全球变化知识资源中心

International trade has become the fastest growing driver of global carbon emissions, with large quantities of emissions embodied in exports from emerging economies. International trade with emerging economies poses a dilemma for climate and trade policy: to the extent emerging markets have comparative advantages in manufacturing, such trade is economically efficient and desirable. However, if carbon-intensive manufacturing in emerging countries such as China entails drastically more CO₂ emissions than making the same product elsewhere, then trade increases global CO₂ emissions. Here we show that the emissions embodied in Chinese exports, which are larger than the annual emissions of Japan or Germany, are primarily the result of China’s coal-based energy mix and the very high emissions intensity (emission per unit of economic value) in a few provinces and industry sectors. Exports from these provinces and sectors therefore represent targeted opportunities to address the climate–trade dilemma by either improving production technologies and decarbonizing the underlying energy systems or else reducing trade volumes.

Despite international efforts to reduce CO₂ emissions^{1, 2}, global emissions have increased by an average of 3.1% per year since 2000 (refs 3, 4). Economic growth has been identified as the main driver of the sharp increase of CO₂ emissions in the 2000s, and in particular the rapid industrialization of China⁵, which has become the world’s largest carbon emitter since 2006 (ref. 6). However, China is also the world’s largest net exporter of CO₂ emissions embodied in goods and services. In 2007, emissions in China were 7.3 GtCO₂, (production-based emissions), of which 1.7 Gt (23%) were related to goods exported and ultimately consumed in other countries^{7, 8}. In contrast, only 0.2 GtCO₂ emissions were embodied in products imported to China from other countries. As of 2008, Chinese trade accounted for a third of all emissions embodied in global trade, and these traded emissions have been growing faster than global emissions⁹. The magnitude and growth of emissions embodied in Chinese trade pose a dilemma for trade and climate policy: to the extent China and other emerging markets have comparative advantages in manufacturing, international trade is economically efficient and desirable¹⁰. However, if carbon-intensive manufacturing in China entails drastically more carbon emissions than making the same product elsewhere, then trade increases global carbon emissions. Yet, although previous studies have quantified emissions embodied in China’s trade^{7, 11, 12, 13}, none have quantified the underlying factors driving these emissions, leaving open the question of how to mitigate such embodied emissions.

Here, we decompose the key factors contributing to the prodigious imbalance of emissions embodied in China’s international trade (see Methods for details): the large trade surplus between China and its trading partners; the structure of the Chinese economy (that is, specialization in energy-intensive production); the energy mix of China’s production (that is, energy mainly supplied by fossil fuels); and the emissions intensity of Chinese production (that is, the emissions produced per unit of economic output)^{10, 11}. China is a country with substantial regional differences in technology, energy mix and economic development, as well as large volumes of interprovincial trade^{8, 14, 15, 16, 17}, our analysis assessed the magnitude and intensity of emissions from 46 industry sectors (Supplementary Table 1) traded among 30 Chinese provinces/cities and 128 other countries/regions.

Details of the analytical approach are presented in Methods. We track emissions embodied in trade among 158 regions using a global multiregional input–output (MRIO) model of emissions and trade as of the year 2007. The trade and emissions data supporting the model are a combination of the Global Trade Analysis Project (GTAPv8) and province-level input–output tables of China that we have previously constructed^{8, 15, 18}. We analyse the driving factors of emissions embodied in international trade using an improved index decomposition approach (IDA; refs 15, 19). The results presented below and in the figures reflect only international trade. Our model links physical production of emissions with the consumption of final goods without regard for the location of intermediate consumption. For example, emissions related to components manufactured in Inner Mongolia that become part of a product assembled in Beijing and are exported to another country are assigned to Inner Mongolia. If the same final product was exported to another Chinese province, the embodied emissions are consumed domestically and are therefore excluded from our analyses.

Figure 1 shows the top five countries and the top five Chinese provinces whose exports, imports and net trade embody the greatest CO₂ emissions, including the greatest emissions per unit of economic output and per capita. China is the largest net exporter of embodied emissions, by a large margin (Fig. 1g) with eight times more emissions embodied in its exports than its imports (Fig. 1a, d). In contrast, this ratio of emissions embodied in exports to imports is much less in other major exporting nations (for example, 0.5 in the US, 0.5 in Japan, 1.3 in India, 1.2 in Canada, 0.5 in Germany and 1.5 in Australia). All of the 30 Chinese provinces assessed are net exporters of embodied emissions, meaning that in all cases the emissions embodied in exports exceed the emissions embodied in imports Fig. 1 also highlights the significance of particular Chinese provinces; seven of the top ten net exporting regions are Chinese provinces—larger than many large nations (Fig. 1g). Furthermore, the ratio of emissions embodied in exports to imports in these Chinese provinces is immense: 11 of China’s 30 provinces export more than ten times as much emissions as they import, including Xinjiang, Shanxi and Hebei, whose export–import ratios are the largest of any region in our model: 25, 19 and 16, respectively. Five provinces account for 46% of the 1,671 MtCO₂ embodied in China’s exports in 2007: Shandong (178 MtCO₂), Jiangsu (173 MtCO₂), Guangdong (161 MtCO₂), Hebei (139 MtCO₂) and Zhejiang (111 MtCO₂; Fig. 1a).

Top ten regions (including top five countries and top five Chinese cities/provinces) by emissions embodied in exports (a–c), imports (d–f) and net trade (g–i), shown in absolute numbers (a,d,g), per dollar of output (b,e,h) and per capita (c,f,i). Data is for 2007.

Several factors can contribute to the observed differences in the magnitude and intensity of emissions embodied in exports and imports. First, in recent years China has become a ‘factory for the world’, with high concentrations of global heavy industry and manufacturing. For example, China produces 60%, 51% and 65% (by mass) of the world’s cement, steel and coke, respectively²⁴. Such large imbalances in the volume of traded products may correspond to similarly large imbalances in the emissions embodied in traded products. Figure 3 compares emissions embodied in imports and exports by industry sector in China (Fig. 3a) and Europe (Fig. 3b). For example, 34% (26 MtCO₂) of emissions produced by the European metal production industry are embodied in products exported from Europe in 2007, but emissions embodied in all metal products consumed in Europe were 140 MtCO₂, 64% of which (90 MtCO₂) were imported from outside Europe (Fig. 3a; red circle labelled ‘Metal’). In comparison, the share of emissions produced by China’s metal production sector that is exported is similar to Europe’s (33%; Fig. 3b), but the share of emissions related to Chinese consumption of metals that is imported is much lower (11%).

Circles indicate the share of consumed emissions that are imported and the share of produced emissions that are exported for a range of industry categories in Europe (a) and China (b). The size of each circle denotes the sector’s total production emissions, providing an indicator of the relative importance of different sectors. The colours of the circles indicate whether the industries are primary (yellow), secondary and energy intensive (red), secondary and non-energy intensive (purple) or tertiary (green). For a list of sector abbrevations see the Supplementary Information. It should be noted that, although the marker area scale is common across both charts (to aid comparison), the x- and y-axis scales differ. A line representing equal import and export share is shown in each chart. Data is for 2007.

We show that the very large quantities of net emissions embodied exported from China are probably due primarily to Chinese reliance on coal energy and the very high energy intensity of the exporting industries, which are in turn geographically concentrated in a small number of less-developed provinces.

Our analysis is based on aggregated sectors (for example, electronic equipment and machinery) rather than the specific products (for example, iPhones), such that we may underestimate the effect of economic structure on net trade of emissions if differences in production are too specialized to be reflected by the 46 sectors in our model (Supplementary Table 1). The comprehensive data necessary to support product-level analysis are not yet available. However, we also used up-to-date and independent life cycle analysis data sets (PRé SimaPro LCA 7.3 data set²⁶ for Europe and RCEES 2012 database²⁷ for China) to investigate the carbon emission per unit product of the production process for a sample of 15 industrial products made in Europe and China (Table 1). Doing so revealed that the emissions per unit mass of each product (kgCO₂ kg⁻¹) for Chinese products was on average 4.4 times higher than the same products made in Europe, ranging from 1.4 times as high for copper production to 18.4 times as high for propylene production (Table 1).

Full table

Production-based accounting of emissions.

Emissions resulting from combustion of fossil fuels or cement production within a territory, or production-based emissions, are the primary basis for national emission inventories^{30, 31}. For example, the methodology prescribed in IPCC guidelines for greenhouse gas (GHG) emission inventories calculates production-based emissions based on activity data in the region (that is, the amount of energy consumption) and the associated emission factors (that is, GHG emissions per unit energy consumption), the emission factors are based on in situ measurements in which the value is lower than IPCC suggested³⁰.

where i is fuel type, j is sector and k is technology type.

Emission factors can be further disaggregated into the net heating value of a certain fuel ‘V’, its carbon content ‘F’ and the oxidization rate ‘O’.

The detailed calculation process can be found in ref. 30.

Consumption-based accounting of emissions.

An alternative to production-based accounting of CO₂ emissions is to compile inventories according to where related goods and services are ultimately consumed. Such a consumption-based method accounts for inter-regional exchange of energy supply, goods and materials by adding emissions embodied in imports to the production-based total and subtracting emissions embodied in exports.

The emissions embodied in a region’s imports and exports can be calculated using environmentally extended input–output analysis (EIO). Environmentally extended multiregional input–output (MRIO) analysis has been widely developed for calculating the embodied carbon emission^{8, 11, 23}, virtual water^{32, 33}, material use³⁴, biodiversity loss³⁵ and land use^{36, 37} associated with international trade.

In the MRIO framework, different regions are connected through inter-regional trade, Z^rs. The technical coefficient sub-matrix A^rs consists of the elements [a_ij^rs], derived from a_ij^rs = z_ij^rs/x_j^s, where z_ij^rs is the inter-sector monetary flow from sector i in region r to sector j in region s; x_j^s is the total output of sector j in region s. The final demand matrix Y consists of the elements [y_i^rs], where y_i^rs is the region’s final demand for goods of sector i from region r. Therefore, MRIO analysis can be represented as

Using familiar matrix notation and dropping the subscripts, equation (1) can be written as: x = Ax + y or x = (I–A)⁻¹y, where (I–A)⁻¹ is the Leontief inverse matrix that captures both direct and indirect inputs required to satisfy one unit of final demand in monetary value; I is the identity matrix. To calculate the consumption-based CO₂ emissions, we then extend the MRIO table with sector-specific CO₂ emissions: E = k(I–A)⁻¹y, where E represents the total CO₂ emissions embodied in goods and services used for final demand and k is a vector of CO₂ emissions per unit of economic output for all economic sectors in all regions.

Index decomposition analysis of emissions embodied in trade.

The index decomposition of trade embodied CO₂ emissions is given by

where E describes CO₂ emissions embodied in imports or exports, Q is the GDP value of imports or exports, S_i refers to the share of the GDP value for sector i, I_i refers to the energy intensity of sector i and F_i refers to the emission per unit of energy consumption of of sector i (i for 46 sectors). Thus, the factors contributing to a net trade in embodied emissions can be expressed based on the logarithmic mean divisia index (LMDI) approach (additive form) ¹⁹as:

where ΔE is the difference between the CO₂ emissions embodied in exports (E^export) and the CO₂ emissions embodied in imports (E^import); ΔE_act, ΔE_str, ΔE_int and ΔE_str refer to economic scale effect, economic structure effect, sector intensity effect and energy mix effect, respectively, where ΔE_act, ΔE_str, ΔE_int and ΔE_str are expressed as:

Q^t, S^t, Iⁱ and F⁰ are the GDP, GDP share, energy intensity and emission coefficient of exports, respectively. Q⁰, S⁰, I⁰ and F⁰ are th GDP, GDP share, energy, intensity and emission coefficient of imports, respectively.

Estimates of sectoral level imported and exported CO₂ emissions.

In a region IO model, a regional economy is considered as its system boundary; thus exports are treated as final products in a region’s economy. Let G_i^r be the total CO₂ emissions in economic sector i and region r, thus Σ_iG_i^r represents the production-based emissions in region r. In each region r, there is an intermediate consumption, denoted Z_ij^r, which represents the domestic purchases of sector i by sector j in region r, and a final consumption, denoted y_i^r, which represents the domestic purchases of sector i by final consumers in region r, which includes households, government, capital investments. In the single region IO model, exports, e_i^rs, from region r to region s are also treated as final consumption. By summing intermediate and final consumption, we can obtain the total output in each region:

By assuming fixed production ratios, we obtain the technical coefficients, A_ij^rr, the ratio of input to output, by dividing Z_ij^rr by x_j^r:

Thus, equation (2) can be re-written as:

where (I–A^rr)⁻¹ is Leontief inverse matrix for region r.

CO₂ emissions are estimated based on the direct emission intensity k^r in each sector in region r.

Therefore, the total embodied emissions (direct and indirect) in exports from region r to region s can be calculated by:

where Exp^r is a vector of embodied CO₂ emissions in sectoral exports of region r to region s; k^r is a row vector of sectoral emissions intensities in region r; is a matrix with sectoral export from region r to region s on diagonal.

In turn, the total embodied emissions in imports from region s to region r can be estimated by:

URL: