Last year, I began writing regular articles on the Bakken, one of the largest oil-producing regions in the US. In those pieces, I looked at data provided by the EIA (Energy Information Administration) in its Drilling Productivity Report in order to estimate what the year-end production figures for oil should be there. Seeing as how 2016 is now over, that 2017 is upon us, and that the EIA has given its updated forecast for December's output, I decided that it would be a good idea to go over my work and see if and how the data provided was useful for long-term oil investors.
In my prior works on the Bakken, I had three different forecasts for oil production every month, a conservative scenario, a moderate scenario and a liberal scenario. To stay in the middle, all of my numbers below, unless otherwise stated, refer to the moderate scenario data I looked at, which was, in my opinion, the most realistic outcome for production.
Some pretty accurate data
In the graph below, I decided to compare my year-end production estimates (estimates for December of 2016) for the Bakken for each individual month that I made the said estimate. By doing so, I can see how my end result changed from month to month. Some of you might recall that in my recent article on the Permian, I arrived at the conclusion that my numbers were pretty accurate and it was due not only to my own model but also due to reliable and consistent data from the EIA. The opposite was true in the case of the Eagle Ford, where I showed that my numbers were meaningfully off. I attributed some of that to overly-aggressive assumptions regarding the change in rig count, but it was mostly driven by significant errors made by the EIA.
*Created by Author
Now, in the case of the Bakken, the data appears to be more similar to what was shown in the Permian. My worst month on the downside (bullish side) showed oil production falling to as low as 772,392 barrels per day, while my upside (bearish side) showed it ending the year at 926,025 barrels per day. Compared to the 918,518 barrels per day forecasted by the EIA, we can conclude that the downside forecasting error came out to 146,126 barrels per day, or about 15.9% of the year-end numbers figured by the EIA, while the upside error came out to a paltry 7,553 barrels per day, or about 0.8% of the year-end total.
During this time frame, the average figure forecasted there came out to a miss of about 9.1%. Given the fact that there has been a great deal of volatility in the Bakken, what with changes in production months later, changes in the decline rates and changes in the drilling productivity improvement rates, I must say that I'm content with this forecast; but was there a more accurate measure?
Even better data
Instead of looking solely at the year-end figure, an alternative approach is to look at my average production forecast throughout the entire year and comparing that to the average figure given using the EIA's data. Employing this approach, I was able to create the following graph below, which shows a similar relationship to my prior graph but with less uncertainty.
*Created by Author
What you can see from this graph is that using the average production throughout the year gave an estimate of between 971,604 barrels per day and 1.04 million barrels per day. Breaking this down, I came up with a forecast error on my end totaling about 64,179 barrels per day on the bullish (less production) side, which boils down to a forecast error of about 6.2%, while on the bearish (more production) side I calculated a forecast error of just 0.4%.
A note of significance
There is one point I should touch on here. What I found is that just as in the case of the Permian but the opposite of the case regarding the Eagle Ford, the decline rate forecasted by the EIA throughout the year was in a pretty narrow range. Admittedly, it did drop over time, falling from 6% per month down to 5.20% (it bottomed at 5% though), but that's far better than the tremendous change seen in the Eagle Ford. This can be seen in the graph below.
*Created by Author
Based on the data provided, it appears as though the EIA's data, combined with my model, for the Bakken region throughout last year has been pretty accurate. No model is perfect by any means and while I would have preferred a smaller forecast error (especially with the year-end figure over the average figure), it's all, on the whole, in a range that I find to be acceptable. Moving forward, this suggests that the Bakken data provided by the EIA will likely be a reasonable indicator of the region's oil output trends.
Disclosure: I/we have no positions in any stocks mentioned, and no plans to initiate any positions within the next 72 hours.
I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.