Elsevier

Energy Policy

Volume 35, Issue 5, May 2007, Pages 3074-3088
Energy Policy

Testing Hubbert

https://doi.org/10.1016/j.enpol.2006.11.004Get rights and content

Abstract

The Hubbert theory of oil depletion, which states that oil production in large regions follows a bell-shaped curve over time, has been cited as a method to predict the future of global oil production. However, the assumptions of the Hubbert method have never been rigorously tested with a large, publicly available data set. In this paper, three assumptions of the modern Hubbert theory are tested using data from 139 oil producing regions. These regions are sub-national (United States state-level, United States regional-level), national, and multi-national (subcontinental and continental) in scale. We test the assumption that oil production follows a bell-shaped curve by generating best-fitting curves for each region using six models and comparing the quality of fit across models. We also test the assumptions that production over time in a region tends to be symmetric, and that production is more bell-shaped in larger regions than in smaller regions.

Section snippets

Introduction and context

Since the beginning of commercial exploitation of oil, there has been great interest in two related questions: how much oil exists in the world, and when will we run out of oil? This very old discussion has recently resurfaced, as interest in oil depletion has increased along with increasing oil prices. Recent projections of global oil production have been made using a set of methods commonly referred to as the “Hubbert theory” of oil depletion, but these projections have been rejected by those

Methods of analysis

In this paper we test three assumptions of the Hubbert theory. We first ask if bell-shaped models fit historical production data better than other simple models. We then ask if regional oil production curves have been historically symmetric. Lastly, we test a commonly made assertion about oil depletion: that the Hubbert model fits larger regions better than smaller regions, due to a “smoothing” behavior resulting from summing smaller production curves. We emphasize that we do not test the

Best-fitting model results

One important general result is that 16 regions were disqualified from comparison due to extremely poor fit, and six more were classified as borderline nonconforming. The disqualified regions are not a significant portion of global production (about 3% of 2004 production), but the borderline-nonconforming regions represent 36% of global production. Thus, fully a third of global production is not well characterized by models with a single up–down cycle.

Discussion and conclusion

It is clear from the results of this analysis that no simple, single cycle model fits all historical production curves from oil producing regions. We illustrated that when comparing the three symmetrical models, the Hubbert model is the most widely useful model, but that somewhat less than half of the regions are well-described by the linear and exponential models. We also showed that when asymmetry is allowed in our oil production curves, that the asymmetrical exponential model becomes the

Acknowledgments

I would like to gratefully acknowledge the assistance of Anand Patil, Alex Farrell, Andrew Mills, and Jim Kirchner in the preparation and revision of this paper.

References (34)

  • BP, 2005. BP statistical review of world energy. Technical Report, British...
  • BP, 2006. BP statistical review of world energy. Technical Report, British...
  • C.J. Campbell

    The Coming Oil Crisis

    (1997)
  • C.J. Campbell

    Oil Crisis

    (2005)
  • C.J. Campbell et al.

    The end of cheap oil

    Scientific American

    (1998)
  • Campbell, C.J., Sivertsson, A., 2003. Updating the depletion model. In: Second International Workshop on Oil Depletion,...
  • K.S. Deffeyes

    Hubbert's Peak: The Impending World Oil Shortage

    (2001)
  • Cited by (161)

    • European carbon storage resource requirements of climate change mitigation targets

      2022, International Journal of Greenhouse Gas Control
      Citation Excerpt :

      Several curve-fitting models exist to describe the S-Shaped pattern, but the logistic model is the most widespread. The logistic model has been widely used to predict peak production in oil and coal consumption, and projecting long-term trends in energy systems, infrastructure, and technology development (Hubbert, 1956, Grübler et al., 1999, Brandt, 2007, Rutledge, 2011, Höök et al., 2011) Recently, the logistic model was applied to the analysis of global carbon storage resources (Zahasky and Krevor, 2020).

    View all citing articles on Scopus
    View full text