Actuarial Review Return to Main Page

Quarterly Review

Cheap Oil for How Long?

Hubbert's Peak: The Impending World Oil Shortage by Kenneth S. Deffeyes
(Princeton University Press, 2001, $24.95)

Reviewed by Curtis Gary Dean

In 1956, a geologist named M. King Hubbert, working at the Shell research lab in Houston, made an unsettling prediction: U.S. oil production would peak in the early 1970's and then begin to decline. Of course this controversial prediction was challenged, but according to the author, the actual peak production year turned out to be 1970. Foreign oil became essential to the U.S. economy.

The author of this book, a professor emeritus at Princeton University, and others have applied Hubbert-type methods to global oil production and forecast that world oil production will peak sometime during 2003-2009. Take a moment to ponder the economic effects that such an event would have on our oil-hungry world! The author says that there is nothing we can do to delay this reckoning. Aggressive new initiatives, whether for more conservation, alternative energy sources, or more expensive oil discovery and retrieval methods, will take years to bear fruit. We will be extracting oil from the ground for decades to come, but the peak production year will come long before the last drop is recovered.

The author was a colleague of Hubbert in Shell's Houston laboratory and later joined the Princeton faculty. He has been close to the oil business his entire life and covers a lot of material in this concise book of 208 pages. Much of the book is devoted to explaining how oil was created, the geological conditions necessary for the formation of underground oil reservoirs, and how to find it and extract it. Given the importance of oil in modern life, I would highly recommend this book. One shortcoming of the book is that some of the explanations are so brief that I found them hard to follow, but I did grasp the general concepts.

In a chapter titled "The Origin of Oil" the author explains that more complex hydrocarbons from organic matter deposited millions of years ago were broken into oil's shorter molecules (commonly referred to as the cracking process) given the proper temperatures. Ground temperatures increase by about 14°F per 1,000 feet and the right temperatures can be found in the "oil window" from 7,500 feet (180°F) to 15,000 feet (295°F) underground. For oil to be created, organic-rich source rocks had to be buried for a million years or more in this oil window. If the material is buried below 15,000 feet, the higher temperatures break the molecules into the smaller molecules of natural gas. Above 7,500 feet the temperatures are too low to crack the organic molecules into oil.

After oil is created, the necessary geological formations have to exist for oil fields to survive. Without the right conditions the oil would leak away or be devoured by microbes. Understanding the requisite geological formations can also provide clues on where to find oil fields.

The author describes some of the discovery and extraction technologies developed to obtain this valuable and highly profitable natural resource.

The author outlines the methods that he, Hubbert, and others have used to estimate peak production years. A key assumption is that a bell-shaped distribution (a smooth curve symmetric about the mode) fits annual oil production. The x-axis represents the production year and the y-axis the amount of oil produced. Normal, Lorentz, and logistic curves are all bell-shaped and the author explains why he believes that a normal curve is the best choice. He displays a graph showing that the normal curve fits U.S. oil production quite well. Because U.S. oil production peaked in 1970, there are data points on both the ascending and descending sides of the curve. For global oil production, only data points on the ascending side of the curve are available and the trick is to predict the peak, that is, the turning point. As actuaries, we know well that identifying turning points is a risky endeavor.

One variation of the forecast methods can be summarized as follows:

  1. Collect the historic data on annual oil production by year,
  2. Estimate the total oil that will ultimately be produced by adding cumulative past production to recoverable reserves, both known reserves plus estimated future discoveries, and
  3. Find the normal curve that best fits the data points from (1) such that the area under the curve matches (2).

Using 2.1 trillion barrels of oil as the ultimate cumulative global oil production, this method gives a peak production year of 2009. The author states that 2.1 trillion barrels is a "reasonably generous upper guess."

Note how (2) resembles incurred losses: paid losses [past oil production] + case reserves [discovered oil reserves] + IBNR [future discoveries]! Do any readers want to try our reserving methodologies on this problem?

Another method is to fit two cumulative normal curves. The lower cumulative normal is fit to cumulative production while the higher curve is fit to cumulative production plus current reserves (i.e., already discovered oil). With an additional assumption about the spacing between these two curves, a peak production year of 2003 and ultimate cumulative oil production of 2.12 trillion barrels result.

These forecasts are certainly open to debate. First, why should a normal curve fit the data? Second, the ultimate cumulative global oil production figure of 2.1 trillion barrels is questionable. The U.S. Geological Survey has estimated that 3.012 trillion barrels can be produced globally, a number that the author considers to be way too high. Jon Evans directed me to two Web sites that readers may want to check out. One is, which shows estimated oil reserves by country. The other is, which is managed by a group of people who are studying questions about peak and ultimate oil production.

Do I believe these forecasts? Not being an expert in the field, I have to reserve judgment. I am intrigued enough to look for ways to put some money into the game by finding profitable ways to invest, possibly in companies with large oil or gas reserves. Hey, what good is a forecast if you can't make a few bucks from it?!