Podcast: What has the U.S. shale oil and gas boom taught us about labor markets?
Associate Professor Gregory B. Upton, Jr.
March 25, 2021
CES Associate Professor Gregory B. Upton, Jr., presented a podcast for the International Association for Energy Economics (IAEE) podcast series. Upton discusses how, over the past decade, the advent of oil and natural gas production from shale geological formations fundamentally changed not only global energy markets but also the communities that reside above these formations. He examines how economists might gain insights from this natural experiment about labor markets and business cycles more broadly. (Transcript below.)
This discussion was motivated by three working papers:
Hello. My name is Greg Upton. I am associate research professor at the Louisiana State University Center for Energy Studies. Today, I’m going to talk about what economists have and can learn from the U.S. shale boom about labor markets. My discussion is motivated by current research with colleagues including Ryan Decker at the Federal Reserve Board of Governors, Meagan McCollum at the University of Tulsa, Bulent Unel at LSU and Han Yu at Dalton State University. You can find links to these working papers on the USAEE website.
[0:33] To begin, what is the U.S. shale boom? Shale oil and gas refers to oil and natural gas trapped in shale formations beneath the earth’s surface. Geologists have been aware of these resources for decades, but extracting them economically has been challenging. Shale formations are typically deeper than “conventional” reservoirs. In fact, most conventional reservoirs are the result of oil and natural gas that has naturally migrated upwards from underlying shale formations until being trapped closer to the surface. They are also technically more difficult to produce as shale has “poor permeability” as referred to by geologists.
It was not until the mid-to-late 2000s that the oil and gas industry was able to economically produce from these formations. The process of combining horizontal drilling and hydraulic fracturing (sometimes informally referred to as “fracking”) to produce from shale involves injecting water, sand and chemicals into rock thousands of feet below ground at high pressures allowing the oil and gas to flow to the surface. Due to these technological innovations, U.S. oil and natural gas production had experienced a decade of growth and were at historic levels before the COVID-19 pandemic began. I will generally refer to this increase in oil and natural gas production, and the economic activity associated with it, as the “shale boom.”
[1:59] My research argues that the shale boom creates a unique opportunity to study labor markets for a number of reasons.
First, the shale boom originated from a plausibly exogenous technological shock that impacted local labor markets. Local labor markets were impacted through two channels.
Channel 1 was a productivity shock. Specific workers suddenly became much more productive because of a technological advancement. These workers with the productivity shock were concentrated in the oil and gas extraction and services sectors. Further, most of these directly impacted workers in the areas where the resources were actually extracted were males with a high school education.
The second channel through which local areas were economically impacted was through bonus and royalty payments paid to local landowners. Landowners with oil and gas reserves below their land sometimes received large payments. The word “shaleionaire” was often used to describe someone who became a millionaire from the shale play. Estimates by Jason Brown, Tim Fitzgerald and Jeremy Weber suggest that six of the major shale plays generated $39 billion in private royalty payments in 2014 alone.
The shock was also conveniently concentrated in specific geographic areas that happened to have specific geological formations thousands of feet below the earth’s surface. And the timing of these shocks all coincided with the technological advancements alongside high oil and natural gas prices that allowed for extraction from these formations. This allows researchers to identify areas that received a treatment and still have access to plausible control areas with similar pre-treatment characteristics.
[3:45] Another desirable factor of these local shocks is that each of these formations has different compositions of oil and natural gas that are as good as randomly assigned to areas. For example, the Haynesville and Appalachia areas are overwhelming “dry” natural gas, as opposed to the Bakken which is overwhelmingly liquids. In between, the Eagle Ford has a mix of both liquids and natural gas, and the ratio of oil and gas naturally changes geographically across the play. Exploiting variation in oil and natural gas prices alongside these plausibly exogenous endowments of different resource types (i.e., oil vs gas) allows researchers to measure the timing and intensity of these shocks in both the boom and bust across areas. In other words, this identification strategy provides plausibly exogenous variation both spatially and and over time.
[4:40] Unsurprisingly, a growing literature has documented the economic impacts of oil and gas booms, and resource booms more broadly. For example, work by James Feyrer, Erin Mansur and Bruce Sacerdote in 2017 estimates that shale booms increased aggregate U.S. employment by 640,000 workers and decreased the national unemployment rate by 0.43 during the Great Recession.
While the effect of the resource boom itself is interesting, the question being addressed today is what economists can learn more broadly about labor markets exploiting this plausibly exogenous natural experiment. I will pose two broad areas of interest.
Area 1: How do labor markets adjust to shocks?
When economic growth occurs, who does the growing? In response to positive economic shocks, firms can either expand their existing business operations or create new locations, which economists refer to as “greenfield” establishments. To better understand this concept, consider the following anecdote of a steakhouse. When an economy grows, perhaps existing steakhouses become busier, therefore employing new workers. I will refer to this as the intensive margin of employment adjustment.
But at some point, even with full staff, one steakhouse location can only serve so many customers. This can prompt the opening of new steakhouses in the area. I will refer to these new establishments as the extensive margin.
But these new steakhouses, too, can be categorized. Perhaps Outback Steakhouse might reach capacity and therefore decide to open up new locations in response to the shale boom. These new Outbacks would be classified as a new establishment of an existing firm.
But perhaps a local entrepreneur also sees that Outback is busy and decides to open up a steakhouse of her own. This locally owned and operated business would be classified as both a new establishment and a new firm.
Analyzing establishment level data from the Census Bureau’s Longitudinal Business Database (or LBD), with Ryan Decker and Meagan McCollum, we compare county level employment growth in shale boom areas to plausible counterfactual areas that were not impacted by the boom. We find that approximately two thirds of the cumulative employment growth in shale boom areas eight years after the onset of the boom can be explained by the extensive margin, i.e. new establishments. What makes this perhaps most interesting is that new establishments make up only about 6 percent of the total employment in any given year. Thus, the entry margin is extremely important when analyzing employment fluctuations. Further dissecting this result, we find that 44 percent of the cumulative employment adjustment comes from entirely new firms started by entrepreneurs, while 24 percent comes from new establishments of firms that already existed before the shale boom began. Thus, if economists track the economy by looking only at what existing establishments or existing firms are doing, much of the business cycle adjustment might be missed.
[7:53] Now let’s take this a step further. In the prior mentioned example of a restaurant, a physical establishment and employees will likely be needed to serve customers at a steakhouse. But not all entrepreneurs open up a physical establishment nor hire employees. Some entrepreneurs simply become self-employed. An example would be a truck driver who decides to purchase and operate her own truck. If this business owner does not hire formal employees who receive a W-2 and pay payroll taxes, they will not be observed in establishment level employment, such as in the LBD. In fact, this person could exit establishment level employment to become self-employed.
During a business cycle expansion, the opportunity cost of starting a business is likely higher as employment and earnings are increasing, but also perhaps the chances of success and income potential are higher in self-employment. Thus, the net effect of an economic boom or bust on self-employment is a priori ambiguous and is itself an empirical question.
[8:58] Focusing on micro data from the U.S. Census American Community Survey (ACS), with Bulent Unel, we find a positive and contemporaneous impact on self-employment in shale boom areas, but the impact is short lived, i.e. once the boom subsides, the self-employment adjusts closer to pre-boom levels. Importantly, point estimates suggest that up to a quarter of the total employment adjustment comes from unincorporated self-employed individuals, a group that makes up just 6% of total employment.
Results of these two papers suggest that the extensive margin of employment is important, and must be tracked carefully if economists are to pick up on boom-and-bust cycles. Further, policies aimed at mitigating business cycle contractions, such as monetary or fiscal policy, might pay particular attention to the extensive margin of employment adjustment.
[9:51] Area 2: How do shocks to a subset of workers impact workers in seemingly unrelated
The second research area of interest is how economic shocks impact workers across sectors, gender, and education level. As mentioned earlier, the shale boom created a productivity shock to the oil and gas extraction and services sectors, which makes up less than one percent of U.S. employment. Interestingly, the workers in shale boom areas that received the productivity shock were overwhelmingly male with a high school education. Thus, researchers are able to analyze the response of earnings differentials between male and female workers as well as workers with a college and high school education within sectors that are seemingly unrelated to the initial productivity shock.
Using data from the Quarterly Workforce Indicators, also produced by the Census Bureau, research with Han Yu finds that the shale boom impacted earnings differentials within many sectors of the economy. Specifically, empirical estimates suggest that the college/high school earnings differentials decreased by 3.0 percent in non-mining sectors while male/female earnings differentials increased by 2.6 percent in non-mining sectors.
Let’s refer back to the steakhouse anecdote. Applied in this context, results suggest that both male and female workers at the steakhouse experienced increases in earnings during the shale boom. This is perhaps unsurprising as the steakhouse was likely much busier due to the boom. But what is perhaps more interesting is that males received a larger increase in earnings, even within the steakhouse, relative to female workers. A perhaps plausible story to explain this phenomenon is that the male worker could have received an outside offer to work in the oilfield, and the steakhouse counters that offer to keep him onboard. The female worker, without this outside offer, does not receive the raise. This is just one example of a mechanism plausibly describing this result and more research on the mechanism for this effect is welcomed.
[12:01] In this work with Han, we also decompose the observed changes in earnings differentials into three channels. (1) earnings differentials can change within the mining sector, (2) earnings differential can change within non-mining sectors and (3) a change in the composition of the labor market due to workers substituting between sectors, labor migration, or a change in labor market participation in response to the shale boom. Results of the decomposition show that the overwhelming share of the change in earnings differentials observed in shale boom areas are explained by the second channel; changes within non-mining sectors. We argue that these results highlight the importance of considering differential effects of technology shocks by education and gender in studying earnings inequality more broadly.
[12:55] Moving away from specific empirical papers, I next discuss a few items that I believe should be considered in reading and interpreting the shale boom literature more broadly.
(1) The Great Recession – The shale boom occurred around the time of the Great Recession; a time of historic slackness in the labor market. For instance, the national unemployment rate peaked at 10 percent during 2009, coinciding closely with when U.S. shale production began to increase from its almost 40-year trough in 2008. Following a recession, aggregate labor market is “slack”, meaning that there are plenty of workers seeking employment. Had this shock occurred at a time with a tighter labor market perhaps the economy’s response would have also been different. This is perhaps especially true when interpreting earlier papers with just a few years of post-boom data.
(2) Barriers to entry- The oil and gas extraction sector has relatively low barriers to entry. A male can plausibly get a job working as a “roustabaout” on a rig out of high school, especially during a boom time. A productivity shock in an industry with higher barriers to entry would perhaps be expected to have a higher earnings effect in the short-run, and less employment response.
(3) Representative Areas – Many of the areas impacted by the shale oil and gas booms are relatively rural, and the response of a rural labor market to a productivity shock might not be representative of the U.S. economy as a whole. Notably, Denver Colorado, and Pittsburgh, Pennsylvania, are included in treatment areas in these prior mentioned empirical papers, but the vast majority of shale counties are relatively rural. This should be considered when extrapolating results of this literature into other contexts.
(4) Direct channels – There are at least two direct channels through which an oil and gas boom can stimulate a local economy. First, there is employment associated with the initial drilling and completion of wells. Second, local landowners received bonus and royalty checks for oil and gas production that occurs beneath their land. Much of this literature, including my research, is unable to distinguish between these two channels.
[15:13] In conclusion, exogenous labor market shocks that are concentrated in specific areas, the timing is well understood, and overwhelming impacted specific subsets of workers in a specific sector are rare. Thus, the shale boom offers a rare opportunity to study labor markets. While drawing broad conclusions from any specific shock begs questions of whether results can be generalized, I believe that valuable insights have been garnered from this literature and that there is still much to be learned about labor markets and human behavior from this natural experiment.