Energy Policy 40 (2012) 147–158

Contents lists available at SciVerse ScienceDirect

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

Energy return on (energy) invested (EROI), oil prices, and energy transitions
Matthew Kuperus Heun a,n, Martin de Wit 1,b
a
Calvin College, Engineering Department, 3201 Burton St. SE, Grand Rapids, MI 49546, USA
b
Stellenbosch University, School of Public Leadership, P.O. Box 610, Bellville 7535, South Africa

a r t i c l e i n f o abstract

Article history: Very little work has been done so far to model, test, and understand the relationship between oil prices
Received 17 January 2011 and EROI over time. This paper investigates whether a declining EROI is associated with an increasing
Accepted 5 September 2011 oil price and speculates on the implications of these results on oil policy. A model of the relationship
Available online 5 November 2011
between EROI and oil market prices was developed using basic economic and physical assumptions and
Keywords: non-linear least-squares regression models to correlate oil production price with EROI using available
EROI data from 1954–1996. The model accurately reflects historical oil prices (1954–1996), and it correlates
Oil prices well with historical oil prices (1997–2010) if a linear extrapolation of EROI decline is assumed. As EROI
Energy transitions declines below 10, highly non-linear oil price movements are observed. Increasing physical oil scarcity
is already providing market signals that would stimulate a transition away from oil toward alternative
energy sources. But, price signals of physical oil scarcity are not sufficient to guarantee smooth
transitions to alternative fuel sources, especially when there is insufficient oil extraction technology
development, a declining mark-up ratio, a non-linear EROI–cost of production relationship, and a non-
linear EROI–price relationship.
& 2011 Elsevier Ltd. All rights reserved.

1. Introduction technology being equal, with a increasing marginal costs of

The economical and sustainable provision of energy to run
modern economies and meet human development goals (UN, 2000)
is one of the Grand Challenges facing the world today (NAE, 2008).
However, an emerging consensus indicates that physical scarcity of
energy resource supplies is, or soon will be, upon us, in the absence of
a concerted effort to develop and implement fossil fuel substitutes
(Brandt, 2007; Hubbert, 1956; Nashawi et al., 2010; Sorrell et al.,
2010). Friedrichs (2010) reviewed case studies of nation-states that
experienced energy supply shortages and noted many possible
responses, including predatory militarism, totalitarian retrenchment,
and socioeconomic adaptation. The energy supply shortages that
trigger such responses deserve closer scrutiny.
One proposed physical indicator of energy supply scarcity is
energy return on (energy) invested (EROI), which, like its well-
known financial counterpart return on investment (ROI) is a ratio
of outputs to inputs (Cleveland et al., 1984). We define EROI as the
ratio of gross energy delivered by an energy production process to
input energy required to obtain that gross energy. Basic economic
theory leads to the expectation that a declining EROI may be
associated, all other things such as costs of capital, labour, and
n
Corresponding author. Tel.:þ 1 616 526 6663.
E-mail addresses: heunm@calvin.edu (M.K. Heun),
martin@sustainableoptions.co.za (M. de Wit).
1
Tel.:þ 27 21 982 7862.
production and, ultimately, an increasing price at which the
commodity (energy) is traded.
Despite some very useful work on EROI (as discussed below),
investigations of the interactions between physical indicators
(such as EROI) and economic indicators (such as oil prices) are
in short supply. Furthermore, very little has been done to
determine if price signals of physical scarcity will be sufficient
to cause transitions to alternative fuel sources. Finally, the speed
and orderliness of such transitions is of the utmost importance
for policy makers, yet there is little work available on the topic
of EROI and energy transitions. The questions addressed in this
paper are as follows:
 ‘‘How is EROI related to energy prices?’’
 ‘‘What implications do EROI trends over time have for eco-
nomic and energy policy?’’ and
 ‘‘What is required to ensure a smooth transition away from oil
toward substitutes?’’
After reviewing the available literature on these topics,
we analyse the interactions between EROI and oil pricing in the
markets, through a combination of historical data analysis,
mathematical modelling, and statistical analysis. We then com-
pare the models with recent market price data to show good
correlation. We end by indicating policy implications and dis-
cussing transitions.

0301-4215/$ - see front matter & 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.enpol.2011.09.008
148 M.K. Heun, M. de Wit / Energy Policy 40 (2012) 147–158
2. Literature review
We begin with the relationship between physical scarcity of
energy resources and market prices of energy commodities, as
documented in the literature. We focus on oil, because it is most
often cited as facing physical scarcity in the coming decade. The
resource economics literature emphasises the important roles of
supply, demand, and prices in the allocation of resources (Sweeney,
1993). Every market has different factors that affect supply and
demand, but generally it is expected that increasing physical
scarcity is associated with rising prices, all other things being equal
(Sweeney, 1993). This means that current prices play an important
role in the evaluation of the future of the energy market. The
EROI literature, however, doubts this claim and argues for the use
of a separate physical indicator in the evaluation of the energy
market future. The rationale in the EROI literature for the use of
EROI as an indicator for physical scarcity is a deep suspicion that
market signals and cost-benefit analyses based on current prices
are insufficient to guide decisions about future energy markets
(Hall et al., 2009).
In the next section we briefly review the literature on the oil
market, focused on the forces of demand and supply, and
the interactions between them. We continue with an overview
of the problems of these markets and review the arguments for
the development of EROI as a physical indicator of the contest
between scarcity and technology. We conclude that the relation-
ship between prices and EROI needs to be empirically tested as
this has not been done to date.
2.1. The oil market and signals of physical scarcity
In the oil market specifically, global oil production rates were
stable in the 2000s and failed to grow as demand increased and
prices rose during the period 2005–2007 (Hamilton, 2009). This
combination of increasing demand and flat production led to oil
price increases in recent years (Hamilton, 2009). When price
increases are the result of rising demand rates in the context of
stagnant production rates, the relationship between physical
scarcity and the rising marginal cost of production may be an
important driver. This warrants further research into the perma-
nence and expected impact of this scarcity over time.
Before turning to the literature on EROI addressing this issue of
physical scarcity, the question remains whether oil prices are a
reliable indicator of what is happening in oil markets. In the
market for gasoline and crude oil, for example, price is strongly
influenced by both supply and demand factors. The present value
of depletable resources such as oil is based on price expectations
going forward. Price expectations are based on historical and spot
prices. Efficient markets may be a sufficient guide for action, but
when no or only patchy information is available and expected
prices do not reflect scarcity (user costs), expected extraction
costs, and search costs, prices are not a sufficient guide for making
decisions. Energy markets today exhibit significant distortions.
Victor (2009), for example, points out that global fossil fuel
subsidies amount to US$500 billion annually thereby stimulating
over-consumption, undermining energy security, and worsening
environmental impacts. Such energy market distortions increase
the risk of sudden shocks to the market as new, undiscounted
information becomes available.
2.2. EROI as a physical indicator of scarcity
Because it takes energy (usually in the form of diesel fuel and
electricity) to make energy available for consumption (as a liquid
fuel or electricity), the relationship between energy input and
energy made available for use is an important indicator. EROI is
defined for an energy production process such as a well or a mine
as the ratio of gross energy produced by the process to energy
required to run the process. EROI functions as a proxy for the
contest between the depletion of fossil fuel energy sources
and the development of technology for fossil fuel extraction
(Hall et al., 2009) because the ‘‘easiest’’ to reach (in an energy
sense) oil and coal deposits are the first to be exploited with
minimal energy input. As wells and mines go offshore and deeper,
it takes increasingly more input energy to make the same amount
of energy available to society, and EROI declines.
Hall et al. (2009) contend that today’s market prices do not
adequately discount geological (depletion) or political factors, are
greatly influenced by various subsidies, and do not adequately
account for externalities. Hence, to Hall et al., markets cannot be
trusted as an indicator for decision-making. The presumption is
that additional decline of EROI will have large economic effects in
the future (Hall et al., 2008), especially with looming peak oil
dynamics. The important question for this paper is whether the
relationship between EROI and market signals, notably prices, has
been modelled and tested. This question is especially relevant in
the aftermath of the meteoric rise and subsequent fall of oil prices
in recent years. King and Hall (2011) started addressing this
question by analysing the financial and energy return on invest-
ment of energy businesses. Murphy and Hall (2010) reviewed
empirical work in the field of EROI during the past few years, but
none had a focus on the possible relationship between EROI and
market prices. Murphy and Hall speculate on the unlikely pro-
spect of indefinite economic growth in the face of declining EROIs
and fossil energy supplies and question the impact of current
debt-based fiscal stimulus by governments around the world.
Such speculation places further emphasis on the critical impor-
tance of understanding the role of market signals for future
energy availability and use.
There are very few estimates of global or country-specific EROI
values. One of the first estimates of EROI for a country was
performed by Cleveland et al. (1984). Cleveland (2005) later
estimated the EROI for oil and gas in the United States during
the period 1954–1997 and found that a U.S. EROI peak occurred at
approximately EROI¼25 in about 1970. Gagnon et al. (2009)
estimated a worldwide EROI for oil and gas production during
the period 1992–2006 and found that a worldwide EROI peak
occurred in 1999 at approximately EROI¼32. Although EROI
cannot be measured easily and little data is available, there is
some consensus that EROI for oil and gas has declined substantially
from the 1930s up to now (Cleveland, 2005; Gagnon et al., 2009).
Hall et al. (2009) estimate that an EROI of 3 or greater is required at
the well-head or mine mouth to provide fossil fuels to society,
because delivery and use of the fuels consume substantial energy.
A similar but different way to assess the depletion-technology
contest comes from a related area of research that focuses on the
concept of ‘‘Emergy’’ (Odum, 1996). Emergy is a contraction of the
phrase ‘‘embodied energy’’, and its use as a physical concept appears
in a wide variety of contexts including ecosystems, material flows,
recycling, thermodynamics and policy (Center for Environmental
Policy, 2010). In contrast to EROI (which typically focuses on
human-produced energy), emergy is quantified by equivalent solar
energy. One metric for analysing an energy production system is
emergy yield ratio (EYR), which, for an energy production process, is
an energy-quality-corrected ratio of energy output to energy input.
Brown (2009) shows that EYR has been decreasing for the U.S.
economy from 1949 to the present and the EYR for U.S. oil
production has decreased from 17 in 1953 to 6 in 2000. Both areas
of research are saying the same thing: over time it is costing more
energy to get energy. For this paper, we use EROI instead of emergy,
because it focuses on human-chosen energy rather than natural
embodied energy.
 M.K. Heun, M. de Wit / Energy Policy 40 (2012) 147–158 149

2.3. The economics of physical scarcity deposits and technological progress, and the importance of
transitions in energy systems.
The economic implications of physical scarcity of natural We have found no study that empirically tests the relationship
resources are not a new area of research. Classical economists between EROI and price, and there is only one other paper
have focused on the value of productive resources, most notably (King and Hall, 2011) that attempts to develop a theoretical
land and labour. More recently ecological economists, and most relationship between EROI and price. We aim to contribute to
notably biophysical economists, have focused on the primary the literature by proposing a physically based model of the
importance of energy, materials, and ecosystem services to func- interaction between physical scarcity and market prices, with a
tioning economies (Cleveland and Costanza, 2008). The starting focus on the behaviour of EROI and oil prices over time. Results
point of this approach is that so-called objective natural laws from the model can be used to stimulate further discussion of
(later interpreted as thermodynamic laws) operate independent of future energy options, possible energy transitions, and policy
human free will and that human actions should conform to these implications.
physical realities for the good of humankind (Cleveland, 1999).
This interpretation of reality underlies the development of and
reliance upon physical indicators such as EROI instead of economic 3. A model of the relationship between EROI and price
indicators.
For the purpose of this paper it is important to establish In this section, we clearly define EROI from a mathematical
whether there is any empirical evidence on the relationship and thermodynamic point of view. Next, we develop a model of
between physical indicators (EROI) and the outcome of decisions the relationship between EROI and price. Then, we use the model
by the buyers and sellers of oil around the world, as measured by and available historical data to assess interactions among the
oil prices (an economic indicator). terms in the model.
In addition to the attention of biophysical economists, the
physical scarcity of depletable resources such as fossil fuels has
also long been an area of study in resource economics (Kneese 3.1. Energy return on investment (EROI)
and Sweeny, 1993; Solow, 1974). In this field of study, according
to the Hotelling (1931) rule, the asset value of non-renewable We define energy return on investment (EROI) to be the ratio
resources increases at the real rate of interest over an optimal of gross energy output (Egross,t) obtained from an energy produc-
resource depletion path. Subsequent empirical work has, how- tion activity, such as drilling for oil, mining coal, or building wind
ever, failed to support this ‘‘rule’’. Krautkraemer and Toman turbines, to energy input (Einput,t) for the energy production
(2003), after reviewing the literature, conclude that apart from process (regardless of the source of the input energy and includ-
finite availability, many other important factors including ing the production machinery’s embodied energy) during a period
exploration, capital investment, and heterogeneous reserve qual- of time (t).
ity are important for the economics of energy depletion and Egross,t
development. In other words, the marginal extraction costs, the EROIt ¼ ð1Þ
Einput,t
marginal direct costs of exploration, and the user costs of reduced
long-term prospects need all be included in explaining how non- The energy terms (Egross,t, Einput,t) are in units of joules or BTUs.
renewable resource prices behave over time. Based on empirical By this definition, the break-even point for energy production
evidence, Krautkraemer and Toman (2003) argue that both dis- occurs when EROI¼ 1.0. Our definition of EROI means that
covery of new deposits and technological progress, which lower 0.0rEROI oN. A process with EROI 41.0 is an energy source;
the costs of extraction and processing, have greatly mitigated the a process with 0.0oEROI o1.0 is an energy sink.
impacts of finite availability on costs and supply of depletable Fig. 1 shows (a) high EROI wells or mines, (b) wells or mines
fossil fuel resources. where EROI is greater than but very close to 1.0, and (c) low EROI
A related point is the importance of energy transitions.
According to resource economic theory, when the marginal
cost of using a traditional fuel type exceeds the marginal costs
of alternative fuel options, a transition from traditional fuels
to alternatives occurs, assuming that the alternatives provide
the same energy services as traditional fuels. The transition is
expected to be smooth in efficient markets, but could be difficult
and disruptive in markets that are failing owing to reasons such
as monopolistic organisations and information deficiencies
(Tietenberg, 1992) and imperfect substitutability. This means that
the focus should not necessarily be on the depletion path of one
non-renewable resource (such as oil) only, but on the timing and
rate of transitions to alternative energy sources as well (Greene
et al., 2006).
In summary, it can be concluded from this review that, since
the early 2000s oil production has not been able to cope with
increased oil demand mainly owing to economic growth and
increased income. This review further points out that the results
from biophysical and resource economics literature differs on the
relative importance attached to depletion and the ability of
markets to discount geological factors over time. Biophysical
economists point to market failures and imminent depletion
while the latest research in resource economics emphasise that Fig. 1. Graphical representation of EROI where line widths figuratively represent
markets include not only depletion, but also discovery of new magnitude of energy flows.
150 M.K. Heun, M. de Wit / Energy Policy 40 (2012) 147–158
wells or mines with energy import required. Line widths repre-
sent (figuratively) the magnitudes of the energy flows.
Considering the well or mine as our system, the system can run
itself (figuratively) when EROI41.0 by investing (Einvest,t) some of its
gross energy output (Egross,t) as input energy (Einput,t) to the energy
production process (see Fig. 1a). When EROI approaches 1.0, the
energy delivered to society (Edelivered,t) becomes vanishingly small
(see Fig. 1b). If the system is not self-sustaining (i.e. EROIo1.0),
energy must be imported (Eimport,t) into the production process from
beyond the well or mine (see Fig. 1c). (Imported energy can come
from a different well or mine.) An energy accounting at the outlet of
the well or mine gives
Edelivered,t ¼ Egross,t Einvest,t ð2Þ
Similarly, an energy accounting at the input to the well or
mine gives
Einput,t ¼ Einvest,t þ Eimport,t ð3Þ
Equating Eqs. (2) and (3) after solving both for Einvest,t gives
Edelivered,t ¼ Egross,t Einput,t þ Eimport,t ð4Þ
The energy delivered to society can be expressed in terms of
EROI (by combining Eqs. (1) and (4))
 
1
Edelivered,t ¼ Egross,t 1 þEimport,t ð5Þ
EROIt
We also note that by dividing each term in Eq. (5) by the
duration of a time period (Dt) and by taking the limit as Dt-0, we
obtain the energy accounting equation in rate form
 
1 1ðEinput,t =Egross,t Þ þðEimport,t =Egross,t Þ
E_ delivered,t ¼ E_ gross,t 1 þ E_ import,t ð6Þ
EROIt
where the E_ t terms are in units of energy per unit time (power) at
time t. For simplicity, we use energy (not power) units for our
derivations, although that restriction is not necessary, because the
derivations could also be accomplished in units of power. From
Eq. (6) it can be seen that as EROIt falls below 1.0, Eimport,t 40 is
required to maintain Edelivered,t 40.
Returning to the definition of EROI to include the energy flows
on either side of the well or mine, we obtain
Egross,t E þ Einvest,t
EROIt ¼ ¼ delivered,t ð7Þ
Einput,t Eimport,t þ Einvest,t
where we observe that as Einvest,t becomes very large relative to
Edelivered,t, EROI approaches 1.0 in the absence of imported energy.
Although unwise from an energy production point of view, a mine
or well that operates with large Eimport,t can have 0.0oEROI o1.0.
We define EROI at the well-head (or mine mouth) although
that restriction is not necessary for the conclusions we draw
herein. We use oil as an example of an energy source in the
following sections, but this analysis could also apply to other non-
renewable fuels such as coal and natural gas.
3.2. Physical–economic model
To develop a model of the interactions between EROI and
energy prices, we move beyond specific wells or mines and
consider the global picture for energy. When considering the
earth as a system, we assume no imported fossil fuel energy, as
the time scale for converting the sun’s energy into fossil fuels is
very long.
To link the price of delivered energy in the economy to the
physical measure of EROI, we developed a model that includes
both production cost and a mark-up. We begin by saying that the
cost of energy (to energy buyers) for the whole world is obtained
by the following equation:
C E,t ¼ P E,t Edelivered,t ð8Þ
where CE,t is the total cost of energy sold in the world in period t
[$/year]; PE,t is the aggregate average market price of energy
produced by all wells and mines in period t [$/GJ]; and Edelivered,t is
the net energy available to the economy delivered by wells and
mines in period t [GJ/year].
Next, we develop a simple model for the relationship between
production costs and market prices by saying that the cost to the
world economy of net energy (CE,t) is equal to the cost to
producers for making gross energy (on a per GJ basis, cE,prod,t)
multiplied by the gross energy produced (Egross,t) multiplied by a
mark-up ratio (mt).
C E,t ¼ mt cE,prod,t Egross,t ð9Þ
where CE,t is the cost to the economy of energy purchasing during
period t [$/year]; mt is the mark-up ratio above production costs
for the selling price of energy in the market in period t [  ];
cE,prod,t is the cost to producers for making one GJ of gross energy
in period t [$/GJ]; and Egross,t is the gross energy produced in
period t [GJ].
Substituting Eq. (8) gives
PE,t Edelivered,t ¼ mt cE,prod,t Egross,t ð10Þ
Next, we substitute Eq. (4) into Eq. (10) to obtain
PE,t ðEgross,t Einput,t þ Eimport,t Þ ¼ mt cE,prod,t Egross,t ð11Þ
Next, we solve for PE,t and rearrange to obtain
mt cE,prod,t
PE,t ¼ ð12Þ
Substituting Eq. (7) into the above equation we obtain
mt cE,prod,t
PE,t ¼ ð13Þ
1ð1=EROIÞ þ ðEimport,t =Egross,t Þ
If we now restrict ourselves to wells and mines where
EROI41.0 and Eimport,t ¼0.0, which is consistent with both the
global situation today and the purposes of this paper (to assess
what happens when global EROI declines from its present status
of EROI41.0), we find that
mt cE,prod,t
PE,t ¼ ð14Þ
1ð1=EROIÞ
Eq. (14) allows us to assess the relationships among market
price (PE,t), EROI, mark-up (mt), and production costs (cE,prod,t). It
also allows us to assess what would happen if only one or two
variables change over time, although none of these terms is
independent, as discussed in the next section.
3.3. Interactions
There are certain to be interactions among PE,t, mt, cE,prod,t, and
EROI as time progresses. Where we address these interactions, we
will identify them as Ix, where x is in the range 1–4. At this point,
we simply identify and describe the interactions. The following
section assesses the interactions using historical data.
One interaction (I1) may appear when drilling and mining
technology improvements decrease energy input to the energy
sector, thereby increasing EROI and decreasing cE,prod,t leading to a
price decrease (Fig. 2).
A second interaction (I2) may occur when fossil fuel resources
deplete relative to demand. Increasing energy input rates are
required to extract increasingly marginal resources, raising pro-
duction costs, and decreasing EROI. It is interesting to speculate
that producers may be able to charge higher mark-ups (mt) in
such a climate. The net effect is that market prices may rise as a
 M.K. Heun, M. de Wit / Energy Policy 40 (2012) 147–158
Fig. 2. Graphical representation of interaction 1 (I1).
Fig. 3. Graphical representation of interaction 2 (I2).
Fig. 4. Graphical representation of interaction 3 (I3).
result of physical depletion effects, although price increases could
be mitigated by market competition or price controls (Fig. 3).
Interactions 1 and 2 describe the contest between technology
and depletion identified by Hall et al. (2009) who argued that
observed declining EROI in the oil sector indicates that depletion is
having a greater effect than technology improvements at the
present time.
A third interaction (I3) is observed by reversing the implied
causality in I1 and I2. Using oil price as the starting point, a price
increase can lead to lower EROI if (a) high prices lead to increased
mark-up and provide capital for more exploration, thereby
increasing energy input to the production process (increasing
the denominator of Eq. (1)); and (b) physical scarcity limits
energy production (the numerator in Eq. (1)). (The I3a row in
Fig. 4 represents these effects.) A decrease to ‘‘normal’’ prices may
lead to an increase in EROI by a reverse process (I3b).
A fourth interaction (I4) is posited to occur if the value share of
energy expenditures in the economy rises too high. Under those
conditions, disposable income becomes unavailable for purchasing
goods that stimulate economic growth (such as cars, computers,
and homes) thereby providing recessionary pressure, decreased
energy demand, and lower energy prices. Reduced energy prices
and possible government stimulus can lead to economic recovery.
With each subsequent trip around this cycle, an ever-declining oil
supply rate leads to deepening oil scarcity. The fourth interaction
(I4) supplies a mechanism by which an upper bound on energy
prices can be obtained in the long run. The vicious economic cycle
that represents this fourth interaction (I4) is illustrated in Fig. 5.
In sum, I1  I3 suggest that EROI (a physical indicator) may be
highly correlated with movements in market prices, mark-up, and
production costs (economic indicators). And, the model presented
in Eq. (14) allows us to assess I1–I3 by using time series data for
EROI, market prices, and producer prices. In the next section,
we will correlate producer prices with EROI directly, evaluate
long-term trends of mark-up, and simplify Eq. (14) to relate
market prices and EROI directly. Then, we will use Eq. (14) to
speculate on future market price movements under various future
scenarios for EROI trends. The fourth interaction (I4) cannot be
tested by the proposed model: doing so will require an integrated
model of national and/or world economies and their energy
sectors.
151
Fig. 5. The vicious cycle of relative energy scarcity (14).
100
95
World, Heat Equivalent (Gagnon, 2009)
90 U.S., Thermal Equivalent (Cleveland, 2005)
U.S., Divisia (Cleveland, 2005)
40 World (Cleveland, 2005)
35
EROI [-] 30
25
20
15
10
5
0
1930 1960 1970 1980 1990 2000 2010
Fig. 6. All known estimates for oil production EROI.
3.4. Data and assessment of interactions
Data for worldwide oil production EROI is in short supply. Fig. 6
shows all EROI data known to the authors. Note that some data are
for the entire world while other data are for the U.S. only. References
are provided in the legend. As can be seen in Fig. 6, Divisia-corrected
EROI for the U.S. remained approximately constant from the mid-
1950s to early 1970s, declined rapidly in the 1970s and increased
again in the mid-1990s after which it started decreasing again. The
thermal-equivalent EROI for the world oscillated during the 1990s
and decreased rapidly during the 2000s.
Fig. 6 allows us to assess the contest between technology and
depletion (I1 and I2). The available data indicates that EROI for the
world (Gagnon et al., 2009) and the U.S. (Cleveland, 2005) are
falling at the present time. The present rate of technology
improvement appears to be insufficient to put oil production
EROI on a positive slope with respect to time.
The Cleveland (2005) U.S. EROI time series are particularly
relevant to the present study, because they use actual reported
direct energy consumption and estimates of indirect energy
consumption to determine EROI. The Divisia-corrected time series
is different from the thermal time series, because the Divisia-
corrected series accounts for both the economic utility of different
input energy sources (electricity, for example, is more economic-
ally useful than diesel fuel) and the economic utility of petroleum
products (crude oil is more economically useful than natural gas).
The Divisia series increasingly diverges from the thermal series
because, over time, the increasing electricity usage share of
the U.S. oil industry results in higher weighting for energy inputs
152 M.K. Heun, M. de Wit / Energy Policy 40 (2012) 147–158

(the denominator of the EROI calculation) while increasing natural 3

gas production share results in lower weighting for energy outputs
(the numerator of the EROI calculation). For this paper, the Divisia-
corrected EROI series is preferred, because, as Cleveland (2005) says:

Oil Price Markup, mt [-]

2
To the extent that EROI does reflect the scarcity of petroleum in
some meaningful way, then energy quality is an important
consideration. y [N]ot all ‘units’ of energy are equally useful to
society, particularly in regards to their ability to perform specific
tasks in the production of goods and services. A more appropriate 1
indicator is a quality-corrected EROI that reflects the net avail-
ability of energy to actually produce goods and services that
reflect choices people make about how to use energy. y The
quality-corrected EROI is consistently lower than the uncorrected
version. This suggests that in a more meaningful economic sense, 0
1940 1950 1960 1970 1980 1990 2000 2010
petroleum is more scarce than we might otherwise think.
Fig. 9. Time series of U.S. oil price mark-up.

Time series data for oil prices are available from many sources.
Fig. 7 shows inflation-adjusted data for the average U.S. oil price
($/barrel) for the period 1946–2010. These average prices are
based on historical free market (stripper) prices of Illinois Crude.
Producer prices for U.S. domestic oil are available from the U.S.
Energy Information Agency (EIA, 2010) and are shown in Fig. 8.
Dividing the U.S. oil price (see Fig. 7) by the U.S. producer
prices (see Fig. 8) provides an estimate of the oil price mark-up
ratio mt.
120
U.S. Oil Price [2010$/barrel]
Fig. 9 allows us to assess the speculation provided in the
discussion of the second interaction (I2) above: oil producers
might charge a premium for scarce oil resources. Over much of
the period for which data are available, the mark-up ratio has
been relatively constant, except for the spike in the 1970s and the
recent decline since 1990. It does not appear that oil producers in
the U.S. are (currently) in a position to charge a premium for ever-
scarcer oil. In fact, just the opposite appears to be happening: the
mark-up ratio has been trending downward over the last two
decades. A declining mark-up ratio reduces the incentives for oil
producers to continue operations and exploration, and a greater
incentive to cut costs, increase efficiencies, gain market share, and

100
switch to other sources of income. As such, the declining mark-up
ratio may be a driver for future transitions, an effect that is
80
discussed later. For convenience, we will take mt to be constant at
its average value during the period from 1949 to 2009, namely
60
mt ¼ 1:62 ð15Þ
40
Fig. 10 presents data from Figs. 6–9 in an easily comparable
way for further discussion.
20
Upon reaching the peak of domestic production, U.S. oil EROI
started declining in the 1970s. At the same time, and probably
0 driven mostly by geopolitical factors, market prices started rising
1940 1950 1960 1970 1980 1990 2000 2010 after a few decades of oil price stability. From the late 1970s
Fig. 7. Market price for oil. through the early 1980s, high oil prices led to increased oil
Source: Inflation Data (2010). company mark-up that provided the opportunity for increased
exploration and drilling. However, in the wake of the U.S. oil
production rate peak in 1970, increased exploration had very little
90 payoff: the oil production rate stayed nearly constant, despite the
U.S. Oil Producer Price [2010$/barrel]

increased energy input into the oil production sector. So, in the
80
1970s, the U.S. EROI denominator increased while the numerator
70 essentially stayed constant, and EROI declined (Cleveland, 2005).
The 1970s exhibit the dynamics expected from I3a.
60
Interaction 3b (I3b) can be observed in the 1980s. As prices
50 began falling and mark-up returned to former levels, the unpre-
cedented exploration and drilling subsided, and EROI recovered
40 somewhat to a minor peak around 1990.
30 From 1990 onward, I2 is playing out. Given a constant level of
exploration and drilling, the EROI denominator (energy input to
20 the oil production process) will remain essentially constant. But,
10 physical depletion effects mean that less oil is produced for the
energy expended in drilling and exploration. In this scenario, EROI
0 will decline over time, because the EROI numerator (energy
1940 1950 1960 1970 1980 1990 2000 2010 delivered to society) declines. Alternatively, struggling to main-
Fig. 8. Producer prices for U.S. domestic oil. tain a constant production rate as oil extraction increasingly
Source: U.S. EIA (2010). occurs in locations where more energy is required for production
 M.K. Heun, M. de Wit / Energy Policy 40 (2012) 147–158 153

40 World EROI [-]

Gagnon (2009)
30

20
U.S. EROI (Divisia) [-]
10
Cleveland (2005)
0

100
U.S. Oil Price
80 [2010$/barrel]
60
40
20
0
100
80 U.S. Oil Producer Price
[2010$/barrel]
60
40
20
0
3
U.S. Oil Price Markup [-]
2

0
1940 1950 1960 1970 1980 1990 2000 2010

Fig. 10. Summary of U.S. time-series data.

(e.g. deepwater, offshore, or tar sands) means that EROI will 200
decline.
Although EROI was declining in both the 1970s and from 1990 180
onward, there is a significant difference between the dynamics in
PE,t [2010$/barrel]

those two periods. In the 1970s, mark-up increased rapidly with 160
market prices, but since 1990 markup has been decreasing, as
140
market prices rise. Interaction I3a is not happening, i.e., EROI is
not falling because price is rising. Thus, the ‘‘sweet spot’’ in which 120
oil producers found themselves in the 1970s is not being
repeated. The differences between the 1970s and the period from 100
1990 onward must be taken into account when considering policy
options, a topic to which we will return later. 80
mtCE,prod,t = 80.3 2010$/barrel
The denominator of Eq. (14) indicates that energy investment
from output to input will cause prices to rise, because the net 60
0 10 20 30 40 50 60 70 80 90 100
energy delivered to society will decrease as EROI declines. We can
assess what will happen to oil price if energy investment from EROI [-]
output to input is the only factor. Assuming an oil price of $85/ Fig. 11. Price of oil as a function of EROI with mt cE,prod,t ¼ 80.3 2010$/barrel.
barrel (as of March 2010) and an EROI of 18 (Gagnon et al., 2009),
Eq. (14) provides a value for the product of mt cE,prod,t of 80.3
2010$/barrel, the expected price of oil if EROI were infinitely large
in the absence of other effects. Fig. 11 shows the relationship We note that both I1 and I2 (Figs. 2 and 3) indicate that EROI
between price and EROI under those circumstances. and producer prices are indirectly and inversely connected. To
Fig. 11 allows us to investigate the second interaction (I2) investigate the interaction between producer prices and EROI
discussed above. In the absence of other factors, the effect of (I1 and I2), we can plot inflation-adjusted U.S. oil producer prices
increased energy investment (due to depletion effects) on prices (U.S. EIA, 2010) from Fig. 8 against U.S. Divisia-corrected EROI
is nearly inconsequential down to an EROI of approximately 18. In (Cleveland, 2005) from Fig. 6. The data show an inverse correla-
fact, the energy investment effect will not significantly affect tion between producer price and EROI (see Fig. 12). The observed
price until EROI approaches 10. At EROI ¼10, oil price has correlation between EROI and cost of oil production (using U.S.
increased by only 10% compared to prices at very large EROI. As data for 1954–1996) is a weak function of energy investment
EROI declines below 10, however, the effect of energy investment from output to input, because the U.S. EROI is mostly above 10
on price becomes substantial. during this time period (as discussed above).
154 M.K. Heun, M. de Wit / Energy Policy 40 (2012) 147–158

100 90

U.S. Producer Price [2010$/barrel]

90 Producer Prices
80
Producer Price – EROI Correlation
CE,prod,t [2010$/barrel]

80
70
70
60 60
50 50
40
40
30
20 30
10 20
0
0 5 10 15 20 25 30 10
U.S.EROIt [-] 0
1940 1950 1960 1970 1980 1990 2000 2010
Fig. 12. Correlation between producer prices (cE,prod,t) and Divisia-corrected EROI
(Cleveland, 2005) for U.S. domestic oil production for the years 1954–1996. Fig. 13. Comparison of actual producer prices with those generated by a Divisia-
corrected EROI correlation with producer prices (Eq. (16)) and Eq. (14).

Table 1 200
Non-linear least-squares regression models. 180
Model form Model coefficients 160
PE,t [2010$/barrel] 140
a [2010$/barrel] b [] c [2010$/barrel] R2
120
1 cE,prod,t ¼ a EROI  b 751.46 1.43 n/a 0.811
100
2 cE,prod,t ¼ ae  b EROI þc 466.56 0.359 12.71 0.848
3 cE,prod,t ¼ ae  b EROI 92.72 0.117 n/a 0.737 80
4 cE,prod,t ¼ aEROI  b þc 10,660 2.86 10.63 0.849
60
40
Non-linear least-squares regression was performed on the data
20
using the program ‘‘R’’ (R Development Core Team, 2010). Four
correlation model options were evaluated. Results are shown in 0
0 10 20 30 40 50 60 70 80 90 100
Table 1.
All coefficients make contributions to all models with a EROI [-]
confidence level of 99.9%, except for a in model 2 (contribution Fig. 14. Model for oil price as a function of EROI using present-day correlations
with a confidence level of 95%) and a in model 4 (not contributing (from Eq. (17)).
at any confidence level, meaning that one could choose a different
value for a and find new values for b and c that provide essentially
the same quality of fit). For this reason, we reject model 4. Fig. 12 EROI directly:
shows the raw data and the four statistical correlations.
1:624½466:6e0:359 EROIt þ 12:72ð2010$=barrelÞ
The correlation models proposed above can be categorised in PE,t ¼ ð17Þ
1ð1=EROIÞ
two ways: (a) whether the model assumes exponential or power
law dependence and (b) whether the model allows for non-zero Fig. 14 shows that model. It is interesting to note that at
producer price at high EROI. Correlations that contain parameter c EROI¼18 (approximately today’s value), oil price has increased by
(models 2 and 4) allow non-zero producer price at high EROI and only 10% from the value at EROI¼100 owing to EROI effects.
provide a better fit with respect to historical data (higher R2 However, highly non-linear oil price increases are observed when
value). For the remainder of this paper, we use model 2 (the EROI declines below 10.
exponential expression with c) because of its superior R2 value.
Thus,
4. Models of EROI vs. time
  2010$
cE,prod,t ¼ 466:6e0:359 EROIt þ12:72 ð16Þ
barrel
Fig. 14 raises the question of how EROI will evolve with time
We can compare the producer price correlation of Eq. (16) into the future, as that may drive the rate at which EROI-related
against actual producer prices in a time series. In Fig. 13, we note price signals reach the economy. Gagnon et al. (2009) presented
excellent agreement over the period of time that the Cleveland three linear scenarios for worldwide EROI decline with time
(2005) Divisia U.S. EROI data are available (1956–1997). Actual (steep, nominal, and gradual). Table 2 provides the worldwide
producer price data for 1997–2010 are shown for comparison EROI in 1992 (the first year of Gagnon et al.’s 2009 models), the
purposes. slope of the worldwide EROI decay line and the year at which
The data above show that low EROI and high producer prices worldwide EROI will reach 1.0 for the three Gagnon et al. (2009)
are correlated, but the reasons for this correlation are complex. scenarios. (The EROI in 1992 is different for each model, because
It is interesting to consider what would happen if the present Gagnon’s objective was capturing steep, nominal and gradual
trends of mark-up (mt) and producer prices (cE,prod,t) hold as EROI slopes. Also see Fig. 15.)
changes. By substituting Eqs. (15) and (16) into Eq. (14), we Gagnon et al. (2009) suggest that a more gradual exponential
obtain a model for the relationship between oil price and U.S. EROI decay model may better reflect expected trends, but they did
 M.K. Heun, M. de Wit / Energy Policy 40 (2012) 147–158 155

Table 2 Table 3
Linear models of worldwide EROI decay. Exponential models of worldwide EROI decay for steep, nominal, and gradual
Source: Gagnon et al. (2009). initial slopes.

Gagnon et al. (2009) EROI in Constant Year for Model EROI1992 d(EROI)/dt91992 [1/year] t [years]
scenario 1992 slope EROI ¼1.0
[1/year] Steep 40  1.333 30.0
Nominal 32  0.744 43.0
Steep linear 40  1.333 2022 Gradual 29  0.250 116.0
Nominal linear 32  0.744 2034
Gradual linear 29  0.250 2104

25
45
40 20
35
Worldwide EROI [-]

U.S.EROI [-]
30 15

25
10
20
15
5
10
5 0

1930
1940
1950
1960
1970
1980
1990
2000
2010
2020
2030
2040
2050
2060
2070
2080
2090
2100
2110
2120
2130
2140
0
1990
2000
2010
2020
2030
2040
2050
2060
2070
2080
2090
2100
2110
2120
2130
2140

Fig. 16. Linear and exponential U.S. Divisia-corrected EROI trends for decline after
Fig. 15. Six models of worldwide oil EROI decay over time. the 1968 EROI peak.

not develop such a model. A model that (a) has 0.0 as the lower- We now turn to the U.S. Divisia-corrected EROI data from
bound EROI asymptote and (b) matches Gagnon et al.’s (2009) Cleveland (2005), which shows that U.S. EROI for oil reached a
worldwide EROI in 1992 can be constructed with an equation of significant peak around 1970, a local maximum around 1990 and
the form declined from 1990 onward. Following the approach of Gagnon
et al. (2009) to linearly extrapolate U.S. EROI decay as a function
EROI ¼ EROI1992 eðt1992=tÞ ð18Þ of time, we generate the graph (shown in Fig. 16) that shows
linear and exponential decay from the 1968-peak onwards.
where EROI1992 is the EROI in 1992, the first year in the Gagnon
Historical data are shown for context only and we are not trying
et al. (2009) linear decay models; t is the time [years AD]; and t is
to fit the historical data at this point. The aim is to encompass the
an exponential decay constant [years].
range of possible future trends for EROI beyond the date where
To determine t such that the initial decay rate for the
historical data are available.
exponential model matches the slope of the Gagnon et al.
The linear decay model can be expressed as
(2009) linear models on an EROI vs. time graph, we take the time
derivative of EROI from the above equation 0:275
EROIt ¼ 19:5 ðt1968A:D:Þ ð21Þ
year
dðEROIÞ EROI1992 ðt1992=tÞ
¼ e ð19Þ and the exponential decay model can be expressed as
dt t
solve for the time constant t EROIt ¼ 19:5eðt1968 A:D:=70:909 yearsÞ ð22Þ

EROI1992 where t is in years AD.

t¼ ð20Þ
dðEROIÞ=dt91992

and substitute the linear model slope for the time derivative 5. Discussion
d(EROI)/dt91992.
The resulting exponential decay model parameters are shown It is instructive to combine the price vs. EROI model (from
in Table 3. Fig. 14) and the U.S. EROI vs. time models (from Fig. 16) to assess
Fig. 15 shows six models of worldwide oil EROI vs. time. Data the effect of EROI on oil prices going forward, assuming that
points and the linear extrapolations are from Gagnon et al. (2009). existing correlations hold into the future. Fig. 17 shows price vs.
All three exponential models have the same slope in 1992 as time results from the model. Actual oil prices are shown as data
the corresponding linear models from Gagnon et al. (2009). The points. Prior to and including 1996, actual Divisia-corrected U.S.
exponential models decay to an asymptote of EROI¼0.0. At this EROI data (Cleveland, 2005) are used with Eq. (17) to calculate oil
point, there has been very little work done to predict the prices. Agreement with actual oil price data is excellent for this
evolution of EROI into the future, but we believe that these six time period.
models encompass a wide range of possibilities as fossil fuel For 1996 and beyond, linear and exponential EROI extrapola-
extraction becomes increasingly energy intensive owing to deple- tions from Fig. 16 are used with Eq. (17) to project oil prices into
tion effects. the future.
156
200
1996
180
160
PE,t [2010$/barrel]
140
120
100
2008
80
2007
60
40
20
0
1940
1950
1960
1970
1980
1990
2000
Fig. 17. U.S. oil price projections as a function of time.
Fig. 17 shows that the linear model of U.S. EROI decay (from
Fig. 16) exhibits a sharp price spike whereas the exponential
model (from Fig. 16) exhibits a more gradual price increase over
time. Actual oil price trends since 2004 appear to correlate
roughly along the linear model line. The price run-up in the
2003–2008 timeframe is steeper than the linear model suggests.
The 2009 price drop and subsequent 2010 rise oscillate about the
linear EROI decay line.
From the preceding discussion on the correlation between oil
prices and EROI, one might conclude that no policy changes are
necessary, the implicit assumption being that declining oil stocks
and, therefore, decreasing EROI are already providing market
signals to create a transition away from oil toward whatever the
market deems to be an appropriate substitute energy source. In
short, one might say that because of strong market signals of
the contest between depletion and technology, the markets will
provide a transition for us through the pricing mechanism. Unfor-
tunately, traditional equilibrium-focused market theory does not
tell us what that transition will look like, how it will proceed, or
how smooth or abrupt it will be. Our value judgement is that a
smooth transition from oil to substitutes is preferred over an
abrupt transition. That judgement has significant implications for
policy, the subject of the next section.
6. Policy implications
We evaluated the interactions among EROI, mark-up ratio, cost
of production, and oil producer prices. We also evaluated three
interactions among these variables, and this section elaborates on
the policy implications of these interactions.
As discussed above there are at least four risk factors to a
smooth transition away from oil namely (i) insufficient technol-
ogy improvements to counter effects of depletion, (ii) declining
mark-up even when oil prices are rising, (iii) non-linear EROI–cost
of production relationships and, (iv) non-linear EROI–oil price
relationships. Each of these factors has implications for policy as
further discussed below.
6.1. Insufficient technology improvements
In the contest between oil-drilling technological improvement
(I1) and resource depletion (I2) as measured by EROI, depletion
effects started becoming dominant for U.S. oil production as far
back as the early 1990s. (In the 1970s and 1980s, technology and
depletion effects were masked by external effects, I3.) A short-
sighted potential policy implication would be to incentivize
M.K. Heun, M. de Wit / Energy Policy 40 (2012) 147–158
additional research and development in oil extraction technolo-
gies. However, such a policy suggestion is difficult to implement
in a sector with a consistent longer-term decline in mark-up, and
it runs counter to the signals that the markets are providing. Thus,
a preferred longer-term policy approach will be to incentivize
2010 technology development that accelerates the movement away
from oil toward other energy sources.
6.2. Declining mark-up
2009
The second factor to be understood is the impact of the
declining mark-up in U.S. oil industry from 1990 onwards, despite
rapidly rising oil prices in the 2000s. Discussion of the second
interaction (I2) above indicates that declining mark-up provides
less incentive for producers to stay in the oil business: it is an
2010
2020
2030
2040
internal, market-related signal for a transition to more cost-
effective forms of energy. Assuming that existing energy compa-
nies are best positioned to provide energy in an era of decreasing
oil supplies, a key policy implication is that existing energy
companies need financial incentive to diversify to alternative
and renewable energy sources themselves. Whether strong incen-
tives or soft policies are needed to move oil producers into
alternative and renewable energy sources will depend on the
economic viability of such energy alternatives, a point to which
we return in our discussion on the non-perfect substitutability
of oil.
6.3. Non-linear relationship between EROI and cost of production
The third factor important for policy makers is found in the
non-linear relationship between EROI and the costs of oil produc-
tion. The costs of production continued to rise from the end of the
1990s onwards. Producer costs of between $50 and $80 per barrel
as seen in recent years imply an EROI of roughly between 6 and 8,
a low level not seen since the 1970s. The startling difference this
time is that the declining EROI is strongly and non-linearly
associated with an increase in production costs, and not with
increasing mark-up as was the case in the 1970s. Assuming that
mark-up and EROI decline trends seen in U.S. data are playing out
worldwide, the importance for policy makers is that rapidly rising
production costs constrain the financial viability of oil producers,
and policy makers would do well to support steering oil compa-
nies to non-oil energy production activities.
6.4. Non-linear relationship between EROI and oil price
The fourth factor important for policy makers is found in the
non-linear relationship between EROI and the oil market price.
The analysis above indicates that oil price shocks are expected
when EROI falls low enough (e.g. below 10, as shown in Fig. 14),
illustrating that the best fit for the correlation between oil
producer price and EROI is highly non-linear. Analogous to what
Rosser (1999) argues for the whole economy, the oil market
specifically can be stable in certain ‘‘corridors’’ (an EROI larger
than 10, for example), but is ‘‘dysfunctional’’ when exceeding
such corridors (Leijonhufvud, 1981). If the time required for the
oil price to double from $100/barrel to $200/barrel (8 years for
linear EROI decay and 32 years for exponential EROI decay as seen
in Fig. 17) is taken as indicative of the time available for transition
from oil to alternatives before the effects of declining EROI
become economically overwhelming, the time span available to
steer this transition is likely to be shorter than any previous
significant energy transition (Fouquet, 2010), possibly a few
decades at best. With EROI values tending towards values less
than 10, the chance of significant increases in oil prices, all other
factors constant, is very high.
 M.K. Heun, M. de Wit / Energy Policy 40 (2012) 147–158
The policy implication from EROI–price non-linearity is that
more research is required to understand how EROI is likely to
evolve over time and much more work is needed to better
understand the interaction between EROI and the broader econ-
omy in an era of relative oil scarcity. We speculate that oil prices
will not continue to rise ad infinitum as corrective, recessionary,
economic feedback mechanisms may occur (I4). The vicious cycle
does not provide an economic climate that is conducive to market
investment in either enhanced fossil-fuel production technology
or alternative energy technologies. Such interactions need to be
studied further in a broader energy-economy model.
Taken together, the above four factors indicate that one cannot
take a smooth transition from oil to other forms of energy for
granted. In the case of a transition away from oil, existing
alternatives are neither cheaper nor better in the eyes of energy
consumers, nor are they necessarily perfectly substitutable from a
technological point of view, the subject of the penultimate policy
implication section.
6.5. Non-perfect substitutability
There are no perfect and scalable substitutes for oil at the
present time. Alternative energy options on the horizon include
wind, solar, natural gas, and biofuels. Both wind and solar suffer
from intermittency, and neither are substitutes for oil in the
transportation sector without electrification of the vehicle fleet
and an enhanced electrical grid infrastructure that includes energy
storage. Natural gas is a potential near-term substitute for oil, but
oil and natural gas are often extracted together, and their prices are
highly correlated, historically (EIA, 2005). If oil prices are too high
for the economy to withstand, natural gas prices may also be too
high. Furthermore, from a technological point of view, natural gas
shares a drawback with wind and solar: it is not a substitute for oil
in the transportation sector without a vehicle fleet retrofit. Liquid
biofuels could be a near-perfect substitute for oil but they have
very low EROI (Murphy et al., 2010).
This is not only a race between oil exploration technology and
oil depletion, as Gagnon et al. (2009) suggested, but also a race
against time between oil depletion and the deployment of
whatever oil substitutes may arise. Unfortunately, significant
transitions in energy technologies are not fast. Fouquet (2010),
in a study of previous energy transitions, indicates that, histori-
cally, the time-scale for diffusion to dominance is in the order of
three decades for the fastest energy transitions.
The policy implications arising from non-perfect substituta-
bility and the time required for energy transitions are that
(i) significant funding for research on alternatives is necessary
now (Heun and VanderLeest (2008) indicate that approximately
$10 billion/year of funding is required for ‘‘Grand Challenges’’),
(ii) that more research is needed on how to achieve society-wide
energy efficiencies without falling prey to the rebound effect in an
effort to buy some time for a transition, (iii) that the development
of an enhanced electric grid and electrification of the vehicle fleet
may need to be incentivised to allow for the use of electricity as a
transportation fuel, and (iv) that research into maximising the
EROI of oil substitutes needs to be prioritised. Thus, we argue that
some form of steering the transition is required.
6.6. Steering the transition
The emerging field of transition management indicates that
sensitivity to existing dynamics and a regular adjustment of goals is
essential and that structural changes should be attempted in a step-
wise manner (Rotmans et al., 2001). Transition management
was adopted as Dutch policy in 2001 including the energy-supply
sector (Loorbach and Rotmans, 2006). Kern and Smith (2008) note
157
that despite some achievements, such as encouraging longer-term
thinking in energy policy and the energy sector, aspects of
entrenched power and established organisational routines need
to be taken seriously in transitional policies. In the Dutch case, the
dominance of existing energy regime actors impeded progress in
the transitions and broader societal processes are recommended
(Kern and Smith, 2008). Smith et al. (2010) further call for a
broader analytical framework to reassess innovation and techno-
logical change, crucial factors for a successful transition.
Given the structural, economy-wide impacts of energy transi-
tions, the current declining trends in EROI, and the relatively
short time estimated to be available for energy transitions (aspects
demonstrated in this paper), the question of whether some form of
transition management from policy makers is required becomes
more urgent. The Dutch example provides some experience, but
more research is needed to understand how policy makers can
better steer fundamental energy transitions.
7. Conclusion
We conclude that physical scarcity of depletable resources
such as oil is a reality that cannot be ignored because of key
interactions between the depleting resources and the economic
system (see also Mohr, 2010). In this analysis, we identified four
interactions between EROI and prices. With Gagnon et al. (2009),
we noted that the first interaction (I1, which states that drilling
technology could overcome depletion effects) is not occurring,
using evidence from previous research to show that EROI is
declining for both the U.S. and the world at the present time.
We noted when discussing the second interaction (I2, which
states that depletion effects can lead to both declining EROI and
upward pressure on oil prices) that mark-up data shows a
downward trend for U.S. oil producers over the last two decades,
despite rising oil prices. We noted that the second interaction (I2)
becomes significant only when EROI declines beneath a low value
of 10. We noted that a third interaction (I3, which states that
externally driven price increases can drive EROI down, and vice-
versa) was only observed in the 1970s and is not being
repeated today.
We posited a fourth interaction (I4), which states that as the
value share of energy expenditures in the economy rises, dis-
posable income declines, thereby providing recessionary pres-
sures that lead to decreased energy demand and falling energy
prices. The fourth interaction (I4) cannot be tested with the
present model: a full-economy model with a detailed energy
sector is required. We agree with Brandt (2010) that ‘‘future
modelling efforts will only be successful to the extent that they
include both economic (e.g. resource substitution) and geological/
physical (e.g. resource depletion) effects.’’ The analysis presented
herein may provide a starting point for addressing I4.
From these interactions, we highlighted four factors indicating
that a smooth transition away from oil is unlikely: insufficient oil-
sector technological development to overcome depletion, declin-
ing mark-up, non-linear relationship between EROI and costs of
production, and the non-linear relationship between EROI and oil
price. Several policy implications (notably to diversify away from
U.S.-produced oil toward alternative energy sources and to be
ready for depletion-induced oil price shocks) lead to the conclu-
sion that some form of management and support will be required
to achieve a smooth transition away from oil. To date, the best
example of transition management is supplied by the Dutch
experience since 2001. However, significant work remains
to develop coherent policies sufficient to meet the challenge
provided by oil depletion.
158 M.K. Heun, M. de Wit / Energy Policy 40 (2012) 147–158
Acknowledgements
The authors are grateful for the support provided by Josephine
Musango (Stellenbosch University) and Loren Heun (Western
Michigan University) for statistical analysis. The authors acknowl-
edge the multitude of constructive comments from colleagues
who reviewed early drafts of this manuscript and the helpful
comments provided by anonymous reviewers.
References
Brandt, A.R., 2007. Testing Hubbert. Energy Policy 35 (5), 3074–3088.
Brandt, A.R., 2010. Review of mathematical models of future oil supply: historical
overview and synthesizing critique. Energy 35, 3958–3974.
Brown, M.T., 2009. Predicting national sustainability: the convergence of
energetic, economic and environmental realities. Ecological Modeling 220,
3424–3438.
Center for Environmental Policy., 2010. In: Sixth Biennial Emergy Research Conference.
Available at: /http://www.cep.ees.ufl.edu/conference6schedule.aspS (accessed
16 July 2010).
Cleveland, C.J., Costanza, R., Hall, C.A.S., Kaufmann, R., 1984. Energy and the U.S.
economy: a biophysical perspective. Science 225, 890–897.
Cleveland, C.J., 1999. Biophysical economics: from physiocracy to ecological
economics and industrial ecology. In: Gowdy, J., Mayumi, K. (Eds.), Bioeco-
nomics and Sustainability: Essays in Honor of Nicholas Georgescu-Roegen,
Edward Elgar Publishing, Cheltenham, pp. 125–154.
Cleveland, C.J., 2005. Net energy from the extraction of oil and gas in the United
States. Energy 30, 769–782.
Cleveland, C.J., Costanza, R., 2008. Biophysical economics. In: Cleveland, C.J. (Ed.),
Encyclopedia of Earth, Environmental Information Coalition, National Council
for Science and the Environment, Washington, D.C. Available at /http://www.
eoearth.org/article/Biophysical_economicsS (accessed 3 June 2010).
Energy Information Agency, Office of Oil and Gas. 2005. The relationship between
crude oil and natural gas prices. Available at /ftp://ftp.eia.doe.gov/pub/
oil_gas/natural_gas/feature_articles/2006/reloilgaspri/reloilgaspri.pdfS
(accessed 23 August 2011).
Energy Information Agency. 2010. World crude oil prices. Available at /http://
www.eia.doe.gov/dnav/pet/pet_pri_wco_k_w.htmS (accessed 1 May 2010).
Fouquet, R., 2010. The slow search for solutions: lessons from historical energy
transitions by sector and service. Energy Policy 38, 6586–6596.
Friedrichs, J., 2010. Global energy crunch: how different parts of the world would
react to a peak oil scenario. Energy Policy 38 (8), 4562–4569.
Gagnon, N., Hall, C.A.S., Brinker, L., 2009. A preliminary investigation of energy
return on energy investment for global oil and gas production. Energies 2,
490–503.
Greene, D.L., Hopson, J.L., Li, J., 2006. Have we run out of oil yet? Oil peaking
analysis from an optimist’s perspective. Energy Policy 34, 515–531.
Hall, C.A.S., Balogh, S., Murphy, D.J.R., 2009. What is the minimum EROI that a
sustainable society must have? Energies 2 (23), 25–47.
Hall, C.A.S., Powers, R., Schoenberg, W., 2008. Peak oil, EROI, investments and the
economy in an uncertain future. In: Pimentel, D. (Ed.), Renewable Energy
Systems: Environmental and Energetic Issues, Elsevier, London, pp. 113–136.
Hamilton, J., 2009. Causes and consequences of the oil shock of 2007–08.
In: Romer, D., Wolfers, J. (Eds.), Brookings Papers on Economic Activity,
Conference Series, Brookings Institute, Washington DC, pp. 1–68.
Heun, M.K., VanderLeest, S., 2008. Why a liberal and multidisciplinary education is
needed to solve the energy crisis. In: Proceedings of the American Society for
Engineering Education Annual Conference and Exposition, Pittsburgh, PA,
Paper number AC 2008-2264.
Hotelling, H., 1931. The economics of exhaustible resources. Journal of Political
Economy 39, 137–175.
Hubbert, M.K., 1956. Nuclear Energy and the Fossil Fuels. Presentation before the
Spring Meeting of the Southern District Division of Production, American
Petroleum Institute, Plaza Hotel, San Antonio, Texas, 7–9 March.
Inflation Data. 2010. Historical crude oil prices. Available at /http://inflationdata.
com/inflation/Inflation_Rate/Historical_Oil_Prices_Table.aspS (accessed 31
August 2010).
Kern, F., Smith, A., 2008. Restructuring energy systems for sustainability? Energy
transition policy in the Netherlands. Energy Policy 36 (11), 4093–4103.
King, C.W., Hall, C.A.S., 2011. Relating financial and energy return on investment.
Sustainability 3 (10), 1810–1832.
Krautkraemer, J., Toman, M., 2003. Fundamental economics of depletable energy
supply. Discussion Paper 03-01 RFF, November.
Kneese, A.V., Sweeny, J.L., 1993. Handbook of Natural Resource and Energy
Economics, vol. 3. Elsevier, Amsterdam, pp. 759–1357.
Leijonhufvud, A., 1981. Information and Coordination: Essays in Macroeconomic
Theory. Oxford University Press, New York.
Loorbach, D.A., Rotmans, J., 2006. Managing transitions for sustainable develop-
ment. In: Olsthoorn, X., Wieczorek, A.J. (Eds.), Understanding Industrial
Transformation, Springer, Dordrecht, pp. 187–206.
Mohr, S., 2010. Projection of world fossil fuel production with supply and demand
interactions. Ph.D. Thesis, University of Newcastle, Australia.
Murphy, D.J., Hall, C.A.S., 2010. Year in review—EROI or energy return on (energy)
invested. Annals of the New York Academy of Sciences 1185, 102–118.
Murphy, D.J., Hall, C.A.S., Powers, B., 2010. New perspectives on the energy return
on (energy) investment (EROI) of corn ethanol. Environment, Development
and Sustainability, 1–24.
National Academy of Engineering (NAE). 2008. Grand challenges for engineering.
Available at /http://www.engineeringchallenges.org/S (accessed 21 Septem-
ber 2010).
Nashawi, I.S., Malallah, A., Al-Bisharah, M., 2010. Forecasting world crude oil
production using multicyclic Hubbert model. Energy Fuels 24 (3), 1788–1800.
Odum, H.T., 1996. Environmental Accounting: Emergy and Environmental
Decision Making. Wiley and Sons, New York.
Rosser Jr, J.B., 1999. On the complexities of complex economic dynamics.
The Journal of Economic Perspectives 13 (4), 169–192.
Rotmans, J., Kemp, R., Van Asselt, M., 2001. More evolution than revolution:
transition management in public policy. Foresight 3 (1), 15–31.
R Development Core Team. 2010. R: A language and environment for statistical
computing. R Foundation for Statistical Computing, Vienna, Austria. /http://
www.R-project.org/S.
Smith, A., Voß, J.-P., Grin, J., 2010. Innovation studies and sustainability transi-
tions: the allure of the multi-level perspective and its challenges. Research
Policy 39 (4), 435–448.
Solow, R.M., 1974. The economics of resources or the resources of economics.
American Economic Review 64 (2), 1–14.
Sorrell, S., Speirs, J., Bentley, R., Brandt, A., Miller, R., 2010. Global oil depletion:
a review of the evidence. Energy Policy 3 (9), 5290–5295.
Sweeney, J.L., 1993. Economic theory of depletable resources: an introduction.
In: Kneese, A.V., Sweeney, J.L. (Eds.), Handbook of Natural Resource and Energy
Economics, Volume 3. , Elsevier, Amsterdam.
Tietenberg, T., 1992. Environmental and Natural Resource Economics, fourth ed.
Harper Collins, New York.
United Nations (UN). 2000. United Nations development goals. Available at
/http://www.un.org/millenniumgoals/S (accessed 21 September 2010).
United States Energy Information Agency (US EIA). 2010. Annual energy review
(AER). Available at /http://www.eia.doe.gov/aer/finan.htmlS (accessed 31
August 2010).
Victor, D., 2009. The Politics of Fossil-Fuel Subsidies. Global Subsidies Initiative,
IISD, Geneva.