“Natural gas output from US’ Marcellus edges closer to 15 Bcf/d: EIA” declared the headline in Platts that attracted my attention, since the latest data on the Marcellus shale gas play of PA and WV indicated production was less than 12 bcf/d. This headline was based on the latest issue of the EIA’s new monthly Drilling Productivity Report published April 14. Reading further, the article claimed that the Haynesville shale play “peaked at about 10 Bcf/d in 2012”, when in fact it had peaked at closer to 7 bcf/d in 2011. These errors are serious exaggerations of reality and bear further investigation, as the EIA Drilling Productivity Report is widely read and quoted in the media.

Fortunately the EIA also publishes independent production data by shale play in its Natural Gas Weekly Update. A check of production data for the Marcellus revealed that it was at 11.8 bcf/d in February and that the Haynesville had indeed peaked at 7.2 bcf/d in November 2011. These figures are also corroborated by Drillinginfo, a commercial database which is used by the EIA.

There are four shale plays in common between the two EIA reports: the Marcellus, Haynesville, Bakken, and Eagle Ford. The actual shale gas production from them, as provided in the EIA weekly update and confirmed by Drillinginfo, is illustrated in Figure 1.

Figure 1. Actual shale gas production from four plays from 2007 through February 2014 (data here).

The production for the same plays over this period reported in the EIA Drilling Productivity Report is illustrated in Figure 2. This amounts to revisionist history as these plays produced essentially nothing in 2007 yet are listed as producing nearly 7 bcf/d then by the EIA.

Figure 2. Shale gas production from the four plays in Figure 1 as reported by the EIA Drilling Productivity report (data here).

The aggregate error in reporting production from these four shale plays is illustrated in Figure 3. Production in February, 2014, is stated to be more than 7.8 bcf/d higher in the EIA Drilling Productivity Report than it actually is—an error of 38% on the upside, equivalent to more than 10 percent of the total gas production of the U.S.

Figure 3. Overestimation of shale gas production in the EIA Drilling Productivity Report compared to actual production for the plays and time period illustrated in figures 1 and 2.

Real production data usually lags two months behind, and the most recent months are subject to revisions. Yet the EIA’s Drilling Productivity Report confidently reports production for the current and following month to the nearest mcf (or barrel, for oil), along with aggregate depletion. Given the errors illustrated above, this is a gross distortion of the facts—and always on the upside.

Furthermore, the EIA’s decision to start putting out a report highlighting production per rig, rather than production per well, implies that rigs, not wells, are what is important. But the physical footprint on the environment is wells, not rigs, and wells determine the capital input required, plus the actual flows to markets. Presumably, given the efficiency improvements of pad drilling and other innovations allowing a rig to drill more wells per unit of time, the growth statistics per rig would make shale look ever rosier. And inflating the actual production numbers would make shale look rosier still.

The EIA is the elephant in the room when it comes to energy statistics. Its data and forecasts are widely used by analysts and the media and influence energy policy. There is no room for the significant scale of errors and distortions reported herein.