ARTICLE IN PRESS

Energy Policy 37 (2009) 1267–1276

Contents lists available at ScienceDirect

Energy Policy

0301-42

doi:10.1

� Corr

E-m

journal homepage: www.elsevier.com/locate/enpol

The peak of oil production—Timings and market recognition

Pedro de Almeida a,�, Pedro D. Silva b

a Computer Science Department, University of Beira Interior, Covilha, Portugalb Electromechanical Department, University of Beira Interior, Covilha, Portugal

a r t i c l e i n f o

Article history:

Received 25 September 2008

Accepted 6 November 2008Available online 13 January 2009

Keywords:

Energy markets

Peak oil production

Oil prices

15/$ - see front matter & 2008 Elsevier Ltd. A

016/j.enpol.2008.11.016

esponding author. Fax: +351 275319899.

ail address: palmeida@di.ubi.pt (P. de Almeid

a b s t r a c t

Energy is essential for present societies. In particular, transportation systems depend on petroleum-

based fuels. That world oil production is set to pass a peak is now a reasonably accepted concept,

although its date is far from consensual. In this work, we analyze the true expectations of the oil market

participants about the future availability of this fundamental energy source. We study the evolution

through time of the curves of crude oil futures prices, and we conclude that the market participants,

among them the crude oil producers, already expect a near-term peak of oil production. This agrees with

many technical predictions for the date of peak production, including our own, that point to peak dates

around the end of the present decade. If this scenario is confirmed, it can cause serious social and

economical problems because societies will have little time to perform the necessary adjustments.

& 2008 Elsevier Ltd. All rights reserved.

1. Introduction

Energy is the lifeblood of present human societies. Withoutcommercial energy (e.g. electricity, natural gas, crude oil and itsrefined products, coal, and biomass energy products that enterformal commercial circuits) societies as we know them wouldcrumble. In particular, fossil fuels are essential for electricitygeneration and also to propel modern transport systems. How-ever, in the last few years serious concerns about the futureavailability of those non-renewable fuels have been brought topublic discussion. In particular, the production of crude oil seemsunsustainable in a relatively short time frame, and even if thepresent levels of production can still be somewhat increased inthe next few years, that increase will probably be insufficient tomatch the rapidly growing consumption in countries like China,India, Iran, or Saudi Arabia. At present, the growing tightnessbetween world production and demand has already caused asignificant rise in oil prices, seriously affecting most worldeconomies. These facts, together with growing concern aboutCO2 emissions, are inducing most countries to increase theproduction of renewable energies. However, most of the relevantrenewable energy production techniques are related to theproduction of electricity, not liquid fuels, and so their impact inthe transport sector is limited.

The most recent statistical data from IEA (2007) indicates thatthe fossil energy is about 81% of the total commercial energy

ll rights reserved.

a).

consumed in the world1 and about 98% of the energy used in thetransport sector. These percentages show that, at present, thecombined alternatives to fossil energy represent a relatively smallproportion of the total energy consumed, and a negligibleproportion of the energy used in the vital transportation sector.The very limited present share of the renewable energies is due toimportant problems that most of them still face, in terms ofeconomic competitiveness of the present technology. Since theseproblems still prevent a fast ramp-up of the most promisingalternatives, there seems to be no escape from the dependence onliquid fossil fuels for the foreseeable future. However, theproduction of those fuels (mainly crude oil) is approaching amaximum, and will then begin to decline (Hubbert, 1949;Campbell and Laherrere, 1998; Campbell and Heapes, 2008), aphenomenon termed ‘‘Peak Oil’’ (PO). Although disputed whenfirst presented, the general idea of a future peak of oil productionis now well accepted. The time frame for that peak, however, isstill under discussion. Although most of authors that study thissubject already expects a near-term PO, some of them stillmaintain that the peak of oil production is so far away in time asto be irrelevant (Jackson, 2006). In addition, several authors arealready pointing to concerns about a relatively near-term ‘‘peakgas’’ (Laherrere, 2003a; Simmons, 2007), as well as a somewhatmore distant ‘‘peak coal’’ (Hubbert, 1971; Zittel and Schindler,2007). At present, conventional crude oil is the most important ofthe fossil fuels and it is all but certain that its depletion process is

1 The other (non-fossil) sources of energy are nuclear (6.3%), hydro (2.2%),

geothermal and solar (0.5%), combustible renewables and waste (10.0%). Electricity

is not an energy source as such but depends on the conversion of the previously

referenced energy sources.

ARTICLE IN PRESS

P. de Almeida, P.D. Silva / Energy Policy 37 (2009) 1267–12761268

more advanced. It should be noted that, when talking about thePO problem, the main concern is not the end of oil as an importantenergy source, but only the peak of its production (and theresulting reduced availability, price increase, and associatedeconomic and social effects).

At present, the available data covering the world’s oil reservesand the production and depletion rates of most of the oil fields arescarce and uncertain. Inevitably, the various approaches to predicta PO date suffer from this limited information and the resultingestimates developed by different experts tend to show a widescatter. This uncertainty about the PO date diminishes the sense ofurgency that the problem should command. In fact, althoughseveral important recent studies (e.g., Hirsch et al., 2005) point tothe need of urgent mitigation efforts to limit the economic andsocial impacts of the PO, it seems clear that many world leadersand government officials are badly informed and that almost nocountry is acting consistently to implement effective mitigationchanges.

This pervasive indifference about the PO problem is not justthe result of direct ignorance, since it is shared by someindividuals and organizations that undoubtedly should be wellinformed. In fact, mainstream media (e.g., New York Times—seeMouawad, 2007), dedicated information organizations (e.g., IEAand EIA—see IEA, 2006; EIA, 2006), energy consulting firms (e.g.,CERA—see Jackson, 2006), and even oil extraction companies (e.g.,Exxon—see Tillerson, 2007) and some leaders of countries likeSaudi Arabia (Jum’ah, 2007) still try to defend publicly the ideathat the PO problem is not very important, or at least that it is nota short-term problem of serious concern. However, these publicstatements contradict the (growing) majority, among those thatstudy this issue that predicts a relatively near-term PO, and somesignal that they may be partially explained by direct marketingconcerns.

These lingering public ‘‘not-a-problem’’ opinions from organi-zations related to the oil industry, and the confusion andmisunderstandings they still cause, prompted us to try to derivean innovative approach to evaluate the effective beliefs of theparticipants in the crude oil markets, among which the oil-producing companies are extremely significant. As such, theobjective of this paper is to evaluate the evolution of the beliefs(in terms of acknowledgement of the PO problem) of thepetroleum market participants, through the analysis of theevolution in time of the crude oil futures price-curves.

2 Association for the Study of Peak Oil and gas (ASPO) (http://www.peakoil.-

net/).

2. The peak oil problem

The problem of the peak of oil production (PO) was introducedin Hubbert (1949). Before that, concerns about a quick exhaustionof crude oil were occasionally formulated, but none of them wasbased on a scientific analysis of the problem.

After that first presentation of the problem, Hubbert, a highlyrespected oil geologist, went on studying this subject and, in 1956,presented an extremely important paper (Hubbert, 1956), inwhich he accurately predicted that the date for the peak of oilproduction for the continental USA would be around 1970. Thatpaper also presented the classical approach to the prediction ofthe unconstrained production profile for a region (or for theworld), explaining that the production should follow a bell-shaped curve. At first, that paper raised significant discussion,since the USA was by then the biggest producer of crude in theworld, and its production was increasing without apparentproblems—in fact, it had to be limited by Government regulationto prevent the ‘‘flooding’’ of markets. However, the ultimatesuccess of that prediction established Hubbert’s methodology. In1971, Hubbert published a paper in which, within some

constraints, he predicted the world conventional crude productionwould peak around 2000 (Hubbert, 1971). Hubbert’s model didnot incorporate the artificial limitations to oil production due topolitical constraints imposed by OPEC in 1973 and 1980. Inabsence of those occurrences, it can be argued that this 1971prediction again would seem close to be correct.

Hubbert can rightly be considered the father of the researchfield concerned with the sustainability of the production ofnatural resources, and of the PO theory. His prediction technique,based on the fitting of a bell-shaped curve to the historicalproduction and to the ultimately recoverable reserves (URR),remains the most used approach to predict future production ofexhaustible natural resources and, in particular, of fossil resources.

Mainly using Hubbert’s approach, Campbell and Laherrerepresented an extremely influential paper in 1998 (Campbell andLaherrere, 1998), discussing the near-term expected peak of oilproduction, and predicting it to occur ‘‘before 2010’’. This paperinfluenced the ‘‘modern’’ discussion of the PO problem, and itsglobal impact, leading the authors to create the ASPO,2 probablythe first and presently the most relevant international organiza-tion dedicated to the study of the PO problem.

In terms of direct indicators to a relatively near-term PO, one ofthe most evident is the fact that in the last few decades thediscovery of new oil fields has been lagging the extraction of oilfrom previously discovered fields. In the last few years, thistendency has become much worse, so that today new fields beingdiscovered represent only about a fourth of the oil being extracted(ASPO, 2007). Another indicator pointing to a short-term PO is thestrong reduction in the OPEC spare production capacity, clearlyillustrated in Fig. 1. The non-OPEC oil-producing countries havebeen producing at almost full capacity since 1990. The evolutionof oil prices through time, illustrated in Fig. 2, is itself a furtherindicator of a fundamental tightening of the production/demandbalance and, in fact, for many observers it is the most ominous ofthem.

In this context, the depletion of the presently producing oilfields constitutes a growing problem. Depletion rates typicallyrange between 2% and 4% in ‘‘well behaved’’ big onshore oil fieldsand up to 18% in some deep-water fields (Jackson and Eastwood,2007). With 86 Mb of daily production, it is easy to understandthat a permanent effort is needed just to maintain the presentlevel of production. This effort includes the opening of new oilwells in presently producing fields, increased water injection, theuse recent technologies like 4D analysis and horizontal andmultilateral wells, and the use of several techniques of enhancedoil recovery (such as nitrogen, CO2, or steam injection). Even so,the discovery of new oil fields is necessary to compensate for thereduction of the production rates of the present oil fields. Thedepletion problem is being compounded by the recent peak ofproduction of some the biggest oil fields in the world (such asCantarell and Burgan), and by the growing proportion ofproduction from deep-sea oil fields that tend to be depletedextremely fast.

These various indicators that point to a near peak of world oilproduction also illustrate the production problems in individualcountries. Many countries, among them some important produ-cers like the USA and Indonesia, had their individual peaks yearsago, and go on losing production in spite of the technologicadvances in exploration and production and of the increaseddrilling efforts brought about by the present ‘‘high’’ crude oilprices. Additionally, in the last few years other importantproducing countries like Venezuela, Nigeria, Norway, and Mexico

ARTICLE IN PRESS

Fig. 2. Monthly average values for the West Texas Intermediate crude oil spot prices, in USD/barrel, from 1986 to 2007 (from the EIA website: http://www.eia.doe.gov).

Fig. 1. OPEC unused production capacity (from Mawdsley, 2006 and IEA, oil market report website: http://omrpublic.iea.org).

P. de Almeida, P.D. Silva / Energy Policy 37 (2009) 1267–1276 1269

have entered an initial phase of diminishing production, whileother countries that were already post-peak, such as the UK, haveincreased the downward slope of the production curve. Also, thetwo most important producers in the world, Russia and SaudiArabia, seem to be near the peak production or even already attheir peaks (Simmons, 2007; Krane, 2007; Stuart, 2007).

Given this situation and the past effort to find petroleum in allthe ‘‘easy places’’, it is an accepted fact that almost all of the ‘‘big’’oil fields (i.e., above 500 Mb) that still remain to be found lie inextremely hard to explore ‘‘final frontiers’’ (Bakhtiari, 2006).Those are the deep-sea offshore, the Artic, and some limitedplaces on land or in shallow waters, where drilling is very difficultbecause of extreme natural conditions (i.e., Siberian regions wherethe ground freezes and melts on yearly cycles, Kashagan, Rub alKhali).

Apart from ‘‘liquid’’ petroleum (i.e., crude that is able to flownaturally through pipelines) there are alternative fossil productsthat, with significant efforts (in terms of investment, technology,energy inputs, environment impact, etc.), can be converted tocrude-like oil or to petroleum products. Among them, the Albertatar sands and the Orinoco bitumen are already in production andrepresent very significant reserves, but the increase in theirproduction rates is very slow, and the maximum productioneventually achieved will always be limited. Even harder toproduce is the oil shale for which, at present, there is no viableproducing technology (Jiang et al., 2007). This represents a

petroleum source-rock, which has not been heated sufficientlyin nature to give up its oil, and has to be retorted to do so with alow net energy yield. Shale reserves are extremely large and existin many countries (the USA having the largest known reserves),but eventual production is dependent on uncertain technologicadvances and in any case the ramp-up in oil shale production willalways be very expensive and relatively slow. Those unconven-tional resources may have an important mitigation role near thepeak oil date, and afterwards, but they will only be able torepresent a limited fraction of today’s production and consump-tion of petroleum.

2.1. Peak oil date prediction

There are several methods used to predict a date for the peak ofthe world oil production.

The ‘‘business as usual’’ method does not incorporate thephysical limits of the oil production. It is mainly used byorganizations that, for several reasons (political, commercial,and so on), do not want to admit the reality of the PO and so‘‘predict’’ that the oil consumption (and, by necessity, itsproduction) will simply go on following the historical growthtrends. A short but interesting discussion of these approachesbased purely on economic models, and their limitations, ispresented in Hallock et al. (2004).

ARTICLE IN PRESS

P. de Almeida, P.D. Silva / Energy Policy 37 (2009) 1267–12761270

Another approach is based on fitting a parametric oil produc-tion curve directly to the observed production profile. The curve isadjusted to represent a total final production that is expected toequal the ultimately recoverable reserves (URR). The mostfrequent methodology is the fitting of a bell curve, following thepioneer works of Hubbert. However there are two importantlimitations with this approach. The first one is the lack ofagreement on a value for the URR.3 The second one is related tothe limitations of the Hubbert curve approach, which is fullyapplicable only to unconstrained production. Despite thoselimitations, this general approach is still the most commonlyused for the prediction of a PO date.

The so called ‘‘bottom-up analysis’’ of crude oil production isbased on specific data about the predicted production of all therelevant new wells/fields, due to enter production and alreadyannounced in the specialized publications. This new production isthen subtracted by the expected depletion rates of the presentlyproducing fields. This methodology was used by Chris Skrebowskiin his well-known ‘‘MegaProjects’’ articles (e.g., Skrebowski, 2007)and also by Rembrandt Koppelaar (e.g., Koppelaar, 2006). This isprobably the most powerful approach that can be used topredict future oil production, but it also has some importantlimitations. One of them is that it is relatively accurate only for thenext four to five years, since that is the typical time horizon of thepublicly announced firm new developments. Also, in this kind ofapproach it is almost impossible to find reliable informationabout all the future field developments, and so it is common touse information relating only to biggest (and better known)future projects (Skrebowski, 2006). Additionally, predicted datesfor new production almost invariably tend to be too optimistic.Finally, the most important limitation of this approach is the lackof a reliable value for the global decline rate of existing oilproduction.

To try to mitigate the effect of these well-recognized problems,in 2005, we have performed an analysis following a modifiedbottom-up methodology (Almeida and Silva, 2005). Our approachwas based on conducting this kind of analysis on a country-by-country approach instead of in a direct global scale. This seems arelatively simple way of limiting some of problems describedabove. This variant of the traditional bottom-up analysis is basedin individual production curves for each country and presentssome significant advantages: it eases the task of consideringindividual depletion rates for largest oil fields; allows theaccounting for spare production capacity at the level of individualcountries; supports a better evaluation of potential projectslippage; and allows the inclusion of other particularities of eachcountry’s production. We believe that, since our approach reachesa global result from adding the more detailed individualcontributions from each country, it allows the incorporation ofmore available information, and gives a more precise end result.Our approach also has the practical advantage of being easy toupdate, since instead of global revisions it allows for individualcountries updates that can be performed when relevant newinformation is available for each country’s production.

For our analysis, we used the countries’ production informa-tion from the IEA and EIA and the new projects data fromSkrebowski’s ‘‘Megaprojects’’ database (Skrebowski, 2004, 2006)and from Koppelaar’s ‘‘Oil Supply Analysis’’ (Koppelaar, 2006). Our2005 evaluation, for total fossil liquids production (i.e., crudeoil+lease condensate+NGL+liquids from tar sands and from bitu-men), predicted a peak situated between 2008 and 2012, with a

3 Even disregarding some less credible (and more radical) sources, the world

URR can be something between 2500 Gb (ASPO, 2007) and 4821 Gb (Laherrere,

2003b).

maximum probability at the center of that interval (Almeida andSilva, 2005). However, the exact peak date was not very relevant,since the production profile predicted for that time interval wasbasically a plateau. In January 2007, we updated this analysis andobtained a slightly earlier peak date probably situated between2007 and 2011.

Recent revisions, using the same methodology and updateddata, again point for a peak date situated between 2008 and 2012.The production profile, we foresee is still basically a productionplateau followed by a slow production decay, initially of about0.5%/yr, starting around 2012. This rate of decay will almostcertainly increase in the following years, but our model does notgenerate reliable estimates for production decrease rates forlonger time frames, since it is limited by the availability ofconcrete new projects’ data.

We should note that, in our prediction, most of the uncertaintyof the actual PO date is dependent on the behavior of a singlecountry: Saudi Arabia. The future production capacity of thiscountry is at present very uncertain, and it is absolutely critical fordefining a more exact world peak date.

This lack of detailed information about some important oil-producing countries that limits the precision of our prediction hasthe same effect in all the similar works done by others. Searchingthe literature, one can find multiple date predictions for peak oil.A list of the PO dates estimated by some of the best-knownexperts in this field is presented in Table 1. However, in spite of theuncertainties, Table 1 shows that the predictions point to PO datesfocused around 2010 or, at most, up to the middle of the nextdecade. This should certainly be an important aspect to factor intothe decisions taken by those involved in the oil markets, since itrepresents the best information available on this subject.

3. Analysis of futures prices

Considering the uncertainty of the peak oil predictions and thedivergence of our own estimate for a peak production date withsome more distant dates presented by others, we tried to validateour early prediction using an independent methodology. Thismethodology is based on the analysis of the evolution in time ofthe crude oil futures prices curve.

3.1. Oil markets and futures prices curves

In today’s market, futures contracts are one of the mostimportant instruments for the trading of oil (Haubrich et al.,2004). By buying or selling futures contracts, the trader assumesboth the right and the obligation to execute the (buying or selling)trade at the predefined future date. However, the trader canterminate this obligation before the expiration date by closing thefutures position in the market. This simply requires the acquisi-tion of an opposite position to compensate the one previouslyheld. While open, the futures contracts involve a continuousadjustment to the changing market price, for each particularcontract date. This adjustment is processed through paying orreceiving cash corresponding to the change of the value of theopen positions of each trader that result from the oscillations ofthe market prices.

This way, futures contracts can be used to guarantee the sellingof future production at a specified price or to guarantee thebuying of products needed in the future at specified prices.Futures contracts can also be used to invest/speculate by traderswho believe that the presently specified price for a future date istoo high or too low. If he buys a future contract, the trader willgain if the price rises above the present price for that date, andwill lose if the price drops. If he sells a contract, he will gain if the

ARTICLE IN PRESS

Table 1Peak oil projected dates from several authors.

Date of

forecast

Source PO date Reference

2000 Bartlett 2004–2014 Bartlett (2000)

2000 EIA 2021–2112 Wood and Long (2000)

2000 IEA Beyond 2020 IEA (2000)

2001 Deffeyes 2003–2008 Deffeyes (2001)

2002 Nemesis 2004–2011 Nemesis (2002)

2002 Smith 2011–2016 Smith (2002)

2003 Simmons 2007–2009 Simmons (2003)

2003 Deffeyes Before 2009 Deffeyes (2003)

2003 Campbell Around 2010 Campbell (2003)

2003 World Energy

Council

After 2010 WEC (2003)

2003 Laherrere 2010–2020 Laherrere (2003a, b)

2003 Shell 2025 or later Davis (2003)

2003 Lynch No visible peak Lynch (2003)

2004 EIA 2021–2112 Wood et al. (2004)

2004 Bakhtiari 2006–2007 Bakhtiari (2004)

2004 Skrebowski After 2007 Skrebowski (2004)

2004 Goodstein Before 2010 Goodstein (2004)

2004 CERA After 2020 Jackson and Esser (2004)

2005 Koppelaar After 2010 Koppelaar (2005)

2006 Skrebowski After 2010 Skrebowski (2006)

2006 Smith 2011 Smith (2006)

2006 Koppelaar After 2012 Koppelaar (2006)

2006 IEA After 2030 IEA (2006)

2006 CERA 2035 Jackson (2006)

2007 Robelius 2008–2018 Robelius (2007)

2007 Koppelaar 2015 Koppelaar (2007)

2007 Laherrere About 2015 Laherrere (2007)

2008 CERA After 2017 CERA (2008)

2008 Shell 2020 or later Shell (2008)

P. de Almeida, P.D. Silva / Energy Policy 37 (2009) 1267–1276 1271

price drops and will lose if the price rises. It is possible to tradefutures contracts for dates ranging from the next month up toseveral years.4 At each specific moment, there are different pricesfor the contracts for each future date. Plotting these pricesaccording to their distance in time, forms what is usually calledthe ‘‘futures prices curve’’.

When analyzing the expected behavior of futures prices inrelation to present (spot) prices, standard economic theorypresupposes that in the absence of changing production costs orrisks, the futures prices should rise as the date of the contractsgets more distant in the future (a situation called ‘‘contango’’).This is because in an equilibrium economic setting (in which thereare no imbalances that allow a correcting arbitrage activity), thecost of maintaining products in storage and the inflation should beincorporated in future prices (or there would be no incentive at allto maintain stocks). This classical theory was first presented byHotelling (1931) and argues that the futures prices should simplyrise to match the nominal interest rate. However, some authorsdefend that in ‘‘normal’’ circumstances, the prices of crude futurescontracts should be in backwardation (the opposite of contango:the situation when present prices for the futures contracts getlower as their dates get more distant in the future). This point ofview is validated by the direct analysis of historical prices. In fact,the oil futures markets are frequently in backwardation, as notedby several authors (e.g., Haubrich et al., 2004; Litzenberger andRabinowitz, 1995; French, 2005; Tabak, 2003). These authorspresent several reasons to justify this frequent backwardation.Most of the explanations are based on theoretical formal models.

4 At present, in the New York mercantile exchange (NYMEX), the world’s

largest futures exchange, the most distant futures contract that can be traded is for

December 2016.

Possibly, the most influential of authors that defend this point ofview are Litzenberger and Rabinowitz (1995). They computestatistics based on real price data and conclude that oil futuresprices are in backwardation most of the time, and try to explainthat using a two-period equilibrium model. Some other authorsexplain this frequent backwardation through more fundamentalreasons based on intuitive market dynamics. Among them, Tabak,an economist from the Brazilian Central Bank, suggests aninteresting explanation (Tabak, 2003). He notes that the mostimportant participants in the trading of long-term futures oilcontracts are producers hedging future production. Those parti-cipants usually ‘‘pre-sell’’ part of their future production (toensure, at present, the prices of that fraction of the expectedproduction). To do that, they sell futures contracts for the desiredfuture time frames and doing that they create a downwardspressure over the prices of those futures contracts. That forcesthose hedgers to pay a premium to induce the investors andspeculators to take the opposite side in those trades (that is, tobuy the corresponding futures contracts). Thus, the futures pricesneed to be below the (discounted) value of the present spot prices,to give the speculators an expected profit—high enough to inducethem to assume the risk that the producing companies want totransfer to others.

Some authors defend that, for crude oil, the price of futurescontracts is a poor predictor for future spot prices. However, thereare several studies that show that futures prices can be used topredict the spot prices (e.g., Fama and French, 1987; Wu andMcCallum, 2005; Chinn et al., 2005). This reliance on presentfutures prices as predictors for future spot prices is based in thebelief that the mechanism of price formation in the futuresmarkets is formed by the interaction of a large number of marketparticipants that base their decisions on informed analysis of thepresent (and future) trends of supply and demand. Also, someauthors (e.g., Hamilton, 2005) point out that the main objective ofthe futures markets is to allow trading based of the differencebetween individual opinions and the actual market price of thefutures contracts (in itself, a consensus value formed from theaggregated opinions as to where traders expect spot oil prices togo in the future). We also believe that the prices in the realcommodities futures markets are mostly formed by the partici-pants’ beliefs about the future balance between offer and demand,and that direct accounting projections about future productioncosts, or about the cost of keeping stocks, seem to be much lessrelevant for actual futures prices formation. This point of view isreinforced by the market behavior for the nearest futures dates inface of speculative news stories (e.g., terrorist threats, powerstruggles in important oil producers, etc.) with no direct impacton production or other direct costs. Even speculative stories,eventually unfounded, sometimes originate important pricefluctuations, because they change the market participants’expectations. This obvious characteristic of the futures marketsis not usually incorporated in the formal theories about the pricesin these markets. That happens mainly because this kind ofmechanism for price formation, based on the aggregation ofinformal beliefs, is difficult to quantify and to incorporate inmathematical models.

3.2. Futures prices as a predictor for peak oil

As do some of the authors referenced above, we believe thatthe behavior of the oil futures markets is mainly governed bypredictions about offer/demand balances and that the price curvereflects the true expectations of the most important marketparticipants: oil producers (mostly large international companiesand exporting countries), oil consumers, and also specialized

ARTICLE IN PRESS

Fig. 3. March 2005 futures prices curve, in USD/barrel (from the NYMEX website: http://www.nymex.com). The first contract available for this curve is for April 2005 and

the last is for December 2011.

P. de Almeida, P.D. Silva / Energy Policy 37 (2009) 1267–12761272

investors and speculators.5 Those investors are sophisticated andwell informed and, although they sometimes overreact to short-term news, they are in a very good position to build stableestimates about the future behavior of prices. In fact, if they didnot believe that they are able to predict the future trend for oilprices, most of them would not have a reason to participate inthose markets. At most, these investors can be accused of beingtoo prudent and rely too much on the ‘‘common sense’’ lessons ofhistory, thus tending to be slow to accept fracturing new realitieslike a near-term PO. This way, the aggregate of the marketparticipants tends to behave very conservatively6 for the longertime frames and tends to believe that the future will repeat thepast: given time, both the inordinately high and the inordinatelylow crude oil prices that will tend to correct to ‘‘normal levels’’.This behavior is illustrated by the fact that distant crude futuresprices show a very strong tendency to lag the most violentmovements of the near-term futures prices, usually being abovethem when they are very low and being below when they are veryhigh (e.g., see Haubrich et al., 2004).

During the price increase of the last few years, the POexplanation was disregarded by most of those prudent andconservative market participants, because it seemed an unneces-sary ‘‘out of the box’’ theory, defended only by a fringe of oilindustry outcasts. This ‘‘conventional’’ belief was reinforced byerroneous previous announcements of ‘‘the end of oil’’ made bybadly informed individuals during previous price increases and byoptimistic predictions presented by organizations like EIA, IEA, orCERA, which are still reluctant to recognize a near-term PO.

Given this background, it should indeed be expected that theconservative futures market participants would recognize theradical idea of a near-term PO only through a very slowadaptation. We have defended this point of view for severalyears, and have been predicting that the prices of the most distantfutures contracts would have to rise to conform with theprogressive acceptance of the reality of a near-term PO.

In 2004, when we started to analyze the curve of the futuresprices, we detected a very strong backwardation. At that time, thelonger term future contracts available, for December 2010, werepriced well below 30 USD dollars and the near-term futures prices

5 It should be noted that at least until very recently (with the appearance of

the USO exchange traded fund, and with the launching of electronic trading for the

NYMEX long-term CL contracts) the long-term crude futures and options were not

easily available for trading by the general public, and in any case until the price

rises of the last few years, the general individual investor did not feel an appeal to

enter that kind of investment.6 This is also explained by the aversion to risk that should be expected from

companies that use the futures markets not to speculate, but to hedge future

production/consumption. Until the recent ‘‘explosion’’ in the ‘‘hedge funds’’

industry, those crude commercial operators clearly dominated the markets for

distant crude oil futures.

almost doubled this value. This futures prices curve was clearly atodds with an expected PO occurrence in the vicinity of 2010, asdefended at the time by some of the leading oil geology experts(e.g., Campbell and Laherrere, 1998). Our evaluation agreed withthis time frame for PO, and so we considered that the most distantfutures contracts available at the time did not reflect the growingshortfall of production that people well informed about the POproblem could already expect to arise by the end of the presentdecade.

With these facts in mind, in the beginning of 2005, we wrote ashort paper arguing that the rise of the pricesof the long-termfutures crude oil contracts (that, by then, had already started)would have to go on much further to incorporate the PO reality(Almeida and Silva, 2005). At that moment, the strong back-

wardation still present in the futures prices curve remained a clearsymptom of market inefficiency, and we argued that thisinefficiency was due to slow recognition, on the part of themarket participants, of the information about future productionconstraints that was already available. Following our previouswork, we studied the historic evolution of the futures pricescurves from March 2005, until October 2008 and discovered that,in this period, these curves evolved from the previous very strongbackwardation to a new situation in which they exhibit asignificant contango. Considering the traditional tendency of thecrude oil futures prices curve to be in backwardation, the presentsituation suggests that the participants in the futures marketshave improved their understanding of the PO problem, and are nolonger expecting the crude oil prices to return to the much lowerhistorical average levels. For this analysis, we used a historicalfutures prices curve from Almeida and Silva (2005) and severalsimilar more recent prices curves spanning the following years,obtained from the NYMEX website. These data are shown ingraphical form in Figs. 3–7.

The curve in Fig. 3, from March 2005, shows that a future peakof production is not contemplated by the market participants,since the price of futures drops continuously almost from the firstmonth and finish significantly below the first prices, in December2011.

The curve on Fig. 4, dating from November 2006, starts with astrong price rise up to two years ahead. From that point on, thecurve drops clearly, but it still ends (some six years from thebeginning) with a price significantly higher than it had for the firstmonth. The strong contango visible in the left portion of this curvewas mainly due to a recent price decrease in the spot price (from76 USD in the beginning of August to around 59 USD at middle ofNovember, when the curve values were obtained). This fast dropin the spot price and in futures prices for the first few months wasnot followed by the longer term contracts. This is a normalbehavior for the futures prices: the frequent (and sometimes verystrong) short-term oscillations in the prices of the futures

ARTICLE IN PRESS

Fig. 4. November 2006 futures prices curve, in USD/barrel (from the NYMEX website: http://www.nymex.com). The first contract available for this curve is for December

2006 and the last is for December 2012.

Fig. 5. July 2007 futures prices curve, in USD/barrel (from the NYMEX website: http://www.nymex.com). The first contract available for this curve is for August 2007 and

the last is for December 2012.

Fig. 6. January 2008 futures prices curve, in USD/barrel (from the NYMEX website: http://www.nymex.com). The first contract available for this curve is for February 2008

and the last is for December 2016.

P. de Almeida, P.D. Silva / Energy Policy 37 (2009) 1267–1276 1273

contracts for the first few months tend not to be followed by thelonger term futures or, at most, to be followed with a significantdelay and with a reduced amplitude. This behavior, that wenoticed empirically by the observation of the futures curvesduring the last few years, was also already been noted by others(e.g., Haubrich et al., 2004). The evolution of the prices during the20 months that separate the two curves suggests that the marketparticipants (among whom the most relevant are the producersthemselves) have adapted their perspectives to recognize that thecrude oil supply constraints would not be resolved in the next fewyears. Relating this with the concept of an inevitable peak of oilproduction, it seems reasonable to conclude that the aggregate ofthe market participants had started to believe that, in the next fewyears, the oil production curve would, at least, reach a plateau ofincreasingly insufficient production.

Fig. 5 shows a futures price curve from February 2007, onlythree months after the curve shown in Fig. 4. These curves arevery similar and start at almost the same value (around 59 USD).However, the comparative analysis of these two curves, illustratesvery clearly the continuing tendency for a price increase in mostdistant contracts in relation to the prices for shorter time frames.

Fig. 6 presents a prices curve from January 2008. This curve,made 11 months after the curve presented in Fig. 5, starts with asignificant backwardation in the first two years. From thatmoment on, it shows a consistent price increase up to the lastavailable contract—now for December 2016. Again, the initialbackwardation of this curve results from a very strong and regularprice increase in the prior twelve months. This price increase,from a monthly average of 59 USD in February 2007 to 93 USD inJanuary 2008, is well illustrated in Fig. 2. As could be expected, the

ARTICLE IN PRESS

Fig. 7. October 2008 futures prices curve, in USD/barrel (from the NYMEX website: http://www.nymex.com). The first contract available for this curve is for November 2008

and the last is for December 2016.

P. de Almeida, P.D. Silva / Energy Policy 37 (2009) 1267–12761274

very fast increase in the short-term crude oil prices again resultedin a significant time lag in the reaction of prices for longer termcontracts. However, it is very interesting to note that after thisinitial backwardation, the curve starts to rise, ending in 2016 at aprice that is quite close to the short-term prices—in spite of theseprices being at almost record highs.

Finally, Fig. 7 shows a more recent futures price curve, fromearly October 2008. After an initial backwardation during only asingle month (due to acute scarcity of crude on the USA spotmarket resulting from the impact of hurricanes Gustav and Ike),this figure shows a very intense contango until 2010. The strengthof this contango tendency is partially due to the recent drop incrude oil prices (from more than $140, attained three monthsbefore, to less than $90, at the time this curve was registered). Aspreviously explained, a fast change of spot and short-term pricestends to ‘‘leave behind’’ the longer term futures prices, since thoselong-term prices are viewed by the market as less affected byshort-term changes in the supply/demand balance. After 2010, Fig.7 still exhibits a contango tendency. This tendency is not as strongas the one shown up to 2010, but is still quite significant andconsistent up to the most distant contracts.

Taken globally, this curve shows an intense contango that iseven more significant, since the crude oil market tends to presenta common tendency for backwardation. The consistent contango

illustrated in Fig. 7 can only be explained by a change in thebeliefs of the aggregate of the market participants. Today, thesemarket participants have come to recognize that in the future thecrude oil price will tend to be significantly higher, even if thepresent prices are already well above the historical average.

This is an important new reality. The crude oil futures pricescurve is mainly based on the expectations and estimates of themarket participants for the future prices of that commodity.However, it also incorporates a significant component of priceformation that results directly from hedging and arbitrage, andthat component tends to produce a backwardation. Even withoutdiscounting this traditional mechanical backwardation effect fromthe October 2008 prices’ curve, illustrated in Fig. 7, this curvealready reveals the belief in a progressive tightening of the futurecrude oil supply/demand situation. In October 2008, this belief ina supply/demand tightening is occurring in a context of severeworld economic problems that no one expects to be quicklysolved, and that will tend to significantly compress the demandfor oil products. In this context, this unprecedented belief in afuture supply/demand tightening cannot be justified by thedemand side, and can only be due to expected supply constraints.

From the strong contango exhibited up to 2010 by the pricescurve of Fig. 7, it can only be concluded that the marketparticipants are expecting the occurrence of extraordinary oilsupply problems starting at the beginning of the next decade. This

prospect is in line with the supply difficulties that should resultfrom most of the peak production predictions shown in Table 1,and from our own analysis. In fact, those expected supplyproblems are difficult to explain without a PO scenario, sincethey are being predicted several years before their occurrence(which is odd in itself), and since the futures curve does not showany sign that the market expects an ulterior solution for thoseproblems. This can be argued to represent an indirect recognition,by the aggregate of the market participants, of a near-term POscenario. This recognition and the most recent futures pricescurve correspond to a new situation in which the importantmarket inefficiency of the last few years, signaled in Almeida andSilva (2005), has all but disappeared.

The crude oil market participants are usually well informedbut tend to be very conservative and very hard to convert toradical new realities, like an emergent near-term PO. However, ascould be expected, they ended up accepting the many pieces ofevidence that point to the occurrence of a peak of oil production inrelatively short-time frame. We believe that this slow andreluctant evolution of the point of view of the crude oil marketparticipants can be construed as an alternative validation of thisnew reality: a near-term PO.

4. Conclusions

The general idea of a future peak of oil production (PO) is nowwell accepted. The time frame for that peak, however, is still muchcontested, and there are multiple, and very different, predictionsfor the date of PO. Nevertheless, most of the relevant publishedresearch on this subject points to a significant probability of thePO occurring prior to 2015. In this paper, we tried to validate thisrelatively early peak using an independent methodology based onthe analysis of the evolution in time of the crude oil futures pricescurve. We believe that this prices curve is mainly based on thetrue expectations of the market participants for the future oilprices. Among these market participants are the oil productioncompanies (including both the independent and the national oilcompanies), the largest fuel oil users (e.g., air transport compa-nies) and specialized funds and individual investors. Almost bydefinition, those traders of long-term futures are very wellinformed about the oil production issues. Even departing fromthe present, relatively high crude oil prices, when compared withthe average of its 150 years history, the aggregate of those marketparticipants still expect a continued strength in future prices. Infact, the price strength expected by those well-informed marketparticipants for the next few years clearly demonstrates aprogressive acceptance of the reality of a near-term PO, which is

ARTICLE IN PRESS

P. de Almeida, P.D. Silva / Energy Policy 37 (2009) 1267–1276 1275

in accordance with most of the predicted dates for PO shown inTable 1.

Although at present, many of the oil producers still maintain apublic denial of near-term production problems, their marketactions, as reflected by the futures price curve, contradicts thatofficial public position. We believe that the recent evolution of theNYMEX futures prices for crude oil means that the marketparticipants already agree with the technical predictions thatpoint to the occurrence of peak oil around the beginning of thenext decade. Additionally, we believe that this, in itself, constitu-tes a further validation for a relatively near-term date for peakproduction.

The known analysis about the effects of the PO and aboutforms of mitigating the ensuing problems (e.g., Hirsch et al., 2005;Bezdek et al., 2006) suggest the necessity of a relatively lengthyadaptation phase during which several efforts must be performedsimultaneously (e.g., increasing energy efficiency, developingalternative short-term transitional energy sources, and developinglong-term sustainable energy solutions). If the early datessuggested by several important analysts and by our ownpredictions (both the bottom-up approach and the futures pricesanalysis) are confirmed, the time available for pre-peak mitigationefforts is already very limited.

In fact, we believe that the mere possibility of those peak datesproving real should be enough to mobilize immediate efforts tomitigate the energy crises that will, almost inevitably, result fromthe peak of the world production of liquid fossil fuels.

References

Almeida, P., Silva, P., 2005. Peak oil and the NYMEX futures market: Do investorsbelieve in physical realities? In: Proceedings of the ASPO IV InternationalWorkshop on Oil and Gas Depletion, May, Lisbon.

ASPO, 2007. The growing gap. Association for the Study of Peak Oil and gas (ASPO),Newsletter 74. /http://www.peakoil.net/S.

Bakhtiari, A., 2004. World oil production capacity model suggests output peak by2006–07. Oil and Gas Journal 102 (16), 18–19.

Bakhtiari, A., 2006. The last frontier. Online article. /www.sfu.ca/�asamsamb/sb.htmS.

Bartlett, A., 2000. An analysis of US and world oil production patterns usingHubbert-style curves. Mathematical Geology 32 (1), 1–17.

Bezdek, R., Wendling, R., Hirsch, R., 2006. Economic impacts of US liquid fuelmitigation options. Report for the US Department of Energy, National EnergyTechnology Laboratory, USA.

Campbell, C., 2003. Industry urged to watch for regular oil production peaks,depletion signals. Oil and Gas Journal 101 (27), 38–45.

Campbell, C., Heapes, S., 2008. An Atlas of Oil and Gas Depletion. Jeremy MillsPublishing, Lindley, Huddersfield, UK.

Campbell, C., Laherrere, J., 1998. The end of cheap oil. Scientific American 278 (3),78–83.

CERA, A., 2008. No Evidence of Precipitous Fall on Horizon for World OilProduction: Global 4.5% Decline Rate Means No Near-Term Peak. CERA/IHSStudy, Cambridge Energy Research Associates, Inc., Cambridge, MA /http://www.cera.com/aspx/cda/public1/news/pressReleases/pressReleaseList.aspxS.

Chinn, M.D., Leblanc, M., Coibion, O., 2005. The Predictive Content of EnergyFutures: An Update on Petroleum, Natural Gas, Heating Oil and Gasoline.National Bureau of Economic Research-Working Paper 11033, Cambridge, MA.

Davis, G., 2003. Meeting future energy needs. The Bridge (National Academy ofEngineering) 33 (2), 16–21.

Deffeyes, K.S., 2001. Hubbert’s peak: The impending world oil shortage. PrincetonUniversity Press, Princeton.

Deffeyes, K.S., 2003. Hubbert’s Peak—The impending world oil shortage. PrincetonUniversity Press, Princeton.

EIA, 2006. International energy outlook 2006. US Department of Energy, EnergyInformation Administration, USA.

Fama, E., French, K., 1987. Commodity futures prices: some evidence on forecastpower, premiums and the theory of storage. Journal of Business 60 (1), 55–73.

French, M.W., 2005. Why and when do spot prices of crude oil revert to futuresprice levels? US Federal Reserve Board /http://www.federalreserve.gov/S,Finance and Economics Discussion Series-FEDS Working Paper 2005-30, USA.

Goodstein, D., 2004. Out of gas—the end of the age of oil. W.W. Norton, New York.Hallock, J., Tharakan, P., Hall, C., Jefferson, M., Wu, W., 2004. Forecasting the limits

to the availability and diversity of global conventional oil supply. Energy 29,1673–1696.

Hamilton, J.D., 2005. Oil futures and the future of oil. Econbrowser, June 09./www.econbrowser.comS.

Haubrich, J.G., Higgins, P., Miller, J., 2004. Oil prices: Backward to the future?Federal Reserve Bank of Cleveland, Economic Commentary, December, Cleve-land, OH.

Hirsch, R., Bezdek, R., Wendling, R., 2005. Peaking of world oil production: Impacts,mitigation and risk management. Report for the US Department of Energy,National Energy Technology Laboratory, USA.

Hotelling, H., 1931. The Economics of Exhaustible Resources. Journal of PoliticalEconomy 39 (2), 137–175.

Hubbert, M.K., 1949. Energy from fossil fuels. Science 109 (2823), 103–109.Hubbert, M.K., 1956. Nuclear energy and fossil fuels. Publication Number 95, Shell

Development Company, Houston, Texas, 1956. Presented on the SpringMeeting on Drilling and Production Practice, March 7–9, American PetroleumInstitute, San Antonio, Texas.

Hubbert, MK., 1971. The energy resources on the Earth. In: Energy andPower, a Scientific American Book. W. H. Freeman & Co., San Francisco,pp. 31–40.

IEA, 2000. World energy outlook 2000. International Energy Agency, Paris.IEA, 2006. World energy outlook 2006. International Energy Agency, Paris.IEA, 2007. 2005 Energy Balances for World. International Energy Agency, Paris.Jackson, P., 2006. Why the Peak Oil Theory Falls Down—Myths, Legends, and the

Future of Oil. CERA, Client Services Cambridge, MA, USA.Jackson, P., Eastwood, K., 2007. Finding the critical numbers: what are the real

decline rates for global oil production? CERA, Private Report–Client Services,Cambridge, MA, USA.

Jackson, P., Esser, R., 2004. Triple witching hour for oil arrives early in 2004—but,as yet, no real witches. CERA, Client Services, April 7, Cambridge, MA, USA.

Jiang, X., Han, X., Cui, Z., 2007. New technology for the comprehensive utilization ofChinese oil shale resources. Energy 32, 772–777.

Jum’ah, A., 2007. The Road to Our Energy Future: A Candid Assessment. A Future ofEnergy Lecture, Harvard University, Center for the Environment, January 31,Cambridge, USA.

Koppelaar, R., 2005. World oil production and Peaking outlook. Peak Oil Nether-lands Foundation/ASPO Netherlands. /http://www.peakoil.nlS.

Koppelaar, R., 2006. Oil supply analysis 2006–2007. Peak Oil NetherlandsFoundation/ASPO Netherlands, Newsletter 5. /http://www.peakoil.nl/wp-content/uploads/2006/10/asponl_newsletter_5_2006.pdfS.

Koppelaar, R., 2007. The shape of oil to come. The Oil Drum /www.theoildrum.comS, October 12. /http://www.theoildrum.com/node/3060S.

Krane, J., 2007. Texas oilman Pickens says global oil production at its peak.Associated Press Interview, March 1. /http://www.iht.com/articles/ap/2007/03/01/africa/ME-FIN-Pickens-Oil-Peak.phpS.

Laherrere, J., 2003a. Will the natural gas supply meet the demand in NorthAmerica? International Journal of Global Energy Issues 19 (1), 1–62.

Laherrere, J., 2003b. Future of oil supplies. Seminar at the Swiss Institute oftechnology, Center of Energy Conversion, May 7, Zurich. /http://www.oilcrisis.com/laherrere/zurich.pdfS.

Laherrere, J., 2007. Limits to Growth Updated. Association for the Study of Peak Oiland Gas—Ireland, July. /http://www.aspo-ireland.org/index.cfm?page=speakerArticles&rbId=9S.

Litzenberger, R., Rabinowitz, N., 1995. Backwardation in oil futures markets: theoryand empirical evidence. Journal of Finance 50 (5), 1517–1545.

Lynch, M.C., 2003. Petroleum resources pessimism debunked in Hubbert modeland Hubbert modelers’ assessment. Oil and Gas Journal 101 (27), 38–48.

Mawdsley, J., 2006. When energy demand exceeds supply: the marketplaceperspective. In: Proceedings of the Symposium at the University of Winnipeg,April 19, Centre for Sustainable Transportation and Institute of Urban Studies,Winnipeg, USA. /http://cst.uwinnipeg.ca/documents/oil_symposium/mawdsley.pdfS.

Mouawad, J., 2007. Oil innovations pump new life into old wells. New York Times,March 5.

Nemesis., 2002. The impact of falling discovery on peak production. ASPO-ODACNewsletter 15. /http://www.peakoil.net/IWOOD2002/Newsletter/Newsletter15.docS.

Robelius, F., 2007. Giant Oil Fields—The Highway to Oil: Giant Oil Fields and theirImportance for Future Oil Production. Doctoral thesis, Uppsala University,Sweden.

Shell., 2008. Shell energy scenarios to 2050. Royal Dutch Shell plc, Netherland./www.shell.com/scenariosS.

Simmons, M., 2003. Is the Glass Half Full or Half Empty? In: Aleklett, K., Campbell,C., Meyer, J., (Eds.), Proceedings of the Second International Workshop on OilDepletion, May 26–27, Paris.

Simmons, M., 2007. Is the world supply of oil and gas peaking? InternationalPetroleum Week 2007, February 13, London.

Skrebowski, C., 2004. Oil field mega projects 2004. Petroleum Review January18–20.

Skrebowski, C., 2006. Prices holding steady, despite massive-planned capacityadditions. Petroleum Review April 28–31.

Skrebowski, C., 2007. New capacity fails to boost 2006 production—delays ordepletion? Petroleum Review February 41–45.

Smith, M.R., 2002. Analysis of global oil supply to 2050. Consultancy Report fromThe Energy Network/Energyfiles. /http://www.energyfiles.com/S.

Smith, M.R., 2006. Modeling the oil supply gap. Energy Institute, London(Presentation in the conference Oil depletion—Dealing with the issues,November).

Stuart, S., 2007. Water in the Gas Tank. The Oil Drum /www.theoildrum.comS,March 26. /http://www.theoildrum.com/node/2393S.

ARTICLE IN PRESS

P. de Almeida, P.D. Silva / Energy Policy 37 (2009) 1267–12761276

Tabak, B.M., 2003. On the information content of oil future prices. Banco Central doBrasil /http://www.bcb.gov.br/S, Working Paper Series 65, Brasil.

Tillerson, R., 2007. The State of the Energy Industry: Strengths, Realities andSolutions. CERA Week 2007 Conference, Houston, Texas.

WEC., 2003. Drivers of the energy scene. World Energy Council, London. /http://www.worldenergy.orgS.

Wood, J., Long, G., 2000. Long-term world oil supply (a resource base/productionpath analysis). Report from the US Department of Energy, Energy InformationAdministration, presented in the Meeting of the American Association ofPetroleum Geologists, April 18, New Orleans, Louisiana. /http://tonto.eia.doe.gov/FTPROOT/presentations/long_term_supply/index.htmS.

Wood, J., Long, G., Morehouse, D., 2004. Long-Term World Oil Supply Scenar-ios—The Future is neither as Bleak or Rosy as some assert. Feature article, USDepartment of Energy, Energy Information Administration, USA.

Wu, T., McCallum, A., 2005. Do oil futures prices help predict future oil prices?Federal Reserve Bank of San Francisco, Economic Letter 2005–38, SanFrancisco, CA.

Zittel, W., Schindler, J., 2007. Coal: resources and future production. Energy WatchGroup, EWG-Series No 1/2007. /http://www.energywatchgroup.org/fileadmin/global/pdf/EWG_Report_Coal_10-07-2007ms.pdfS.