IEA Underestimates U.S. Oil Production Growth; Important Ramifications For Prices

The International Energy Agency (IEA), and other major oil forecasting organizations such as OPEC and the U.S. Energy Information Administration (EIA), have consistently and seriously underestimated U.S. oil production since the Shale Oil Revolution took off in earnest early this decade.

Forecasters at these organizations simply do not understand that we are in a new age of rapid technological advancement and innovation. This new age has been pushing down the cost of U.S. light tight oil (LTO) faster and lower than expected, expanding ultimate recoverable resource and so-called sweet spots, and pushing U.S. oil production — both crude and NGLs — substantially higher at lower than anticipated prices.

In the first part of this series we discussed potential headwinds global oil consumption is likely to face through 2024, and then provided what we at Geopolitics Central (GPC) believe is a more plausible and moderate oil demand forecast than provided by the IEA in its recently released medium-term outlook. In the second, and final, part of this series, we explore the agency’s outlook for U.S. oil production, and again provide what we believe is a more plausible forecast, with important ramifications for crude oil prices over the period.

IEA’s outlook

The IEA is now forecasting U.S. oil production to increase 4.08 million bbls/d, or more than two-thirds of total non-OPEC supply growth, to 19.56 million bbls/d by 2024. This represents a substantial upgrade to U.S. production compared to last year’s medium-term outlook, when the agency was forecasting 16.9 million bbls/d for 2023 — the final year of that year’s outlook —2.6 million bbls/d less than this year’s forecast for the same year.

In terms of crude oil price assumptions, as a political organization, the IEA simply assumed the Brent futures curve at the time of modeling and writing their outlook, despite acknowledging “the futures curve is not a price forecast and an imperfect modeling tool.” At the time, prompt month Brent futures were trading at just over US$60/bbl with the curve relatively flat around that level through 2024 — prompt prices currently are closer to US$70.

Forecasting error

Forecasting errors for U.S. oil production and oil prices in recent years have tended to be large. For example, the IEA’s medium-term outlook released in 2011 — once the Shale Oil Revolution was showing solid signs of takeoff — pegged U.S. oil production at 8.35 million bbls/d in 2016, an increase of just 550,000 bbls/d compared to 2010 production based on an IEA average import price of US$103/bbl over the period. Actual U.S. oil production in 2016 was 12.54 million bbls/d for a forecasting error of 4.19 million bbls/d, despite oil prices collapsing to roughly half of their forecasted level for the last two years of the projection period. The IEA’s three medium-term outlooks from 2011 to 2013 have underestimated U.S. production in their final year — 2016 through 2018 — by an average of 3.18 million bbls/d.

Generally speaking, even when the IEA has forecasted relatively strong growth in U.S oil production early in the period, they have had it petering out at the end. In the agency’s latest medium-term outlook, growth slows from 1.46 million bbls/d in 2019 to a mere 60,000 bbls/d in 2024. The agency argues that “[w]ithout a material increase in spending,” and/or a material increase in productivity, we’d assume, “growth will slow as an increasing number of new wells are needed to offset the steep decline from the existing production base.”

Second machine age

The production of LTO and shale gas in the U.S. has often been compared to an industrial process, with companies constantly attempting to push the technological and innovation envelope to produce greater quantities from lesser quality resource at lower costs. But the combination of horizontal drilling and hydrofracking was just the beginning for the industry.

In their 2014 book “The Second Machine Age,” MIT professors Erik Brynjolfsson and Andrew McAfee convincingly argued that we are at a second inflection point in terms of human progress that is supercharging technological advancement and innovation already propelled by the Industrial Revolution, leading to even greater productivity improvements than in the past. They wrote:

 “Now comes the second machine age. Computers and other digital advances are doing for mental power — the ability to use our brains to understand and shape our environment — what the steam engine and its descendants did for muscle power. They're allowing us to blow past previous limitations and taking us into new territory.”

Two major examples that Brynjilfsson and McAfee gave to support their second machine age thesis were autonomous vehicles and artificial intelligence. Both appeared distant possibilities, if not pipe dreams, early last decade. But due to incredibly rapid technological progress in recent years they are both now on the verge of commercialization — and with relatively rudimentary versions of each already commercialized.

Costs, productivity and resource expansion

Another reason the IEA has the rate of growth of U.S. oil production rapidly declining over the medium-term — again and again like a broken record — is because they assume the cost of production will climb over time as producers burn through prime drilling targets and the amount of recoverable oil resource declines. But this sort of thinking, which forecasters from other organizations also suffer, remains Industrial Revolution, not second machine age, and is likely to be wrong yet again.

The average breakeven wellhead price per bbl for U.S. LTO production was cut almost in half to US$35/bbl over the 2014-16 downturn — add US$12 for overhead and interest charges, return on investment and transportation cost to market to calculate breakeven market price — based on producer survey data by oil and gas consultancy Rystad Energy. At the time, only one-half to two-thirds of the cost decline was attributed to structural factors, including rapid technological and operational improvements such as walking rigs, intelligent drill bits and predictive underground analytics.

But improvements in LTO rig productivity have been truly second machine age-like since 2007. For example, the average monthly growth rate in oil production per rig in U.S. LTO basins grew at a rate of 2.7 per cent per month between January 2007 and August 2016. This incredible rate of productivity improvement led average monthly oil production for a rig in these areas to skyrocket from 36 bbls/d to 751 bbls/d over this period. After a productivity reversal from September 2016 to July 2017, as the rig count more than doubled to over 800 and production per rig declined an average of 2.2 per cent per month to 585 bbls/d, productivity has since rebounded 0.8 per cent per month to 674 b/d.

These productivity improvements were gradually bringing down breakeven prices even before the 2014 oil price collapse, and this was despite frenetic activity in the industry, and have been countering rising cyclical costs in the U.S. LTO patch since the industry rebounded in 2017.

As of now, even the IEA indirectly acknowledges this to be the case, despite failing to translate this factor into their medium-term outlook for U.S. oil production. The agency wrote: “Companies maintain that a significant share of the cost deflation seen over the past four years is structural and here to stay due to great efforts to improve operational efficiency, including digitalization.”

And as more digitized drilling information becomes available we will continue to see both an expansion of overall technically recoverable resource and sweet spots in major LTO basins due to improved drilling plans. For example, the EIA has had to ramp up its assessment of technically recoverable U.S. crude oil resource (TRR) each year this decade on the back of rapidly rising LTO resource, a trend that is likely to continue for the foreseeable future. Between the beginning of 2008 and 2017 — with the January 2017 estimate from Annual Energy Outlook 2019 — the EIA has increased its TRR estimate for U.S. crude oil by 105 billion barrels to 303 billion barrels, despite the U.S. producing 23 billion bbls of crude in the interim.

The EIA did not release a formal TRR estimate for LTO early in this period, given the newness of the Shale Oil Revolution, but LTO appears to have contributed most, if not all, of the increase in total crude oil resource. LTO resource accounted for 113 billion bbls of the 2017 estimate, more than a third of the total, with 48 billion bbls of that located in the prolific, low-cost Permian Basin.

GPC’s outlook

GPC is far more bullish on U.S. oil production assuming US$60/bbl Brent crude — equivalent to around US$55 for WTI at both Cushing, OK and Midland, TX, once pipeline and export constraints are resolved by the end of this year — to the extent that we don’t believe these prices are sustainable through 2024 when combined with our more moderate outlook for global oil consumption.

U.S. oil production increased a record 2.17 million bbls/d to 15.3 million bbls/d last year, with crude contributing almost three-quarters of this growth and NGLs the remainder. The IEA implied a major reason for this growth was relatively high crude prices, with spot Brent averaging over US$71/bbl last year. But in fact, U.S. crude prices weren’t really that high, with spot WTI at Midland, the pricing hub for the Permian Basin, the major driver for the country’s crude growth, averaging just US$57.64 per barrel due to regional pipeline constraints.

In the four years that LTO production was firing on all cylinders — 2012, 2013, 2014 and 2018 — annual growth in U.S. oil production averaged 1.52 million bbls/d, and that was before international oil companies (IOCs) got serious about developing short-cycle U.S. LTO. As highlighted by the IEA in their recent outlook, “Exxon and Chevron have made the Permian a centrepiece of their strategies, while Shell and BP are increasing their positions … As a result, 2019 might be the first year where investment growth in shale assets passes from independents to big oil companies.”

Based on US$60 Brent, GPC would expect U.S. oil production to increase an average of 1.38 million bbls/d per year — over twice the growth forecast by the IEA — between 2018 and 2024 to 23.73 million bbls/d, with annual growth declining from 1.75 million bbls/d in 2019 to one million bbls/d in 2024.

Despite their relatively optimistic outlook for global oil consumption and relatively pessimistic one for non-OPEC supply, the IEA warned that some form of market management would have to remain in place by OPEC+ — OPEC members and their non-OPEC allies such as Russia — to support prices for a good chunk of the 2019-24 period. The call on OPEC crude declines from 31.07 million bbls/d in 2018 to a low of 30.1 million bbls/d in 2020 before gradually rising to 32.04 million bbls/d in 2024.

In fact, OPEC+ market management efforts are likely to break down in 2021 — if not sooner — based on GPC’s more moderate outlook for global oil consumption and more optimistic outlook for U.S. oil production, assuming US$60/bbl Brent. The call on OPEC crude drops 890,000 bbls/d in 2019, another 1.63 million bbls/d in 2020, and a cumulative total of 3.52 million bbls/d by 2021, roughly twice as much as the hard fought production cuts OPEC+ agreed upon for the beginning of 2017 to resurrect crude prices.

As a result, GPC believes a far more plausible scenario for U.S. oil production and crude prices over the medium-term is rapid growth in U.S. output into 2021 and then another oil price crash, akin to 2014-16 when then Saudi energy minister Ali al-Naimi tanked the world oil market in a vain attempt to sabotage the U.S. LTO industry. In this case, U.S. oil production increases 3.35 million bbls/d between 2018 and 2020 with Brent crude at US$60/bbl, 800,000 bbls/d in 2021 as Brent slips to an average of US$50, and U.S. output flatlines from 2022 to 2024 with Brent at US$40/bbl each year.

The U.S. Shale Revolution isn’t called a revolution for nothing.