The Texas Railroad Commission (RRC of TX) updated its online database a few weeks ago. The best estimates I have seen for Texas C+C and natural gas output are produced by Dean. Thank you Dean for sharing your analysis with us.
The following 4 charts were produced directly by Dean, output for Oil + Condensate, Oil, and Condensate are in barrels per day, and Total Natural Gas is in thousands of cubic feet per day.
The chart above shows how Dean’s estimate for Texas Oil and Condensate has changed from August 2015 to Feb 2016, the most recent estimate is a dashed line.
The chart above shows how the RRC estimates have changed from Jan 2013 to April 2016, generally the data is incomplete for about 18 months prior to the most recent data point. Look closely at the May 2015 dataset (most recent data point in March 2015) and you will see that it does not get close to the April 2016 data until September 2013. Typically when considering RRC data from the Production Data Query (PDQ) System one has to go back 18 months to find data that is within 1% of actual output.
In the chart above the upper and lower bound are each 1 standard deviation (sigma) above and below the “CORRECTED” estimate, this is different from Dean’s chart where the usual 2 sigma upper and lower bound is used. If the probability distribution is normal, the 2 sigma bounds covers 95% of the probability (there is a 2.5% chance output is higher than this range and a 2.5% chance it is lower.) The 1 sigma upper and lower bounds should encompass 68% of the probability assuming a normal probability distribution, with a 16% probability that output might be below the lower bound.
The chart above shows an “Estimate” of Texas C+C output that is the average of the EIA estimate and Dean’s estimate. The dotted line is average output based on this estimate from Dec 2014 to Jan 2016.
Dean recently sent Ron and I the following e-mail:
I have sent you a couple of additional analyses of Texas oil data:
1) A plot of the time evolution of my correcting factors for the first 12 (corrected) months
Plot is below.
2) The EIA Texas data C+C vs three alternative corrected RRC data:
- my usual corrected data using the correcting factors averaged over the whole time sample available
- my corrected data using the correcting factors averaged with data up to November 2015
- my corrected data using the correcting factors averaged with data up to March 2015
The second plot was based on the evidence that some statistical tests highlighted a possible structural break in the dynamics of the correcting factors in April 2015, that is the correcting factors from April 2015 onward belong to a different statistical regime than the factors before April 2015.
I also added the case using the correcting factors averaged with data up to November 2015, because the data released by the RRC in the last 3 months are quite anomalous.
(Second chart is below)
I asked Dean which of these 3 “corrected” estimates was “best” in his view.
From the limited data available he is not sure, but his “best guess” (my words) is that somewhere between “Corrected Break Dec15” and “Corrected Break Apr15” may be close to correct.
The Chart above shows the average of the “Break Dec15” and “BreakApr15” estimates from Dean’s recent analysis (previous chart) from Feb 2015 to Feb 2016 with exponential trend lines fit to different periods. The decline rate from the peak is 4.7%/year, for the whole range shown (blue dots) the decline rate is 4.4%/year, and from June 2015 to Feb 2015 the decline rate has slowed to 1.1%/year.
The future decline rate will depend on many factors including the price of oil, my guess is that it will be between 1.1%/year and 4.7%/year. At 4.7% output in Texas would fall 210 kb/d over the next 12 months, the Bakken may fall about 200 kb/d, the rest of the lower 48 onshore maybe 100 kb/d, for a total decrease in US L48 onshore C+C output of 500 kb/d in 2016.
An alternative estimate can be found by considering US lower 48 onshore (L48 OS) C+C annual decline rates.
The chart above replaces the EIA estimate for Texas C+C with Dean’s “best guess” for Texas C+C to estimate US L48 OS C+C output from Feb 2015 to Jan 2016, the decline rate is 7%/year, which would result in a 500 kb/d drop in US L48 OS C+C output in the next 12 months if the decline rate remains 7%/year and Dean’s estimate is correct for Texas.
The decline rate from the peak in March 2015 is about 8.1%, if that trend continues for all of 2016, L48 OS C+C output will fall by 560 kb/d in 2016, increases in output in the Gulf of Mexico may offset this decline by 100 kb/d, if so US C+C output would fall by 485 kb/d in 2016, assuming Alaska continues its 5%/year decline rate.