Like many in this industry, we spend a lot of time looking at the weekly data seeking a clue to the end of the industry’s current down-cycle and the start of the next up-cycle. Throughout our career we have watched weekly data series become a cause célèbre for forecasters. In the past, it might have been weekly Baker Hughes (BHI-NYSE) rig count numbers, natural gas storage injections or withdrawals, oil storage volumes, weekly gasoline production figures, gasoline pump prices, or possibly even daily crude oil and natural gas prices that drew everyone’s attention. However, trying to draw meaningful conclusions about these short-term trends often became not only difficult, but often impossible.
The number one conundrum confounding industry forecasters today is to understand how the continuing growth in crude oil and natural gas production in the face of the 900+ rig count decline can still be happening. As a result, we have been monitoring the weekly changes in the estimate of domestic crude oil production announced every Wednesday by the Energy Information Administration (EIA) along with their Thursday weekly estimates of the change in natural gas storage volumes. We consider these weekly changes in conjunction with the weekly count of drilling rigs targeting oil and natural gas prospects.
Last week provided an interesting perspective on this ultra-short-term market focus when oil production numbers surprised observers by falling compared to the prior week’s estimate and natural gas storage withdrawals were twice forecasters’ expectations. In both cases, the immediate price reactions were significant – crude oil prices increased by 5% Wednesday while the front-month natural gas futures contract price rose over 4%.
In Exhibit 1, we show the pace of the current rig count decline compared to the one of 2008-2009. As seen, the current rig decline has been sharper than before. However, the pace of the current rig decline appears to be slowing, similar to the pattern observed during the earlier decline, and frankly, something that usually happens with all declines as sharp reversals are not typical. The question on the minds of many oil industry participants and forecasters is: Does this slowing mark that we are at the bottom?
It is interesting to note when examining the two rig declines that we are slowing at virtually the same point as in 2008-2009. As a caution to those believing the current slowing rig count decline marks an approaching bottom, in 2008-2009, following a couple of weeks of flat rig counts, the decline continued. The bottom of the 2008-2009 decline came after roughly 150 more rigs were lost or about 14% of the rig count at the time of the pause.
An interesting chart (Exhibit 2, next page) we have been maintaining is a plot of the weekly rig count changes compared with the total rig count and WTI oil price since 1987. We suggest viewers focus on the weekly rig count changes. In looking at 1987-2000, we notice that there was a high degree of variability in weekly rig count changes. The industry downturns in 1990 and 1997-1998 can clearly be seen. Interestingly, those downturns show width but limited depth, meaning that the rig correction lasted for a while but each week’s decline was modest. The entire 1987-2000 period showed sporadic weeks with very sharp declines or gains.
Shifting our focus to post-2000, we note that after the 2001 recession’s impact on drilling, the preponderance of weekly rig count changes was to the upside. That again is not a surprise when one examines the slow, steady climb in the oil price toward the $147 peak in 2007. Following the 2008-2009 downturn, the pattern of mostly positive weekly rig count changes returned until the current downturn. Again, that pattern was supported by the steady march higher in oil prices into an extended period of stable and high prices. Our last observation is that if we look at the growing mass of dark blue reflecting the current weekly rig count declines, its shape mirrors the 2008-2009 decline, which is consistent with the rig count chart in Exhibit 1 above.
Another interesting industry pattern to observe is the weekly change in oil production since 2005 and the weekly oil rig count. In the earlier years, there were several episodes of sharp weekly oil production declines that mark the impact of hurricanes that shut in production offshore and hurt Gulf Coast onshore oil output. We would direct the readers’ attention to the past several years. There have been very few declines in the 4-week average production figures, and the falling volumes were minor.
We will now shift our focus to a more recent time period and the difference in the results of the weekly production estimates and the 4-week average of weekly production estimates. The chart in Exhibit 4 shows the weekly production estimates since the start of 2014. In the chart, we have marked the point at which the highest WTI price of $107 a barrel was recorded. We also have noted the week when the U.S. drilling rig count reached its peak.
Many people became quite excited last Wednesday when the EIA released its weekly report showing that domestic production declined by -36,000 barrels a day (b/d), following the prior week when production rose by only 3,000 b/d. These figures prompted comments about this marking a bottom in the downturn because we were now seeing definitive evidence that the fall in the rig count was causing production to stop rising. Obviously, one could make that case based on the latest weekly numbers. However, one needs to only go back to the last half of January when the three weekly production reports showed the following: -6,000 b/d, +27,000 b/d and -36,000 b/d. There was also the week of February 20th when oil production only increased by 5,000 b/d. That small weekly increase was sandwiched between two weeks of +49,000 b/d and +54,000 b/d changes in production. The small weekly gain was followed by three weeks of production gains of +39,000 b/d, +42,000 b/d and +53,000 b/d before the +3,000 b/d gain. Because of this volatility, we are very careful about drawing inferences about a shift in the trend of domestic production. Instead, we prefer to focus on changes in the 4-week average production estimates.
When we examine the pattern of weekly data based on the 4-week production figures, we see what appears to be a visual pattern of lower weekly gains following the peak in the rig count compared to the period between the peak rig count and the peak oil price and prior to the oil price peak as noted in Exhibit 5.
Prior to the past four weekly reports, we were beginning to feel optimistic that we might be approaching a bottom in the growth in domestic oil production. Then we experienced four consecutive weeks when the 4-week average production increases were well above what they had been averaging during the prior 12-week span, a time encompassing the winter months, when production in major basins such as the Permian, Bakken and Marcellus was impacted by the weather. There is also a question about what impact on oil production, if any, occurred in the Bakken area due to North Dakota’s new natural gas flaring regulations.
We are hopeful that the latest domestic oil production estimates reflect the start of a sustainable decline in output that will lead to a stable oil price, and in turn, sustained drilling and completion activity, at least at current levels. If total domestic production continues falling, as suggested by the typical shale oil well production profile, then the industry will find itself on a drilling treadmill in order to sustain output. We may be much closer to the oil industry stepping on that drilling treadmill than everyone suspects.