Resource booms and the macroeconomy: The case of U.S. shale oil

Abstract

We examine the implications of the U.S. shale oil boom for the U.S. economy, trade balances, and the global oil market. Using comprehensive data on different types of crude oil, and a two-country general equilibrium model with heterogeneous oil and refined products, we show that the shale boom boosted U.S. real GDP by a little more than 1 percent and improved the oil trade balance as a share of GDP by about 1 percentage point from 2010 to 2015. The boom led to a decline in oil and fuel prices, and a dramatic fall in U.S. light oil imports. In addition, we find that the crude oil export ban, a policy in effect until the end of 2015, was a binding constraint, and would likely have remained a binding constraint thereafter had the policy not been removed.

Introduction

Technological advances in horizontal drilling and hydraulic fracturing have led to an unprecedented increase in U.S. oil production. Often referred to as the shale or fracking revolution, the boom in U.S. oil production has renewed interest in the long-standing question on the link between resource booms and economic performance. Several recent papers have focused on the local or regional implications of the U.S. shale boom, suggesting positive economic effects (see, for example, Feyrer et al., 2017; Allcott and Keniston, 2018). However, little is known about the implications of this boom for the U.S. aggregate economy and trade. In this paper, we study the importance and implications of the U.S. oil boom for the U.S. economy, trade balances, and the global oil market. We do so using a dynamic stochastic general equilibrium model of the world economy that takes into account unique characteristics of the U.S. experience: a large increase in production of a certain type of crude oil with an oil export ban in place.

Oil has often not been explicitly modeled in many leading macro and international trade models, partly because the share of oil in aggregate production is small and primary commodities overall account for a modest fraction of global trade. However, recent research has shown that shocks to economically small sectors, such as oil and gas, that feature complementarities with other inputs of production can have disproportionate aggregate effects (Baqaee and Farhi, 2019) and that trade in primary commodities is important with notably large gains from trade compared to gains from trade in standard models (Farrokhi, 2020; Fally and Sayre, 2018).

The relatively few general equilibrium models that do feature oil generally assume that it is a homogeneous good. This is a strong assumption since the characteristics of oil can differ across several dimensions, one of which is density. A key feature of the recent U.S. oil boom is that oil produced from shale deposits via the application of horizontal drilling and hydraulic fracturing is predominantly of one type: light crude. Different types of crude oil are imperfect substitutes for each other in the refining process and refining sectors tend to specialize in processing certain types of oil. The U.S. refining sector is specialized in processing heavier crude oils relative to the rest of the world. This mismatch of increased supply of light oil and existing refining capacity for heavier oil in the U.S. has important implications for the use and trade of various types of crude oil. The importance of this mismatch was potentially magnified by the U.S. export ban on crude oil, a policy in effect until the end of 2015.

In assessing the implications of the U.S. light oil boom quantitatively, we make two contributions to the literature. First, we introduce two previously unexamined sources of heterogeneity into a general equilibrium model with endogenous oil prices. The first source of heterogeneity arises from the different types of oil produced that are imperfect substitutes into the refining process. The second stems from the difference between refineries in the U.S. and the rest of the world (ROW). Our model also features an occasionally binding export ban on U.S. crude oil.

Our second contribution is to assemble a comprehensive data set that contains information on crude oil quality in order to build our model on solid microeconomic foundations. These data inform the building of our model in two ways. First, we use simulated method of moments (SMM) to estimate a number of model parameters to match moments from oil-related and macro data. Of particular note, we estimate three key parameters related to the refining sector: the elasticities of substitution across different oil types and the elasticity of substitution between oil and other factors of production. Second, we carefully calibrate other model parameters targeting a set of first moments for oil-related and macro variables. One key point to highlight is the importance of examining detailed oil data and introducing heterogeneity in crude oil types and in refining technology. If we were to only use aggregate data and pool different types of crude oil into one single oil sector, we would not be able to assess the implications of the shale oil boom for trade in different types of oil, relative prices of oil, and specialization of the refining sector. We also show these heterogeneities are key to properly understand the impacts of the crude export ban.

An essential initial step for our analysis is to document how some important oil market variables have changed during the U.S. shale oil boom. Using various sources, we gather data on production and prices of different types of crude oil as well as trade flows and refiner use of different types of oil. We document that from 2010 to 2015 U.S. light oil production more than tripled, while production increases outside the U.S. were from medium and heavy crudes. In addition, the U.S. refiners' use of light oil increased substantially from 2010 to 2015. Meanwhile, their medium crude use declined and heavy crude use increased. We document dramatic shifts in the quantity and types of oil being imported as well: U.S. light oil imports dropped sharply, medium oil imports declined and heavy oil imports increased since the shale boom. These facts help motivate the features of our model.

Our two-country (U.S. and ROW) general equilibrium model with heterogeneities and an export ban also has the following features. In addition to oil and refined products (fuel), both countries produce a non-oil good.¹ The non-oil good is used for consumption and investment, which is costly to adjust, and as an input in the production of oil. Oil is only used to produce fuel, while fuel is consumed by households and also used as an input to produce the non-oil good. An internationally traded bond allows for the possibility of trade imbalances. To simplify, we abstract from distinguishing between traded and non-traded goods, a common feature in many commodity and resource papers that discuss the Dutch disease effect. The reasons are that, contrary to even other advanced (small) open economies that are net commodity exporters, where the commodity sector can be relatively large, the share of oil and gas in U.S. GDP is relatively small, having not exceeded 3 percent since the early 1980s. The U.S. also remains a net importer of crude oil.

We model the shale boom as a series of positive technology shocks that replicate the increase in U.S. light crude production from 2010 to 2015 and then illustrate the general equilibrium outcomes. Our main findings are three. First, we find that the shale oil boom had important impacts on several U.S. macroeconomic variables, notably on real GDP and trade balances. According to our model, the boom boosted U.S. real GDP by a little over 1 percent from 2010 to 2015, which accounts for about one tenth of actual GDP growth over this period. This suggests that the boom has contributed to the recovery from the Great Recession. There is also major improvement in the U.S. oil trade balance, by about one percentage point (as a share of GDP), in line with the data.

Second, our model can match several important aspects of U.S. oil market data during the boom despite relying on a single shock, a light oil technology shock. This includes the sharp drop in U.S. imports of light crude oil, the increased use of light oil by U.S. refiners, and the drop in both the use and imports of medium crude oil by U.S. refiners. On the other hand, we find a counterfactual decline in U.S. refiner use and imports of heavy crude oil. However, we show that the model can explain the changes in the heavy crude data successfully if we add a second shock, a ROW heavy oil supply shock, into the model.

Third, we find that the U.S. crude export ban was a binding constraint, particularly in 2014 and 2015, and that it primarily distorted the upstream and downstream oil sectors and petroleum trade, with negligible impacts on macroeconomic aggregates. We find the ban artificially depressed U.S. light crude oil prices, inflated light crude oil prices outside the U.S., and distorted the relative price of light crude to other types of crude oil. Oil producers, with the exception of light crude producers outside the U.S., were negatively impacted by these price distortions. We find U.S. refiners benefited from those price distortions, as they provided a cost-advantage, leading them to over-process light crude oil and take market share from refiners elsewhere. However, the ban had little impact on the global fuel supply or fuel prices as there was no ban on trade in refined products. Impacts on households, both in the U.S. and ROW, were negligible. Finally, we show that taking into account the heterogeneity in crude quality, and properly modeling and calibrating the refinery sector are key to examine the effects of the ban.²

Our paper analyzes the implications of the U.S. oil boom for the U.S. economy, trade, and the global oil market, taking into account specific characteristics of the U.S. experience. The model and calibration approach we use are similar to a number of other papers that focus on oil and international real business cycles. More generally, our work draws on and has connections with several literatures that focus on commodities and the macroeconomy.

First and foremost, our paper relates to a large literature that uses real business cycle models to analyze the effects of oil price fluctuations and other oil shocks on the economy.³ One distinguishing aspect of our work from much of this prior literature is that we analyze the impact of an oil boom brought about by a technology shock in the domestic oil sector. The DSGE literature has tended to focus on how the domestic economy is affected by exogenous oil price shocks or oil supply shocks that originate outside the economy. The model developed in this paper is closely related to Bodenstein et al. (2011) with some key differences regarding the distinction between different types of oil, oil production and the inclusion of a refining sector. Careful modeling of oil production and refining are theoretical contributions of our paper, allowing a granular analysis of the oil market. Our paper also relates to Manescu and Nuno (2015) who analyze the international effects of the U.S. shale oil boom using the three-country model of Nakov and Nuno (2013). In addition to differences in modeling, such as different types of oil, a refinery sector and explicit components of the U.S. economy, our discussion focuses heavily on U.S. macroeconomic aggregates and disaggregated oil market variables.

Our paper complements two recent studies, Farrokhi (2020) and Fally and Sayre (2018), who use detailed data on commodities and incorporate oil and other primary commodities into multi-country models of trade to address the role of commodities in trade. Farrokhi (2020) develops a static, multi-country, general equilibrium framework that incorporates a detailed model of global sourcing of oil inputs by refineries, and downstream demand for refined petroleum products. He then conducts several policy experiments, including the U.S. oil boom from 2010 to 2013 and the implications of lifting the U.S. crude oil export ban.⁴ Some key differences between our work and Farrokhi (2020) are that we study the dynamic effects of the shale oil boom on macroeconomic aggregates, upstream and downstream oil sectors, and trade, through the lens of a two-country DSGE model. Our model also allows for capital accumulation and trade deficits.

More broadly, our work relates to an extensive literature on resource booms. Earlier studies in this literature have focused on small countries with a large dependence on resource production, and many suggest an adverse impact on economic growth (see van der Ploeg, 2011 for an overview). More recent papers, however, find that a resource boom can have a positive effect on the overall economy if it is driven by increasing resource activity (see, for example, Bjornland and Thorsrud, 2016; Bjornland et al., 2019; Allcott and Keniston, 2018).⁵

The non-traded sector has been a key component of the analysis done in this line of research.⁶ Our work, while linked with this literature given our focus on U.S. oil boom, does not consider the non-traded sector and is, therefore, silent on any issues related to it. There are two other important differences between our paper and this literature. The first is the relative importance of the commodity sector for the domestic economy. Even in the developed small open economies studied in the resource boom literature, the GDP share of the commodity under study can be as high as 12 percent, with the share in exports up to 30 percent. The share of the oil and gas sector in U.S. GDP, on the other hand, is relatively small, having a recent peak of 2.2 percent in 2008, while the share of crude oil in U.S. exports was 3.7 percent in 2019. A second difference is the net importer status of the U.S.: the U.S. was a major net importer of crude oil at the beginning of the resource boom and remains a net importer now.⁷

Several recent papers analyzing the regional effects of shale oil and gas booms in the U.S., provide useful insights on the impact of natural resources on economic performance. Using detailed micro data, these studies show that oil and gas booms increased overall income and wages and non-mining jobs in the producing regions (see for example, Weber, 2014; Feyrer et al., 2017).⁸

In addition, a line of empirical research has discussed the impact of the U.S. oil boom on crude oil price differentials between U.S. and international crude oil prices (see for example, Bornstein et al., 2018; Agerton and Upton, 2019; Plante and Strickler, 2020). Kilian (2016) also examines how the oil boom has affected the evolution of crude oil and gasoline prices.

Our work complements these papers by analyzing the implications of the shale oil boom, a resource boom stemming from a resource (technology) shock, for the global oil market and the U.S. using a DSGE model. We show that the oil boom contributed importantly to economic activity despite the U.S. being a net-importer of oil with the resource (oil) sector representing a relatively smaller share of the aggregate economy.

The remainder of the paper is organized as follows. Section 2 presents data. Section 3 develops the model. Section 4 presents the calibration of the baseline model. Our main results are discussed in Section 5 while Section 6 discusses sensitivity analysis with respect to a number of model features. We conclude in Section 7. An online appendix presents additional results (including extensive figures related to the sensitivity analysis), a variety of analytical results derived from simpler stylized models, and details on our data, calibration, and estimation procedures.