The Numbers Game: Inside Crude Oil
NEW YORK (
) -- Every day, a gusher of economic data comes pouring out from one source or another.
Unemployment figures, manufacturing numbers, retail sales, consumer-confidence, gross-domestic product, spending estimates, orders for durable goods. This dizzying swath can be deep and confusing. Preliminary numbers are reported, only to be revised weeks later. Meanwhile, the markets rise and fall on it all.
Then there is the fairly simple question that two groups try to answer each week: How much oil exists in U.S. inventories?
The attempt is hardly an academic exercise -- that is, of course, if you subscribe to the belief that oil prices are actually determined by supply and demand.
The Energy Department's own Energy Information Administration typically releases inventory and production statistics for the prior week on Wednesday mornings. The trade industry's own American Petroleum Institute usually release their numbers on Tuesday afternoons.
Then, there's the weekly routine: Figures come out. Investors digest the numbers and trade on them. Crude futures immediately bound up or fall off in the immediate aftermath. Move on. Rinse. Repeat.
But, every so often, you have days like last week, when the EIA reported a 4.7 million barrel drop in oil inventories compared to the API's own showing of 631,000 barrel increase.
That's when this comment trickled in from a reader:
"Were you totally surprised about the huge disconnect between EIA and API numbers today? I was ... does anyone have any thoughts on that?"
-- TradingNymph
Of course, in order to be surprised, you must first have expectations. But in order to have expectations, it helps to understand what the EIA and API do to come up with these numbers. In truth, their approaches are remarkably similar.
The process starts each Friday, that day being the cutoff for weekly data. At that point, refiners, importers, blenders, pipeline operators, barges and terminals, to name a few, begin reporting the number of crude stocks they're holding.
The operations can fill out an EIA survey and e-mail it to the agency. Interestingly, they often cc the API on those same e-mails in order to catch two birds with one stone. Both groups also accept the info via fax, or even through a data-entry-and-distribution system called PEDRO, which can push out the same inventory and production data to both sources at once.
The EIA gives the operations till Monday afternoon to report. The API can take data till Tuesday. Once the numbers begin to role in, each organization attempts to achieve a threshold for their sample size. In the case of the API, it attempts to get between 80% to 90% of volumes, while the EIA shoots for 90%.
In the EIA's case, for instance, it will begin looking at numbers starting with the largest operations -- like an
Exxon Mobil
(XOM) - Get Report
,
ConocoPhillips
(COP) - Get Report
,
Chevron
(CVX) - Get Report
or
Valero Energy
(VLO) - Get Report
-- and then move down the line to the medium-sized and smaller-sized operations until they hit that magical 90% level. Since there are so many small producers and operations, that 90% volume figure typically represents about half of the active oil operations.
When the data rolls in, both the EIA and the API set out to scrub the information, looking for inconsistencies, errors or odd results in the various reports. When a red flag is raised, both groups may contact the reporting organization in question to resolve the problem. In rare cases, the data may be excluded all together.
To account for those 10% or 15% of remaining, unreported volumes, both the EIA and the API use statistical estimating methodologies to determine the likely remaining results.
From afar, the processes look fairly straightforward and nearly identical. So, why the discrepancies? There are a host of reasons.
First, there is the subtle differences in the samples. API uses around 700 respondents, while the EIA sources somewhere around the mid-800 to 900 range. Thus, the EIA typically captures many more smaller players.
There's also this fact: Data sent to the EIA is mandatory and can result in penalties if requests are ignored. Providing those same numbers to the API, on the other hand, is voluntary.
Ron Planting, the API's head of statistics, is adamant that that is not a major factor in the differences between the results. Yet talk to enough players in the industry and the same concerning hypothetical emerges: A small oil operation, say, is required by law to show their numbers to the EIA though they may not want to, but that same operation can easily bypass the API altogether.
Divergences also arise from the treatment of the data itself. Though both groups may flag quirky data or use estimating methodologies to make up what volumes they don't have, how they treat those quirks and exactly what methodologies to use will differ between the two.
For instance, if a red flag is raised on a reporting outfit, a reviewer could ignore it, could choose not to take a company's explanation for the discrepancy, or could use any of an assortment of statistical methods to fill in the hole.
Even their re-benchmarking processes differ.
In other words, for all the cold comfort that the numbers seem to provide, there is still a significant amount of human decision-making at play in determining how best to treat the figures and reflect the reality of the world we live in. In truth, the same can be said for a number of economic reports that whip past the markets everyday. And while all of this is statistically sound and above board, there is no dispute that the EIA and the API may look at the same numbers and come to wildly different conclusions about how best to treat them.
Thus, last week's discrepancy.
Granted, for all the ballyhooed swings and differences between weekly draws and builds when comparing the two reports, the general longer term trends that can be gleaned from the results tend to compare better. Even the weekly aggregate figures are somewhat similar. Just this week, the API reported outsized total crude stocks at 337.2 million, while the EIA said oil inventories grew to a similarly high 335.6 million.
Still, analysts at the EIA and API agree that it's better to think of their weekly numbers as preliminary estimates. The monthly supply data, on the other hand, is considered a more final say. Jim Beck, a senior analyst on the EIA's reports who used to work as an API number cruncher, likens the process to a political campaign. The weeklies serve as a kind-of Gallup or Rasmussen poll, with the EIA's Petroleum Supply Monthly reflecting a final election of sorts.
The irony in all of this, of course, is how little these numbers seem to get looked at today compared to yesteryear. Though renewed focus on inventory data this week may have helped drive prices down on the front-month contract for NYMEX crude, the typical trend for the past few months has been to give scant attention to the weeklies.
If they are to be reviewed, the EIA numbers are the data of choice among most. Many adhere to the belief that there's more veracity associated with the federal government's numbers rather than the industry's own statistics.
Still, Mike Fitzpatrick, an analyst at MF Global, admits that while he may look at the weekly EIA numbers when they come out, he typically won't think much of them past the first hour of release.
So, what is this all for? Well, talk to analyst Stephen Schork, who edits
The Schork Report
on commodity movements, and he'll give you another view about inventory numbers.
"The Labor Department can't even count the number of people who are losing their jobs consistently three months after the fact," Schork said. "To sit there and say 'now we're going to be able to count the amount of crude oil and all of the chemical derivatives you can refine from that and get an accurate reading of the ebb and flow on a weekly basis'... it's a fools errand. You can't do it consistently."
Schork offers his caveats -- that the EIA does do the best job it can and that the numbers are helpful in a general sense -- but in the final analysis, there's much to be wary of when those reports come out.
"There are just too many variables in the equation to get an accurate reading," he added. "The market values quantity of information over quality of information."
Indeed, it's no surprise, then, that many oil traders these days often pay little attention to the weekly ebbs and flows, despite the fact that weekly U.S. oil inventory numbers used to be scrutinized closely, say, back in the 90s. Gene McGillian, a broker analyst for Tradition Energy, contends that the size of the U.S. economy relative to the rest of the world was so great at the time, and the consequent amount of oil consumed so impressive, that it would have been foolhardy to ignore the weekly figures then.
But those days are long gone. In their place is a new world oil order that views oil as an asset class unto itself, no longer the world of just physical traders and commodity experts, but now also including a swell of general investors looking to diversify -- or speculators looking for a quick buck.
It's no wonder, then, that the status of inventory data has fallen by the wayside, in a world where dollar swings, equity prices and other macro trends are just as closely watched, if not more so. McGillian also notes that U.S. inventory data just don't carry the same gravitas given advancements around the world, too.
"People are realizing it's a global commodity and you've got to be aware of inventories not only in the states," McGillian said, "you have to be aware of global demand levels and import levels in China and Western Europe and even in India."
In the end, many traders use the weekly numbers to get a snapshot or to spot trends or find something unpredictable these days: nothing more, nothing less. Still, it's not surprising when you see the contract swing wildly following an oil inventory report only to move again on the next economic indicator of the day or the week.
As Schork notes, somewhat ruefully: "The market isn't trading on the historicals. It's trading on the weekly data."
-- Written by Sung Moss in New York
Follow TheStreet.com on
and become a fan on
