Question: How much of world carbon dioxide emissions are from international upstream trade (fossil fuels exported, imported, then burned)?
The EIA provides easily accessible international energy export data, reproduced below. (I assume total world imports should be identical, because no country is trading with space aliens, yet.) The fossil fuels in question are coal, oil, and gas. A nuance is that some of the traded petroleum has been refined while most is in crude form; moreover, the traded crude itself would be later refined into an assortment of end products. In addition, while the coal number seems simpler, it actually encompasses very different coal types. Finally, natural gas has one main category, but not all of it is combusted. These subtleties will be important in the calculation later.
TABLE 1. 2010* World Fossil Fuel Exports 
|Distillate fuel oil||6,209.17||”|
|Residual fuel oil||5,235.60||”|
|All coal||1,211,835||1000 short tons|
|Natural gas||35,405||billion cubic ft.|
*I focus on 2010 because it is the most recent year with complete international data (I think EIA is still collecting oil data from some countries for more recent years, as they show “NA” now), but the method should work for any year.
The EIA helpfully provides a table of conversion coefficients for various fuel types, seen in Table 2. Notice some naming incongruities with Table 1. “Distillate fuel oil” includes fuels used for “home heating” as well as “diesel fuel,” so I am assuming these two types are equivalent. The same is true for “residual fuel oil” and “residual heating oil.” Finally, liquefied petroleum gas (“LPG”) contains butane and propane, so I am also assuming the coefficient for “butane/propane mix” is representative.
These numbers represent emissions that occur during the combustion of a given volume of fuel. They do not include energy emissions during production or transport, i.e. energy used to operate drills or ships or pumps, which can be inordinate for unconventional sources like tar sands. This omission is actually necessary to avoid double counting, because any fossil fuels used for these purposes have already been accounted for at the point of extraction (they’ve already been dug up). However, another omission involves fugitive emissions; since these gas releases are not included elsewhere as part of fossil fuel production, this exclusion would make my number slightly too low.
TABLE 2. Carbon Dioxide Emission Coefficients for Fossil Fuels 
|Type||Emissions||kg CO2 per unit|
|Home heating, diesel fuel||10.2||”|
|Residual heating fuel||11.8||”|
||1000 cubic ft.|
Now I use production data (along with the Table 2 numbers above), to find conversion rates corresponding to the crude, coal, and gas rows in Table 1.
Again with the same EIA database, Table 3 shows what a “global average barrel” of crude would have been refined into in 2010. While different grades of crude (sweet vs. sour, light vs. heavy, etc.) can produce vastly different ratios of refined products, these geographical/source heterogeneities do not matter because I am finding the global average. Notice the total figure represents total refined outputs, rather than crude inputs, to avoid “refinery gain” errors – these end products overall are less dense than crude, so the input and output volumes are not comparable. Also, the individual volume ratios do not sum to 1, because they exclude petrochemicals that are not used as fuel and so contribute no emissions.
With appropriate values from Tables 2 and 3, a dot product finds that one gallon of crude oil leads to 7.9 kg CO2. This is 330 kg CO2 per barrel (42 gallons in a barrel ); this seems similar to the “minimum figure” of 317 kg CO2 per barrel from Jim Bliss’s blog .
Table 3. 2010 World Refined Petroleum Production 
|Type||k barrels/day||Volume ratio of total|
|Distillate fuel oil||25,445.90||30.37%|
|Residual fuel oil||9,804.50||11.70%|
EIA gives following production ratios for coal in Table 4. Unlike for oil, where the differentiation occurs in refineries, these differences for coal occur in the geological processes producing the coal itself. Using the same method as above, with numbers from Tables 2 and 4, I find one short ton of coal produces 2.134 kg CO2.
Table 4. 2010 World Coal Production 
|Type||1000 short tons||Ratio by mass|
Now, natural gas is a major raw material for plastics production and other industrial processes, so not all of it that gets drilled/pumped/fracked undergoes combustion to release CO2. Using 2010 data for consistency: 24,086,797 million cubic feet of natural gas was consumed in the US, of which 6,826,192 million cubic feet were used by industrial consumers and everything else for explicitly fuel applications . Assuming that US’s gas consumption profile is representative of the world, 28.34% of all natural gas is potentially prevented from being burnt. For simplicity, assume that all of the natural gas consumed by the industrial sector is used to make plastic bags or some other non-combustive application. (This is obviously not true because many applications labeled “industrial” actually do yield CO2, e.g. production of hydrogen gas; however, this would again make my number too low.) This brings the natural gas coefficient down to 38.05 kg CO2 / thousand cubic feet.
Finally, I can return to Table 1. With some simple dimensional analysis: in 2010, international petroleum trade accounted for 8.1 billion tons CO2, coal trade 2.6 billion tons, and gas trade 1.3 billion tons, for a total of 12 billion tons. This is in the context of 33.62 billion tons of total world emissions in 2010 (I had to multiply the raw figure by 3.667, as instructed by the source, to convert to units of CO2) .
Conclusion: In 2010, around 36% of global carbon dioxide emissions can be attributed to upstream fossil fuel trade.
 U.S. Energy Information Administration. “International Energy Statistics.” Web Aug. 3, 2014. Web. 3 Aug. 2014. <http://www.eia.gov/cfapps/ipdbproject/IEDIndex3.cfm?tid=90&pid=44&aid=8>.
 U.S. EIA. “Carbon Dioxide Emissions Coefficients.” Environment. Web. Aug. 3, 2014. <http://www.eia.gov/environment/emissions/co2_vol_mass.cfm>.
 U.S. EIA. “How many gallons of gasoline does one barrel of oil make?.” Frequently Asked Questions. Web. Aug. 3, 2014. <http://www.eia.gov/tools/faqs/faq.cfm?id=24&t=10>.
 Jim Bliss. “Carbon dioxide emissions per barrel of crude.” The Quiet Road Blog, 20 Mar. 2008. Web. 3 Aug. 2014. <http://numero57.net/2008/03/20/carbon-dioxide-emissions-per-barrel-of-crude/>.
 U.S. EIA. “Natural Gas Consumption By End Use.” Natural Gas, 31 Jul. 2014. Web. 3 Aug. 2014. <http://www.eia.gov/dnav/ng/ng_cons_sum_dcu_nus_a.htm>.
 Oak Ridge National Laboratory. “Global CO2 Emissions from Fossil-Fuel Burning, Cement Manufacture, and Gas Flaring: 1751-2010.” Carbon Dioxide Information Analysis Center, 30 Jul. 2013. Web. 3 Aug. 2014. <http://cdiac.ornl.gov/ftp/ndp030/global.1751_2010.ems>.