US GDP, Energy Consumption and CO2 Emissions

A review of the structure of US GDP, imports and exports shows that none of these variables has contributed to the fall in US CO2 emissions post-2008 finance crash. The main contributions to reduced CO2 come from high energy prices and recession (36%), gas substitution for coal (20%) and growth in wind and solar (15%) which more or less corroborates the findings of Roger Andrew’s in his recent post on this topic.

At the review stage of his recent post The causes of the recent decrease in US greenhouse gas emissions, I suggested to Roger that he may wish to look into US exports and imports and the changing shape of US GDP as possible additional causes for the recent fall in US CO2 emissions. Roger suggested that maybe I could do this 😉 We have heard a lot recently about offshoring US industry, and the need to account for CO2 embedded in traded goods and so I decided to have a look.

US GDP, Imports and Exports

Figure 1 The makeup of US GDP according to  UN Statistics. The description of the category codes are given in the Appendix. The main feature is the strong recovery post-2008 crash, although the rate of growth has slowed. To make trends easier to see, the data are plotted to 100% in Figure 2.

Figure 2 The data from Figure 1 plotted as %.

There are a number of interesting and key observations to be made from Figure 2. 1) “Services” that includes things  like government, healthcare, education, defence and finance accounts for over 50% of US GDP but the proportion has changed little over the 45 year period, 2) transport etc and retail etc have both expanded steadily. The consumption part of the economy has grown and now accounts for 25%; 3) Construction has declined steadily and now accounts for only 4% (Figure 3); 4) manufacturing accounted for 13% of the economy in 2015 (Figure 3) and has changed little over 45 years, which is difficult to reconcile with the rhetoric emanating from the Trump camp. One must presume that manufacturing jobs lost in the rust belt have been replaced in other parts of the country; 5) Mining etc appears to have contracted with time, with perhaps a minor expansion in the last decade marking higher energy prices and the shale boom; and finally 6) agriculture etc, comprises only 1% of the US economy which I find to be astonishingly low.

There are two further points to make. First, I can see nothing in these data to account for the decline in CO2 emissions since 2008. And second, agriculture accounts for around 7% of US emissions, but only 1% of the economy.

Figure 3 The makeup of US GDP in so15.

If the slowly changing face of the US economy does not seem to account for the change in CO2 emissions then perhaps a change in the volume of imports does?

Figure 4 US imports and exports were roughly in balance until 1996, but since then they have diverged with imports running well ahead of exports.

Both imports and exports fell in 2008 as global trade shrank. In 2015, imports were running at $2.5 trillion out of a total economy of $14.6 trillion (17%). More significant is the observation that US manufacturing was $1.8 trillion in 2015 and is dwarfed by imports.

Figure 5 Looking at the difference between imports and exports we see that this actually contracted post-2008 crash.

The idea that the energy and CO2 embedded in traded goods should be accounted in estimating a country’s emissions is often cited, and while this is undoubtedly true, there is no evidence from the US GDP data that this has anything to do with the fall in CO2 emissions post-2008. If anything, US emissions should be adjusted upwards from 1996 which is when the trade gap began to yawn. This has been somewhat a wild goose chase (perhaps Roger already knew 😉 but I hope the data have been enlightening none the less.

So if not the structure of GDP, what else can we look at? Energy consumption and its structure is the obvious choice.

Primary Energy Consumption

Figure 6 Total primary energy consumption fell post-2008 crash and has remained flat ever since.

The fall in primary energy consumption is a principal cause for the decline in CO2 emissions. But what caused energy consumption to fall? We can see that the rising trend began to flatten in the noughties as energy prices began to rise. And this brings us to the conundrum of whether the crash was caused by high energy prices or defaults on sub-prime mortgages. Perhaps both were caused by reckless expansion of credit? A notable feature is the decline in coal production with expansion of gas production flowing from the shale boom (Figure 7). I’m unsure if the substitution of coal by gas came about because of the shale boom or because of Obama environment policy.

Figure 7 Looking at the percentages shows more clearly how coal has fallen out of favour with the expansion of gas.

Figure 8 Just to ram this point home, post-2008, coal consumption has declined and gas consumption has risen.

As Roger documented, substitution of coal by gas is a primary reason for the fall in US CO2 emissions. There are two reasons for this. First, gas burned in a CCGT is far more energy efficient than coal fired electricity generation. Roughy 50% versus 35%. Second, in methane it is C-H bonds that yield energy producing CO2+H2O while in coal it is C-C bonds that yields CO2+CO2. Combined, this results in gas producing roughly half of the CO2 produced by coal per unit of electricity generated.

In rough terms between 2008 and 2015, 168 Mtoe of coal was substituted by 113 Mtoe gas for a net saving of 55 Mtoe.

From Figure 8, we also need to note that oil consumption dropped by roughly 100 Mtoe post-2008 and has stayed down. It is this drop in oil consumption that accounts for the fall in primary energy consumption and is major factor that contributes to the fall in US CO2 emissions.  In the 1970s, the drop in oil consumption was mainly due to the closure of oil-fired electricity generation (Figure 9). Figure 9 shows that fuel oil consumption has declined steadily since the mid-80s, and there is a fall of about 500,000 bbls/day post 2005 suggesting that a further reduction in fuel oil consumption for electricity generation and home heat may account for 25 Mtoe.

Figure 9 CO2 emissions/mile and fuel economy for US cars and trucks, 1975-2016

That leaves transport and the low hanging fruit of improving vehicle efficiency in a land of gas guzzlers. And that brings us back to the fuel economy charts posted by Roger (Figures 10 and 11).

Figure 10: US deliveries of gasoline and diesel fuel and total miles traveled, 1984-2015

Figure 11 CO2 emissions/mile and fuel economy for US cars and trucks, 1975-2016

Figure 12 US oil and biofuel consumption. The fall in oil consumption can be divided into two parts. 100 Mtoe that is below-trend can be allocated to the recession and a further 90 Mtoe to fuel efficiency in vehicles and further substitution of fuel oil with other energy sources.

Figure 12 provides an interpretation of US oil consumption . Had it continued on trend it would arguably be 190 Mtoe higher than it is now. Figure 10 shows that vehicle miles travelled have returned to pre-2008 crash levels and all other things being equal we might have expected oil consumption to have returned to the same pre-crash level. We can allocate 100 Mtoe savings to Americans driving less and put that down to high energy prices and the recession. In 2015, the USA consumed 3.34 billion bbls of gasoline, and looking at Figure 10 I estimate 1.48 billion bbls of diesel. Looking at Figure 11 I estimate efficiency improvements of 4% 2008-2015. I’m unsure if that is a figure that can be applied to the whole vehicle fleet, but that’s what I’m going to do. 4% of 4.82 billion bbls (gasoline + diesel) = 193 million bbls or 26 Mtoe.

The remaining 90 Mtoe deficit (Figure 12) can therefore be accounted for as follows: 25 Mtoe on less fuel oil used; 26 Mtoe on improved energy efficiency of cars leaving 29 Mtoe unaccounted for that must surely lie in aviation and shipping.

Growth in bio-fuels 2008-2015 was 8.4 Mtoe. I’m unsure how to handle this since the ERoEI for corn ethanol is close to unity and any energy savings imagined in the vehicle fleet is matched by energy used to produce the biofuel.

This leaves CO2 savings from wind and solar combined which have increased by 40 Mtoe 2008-2015.

Summary and Conclusions

There is no evidence to support the notion that the changing face of US GDP may have contributed to the fall in CO2 emissions post-2008 finance crash. Nor is there evidence to support the notion that offshoring of manufacturing with attendant increase of imports with embedded energy and CO2 emissions has contributed to the fall in US emissions. In fact, to the contrary, US imports have fallen. These observations are difficult to reconcile with rhetoric from the recent US presidential race.

In his recent post, Roger summarised CO2 emissions reductions in the USA as follows:

  • Gas replacing coal in electricity generation:     40%
  • Decrease in gm/mile vehicle CO2 emissions:   30%
  • Growth in low-carbon renewables generation: 30%

My estimates are as given below:

  • Gas replacing coal 55 Mtoe – 20%
  • Fall in oil consumption owing to high price and recession 100 Mtoe – 36%
  • Fall in fuel oil consumption 25 Mtoe – 9%
  • Improved vehicle efficiency 26 Mtoe – 9.5%
  • Aviation, shipping and unallocated oil savings 29 Mtoe – 10.5%
  • [Biofuels 8.4 Mtoe]
  • Growth in wind and solar 40 Mtoe – 15%

Total excluding biofuels = 275 Mtoe

Assuming CH2 as a general formula for oil that translates to 931 MtCO2 which is darned close to Roger’s estimate of 1 billion tCO2 that has been saved in the USA.

The main material difference between my estimates and Roger’s is that I allocate 36% savings to reduced oil use as a result of high price and the recession.


Codes applied to the various categories of economic activity shown in Figure 1. Source is UNstats.

  • A – Agriculture, hunting and forestry
    • 01 – Agriculture, hunting and related service activities
    • 02 – Forestry, logging and related service activities
  • B – Fishing
    • 05 – Fishing, aquaculture and service activities incidental to fishing
  • C – Mining and quarrying
    • 10 – Mining of coal and lignite; extraction of peat
    • 11 – Extraction of crude petroleum and natural gas; service activities incidental to oil and gas extraction, excluding surveying
    • 12 – Mining of uranium and thorium ores
    • 13 – Mining of metal ores
    • 14 – Other mining and quarrying
  • D – Manufacturing
    • 15 – Manufacture of food products and beverages
    • 16 – Manufacture of tobacco products
    • 17 – Manufacture of textiles
    • 18 – Manufacture of wearing apparel; dressing and dyeing of fur
    • 19 – Tanning and dressing of leather; manufacture of luggage, handbags, saddlery, harness and footwear
    • 20 – Manufacture of wood and of products of wood and cork, except furniture; manufacture of articles of straw and plaiting materials
    • 21 – Manufacture of paper and paper products
    • 22 – Publishing, printing and reproduction of recorded media
    • 23 – Manufacture of coke, refined petroleum products and nuclear fuel
    • 24 – Manufacture of chemicals and chemical products
    • 25 – Manufacture of rubber and plastics products
    • 26 – Manufacture of other non-metallic mineral products
    • 27 – Manufacture of basic metals
    • 28 – Manufacture of fabricated metal products, except machinery and equipment
    • 29 – Manufacture of machinery and equipment n.e.c.
    • 30 – Manufacture of office, accounting and computing machinery
    • 31 – Manufacture of electrical machinery and apparatus n.e.c.
    • 32 – Manufacture of radio, television and communication equipment and apparatus
    • 33 – Manufacture of medical, precision and optical instruments, watches and clocks
    • 34 – Manufacture of motor vehicles, trailers and semi-trailers
    • 35 – Manufacture of other transport equipment
    • 36 – Manufacture of furniture; manufacturing n.e.c.
    • 37 – Recycling
  • E – Electricity, gas and water supply
    • 40 – Electricity, gas, steam and hot water supply
    • 41 – Collection, purification and distribution of water
  • F – Construction
    • 45 – Construction
  • G – Wholesale and retail trade; repair of motor vehicles, motorcycles and personal and household goods
    • 50 – Sale, maintenance and repair of motor vehicles and motorcycles; retail sale of automotive fuel
    • 51 – Wholesale trade and commission trade, except of motor vehicles and motorcycles
    • 52 – Retail trade, except of motor vehicles and motorcycles; repair of personal and household goods
  • H – Hotels and restaurants
    • 55 – Hotels and restaurants
  • I – Transport, storage and communications
    • 60 – Land transport; transport via pipelines
    • 61 – Water transport
    • 62 – Air transport
    • 63 – Supporting and auxiliary transport activities; activities of travel agencies
    • 64 – Post and telecommunications
  • J – Financial intermediation
    • 65 – Financial intermediation, except insurance and pension funding
    • 66 – Insurance and pension funding, except compulsory social security
    • 67 – Activities auxiliary to financial intermediation
  • K – Real estate, renting and business activities
    • 70 – Real estate activities
    • 71 – Renting of machinery and equipment without operator and of personal and household goods
    • 72 – Computer and related activities
    • 73 – Research and development
    • 74 – Other business activities
  • L – Public administration and defence; compulsory social security
    • 75 – Public administration and defence; compulsory social security
  • M – Education
    • 80 – Education
  • N – Health and social work
    • 85 – Health and social work
  • O – Other community, social and personal service activities
    • 90 – Sewage and refuse disposal, sanitation and similar activities
    • 91 – Activities of membership organizations n.e.c.
    • 92 – Recreational, cultural and sporting activities
    • 93 – Other service activities
  • P – Activities of private households as employers and undifferentiated production activities of private households
    • 95 – Activities of private households as employers of domestic staff
    • 96 – Undifferentiated goods-producing activities of private households for own use
    • 97 – Undifferentiated service-producing activities of private households for own use
This entry was posted in Energy and tagged , , , , , , . Bookmark the permalink.

15 Responses to US GDP, Energy Consumption and CO2 Emissions

  1. Lars says:

    “Nor is there evidence to support the notion that offshoring of manufacturing with attendant increase of imports with embedded energy and CO2 emissions has contributed to the fall in US emissions. In fact, to the contrary, US imports have fallen. These observations are difficult to reconcile with rhetoric from the recent US presidential race.”

    True, US imports have fallen but only if you compare to the peak level around 2006. Going further back to around 1996 we see from figure 5 that the trade deficit now is still much higher than back then.
    At the same time we see from figure 6 that US energy consumption stopped rising significantly around 1998 and has been essentially flat ever since. Coincidense?

  2. Thinkstoomuch says:


    Thank you for the post.

    A comment not really addressing the subject of the post. For the rise of President Trump and manufacturing GDP isn’t the real way to measure it. Dollars don’t vote, well technically. 😉

    This is a graph of what the average US Citizen sees on the streets.

    Hopefully the link works. But it isn’t the dollar amount but the jobs. Is it fair in today’s world and how we got here?

    To a certain extent no.

    But in a way it does relate. A question that has been bugging me about EROI, CO2 and _a_ measure of energy efficiency. Does actually replacing a human manufactured item with a machine manufactured item save energy?


    • Euan Mearns says:

      With manufacturing jobs in decline and manufacturing GDP constant, then its clear that efficiency (automation) has improved. So its important to distinguish between jobs lost to Mexico and China from jobs lost to machines. I believe automation is pursued to make more money, and if it is more economical then it is likely more energy efficient too.

      • Thinkstoomuch says:

        I agree on the which jobs are lost. For example. Suniva makes the silicon wafers in the US then ships them overseas to be assembled into panels. To be shipped back to the US. IIRC.

        But economic efficiency does not have to be related to energy efficiency.

        For example.

        Referring to the Cash Flow Analysis chart for a typical material handling system with two robots, you can see the nominal maintenance costs through year 4, then the $10,000 incremental cost for replacement of worn components in year 5, and the $60,000 expenditure to refurbish the robots in year 10.

        Also part of TCO, the operating costs include the power consumption noted previously and the practically negligible cost for compressed air to operate the robot end-of-arm tooling or gripper in a typical material handling application. For our example, the $6,000 operating cost in year 1 accounts for two robots working two shifts per day, for 5 days a week, 50 weeks a year, at 75 cents per hour.
        End Quote

        The cost of electricity is stated as 10 cents an hour earlier, 60 MWH a year in this example. If people were doing the same task that electricity would not be used.

        People in those jobs would entail other energy costs (most likely higher US CO2 and primary energy costs) but not on the “factory floor”.


      • Andre says:

        One sector that remains robust in terms of employment, although one might expect decline in relative terms due to productivity gains from technology/automation, is government employment, which has remained doggedly at around 14-15% of total employment from 1939 to the present.

    • Willem Post says:


      The US should have local content laws, such as 85% of the VALUE of all parts (including the DESIGN value) in a car must be locally sourced.

      That way engines, transmissions, etc., for domestic and foreign LABELS, would be designed and built in the US.

      US workers need to be vertically employed, not just as assemblers.

      Is there a similar graph for agricultural employment?

      • Thinkstoomuch says:

        “Is there a similar graph for agricultural employment?”

        You can make your own graph on the site.

        Used to be more graph options … but another improvement. 🙁 I am not registered might get more that way.

        Search for “Agriculture, hunting and related service activities” from Euan’s appendix above. Second item on the list. Not seasonally adjusted for some reason. 😉

        But notice the scale max is at 1,500 thousand persons versus 160,000 thousand persons on the previous graph.

        Don’t copy and paste from the browser get the image url from “share links” button. You don’t want to know … 🙂

        You can plot all the line items from the US Jobs report and a _whole_ lot more. Which can make finding the right line item … interesting. Experience helps.

        For that matter the whole data tables are available from thru interactive charts/tables or it used to be, anyway. Takes a while until you get the hang of it.


    • “Does actually replacing a human manufactured item with a machine manufactured item save energy?”

      Certainly the level of automation over the years in the steel industry has increased output and lowered the energy cost per ton of a specific product/grade so it depends on what exactly do you mean and at what level? Certainly you do not want to return to manually drilling of blast furnaces…

      • Thinkstoomuch says:

        I do not want to return to the past!

        Though those were always the glorious days. 😉

        I asked a question and gave one example with numbers in a later comment.

        I “generally” think that any time a machine replaces a person it does cause more “measurable” energy usage.

        The energy requirements can be reduced by more energy efficient motors, improved processes that can better utilize the capabilities of the new machines and the like.

        For example using less material because machine is less likely break it than my fat, clumsy fingers.

        Or apparently the steel industry.


  3. Dave Rutledge says:

    Hi Euan,

    It is fascinating to me that you can see the greater slope on the coal decline compared to the smaller slope of the gas rise, evidence of the greater efficiency of combined-cycle generation compared with a steam turbine.

    The ongoing slide in fuel oil consumption is impressive. This too, is a direct switch to natural gas, although some of it may be indirect through installing electrical heat pumps.

    “Growth in bio-fuels 2008-2015 was 8.4 Mtoe. I’m unsure how to handle this since the ERoEI for corn ethanol is close to unity and any energy savings imagined in the vehicle fleet is matched by energy used to produce the biofuel.”

    A thought I had was that most of the energy use in making corn ethanol is for natural gas in the fertilizer and in the distillation.For the purpose of this post, simply view biofuels a way to convert natural gas to liquid fuel.

    From this perspective, we have natural gas displacing coal in electricity generation, fuel oil in heating, and the oil component of gasoline. Is there anything natural gas can’t do?


  4. meliorismnow says:

    “One must presume that manufacturing jobs lost in the rust belt have been replaced in other parts of the country” While this was probably true in the 90’s (as tech ramped up output to the world and as the US population generally felt wealthier) it certainly wasn’t true after. The US generally offshored any manufacturing that couldn’t be heavily automated, reducing manufacturing employment by 1/3 from 2000 to 2010. After that we see a slow rebound as pay between unskilled labor in the US and foreign markets begin to narrow, oil/transport prices rise, and as technology continues to replace unskilled foreign labor (with medium skilled domestic labor).

  5. Hi Euan:

    Vindicated again 🙂

    There’s an article on this subject over at the Energy Collective that Ryan Paulsmeyer linked to in my emissions post. It reaches conclusions similar to yours:

  6. David B. Benson says:

    Eastern Washington now has at least 4 carbon fiber manufacturers, relatively new. Here in Pullman, Washington, Schweitzer Engineering Laboratory continues to expand the engineering and manufacturing of computerized equipment for the electrical power industry. Truck and aircraft manufacturing is up in the western part of the state.

  7. jacobress says:

    If you look at Roger’s graph of emissions (Fig 1 in that post) and at the graph for total primary energy use (fig.6 here ) you see one clear thing: the reduction in emissions and in energy use (practically the same thing) was a one-time drop between 2007-2009.
    From 2009 onward there there is no energy use reduction trend and no emission reduction trend. Both graphs are rather flat.

Leave a Reply