Quantitative Estimation of the Climatic Effects of Carbon Transferred by International Trade

Abstract

Carbon transfer via international trade affects the spatial pattern of global carbon emissions by redistributing emissions related to production of goods and services. It has potential impacts on attribution of the responsibility of various countries for climate change and formulation of carbon-reduction policies. However, the effect of carbon transfer on climate change has not been quantified. Here, we present a quantitative estimate of climatic impacts of carbon transfer based on a simple CO₂ Impulse Response Function and three Earth System Models. The results suggest that carbon transfer leads to a migration of CO₂ by 0.1-3.9 ppm or 3-9% of the rise in the global atmospheric concentrations from developed countries to developing countries during 1990-2005 and potentially reduces the effectiveness of the Kyoto Protocol by up to 5.3%. However, the induced atmospheric CO₂ concentration and climate changes (e.g., in temperature, ocean heat content, and sea-ice) are very small and lie within observed interannual variability. Given continuous growth of transferred carbon emissions and their proportion in global total carbon emissions, the climatic effect of traded carbon is likely to become more significant in the future, highlighting the need to consider carbon transfer in future climate negotiations.

Quantitative Estimation of

OPEN

R A

P

TingWei, Wenjie Dong,, ,,, QingYan, Yi Song, , WenpingYuan,, Jieming Chou, Xuefeng Cui , ZhigangWei, Yan Guo, ShiliYang, DiTian, Pengfei Lin, SongYang,, ZhipingWen,, Hui Lin, Min Chen,, Guolin Feng , Xian Zhu, Juan Chen, XinWei, Wen Shi, Zhiguo Zhang, Juan Dong, Yexin Li & Deliang Chen

CO

Humans have for centuries been changing the composition of the Earths atmosphere, leading to signicant climate change and air pollution, but the process has been particularly rapid since the 1950s1. To avoid the serious threat to the environment posed by exponential growth of greenhouse gas emissions, the international community has tried for 20years to reduce global carbon emissions through sovereign state-level negotiations. One critical issue in these negotiations is to dierentiate the historical responsibility for climate change and make a fair emission reduction program for dierent countries. Previous attribution studies of responsibility for climate change26 have been based on production-based emissions, i.e., accounted for using territorial boundaries7.

Zhuhai Joint Innovative Center State Key Laboratory of College of Global Arctic Centre, University of Nansen-Zhu International Research Centre, Institute of Atmospheric Marine Weather Forecast Division, National Marine The State Key Laboratory of Numerical Modelling for Atmospheric Sciences and Geophysical Fluid Dynamics, Institute of Atmospheric Physics, Chinese Academy of Institute of Space and Earth Information Science, Chinese University of Hong National Zhuhai Meteorological Bureau, Zhuhai

SCIENTIFIC REPORTS

1

www.nature.com/scientificreports/

Figure 1. Temporal evolution of the simulated atmospheric CO2 concentration changes relative to 1990 using (a) IRF, (b) CESM, (c) BNU-ESM and (d) FGOALS-s2 under the PAX1, PNX1, CAX1 and CNX1 scenarios.

(b) Shading shows the range of CO2 changes due to dierent initial conditions and lines are the ensemble mean.

These production-based emissions allow for convenient monitoring and regulation. However, international trade creates a geographic separation between the products final consumers and the carbon emitted in the production process, eectively shiing the CO2 associated with their consumption to distant lands811. This challenges the traditional principle of the polluter pays. One way of rectifying this problem is that responsibilities for climate change should be attributed in accordance with consumption-based accounting of carbon emissions that is dened as adding the emissions associated with imports and subtracting the emissions associated with exports, from production-based emissions1214. Therefore, it has been argued that the current production-based carbon emission inventories should be replaced by consumption-based system in formulating emission reduction policies in post-Kyoto frameworks1517.

Carbon emissions embodied in international trade rose by ~38% from 1990 to 200818, and the trend has continued in recent years11,1921, motivating our quantication of its impact on the climate change attribution and responsibilities for mitigation. We present estimates of the role of carbon emissions in international trade using both a simple model (allowing calculations over a longer time interval) and three state-of-the-art Earth System Models (limited to a shorter study period by data availability), and explore the potential impact of transferred carbon emissions on the Kyoto Protocol (KP)22. We believe that the results will be useful for international negotiations in the future.

Carbon emissions via international trade potentially reduce the gap in historical responsibilities for CO2 loading between developed and developing countries. To investigate the inuence of transferred carbon on historical responsibility for climate change, four experiments were designed and executed with a simple CO2

Impulse Response Function (IRF) model and three Earth System Models (Methods). The experiments are (i) PAX1: production-based CO2 emissions only allowed from developed countries (i.e., Annex I countries); (ii) PNX1: production-based CO2 emissions only allowed from developing countries (i.e., Non-annex I countries); (iii) CAX1: consumption-based CO2 emissions only allowed from the developed countries; and (iv) CNX1: consumption-based CO2 emissions only allowed from the developing countries. The simulations show that atmospheric CO2 concentrations would increase by 8.614.8ppm (11.212.5ppm) from 1990 to 2005 on conditions that production-based CO2 emissions are only allowed from the developed (developing) countries (Fig.1). If consumption-based CO2 emissions are only allowed from the developed countries or the developing countries, atmospheric CO2 concentrations show an increase of 9.615.6ppm and 8.911.5ppm, respectively. Therefore, over the period 19902005, 0.82.3 ppm CO2 was transferred from the developed world to the developing world via international trade. This indicates that 39% of responsibility for the increased atmospheric CO2 concentration was shied from the developed countries to the developing countries between 1990 and 2005 based on the normalized proportional method5(Table1). These results suggest that transferred carbon emissions reduce the dierence in historical responsibilities for CO2 loading between the developed and the developing countries, though these amounts are small.

Over the longer period (19902012), carbon emissions via international trade resulted in an increase of CO2 by ~1.4 ppm and hence a shi of historical responsibility by ~4% based on the IRF model. These numbers are quite similar to results for the 19902005 period from the Earth System Models. It should be noted that transferred carbon emissions account for a considerable proportion in production-based emissions for some regions and countries (e.g., China, USA and EU28). Based on the IRF model, carbon transfer (19902012) leads to a migration of CO2 by ~1.08ppm (accounting for ~17.2% of CO2 rise that results from consumption-based emissions) from other countries to China, whereas a transfer of CO2 by ~0.33 and 1.17 ppm (4.6% and 19.3%) from USA and EU28 to other countries, respectively (Supplementary Fig. S1).

SCIENTIFIC REPORTS

2

www.nature.com/scientificreports/

IRFa CESMa BNU-ESMa FGOALSa IRFb

AX1 production-based contribution 57% 53% 50% 43% 55%

NX1 production-based contribution 43% 47% 50% 57% 45% AX1 consumption-based contribution 60% 59% 53% 52% 51% NX1 consumption-based contribution 40% 41% 47% 48% 49% Transferred contribution 3% 6% 3% 9% 4% Eectiveness of AKNP 1.7% 0.7% 1.1% 8.9% 2.3%c Eectiveness of AKNC 5.0% 4.6% 5.2% 8.9% 6.0%c Transferred eectiveness 4.3% 5.3% 4.1% 0% 3.7%c

Table 1. Contributions of the developed (AX1) and developing (NX1) countries to the rise in atmospheric CO2 concentration and the eectiveness of AKNP and AKNC scenarios (see text for denition). aFrom 1990 to 2005. bFrom 1990 to 2012. cFrom 1990 to 2008.

As may be expected given the relatively small levels of CO2 involved, the climate system shows little response to the carbon transferred via international trade. The modeled warming of global atmosphere and oceans, and the melting of sea-ice in Northern Hemisphere are similar under all scenarios between 19902005 (Fig.2); also borne out by dierences in initial conditions being comparable with dierences between the experiments (Figs1 and 2). If a longer history of trade was available then climate eects due to trade may be more discernable, although trade has only grown rapidly in recent decades. The amount of transferred carbon emissions and their proportion in global total carbon emissions are gradually increasing18,19,21, and so traded carbon is likely to become more signicant in future.

Transferred carbon emissions will, to an extent, aect the eectiveness of the Kyoto Protocol. To investigate the impact of transferred carbon on the KP, we construct three CO2 emission pathways for 19902005 depending on whether carbon transfers are allowed between the developed and the developing countries while following CO2 mitigation protocols in KP (Methods). Under the scenario that the developed and the developing countries ignore their pledges and follow their production-based emissions (APNP; equivalent to the CMIP5 historical experiment), the simulated CO2 concentration in 2005 is 23.530.6ppm higher than in 1990 (Fig.3). If the developed countries follow the KP and the developing countries pursue their production-based emission (AKNP; as is specied by the KP), the increase of CO2 is simulated as 23.230.8ppm. Therefore, actual global carbon emissions are seemingly in keeping with the KP. When the developed countries follow the KP and the developing countries pursue their consumption-based emission (AKNC; equivalent to the KP without counting carbon transfers from the developed to the developing countries), simulated CO2 concentration increases by 22.429.2ppm, 01.6ppm less than that simulated by AKNP. We now dene the relative change of CO2 concentration under a mitigation scenario (i.e. AKNP or AKNC) to that under the observed emission scenario (i.e. APNP) as the eectiveness of the mitigation scenario. These 19902005 simulations indicate that the eectiveness of AKNP and AKNC is 0.78.9% and 4.68.9%, respectively (Table1). This result indicates that the trade between the developed countries and the developing world contributed up to 5.3% of CO2 concentration increases from 19902005 (Table1).

This is the contribution to CO2 rise from items actually used in the developed world but which were produced in the developing world, and hence escaped the limitations of the KP. Over the whole rst commitment period of KP (19902008) simulated by the IRF model, 3.7% of CO2 increase can be similarly attributed. The accumulated sum of transferred emissions (01.6ppm) from 1990 to 2005 is, however, small: less than the annual increase of CO2 (~1.7ppm/yr) over the same period. The climate system hence shows little response to the transferred emissions (Fig.4). Overall, the eectiveness of the Kyoto Protocol may have been potentially increased during 19902005 if the transferred carbon emissions are taken into account, though the resulting CO2 concentrations reduction and climate responses are tiny.

Numerical simulations with the IRF and three Earth System Models reveal that including carbon in international trade reduces the gap of historical responsibilities between the developed and the developing countries and the eectiveness of the KP. Although the climate change caused by the transferred carbon emissions (19902005) is almost negligible, the climatic eects of embodied emissions is expected to be more profound in future as global trade appears set to continue to grow. International trade also results in transfer of polluting gases which has additional environment and health hazards to the regions where goods are produced. For example, we estimate that the developed countries transferred 2.26 teragrams of SO2 to the developing world in 1990, which grew to 3.28 teragrams by 2005 (Supplementary Fig. S2). In addition, international trade potentially increases global carbon emissions as carbon-intensive manufacturing in emerging countries (e.g., China) entails more carbon emissions than would making the same product in the developed (importing) countries23. Given continuous growth of transferred carbon emissions and likely more signicant impact on climate change, future climate negotiations should take into consideration embodied emissions in international trade. This entails accurate national carbon emissions accounting24 and implementation of incentives to make a feasible, fair emissions reduction policy.

It is undeniable that international trade aects global carbon emissions, air pollution and countries historical responsibility by redistributing emissions related to production of goods and services. But countries with net exports prot while bearing the extra climatic and environmental burden. Whether the prots compensate for the damage, especially over the long run, is still an open question which has many other dimensions and cannot be properly addressed by simple measurement or models.

SCIENTIFIC REPORTS

3

www.nature.com/scientificreports/

Figure 2. Temporal evolution of annual mean surface air temperature, upper ocean heat content (0700m) and Northern Hemisphere sea ice fraction simulated by CESM (le panel), BNU-ESM (middle panel), and FGOALS-s2 (right panel) under the PAX1, PNX1, CAX1 and CNX1 scenarios. Le panel: shading shows the range of values due to dierent initial conditions and lines are the ensemble mean.

Figure 3. Same as in Fig. 1 but under the APNP, AKNP, and AKNC scenarios.

We use a CO2 Impulse Response Function (IRF) and three Earth System Models that have participated in the Coupled Model Intercomparison Project Phase 5 (CMIP5). The IRF is used to calculate CO2 concentration by a sum of exponentially decaying functions, one for each fraction of the additional concentrations, which should reect the time scales of dierent sinks25.

t

t t

=

CO CO CO ,0

1

t C E t f f e dt

( ) ( ) ,

(1)

+

CO S

2 2 2 2 2

2,

=

CO

CO S

,

S

n

SCIENTIFIC REPORTS

4

www.nature.com/scientificreports/

Figure 4. Same as in Fig. 2 but under the APNP, AKNP, and AKNC scenarios.

where CO2 is CO2 concentrations, CCO2 is a constant and set to approximately 0.47 ppmv/GtC, ECO2 is the emission of CO2, CO S

,

2 is the atmospheric exponential decay time of the sth fraction of the additional concentration CO2 (171.0, 18.0 and 2.57years), f CO ,0

2 is the rst fraction (0.152), and f CO S

,

2 is the respective fraction (0.253, 0.279 and 0.316). The coefficients are based on the impulse response of the Bern model26 as used in the IPCCSAR and IPCCTAR.

The three Earth System Models are the Community Earth System Model (CESM)27, the Beijing Normal University-Earth System Model (BNU-ESM)28 and the Flexible Global Ocean-Atmosphere-Land System model (FGOALS-s2)29. Each of the three Earth System Models contains an interactive carbon cycle module in the land component and an ecosystem-biogeochemical module in the ocean component. The simulated atmospheric CO2 concentrations are fully coupled to the land and ocean surface CO2 uxes and are used directly to compute the radiative forcing, hence forming a complete carbon cycle process. In this study, the atmospheric horizontal resolution of the CESM, BNU-ESM and FGOALS-s2 is ~0.91.25, ~2.8 2.8, and ~2.811.66, respectively. The ocean component has a nominal 1 resolution for the CESM and 1 1 for the BNU-ESM and FGOALS-s2.

Two groups of numerical experiments were designed to investigate the inuence of transferred carbon emissions. In group I, the inuence of transferred carbon emissions on historical climate change is examined. We design four scenarios in which production-based/consumption-based emissions are allowed only from either the developed countries or the developing countries (Supplementary Table S1). The production-based carbon emissions uxes are available at 1 1 spatial resolution from 1751 to 1949 at annual resolution and from 1950 to 2007 at monthly resolution30. The national inventories of consumption-based carbon emissions18 cover 113 regions and extend from 1990 to 2008. We use the regional distribution of production-based carbon uxes to construct gridded consumption-based carbon uxes at monthly and 1 1 spatial resolution. The cumulative transferred carbon emissions in the developed and the developing countries are shown in Supplementary Fig. S3. The CESM was rst integrated over the period of 18501990 under the PAX1 and PNX1 scenarios (Supplementary Table S1), respectively. Initialized from the year at 1990 in the PAX1 (PNX1) experiment, CESM was further run from 1990 to 2005 under the PAX1 and CAX1 (PNX1 and CNX1) scenarios. Other forcings varying over the historical period (18502005) include CH4, N2O, halocarbons, aerosols, solar irradiance, and volcanoes. The same method is used for the BNU-ESM and FGOALS-s2. Note that we run all the sensitivity experiments (groups I and II) with three dierent initial conditions using CESM. For the IRF, we construct four time series of carbon emissions (18502012) based on the designed emissions scenarios (Supplementary Table S1).

In group II, the eect of mitigation of production-based and consumption-based counting on the KP is investigated. We assume that each developed country decreases (or increases) its annual carbon emissions linearly and achieves its reduction commitment in 2008 according to the KPwhose purpose is reducing the overall emissions of anthropogenic greenhouse gases of the developed world by at least 5% below the 1990 levels in the commitment period from 2008 to 2012. We construct three emission inventories (Supplementary Table S1) for each developed country from 1990 to 2005 at monthly and 1 1 spatial resolution (Supplementary Fig. S4). For the CESM, BNU-ESM, and FGOALS-s2, each model is rst integrated over the period of 18501990 under the

SCIENTIFIC REPORTS

5

www.nature.com/scientificreports/

scenario that all countries follow their production-based carbon emissions (equivalent to the CMIP5 historical experiment). Starting from the end of this experiment, each model was then run from 1990 to 2005 under the APNP, AKNP and AKNC scenarios (Supplementary Table S1). For the IRF, we create three time series of carbon emissions (18502008) based on the designed emissions pathways (Supplementary Table S1).

1. IPCC. Summary for Policymakers. In: Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fih Assessment Report of the Intergovernmental Panel on Climate Change. Stocker T et al. (eds). Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA, 130 (2013).

2. Hhne, N. & Blok, K. Calculating historical contributions to climate changediscussing the Brazilian Proposal. Clim. Change 71, 141173 (2005).

3. Allen, M. R. et al. Warming caused by cumulative carbon emissions towards the trillionth tonne. Nature 458, 11631166 (2009).4. Prather, M. et al. Tracking uncertainties in the causal chain from human activities to climate. Geophys. Res. Lett. 36, L05707, doi: 10.1029/2008GL036474 (2009).

5. Wei, T. et al. Developed and developing world responsibilities for historical climate change and CO2 mitigation. Proc. Natl. Acad. Sci. USA 109, 1291112915 (2012).

6. den Elzen, M., Olivier, J., Hhne, N. & Janssens-Maenhout, G. Countries contributions to climate change: eect of accounting for all greenhouse gases, recent trends, basic needs and technological progress. Clim. Change 121, 397412 (2013).

7. IPCC. IPCC Guidelines for National Greenhouse Gas Inventories, Prepared by the National Greenhouse Gas Inventories Programme. Eggleston, H. S. et al. (eds). IGES, Japan (2006).

8. Kondo, Y., Moriguchi, Y. & Shimizu, H. CO2 emissions in Japan: inuences of imports and exports. Appl. Energy 59, 163174 (1998).9. Munksgaard, J. & Pedersen, K. A. CO2 accounts for open economies: producer or consumer responsibility? Energy Policy 29, 327334 (2001).

10. Davis, S. & Caldeira, K. Consumption-based accounting of CO2 emissions. Proc. Natl. Acad. Sci. USA 107, 56875692 (2010).11. Peters, G. P. et al. Rapid growth in CO2 emissions aer the 20082009 global nancial crisis. Nat. Clim. Change 2, 24 (2012).12. Ferng, J. Allocating the responsibility of CO2 over-emissions from the perspectives of benet principle and ecological decit. Ecol. Econ. 46, 121141 (2003).

13. Bastianoni, S., Pulselli, F. M. & Tiezzi, E. The problem of assigning responsibility for greenhouse gas emissions. Ecol. Econ. 49,

253257 (2004).

14. Rodrigues, J., Marques, A. & Domingos, T. Carbon responsibility and embodied emissions theory and measurement. Routledge, London, UK (2010).

15. Wycko, A. W. & Roop, J. M. The embodiment of carbon in imports of manufactured products: Implications for international agreements on greenhouse gas emissions. Energy Policy 22, 187194 (1994).

16. Peters, G. & Hertwich, E. Post-Kyoto greenhouse gas inventories: production versus consumption. Clim. Change 86, 5166 (2008).17. Barrett, J. et al. Consumption-based GHG emission accounting: a UK case study. Climate Policy 13, 451470 (2013).18. Peters, G., Minx, J., Weber, C. & Edenhofer, O. Growth in emission transfers via international trade from 1990 to 2008. Proc. Natl. Acad. Sci. USA 108, 89038908 (2011).

19. Lenzen, M., Moran, D., Kanemoto, K. & Geschke, A. Building Eora: a global multi-region inpute output database at high country and sector resolution. Econ. Syst. Res. 25, 2049 (2013).

20. Kanemoto, K., Moran, D., Lenzen, M. & Geschke, A. International trade undermines national emission reduction targets: New evidence from air pollution. Glob. Environ. Change 24, 5259 (2014).

21. Yang, Z. et al. Constructing long-term (19482011) consumption-based emissions inventories. J. Clean Prod. 103, 793800 (2014).22. United Nations Framework Convention on Climate Change, Kyoto Protocol. (2010) Available at: http://unfccc.int/kyoto_protocol/ items/2830.php. (Date of access: 15th March 2016).

23. Liu, Z. et al. Targeted opportunities to address the climatetrade dilemma in China, Nat. Clim. Change 6, 201206 (2016).24. Kander, A., Jiborn, M., Moran, D. D. & Wiedmann, T. O. National greenhouse-gas accounting for eective climate policy on international trade. Nat Clim. Change 5, 431435 (2015).

25. den Elzen, M. et al. The Brazilian proposal and other options for international burden sharing: an evaluation of methodological and policy aspects using the FAIR model, RIVM Report No. 728001011 (1999).

26. Siegenthaler, U. & Joos, F. Use of a simple model for studying oceanic tracer distributions and the global carbon cycle. Tellus B 44, 186207 (1992).

27. Gent, P. et al. The community climate system model version 4. Bull. Am. Meteorol. Soc. 24, 49734991 (2011).28. Ji, D. et al. Description and basic evaluation of BNU-ESM version 1. Geosci. Model Dev. 7, 20392064 (2014).29. Bao, Q. et al. The Flexible global ocean-atmosphere-land system model, spectral version 2: FGOALS-s2, Adv. Atm. Sci. 30, 561576 (2013).

30. Andres, R. J., Boden, T. A. & Marland, G. Annual fossil-fuel CO2 emissions: mass of emissions gridded by one degree latitude by one degree longitude. CDIAC. doi: 10.3334/CDIAC/e.AnnualIsomass (2013).

We thank Glen Peters and Steven J. Davis for providing carbon transfer emissions inventories. Many thanks to Jason Cohen, J. Hang, G.X. Chen, and J. Zhao for their help in revising the manuscript. This study was jointly funded by the National Natural Science Foundation of China (41330527), National Key Basic Research Program of China (2015CB954100), National Natural Science Foundation of China (41505068 and 41402158), and National Key Program for Global Change Research of China (2010CB950500).

W.D. conceived this study. W.T. performed all the analysis, plotted the gures and wrote the initial version of the manuscript with great input from J.M. and Q.Y. T.W. performed the CESM and BNU-ESM simulations with W.Y., X.C., X.Y., Z.W., Y.G., Shili Y., D.T., Song Y., Z.W., H.L., M.C., F.G., Y.J., X.Z., J.C., X.W., W.S., Z.Z., J.D., Y.L. and D.C. Y.S. and P.L. performed the FGOALS simulations. Z.Y. and J.C. contributed to the long-term consumption measurement model. All authors contributed to discussion of the results and writing of the paper.

Supplementary information accompanies this paper at http://www.nature.com/srep

Competing nancial interests: The authors declare no competing nancial interests.

SCIENTIFIC REPORTS

6

www.nature.com/scientificreports/

How to cite this article: Wei, T. et al. Quantitative Estimation of the Climatic Eects of Carbon Transferred by

International Trade. Sci. Rep. 6, 28046; doi: 10.1038/srep28046 (2016).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

SCIENTIFIC REPORTS

7