Carbon Emissions and Carbon Intensity in China’s Exports: A Contrast of SRIO and GIRIO Methods

2.1 Carbon Emissions and Carbon Intensity in Exports in SRIO Model

In standard SRIO model, suppose S as object country, with n industries. Assume A_D as direct input coefficients matrix of domestic products, A_M as direct input coefficients matrix of imported products, v_S as value-added coefficient vector. Exports Ex_S includes domestic value-added in exports DVSSRIO, foreign value-added in exports FVSSRIO, i.e., foreign value-added embodied in imported intermediate inputs required for export production.

Similarly, carbon emissions in exports include domestic carbon emissions in exports DCSSRIO and foreign emissions in exports FCSSRIO, where c_S is carbon emissions coefficient vector.

Environmental cost per GDP created in export production is calculated by combining corresponding value-added terms and carbon emissions terms: embodied domestic carbon intensity in exports DESSRIO, embodied foreign carbon intensity in exports FESSRIO,

2.2 Carbon emissions and carbon intensities in exports in GIRIO model

Assume there are G countries, each with n industries. Suppose S as object country (exporter), and R, T as any other countries. GIRIO model covers worldwide interregional, inter-sector supply-demand linkages, that is, imported production distinguished by country source. Suppose Y_SR as country R’s final demand of country S's products, A as direct input coefficient, B as Leontief coefficient matrix. v is direct value-added coefficient vector, c is direct carbon emissions coefficient vector.

Under global intermediate trade, direct importer is not always the final destination of exported goods, and direct exports are not always the origin of exported products. By transforming endogenous intermediate trade in GIRIO model to terms induced by exogenous final demand, a key technical step by (7) decomposed gross exports flows to value-added flows by origins by final destinations[10]: (a) Domestic value-added exports DVXSGIRIO, i.e. domestic value-added finally absorbed abroad; (b) Domestic value-added in intermediate exports that finally returns home DRVSGIRIO; (c) Foreign value-added in supplying imported inputs FVSGIRIO. It bridges official direct trade data with value-added flows.

Similarly, carbon emissions in S′ exports can also be decomposed by original producers by final destinations along global supply chains. Carbon emissions in S′ exports CEE_S*:

1. Domestic carbon emission exports DCXSGIRIO, namely domestic emissions finally absorbed abroad:

   (9)DCXSGIRIO=cS∑R≠SGBSSYSR+cS∑R≠SGBSRYRR+cS∑R≠SG∑T≠S,RGBSRYRT

   (9) include S′ domestic emissions in final exports to R absorbed by R (1st term), in intermediate exports to R absorbed by R (2nd term), and in intermediate exports to T for production of final products absorbed by R (3rd term), involving indirect carbon emissions exports via third parties.

   Specifically S′ domestic emissions exports to country R:

   (10)DCXSRGIRIO=cSBSSYSR+cSBSRYSR+cS∑T≠S,RGBSTYTR

2. Domestic emissions in intermediate exports but finally returns home DRCSGIRIO.

   (11)DRCSGIRIO=cS∑R≠SGBSRYRS+cS∑R≠SGBSRARS(I−ASS)−1YSS+cS∑T≠SGBSRARS(I−ASS)−1YSSES∗

   Specifically R′s contribution in returned emissions in S′ exports:

   (12)DRCSRSGIRIO=cSBSRYRS+cSBSRARS(I−ASS)−1YSS+cSBSRARS(I−ASS)−1ES∗

   (11) and (12) include S′ domestic emissions in intermediate exports to R but returns home via final imports (1st term), via intermediate imports (2nd term), and domestic emissions in intermediate exports to R, then return home via intermediate imports but then exported, also double counted term but belongs to domestic emissions in exports (3rd term).

   Domestic emissions in exports DCSGIRIO include domestic carbon emissions exports DCXSGIRIO and domestic emissions firstly exported but finally returned home DRCSGIRIO.

   (13)DCSGIRIO=DCXSGIRIO+DRCSGIRIO

3. Foreign carbon emissions in S′ exports FCSGIRIO:

   (14)FCSGIRIO=∑R≠SGcTBTSYSR+∑R≠SGcTBTSYSRASR(I−ARR−1)YRR+∑T≠SG∑T≠S,RGcTBTSASR(I−ARR−1)ER∗

   Specifically T′s carbon emissions in S′ exports FCTSGIRIO:

   (15)FCTSGIRIO=∑R≠SGcTBTSYSR+∑R≠SGcTBTSYSRASR(I−ARR−1)YRR+∑T≠SGcTBTSASR(I−ARR−1)ER∗

(14) and (15) include T′ emissions in intermediate exports to S for producing S′ final exports absorbed abroad (1st term), for producing S′ intermediate exports absorbed abroad (2nd term), for producing S′ intermediate exports then used by R to producing R′ exports, also double-counted but belongs to foreign emissions in exports (3rd term).

Above decomposition shows value-added distribution and carbon emissions transfer among countries in exports production. Combing value-added and emissions accounting by origin by destination, calculate each country’s potential environment cost per value-added gained, i.e. embodied carbon intensity of each component, which indicates different participation in exports production fragmentation.

Then, embodied domestic carbon intensity in exports absorbed abroad DEXSGIRIO, embodied domestic carbon intensity in return part DRESGIRIO, and embodied foreign carbon intensity in exports FESGIRIO.

Overall domestic carbon intensity in exports DESGIRIO are shown as follow.

Figure 1 show the overall extending KWW framework of value-added, carbon emissions, and carbon intensities accounting of gross export.

Figure 1

Value-added, carbon emissions, and carbon intensities accounting of gross exports

2.3 Contrast of SRIO and GIRIO Model in Exports Analysis

SRIO model mainly specifies domestic production while GIRIO model captures global interregional inter-sector linkages. Different scopes in Leontief matrix, intermediate and final trade flows thus affect estimation of carbon emissions and carbon intensities embodied in exports.

Specially, based on the assumption that all imported intermediates contain 100% foreign content, i.e., no domestic emissions that initially exported but returned home via intermediate imports DRCSGIRIO, SRIO model underestimates domestic emissions in exports DCSGIRIO, confuses carbon emissions exports DCXSGIRIO with DCSSRIO, and mistakes foreign emissions in exports FCSGIRIO with domestic emissions avoided by imports FCSSRIO, compared with GIRIO analysis.

Furthermore, when more than one country export intermediate products, direct importer is not always the final destination, and direct exporter is not always the origin of value-added and carbon emissions in exports. SRIO model merely considers direct trade. By contrast, GIRIO model estimates bilateral carbon emissions flows DCXSRGIRIO,DRCSRSGIRIO,FCTSGIRIO in direct trade and indirect trade via third parties by tracing their origins and destinations of emissions. As domestic emissions in intermediate exports to developing countries may be finally absorbed by developed countries, and imports from developed countries may contain quantitative emissions from developing countries, SRIO analysis potentially underestimates carbon emissions exports to the developed countries while overestimate embodied emissions in exports from the developing countries.

As for environment losses per GDP gained in exports production fragmentation, SRIO model estimates DESSRIO in (5) and FESSRIO in (6), the last of which is actually domestic emissions avoided per value-added reduced by imports. Combining “trade in value-added” and “trade in CO2 emissions” accounts, GIRIO model estimates embodied carbon intensity in each component DEXSGIRIO,DRESGIRIO,FESGIRIO by origins by destinations, which precisely shows each country’s different roles in S′ exports production fragmentation.

Accurate calculation of carbon emissions and embodied carbon intensity in exports is important for understand carbon emissions responsibilities among producers and consumers in different countries, especially for large export-oriented economy. We further empirically compare the two approaches by decomposing carbon emissions and carbon intensity in China’s exports.

3.1 Data

We use the data from national input-output table (SRIO), world input-output table (GIRIO) and Environmental Accounts of WIOD database. Although household emission data are available, we focus on CO₂ emission from production. To facilitate analysis, we merger 41 economies into 8 regions: Mainland China, the EU, USA, Canada (CAN), Australia (AUS), other emerging countries (Brazil, Russia, India, Indonesia, Mexico, Turkey; EMR), other East Asia countries or regions (Japan, Korea, Taiwan; OEA), and the rest of the world (ROW). Also, merge original sector 19, 20, 21 into retail and wholesale, merge original sector 34, 35 into Social and Private Service. New SRIO table contains 1 economy, 32 sectors; new GIRIO table contains 8 economies, each with 32 sectors. Then we use SRIO model and GIRIO model to analyze carbon emissions and carbon intensity in China’s exports in 1995, 2000, 2005 and 2009.

3.2 Carbon Emissions in China’s Exports at National Level

Shown by Table 1 and Table 2, there are differences by SRIO model and GIRIO model in estimation of carbon emissions in China’s exports. And the discrepancies generally increased from 1995 to 2009. Specifically, by considering emissions firstly exported but finally returned home, and different technology in domestic and imported inputs, compared with SRIO model, GIRIO model yields more domestic missions in exports DCSGIRIO, rising from 0.76MT (0.13% more than DCSSRIO) in 1995 to 12.39MT (0.82%) in 2009; and less domestic emissions absorbed abroad DCXSGIRIO, from 0.45% less than DCSSRIO in 1995 to 2.11% in 2009; bias of foreign emissions in exports FCSGIRIO reaching to 85.2MT (over 5% of total emissions in exports) in 2009.

Of carbon emissions in China’s exports by GIRIO model (Table 2), the majority is absorbed abroad DCXSGIRIO, while embodied foreign emissions FCSSRIO still occupy a little share. But the structure is getting optimized, with rising domestic carbon emissions that finally return home DRCSGIRIO and embodied foreign emissions FCSGIRIO. In detail, domestic carbon emissions exports DCXSGIRIO multiplied 1.5 times from 593.77MT in 1995 to 1480.60MT in 2009. Though with declining ratio, DCXSGIRIO still takes up 90% of total carbon emissions in exports. Domestic emissions firstly exported but finally absorbed by home via imports DRCSGIRIO, increased in absolute quantity and relative share, from 3.45MT (0.56%) in 1995 to 44.29MT (2.66%) in 2009. The share of foreign emissions FCSGIRIO induced by China’s exports production also rose sharply, from 41.99MT (6.51%) in 2000 to 147.03MT (9.24%) in 2009, indicating improvement of carbon emissions distribution in China’s exports production fragmentation.

3.3 China’s Carbon Emissions Exports at Bilateral Level

Table 3 illustrates China’s domestic emissions absorbed by each country directly and indirectly DCXSRGIRIO. Although USA and the EU remained major absorbers of carbon emissions in China’s exports, rising demand from other emerging countries (EMR) in absolute amount and relative share needs to be emphasized.

In 1995, about 157.34MT (25.47%) domestic emissions are “consumed” by USA, and 138.16 MT (22.37%) by the EU. The amounts increased rapidly to 343.99MT (20.64%), 327.02MT (20.64%) respectively in 2009. It constitutes major parts of China’s carbon emissions exports. Emissions exports to other East Asia (OEA) also increased in absolute amount but reduced in relative share, from 134.62MT (21.79%) in 1995 to 201.13MT (12.07%) in 2009. On the contrary, carbon emissions absorbed by other emerging countries (EMR) and rest of the world (ROW) increased steadily from 28.38MT (4.59%), 103.95 (16.83%) in 1995 to 176.99MT (10.62%), 335.83MT (20.15%) in 2009. It can be seen final destinations of China’s carbon emissions in exports are slowly shifting from developed countries to other emerging countries. If China is to tackle the problem of carbon leakage, other emerging countries’ demand cannot be neglected.

Figure 2 shows shares of China’s carbon emissions exports to each country by GIRIO model, and corresponding direct bilateral exports shares. Contradiction of exports share change and carbon emissions exports share change indicates indirect emissions flows are concealed in direct trade relationships. Impressively, despite China’s direct exports to EU reduced dramatically from 26.66% to 21.80% from 2000 to 2009, share of carbon emissions exports to EU remained stable. Also, although exports share to USA remained the same from 2000 to 2009, share of carbon emissions exports to USA decreased dramatically. Possibly because besides carbon emissions leakage by direct exports, USA and the EU also absorb China’s emission by importing from other countries; and carbon emissions in direct exports to USA and the EU may finally absorbed by other countries, all of which are indirect carbon emissions exports by third countries. It highlights our work in tracing emissions back to origins and final destinations in GIRIO model.

Figure 2

Share of China’s domestic carbon emission exports to each country and exports to each country

3.4 China’s Returned Carbon Emissions at Bilateral Level

Table 4 shows domestic carbon missions firstly exported to each region but finally returned home via imports DRCSRSGIRIO. Developing countries are making more and more contribution to China’s exports production fragmentation and returned carbon emissions. In 1995, only 1.92MT domestic carbon emissions were firstly exported to OEA and EMR but finally returned back to China. Surprisingly, the amount rose sharply to 20.14MT in 2009, exceeding contribution by USA and the EU (7.93MT). As developing countries are usually at processing in global supply chains, carbon emissions in exports to these countries are likely to be re-exported back to China or other countries.

3.5 Foreign Carbon Emissions in China’s Exports at Bilateral Level

Table 5 shows foreign emissions in imported inputs in the production of China’s exports, resembling China’s carbon emission outsourcing. China mainly transfers environmental pressure in exports production to developing countries, with increasing trend during the period. Specifically, embodied emissions in China’s exports from developing countries (EMR, OEA, ROW) amounted to 14.42MT in 1995, then rose sharply to 117.45MT in 2005, but slightly fell down to 111.2MT (about 6.68%) in 2009. Emissions from developed countries (USA, the EU, AUS and CAN) also increased to 30.49MT in 2009, but with low share (merely 1.83%). Foreign emissions in China’s exports reflect extension of global supply chains in China’s production process, together with China’s ability to transfer carbon emissions.

Figure 3 compares share of foreign emissions in China’s exports based on GIRIO model, and share of imported intermediates goods. Different trends between the two estimations suggest origins of embodied foreign emissions in direct and indirect trade be traced under GIRIO model. Of foreign emissions in China’s exports, carbon emissions from USA and the EU decreased from 11.34%, 11.29% in 1995 to 7.52%, 8.07% in 2009 respectively. However, share of intermediate imports from USA remained relatively stable, and intermediate imports from EU even rose from 6.8% in 2000 to 9.57% in 2009. Also, carbon emissions from OEA in China’s exports maintained at 24%, while imported inputs from EA declined from 43.06% in 1995 to 28.76% in 2009.

Figure 3

Share of China’s carbon exports contain foreign parts from each country and intermediate import from each country

Take EU for example, its exports may contain much foreign emissions, especially emissions from developing countries. It is actually other developing countries’ carbon emissions exports to China via the EU. Therefore, despite growing direct exports, embodied carbon emissions from the EU in China’s exports may even decline. Similarly, despite descending imports from EU, imported emissions in China’s exports could probably increase by growing indirect carbon emissions exports. Under global production fragmentation, bilateral carbon emissions flows should be analyzed by tracing its origins in GIRIO model.

3.6 Embodied Carbon Intensities in China’s Exports

Combing value-added gains and carbon emissions accounts in each component in exports, Table 6 compares environmental cost per value-added in exports production fragmentation in GIRIO and SRIO framework. Generally, domestic carbon intensity in exports in GIRIO model DESGIRIO is slightly higher than DESSRIO in SRIO model but with decreasing differences. The discrepancy decreased from 0.05 MT CO₂ per billion dollars in 1995 to 0.02MT CO₂ per billion dollars in 2009. FESSRIO is meaningless in SRIO model.

Despite dramatic decrease, DESGIRIO is 1.52 times more than FESGIRIO, indicate in exports fragmentation, China specialized in high emissions but low value-added process, while foreign countries in “low-carbon but high value-added” production process.

Specifically, in 1995, DESGIRIO was 4.28 MT CO₂ per billion dollars, over 5 times of FESGIRIO. With technology improvements and official emphasis, DESGIRIO dropped to 1.44 MT CO₂ per billion dollars, and FESGIRIO decreased to 0.57MT CO₂ per billion dollars in 2009. Decrease of DESGIRIO was especially significant during 2005 to 2009, indicating China’s progress in emission reduction during the 11th Five-Year Plan periods. To improve carbon emissions distribution, China still needs to move upstream and reduce emissions cost per GDP gained in exports production fragmentation.