate stocks were almost 8m barrels higher than a year earlier, at over 152m barrels.