The Unconventional Oil Supply Boom: Aggregate Price Response from Microdata.

AuthorNewell, Richard G.

    Crude oil is the largest commodity market and the shale revolution has dramatically altered U.S. oil supply and markets in recent years. New technological developments such as seismic imaging, hydraulic fracturing, and horizontal drilling have unlocked vast "tight oil" reserves previously thought to be nonviable. This has led to the largest and most rapid surge in oil production in U.S. history, amounting to several million barrels per day (BPD) of additional supply in just a few years (see Kilian 2017 and Prest 2018 for an overview of the impact on oil and gasoline prices). This dramatic expansion in the United States' role in oil supply suggests an enhanced responsiveness of oil production to price fluctuations, increasing supply responsiveness. Along with one of the most significant downturns in global oil prices and questions surrounding OPEC's interest in acting as a market stabilizer, the tight oil boom has also raised questions about whether U.S. unconventional oil might play the role of "swing producer". In this paper, we extend the methods from Newell, Prest and Vissing (2019) to investigate the relative price responsiveness of unconventional and conventional oil drilling in the United States, estimating the degree to which these supply dynamics have fundamentally shifted. We extend the methods from that paper to additionally estimate how well productivity (i.e., oil production per well) responds to price changes, given the competing forces of declining reservoir quality and induced investment in improving well output.

    We estimate the differences in price responsiveness for oil wells using a detailed dataset on more than 150,000 oil wells in the five major oil-producing states of Texas, North Dakota, California, Oklahoma, and Colorado. We estimate the price responsiveness at three key stages of production: drilling, spud-to-production time, and production from existing wells. We find that the important margin for the price response is drilling activity, estimating drilling responses of 1.3 for conventional wells and 1.6 for unconventional wells. (1) However, declining resource quality/diminishing returns imply that oil production per well declines somewhat with increased drilling activity, with price elasticities of productivity (output per well) of -0.2 and -0.4 for conventional and unconventional wells respectively. As a result, the supply elasticities are somewhat smaller than the drilling elasticity. Our simulations suggest long-run supply elasticities of about 1.1 and 1.2, which are approximately equal to the sum of the drilling and productivity elasticities.

    Despite these similar elasticities, we find that the much higher overall productivity of unconventional oil wells (which is about 10 times larger initially and 5 times larger cumulatively) magnifies the unconventional drilling responsiveness many times over. We conduct simulations to combine the different stages to show how the rise of unconventional drilling has affected the responsiveness of U.S. oil supply over the past decade. Altogether, the price responsiveness of U.S. oil supply is many times larger due to the rise of tight oil production, primarily owing to higher output per well. For example, simulations calibrated to 2017 market conditions suggest that, in response to a $10 price increase, firms could supply an incremental 1.5 million barrels per day in production (by the end of a multi-year ramp up period). By contrast, simulations calibrated to 2006 conditions (before the tight oil boom began in earnest), suggest the same price increase would spur only about 110,000 incremental barrels per day. This represents an approximately 13-fold larger price response of U.S. supply compared to the pre-shale era.

    We use our simulations to approximate an aggregate U.S. oil supply curve based on our estimates derived from microdata. We run simulations calibrated to the market situation as of 2017, estimating incremental oil production at different price levels ($50 to $80 per barrel) and time frames (6 months, 1 year, 2, years, and 5 years). The simulations suggest that a price rise from $50 to $80 per barrel could induce incremental U.S. production of 0.6 million barrels per day in 6 months, 1.4 million in 1 year, 2.4 million in 2 years, and 4.2 million in 5 years. These magnitudes are significant in the context of the global market, suggesting a significantly larger role for the United States as an incremental producer. However, the time needed to drill and complete wells imply that the production response takes longer than is typically considered for a "swing producer", which has typically been taken to mean a supplier that can increase oil production substantially (say, 1 million barrels per day) in a short period of time (within 30 to 90 days).


    This paper contributes to a growing line of research studying the shale revolution (e.g., Joskow 2013; Covert 2015; Kilian 2017; Steck 2018), price formation in oil markets in general (Hamilton 2009; Kilian 2009; Anderson, Kellogg and Salant2018; Baumeister and Kilian 2016a,b), and firm behavior in the oil and gas industry (Kellogg 2011, 2014; Agerton 2018).

    The literature on price responsiveness of oil supply often compares results for both the short-run and long-run, typically finding smaller short-run supply responses, as expected. Nevertheless, Baumeister and Hamilton (2015) find evidence for a small, positive short-run supply elasticity. Papers analyzing oil extraction elasticities include Griffin (1985); Hogan (1989); Jones (1990); Dahl and Yucel (1991); Ramcharran (2002); and Guntner (2014). Much of this work pre-dates the shale revolution, and even the more recent literature does not distinguish between unconventional and conventional supply.

    While a number recent studies on the shale gas boom have estimated drilling elasticities for gas wells (Newell, Prest and Vissing 2019; Mason and Roberts 2018, Hausman and Kellogg 2015), relatively few studies have touched upon U.S. oil drilling or supply elasticities. A couple of recent studies do estimate oil drilling elasticities, but they tend to analyze conventional wells pre-dating the shale boom of the 2010s. Anderson, Kellogg and Salant (2018) estimate drilling elasticities in the state of Texas using time series data during the 1990-2007 period, finding an elasticity of 0.7. Brown, Maniloff and Manning (2018) also estimate a drilling elasticity of 0.5. (2) Both of those papers are more relevant for conventional drilling; the dataset from the former paper ends in 2007, before the era of shale oil, and the dataset in the latter paper is predominantly composed of conventional wells. (3) Another difference is that both studies limit the length of responses to lagged prices to no more than 1 to 3 months, whereas we allow for potential responses with up to a year lag.

    Relatedly, Rao (2013) finds production responses from existing wells due to large windfall profits taxes enacted in California in the 1980s. Metcalf (2018) explores the impacts on oil and gas production and prices of removing major U.S. tax preferences for the industry, finding modest impacts for oil but somewhat larger impacts for gas prices.

    Other studies have considered other aspects of the shale boom, often with a focus on estimation of the resource base and depletion. Smith and Lee (2017) estimate the elasticity of oil reserves (not to be confused with the elasticity of supply), particularly in context of wide heterogeneity across basins. They find reserves to be fairly inelastic, since marginal, low-productivity wells suddenly made economic by a price increase add little to overall reserves. This highlights the importance of considering not only how drilling responds to prices, but how average well productivity changes as well.

    Smith (2018a) also uses sequential sampling combined with drilling cost estimates to estimate economic recoverable resources in the Bakken shale formation at different price levels. Further, Smith (2018b) uses real options theory to show why some firms have incentives to drill uneconomic wells in order to hold the lease before it expires. Collins and Medlock (2017) provides a qualitative overview of the ability of shale producers to scale-up activity, suggesting that high and rising productivity is a major factor.

    Our paper applies and builds on the methodology established in Newell, Prest and Vissing (2019), which focused on natural gas supply, to estimate the U.S. oil supply response at different stages of the production process. This paper extends Newell, Prest and Vissing (2019) in three major respects (apart from analyzing oil, rather than natural gas, supply). First, we use a much broader dataset, covering a majority of U.S. oil production, whereas that paper only covered Texas gas wells. Second, we further analyze how well productivity responds to price increases, as lower-productivity wells are drilled. Third, we show how production in barrels responds at different price levels and time horizons, sketching out simulated time-varying supply curves.

  3. DATA

    We use well-level data assembled by Drillinginfo, a company that provides information services on upstream oil and natural gas activity. We use Drillinginfo data on oil wells in five states that collectively account for nearly 60 percent of total U.S. oil production: Texas (33 percent), North Dakota (11 percent), California (7 percent), Oklahoma (4 percent), and Colorado (3 percent). (4) These states account for an even larger share of U.S. drilling activity at 73 percent. These states also account for virtually all tight oil plays in the United States, including the Permian, Eagle Ford, Niobrara, Bakken, and Monterey formations. After cleaning the data (discussed below), the wells in our dataset account for nearly half of all U.S. oil production in 2016, the last full year of our data. (5)

    While the Drillinginfo data contains a wealth of information on...

To continue reading

Request your trial

VLEX uses login cookies to provide you with a better browsing experience. If you click on 'Accept' or continue browsing this site we consider that you accept our cookie policy. ACCEPT