Is it really possible to decouple GDP Growth from Energy Growth?

In recent years, we have heard statements indicating that it is possible to decouple GDP growth from energy growth. I have been looking at the relationship between world GDP and world energy use and am becoming increasingly skeptical that such a decoupling is really possible.

Figure 1. Growth in world energy consumption (based on BP data) and growth in world real GDP

Prior to 2000, world real GDP (based on USDA Economic Research Institute data) was indeed growing faster than energy use, as measured by BP Statistical Data. Between 1980 and 2000, world real GDP growth averaged a little under 3% per year, and world energy growth averaged a little under 2% per year,  so GDP growth increased about 1% more per year than energy use. Since 2000, energy use has grown approximately as fast as world real GDP–increases for both have averaged about 2.5% per year growth. This is not what we have been told to expect.

Why should this “efficiency gain” go away after 2000? Many economists are concerned about energy intensity of GDP and like to publicize the fact that for their country, GDP is rising faster than energy consumption. These indications can be deceiving, however. It is easy to reduce the energy intensity of GDP for an individual country by moving the more energy-intensive manufacturing to a country with higher energy intensity of GDP.

What happens when this shell game is over? In total, is the growth in world GDP any less energy intense? The answer since 2000 seems to be “No”.

It seems to me that at least part of the issue is declining energy return on energy invested (EROI)–we are using an increasing share of energy consumption just to extract and process the energy we use–for example, in “fracking” and in deep water drilling. This higher energy cost is acting to offset efficiency gains. But there are other issues as well, which I will discuss in this post.

If GDP growth and energy use are closely tied, it will be even more difficult to meet CO2 emission goals than most have expected. Without huge efficiency savings, a reduction in emissions (say, 80% by 2050) is likely to require a similar percentage reduction in world GDP. Because of the huge disparity in real GDP between the developed nations and the developing nations, the majority of this GDP reduction would likely need to come from developed nations. It is difficult to see this happening without economic collapse.

Real GDP Growth and Energy Growth for Several Countries

I started this analysis by looking at trends (1) in real GDP and (2) in total energy consumption for a number of countries, and was struck by how different the patterns appeared.

Figure 2. USA's real GDP and total energy consumption

Up until 2005, the USA was able to increase real GDP by 3% per year, while increasing energy use by only 1% per year. The 2% savings would seem to come from some combination of offshoring and energy efficiency. Since 2005, the relationship between GDP growth and energy growth has been closer.

Figure 3 - Germany's total energy consumption and real GDP.

Germany is another example with much higher GDP growth than growth in energy use. Between 1980 and 2005, energy use was close to flat, while GDP rose by an average of 1.7% per year.

Figure 4. Italy's energy consumption and real GDP

Italy’s real GDP grew by about 1.7% a year between 1980 and 2005, while its oil use grew about 1.0% per year, so it too shows energy efficiency/offshoring gains, amounting to about 0.7% per year. This gain is less than that of the USA and Germany, but Italy was also less industrial to start with, so offshoring was less of an option.

Figure 5. Japan's energy consumption and real GDP

The tie between energy consumption and GDP growth has been much tighter for Japan, especially since about 1987. Since 1987, the two have grown at about the same rate.

Figure 6. Spain's energy consumption and real GDP

Spain has also showed a very close tie between growth in energy consumption and growth in real GDP, with both growing by close to 3% per year between 1980 and 2005.

Figure 7. Greece's energy consumption and real GDP

Greece managed to grow its energy use faster than real GDP during most of the 1980 to 2005 period. This may contribute to its current economic problems.

Figure 8. China's energy consumption and real GDP.

China shows more rapid growth in real GDP than in energy consumption. Its real GDP grew by about 10% a year between 1980 and 2005, while energy use grew by a little less than 6% per year. Between 2005 and 2010, real GDP continued to grow by about 10% per year, while energy use grew by about 7.5% per year.  The country has been changing so rapidly that a person wonders how accurate the early GDP numbers are.

Figure 9. Energy consumption and real GDP for the Former Soviet Union

Figure 9 shows that the pattern for the Former Soviet Union (FSU) has been very unusual. Both energy use and real GDP collapsed after the collapse of the Soviet Union, but the drop in GDP was greater than the drop in energy use. Recently, real GDP has been soaring while energy use remains flat, suggesting that outsourcing of high energy manufacturing is occurring, or that new sources of GDP that do not require much energy use are being created.

Changes in Energy Intensity

The usual way of measuring energy intensity is as the ratio of energy consumed to real GDP (the red line divided by the blue line in the above graphs), and these ratios vary greatly. In fact, it is hard to even put energy intensities for different countries on the same graph, because the amounts are so different.

Figure 10a. Energy intensities for the world, the Former Soviet Union, and China

Figure 10b. Energy intensities for the world minus the Former Soviet Union, the USA, Japan, and the EU-15 countries

It seems to me that what we are really interested in is the energy intensity of the world, (or perhaps of the world less the Former Soviet Union, if Former Soviet Union data is totally bizarre, reflecting very high energy usage in the past that is now disappearing, and cannot be replicated elsewhere). We show these in Figure 11:

Figure 11. Historical energy intensities for the world, and for the world excluding Former Soviet Union

Figure 11 shows that energy intensity on a world basis has been flat since 2000. This is also the case when FSU data is excluded. We expected recent world energy intensity to be flat, based on Figure 1 at the top of the page.

Why does world energy intensity remain flat, while energy intensity for many individual countries has been decreasing?

We are dealing with a large number of countries with very different energy intensities. The big issue would seem to be outsourcing of heavy manufacturing. This makes the energy intensity of the country losing the manufacturing look better. Outsourcing transfers manufacturing to a country with a much higher energy intensity, so even with the new manufacturing, its ratio can still look better (lower). It is hard to measure the overall impact of outsourcing, except by looking at world total energy intensities rather than individual country amounts.

Looking at world energy intensities, it appears that the huge amount of outsourcing is resulting in pretty much comparable energy use to the original energy that was outsourced. It is hard to make a direct calculation of the difference in energy use, because much of the new energy use is indirect. For example, the government of the developing nation may build huge amounts of new paved roads and concrete homes with tax revenues, and individual workers may buy new cars with their salaries. These amounts are not captured in a simple comparison of the energy used in making a widget in the USA relative to the energy used in making a widget in China, for example.

Another issue is that the energy use of interest is per dollar of real GDP, and a savings in energy that results in a cost savings may not be very helpful in lowering energy intensity of GDP. For example, suppose that a manufacturer creates a new, smaller car, that is 20% cheaper and uses 20% less gasoline on an ongoing basis. More workers will be able to afford this car. Furthermore, a well-off worker who can afford this new cheaper car (and who could also have afforded a more expensive car) will have left-over money. With this left-over money, the well-off worker can purchase something else, such as an airline trip, food flown in from overseas, or a new iPod. All of these extra purchases take energy as well. So when the overall picture is viewed, the fact that more energy-efficient cars are being manufactured does not necessarily translate to lower energy intensity of GDP.

One issue mentioned in the introduction to this post is the fact that EROI for fossil fuels is declining because the easy-to-extract fossil fuels have mostly been extracted. As a result, we are now extracting the more difficult to extract fossil fuels, requiring more energy.

A similar situation occurs in many other endeavors, because we live in a finite world, and we are reaching limits. In mining, the quality of ores is getting poorer, meaning than more energy needs to be used in extraction. In farming, we are stretching our resources tighter, requiring more fertilizer, pesticides, and more irrigation, all requiring energy. We are running short of fresh water in some places, so water is pumped from greater distances or desalination is used, adding to energy usage. Pollution is an issue, so we require utilities to add scrubbers to old coal plants. All of these efforts require energy, and likely contribute to an upward trend in energy usage, offsetting efficiency savings elsewhere.

Another issue that tends to raise energy intensity of GDP is the long-term trend toward using machines and additional energy to do jobs, rather than simple human labor. For example, if a person chops down a few trees and builds his own house, most calculations would say that there is neither GDP nor (outside) energy used. If a person hires a builder to build a house, and the builder uses hand tools to chop down trees and human labor to build the house, the result is an increase in GDP, but little fossil fuel energy use. If the builder becomes more “modern” and uses earth movers and concrete to build homes, then energy use rises relative to GDP created.

Carbon Dioxide Emissions

As might be expected, carbon dioxide emissions per unit of GDP are closely related to energy intensity. In fact,

(Co2 Emissions/GDP ) = (CO2 Emissions / Energy Used) x (Energy Used / GDP)

The ratio (Energy Used /GDP) is simply energy intensity, which was graphed in Figures 10a, 10b, and 11. The other ratio is (CO2 Emissions / Energy Used) graphed below in Figure 12. It shows a similar pattern: declining prior to 2000, and then leveling.

Figure 12 - Carbon dioxide emissions per barrel of oil equivalent energy, based on BP Statistical Data.

The ratios in Figure 12 reflect changes in energy mix over time, and their relative propensity to generate CO2. Since 2000 these emission per unit of energy use have, in fact, started to rise a little, because of the greater use of coal in the energy mix. The CO2 measurements used in this analysis are BP’s calculations, based on the types of energy used each year (including renewables*). They do not reflect actual measured CO2 in the atmosphere.

The ratio of new CO2 emissions to Real GDP reflects a combination of these ratios (C02/Energy and Energy/Real GDP) and is shown in Figure 13.

Figure 13: Ratio of New Energy Carbon Dioxide Emissions to Real GDP

Figure 13 indicates what we would expect from Figures 11 and 12: A declining ratio of CO2 emissions to real GDP until about 2000, then fairly flat thereafter. In fact, there is a distinct upturn in 2010. Thus new CO2 emissions from energy sources have been rising about as fast as real GDP since about 2000, and a little faster than real GDP in 2010. This is no doubt discouraging news to those who adopted the Kyoto Protocol in 1997, thinking it would reduce CO2 emissions.

A Few Thoughts on Energy Policies

Carbon taxes and cap and trade policies seem to encourage outsourcing of manufacturing. The primary benefits of outsourcing would seem to be (1) a reduction in imported fossil fuels, (2) lower cost of manufactured goods to the consumer, because of lower labor costs, and (3) possibly higher profits to the company selling the new cheaper product. Offsetting these benefits are a loss of jobs for the country doing the outsourcing and a loss of control over what types of energy are used in the manufacturing process. It seems to me that we would be better off not encouraging this outsourcing, especially when essential goods are being created.

One misconception that seems to guide much energy policy is the view the biofuels will substitute for oil, and that use of additional electricity use will substitute for oil. Oil use is pretty much maxed out. Oil supply is very close of inelastic, regardless of price. Someone, somewhere, will use any oil pulled out of the ground, perhaps at a slightly lower price, even if a particular country can reduce its oil consumption through the use of biofuels, or if a car can run on electricity.

This means that any biofuel that is created will add to world energy supply, by using natural gas and coal supplies more quickly, since their use is still somewhat elastic. Similarly, by moving energy demand from oil to electricity, we what we are really doing is expanding total energy usage, by burning more coal and natural gas to make more electricity.

Thus, from a world CO2 perspective, biofuels and increased electrical usage are not helpful. Individual countries may still find biofuels and expanded electrical use helpful, because they can reduce oil imports, if oil use can be shifted to another country. There is also the hope that we can continue our motoring  lifestyle longer, using electric cars.

If our intent is really to reduce CO2 emissions, it seems to me that we need to look much more broadly at the issue. Maybe the issue should be viewed in terms of (1) fossil fuel resources that we are willing to use in each future year, and (2) how much real GDP can be created from those resources, given the issues we are facing. The quantity of fossil fuels to be used each future year might consider CO2 goals as well as limits on the amount of oil that can be extracted each year because the “easy oil is gone”. The amount of real GDP that can be created from these fuels would depend on a number of factors, including declining EROI and increasing efficiency.

If the plan is to reduce fossil fuel consumption, then we may very well be expecting real GDP to also decrease, perhaps by a similar percentage. In fact, looking at the experience of FSU in Figure 9, the GDP decline may even be greater than the energy decline.


We are facing a challenging time. This post seems to suggest that there is yet another story that we are being told, that isn’t quite true. It seems to me that we need to examine the issues ourselves, come to our own conclusions, and start telling the real story.

*I have not attempted to discuss the impact of renewables, since to date their impact has been small. The front-ending of energy use of renewable makes their impact on energy intensity of GDP less beneficial than standard comparisons would suggest.

This entry was posted in Energy policy, Financial Implications and tagged , , , by Gail Tverberg. Bookmark the permalink.

About Gail Tverberg

My name is Gail Tverberg. I am an actuary interested in finite world issues - oil depletion, natural gas depletion, water shortages, and climate change. Oil limits look very different from what most expect, with high prices leading to recession, and low prices leading to financial problems for oil producers and for oil exporting countries. We are really dealing with a physics problem that affects many parts of the economy at once, including wages and the financial system. I try to look at the overall problem.

48 thoughts on “Is it really possible to decouple GDP Growth from Energy Growth?

  1. Gail, very nice article, although I am a bit confused by your Figure 12.

    Right above the figure, you wrote, “if we look at carbon dioxide emissions per unit of GDP, they show a similar pattern to energy intensity.” But the caption to Figure 12 itself is described as “Carbon dioxide emissions per barrel of oil equivalent energy.” So you are not actually showing carbon dioxide emissions per unit of GDP in Figure 12. Was that your intent?

    Recently, I pointed out that there seems to be a fundamental difference between how one’s view of the future is shaped by whether or not has a “PO” versus “AGW” view ( and how these two views seem to be dissonant with each other (

    In view of this, I have two, perhaps leading, questions:

    If carbon dioxide emissions are highly correlated with fossil fuel use, and, if the rates of fossil fuel use are, or will soon, peak and decline in the coming decades, then won’t carbon dioxide emissions (and along with it, GDP) also decline, no matter what we do? And, aren’t we just kidding ourselves to think that “we” can really do anything to affect this trend?

    • That CO2 will decline as fossil fuels decline is a point that Aleklett at Uppsalla University in Sewden has made. He says the UN model assumes exponential increase all the way to 2100 and there is simple not enough fossil fuel to meet that assumption.

      Though I am concerned by an article in TOD that claims 5000 gigatons of coal in Alaska.

    • You are right–the-write up about Figure 12 is not correct. I went back and made some changes. Now the text says:

      As might be expected, carbon dioxide emissions per unit of GDP are closely related to energy intensity. In fact,

      (Co2 Emissions/GDP ) = (CO2 Emissions / Energy Used) x (Energy Used / GDP)

      The ratio (Energy Used /GDP) is simply energy intensity, which was graphed in Figures 10a, 10b, and 11. The other ratio is (CO2 Emissions / Energy Used) graphed below in Figure 12. It shows a similar pattern: declining prior to 2000, and then leveling.

      (Figure 12)

      The ratios in Figure 12 reflect changes in energy mix over time, and their relative propensity to generate CO2. Since 2000 these emission per unit of energy use have, in fact, started to rise a little, because of the greater use of coal in the energy mix.

      Sorry about that! I think I was too anxious to get finished, and stopped thinking very clearly.

      I didn’t explain why the two ratios should shift from downward to flat at the same time. It may be that the big push to coal in China and India affected both, and these in turn were related to the same other issues–reaching limits in many things, including oil EROI.

      Edit: I also added a Figure 13, that looks a lot like Figure 11 and Figure 12, showing the ratio of CO2 emissions to Real GDP. You might check the post.


      If carbon dioxide emissions are highly correlated with fossil fuel use, and, if the rates of fossil fuel use are, or will soon, peak and decline in the coming decades, then won’t carbon dioxide emissions (and along with it, GDP) also decline, no matter what we do? And, aren’t we just kidding ourselves to think that “we” can really do anything to affect this trend?

      I think this has been my view as well. But I didn’t really thing I could put it that way in this post.

  2. Pingback: Casa Food Shed » Blog Archive » EIA projections: numbers don’t add up

  3. Gail:
    The Fed’s have been trying to decouple GDP growth and energy growth by quantitative easing or printing money. It will only work for awhile and it’s probably not working anymore. Martensen has a full explanation energy and GDP growth in Chapter 16 of his book “The Crash Course”. It confirms what you are saying or visa versa.

  4. Perhaps I missed it, but could some of the discrepancies you note be the result of financial capital effects on GDP?

    For example, banks “loan money into existence,” but that is really borrowing from the future, with no energy backing the new money. So this “new” money contributes to GDP, but there’s no “there” there — it’s out of thin air. When the piper comes a callin’ (as in 2008) then GDP “crashes” back to sync with energy.

    • The growth of the financial industry, medicine, education, and entertainment has helped reduce the energy intensity of the United States’ economy. And the all of the debt has helped build demand for these services. I expect we will see a lot of debt defaults, as well as a crash in the discretionary section of the economy. So, directly or indirectly, I think you are right.

      • The real things require energy food, housing, heating, lighting, transportation.
        The services that we can live without are low energy entertainment, education, medicine, finance.

        • Ed and Gail have a good point. I would like to see the energy intensity history of resource extraction and material goods manufacture/distribution rather than GDP, which can include all sorts of low energy fluff.

          Concentrate on a few key parameters: steel, concrete, lumber, water, food staples; all as delivered to the end user. Use mass quantities rather than dollar value. This would really tell us how energy “efficient” we have become.

          • There would still be a “mix” issue. Suppose wood is not energy intensive at all, but concrete is. If a country switches from wood to concrete buildings (as China seems to have) under these circumstances, the energy intensity would go way up.

  5. For most of us (99%), maybe the question should be “Does quality of life have anything to do with GDP and energy growth or can quality of life improve while GDP and energy growth decrease ?”.

    Then again, maybe GDP and energy growth will just destroy all of human life by global climate change or just some of it like the Iraq war, coal miners, chernobyl and air pollution, etc.

    I would prefer quality over quantity (GDP and Energy growth). I’m going to go for a bike ride now because I don’t need a job to pay for a car, gasoline, insurance and auto repairs. What are you doing ?

  6. @DownToTheLastCookie, ran an article about the relationship between energy and happiness back in 2006. The author’s results agreed with work done by Vaclav Smil: past 100 MJ/capita, happiness doesn’t increase very much.

    You might also be interested in Ivan Illich’s essay “Energy and Equity.” At the risk of oversimplification, he argued the real energy crisis of the 1970s was that we were consuming too much energy per capita (he uses Buckminster Fuller’s concept of “energy slaves” to great effect and discusses the wonders of the bicycle).

  7. Could it be that the recoupling of the GDP to energy use is that after March 2000 when the dot-com bubble crash, a significant part of the GDP growth after 2000 is due to the real estate booming period. Thus before 2000, the energy intensity was able to decrease significantly because the tech boom does not require much energy but after the crash and the FED and the whole world starts on the real estate boom which requires large amount of energy, the GDP growth is thus directly dependent on the energy group.

    • I suppose that more emphasis on real estate after 2000 could have raised energy intensity (although, as calculated, the flattening in US energy intensity happened more at the end of the decade). China has certainly put a big emphasis on real estate development in the last few years. I am not certain about the rest of the world–I would expect that Europe and Japan probably have not seem as much real estate growth (except in prices), because most lots have already been built.

  8. Pingback: Is It Really Possible To Decouple GDP Growth From Energy Growth? | Bailout and Financial Crisis News

  9. Pingback: Is It Really Possible to Decouple GDP Growth from Energy Growth? | Crude Oil News

Comments are closed.