NEW YORK Energy investors have taken bets for years on what they thought was an important indicator of future energy production: the weekly rig count data provided by oil service firms.
They may want to be careful about how much money they put on the table.
A Reuters analysis of the data, and interviews with officials at companies involved in collecting and compiling it, shows that it may sometimes be an arbitrary and misleading gauge subject to revisions.
The culprit appears to be the fracking boom and the complex geology that has made it much more difficult to decide whether a rig is likely to discover oil or gas in large quantities, often leading companies to rely on guesswork when drilling begins.
At stake is not only the direction of U.S. natural gas prices, but the credibility of U.S. energy companies desperate to show investors that they are drilling for more oil -- which is near $100 a barrel -- and less gas, the price of which remains depressed at near a decade low.
Equity analysts need to know what a company is likely to produce to predict its profits; gas traders are desperate to anticipate any let-up in the unrelenting gush of supplies.
To understand potential flaws in the data, a glance at Chesapeake Energy's rigs in west Oklahoma provides some clues. On January 8 this year, the energy company's subsidiary Nomac began drilling Ogle well 9-11-18 in the Granite Wash Basin, one of many shale patches across the country that are now gushing an often unpredictable mix of oil, liquids and natural gas.
Chesapeake had listed the well as "oil/gas" with state regulators, a common practice. But that was not the assessment of Smith Bits, a Schlumberger subsidiary that is one of two main firms that gather data on nearly 2,000 U.S. drilling rigs and provide it on a weekly basis to the industry and markets.
Based on their own assessments, Smith Bits determined the Ogle rig, and 21 other Chesapeake rigs drilling new wells in the area, to be gas rigs, and marked them as such in their January 20 report, which is used by traders and company analysts as an important signal of oil and gas output in the coming months.
But in the week to January 27, Smith Bits changed its mind and reclassified all 22 rigs - which had begun drilling between November and January - as "oil", according to data compiled by U.S. Rig Activity and confirmed by Smith Bits.
The reclassifications - which are hard to spot without closely researching each rig - made up the majority of the 32 gas rig decline in Smith Bits data that week. It also contributed to a fall of 78 gas rigs in January, the largest monthly drop since May 2009 when companies were paring back on drilling during the financial crisis, according to Smith Bits figures.
To be sure, such a sweeping change appears to be a rarity, with no other obvious reclassifications of that size, according to a Reuters analysis of thousands of wells and the rigs that drilled them. But analysts say that relabeling on a rig-by-rig basis clearly happens frequently, making it difficult to be sure about the weekly data, especially when it comes to new shale developments.
The defects of the rig count have grown increasingly apparent over the past 12 months, as a long-standing correlation between natural gas rigs and natural gas output broke down. The number of rigs drilling for natural gas has halved since it peaked in October last year, but U.S. natural gas production has remained near record highs. Between October and May, production fell just 6 percent, according to the latest government data.
Experts say the problems run even deeper than the complexity of geology. A lack of industry standards, patchy oversight and a mismatch between what companies say they are drilling for and what shows up in official state production data, is exacerbating the issue.
"You need to be careful of using the rig count as an indicator. There is no industry standard on methodology for classifying these rigs," said Bob Williams, director at Land Rig Newsletter, which analyses the rig count data. "It is going to become increasingly complicated to classify a well."
Chesapeake was not involved in the decision to classify or reclassify the rigs, according to Smith Bits, and Oklahoma-based Chesapeake cautions against relying on the rig data for detailed analysis. Yet Chesapeake still used Smith Bits data in an August 7 presentation to shareholders showing that it was running 116 oil-based rigs, nearly 8 times more than at the beginning of 2010.
"These reports are pretty perfunctory, not audited and as I understand it, intended to provide fairly broad trends versus specificity," said Jim Gipson, spokesman for Chesapeake. He declined to comment on the company's use of the data in investor presentations.
DATA "CLEAN UP"
The rig changes in January were the result of a data "clean up" in which Smith Bits revised the rig target based on information from teams out in the field, according to Schlumberger operations supervisor Sheila Lewis, who helps compile the data.
"Because this is such a dynamic process and we use a variety of sources to update our information, we inevitably, at a later date and in light of new information, revise this data to ensure the highest possible accuracy," Lewis said.
Smith Bits' timing of the change raised eyebrows in the natural gas markets. The data were released on Jan 27, just four days after Chesapeake had announced an 8 percent cut in natural gas production, the first of several efforts by gas producers to revive prices that had just touched a new 10-year low under $2.25 per million British thermal units.
The saga of well Ogle 9-11-18 did not end there. When Chesapeake filed the results of its drilling test with state regulators months later, it showed output of 7.3 million cubic feet per day of natural gas and 314 barrels of oil per day.
Ogle and another of the reclassified wells were given a final designation by Oklahoma regulators: natural gas.
The effort to divine production typically begins with official state data. Before drilling, energy companies must apply for a permit with the state regulator, including information about whether it plans to drill for oil or gas. In the past there was little doubt about the target, with most on-shore fields typically well-documented as one or the other.
The advent of substantial production from shale has changed that. Thanks to the rise of hydraulic fracturing and horizontal drilling, energy companies are now extracting a cocktail of hydrocarbons from what the industry calls "windows" of gas, oil and liquids, many of which overlap or are stacked on top of each other.
Natural gas, while historically extracted alone in pure gas wells, is now produced in large quantities from rigs officially classified as drilling for oil. And wells that were initially probing for gas may instead be more valuable probing for oil.
Thousands of well permits each week are labeled as "oil/gas", data specialists say.
Both Smith Bits and Baker Hughes, a rival services firm that also compiles rig data, use the state permits as a reference point, while also using agents in the field to clarify details on specific wells. Once drilling begins, both receive weekly updates from the operator about their rigs.
But their methodology differs. Baker Hughes requires companies to clarify the drilling target; Smith Bits does not.
When a rig's drilling target is not clear, Smith Bits will carry out its own modeling depending on the location of a rig and how it is labeled in state permit, according to Schlumberger's Lewis.
Baker Hughes officials declined to comment on rig assessments for specific companies. "The data provides outstanding directional historical information about rig activity," said Trey Clark, vice president at Baker Hughes in Houston.
ECONOMICS, NOT SUPPLIES
Even collecting a company's intended drilling target, like Baker Hughes, does not offer a clear indication of what the well will actually produce, given that both oil and gas are present in most unconventional plays like the Granite Wash.
Thousands of rigs currently drilling in Texas, home to a portion of the Granite Wash and the Eagle Ford shale formation - another liquids-rich play that producers have flocked to over the past year - changed classification between permitting and when production was officially logged months later based on actual well data, according to state records.
Classifications do not necessarily depend on the contents of a well, but on what is driving the economics. A rig producing, on an energy basis, 50 percent natural gas, 20 percent natural gas liquids and 30 percent oil, can still be considered an oil well if that is the major revenue driver, Baker Hughes said.
The number of rigs may now be less telling than where they are located. Efficiencies in new oil and gas fields vary widely depending on the basin in which they are operating.
Gas produced from one rig in the Haynesville shale play in Louisiana and Texas produces about as much gas as 50 wells in the Permian, according to Adam Bedard, formerly at Bentek Energy in Colorado, which closely monitors production figures.
"There is an illusion that production is going to fall this year, an illusion driven by the declining rig count," he said.
(Editing by Leslie Gevirtz)