全球变化知识资源中心

There are three important conditions that a national carbon accounting scheme ideally should satisfy to provide useful and reliable feedback for global and national climate policy.

First, it should be responsive to factors that nations can influence, for example the level and composition of their consumption, and their domestic carbon efficiency (sensitivity). Second, countries should not be able to reduce their national carbon footprints in ways that contribute to increased global carbon emissions (monotonicity). Third, the sum of national emissions for all countries should equal total global emissions (additivity). For a further elaboration of these conditions see Supplementary Section 1.

An accounting method that does not satisfy Sensitivity will leave some relevant and manageable sources of emissions out of its scope. A method that does not satisfy Monotonicity may sometimes yield misleading feedback on the effects of national policies: a country that tries to contribute to global climate targets by reducing its own carbon emissions may inadvertently contribute to increased emissions globally. Finally, a method that does not respect Additivity is inadequate as a baseline for allocating responsibilities for emission reductions because it cannot guarantee that global reduction targets are reached even if all countries meet their individual targets.

Neither production-based accounting (PBA) nor consumption-based accounting (CBA) satisfies the first two conditions. The problems with PBA have been widely recognized^{2, 3} but the problems with CBA are less frequently noted^{4, 5}. One weakness of CBA is that it is not responsive to changes in the carbon efficiency of export sectors, because all export-linked emissions are passed on to final consumers.

A second weakness is that CBA fails to encourage certain kinds of specialization and trade that might contribute to a more carbon efficient use of global production resources. This argument is analogous to classical economic arguments about the possibility of welfare gains through trade between parties with different comparative advantages. Even if a country has more carbon efficient production than its trading partners in all production sectors, it might be possible to achieve welfare gains in terms of reduced global carbon emissions by exploiting differences in sectoral carbon efficiency through international trade.

Under CBA, however, countries with more carbon efficient production technologies than their trading partners can be punished for participating in such trade—in the sense of having higher national emissions—even if the overall result is a reduction of global emissions. This argument is supported by an example and formal proof in Supplementary Section 2.

Unlike classical comparative advantages, which are defined in terms of production factor costs that are covered by the parties to a transaction, comparative advantages are in this case related to an external cost—physical carbon emissions—and hence do not provide direct economic incentives. For this reason, it is essential that differences in production technologies between countries are recognized in carbon accounting, so that policies can be designed to support carbon-efficient trade patterns.

Because traded goods account for about one quarter of global CO₂ emissions, counterproductive policy implications in this area may affect a substantial fraction of total emissions.

When it comes to the first of these weaknesses, proponents of CBA have suggested regulating export-linked emissions indirectly by using border tax adjustments (BTAs; ref. 6). BTAs have been discussed at least since the early 1990s (ref. 7), but the idea remains controversial and unimplemented^{8, 9, 10, 11}. Here, it is sufficient to note that, although CBA supports some policy options available to import countries, which might in turn create indirect economic incentives to exporters, it does not provide direct feedback to export countries on the carbon efficiency of their export industries. Thereby, some potentially effective policy options are left out of focus.

We propose a new method for carbon accounting—technology-adjusted CBA (TCBA)—that addresses the issue of carbon intensity in exports. Like conventional carbon footprints, technology-adjusted footprints incorporate emissions embodied in trade, but also adjust for differences in carbon efficiency in export sectors of different countries.

CBA is calculated by adding embodied emissions in imports to production emissions and subtracting embodied emissions in exports—in both cases using the average emissions intensity of the relevant production sector in the producer country.

TCBA applies a similar formula, with a twist: export-related emissions are subtracted based on the average carbon intensity for the relevant sector on the world market, rather than the domestic average. The reasoning is this: if carbon footprints are to reflect the effects of a country’s export on global emissions, we must consider not only how a certain exported commodity was actually produced, but also what alternative production it replaces.

Because we here compare, ex ante, the current state of affairs with a potential change—we consider what would be the case if a certain commodity were not to be exported from the country in question—we normally do not know exactly which alternative supplier would provide the substitute.

Given this lack of knowledge, we suggest that the most plausible, and least demanding, assumption is that a similar good would have been produced at the average emissions intensity on the world market for the relevant sector.

Because the global sum of all imports, given the actual emissions intensities of producers, equals the global sum of all exports, given the world market average emissions intensities for each sector, TCBA will satisfy Additivity.

On the assumption that the expected effect of a marginal change in a country’s export can be calculated by comparing it against the world market average emissions intensity for the relevant sector, TCBA would also satisfy the first two conditions. A proof is provided in Supplementary Section 3.

Note that the world market average in this calculation includes the export of the country under study. The reason for doing this (rather than taking just the average of all other countries) is that it is a simple way of ensuring that it satisfies Additivity. As a consequence, however, TCBA will to some extent underestimate the effects on global emissions of technology differences between countries in their export industries, and it will therefore not fully satisfy Monotonicity.

TCBA is designed to account for the effects of small, marginal changes, but is less suitable for tracing the effects of large-scale changes in exports with dynamic impacts especially for larger exporters.

Still, TCBA does constitute a clear improvement of consumption-based carbon accounting in this respect, because standard CBA ignores technology differences in export completely.

We used a global multi-regional input–output table to compute the TCBA for 40 countries. Comparing the results against those of PBA and CBA, it is found that TCBA changes the levels and trends of many regions of the world, but not in a way that systematically benefits either developed or developing regions (Fig. 1).

Complete results are available in Supplementary Section 5. The legend applies to all panels.

Technology-adjusted carbon footprints throw new light on the so-called displacement hypothesis and improve our understanding of the impact of international trade on global emissions.

By accounting for technological differences in production for export TCBA will support a wider range of policy options, available to consumers as well as producers, and also better align incentives with available policy options at a national level, than conventional consumption-based footprints.

All methods have their limitations, however, as TCBA, CBA and PBA do not reflect all kinds of indirect and dynamic climate policy effects. Indirect effects—for example, positive stimulus on other countries’ climate policies—will not be accounted for.

Another limitation with both CBA and TCBA is that all responsibility for consumption is allocated to consumers, even in cases where producers clearly influence it through marketing or innovations. TCBA assumes that export countries are held responsible for the technology in their exports but not for consumption that is induced by such export. If a country has contributed to lower global emissions by an innovation that also triggers a change in world demand for this clean product—and thereby also in the country’s export—this contribution may not always be reflected correctly by TCBA.

A third limitation is that TCBA may to some extent underestimate the effects of technology differences in export between countries, and hence does not fully satisfy Monotonicity. Still, even with these limitations, TCBA does offer clear advantages compared to standard CBA in accounting for the effects on global emissions of international trade.

Complementing conventional national carbon inventories with carbon footprinting according to TCBA would therefore provide useful information on the expected effects of national decisions on global emissions. Together with additional information about demand and supply elasticities, it may assist national decision makers in choosing efficient policies to contribute to global climate goals.

The calculations in this study were executed using the World Input–Output Database (WIOD; ref. 15), one of a new generation¹⁶ of global multi-region input–output (MRIO) databases that document trade flows between countries. WIOD was chosen over the other available MRIO databases (specifically, over Eora¹⁷) because of its homogeneous sector classification, which allows industries in different countries to be directly compared. For the purposes of this initial analysis WIOD offers reasonable resolution (35 sectors per country), country coverage (40 countries responsible for >97% of global GDP), and temporal coverage (1995–2010).

The accuracy of CBA and TCBA accounting is constrained by the level of sectoral resolution available in MRIO tables because sector-average carbon intensities are applied to a potentially diverse set of products within that sector. This may introduce an unintended bias in the balances of embodied emissions if a country’s import and export mix of products within one sector differs significantly in terms of carbon intensity. This ’aggregation’ problem is well known in input–output analysis¹⁸, is generally accepted to introduce manageable uncertainty, and can be addressed by increasing the number of sectors in the model^{19, 20} as done, for example, in the Eora²¹ and EXIOPOL (ref. 22) projects. To give one concrete example of how product-level heterogeneity can affect results, consider a country which exclusively exports pharmaceutical products which it is able to produce with low carbon emissions. In the MRIO tables pharmaceutical products are included within a broader chemicals sector. This country will receive ’bonuses’ in relation to the world average for the total chemical sector (because its export mix is so clean), yet if the MRIO tables had distinguished pharmaceutical products from chemicals more generally, these bonuses would be fewer. The inverse may occur too: a country could export a specific product which is significantly more carbon intensive than is its parent sector as a whole. The question whether TCBA is more susceptible to aggregation error than CBA would merit further empirical analysis which was beyond the scope of this Letter.

Whilst the results presented here demonstrate the principal validity, applicability and relevance of the TCBA method with the 35-sector WIOD model we would recommend more detailed analyses before implementing specific policies in practice.

Consumption-based accounting (CBA) results were calculated in the normal manner using a Leontief demand-pull model^{23, 24, 25}. In this model emissions in producer countries are reallocated to final consumers by following products through multiple trade and transformation steps. A summary of differences between PBA, CBA and TCBA for some typical countries is given in Table 1. In the Leontief model the CBA emissions account can be calculated as in equation (1):

where f denotes direct emissions, q is an emissions multiplier # denotes elementwise multiplication, and x_i^sr is the output from production sector i in country s that is produced for final consumption in country r. The distribution of total output from each sector over final consumers is given by equation (2):

Here y is a collection of final demand bundles and L is the classical Leontief inverse, , where I is the identity matrix, Z is a multi-region input–output table documenting economic flows between countries and sectors, is the diagonal of k, and k_i records gross output of sector i. The emissions multiplier q is calculated by simply dividing direct emissions with total output—allocated amongst all final consumers—for the relevant sector:

To calculate TCBA we begin as before by adding emissions embodied in imports to territorial emissions. But then, when subtracting export-related emissions, instead of using the domestic emissions multiplier q_i^s of the exporter, we use a weighted world market average emissions multiplier for each sector i. World market average is defined as the average emissions multiplier in the part of production that, directly or indirectly, is exported for foreign final consumption. This multiplier is calculated as:

Although it would be preferable to divide emissions linked to export production from those linked to production for domestic use, current emissions and economic databases, and thus current CBA accounts, do not make such a distinction.

Full table

1. Peters, G. P., Minx, J. C., Weber, C. L. & Edenhofer, O. Growth in emission transfers via international trade from 1990 to 2008. Proc. Natl Acad. Sci. USA 108, 8903–8908 (2011).
   • PubMed
   • Article
2. Peters, G. P. & Hertwich, E. G. Post-Kyoto greenhouse gas inventories: Production versus consumption. Climatic Change 86, 51–66 (2008).
   • CAS
   • ISI
   • Article
3. Davis, S. J. & Caldeira, K. Consumption-based accounting of CO₂ emissions. Proc. Natl Acad. Sci. USA 107, 5687–5692 (2010).
   • PubMed
   • Article
4. Steininger, K. et al. Justice and cost effectiveness of consumption-based versus production-based approaches in the case of unilateral climate policies. Glob. Environ. Change 24, 75–87 (2014).
   • Article
5. Jakob, M. & Marschinski, R. Interpreting trade-related CO₂ emission transfers. Nature Clim. Change 3, 19–23 (2013).
   • ISI
   • URL: