Energy Data Explanatory Notes and Resources

Total Production by Shale Formation

The Energy Information Administration's Drilling Productivity Report offers total oil and natural gas production for the Eagle Ford, the Permian Basin, and the Haynesville Shale. However, these data only start in 2007. Thus, total production reported for the Permian Basin and the Haynesville Shale before 2007 is the sum of county-level production data obtained from state agencies. For the Barnett Shale, total dry gas production from 2000 through the present comes directly from the Energy Information Administration's Natural Gas Weekly. The following table summarizes the sources for these data.

State-Level Production

The Eagle Ford Shale and the Barnett Shale both lie entirely within Texas. As a result, the Energy Information Administration's (EIA) data on total production for the region also provides timely estimates of statewide production in each shale formation. The Texas production data for the Eagle Ford shale and the Barnett shale therefore come directly from EIA.

By comparison, the Permian Basin and the Haynesville Shale straddle two states. For both regions, total production estimates beginning in 2007 come directly from the EIA's Drilling Productivity Report, which uses well-level production data to forecast total oil and natural gas production. However, the Drilling Productivity Report does not provide estimates of state-level production. Estimates of state-level production in each formation therefore rely on a combination of data from the EIA and state agencies.

The New Mexico Oil Conservation Division (OCD) provides county-level production data for the four New Mexico counties in the Permian Basin: Chaves, Eddy, Lea, and Roosevelt. New Mexico's statewide production in the Permian Basin is therefore computed by summing production across these four counties. This estimate of New Mexico's production in the Permian Basin may be revised upward as operators file production reports. To minimize the magnitude of these revisions, this site only reports lagged quarterly statewide production for the Permian Basin.

To compute Texas's production in the Permian Basin, the estimate of New Mexico production is subtracted from total production in the Permian Basin as reported by the EIA. Because New Mexico's production estimate may undergo upward revisions, this method risks overestimating production for the Texas portion in the Permian Basin. Nonetheless, the OCD incorporates its revisions more quickly than its Texas counterpart, the Texas Railroad Commission (TRRC). For the most recent quarters, this method therefore offers more reliable production estimates than simply summing the county-level data from the TRRC.

A similar method is used to compute Louisiana and Texas production in the Haynesville Shale. In this case, county-level natural gas production in Louisiana comes from the Louisiana Department of Natural Resources. Louisiana's statewide production in the Haynesville is computed as the sum of production across the Louisiana counties in the formation. To estimate Texas's production in the Haynesville shale, this estimate of Louisiana production is subtracted from total Haynesville production reported by the EIA. The Louisiana data may also undergo upward revisions as new data becomes available. To minimize the magnitude of these revisions, statewide production data for the Haynesville is also reported quarterly and with a lag.

To clarify the methodology, the table below illustrates the computations for second quarter 2013.

County-Level Production

For recent years, the Energy Information Administration provides accurate data on total production within each shale formation. However, it does not provide estimates of county-level production.

The Texas Railroad Commission, the Louisiana Department of Natural Resources, and the New Mexico Oil Conservation Division offer county-level production data for their respective states; however, initial releases of these data are often revised as producers continue to file production reports. In Louisiana and New Mexico, the state incorporates the majority of the revisions in the county-level production data reasonably quickly. Thus, for counties in these states, the data reported on the website come directly from the state agency.

In Texas, oil and gas production numbers move more slowly from the Texas Railroad Commission's (TRRC) so-called "pending file." The chart below illustrates this for oil production in Karnes county, the largest oil-producing county in the Eagle Ford. Each bar represents the level of production for December 2012 as reported by the TRRC at the labeled date. These data were revised nearly a year after their initial release; from the earliest to the latest release date, reported production changed by almost 25 percent.

While the level of oil and natural gas production reported by the TRRC changes significantly across release dates, a county’s share of Texas production in the shale formation is more stable. To compute a county’s share of Texas production in the shale formation at any release date, the county’s level of production as reported in a given release is divided by the sum of production across all the Texas counties in the shale formation. As shown for Karnes county in the chart below, the county’s December 2012 oil production share did not change as much as its oil production level across release dates.

To summarize, the TRRC's reported level of oil and natural gas production undergoes substantial upward revisions across release dates, but a given county’s share of total production in the Texas portion of the shale formation does not change as much. As a result, to estimate the level of production in each Texas county, we multiply its share of total production as observed in the TRRC’s county-level data by the estimate of Texas production in the formation of interest. To clarify the methodology, the table below illustrates this computation for Karnes county in the Eagle Ford, Panola county in the Haynesville Shale, Tarrant county in the Barnett Shale, and Andrews county in the Permian Basin.

Indexing

Economists often compare variables’ rates of change over time. They might, for example, compare retail sales growth in the Texas portion of the Permian Basin with the rest of Texas. One way to do this is to simply plot both variables on a chart. But there’s a problem with this approach: the rest of Texas sells a lot more stuff than the counties located in the Permian Basin. On the plot, both variables would appear as flat lines, separated by great vertical distance. The gulf between their sales volumes makes it hard to observe changes, much less contrast them.

Economists use indexing to filter out these differences in magnitude and focus on relative rates of change. This is how it works: a researcher changes each variable’s first year value to a common number (usually 100), called the index base.

In the original dataset, the researcher notes the percent change in each year relative to the first. The new time-series then arises from a simple rule. Each annual value must mirror the percent change found in the original dataset, and is presented relative to the index base (100). The resulting plot eliminates differences in absolute size, allowing viewers to compare growth rates between big and small variables.

Consider this example. The index of taxable sales in the Permian counties located in Texas, as of 2012Q4, was 243. Its starting value (in 2004:Q1) was 100. The value 243 tells us that taxable sales in those counties have grown by 143 percent since the start of 2004. For the rest of Texas, the most recent index value was 154, so that variable has grown by 54 percent since 2004. Note that this tells us nothing about the level of taxable sales in the Permian counties compared to those in the rest of Texas. The index only shows that taxable sales in the Permian counties have grown much faster.

For more information on Indexing, see Indexing Data to a Common Starting Point.