How to account for emissions of greenhouse gases from imported goods?

Sweden’s greenhouse gas emissions dropped by 22 per cent between 1990 and 2013, as reported to the climate convention. But what if emissions caused by imported goods are included? Would there be no emissions drop or an even bigger drop? Results go in all directions, and are of interest far beyond academia.


1990 million tonnes CO₂eq.

2012 million tonnes CO₂eq.

Emissions from Sweden



From consumption






In April 2015, the Swedish Environmental Agency, SNV, claimed that if consumption-induced emissions are added to the emissions from Swedish territory, the results are as follows:

So in reality, Sweden’s emissions increased instead of decreasing! That, at least, was the Swedish media take on the SNV press release. (There is no actual report, just a few html pages.)

Other reports have come to different conclusions.

According to Glenn Peters et al., Sweden imported less CO₂ in 2008 than in 1990. This means that the emission cuts, so far, were real (although data quality for the early 1990s is poor).

Astrid Kander et al., write: “Several European countries – Sweden, France, Ireland and Austria – are large net importers of embodied carbon emissions according to CBA (consumption-based accounting), but are instead net exporters when the carbon footprint is adjusted for technology differences in exports”. This also means that emission cuts are real. (The technology element is explained below.)

The studies disagree. 

Accounting for CO₂ emissions in a nation is (relatively) simple. Production and import data for fossil fuels are collected for administrative reasons. 

Consumption-based emissions are much harder to measure. Look at a mobile phone. Parts and materials come from all over the world, using thousands of different technologies, each of them ranging from the best available technology to the worst current technology. A serious effort would result in an ocean of data, but still with a high degree of uncertainty regarding emissions. 

A bottom-up approach is impractical. Indirect emissions have to be modelled as products of trade volumes in money and greenhouse gas intensities. The data is aggregated, so for example “chemical industry” can include GHG-intensive production (fertilizer, chlorine) and the low-emitting pharmaceutical industry.

There are three reasons why greenhouse gases are reported as they are to the Climate Convention.  First, the reports present a picture of a nation’s emissions and trends, which is useful even if it is not the whole truth.  Second, there is no cheating because there is an agreed methodology. This is quite an achievement. Third, the emissions within a country are roughly within its own control: a government can influence energy standards for buildings, tax on petrol, push down coal power emissions through environmental legislation etc.

This was good enough for an agreement on the Kyoto protocol in 1997 and later for burden sharing within the EU. In all probability it will be the base for a Paris agreement in 2015.

There are however flaws, which are easy to spot. 

Some countries have carbon-intensive production, and thus high emissions, whereas others have less emissions but import goods produced in the former countries. The geography is not fair. Nations that increase their exports of steel, cement, aluminium and silicon should be encouraged to do so if they can produce them with lower emissions than the world average, as this decreases global emissions. Conversely, nations that close down dirty industries but import more and more materials produced by dirty industries in other countries actually add to global emissions, and should have stricter climate targets.

There is also a historical disparity. Some countries have a tradition of high-emitting industry. Industrial structure can change but usually not very fast.

The consumption-based view as presented by SNV above does not take exports into consideration, but adds territorial emissions to imported emissions. This obviously overstates Sweden’s GHG global footprint.

Less obviously, it misrepresents what kind of economy Sweden is, what has been achieved and what has not. In greenhouse terms, Sweden is not a “services economy”. It produces very large amounts of iron ore, ore-based steel, paper and pulp, oil refinery products, cement, copper, aluminium etc., much of it for export. It is also a big net exporter of power. It imports a mix of products, many of which are not very GHG-intensive.

A popular notion is that dirty industries have moved to China, while consumption of imported goods from China is increasing in the OECD, including Sweden. But is that a fact?

After all, the heavy CO₂ emitters in Sweden – oil refineries and producers of cement, lime, steel, paper and pulp – have not moved abroad. They are still in Sweden. They have invested and have stepped up their production.

The decarbonization of heat and electricity in Sweden, which is the main explanation for the reported emissions decrease, is no statistical artifact. It is real. The increasing export of electricity, from wind and biomass, in recent years has also led to less use of coal power in neighbouring countries. There have also been real improvements in some industries.

If the models do not represent this, they are not very useful. It is hard to conceive that the present system should be switched to consumption-based accounting (CBA),but it could be used to complement it.

The models are so far not good enough, but if they can be improved they would be useful for evaluating the performance of different countries, for their national targets, which in the long run will also influence international negotiations.

Kander et al. propose a “technology adjusted” CBA, which would be a great improvement. An example: the Swedish pulp industry uses less fossil fuel than its competitors, and also less than it used to. Higher exports of pulp and paper will, all other things being equal, decrease world emissions.

It is difficult to change the method of accounting, but it would be welcome to add a note about how consumption and exports contribute to our GHG footprint, and whether our heavy industries are really best-in-class and improving at a reasonable rate. 

International transport is not included in the emission total, though such data are attached. This is of course not good, as trade and travel cause substantial and increasing emissions. But there is a reason: aeroplanes and ships are fuelled where it is convenient. It has little to do with the destination for goods and people. If a simple attribution formula for international transport could be negotiated, it would be a good thing, but this is perhaps not high on the agenda.

The underlying assumption of any climate negotiation is the “contract and converge” notion, aiming for a smaller and more equal per capita footprint. But if every nation aims for the same per capita emissions, some have to work much harder than others, for geographical and historical reasons. Norway can produce all its electricity from hydropower, whereas Denmark has no hydro. On the other hand Denmark is densely populated, which makes cycling and public transport easier than in mountainous Norway. Differences cancel out over larger areas, and if they don’t there is the option to trade between nations, at a probably modest cost.

The geographical problem can be mitigated very much by pooling resources, not necessarily through mechanisms in the Kyoto Protocol. Denmark is the world champion of wind power and gets 40 per cent of its electricity from wind. This is possible because it can balance surpluses and deficits through trade with Norway and Sweden. If there is a will, there is a way.

Anyway, differences can be acknowledged in climate negotiations, as they were in the Kyoto Protocol and in EU burden sharing, and more recently in the US-China climate agreement and other national commitments. These reflect common but differentiated responsibilities. 

It is not possible to capture everything in one formula.

The problem may also be decreasing. The historical and geographical excuses for special treatment and high emissions are becoming less and less credible. New production methods or power plants look pretty much the same everywhere in the world.


Fredrik Lundberg

1. (in Swedish)

2. Growth in emission transfers via international trade from 1990 to 2008, Excel appendix

3. National greenhouse gas accounting for effective climate policy on international trade, Nature Climate Change, May 2015, DOI:10.1038/NCLIMATE2555

In this issue