U.S. carbon emissions 20% greater than official estimates (corrected)

The data in this post has been rendered out of date due to improved methodology and updated results posted here. The description below is valid, but the data is not.

The role of the United States in climate disruption is far greater than most people realize. Not only does the U.S. emit more carbon dioxide (CO2) than any other nation besides China, not only does the U.S. have one the highest per-capita emissions in the world, but the U.S. economy also accounts for a massive amount of emissions released by the rest of the world too. S&R has investigated just how much CO2 the United States economy is actually responsible for, and the results suggest the real emissions are 20% greater than official estimates.

Every product and service requires energy, and thus carbon. Commercial agriculture requires petroleum or natural gas-based fertilizers and diesel fuel for planting and harvesting. Manufacturing requires energy to extract raw materials, petroleum to transport those materials to a factory, energy to convert those materials into products, and yet more petroleum to transport the products to end users. Even services like housecleaning or website hosting have an energy cost, the former in the creation of the chemicals and electric cleaning tools and the latter for the server (a product with its associated energy cost of creation), the electricity used to run the computer, and the energy consumed in constructing the computer center that houses the server have energy costs. And in all cases, the energy cost to create the product or service creates carbon emissions.

Given this, the amount of CO2 that a product or service indirectly emits in its creation, transport, and use can be estimated. And by extension the total amount of CO2 produced by the combined products and services (gross domestic product, or GDP) of a nation can also be estimated. When the total CO2 produced by a country by the country’s GDP, the amount of carbon emitted per unit of economic production can be determined. This is called “carbon intensity.”

The carbon intensity of the United States in 2006 was 0.52 metric tons of CO2 emitted per thousand dollars (indexed for inflation to the value of the dollar in 2000). For comparison, the carbon intensity of Iceland in 2006 was 0.31 metric tons of CO2 emitted per thousand dollars, and the carbon intensity of Russia was 4.54 metric tons of CO2 emitted per thousand dollars.

The fact that carbon intensity varies from country to country is a function of the country’s energy mix and overall productivity – more coal or oil burned for electricity or heating produces higher carbon intensity, and lots of manual labor producing valuable products produces a higher carbon intensity too. Large amounts of manual labor producing inexpensive products produces an extremely low carbon intensity, as witnessed by the very low carbon intensity of 0.10 metric tons of CO2 emitted per thousand dollars from Cambodia.

From carbon intensity, the amount of CO2 produced in the process of creating the goods and services that the U.S. exports – and that other countries export to us – can be estimated as well. The result is the following graph:

Figure 1

Figure 1 shows the 15 nations who “export” the most CO2 to the United States in goods and services that the U.S. use as part of our economy. In essence, anything the U.S buys that says “Made in China” is part of the U.S. economy, and so the carbon emitted in the creation of that product belongs to the U.S. economy as much as the carbon emitted in manufacturing a Ford Focus in Detroit. Figure 1 represents the balance of carbon, imported CO2 from other nations to the U.S. minus the CO2 the U.S. exports to them, as determined from the nations’ carbon intensity. It’s clear from the figure that China contributes by far the most CO2 to U.S. carbon emissions.

In words, Figure 1 says that the U.S. exported over a billion metric tons of CO2 to the rest of the world in 2006.

Figure 2 below illustrates data in a similar fashion, but as a percentage of total U.S. carbon emissions:

Figure 2

In this case, U.S.-generated emissions as a percentage of total emissions attributable to the U.S. economy have fallen steadily since 1985, from a high of 97.8% to 79.3%. This means that the U.S. economy has offshored 20.7% of our CO2 emissions to the rest of the world at the same time the United States has offshored production, services, and jobs.

If the U.S. is no longer generating a significant amount of our CO2 emissions, that means that the official carbon intensity of the Untied States (0.52 metric tons per thousand dollars) is actually much higher. And if this is the case, that means the reduction in carbon intensity that many people are pleased about is at least partly an illusion. Figure 3 below illustrates how much an illusion the regular improvements in U.S. carbon intensity actually is:

Figure 3

Figure 3 shows an unpleasant fact – as U.S. businesses have offshored more and more of the U.S. economy’s CO2 emissions to parts of the world where the carbon intensity is higher but labor is cheaper, the economy’s real carbon intensity has actually worsened since it hit it’s all-time low in 2001.

Correction: Saint from the comments below indicated that I had made two errors. Saint’s first claim is that I forgot to add the monetary value of imports into GDP while adding the CO2 value of imports. The second was that the trade values I was using are valued in current dollars rather than “real” dollars (indexed for inflation to the value of the dollar in the year 2000). I have not been able to verify Saint’s second claim since the link provided in the comment doesn’t point to anything that I can find saying “current” dollars, and I used the Census Bureau’s foreign trade data anyway, but did verify that I had made the first error. However, if I assume that I’ve made the second error that he claims and when I corrected the data for the first error, Figure 4 was the result:

Figure 4

Note that, even corrected for a possible error in using current vs. real dollars, the difference between the two graphs is a drop in carbon “offshored” in 2006 from 20.7% in the erroneous graph to 18.3% in the corrected graph. Not insignificant, but not enough to change the overall conclusions of this analysis.


As a result of comments, I choose to run the numbers for the purchasing power parity (PPP) exchange rate as well as the market exchange rate (all the graphs above). The result is Figure 5 below.

Figure 5

Figure 5 was done under the same assumptions as the corrected Figure 4 above. Note that the official carbon intensity is actually worse than the actual carbon intensity until 2003 or 2004. In fact, in 1991 and 1992, the U.S. exported a small amount of CO2 under this model. This graph shows two periods of significant offshored CO2 growth – 1997 to 2000, and 2001 to 2006, with the latter far outweighing the former. The data file for both Figures 4 and 5 is here (zipped .xls).

These figures illustrate a vitally important conclusion – the U.S. economy demands a huge amount of CO2 emissions beyond it’s borders. The U.S. has essentially offshored its GHG emission problem to the rest of the world, turning their economies into dumping grounds for our own air pollution. Yes, they’ve been paid well for it in U.S. dollars that helped raise the standards of living in the affected countries. But this also means the U.S. has a responsibility to help those countries clean their dirty energy houses.

After all, wouldn’t you want your neighbor to help rake up all the leaves he blew from his yard into yours?


S&R acquired the US GDP data, carbon intensity data related to energy production and gas flaring, and U.S. imports and exports from the data repositories listed below in sources and then performed the calculations that resulted in the graphs above. The Excel file of these calculations is available here (zipped Excel file) for anyone wishing to verify the calculations.

The EIA data on carbon intensity is only from the consumption of fossil fuels and flaring of gas – it does not include agricultural emissions, for example. In addition, the calculations are for carbon dioxide alone, excluding methane, ozone, HCFCs and other long lived greenhouse gases.

This analysis assumes that all units of production for import and export are equivalent in terms of CO2 emissions, while this is certainly untrue. However, given the large variety of imports and exports, and thus a large variety of CO2 emission profiles, S&R believes that this assumption is reasonable.

The estimates of imported (and exported) CO2 are equal the carbon intensity multiplied by the value of the imports, with the net amount of CO2 generated by other nations on behalf of the U.S. defined as the CO2 imports minus the exports.

Finally, all nations have data from 1992 until 2006, but only major trading partners have data from 1985 until 2006. This produces an error in the data from 1985 to 1992. Given that over 90% of all emissions are via major trading parters such as China and Canada, this error is believed to be relatively small.

US GDP information (Excel file)
Energy related carbon intensity (Excel file)
Data on trade for all countries back to 1992, major trading partners back to 1985 (zipped Excel file)

Crossposted:, ClimateProgress

14 replies »

  1. Nice work, on a topic which we (collaborators in the Greenhouse Development Rights project) are very keenly interested in.

    I can’t tell from this whether you used PPP-based GDP figures or not – what choice did you make?



  2. Thanks.

    I did not use PPP-adjusted GDP, only indexed it to 2000 dollars. I have no idea what adjusting the GDP numbers for purchasing power parity would do, but it would be a relatively easy thing to do (I think…).

  3. “After all, wouldn’t you want your neighbor to help rake up all the leaves he blew from his yard into yours?”

    as all libruls, you’re asking the wrong question.. The question is, “didn’t you blow your leaves into your neighbor’s yard in the middle of the night so you could avoid having to actually dispose of them properly?”.

    You are coming from a position of “do the right thing first”.. that’s not a right-wing premise.. “what’s best for ME, personally, regardless of the impact to others” is the right-wing premise. Blowing that CO2 into the air while profiting more, regardless of where it gets created was the “idea”, not an unwanted consequence. Therefore, saying we should help clean up our own mess brings you full circle to why they started “blowing leaves in the middle of the night” in the first place.

    If you were in their inner circles, they would say “SHHHHHHH!”

  4. True — there was a similar article in either FT or the Economist about 6-10 months ago regarding this very point – except it was between China and the EU. China made this very point saying 20 (or 30 or whatever %) of its emissions were caused by EU consumers and so was EU pollution. The guy from Brussels responded and said “yep and its our jobs as well”

    That said — the global slowdown will also reduce China’s pollution dramatically. Factories are being abandon ( and contrary to reports from China’s government regarding growth the economy is actually shrinking — see their Electricity usage for Nov (down 9% or so, again December will likely be much worse.

    The US and UK have both dropped the number of miles driven by 5-6 % YoY and transporation is about 1/3 of Carbon production. Builders, Manufacturers, Railway and Truckers in the US have seen tremendous drops in demand, which corresponds to CO2 output.

    So it looks like “Peak Credit” will buy us a bit of time -probably not much though

  5. Are the import figures also in adjusted 2000 dollars, or are they in nominal dollars? It’s not clear from the spreadsheet (nor is it clear where you got these import figures).

    In calculating the emissions intensity numbers, it seems you increased U.S. emissions by the amount “offshored,” but you did not increase U.S. GDP by the amount “offshored.” That’s analytically unsound. Why should the U.S. be attributed the offshored emisssions but not the offshored economic growth that gave rise to them? This means that the real emissions intensity of the U.S., as you define it, is overstated.

  6. I believe that the import figures are in adjusted 2000 dollars, although I can’t find again where I found that information. I’ll post again with a link one way or another.

    You are correct about the error in my GDP numbers. Given that imports are about 1% of GDP, the offshored GDP not being included in my numbers means that my intensity values have a 1% error. That’s not enough to change the overall conclusion in any significant way, though.

    As I try to be accurate and did mess up, I’ll update the numbers and post an update and explanation of the update. I’ll also post a PPP adjusted rate graph, just to see how it works out.

  7. There are two methodological issues with this analysis.

    First, the economic value attributed to the imports is missing from the U.S. intensity calculation. If we’re responsible for the emissions, we should take credit for the economic value associated with them.

    Second, the import and export trade figures are in current dollars. The EIA international intensity numbers are calculated using real (2000) dollars. Pegging the import and export dollar amounts to 2000 makes a significant difference.

    When you convert the import/export trade figures to 2000 dollars and include the economic activity from the net imports, the amount of CO2 “outsourced” from the U.S. winds up at about 12% in 2006.

    A third issue is whether to use the EIA intensity figures based on market exchange rates (what you used) or purchasing power parity. If you use EIA’s international intensity figures based on PPP instead of MER, the amount outsourced falls to just 7% in 2006.

    And I should add that, whether using either the MER or PPP data, when you put the import/export figures into constant dollars and include the economic activity data in the intensity calculations, what you call the “actual” CO2 intensity of the U.S. economy, broadly defined, does not deteriorate after 2001, but continues to improve.

    (BTW, the trade stats are avaialable at: TradeStats Express – National Trade Data When you go to the download center, it explains that the data are actual values in current U.S. dollars.)

  8. Saint, I’ve corrected the first error and given you the benefit of the doubt on the second with a correction factor for current to real dollars. The numbers I generated do not agree with yours – a reduction from 20.7% to 18.3% (not 12% as you claim) for MER (Figure 4) and about 10% for PPP (Figure 5).

    Figure 4 shows that actual CO2 intensity is flat from 2001 to 2005 to 2 significant digits but does fall again in 2006 – not as bad as the erroneous data, but still bad. You are correct for the PPP data shown in Figure 5, although 2005 to 2006 shows a noticeable flattening that makes me curious what will happen when I integrate 2007 EIA data.