Report Contents

Report#:EIA/DOE-0607(99)

Preface

Trends in Power Plant Operating Costs

Sectoral Pricing in a Restructured Electricity Market

Modeling the Costs of U.S. Wind Supply

Modeling Technology Learning in the National Energy Modeling System

Employment Trends in Oil and Gas Extraction

Price Responsiveness in the NEMS Buildings Sector Models

Annual Energy Outlook Forecast Evaluation

National Energy Modeling System/Annual Energy Outlook Conference Summary

Completed Report in
PDF Format (862 KB)

Related Links

Forcasting Page

EIA Homepage

by
James M. Kendell

Employment in the Upstream Sector

Economic Impacts

A Model of Industry Employment

Forecasting the Level of Employment

Conclusion

Appendix

Introduction

Between October 1997 and December 1998 the imported refiner acquisition cost of a barrel of oil dropped by almost half, from $18.73 per barrel to $9.39 per barrel.¹ According to the Bureau of Labor Statistics (BLS), U.S. Department of Labor, employment in the upstream oil and gas industry had declined by more than 50,000 jobs by February 1999, leaving the United States with less than half the number of oil and gas extraction jobs it had during the early 1980s, when oil prices were more than four times as high in real terms.²

This paper presents an econometric model that can be used to forecast industry employment. On an average annual basis, the number of upstream oil and gas jobs was 325,900 in 1998. Based on the model’s parameters and inputs from the Annual Energy Outlook 1999 (AEO99), the average level of employment is expected to decline to 273,000 in 2000.³ These projections are based on a reference case in which the average imported refiner acquisition cost (world oil price) is expected to be $13.97 a barrel in 2000, and the average lower 48 natural gas wellhead price is expected to be $2.10 per thousand cubic feet for the year, both in 1997 dollars. If prices— and therefore, drilling—turn out to be higher, the level of employment is also expected to be somewhat higher than in the reference case.

Regardless of what happens in the short run, industry employment can be expected to increase between now and 2010. Based on the activity levels expected in the AEO99 reference case, employment in 2010 is projected to equal approximately 350,000, in large part because of increased drilling. In the low world oil price case, employment in 2010 is projected at 329,000 jobs. In the high world oil price case, 2010 employment is projected to grow to 372,000.

Employment in the Upstream Sector

Oil and gas extraction activities are part of a larger petroleum industry, including petroleum refineries, wholesale terminals, and gasoline stations. Since 1972, employment in the entire petroleum industry has ranged from 1.3 to 1.7 million workers (Figure 1). Employment peaked in 1981, when the real imported refiner acquisition cost of a barrel of crude oil was almost $63 a barrel in 1997 dollars.⁴ Since then, petroleum industry employment has declined, as the numbers of oil and gas wells drilled, operating refineries, and wholesale jobbers have declined.

Figure 1.  U.S. Petroleum Industry Employment, 1972-1998 [source]

This analysis focuses on employment in the oil and gas extraction (or upstream) industry. In collecting data on the industry, the BLS considers two primary sectors—a service sector and a production sector. Work in the service sector is performed on a contract basis. Service jobs include drilling, geological services, and well repair work performed under contract. Although some drilling occurs in the BLS production sector, most occurs in the service sector. Employment in both sectors increased in the wake of the sharp rise in real oil prices in the late 1970s but has fallen almost every year since 1982 (Figure 2). Since 1982, the level of employment in both sectors has declined by almost 400,000. As of 1998, the average level of employment in the oilfield service sector was 186,100, and the production sector had an average employment level of 135,000.

Figure 2.  U.S. Production and Service Jobs and Wells Drilled, 1970-1998 [source]

The decline in employment from 1982 to 1998 can largely be attributed to the sharp decline in oil and gas drilling over the same period. From 1982 to 1998 the number of wells drilled annually fell from about 84,400 to 24,200 (Figure 2). The principal factor contributing to the decline was the collapse of oil prices after 1985. For example, the price of oil, as measured by the imported refiner acquisition cost, fell from $53.33 a barrel (in real 1997 dollars) in 1982 to $11.97 in 1997, and the average wellhead price of natural gas declined by half in real terms. Prices affect jobs in the upstream oil and gas industry through their effects on drilling and production. Prices tend to have an immediate effect on exploratory activity, including seismic studies and wildcat drilling, which tends to last only a few months. Their initial impact on production activities—such as developmental drilling and operation of lease equipment, which tend to last for many years—is smaller. Thus, employment increases when oil prices rise, but not as much as prices.

Other factors that have contributed to the decline in drilling and employment include technological advances (such as horizontal drilling), which have made it possible to extract more oil and gas with fewer wells and fewer workers. Factors that have tended to mitigate the downward trend in drilling and employment include lower drilling costs and favorable tax policies. For example, in 1982 the average gas well cost $1.39 million (1997 dollars) to drill, but by 1997 average costs had declined almost continuously to $723,000 per well.⁵ Tax policies such as tax credits for unconventional drilling and royalty relief for deepwater production have also served to moderate the decline in drilling and employment.

Technology has had both positive and negative impacts on jobs over the 1982-1997 period, creating new jobs in data processing and interpretation but reducing the need for oil and gas drilling platforms. Major advances in data acquisition, data processing, and the technology of displaying and integrating seismic data with other geologic data have created computer and analysis jobs but reduced the number of exploratory wells needed. The introduction of subsea well technologies, tension leg platforms, and production spars have opened up vast new and promising areas for exploration in the deepwater areas of the offshore that had been inaccessible, creating more offshore jobs but reducing conventional drilling activity. Another significant cost-saving technology, adopted in the later part of the 1980s, was horizontal drilling. Most reservoirs are wider than they are deep, and drilling a horizontal, as opposed to a conventional vertical well enables more of the reservoir to be exposed to the wellbore. Fewer wells need to be drilled, but more skills are needed in data analysis and directional drilling.

Finally, the decline in employment has been ameliorated by the shift in drilling in favor of natural gas. In 1982, the number of oil well completions was more than double the number of gas completions. By 1993, the number of gas completions had exceeded the number of oil completions. In October 1998 twice as many successful gas wells were drilled as oil wells, and in February 1999 three times as many gas wells were drilled.⁶ This shift in the composition of drilling has tended to slow the rate of decline in employment, because gas wells are generally deeper than oil wells and hence require more labor inputs.

The October 1997 to December 1998 decline in upstream employment was similar to the decline that occurred in the 15-month period from October 1990 to January 1992. Some 31,300 jobs were reported lost in the most recent period and 34,700 in the early 1990s period. During both periods the imported refiner acquisition cost of a barrel of oil dropped by about half.⁷ The October 1997 to December 1998 period was quite different from January to July 1986, another period when oil prices fell by half. During those 6 months in 1986, 127,000 oilfield workers lost their jobs, because some producers and lenders began shifting their expectations for future prices from a view that oil prices would continue to rise to a view that they would hold steady, varying around an average.

This was primarily because in late 1985 Saudi Arabia—facing increasing needs for oil revenue—abandoned the role of swing producer that it had played during the first half of the 1980s. In the most recent two periods of job loss, in contrast, the industry’s fundamental price outlook was largely unchanged.⁸

Over a few months’ time, even a doubling in prices historically has had little immediate effect on upstream oil and gas employment. From June to October 1990, imported refiner acquisition costs of crude oil more than doubled in the run-up to the invasion of Kuwait by Iraq. But in those 4 months only 3,500 oil and gas extraction jobs were added in the United States, less than a 1-percent increase. For rising prices to create new upstream jobs, they must be sustained for a substantial length of time—more than just a few months—because producers need to be confident that they will get an adequate return on their investments.

Economic Impacts

In the 14 months from January 1998 to March 1999, the five largest oil and gas producing States (Texas, Louisiana, Oklahoma, California, and Alaska) suffered proportionally fewer job losses than their production share. In 1997 the five largest oil and gas States (Figure 3) produced 77 percent of U.S. oil and gas; however, they suffered only 41 percent of the total 46,700 job losses from January 1998 to March 1999.⁹ This was because a larger share of marginal oil and gas production is outside these five States and because company headquarters and producing sector employees who contribute to the operation of oil and gas production in many States are located in these five States.

Figure 3.  Oil and Gas Production by State, 1997 [source]

As the overall number of upstream oil and gas jobs has declined, they have become less important to State economies. In Texas, the State that has the most oil and gas extraction jobs and produces the most oil and gas, extraction jobs peaked in January 1982 at 313,700 or 5.0 percent of the total Texas labor market.¹⁰ By March 1999 the preliminary estimate for employment in upstream oil and gas had fallen to 149,800, less than half of the peak, and accounted for only 1.6 percent of the State’s total employment.¹¹ While oil and gas jobs declined, the rest of the Texas economy grew, boosting total employment by more than 2.8 million over 18 years. As a result, the recent low oil prices have not hurt the Texas economy as much as they did in the 1980s.

A similar story can be told at the national level. When oil and gas extraction employment peaked in 1982, 0.7 percent of civilian workers were directly employed by the upstream petroleum industry. By 1998, however, only 0.2 percent of civilian workers were directly employed by the upstream petroleum industry. While the number of oil and gas extraction jobs has been declining, the number of other civilian workers has been increasing (Figure 4).

A Model of Industry Employment

Employment in the oil and gas extraction sectors was econometrically estimated for both the service and production sectors. The regressions contained the following independent variables (see Appendix): the number of total wells drilled in a year (Drilling), the share of successful wells accounted for by natural gas (Gas share), the share of total U.S. oil production accounted for by operations in Alaska (Alaska share), the share of total U.S. oil and gas production accounted for by offshore activity (Offshore share), and a structural change variable (New Era).

The level of drilling activity was included as an explanatory variable because employment is closely tied to the demand for the geological, construction, and geophysical services needed to drill a well. Oil and gas prices provide the foundation for making a decision about whether to drill a well, but job loss and creation are more closely related to the actual level of drilling.¹² For example, between January 1996 and August 1997, when oil and gas prices were largely flat, drilling increased by about 30 percent, in part because of expectations of higher future prices. Reflecting the increase in the level of drilling, the level of upstream service employment increased by almost 20 percent or approximately 30,000 employees.

The level of employment depends not only on the number of wells drilled but also on the type of wells drilled. For this reason, the share of successful wells that are classified as gas wells was included as an independent variable. The Alaska share of oil production and the offshore share of total oil and gas production were included as variables to account for the fact that the level of drilling in those locales is a relatively poor proxy for their overall activity levels. Offshore drilling accounts for less than 5 percent of total drilling, but because of the nature of the wells—substantially more expensive, but also more productive than onshore wells—the offshore is important to upstream employment.

The model also recognizes that the structure of the industry changed between 1970 and 1997. One of the most important changes is the increased importance of both offshore operations and operations in Alaska. Previously, activity in those locales played only a very minor role in determining the overall level of U.S. upstream employment. That is no longer the case, however, given the reduced industry focus on lower 48 onshore drilling, which has become less profitable. Reduced industry drilling in the lower 48 onshore resulted, in part, from the increase in investment opportunities outside the United States after the fall of the Berlin Wall and the increased likelihood of returns on those investments. To incorporate this “regime change,” the coefficients on these variables were permitted to change after 1990 through the inclusion of the variable New Era, which is a binary variable equal to 1 after 1990.

To allow for nonlinearities, the employment levels and level of drilling were represented in terms of their natural logarithms. The equations were estimated with annual data from 1970 through 1997. Employment levels in the “service” and “production” sectors were formulated using two separate equations, but they were estimated together using the seemingly unrelated regression technique, so as to obtain more efficient estimates. The model explains more than 99 percent of the variation in employment levels in both sectors over the sample period. The parameters are reported in the Appendix to this paper. All the parameters were statistically significant.

Other model specifications were considered. For instance, it was hypothesized that employment could be affected by the proportion of drilling that is successful— i.e., that a higher success rate, other things being equal, would result in more wells being completed for production and hence in more employment. It was also hypothesized that the overall level of production as well as wellhead revenues could contribute explanatory power to the model. However, the empirical results did not support any of these hypotheses, and hence the variables were excluded from the forecasting equation.¹³ All the estimated coefficients were tested for evidence of structural change, as discussed above, but the results were included in the forecasting equation only when there was statistically significant evidence of structural change.

This analysis accounts only for direct oil industry jobs, not the additional, associated jobs that would be affected in the oil and gas retail and consumer service industries, at manufacturers working on oil industry projects, or in pending oil company mergers. Local department stores, automobile dealers, and even school districts can be expected to suffer during a period of low oil prices, as incomes and property values decline. Manufacturers far distant from the oilfields might also suffer a business decline in a period of low oil prices.¹⁴ Shrinking revenues also force oil companies as well as individuals to search for ways to cut costs. In addition to the recently completed merger of British Petroleum with Amoco, the largest prospective oil company merger in U.S. history, between Exxon and Mobil, is pending. On the other hand, lower oil prices can be expected to stimulate economic activity in the consuming sectors and, therefore, increase the number of jobs in the rest of the economy. Oil is an input to production for many industrial and manufacturing processes, and lower prices will lead to increased output and more jobs.¹⁵

Forecasting the Level of Employment

Similarly, if the number of wells turns out to be higher or lower than projected in the reference case, the number of upstream jobs will also be higher. If the number of wells drilled is 10 percent higher than the reference case projection over the entire forecast, the number of upstream jobs would be 7 percent higher, all other things being equal. A similar falloff in jobs can be expected if the number of wells drilled is 10 percent lower over the entire forecast. The effect of a difference in drilling from the reference case increases slightly over time as the industry adjusts to the new drilling levels. For example, 10 percent more drilling starting in 1997 yields 5 percent more upstream jobs in 2000.

Conclusion

Appendix