Are We Running Out of Oil?

Policy Backgrounders | Environment

No. 159
Wednesday, January 29, 2003
by David Deming


Hubbert's Prediction of Declining Production

Despite these facts, some environmentalists claim that declining oil production is inevitable, based on the so-called Hubbert model of energy production. They ignore the inaccuracy of Hubbert's projections.

Problems with Hubbert's Model.

Figure IV - Hubbert Predictions - 1956

"Environmentalists now tie their predictions of declining energy supplies to M. King Hubbert's model of energy production -- which has been consistently inaccurate."In March 1956, M. King Hubbert, a research scientist for Shell Oil, predicted that oil production from the 48 contiguous United States would peak between 1965 and 1970.24 Hubbert's prediction was initially called "utterly ridiculous."25 But when U.S. oil production peaked in 1970, he became an instant celebrity and living legend.

Hubbert based his estimate on a mathematical model that assumes the production of a resource follows a bell-shaped curve - one that rises rapidly to a peak and declines just as quickly. In the case of petroleum, the model requires an accurate estimate of the size of the total oil endowment.26 His best estimate of the size of petroleum resources in the lower 48 states was 150 billion barrels. His high estimate, which he considered an exaggeration, was 200 billion barrels.

Based on these numbers, Hubbert produced two curves showing a "best" estimate of U.S. oil production and a "high" estimate. The claimed accuracy of Hubbert's predictions are largely based on the upper curve - his absolute upper limit [see Figure IV].

  • Hubbert set the absolute upper limit for peak U.S. oil production at roughly 3 billion barrels a year, and his best or lower estimate of peak future U.S. crude oil production was closer to 2.5 billion barrels.
  • As early as 1970, actual U.S. crude oil production exceeded Hubbert's upper limit by 13 percent.
  • By the year 2000, actual U.S. oil production from the lower 48 states was 2.5 times higher than Hubbert's 1956 "best" prediction.

Production in the 48 contiguous states peaked, but at much higher levels than Hubbert predicted. From about 1975 through 1995, Hubbert's upper curve was a fairly good match to actual U.S. production data. But in recent years, U.S. crude oil production has been consistently higher than Hubbert considered possible.

Figure V - Hubbert Predition - 1980

"U.S. oil production has been higher than Hubbert thought possile."

Hubbert's 1980 prediction of U.S. oil production, his last, was substantially less accurate than his 1956 "high" estimate.27 In the year 2000, actual U.S. oil production from the lower 48 states was 1.7 times higher than his 1980 revised prediction [see Figure V].

In light of this, it is strange that Hubbert's predictions have been characterized as remarkably successful. While production in the United States is declining, as Hubbert predicted, it is doing so at a much slower rate. Furthermore, lower production does not necessarily indicate the looming exhaustion of U.S. oil resources. It shows instead that at current prices and with current technology, less of the remaining petroleum is economically recoverable.

Hubbert's Prediction for Natural Gas.

In 1998, Peter McCabe of the U.S. Geological Survey showed that energy resources do not necessarily follow Hubbert-type curves, and even if they do a decline in production may not be due to exhaustion of the resource.28

For example, Hubbert also predicted future U.S. natural gas production. This prediction turned out to be grossly wrong. As of 2000, U.S. natural gas production was 2.4 times higher than Hubbert had predicted in 1956.29

The Production Curve for Coal.

Production of anthracite coal in Pennsylvania through the 19th and 20th centuries followed a Hubbert-type curve more closely than any other known energy resource. Production started around 1830, peaked around 1920, and by 1995 had fallen to about 5 percent of its peak value. However, the supply of Pennsylvania anthracite coal is far from exhausted. If production were to resume at the all-time high rate of 100 million short tons per year, the resource base would support 190 years of production. Production declined not because the resource was depleted but because people stopped heating their homes with coal and switched to cleaner-burning oil and gas.30

"U.S. production in 2000 was 1.7 times higher than Hubbert projected in 1980."

The primary problem with a Hubbert-type analysis is that it requires an accurate estimate of the total resource endowment. Yet estimates of the total endowment have grown systematically larger for at least 50 years as technology has made it possible to exploit petroleum resources previously not considered economical. Hubbert-type analyses of oil production have systematically underestimated future oil production. This will continue to be the case until geologists can produce an accurate and stable estimate of the size of the total oil endowment.


Read Article as PDF