Oil markets have had yet another confusing week, this time not just because of the coronavirus and all the media attention that has garnered. But this week, disparate crude oil inventory data from the Energy Information Administration (EIA) and American Petroleum Institute (API) left traders wondering which set of inventory change data–which is supposed to be representative of the balance between supply and demand–was “the right data”.
History suggests that markets will move after hours on Tuesday in response to whatever data the API dishes out. But because inventory moves are almost always priced into the market before the data release, the biggest price moves in response to the data release is not when inventories have moved up or down in a significant fashion, rather, it is when the data released differs significantly from what analysts had predicted.
Eighteen hours later, the markets look toward the EIA data. The moves here are triggered not only by the difference between analyst expectations and the actual data, but by how much it differs from the API data released the afternoon prior.
Often these two industry bodies–the official government data delivered by the EIA and the non-governmental API–are in lockstep. But from time to time, the data shows a signficant discrepancy between the two reports, leaving traders wondering which data set they should use.
The answer isn’t as straightforward as one might imagine.
For 2020, the inventory moves–not the actual inventory, but the change in inventory–reported by the two bodies show clear evidence that the two sets of data are typically in sync.
This should inspire a fair amount of confidence in the data in general–at least over time. But, as the chart also shows, there are sometimes major differences.
In the first week of January, the two reports were more than 7 million barrels off, with the API reporting on January 7 that crude oil inventories fell by 5.945 million barrels, while the EIA on January 8 reported a 1.164 million barrel build. Expectations were that U.S. crude inventories would fall by more than 11 million barrels.
Just a few weeks later, the situation was reversed. On January 29th, the API reported a 4.267 million barrel draw, while the EIA reported a 3.548 million barrel build–a discrepancy of nearly 8 million barrels. This suggests that the discrepancy is likely motivated by a timing issue in the reporting, rather than a difference in how the EIA and API collect and extrapolate data.
There were two other major discrepancies: the week of April 14 and this week, the week of May 12. And again, those two mostly negate the discrepancy cumulatively speaking.
Let’s talk about those methods. Oil companies are required to report their inventory data to the EIA–it is not optional. For data supplied to the API, cooperation is voluntary. They both, however, enjoy similar response rates from companies, despite the fact that many consider the EIA to be the “official” data. Both bodies are thought to receive around a 90% response rate.
Also, since both sets of data are company reported, they should, in theory, produce strikingly similar results. And that is usually the case. But the four major discrepancies this year have caused many to wonder if one data set is superior to the other.
Before we answer that, let’s look at the cumulative inventory changes for 2020. Of course, this is just a snapshot and ignores the data that came before, but it gives a pretty clear picture of how similar–or how different–the data reporting is.
From January 7 to May 12, the cumulative inventory gain for crude oil reported by the API is 102.891 million barrels. For the EIA, the cumulative gains to inventory were 99.672 million barrels. This results in a total difference of just 3.219 million barrels.
Not too shabby.
But this data is not really used in a cumulative way. For the most part, it is the weekly snapshot of losses or gains that moves the oil price needle, and for this, the market is clamoring for more accuracy.
This week’s inventory difference between the EIA and API is striking. But if we look back a few weeks, we can see that the EIA reported a much higher build a few weeks ago–that discrepancy from a few weeks ago is now being righted.
So far in 2020, the big differences came in pairs, three or four weeks apart. This could cue data watchers to anticipate data corrections when the two differ.
But ultimately, the reason for the weeks-long gap before the inventory move discrepancies are rectified is not clear. The takeaway, perhaps, is that there is no clear “right” data, and traders would be wise to use both the API and EIA–as well as the rig count–to form opinions on market fundamentals. Additionally, traders might find value in using any large differences in the data to anticipate more of the same in future weeks.