Skip to content

Learning to Frack

Technological advances and learning-by-doing have made U.S. shale oil profitable even at $55/barrel.

Just ten years ago shale oil was expensive. Global oil prices spiked to $135/barrel in 2008 but shale oil didn’t and couldn’t respond. Now, at only $55/barrel, U.S. oil producers are going all in, announcing billions of dollars of increased investment, particularly in Texas and North Dakota, and energy experts like Daniel Yergin are expecting U.S. production to increase this year by more than 500,000 barrels per day or about 6%.

This remarkable transformation is a perfect example of learning-by-doing.  Cost reductions in solar, wind, and batteries get a lot of attention.  But fossil fuel producers also learn from experience, and the pace of learning in U.S. shale oil over the last couple of years has been impressive. Recent research by Thom Covert and others help us make sense of what happened, and identify several key lessons for other industries.

Shale Oil Ramps Up, Even at Lower Prices

U.S. oil production in 2015 reached its highest level in decades, driven large increases in production from shale oil.  With shale oil and other forms of “tight oil”, producers drill horizontally and then use hydraulic fracturing to reach oil trapped in low-permeability rock like shale, sandstone, and limestone.  Over the last decade, these technologies have provided access to vast areas of oil reserves that were previously out of reach.  The growth has been particularly dramatic in the Eagle Ford, Permian, and Bakken, but other areas have grown rapidly as well, and shale oil now represents almost half of all U.S. production.

shaleoilNote: After slowing down in 2016, U.S. shale oil is expected to ramp up again in 2017, continuing a decade-long surge. Constructed by Lucas Davis (UC Berkeley) using EIA data.

Production slowed in 2016 as oil prices briefly dipped below $30/barrel.  Companies made severe budget cuts, reduced drilling activities, and U.S. oil production fell 12% between July 2015 and September 2016.  Remarkably though, even at $30/barrel, production still continued at a solid pace.  Despite historically low oil prices, the U.S. still produced almost 9 million barrels per day during 2016.

Now with oil prices at $55/barrel, U.S. producers are pushing their chips to the center of the table.  Hess, one of the biggest producers in North Dakota, just announced a $2.25 billion investment budget for 2017, up 18% from last year.  Noble Energy has said they will spend up to $2.5 billion for 2017, up 67% from last year.   Companies believe these investments will be profitable because U.S. production has become so much more efficient.

Practice Makes Perfect

Economists have long studied learning-by-doing. As a company makes more of a good, it tends to become more efficient at production.  That is, the company learns to make output using fewer resources; or, equivalently, the company learns how to make more output for a given set of inputs.  Leaning-by-doing has been shown to happen in virtually all industries, with some of the best-known studies in economics coming from aircraft, semiconductors, and shipbuilding.

These industries may at first seem quite different from hydraulic fracturing, but there is a key similarity.  In all of these industries, companies are performing a repetitive production process.  Boeing builds hundreds of airplanes.  Chesapeake Energy drills hundreds of wells.  Key to learning is repetition, so industries like this are best suited to learning-by-doing.

rigNote: Oil production in North Dakota’s Bakken shale.  Photo from here.

Tens of thousands of oil wells have been drilled in the United States since 2010.  This repetition has allowed producers to experiment with innovative new approaches to drilling and finishing wells; new combinations of drilling fluids, well depths, and other factors. This trial-and-error over thousands of wells has facilitated learning, pushing new techniques forward and reducing the cost per well.

Not all markets look like this. For example look at building nuclear power plants.  Nukes are such massive projects, undertaken so infrequently, that there is less scope for learning.  Not surprisingly, studies have found only limited learning-by-doing for nuclear power plants.  See here, here, and here.  This is exactly the rationale for small modular reactors – proponents argue that in building much larger numbers of smaller reactors they can benefit from learning-by-doing and decrease costs.

Evidence from the Bakken

In the most comprehensive study to date, University of Chicago economist Thom Covert documents rapid learning-by-doing by shale oil producers in North Dakota’s Bakken Shale.  Covert gathered detailed data from all oil wells drilled in the Bakken over an eight year period, including total oil production as well as information about the sand, water, and other “inputs”.  Covert shows that over just a small number of years, producers learned to make much more profitable choices, squeezing up to three times as much profit out of a given well.

One of the interesting patterns is that over time, producers have learned to be much more aggressive, extending the horizontal section of wells much farther and then using much more water and sand. By the end of the period, producers were using five times (!) as much water and three times as much sand per foot of horizontal length.  This “super-size’’ approach has been adopted successfully with natural gas production as well.  Super-size wells cost more upfront, but yield much larger amounts of oil and gas.

bakken.pngNote: Reproduced with permission from, this map shows the expected returns from using additional water and sand.  The black dots are the locations of actual wells.

It is complicated, however.  Covert shows that the optimal set of inputs varies dramatically across locations. The heat map above shows that some locations are much more suited to super-sizing than others.  Using more water and sand increases production in some places (the red areas), while decreasing production in others (the blue areas).

Global Technology Spillovers

Learning-by-doing in U.S. shale oil is important not only for the United States, but also because of the prospects for global technology spillovers.  The United States is not the only country with shale oil formations, and it would be naive to think these techniques will not quickly spread around the world.

That said, there is an important feature of the U.S. oil and gas market that makes it particularly well-suited for learning. Unlike most other countries, in the United States there is a wealth of publicly-available information.  Regulators and third-parties collect and disseminate detailed data on each and every well drilled in the United States.  Other countries hoping to grow their oil production would do well to adopt similar practices for information disclosure.

Moreover, the United States has private mineral rights.  This makes it possible for many firms, including smaller firms, to enter the market and try new drilling approaches.  In contrast, where mineral rights are government controlled, it can be harder for companies to enter the market and we might expect there to be less experimentation. U.S. oil and gas production has long been characterized by ingenuity, innovation, and risk-taking – all typical of competitive markets.

Still, it seems inevitable that these techniques will quickly spread around the world.  Globally there are already an estimated 400+ billion barrels of tight oil reserves, so there are a lot of places where these techniques could be profitably employed. This means the global supply curve for crude oil will continue to shift outward, applying continued downward pressure on global crude oil prices.

Keep up with Energy Institute blogs, research, and events on Twitter @energyathaas.
Suggested citation: Davis, Lucas. “Learning to Frack”, Energy Institute Blog, UC Berkeley, February 13, 2017.





Lucas Davis View All

Lucas Davis is the Jeffrey A. Jacobs Distinguished Professor in Business and Technology at the Haas School of Business at the University of California, Berkeley. He is a Faculty Affiliate at the Energy Institute at Haas, a coeditor at the American Economic Journal: Economic Policy, and a Research Associate at the National Bureau of Economic Research. He received a BA from Amherst College and a PhD in Economics from the University of Wisconsin. His research focuses on energy and environmental markets, and in particular, on electricity and natural gas regulation, pricing in competitive and non-competitive markets, and the economic and business impacts of environmental policy.

11 thoughts on “Learning to Frack Leave a comment

  1. This paper is one of the worse that I have ever seen related to “fracking”. Frankly, Haas School should have its “hand” slapped for publishing it on your site. The facts are false and misleading, including the Univ of Chicago reference. The costs for this are not measured correctly and above all exclude the externalities and other costs — danger, shipping, processing etc that are part of the problem with oil and gas. I hope that Haas reveals who funded this more than just “questionable” paper. I encourage Hass to get accurate and correct research data to publish on this topic — and make it public.

  2. This is excellent. Data published by Rystead, a Norwegian consulting firm suggest the rate of decline in costs may exceed 10% per year. One can even say that Moore’s law applies to oil and gas. Should that be the case those who invest in “old oil and gas” as well as the sovereign debt of oil exporting countries should start to worry.

  3. Great article! Given that public information and well defined property rights help lower barriers to entry and foster learning, it would be interesting to explore what kind of lags exist on a firm’s gains in output based on whether they are operating in a fracking-averse regulatory environment or otherwise. By assuming (a big “if”) that the output potential of shale-rich states is similar once firms have been operating for a certain amount of time, it might then be enlightening to explore any significant difference in output to learn if institutional constraints damper realized gains from learning. Your chart showing shale oil production would seem to indicate that even over the course of an environmental friendly administration and falling oil prices between 7/2015 and 9/2016, these regulatory and market drags on output were not enough to dampen the gains made from learning at a national level. Drilling operations continued to produce more shale oil, cheaper. I would love to see if the same story plays out state-by-state (good luck controlling for “everything else”).

%d bloggers like this: