Home‎ > ‎

Energy and Limits to Growth

Towards the end of The Protestant Ethic and The Spirit of Capitalism, Max Weber wrote:

The Puritan wanted to work in a calling; we are forced to do so.  For when asceticism was carried out of monastic cells into everyday life, and began to dominate worldly morality, it did its part in building the tremendous cosmos of the modern economic order.  This is order is now bound to the technical and economic conditions of machine production which today determine the lives of all individuals who are born into this mechanism, not only those directly concerned with economic acquisition, with irresistible force.  Perhaps it will so determine them until the last ton of fossilized coal is burnt.  (1958:  181).

Not only does this passage eloquently articulate the binding imperatives engendered by the “modern economic order”, it also unambiguously contravenes the view that the classical theorists were unaware of the essential role that energy played in the creation and maintenance of that order.  Thus far I have paid little attention to the role that the natural environment plays in the constitution of social systems.  Thus far, the theme of ecology has remained a part of the environment of this communication.

Below I explicitly address the relationship between natural resources and the modern economic system.   Specifically, I explain the relationship between energy use and economic growth, and the relationship between “peak oil” and the global debt crisis.  My primary finding is that the economic activity, measured by GDP, has to some extent decoupled from energy expenditure, both in the US and globally, since the 1970s.  This does not imply that energy use is not essential for economic activity.  However, there is no one-to-one correspondence between geophysical systems and monetary/economic systems.  I conclude with some plausible explanations for my findings.

Mainstream Economic Theory and Energy

Mainstream economic theory is blind to the realities of ecological scarcity and the laws of thermodynamics.  Nothing illustrates this better than the contemporary, textbook depiction of the economy as a circular flow.

In this model, the economy is disaggregated into individuals (households) and businesses (firms).[1]  The blue line represents the flow of money, and the green line represents the flow of goods and services.  Neither the monetary nor physical origins of surplus are explained.

Figure 57.  Circular Flow.

To account for the former (i.e. profit), neoclassical economic theory invokes the idea of returns to factors of production, labor and capital. In contrast to Marxian and Sraffian theory, neoclassical theory does not regard profit as a material surplus, but rather, as the price paid to the marginal contribution of capital, understood as an abstract quantity of money, which moreover is substitutable for labor.  The system is a closed perpetual motion machine.

Figure 57 depicts the circular flow model from the Bureau of Economic Analysis (BEA).  Importantly, in this model there are no external inputs to the production process, nor are there any waste products (aka “externalities”).   It is a closed system.  In reality, the production of material wealth transforms low entropy inputs into high entropy outputs, and thus requires an entropy gradient.  As Herman Daly presciently observes:   “The high-entropy output cannot be directly used again as an input for the same reason that organisms cannot eat their own excrement” (1991:  16).

Mainstream theory, including neoclassical and Keynesian, regards increased resource extraction as a consequence rather than a cause of growth.  The equilibrium conditions posited by neoclassical theory pertain to utility, a concept derivative of the classical concept of exchange-value.  It ignores use-value, or material wealth.  For instance, in Walras’s model of exchange, market outcomes can generate the best of all possible worlds with respect to utility, but it presumes that the products from which utility is derived are already produced.  In other words, it cannot explain how growth occurs, or how efficiency of production is improved.  Standard calculations show that about a quarter of productivity growth can be accounted for by so-called human capital (population increases, skill improvements, knowledge, etc.); another quarter can accounted for by capital stock accumulation (Mankiw 2002).  This leaves about half of productivity growth unaccounted for in the standard growth models (e.g. the Cobb-Douglas function), sometimes known as the Solow residual, after the economist Robert Solow.  Neoclassical theory attributes productivity growth to technology.

According to the mainstream models, growth derives from the investment of savings.  Quantitatively, the growth-from-savings theory can be expressed in terms of a decomposition of output Y.  As stated above, the output of an economy Y can be decomposed into four components:  personal consumption (C), business investment (I), government spending (G), and net exports (NX).  Assuming a closed economy for simplicity of exposition, national income can be written as:


Subtracting the money spend on consumption and government spending from both sides yields:


The left side of the equation is also equal to national savings (S), so that:


Savings can also be decomposed into private and public savings. Let T equal taxes.  Savings is then equal to:

S=(Y-T-C) + (T-G)

The left hand side is private savings, which is equal to total income minus taxes and consumer spending.  The right hand side is public savings, or the amount of tax revenue collected minus the amount the government spends as transfer payments.  When T exceeds G, the government runs a budget surplus.

This theory is attractive because it links macro-economic growth to the time preferences of individual actors, modeled as utility maximizing monads.  An individual will save when the expected utility of future returns exceeds the present utility of present consumption, taking into account the greater weight attached to present consumption, both because the future is risky, and because some sacrifice of present pleasure is required.  Saving for the future, however, means delaying the gratification derived from consuming/spending in the present.  The weighting factor which discounts the magnitude of expected utility is known as the "discount rate."  The discount rate refers to the tendency, first described by Pigou (1920), for people to prefer present consumption over future consumption.  Future consumption, therefore, has to be sufficiently attractive to compensate for the postponement of preferred consumption in the present. Sacrificing present consumption for future returns is another way of expressing Max Weber's notion of the "the Protestant Ethic."

If growth is assumed to be automatic, why save at all?   Moreover, investment can also derive from credit (i.e. debt) in addition to savings.  Through fractional reserve banking, banks create temporary money to create savings without investments.  The amount of investment that can be generated is some function of the money multiplier, the amount of money the banking system generates with each dollar of reserves.  The money multiplier is the reciprocal of the reserve ratio (Makiw 2002:  331).  For instance, if the reserve requirement is 10 percent, or .10, then the money multiplier will is equal to 1/.01 = 10.  In a fractional reserve system that requires 10 percent deposits, then, 10 times the amount of money originally deposited in a bank can be generated.

This means that, increasingly, investment is financed through debt rather than equity (i.e. savings).  Debt has to be paid back, plus interest.  Therefore, in order to remain solvent, individual firms and households in debt must generate more income than previously:  debt thus fuels the growth imperative.  Debt is generated by and generates growth, but it also compels growth, in the sense that debtors have to grow in order to remain solvent.  Consequently, the notion that credit is created from savings (i.e. savers lend their money) does not imply a one to one correspondence between them.

According to mainstream economic theory, physical resource stocks place no limitations on economic growth.  A typical mainstream view of the role that natural resources play in economic growth can be found in Greg Mankiw’s economics textbook[2]:

[M]ost economists are less concerned about such [environmental] limits to growth that one might guess.  They argue that technological progress often yields ways to avoid these limits.  If we compare the economy today to the economy of the past, we see various ways in which the use of natural resources has improved.  Modern cars have better gas mileage….  More efficient oil rigs waste less oil in the process of extraction…..

In a market economy, scarcity is reflected in market prices.  If the world were running out of natural resources, then the prices of those resources would be rising over time. But, in fact, the opposite is nearly true.  The prices of most natural resources (adjusted for overall inflation) are stable or falling.  It appears that our ability to conserve these resources is growing more rapidly than their supplies are dwindling.  Market prices give no reason to believe that natural resources are a limit to economic growth.  (2002:  246-7).

The statement that “our ability to conserve these resources is growing more rapidly than their supplies are dwindling” does not at all mean that the absolute stock of resources is no longer declining.  The fact that improved efficiency of extraction can coincide with increased rates of extraction is known as the Jevons Paradox.[3]  Mankiw writes as if resources were infinitely substitutable, and moreover, that resources are renewable at any possible rate of extraction.  Both of these assumptions are false.

Although the actual prices of most commodities, including energy, have fallen for most of the past century, this does not mean that they will continue to do so.  There is ample evidence that we are approaching, or have already approached, the peak rates of oil supply (Deffeyes 2001; Simmons 2005; Heinberg 2007).  Moreover, economic growth has always been coupled to increased energy expenditure (cf. McKillop 2005).  Moreover, the costs of extraction are not taken into account by national accounting statistics.  These include, for example, the recent BP oil spill disaster in the Gulf of Mexico.

What does GDP measure?

Growth is the desideratum of economic policy.  What, however, does growth mean?  What sort of growth is desirable, and how should growth be measured?  Herman Daly refers to growthmania as the mindset which assumes that the answer to every problem is growth.  The costs of growth, however, are not measured, and can even be counted as benefits.  

Aggregate economic growth is not something that can be observed directly as can say, the growth of a population of fish, a crystal, or a pile of garbage.  Economic growth can only be inferred through the use of constructed indicators such as Gross Domestic Product (GDP), the standard indicator of economic growth for a country.  There are numerous indicators of national economic growth including Gross National Product (GNP), Net Domestic Product (NDP), Net National Product (NNP), Gross National Income (GNI), and Gross Domestic Income (GDI), among others.    The GDP is calculated as the market value of all final goods and services made within the borders of a country in a given year.  GDP can be calculated according to final products, expenditures, or income, all of which in theory should generate the same results.  There are therefore a number of ways the composite measure can be decomposed.

Gross Domestic Product (GDP) is often used as a proxy for prosperity.   Raising the per capita GDP has consequently been the focus and objective of developmental policies for virtually all governments and International Governmental Organizations (IGOs), such as the International Monetary Fund (IMF) and the World Bank (WB).  Using GDP as a measure of prosperity, however, is misleading for several reasons.  The economist Herman Daly (1973) refers to the belief that economic growth is the panacea for all social problems as "growthmania," which he compares to psychological delusion.  First, average per capita GDP of a nation or region does not measure the inequality of incomes within and between nations or regions.  Using the arithmetic mean of income is not a measure of the most common income level (measured by a mode or by a harmonic mean) or how the income is distributed.  A nation with a GDP of $100 and 10 people where 1 person owns all of the wealth, has the same average per capita GDP as a nation where 10 people each possess $10.  The per capita GDP is $10 in both cases.  In addition, inequality within a nation or area itself can cause a decline in overall social well-being.  To measure inequality, measurements of dispersion (e.g. the 'GINI' index, the Theil index, etc.) are more useful.

According to Amartya Sen, moreover, whether and to what extent increases in per capita GDP, relative either to the past or other regions, translates into more freedom and well-being is influenced significantly by five distinct sources of variation including:  a) personal factors such as sickness and health; b) environmental factors such as climate, air and water pollution, natural resources, etc.; c) social climates (e.g. friendly, corrupt, violent); d) inter-societal income distribution; and e) inter-family income distribution (cf. Deb 2009:  78).  Gross Domestic Product does not take into account unpaid services and activities, such as domestic labor.  Equally disturbing is the fact that natural and anthropogenic disasters can inflate the GDP.  Efforts to clean up pollution or to rebuild after a calamity of some kind are counted as part of a country's expenditure and thus can raise the GDP, even when human well-being has clearly declined.[4]

Numerous cross national studies have also shown that, after about $30,000 or so, rising per capita GDP does not significantly contribute to happiness and well-being.  Nations with higher average GDPs are relatively less happy than their less affluent counterparts.  The last point I will mention is that the relationship between GDP and prosperity has not remained constant across time.  Specifically, both the historic links between GDP growth and employment, and the link between GDP growth and rising wages, have been broken (Ayres and Warr 2009; Rifkin 1995).  Because of these indexes other development indexes that have been devised.  The United Nation's Human Development Index (HDI) and the Genuine Progress Indicator (GPI) are examples.  

The most important point for purposes of this paper, however, is that GDP is theoretically incoherent.  Herman Daly explains:

The problem of [GDP] is that it adds together three very unlikely categories:  throughput, additions to capital stock, and services rendered by the capital stock.  Throughput (the entropic depletion-pollution flow) is the ultimate physical cost.  Services rendered by physical and human capital represent a value estimate of the final benefit, or true psychic income, resulting from economic activity.  Additions to capital stock represent an increased capacity for future service, the net cost of which (throughput) has been incurred in the present, but the net benefits of which accrue only in the future.  These three distinct concepts should be kept in separate accounts.  It makes no sense to add together costs, benefits, and changes in capital stock.  It is as if a firm were to add up its receipts, its expenditures, and its change in net work.  What sense could any accountant make of such a sum? (1991:  30)

An increase in GDP represents an increase in the capacity for making payments and nothing more. Moreover, the flow of economic activity is in part based upon the stocks of capital that have accumulated in prior years.  A factory does not have to be built from scratch every year.  Stocks can still generate a flow of goods that, if sold, can contribute to GDP.  One should not expect, therefore, a point by point correlation between changes in the flow of energy and the flow of economic activity.  Moreover, the accumulation of debt has mollified much of the international competition that would have inflated the prices for scarce resources such as petroleum even higher.  The growth of finance has meant that the rate at which money is accumulated by lending institutions exceeds the rate at which this money is lent.  Instead of using the money to make payments for productive resources, the money is instead stockpiled, thereby postponing the exchange of money for goods and services, all of which depend on energy extraction.

Finally, rising scarcity of raw materials (partially) reflected in declining rates of energy output and rising energy prices has engendered a process by which exchange in the economy is increasingly constituted by the exchange of already extant stocks (e.g. real estate).  Thus unique historical dynamic has been translated into exploding debt, rising finance, and growing inequality.

The economy is ultimately a social construct, inseparable from how we measure it.  Accordingly, GDP does not merely reflect an already existing economy, but steers policy decisions about how best to manage it, and in this sense, in part performs and constitutes that which it measures.  The economy can be regarded as an operationally closed system in the sense that Norbert Weiner and Niklas Luhman have used the term, that is, as closed to information but open to the matter and energy.  In this sense, it is not real, as are biophysical systems, but has presence only as codified communications.  Whether or to what extent economic growth reflects underlying material flows and human welfare depends entirely upon how we measure and understand it.  In short, the economy exists in the medium of ideology, broadly understood .  The ultimate ends of the economy qua economic theory is provided a functional equivalent in the self-referencing operation of its abstract means of self-measurement.

Entropy and the Economic Process

In 1971, Georgescu-Roegen published his seminal work The entropy law and the economic process.  This work represented a paradigm shift in thinking about economic processes and more generally, about the interrelation between natural resources and social order.  The basic gist of his argument is that economic production obeys the Second Law of Thermodynamics, also known as the Entropy Law.  The First Law of thermodynamics, also known as the Conservation Law, stipulates that the total quantity of matter and energy in the universe does not change, but only changes form.  The Second Law of thermodynamics, or the Entropy Law, states that the reduction of entropy in one place is always accompanied by a greater magnitude of entropy in the overall environment.  

The term entropy was first coined in 1868 by the German physicist Rudolf Clausius.  The concept as used in thermodynamics can be understood in three ways:  1) as a transition from a more concentrated to a less concentrated state; 2) as a transformation from available energy (exergy) into unavailable energy; and 3) as a transition from order to disorder.  In short, entropy can be thought of as an inverse measure of order, concentration, or available energy.[5]  Available energy is also known as exergy.

Economics has generally disregarded or ignored the profound implications of the Entropy Law for economic growth and production.  According to the Law of Entropy, all processes of production create more disorder (waste) than they do order. The universe is therefore in a state of continuous qualitative degradation (Rifkin 1980).  Moreover, this process occurs automatically by itself, regardless of whether mechanical work is consciously performed.  Mechanical work, however, does accelerate the entropy process.  The more we produce, the faster the degradation.  Georgescu-Roegen notes that:

Economic processes materially consist of a transformation of low entropy into high entropy, i.e. into waste… [W]aste is an inevitable result of that process and ceteris paribus increases in greater proportion than the intensity of economic activity.  (1972)

Herman Daly recapitulates this most fundamental point:

That low entropy is the common denominator of all useful things is evident from the second law of thermodynamics.  All states of matter and all forms of energy do not have equal potential for use.  Though we neither create not destroy matter-energy in production and consumption, we do transform it.  Specifically, we transform matter from organized, structured, concentrated, low-entropy states (raw materials) into still more highly structured commodities, and then through use into dispersed, randomized high-entropy states (waste)….  All life processes and all technological processes work on an entropy gradient.  In all physical processes the matter-energy inputs in their totality are always of lower entropy than the matter-energy outputs in their totality.  Organisms canot survive in a medium consisting of their own final outputs.  Neither can economies.  (1991:  22)

Economics is exclusively concerned with the relation between intermediate means and intermediate ends, ignoring both ultimate means and ultimate ends.  Daly argues that we will eventually run out of a) “worthwhile ends whose satisfactions depend on further conversionof ultimate into intermediate means”, that is, the increasing depletion of resources, and b) ultimate means, or reach limits to the rate at which ultimate means can be used.  Systems, including economic systems and technologies, work by transforming available into unavailable energy, thus accelerating the entropy process.

In the past 200 years, humanity has become increasingly dependent upon nonrenewable minerals for its survival, including petroleum, coal, and other fossil fuels.  Because these are nonrenewable, humanity is increasingly relying on a phantom carrying capacity.  Carrying capacity refers to the population size of a species that the environment can sustain indefinitely.  Likewise, phantom carrying capacity is described by William Catton Jr. as:

Either the illusory or the extremely precarious capacity of an environment to support a given life form of a given way of living.  It can be quantitatively expressed as that portion of a population that cannot be permanently supported when temporarily available resources become unavailable (1980).

Low entropy energy constitutes the ultimate means and exists in two forms:  a terrestrial stock and a solar flow.  Furthermore, the terrestrial stock takes two forms:  “those renewable on a human time scale and those renewable only over geologic time and which, for human  purposes, must be treated as nonrenewable” (Daly 1991:  21).   Ultimately, the human economy is a subsystem of the steady-state ecosystem.   Accordingly, the human economy must become a steady state either willingly or by extinction.  Trends cannot continue indefinitely.  A growing system will always come up against internal or external barriers (Wallerstein 2004; Meadows 2008).

Empirical findings:  Growth and Energy

Economic growth has always been coupled with growth in energy extraction, production, and depletion.  Figure 58 is a plot of US GDP (in constant dollars) against US energy use (kt oil equivalent), from 1960 to 2009.  World estimates are provided in Figure 59and exhibit the same clearly identifiable relationship.

This relationship does not establish a causal relationship.[6]  Growth and energy production are in a sense two ways of measuring a common phenomenon.  Growth entails energy use and vice-versa.  Neoclassical economics considers growth a cause of energy use.  Certainly this is not incorrect.  Petroleum by itself does not cause its extraction and refinement, for example.  Neoclassical theory thus adopts the agency paradigm of causal attribution, regarding growth qua human agency as the efficient cause of energy consumption.

Although human agency qua intention can be regarded as an efficient cause of energy consumption, human agency cannot be regarded as a material cause.  This is why the “production” of energy is actually a misnomer.  Humans do not produce energy, rather, humans, using their endosomatic and exosomatic instruments, extract and deplete exergy.
Figure 58.  Energy use vs GDP.  United States.  1960-2009.  Source:  World Bank.

Figure 59.  World GDP vs. World Energy Use.  1971-2009.  Source:  World Bank.

In addition, GDP and energy production are flow variables, as opposed to stock variables.  To regard the extraction of resources as entirely determined by human agency is to assume that there are no exogenous constraints placed upon this agency and that there are no limits to growth.  That GDP is a flow variable is also important, because conceptually it means that a “growing economy” only indirectly refers to the stocks that are accumulated and depleted via economic activity.

Peak Oil

Oil contributes to about 40 percent of energy production and supplies 90 percent of all transportation fuel (Korowicz 2010).  It is therefore vitally important to the functioning of the global economic system.   For most of the 20th century, high quality petroleum could be extracted at very little cost.  A barrel of oil, which could be extracted for a dollar, would in turn generate 25,000 hours of labor.  One dollar equals 25,000 hours of labor.   This is essentially free energy, and it has fueled the economic growth of every nation on earth.

Up until the 1950s, the United States was the “Saudi Arabia of oil” in the sense that it was world’s largest exporter.  Its production, however, peaked in 1970 at 10.2 million barrels a day and subsequently declined.  Ten years later, domestic oil production was still in decline, despite the fact that ten times more oil wells had been drilled.  Currently the United States uses 25 percent of the world’s oil but possesses only 2 percent of the world’s known reserves.

Today, there are about 50 countries that are producing less oil today than in the past.  Ironically, more efficient means of extraction petroleum has only expedited its depletion, acting as giant “super straws” sucking the last easy oil out of the ground at faster and faster rates, but without significantly increasing the amount of petroleum that would be produced from any given oil field.  The last great oil discoveries of the 20th century, which effectively postponed the point of peak global production, were fields in Alaska,  Siberia, and the North Sea, discovered in 1967, 1968, and 1969, respectively.   Worldwide discovery of oil peaked in 1964.

Contrary to Mankiw’s assertions, the price of crude oil is rising, not falling.  Figure 60 is a time series depicting world crude oil production, measured as millions of barrels per day, and average oil prices for the US (measured as USD).  Between 1960 and 1970 global petroleum production increased 118.6 percent.  By contrast, between 1971 and 2009, global petroleum production increased only 52.1 percent.

Probably the most important fact, however, is that global petroleum production has remained nearly flat since 2005.   In this year the Energy Information Administration (EIA) estimates that an average of 73.74 million barrels of oil was extracted daily.  This declined slightly until 2008, when it increased to 73.78 million barrels of oil per day, an increase of only .054 percent over four years.  The average annual percent change of production from 1960 to 1970 is 8.139 percent, whereas the average annual percent change of production from 1971 to 2009 is only 1.311 percent.

The rate of global oil production is declining, but the price of oil is rising.  From 1999 to 2008, the average price for a barrel of oil in the US rose by a factor of 5, from $11.27 to $76.82.  These are annual averages.  Oil prices reached a high of $145/barrel on July 3, 2008, only to be followed by a dramatic decline.[7]  A time-series of oil price data from 1960 to 2008 is depicted in Figure 61.  In the late 1970s a series of events in the Middle East sent the price of oil over $40/barrel.  By 1982, oil prices peaked at an annual average of $53.74.  The two most important events disrupting supply were the Iranian revolution in 1978 and Iraq’s invasion of Iran in 1980.  In 1978, world oil production increased by 0.7 percent, a significant deceleration from its previous 1977 growth rate of 4 percent.  By 1980, however, global petroleum extraction declined 4 percent, and continued to decline another 5.8 percent and 4.6 percent in 1981 and 1982, respectively.

Between 1960 and 1974, a stable relationship is exhibited between oil price and oil quantity.  A supply curve is given by plotting price (y) and oil produced (x).  This is illustrated in Figure 62.  Notice the downward slope.  As prices fall, so does the quantity of oil produced.  This is in contradistinction to the standard supply curve in economic theory.  The price elasticity of supply (PES) can also be inferred for this period.  Calculating the percentage change in quantity over the percentage change in price yields a negative slope of approximately 13.[8]

Figure 60.  Global petroleum production 1960-2008.  (Source:  EIA)

Figure 61.  Oil price average USD 1960-2008. (Source:  EIA)

According to the law of supply, supply curves should be positive, not negative.  Note that this could be explained as the result of the interaction of a shifting supply curve and a stable or constant demand curve.   In standard economic analysis, prices are at equilibrium, which is defined as the point where the (falling) demand curve and the (rising) supply curve meet.  Assuming that prices are at equilibrium, this result can be obtained by assuming that technology or other factors make petroleum significantly less expensive to produce.  When this occurs under conditions of relatively stable global demand, rising output can coincide with falling prices.

This relationship changes abruptly in the mid-1970s due to political events in the Mideast discussed above.  Abstracting away from these events, and tracing the relationship between quantity supplied and prices for the years 1987 to 2008, yields the time series depicted in Figure 63.  What is clear is that the relationship begins to break down, and the (linear) slope becomes positive.  The relationship also becomes curvilinear or exponential.

A divergence in the relationship between petroleum and GDP is also exhibited.  Plotting the ratio of oil price to oil quantity (P/Q) against real Gross Domestic Product for the period 1960 to 1974 yields the time series in Figure 64.   The price of oil declines while total economic output rises.  This relationship clearly changes during the period of 1987 to 2008 as depicted in Figure 65.  Limits to growth are implied and the law of diminishing marginal returns is confirmed.  As the price of oil rises relative to its quantity, GDP rises but at a diminishing rate.  A similar relationship holds between petroleum price and production to world per capita GDP, as depicted in Figures 66 and 67.  The relationship dramatically changes during the period of 1987 to 2008.

An important indicator of net energy production is a ratio known as Energy Returned on Energy Invested (EROEI).  Although difficult to measure precisely, it is widely acknowledged that the EROEI is declining substantially for both gasoline and petroleum.  In 1930, one barrel of oil was needed to produce 100 barrels of oil output, that is, the EROEI was 100:1 (Hagens 2008).[9]   By 1970, the EROEI for oil in the United States had declined to an estimated 30:1, and by 2000 the EROEI for oil was estimated to be between 18:1 and 11:1.  

Another important indicator is peak oil production per capita, which represents the largest amount of oil that can be used by each person on the planet if equally distributed.  This has been in decline for the past half century, as indicated in the Figure 68.

Oil production for the United States peaked in 1970 and thereafter began its decline.  For much of the 20th century, the United States was the largest oil exporter in the world.  Petroleum, however, is a nonrenewable resource.  The volume of petroleum that we are able to extract must eventually decline.  The production of oil over time tends to follow a logistic distribution curve, first rising then peaking, and eventually declining.  This has been the production pattern exhibited for individual oil fields and for individual countries.

There are two reasons why oil production exhibits a logistic growth pattern over time.  First, global petroleum extraction tends to lag behind discoveries of oil reserves.  M.K. Hubbert correctly predicted the year that petroleum production would peak in the United States.  Peak oil production is therefore also known as Hubbert’s peak.

Figure 62.  Oil Price and Quantity (1960-1974).

Figure 63.   Oil Price and Quantity (1987-2008)

Figure 64.  Oil Price-to-Quatity versus GDP (1960-1974)

Figure 65.  Oil Price-to-Quantity versus GDP (1987-2008)

Figure 66.  Oil price-to-quantity versus World per capita GDP (1960-1974)

Figure 67.  Oil price-to-quantity versus World per capita GDP (1987-2008)

Oil wells may still be being discovered, but their rate of discovery continues to decline.  Second, oil extraction follows the principle of the lowest hanging fruit.  The cheapest, most energy dense crude is extracted first, followed by oil that is increasingly less energy dense and more expensive to extract.[10]  Moreover, the specific use-values of petroleum (from which plastics, fertilizers and pesticides, pharmaceuticals, and over 90 percent of transportation fuel, are derived) are non-substitutable.

Stagnating supply coupled with increasing global energy demands has translated into rising energy prices.  Although US production peaked in 1970,  world production continued to increase in large part because of new oil field discoveries including the oil reserves in the North Sea and Mexico’s Cantarell Field, which was formerly the world’s second largest producing field (Hamilton 2009:   11).  

Both of these, and many others, are today in decline.  Saudi Arabia is the largest exporter and its stagnation contributed greatly to global stagnation beginning in 2005.  Figure 69 depicts quantity of petroleum production for the United States and Saudi Arabia from 1960 to 2008.

Figure 68.  World oil production per capita.  1960-2003.  Source:  Energy Energy Information Administration (EIA).  Population figures from Ecological Footprint Network.

Figure 69.   Oil Production for US and Saudi Arabia, 1960-2008.   Source:  Energy Information Administration.

Figure 70.  Growth in per capita Energy Consumption.  US, OECD, China.  1990-2006.  Source:  World Bank.

The demand for oil has meanwhile accelerated.  An important source of increased demand is from China, which saw a 16 percent increase in its energy consumption in 2004.  Although still significantly smaller in terms of absolute numbers, its relative growth in oil consumption far outpaces that of the United States and the rest of the OECD countries, as indicated in Figure 70.

In a thorough review of the data and available studies, Hamilton concludes that:  “had there been no oil shock, we would have described the U.S. economy in 2007:Q4-2008:Q3 as growing slowly, but not in a recession” (2009:  40).  The primary reason being that oil price hikes had a significant impact on consumer spending.

Of course, that consumers are spending on something does mean that the money is going somewhere, and that therefore demand should equal supply in the aggregate, according to Say’s Law.  However, the problem is that much of this money ends up being invested in the capital required to obtain increasingly scarce energy resources, and is consequently concentrated in the hands of a few oil exporters, firms, and individuals.

Peak Coal

According to standard neoclassical theory, energy inputs are substitutable.  Even if petroleum production might be stagnating for geophysical reasons, and not entirely because of speculation or political decisions on the part of OPEC, then some might conclude that other energy sources should be able to meet increases in global energy demand.  One likely candidate is coal.

Today, the United States is the second largest producer of coal after China.  However, because coal is nonrenewable, it too will peak.  Although estimates vary regarding when this coal production will peak, the Energy Watch Group and the Uppsala Hydrocarbon Depletion Study Group both predict that coal production will peak by 2020 or 2025.[11]  Moreover, the quality of coal, like petroleum, has been in steady decline as measured by the EROEI.  Consequently, although coal production per weight has not yet peaked, the total amount of energy generated from coal in the United States already peaked in 1998 (Heinberg 2009).

Historically, coal mines only extract about 50 percent of their economically viable coal reserves.[12]  For surface mining, this figure is 85 percent (cf. Heinberg 2009).  Of course, economically viable reserves are tied to price and demand.  Because of the inelasticity of energy goods, prices will tend to rise making marginal quality coal more profitable, rather than making it cost-prohibitive.

In addition, there are the unaccounted costs of coal extraction, including the costs to coal miners and the ecological devastation wrought by strip mining, including “Mountain Top Removal.”  Coal is also the most carbon intensive and polluting energy source.  At best therefore, coal will remain a stop-gap measure to meet increased energy demands.

Energy and Income

Spending on fuel and other energy sources is primarily a function of income, not price.  To see this, examine Figure 71, which depicts personal spending on gasoline (and other energy products) plotted against total income, from 1947-2009.  Figure 71 is significant because it shows that energy spending rises with income, irrespective of price.  The (absolute value of) price elasticity demand (PED) for petroleum is very low, or in other words, the demand for petroleum is highly inelastic.[13]    Hamilton (2009) reports that the PED for oil is 0.07; the PED for gasoline is much higher at 0.26.

Figure 72 shows energy spending as a percentage of both total spending and income for the United States for the years 1947 to 2010.  The price inelasticity of petroleum is significant because it means that higher prices can lead to a significant decline in demand for other goods with higher demand elasticities.  Whereas in the late 1970s and 1980s, high energy prices can be attributed to exogenous political disturbances, the same cannot be said for the most recent oil price hike.   As reported by Hamilton:  “the big story has not been a dramatic reduction in supply … but a failure of production to increase between 2005 and 2007” (2009:  9).    The question is, why has global oil production begun to stagnate?

Figure 71.  Energy spending vs. Income.  Source:  BEA NIPA tables 2.3.5 line 11 and Table 2.1 line 1.

Figure 72.  Energy spending as % of Income and Total Spending. Source:  BEA NIPA Tables 2.3.5 and 2.1

Limits to Growth

Few scholars have linked the current economic crisis to geophysical constraints on energy production. One of the few is Ramon Lopez (2008), who attributes the financial crisis to three interrelated structural factors:

i)      The incorporation of highly populated countries into growth process;
ii)      The increasing scarcity of natural resources; and
iii)     The unprecedented concentration of wealth and income in advanced economies in past 3 decades

Since the industrial revolution, the global “North” (i.e. the developed/'First world' countries) grew under conditions of constant or even declining commodity prices, including important raw materials and natural resources.  In the 1990s, however, this changed. The NIGs (new industrial giants) including China, India, Russia, and Brazil experienced growth rates up to 3 times faster than the advanced countries.  In fact, a 1/3 of total annual world growth is attributed to the NIGs.  Because the NIGs represent over 50 percent of humanity, their industrialization has generated an increased demand for energy, food, and other raw materials accelerating their physical depletion and monetary inflation.

Although global crude oil production rates are nearly 4 times what they were a century ago, increased global demand as well as deteriorating quality generates upward pressure on energy prices.  Although rising prices can make oil that is more difficult to extract profitable and hence market viable, the extraction and production of energy requires energy.  Once the EROEI for petroleum reaches 1 (i.e. whenever it takes one barrel of oil to produce a barrel of oil), petroleum will not be market viable, regardless of how expensive it becomes and regardless of how much petroleum remains in the ground.

Energy “De-Linking”

One of the most frequently cited arguments regarding energy use and economic growth is that economic growth relies increasingly less on energy consumption.  The ratio of energy use to GDP is called energy intensity, and it measures how much energy is required to generate 1 unit of GDP.  In Figure 73 I plot the inverse of this ratio, the GDP to Energy ratio, which measures the amount of GDP generated from 1 unit of energy, for the years 1976 to 2008.

Because GDP is a flow variable that includes expenditures acquired from borrowed income, I also include a measure of GDP that subtracts total dollars borrowed.[14] The trends are essentially the same.  GDP per unit of energy steadily rises.  In Figure 74, I control for both inflation and population by comparing the growth of per capita real GDP and per capita energy consumption.

Figure 73.  Ratio of GDP to Energy.  1976-2008

Figure 74.  Per capita GDP and Energy Use.  United States (1960-2008).[15]

Figure 75.  Global per capita GDP and energy use (1971-2008)

Figure 76.   Energy/Borrowing to Fixed Assets.  1976-2008

Beginning in the mid-1970s the clear growth rates of GDP and energy use per capita begin to diverge.  Indeed, the growth rate of use per capita seems to flat line.  World per capita GDP and Energy use are less striking, but still indicate that GDP growth is outpacing the growth of per capita energy consumption, as depicted in Figure 75.

Finally, I use an unconventional measure, plotting the ratio of energy use to borrowing (Energy/Borrowing) against the historical cost of net fixed private assets.  Private “fixed assets” is a stock variable that represents the accumulation of capital derived from investments.  Because this variable represents the sum total of historical capital accumulation less depreciation, it is much larger than GDP.  Figure 76 is intended to capture the relation between energy consumption and capital stock growth, holding borrowing constant.  Again, the data show a declining amount of energy required to sustain a unit increase in fixed assets.

Efficiency improvements have undoubtedly contributed to the decoupling of economic growth and energy consumption.  According to figures calculated by Ayres and Warr (2009), the ratio of useful to work to total exergy, or available energy, has risen from approximately 9 percent in 1960 to 13 percent by 2005.  This represents an efficiency improvement of about 4 percent.  Although this is significant, it does not explain the apparent de-coupling of growth rates for GDP and energy use per capita in the United States.  In particular, after 2005, oil production is nearly flat, but world GDP continues to grow at nearly 3.2 percent per year until 2008.

Two additional factors, in my view, explain the apparent decoupling of US GDP and energy: imports and debt.  First, the United States has not de-materialized its consumption.  The United States and, more generally, the global North, has become dependent on the global South for commodities and manufactured goods.   The growth of the North is therefore tied irrevocably to the South, and vice-versa.

Lopez (2009) observes that whereas for many decades the North could grow with constant or even declining commodity prices, this began to change in the 1970s.  The growth of the new industrial giants, such as China, India, Russia, and Brazil has generated increased demands for energy, food and other raw materials, which has helped to inflate their respective prices.  Off-shoring to China, for instance, has generated high growth rates in turn generating greater calls on global raw materials such as oil, cement, copper, steel, and petroleum (Schwartz 2009:  175).  Increased demand for raw materials has helped put an end to the global disinflation that helped fuel the housing bubble during the “long 1990s”, lasting approximately from 1996 to 2005 (Schwartz 2009).

Global demand for raw materials and energy eventually put an end to the disinflationary climate that prevailed from 1996 until 2005.  Inflation provoked a tighter monetary policy, which in the US began in the third quarter of 2004, triggering the end of the boom.   Lopez argues that stricter monetary policy reduces aggregate demand because of a highly concentrated income distribution.  Lopez concludes that:  “"The emergence of the NIG has meant that world economic growth has become more dependent on commodities at a time when commodity supply has become less elastic" (16).

Figure 77.  GDP to historical cost net private Fixed Assets.  1960-2008

Figure 78.   Ratio of personal consumption expenditure to historical cost net private fixed assets. 1960-2008

Because GDP is a flow variable in part dependent upon the accumulation of existing capital stock, it is useful to examine the relationship over time between the size of the flow relative to the capital stock that regulates it.  Figures 77 and 78 present the ratios of US GDP and personal consumption, respectively, to the estimated historical net cost of fixed private assets.  Data are estimated in billions of current dollars.  Data for personal consumption are taken from NIPA Table 1.1.5, and fixed assets from Fixed Asset table 2.3.  

These time series show a clear pattern of diminishing returns to fixed capital, which closely resembles the patterns for global oil production and the relationship between energy consumption and GDP explicated above.  If GDP were directly proportional to the stock of fixed assets, it would exhibit a pattern of exponential growth.  In the mid 1970s, however, this growth halts relative to capital stocks, and begins to reverse.  These data indicate that US growth since the 1980s has depended upon a drawing down of resources and capital stocks and infrastructure previously accumulated.


Can rising energy prices and/or peak petroleum production account for the global economic downturn?  The answer to this question depends on how the latter is defined.  The major stumbling block for those looking for a clear and direct link between energy production and global economic growth is that economic growth has, to some extent, decoupled from energy use.  This does not mean that growth does not require energy, for it clearly does, just as economic growth requires a suitable atmosphere to breath and solar radiation.  A presupposition, however, is not necessarily an element of a system.   The economic system operates according to the code:  to pay or not to pay.   This monetary code facilitates economic transactions and distinguishes what is relevant and irrelevant for the economic system.  A component of this de-coupling comes from improvements in the energy efficiency of production, a de-materialization of production in the United States, borrowing, and above all, a drawdown of existing capital stocks.

The theoretically relevant comparison, moreover, is not between absolute levels of energy use, but the relative rates of growth between energy use, energy efficiency, and total debt.  Figure 79 is a time-series illustrating the relative growth rates between total debt, the ratio of GDP to energy, and fixed assets.  The data show that the growth rate of debt outpaced the growth of energy efficiency.  A higher rate of delinking or a greater quantity of energy production (at prevailing prices) could have sustained this debt.  An over-accumulation of debt is always relative to rates of growth.  In order to demonstrate how a debt-crisis occurs, it is necessary to show both the run up of debt, representing anticipations of future growth, and also the factors that contributed to the negation of these anticipations.

That the modern economic system (aka capitalism) requires no one-to-one correlation between commodity production and material production can also be demonstrated with reference to its history.  Contrary to what is generally thought, the first commodities to supplant subsistence activities were in fact services, not material goods.

Figure 79.  GDP/Energy, Fixed Assets, and Debt Growth (1976=1)

Ivan Illich details, for instance, the switch from un-taught vernacular languages to a single, national ‘mother-tongue’ in the late 15th century.  According to Illich, “the switch from the vernacular to an officially taught mother tongue is perhaps the most significant- and, therefore, least researched- event in the coming of a commodity-intensive society [foreshadowing] the switch from . . . production for use to production for market” (1981: 44).  The explicit purpose of this pedagogy, as detailed by the writings of Elio Antonio de Nebrija in Spain, was the creation of a new type of dependence.  Contrary to the belief that a unified national language is necessary for literacy, the purpose of the proposed standardization of language, was instead to suppress untaught vernacular reading (1981: 40).  This, according to Illich, was the first shot fired in the war against subsistence.[16]

In my view, the differentiation of needs co-occurs but does not necessarily co-relate to the differentiation of the codification of the means for distributing these needs.  The monetary code, regulating the distribution of the desires (i.e. the meanings) of the socius, must therefore be analytically distinguished from the organizations that regulate and distribute its material and energetic flows.  The lack of a direct correspondence between economic growth and energy, however, does not mean that there are no causal relations between them.

[1]  The BEA’s depiction has an added advantage of realism.  In most economic textbooks (e.g. Mankiw 2010), households are depicted as providing firms not only labor, but also land and capital.  Capital is an ambiguous term, but when distinguished from mere money, it means physical equipment, such as factories.  Yet it is difficult to understand how households could provide firms with factors of production other than labor.  It is necessary, however, in neoclassical theory to depict “capital” as a separate factor of production, so that profits are derived from the contribution of capital rather than from the exploitation of labor.
[2]  According to Amazon.com, Mankiw’s macroeconomics textbook is the “ #1 bestselling intermediate macroeconomics text” in the United States, beating out Paul Krugman, who ranks 3rd.
[3]  This is named after Alfred Jevons, who in the 1800s observed that efficiency improvements in energy extraction from coal coincided with greater absolute levels of coal extraction.  A modern day example of the Jevons Paradox is fuel efficiency.  Greater fuel efficiency for automobile has not diminished the demand for gasoline.  On the contrary, people tend to drive their cars greater distances the more fuel efficient their cars are.  In addition, more and more cars are being produced and increasingly more drivers are on the road due to population and economic growth.
[4]  For example, both the production of asbestos and the subsequent medical care required by victims of asbestosis, a lung disease resulting from asbestos exposure, positively affected GDP in the United States.
[5]  The concept of entropy has also been employed and given a different but related meaning in probability terms by Claude Shannon (1948) as a measure of information. Entropy is a measure of the likelihood or probability of a “message” given a set of possible messages. In this theoretical context, entropy is identified as the inverse of probability or certainty: the more surprising or unlikely something is, the greater its entropy, or information value.   Information is the inverse of the probability:   “the less probable a message is, the more meaning it carries” (Wiener 1950: 8 in Marcus 1997: 27)
[6] It should be mentioned that determining a causal relationship between economic growth and energy use by econometric means is not helpful.  The two series vary concurrently.  Granger tests are therefore not meaningful in this case (Wooldridge 2006).
[7]  Note that these prices are in nominal dollars and do not take into account inflation.
[8] I calculate this as follows:  ((Q1974-Q1960)/Q1960)/((P1974-P1960)/P1960), where Q and P are quantity and price, respectively, and where 1974 and 1960 numbers indicate the years.
[9]  Estimates available at http://www.theoildrum.com/node/3810  
[10]  Production of nonrenewable resources tends to follow more closely the cost of production curve depicted by neoclassical economic theory than do normal commodities, which, contrary to neoclassical theory, tend to have flat supply curves.
[11]  Relevant reports are available at http://www.energywatchgroup.org/ and http://www.tsl.uu.se/uhdsg/
[12] Reserves are not equal to resources.  Resources refer to the total amount of coal present, whereas reserves refer to the portion of the resource expected to be profitably extracted.
[13]  Price elasticity of demand (PED) is calculated as the ratio of the percentage change in quantity to the percentage change in price:  %∆Q/%∆P.  A good is inelastic whenever its PED is below 1, which means that a 1 percent change in price is accompanied by a smaller percentage change (<1%) in the quantity demanded.
[14]  Gross domestic product data are in current (not constant) dollars.  They represent total GDP rather than per capita figures, which will be used below.  Data are taken from NIPA Table 1.1.5.   The second series is calculated as:  (GDP-Borrowing)/Energy.  Borrowing data are collected from the Federal Reserve’s Flow of Funds Accounts.   Energy data are acquired from the World Bank and are measured as kt of oil equivalent.
[15] Data are taken from the World Bank’s World Development Indicators (WDI) database.   Gross Domestic Product is measured at constant 2000 US dollars.  Energy use is measured as kg of oil equivalent per capita.  The WDI database can be located online at:  http://data.worldbank.org/data-catalog
[16]  While militating against a productivist, orthodox reading of Marx, Illich’s views, I think, are consonant with the idea, that the process of commodification is one which “reproduces the needy individual” (Marx 1977: 719).
Subpages (1): Energy Resources