Pattern changes in determinants of Chinese emissions

The environmental input–output analysis (IOA) method is an established life cycle assessment (LCA) and carbon emissions analysis approach (Meng et al 2015, Mi et al 2017a, Mi et al 2017b). It can be mathematically expressed as follows:

$$\begin{equation} C = F(I - A)^{ - 1} y, \end{equation} \tag{ 1 }$$

where (I− A)⁻¹ is the Leontief inverse matrix, C are the total CO₂ emissions, F is a row vector of carbon emissions intensities (CO₂ emissions per unit of economic output) for each economic sector, I is the identity matrix, A is a matrix denoting the monetary relationship between different sectors of the economy, and y is a column vector of final demand for each economic sector (Zhang et al 2015).

A country's CO₂ emissions change over time for various reasons. Five factors (population, efficiency, production structure, consumption patterns and consumption volume) are considered to assess changes in China's CO₂ emissions via an SDA, which is represented as follows:

$$\begin{equation} \begin{array}{*{20}c} {\Delta C} \hfill & = \hfill & {\Delta pFLy_s y_v + p\Delta FLy_s y_v + pF\Delta Ly_s y_v } \hfill \\ {} \hfill & + \hfill & {pFL\Delta y_s y_v + pFLy_s \Delta y_v,} \hfill \\ \end{array} \end{equation} \tag{ 2 }$$

where

∆: the change in a factor;

p: a scalar that represents the population;

F: a row vector that represents carbon emissions intensities;

L: the Leontief inverse matrix, L = (I− A)⁻¹;

y_s: a column vector that represents per capita consumption patterns; and

y_v: a scalar that represents the per capita consumption volume.

Each of five terms in equation (2) denotes the contributions to CO₂ emissions changes that are triggered by one driving force if the rest of the variables are kept constant. When conducting an SDA, evaluating different terms for the start or end point of the investigated time period is possible. There are several methods for dealing with this issue. For example, Su and Ang (2012) summarized four SDA methods and pointed out their pros and cons. In this study, we take the average of all possible first-order decompositions to address this issue. Five factors are considered in this paper, so there are 5! = 120 forms. For a detailed discussion, see Dietzenbacher and Los (1998) and Hoekstra and Van Den Bergh (2002).

Based on our previous studies (Mi et al 2017b, Mi et al 2016, Shan et al 2016a, Shan et al 2016b), we adopted the Intergovernmental Panel on Climate Change (IPCC) administrative territorial CO₂ emissions inventories for China's emission accounting in this study. The inventories compiled in this study include two parts: emissions from fossil fuel combustion and from cement production. For CO₂ emissions from fossil fuel, we calculate the emissions by (IPCC 2006, Liang et al 2016):

$$\begin{equation} C_{\hbox{fossil}} = D_{\hbox{fossil}} \times N \times H \times O, \end{equation} \tag{ 3 }$$

where C_fossil are the CO₂ emissions from fossil fuel, D_fossil is the fossil fuel consumption (in physical units), N is the net calorific value that represents heat released when the unit fossil fuel is combusting, H is the carbon content that represents CO₂ emitted when unit heat is released, and O is the oxygenation that represents the oxidization rate of fossil fuel combustion. N × H × O is defined as emission factors. See table S1 available at stacks.iop.org/ERL/12/074003/mmedia in the supplementary information for all emission factors for fuel combustion.

To avoid missing or double accounting, we calculate the fossil fuel consumption by (Peters et al 2006):

$$\begin{equation} \begin{array}{*{20}c} {D_{\hbox{fossil}} } \hfill & { = \hbox{Final consumption} - } \hfill \\ {} \hfill & {\hbox{Used as chemical material } - \hbox{ Loss}} \hfill \\ {} \hfill & { + \;\hbox{Input for thermal power} + \hbox{Input for heating}} \hfill \\ \end{array} \end{equation} \tag{ 4 }$$

The CO₂ emissions from cement production are calculated by (IPCC 2006):

$$\begin{equation} C_{\hbox{cement}} = D_{\hbox{cement}} \times E, \end{equation} \tag{ 5 }$$

where C_cement are the CO₂ emissions from cement production, D_cement is the cement production amount, E is the emission factor of cement production that represents CO₂ emit when producing unit cement. The emission factor for the cement process is 0.2906 tonne CO₂ per tonne cement produced (Liu et al 2015).

Due to the differences in technology, the carbon intensity of China's imports is different from that of its domestic products. Many previous studies assumed that the imports to China were produced with Chinese technology (Guan et al 2008). This approach may cause large errors in estimating the emissions embodied in imports (Meng et al 2016). In this paper, we link China's imports to a global multi-regional input–output (MRIO) model which are based on the Global Trade and Analysis Project (GTAP) database. China's imports in each sector are divided into all other regions of the world according to the GTAP database. The CO₂ emissions embodied in imports are calculated using the global MRIO model:

$$\begin{equation} C_{\hbox{import}} = \overline F (I - \overline A )^{ - 1} y_{\hbox{import}}, \end{equation} \tag{ 6 }$$

where D_import are the embodied CO₂ emissions in imports, $\overline{F}$ is a row vector of carbon emissions intensities for all sectors in all regions, $\overline{A}$ is the direct requirement matrix for the global MRIO model, y_import is a column vector of China's imports from all sectors in all regions.

Several recent studies analyzed supply chain relationships in the export processing sectors which can substantially affect the emissions attributed to exports. The processing exports, which generate relatively little value added but also relatively little emissions, account for a large proportion of China's total exports. Therefore, the CO₂ emissions embodied in China's exports are likely to be overestimated if export processing is not taken into consideration (Dietzenbacher et al 2012).

Two main datasets were examined in this study: time-series IO tables and corresponding CO₂ emissions data. China IO tables for 2005, 2007, 2010, and 2012 were published by the National Bureau of Statistics (NBS) of China and can be freely downloaded from the NBS website (National Bureau of Statistics 2016). NBS regularly publishes China's national input–output tables based on national surveys every five years. In this study, 2007 and 2012 IO tables are the surveyed ones. In between the five-year period, NBS updates extend-tables to reflect new changes in economic production and consumption based on economic upscaling and balancing methods. The adopted 2005 and 2010 IO tables by this study are the extended version. The production structure changes in extended IO tables cannot be fully represented, but these tables capture essential interdependences among different economic sectors.

We deflated all the tables to 2012 prices using the double deflation method (UNSD 1999). The pricing data for China's IO tables were gotten from the China Statistics Yearbook (National Bureau of Statistics 2007, 2011, 2015b), while the pricing data for China's imports and global MRIO tables were obtained from the National Account Main Aggregates Database (UNSD 2016). In addition, the import value of China was deflated by the weighted average price deflators of all regions (excluding China). The weight for one region was in proportion to China's imports from the region. It needs to be noted that there are several drawbacks related to the double deflation method, although this method is widely accepted. Firstly, by adopting this method, most sectors are assumed to produce one homogeneous product, each sector's gross output and intermediate and final demand are deflated by this sector's price index (Dietzenbacher and Hoen 1998). However, most sectors consist of more than one good; therefore to use the price index of certain goods to represent the entire sector is not always appropriate (Sevaldson 1976). Secondly, value-added is obtained as the difference between the total input and intermediate input in each sector. Consequently, it is not accurate to use value-added to balance the IO table after the deflation (Wolff 1994).

Fossil fuel consumption, cement production and emission factors were necessary to compile CO₂ emissions inventories. We used the most recent Chinese energy data published in China Energy Statistical Yearbooks 2014 and 2015 (National Bureau of Statistics 2014, 2015a). In these recent datasets, there are unusually large adjustments on Chinese energy data (EIA 2015). The cement production amount was collected from the China Statistical Yearbook 2015 (National Bureau of Statistics 2015b). We used emission coefficients from our previous studies (Liu et al 2015). The coefficients are measured based on 602 coal samples from the 100 largest coal-mining areas in China and are assumed to be more accurate than the IPCC default value. All the inventories used in this study can be freely downloaded from the website of China Emission Accounts and Datasets (CEADs) (www.ceads.net/) (Mi et al 2017b, Mi et al 2016, Shan et al 2016a). China's CO₂ emission inventory for 47 sectors for 2000–2015 is shown in the online supplementary information. The inventory includes CO₂ emissions from fossil fuel combustion and cement production.

Global MRIO tables were obtained based on version 9 of the GTAP database, which describes bilateral trade patterns, production, consumption and intermediate use of commodities and services among 140 regions for 57 sectors for the years 2004, 2007 and 2011 (Dimaranan 2006, Narayanan et al 2015). See table S2 in the supplementary material for the concordance of sectors for Chinese IO tables, carbon emission inventories and GTAP MRIO tables.

From 2005 to 2010, Chinese production-based CO₂ emissions increased by 47% from 5159 million tonnes (Mt) to 7604 Mt (figure 1, black curve). Clearly, efficiency was a strong factor in offsetting emissions during 2005–2010 (figure 1, blue curve). The efficiency gains offset 39% of emissions by keeping other driving forces constant. In fact, Chinese carbon intensity decreased by 10% over this period (figure 2(a), black curve). From a sectorial perspective, carbon intensity in agriculture and industry decreased by 24% and 30%, respectively, from 2005 to 2010. See figure S1 for changes in the carbon intensity levels of all 20 sectors.

Zoom In Zoom Out Reset image size

Zoom In Zoom Out Reset image size

However, the other four factors drove emissions growth from 2005 to 2010. Per capita consumption alone drove 36% of emission increases with other factors held constant, and production structural changes increased emissions by 34% compared with 2005 levels (figure 1, green). Changing consumption patterns and population growth were relatively weak factors, causing approximately 14% and 4% of emissions increases, respectively.

The global financial crisis had impacts on the determinants of Chinese carbon emissions changes. First, the CO₂ emissions embodied in Chinese exports from 2007 to 2010 dropped (figure 3(b)). Investments, consumption and exports form the triumvirate that supports Chinese economy. China has become a 'factory to the world'; thus, exports play a critical role in promoting its economic development. Export-induced CO₂ emissions increased by 214 Mt from 2005 to 2007, which accounted for about 20% of total emissions growth for that period (figure 3(a), blue section). However, China's exports sharply declined after the global financial crisis. As a result, CO₂ emissions induced by exports from 2007 to 2010 declined (figure 3(b), blue section). Several sectors were affected greatly. For instance, CO₂ emissions induced by metal product exports fell by 26% from 361 Mt in 2007 to 267 Mt in 2010. CO₂ emissions induced by textiles exports also declined by 15% over the same period (figure S2). Chinese export-induced emissions represented 24% of Chinese CO₂ emissions in 2010, decreasing from levels in 2005 (32%) and 2007 (30%) (figure 2(b), blue bar).

Zoom In Zoom Out Reset image size

On the contrary, Chinese CO₂ emissions induced by capital formation from 2007 to 2010 grew rapidly mainly due to the government's response to the financial crisis. Chinese exports were severely affected by the financial crisis; as such, the government increased its investments on capital formation to promote its economic growth. The four trillion yuan stimulus package is mainly focused on fixed assets and construction, such as the high-speed rail network, rural infrastructure, urban electrical grid and post-quake reconstruction. As a result, approximately 71% of the CO₂ emissions growth from 2007 to 2010 was due to capital formation (figure 3(b), yellow section), which was much higher than those during 2005–2007 (61%) and 2007–2012 (56%). From a sectorial perspective, the CO₂ emissions induced by the capital formation in terms of construction rose by 27% from 1762 Mt in 2007 to 2246 Mt in 2010. The CO₂emissions induced by the capital formation with respect to transport equipment also increased by 86% (156 Mt) during the same period, which was much higher than those for 2005–2007 (40 Mt) and 2010–2012 (17 Mt) (figure 3). One of the reasons for the emission increase in transport equipment is the rapid development of the high-speed railway. The length in operation of the high-speed railway increased by 664% from 672 km in 2008 to 5133 km in 2010, and its passenger traffic also grew from 7 million in 2008 to 133 million in 2010 (National Bureau of Statistics 2015b). The proportion of capital formation-induced emissions of the total CO₂ emissions grew from 41% in 2007 to 48% in 2012. It can be seen from the figure 2(b) that the switch from exports to capital formation can be observed through the entire period (2005–2012), but the financial crisis made it faster.

In addition, Chinese direct household CO₂ emissions declined from 2006 to 2008. With improvements in Chinese living standards, direct household emissions increased before 2006. However, direct household emissions declined by 11% from 302 Mt in 2006 to 270 Mt in 2008 (figure 2(c)). It was mainly caused by the decreases of household oil consumption. Moreover, the direct CO₂ emissions per capita of urban residents were greater than those of rural residents in China, but the gap was diminishing. The direct CO₂ emissions per capita of urban residents were 42% higher than those of rural residents in 2005 and 15% higher in 2012.

China has struggled to achieve a 'new normal' since the global financial crisis. According to China's traditional development model, production structure and consumption patterns drive Chinese CO₂ emissions growth, whereas efficiency gains constitute the main factor that offsets emissions (Guan et al 2009, Peters et al 2007). We found that production structure and consumption patterns caused 34% and 14% increases, respectively, in China's CO₂ emissions from 2005 to 2010, whereas efficiency levels induced a 39% decrease in Chinese CO₂ emissions over the same period (figure 1).

However, the effects of these factors on Chinese CO₂ emissions have changed considerably under the 'new normal'. First of all, as the strongest factor offsetting emissions from 2005 to 2010, efficiency drove a 1.4% increase in emissions from 2010 to 2012 (figure 4, blue bar). Almost all studies that focused on China's emission changes during the time periods prior to 2010 have shown the great contributions of efficiency to offset CO₂ emissions (Chang and Lahr 2016, Guan et al 2009). Peters et al (2007) summarized China's CO₂ emission changes as a race between increasing consumption and efficiency gains. However, China's lost its advantages of efficiency improvement during 2010–2012. This effect mainly resulted from efficiency losses in the transport and industrial sectors. Although the mean carbon intensity of the Chinese economy declined slightly from 2010 to 2012, the carbon intensity of the transport and industrial sectors increased by 15% and 2%, respectively.

Zoom In Zoom Out Reset image size

By contrast, changes in production structure (figure 4, orange bar) and consumption patterns (figure 4, purple bar) caused 2.6% and 1.3% decreases in Chinese CO₂ emissions, respectively, from 2010 to 2012. The two factors both drove increases in CO₂ emissions during 2005–2010. From the production perspective, the effects of production structure were mainly driven by the reduction of proportions of carbon-intensive sectors in total inputs, i.e. inputs from transport, electricity, gas, and nonmetallic mineral products sectors to other production sector. The sector of electricity had the highest carbon intensity among all economic sectors, and its proportion as inputs to other sectors declined from 5.1% in 2010 to 4.6% in 2012 (figure S3). Further, the proportion of electricity sector as inputs to mining sector declined from 13.0% to 11.4% during 2010–2012. The proportion of transport in total inputs also declined from 5.6% to 4.6% over the same period. Further, the proportion of transport as inputs to the sector of wholesale, retail and catering declined from 12.7% to 8.1% during 2010–2012.

From the consumption perspective, the effects of consumption patterns were mainly due to the reduction of proportions of electrical equipment and chemicals in final use, because the embodied carbon emission intensity (i.e. direct and indirect CO₂ emissions induced by per unit of final use) in these sectors was relatively higher (figure S4). For example, the proportion of electrical equipment in total consumption-based CO₂ emissions declined from 5.9% in 2010 to 4.7% in 2012, and the proportion of chemicals in total consumption-based emissions also decreased from 3.4% to 3.2% over the same period (table S3).

Another change in the 'new normal' is that the proportion of CO₂ emissions induced by consumption (including household and government consumption) has begun to increase. Prior to 2010, the proportion of consumption induced CO₂ emissions had a declining trend. However, it grew from 28.6% in 2010 to 29.1% in 2012. Although it is a very slight increase, it reflects Chinese efforts of expanding domestic demand and stimulating consumption. This shift was mainly because of changes in China's investment and consumption structure (figure S5). China's consumption rate (i.e. the proportion of final consumption in GDP) had declined since 2000 until reaching the lowest point with 48.5% in 2010. Meanwhile, its investment rate (i.e. the proportion of capital formation in GDP) peaked. After that, its consumption rate increased to 51.6%, while the investment rate declined to 44.9% in 2015. According to the report of the State Information Center, China will shift gradually from an investment to a consumption-driven economy, and its consumption rate will continue to increase until peaking about 60%–70% (State Information Center 2016). This shift will result in increases in the proportion of CO₂ emissions induced by consumption.