G Model
ARTICLE IN PRESS
STRECO-743; No. of Pages 9
Structural Change and Economic Dynamics xxx (2018) xxx–xxx
Contents lists available at ScienceDirect
Structural Change and Economic Dynamics journal homepage: www.elsevier.com/locate/sced
How does foreign trade influence China’s carbon productivity? Based on panel spatial lag model analysis Lulu Zhang, Lichun Xiong, Baodong Cheng ∗ , Chang Yu ∗ School of Economics and Management, Beijing Forestry University, Beijing, China
a r t i c l e
i n f o
Article history: Received 20 March 2018 Received in revised form 19 July 2018 Accepted 28 August 2018 Available online xxx Keywords: Foreign trade Carbon emissions Carbon productivity Spatial lag model
a b s t r a c t Foreign trade is one of the main sources of China’s carbon emissions. Carbon productivity has become an important approach to measure the level of low carbon development, because it can integrate the goals of economic development and carbon reduction. Therefore, it is of great significance to understand the influencing mechanisms of foreign trade on China’s carbon productivity to improve trade quality and low carbon development. This research measured China’s carbon productivity and analyzed the effect of foreign trade on carbon productivity through applying spatial lag model with the data of 30 provinces from 2000 to 2014. The results show that China’s carbon productivity had obviously positive spatial spillover effect. Export-import significantly improved China’s carbon productivity and mainly through import. As for the regional regression results, foreign trade of the western region contributed the most to enhance China’s carbon productivity, while the eastern and central regions were at the stage that the foreign trade inhibited carbon productivity. Policy recommendations were put forward in terms of the emphasis of import trade, the optimization of trade structure, the customized measures in different regions to improve carbon productivity. Furthermore, provinces should coordinate with each other and eliminate inter-regional barriers to promote the market-oriented distribution of production factors, in order to realize the spillover effect of carbon productivity. © 2018 Elsevier B.V. All rights reserved.
1. Introduction As the global environmental problems have become increasingly prominent, the low carbon oriented development model has attracted the attention of the world. China is the world’s largest carbon dioxide (CO2 ) emitter. According to the Report on Development of Low-Carbon Economy in China 2014, 29% of China’s CO2 emission was caused by exports (Zhao and Yan, 2014). Indeed, export, as one of the three engines to drive China’s economy, has made a significant contribution to the economic growth. In 2013, China became the world’s largest trading power in goods. China’s value of foreign trade had reached $3194.4057 billion in 2015. After Paris Agreement, the urgent global situation of climate change also makes China confront enormous challenges and difficult tasks. Although up to 2015, China had overfulfilled the targets of emission reduc¨ tion formulated by 12th Five Year¨, the greater efforts still need to be made to achieve the voluntary commitments of carbon reduction in Paris Agreement: the intensity of CO2 per gross domestic product (GDP) need to drop by 60%–65% until 2030 comparing with 2005
∗ Corresponding authors. E-mail addresses:
[email protected] (L. Zhang),
[email protected] (L. Xiong),
[email protected] (B. Cheng),
[email protected] (C. Yu).
and meanwhile CO2 emission ought to reach a peak around 2030. Moreover, there is still a gap in carbon productivity between China and the developed countries (He and Su, 2011). Under the increasingly stringent constraints for carbon emission, how to achieve low carbon development of foreign trade has become a challenge for China. Carbon productivity is a core criterion of low carbon development, thus improving the carbon productivity of foreign trade is the key issue (He and Su, 2011; Li et al., 2018). Therefore, studying the influencing mechanisms of foreign trade on China’s carbon productivity has great practical significance to improve trade quality and low carbon development. This research aims to understand the relationship between foreign trade and carbon productivity in China. The remainder of this research is as follows. Section 2 is the literature review about the impact of foreign trade on carbon productivity. Section 3 is the research design and introduces the method to calculate carbon productivity, Moran’I index to verify the spatial correlation of China’s carbon productivity. The spatial econometric model is also described as well as the explanations of data and variables. The detailed analysis of China’s national and regional carbon productivity is conducted in Section 4 based on spatial lag model. In the end, we summarize the main conclusions and put forward the policy recommendations for improving China’s carbon productivity from the perspective of foreign trade.
https://doi.org/10.1016/j.strueco.2018.08.008 0954-349X/© 2018 Elsevier B.V. All rights reserved.
Please cite this article in press as: Zhang, L., et al., How does foreign trade influence China’s carbon productivity? Based on panel spatial lag model analysis. Struct. Change Econ. Dyn. (2018), https://doi.org/10.1016/j.strueco.2018.08.008
G Model STRECO-743; No. of Pages 9
ARTICLE IN PRESS L. Zhang et al. / Structural Change and Economic Dynamics xxx (2018) xxx–xxx
2
2. Literature review The concept of carbon productivity was firstly proposed by Kaya and Yokobori (1997). It refers to the ratio between GDP and CO2 emissions. Currently, the research on carbon productivity mainly focuses on two aspects: the first is the importance of improving carbon productivity to low carbon economy; the second is the factor decomposition of carbon productivity. Beinhocker et al. (2008) pointed out that the key to reduce global carbon emissions was to increase carbon productivity. They believed that only the carbon productivity increase by at least 10 times as much as that in 2005 can we achieve the goal of reducing global CO2 emissions by 50% in 2050. He and Su (2011) also stressed the importance of carbon productivity. They decomposed the influencing factors into three parts to improve carbon productivity: adjustment of industrial structure, optimization of energy structure and improvement of energy technical efficiency. Using the improved stochastic frontier model, Zhang and Wang (2014) decomposed the growth rate of carbon productivity into seven factors. The results showed that technological progress, substitution effect between capital and energy and substitution effect between labor and energy were the main factors influencing the growth rate of carbon productivity. Moreover, Chen et al. (2018) analyzed carbon productivity by Logarithmic Mean Divisia Index (LMDI), and explored the growth drivers of carbon productivity in China’s electricity sector from the perspective of production and consumption. With the same decomposition method (LMDI), the source of carbon productivity change in Australian construction industry was also investigated in Hu and Liu (2016)’s study. Given the above, it also can be found that carbon productivity is popularly used over the world in measuring the emission performance of an economy. Based on the definition of carbon productivity, the impact of foreign trade on carbon productivity essentially refers to the impact of foreign trade on environment and productivity. The environmental impact of foreign trade can be traced back to Grossman and Krueger (1994) which indicated that foreign trade would influence environment through scale effect, technical effect and structural effect. The total effect depends on the sum of these three effects. Since then, plenty of scholars have studied the positive and negative effects of international trade on environment. On the one hand, trade has made it possible for trading countries, especially the developing countries, to learn more advanced clean technologies, which may generate positive effect on domestic environment. For instance, Antweiler et al. (2001) decomposed the environmental effects of trade and concluded that foreign trade was beneficial to reduce domestic sulfur dioxide (SO2 ) concentration. On the other hand, trade liberalization may intensify enterprises’ competition, therefore countries are likely to reduce environmental regulations on domestic enterprises, so as to decrease their cost of export products, in order to strengthen their competitiveness in international mar¨ environmental standards, ket. This results in a r¨ ace to the bottomof which undoubtedly exerts a negative influence for environmental performance (Korves et al., 2011; Revesz, 1992). Indeed, the debate about the positive and negative effects of trade on CO2 emissions has obtained continuous attention. Sandrini and Censor (2009) estimated the impact of trade on environmental quality using the instrumental variables with the emission of SO2 and CO2 . It was found that trade was beneficial to the environment in Organization for Economic Co-operation and Development (OECD) countries, but it had a detrimental effect on non-OECD countries. It implies that the effect of trade depends on the different situation of countries. Naranpanawa (2011) investigated the relationship between trade and carbon emissions in the case of Sri Lanka by the approaches of autoregressive distributed lag bounds testing and the JohansenJuselius maximum likelihood. The results showed that there existed a short-term uni-directional causality from trade openness to CO2
emissions, meaning that trade would boost CO2 emissions. In the study of Knight and Schor (2014), they analyzed the impacts of import and export on CO2 emissions over the period 1991–2008 with a balanced panel data set of 29 high-income countries. The results indicated that import had a positive effect on per capita carbon emissions generated by consumption activities, whereas exports had the negative consequence. While Guo and Liu (2011) examined the relationship between China’s export and carbon emissions and concluded that export had driven carbon emissions between 1978–2009. The concerns about the impact of trade on productivity started from endogenous growth theory. Emphasizing the technological spillover of foreign trade, endogenous growth theory takes the research and development (R&D) and foreign trade as the engines of technical progress and productivity improvement (Grossman and Helpman, 1990; Rivera-Batiz and Romer, 1991). In an open economy, technological progress not only depends on the domestic R&D and investment stock, but also the spillover effect of foreign investment and R&D that are generated mainly through foreign trade (Gillard and Lock, 2017; Seck, 2012). Coe and Helpman (1993) proved that the spillover effect of R&D and investment did exist in trade among the OECD countries and Israel. This external spillover effect significantly improved the total factor productivity. It was also discovered by Almodóvar et al. (2014) from the analysis of the manufacturing enterprises in Spain that learning effect would be generated in the process of importing advanced foreign products and services so as to realize technological progress and productivity improvement. In summary, foreign trade certainly has impact on carbon productivity but the quantitative economic relationship is relatively uncertain. At present, a large number of literatures separately pay attention to carbon productivity and the impact of foreign trade on carbon emissions, but few literatures directly discuss the relationship between foreign trade and carbon productivity. Compared with carbon emission, carbon productivity takes both economic development and carbon emissions into considerations. Therefore, studying the effect of foreign trade on carbon productivity is more in line with China’s reality for developing low carbon economy. More in-depth research should be conducted in this direction. Among the few research on the impact of foreign trade on carbon productivity, Zhao and Zhang (2016) constructed the static and dynamic panel models to analyze 88 countries and then concluded that foreign trade was advantageous to improve carbon productivity. This makes important foundation for our research. But their analysis did not consider the significant spatial geographic attribute of carbon emission. Meanwhile, low carbon technology also has regional spillover effect, leading to the spatial correlation of carbon productivity (Poon et al., 2006; Wang et al.,2018). If the effect of spatial correlation is neglected, the econometric models’ estimation may become biased or generate erroneous parameters (Anselin and Griffith, 2010). Consequently, this research involves spatial factors and takes spatial lag model to analyze the impact of foreign trade on China’s carbon productivity. 3. Study design 3.1. Calculating method of carbon productivity Carbon productivity refers to the ratio of GDP and CO2 emissions during the same period (Kaya and Yokobori, 1997). The formula is as follows: y cpit = it (1) cit where cp refers to carbon productivity; y is total output value; c is carbon emissions; i and t refer to province and year, respec-
Please cite this article in press as: Zhang, L., et al., How does foreign trade influence China’s carbon productivity? Based on panel spatial lag model analysis. Struct. Change Econ. Dyn. (2018), https://doi.org/10.1016/j.strueco.2018.08.008
G Model
ARTICLE IN PRESS
STRECO-743; No. of Pages 9
L. Zhang et al. / Structural Change and Economic Dynamics xxx (2018) xxx–xxx Table 1 Carbon emission coefficients of various energies.
3.3. Model specification
Energy Type
NCV (kJ/kg)
CC (kgC/GJ)
COF
e (tC/t)
coal coke gasoline diesel oil fuel oil natural gas kerosene
20908 28435 43070 42652 41816 38931 43070
25.8 29.2 18.9 20.2 21.1 15.3 19.5
1 1 1 1 1 1 1
0.5394 0.8303 0.8140 0.8616 0.8823 0.5956 0.8399
Data Source: China Energy Statistical Yearbook.
tively. To exclude the price factor, GDP of each province in China has been deflated to constant price referring to the year of 2000. Since the volume of CO2 emission of China’s each province cannot be directly derived, we estimate China’s provincial CO2 emission volume based on the method provided by Intergovernmental Panel on Climate Change (IPCC, 2007). Carbon emissions are calculated as: cijt = ijt × ej × 44 , where denotes energy consumption; e denotes 12 CO2 emission coefficient; j denotes the types of energy consumption. The numbers of 44 and 12 represent the chemical molecular weight of CO2 and carbon, respectively. For energy consumptions, we have selected 7 final energies – coal, coke, gasoline, kerosene, diesel oil, fuel oil and natural gas from China Energy Statistical Yearbook.1 The formula of carbon emission coefficient formula is: ej = NCVj × CCj × COFj , where NCV means the net calorific value; CC means the carbon content of energy and COF means the carbon oxidation factor of energy (Pan and Zhang, 2011). The carbon emission coefficients of various energies are as shown in Table 1.
3.2. Moran Index When determining whether to use spatial econometric method, we should firstly verify the spatial dependence. If the spatial dependence exists, the spatial econometric method can be used. Otherwise the regular econometric methods should be used (Anselin and Bera, 1998). Hence, global Moran’s I is adopted to test global spatial autocorrelation on carbon productivity, so as to verify whether the carbon productivity has spatial dependence. The formula is as follows:
n n I=
i=1
w (x − x¯ )(xj j=1 ij i n (x − x¯ )2 i=1 i
3
This paper has established the following regression model based on the above analysis and the related empirical models in the literature (Zhao and Zhang, 2016): cpit = ˛0 + ˛1 trade + ˛2 X + i + t + εit
(3)
where i and t refer to province and period, respectively. ˛0 , ˛1 and ˛2 are the parameters to be estimated. Trade is the core explanatory variable, referring to foreign trade which is expressed by exportimport dependence, import dependence and export dependence. X represents the control variables. i refers to the individual effect not affected by time, while t represents the time effect which has an impact on all provinces in the same period. εit is the random error. Due to the obvious spatial correlation and spatial spillover effect of carbon productivity, the traditional econometric methods may lead to the biased estimation (Anselin and Griffith, 2010). Therefore, spatial econometric models should be adopted. The regular spatial econometric models include Spatial Lag Model (SLM), Spatial Error Model (SEM), and Spatial Durbin Model (SDM). Using which model needs further tests. In this research, the spatial factor has been considered based on regression Eq. (3) to establish the following econometric model: yit =
n
X + Wij yjt + ˇ it
j=1
n
+ Wij X jt i
j=1
(4)
i = W + εi where i and j refer to the different areas; Wij refers to N×N order
is the independent variable vector spatial weights matrix; X it (including the core explanatory variables and the control vari is the regression coefficient of independent variable ables) and ˇ vector; denotes the spatial regression coefficient of the dependent variables; denotes the spatial regression coefficient vector of independent variables. represents the spatial error regression coefficient. / 0 and = = 0, formula (4) is Spatial Lag Model; when When = = = 0 and = / 0, formula (4) is Spatial Error Model; when = / 0, = / 0 and = 0, formula (4) is Spatial Dubin Model. Through Wald test and Lratio test, we have found that Spatial Lag Model is suitable for our econometric model (see Table 4). 3.4. Data sources and variables selection
− x¯ )
(2)
where wij is element (i,j) in the spatial weights matrix to measure the distance between area i and area j. If area inand jnhave common w refers to boundary, then wij = 1, otherwise wij = 0.2 i=1 j=1 ij the sum of all spatial weights. The value of Moran’s I ranges from -1 to 1. If the value is less than 0, it indicates a negative spatial correlation. If the value is greater than 0, it indicates a positive spatial correlation. And there is no spatial correlation if the value is equal to 0.
1 There are nine types of energy consumption involved in China Energy Statistics Yearbook, including coal, coke, crude oil, gasoline, kerosene, diesel oil, fuel oil, gas and electricity. Since most of the crude oil is used for processing and conversion (e.g., oil refining), and electricity does not directly generate carbon dioxide during the usage. Thus we have excluded crude oil and electricity to avoid redundant computation. 2 Hainan province is an island that closely associated with Guangdong province in economy, so this paper assumes that the two provinces are adjacent.
The sample areas include 30 provinces of China.3 The data period is from the year of 2000 to 2014. The data sources mainly include China Statistical Yearbooks, China Energy Statistical Yearbooks, China Compendium of Statistics 1949–2008 and Annual Statistical Yearbooks of each province. The description of all the variables are shown in Table 2 with 450 observations. 3.4.1. Explained variable Carbon productivity (cp) is the explained variable in our model. It is calculated by Eq. (1). 3.4.2. Explanatory variables Foreign trade (trade). It is reflected by three indicators: exportimport dependence (xm) (i.e., the proportion of export-import value in GDP), import dependence (imp) (i.e., the proportion of
3 Due to the missing data, Tibet Autonomous Region, Taiwan Province, Hong Kong Special Administrative Region, Macao Special Administrative Region and South China Sea Islands are not included.
Please cite this article in press as: Zhang, L., et al., How does foreign trade influence China’s carbon productivity? Based on panel spatial lag model analysis. Struct. Change Econ. Dyn. (2018), https://doi.org/10.1016/j.strueco.2018.08.008
G Model
ARTICLE IN PRESS
STRECO-743; No. of Pages 9
L. Zhang et al. / Structural Change and Economic Dynamics xxx (2018) xxx–xxx
4 Table 2 Variable explanations. Variable
Full Name
Unit
Mean
Standard Deviation
cp xm exp imp ind tec peo energy
carbon productivity export-import dependence export dependence import dependence industrial structure technology progress population size energy consumption structure
ten thousand yuan per ton % % % % % ten thousand people %
0.338 0.324 0.165 0.159 39.808 3.435 4,353.682 63.407
0.203 0.403 0.192 0.241 7.589 4.667 2,628.010 17.460
import value in GDP) and export dependence (exp) (i.e., the proportion of export value in GDP). Additionally, trade value in U.S. dollars is converted into RMB by the average exchange rate of RMB against dollars from 2000 to 2014. 3.4.3. Control variables Moreover, we have selected four control variables to more accurately analyze the effect of foreign trade on carbon productivity, including technology progress, industrial structure, population size and energy consumption structure. The reasons are explained as follows. Firstly, we can decompose the carbon productivity as follows, in order to reflect the related factors: cp =
n y
i
C i=1
=
n y
i
i=1
Ei
Ei c × i = ri × si × mi ci C n
×
(5)
i=1
where yi and Ei denote GDP and energy consumption of the i province, respectively. C represents the regional total carbon emissions. ci refers to carbon emissions of the i province. ri is the output value created by the per unit of energy consumption, reflecting energy efficiency. si stands for the energy consumption of i province’s per unit of carbon emission. ri and si depend on lowcarbon technology level of the i province. While mi shows the space structure of carbon emission, which is expressed by the proportion of the i province’s carbon emissions in the regional total carbon emissions. Thus, the overall carbon productivity of one country is decomposed into the technical effects and structural effects of each province. The two effects can be influenced by technological level, industrial structure, population scale and energy consumption structure (Liang et al., 2017; Long et al., 2016; Lu et al., 2014; Meng and Niu, 2012). Consequently, we eventually select four variables as control variables. Technology progress presents technical effect; Industrial structure and energy consumption structure reflect the structure effect. In addition, population size, as a primary factor, may generate positive or negative effect (Guo et al., 2016; Ohlan, 2015). Technology progress (tec) is represented by the proportion of each province’s R&D internal expenditure in the industrial added value. Technology progress is good to transfer the economy into low-carbon development mode, so as to improve the carbon productivity. Hence, we anticipate that technology progress may have positive effect on China’s carbon productivity, and variable coefficient would be positive. Industrial structure (ind) is expressed as the proportion of total output value of the tertiary industry in the gross output value of that year. Among all the production activities, the tertiary industry has the lowest carbon emissions. Therefore, the higher the proportion of tertiary industry is, the smaller the carbon emissions in this area is, and the higher the carbon productivity is. Thus, the anticipated coefficient of industrial structure is positive. Population size (peo) is measured by the number of permanent resident population of each province at the end of the year. Population growth may aggravate the consumption of environment and
resources, such as promoting the growth of carbon emissions. Due to the limited resources and environment carrying capacity, population growth may prevent economic sustainable development (Casey and Galor, 2017; Guo et al., 2016). While other scholars have different view about population growth. They believe that the demand rise resulted from population growth can stimulate technical progress and economic growth (Bucci, 2015; Sasaki, 2017). Meanwhile, the population growth may bring about a continuous accumulation of human capital. Hence, the anticipated symbol for this variable is uncertain. Energy consumption structure (energy) is represented by the proportion of coal consumption (converted into standard coal) of each province in yearly energy consumption. China is a big coal consumer. In 2015, the coal consumption accounted for 63% in the total energy consumption. CO2 (organic carbon and elemental carbon) is mainly generated by fossil energy (e.g., coal) (Qiao et al., 2014). The bigger the proportion of coal consumption in energy consumption is, the more the carbon emissions are, and the lower the carbon productivity is. Therefore, the anticipated coefficient for this variable is negative. In addition, in order to reduce the dimensional effect and increase the data stability, each variable has been represented by logarithm. Before the regression analysis, spearman correlation coefficient analysis has been performed on the variables. The results show that the correlation coefficients among lnxm, lnimp and lnexp are high, while the correlation coefficients between these three variables and any other variables are lower than 0.7, indicating that these three variables should not be simultaneously included into one model. Furthermore, multicollinearity test has been carried out by variance inflation factor (VIF). When putting the three variables into one model, the average VIF is 62.86, which means the explanatory variables have severe multicollinearity. When the three variables are added into model separately, VIF values of all variables are less than 10. Thus, these three variables are supposed to be included into the model separately. 4. Empirical analysis 4.1. China’s carbon productivity and the characteristics of spatial distribution Based on the above calculation method of carbon productivity, we have measured the carbon productivity of 30 provinces in China. Fig. 1 shows the carbon productivity at China’s provincial level in 2000 and 2014. In 2014, most of the provincial carbon productivity revealed the medium and high value, with more than 0.37 ten thousand yuan per ton. Specifically, the high-value regions were dominated by the eastern coastal provinces (except for Liaoning, Hebei and Shandong) whose carbon productivity were above 0.59 ten thousand yuan per ton. The medium-value region included most central provinces (except for Shanxi, Anhui), some eastern provinces (Liaoning, Shandong) and western provinces (Guangxi, Sichuan), where the carbon productivity were between 0.37 and 0.59 ten thousand yuan per ton. The low-value regions were con-
Please cite this article in press as: Zhang, L., et al., How does foreign trade influence China’s carbon productivity? Based on panel spatial lag model analysis. Struct. Change Econ. Dyn. (2018), https://doi.org/10.1016/j.strueco.2018.08.008
G Model STRECO-743; No. of Pages 9
ARTICLE IN PRESS L. Zhang et al. / Structural Change and Economic Dynamics xxx (2018) xxx–xxx
5
Fig. 1. Distribution of China’s carbon productivity in (a) 2000 and (b) 2014.
centrated in the vast western provinces and its carbon productivity were below 0.37 ten thousand yuan per ton. Compared with the year of 2000, carbon productivity of the most eastern and central provinces improved significantly. For instance, in 2000, the carbon productivity of Shanghai was distributed in the low-value region (carbon productivity was between 0.2 and 0.37); Beijing and Guangdong were in the group of medium-value region. While, in 2014, the carbon productivity of the three areas were all in the higher-value region (carbon productivity was above 0.81), whose carbon productivity climbed two and three levels, respectively. Notably, in 2014, the number of the higher-value region broke through zero. As to western region, the change of carbon productivity was little with the exception of Sichuan and Chongqing. In general, there was no significant variation in terms of regional distribution of the provincial carbon productivity from 2000 to 2014. Although all the values had increased, the pattern was relatively stable. The carbon productivity showed a stair-step distribution from the east to the west, revealing that the east had the highest value and the west had the lowest value. It was roughly in line with the gradient development pattern of China. Moreover, Fig. 1 also illustrates another feature that the carbon productivity of neighboring provinces were similar. Therefore, we assume that China’s carbon productivity may have spatial dependence and need for the next-step test. 4.2. Spatial correlation test The spatial autocorrelation test on carbon productivity should be tested before regression analysis. The test results are shown in Table 3. The global Moran’s I of provincial carbon productivity in China from 2000 to 2014 strongly reject the primary hypothesis ¨ values are less than 0.05), and the ¨ spatial autocorrelation(p of no statistics are positive all the time. It indicates that there is obviously positive spatial autocorrelation in carbon productivity among China’s provinces. Therefore, spatial econometric models need to be considered for empirical analysis.
Table 3 Global Moran’s I of provincial carbon productivity in China (2000–2014). Year
Global Moran’s I
P-value
2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014
0.429 0.428 0.448 0.424 0.446 0.430 0.426 0.397 0.395 0.380 0.375 0.308 0.321 0.296 0.305
0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.001 0.001 0.002 0.001
4.3. Empirical results This research uses Stata13 to carry out the maximum likelihood estimation and check whether SDM model can be simplified into SLM model or SEM model through Wald test and Lratio test. The test results show that all models reject SDM model and cannot be simplified as SEM model, while accepting to be simplified as SLM model. Furthermore, we use Hausman test to decide whether to use random effect models or fixed effect models. Table 4 shows that Model 1–6 are suitable for random effect models, while Model 7–12 for fixed effect models. From the regression results of Table 4, it can be seen that: (1) In Model 1–3, the spatial autoregression coefficient (rho) is significantly positive at the level of 1%, which indicates that there are strong spatial spillover effects on China’s carbon productivity. (2) The coefficients of export-import and import are both positive at the significant level of 1%; the coefficient of export is positive but
Please cite this article in press as: Zhang, L., et al., How does foreign trade influence China’s carbon productivity? Based on panel spatial lag model analysis. Struct. Change Econ. Dyn. (2018), https://doi.org/10.1016/j.strueco.2018.08.008
G Model
ARTICLE IN PRESS
STRECO-743; No. of Pages 9
L. Zhang et al. / Structural Change and Economic Dynamics xxx (2018) xxx–xxx
6 Table 4 Results of SLM model. Variable
Model 1
lnxm
0.0836*** (2.59)
Model 2
Model 3
0.104*** (2.87)
lnimp
0.0238 (0.65) 0.368*** (2.77) 0.0350 (0.58) −0.696*** (-5.06) 0.303*** (4.40) 0.503*** (6.98) 0.0142*** (6.30) chi2(4) = 5.72, Prob > chi2 = 0.2208 chi2(5) = 18.53, Prob > chi2 = 0.0024 chi2(5) = 6.52, Prob > chi2 = 0.2587 450
lnexp lnind lntec lnenergy lnpeo rho sigma2 e Wald test Lratio test Hausman test N
0.386*** (2.86) 0.0170 (0.28) −0.700*** (-5.44) 0.311*** (5.12) 0.508*** (7.16) 0.0139*** (6.44) chi2(5) = 6.68, Prob > chi2 = 0.2455 chi2(5) = 19.07, Prob > chi2 = 0.0019 chi2(5) = 6.88, Prob > chi2 = 0.2297 450
0.382*** (2.80) 0.00928 (0.15) −0.720*** (-6.05) 0.306*** (5.68) 0.513*** (7.13) 0.0134*** (6.85) chi2(4) = 5.98, Prob > chi2 = 0.2004 chi2(5) = 20.64, Prob > chi2 = 0.0009 chi2(5) = 6.08, Prob > chi2 = 0.2988 450
Note: ***, ** and * means significant level of 1%, 5% and 10%, respectively.
not significant, which indicates that foreign trade is conducive to raise the carbon productivity and mainly through import. 4.3.1. Effect decomposition of spatial lag model In a spatial econometric model, the regression coefficient of explanatory variable cannot directly reflect its influence on the explained variable, because the parameter effects of space lag model may involve direct effect, indirect effect and inducing effect. Therefore, it is necessary to eliminate the inducing effect and further decompose it into direct effect and indirect effect. 4.3.1.1. Effect decomposition at the national level. Table 5 illustrates the decomposition of the total effect based on the whole samples. It can be found that: the direct effect and indirect effect of export-import and import are significantly positive, while the coefficient of export is positive but not significant. This indicates that foreign trade has obvious positive effects on carbon productivity in the region, especially promoted through import. Besides, carbon productivity of adjacent areas reveals obvious positive impact within the region. The possible explanations are as follows: on the one hand, due to China’s unbalanced trade commodity structure, the export commodities mainly include the low-end manufacturing and low value-added laborintensive products. While the imported goods are dominated by mechanical and electrical products and other high-tech products. The import of capital and technology intensive products (especially pollution-treatment equipment) can not only directly increase carbon productivity, but also facilitate the imitation and innova¨ tion of domestic enterprises through t¨ echnology spillovereffect. The application of new technology and new process leads to the improvement of productivity. The higher the productivity, the lower the resources input and energy consumption, thus reducing the carbon content of the unit product. This may indirectly improve the carbon productivity (Fouquet, 2017; Zhao and Zhang, 2016). Additionally, importers benefit from the embodied carbon trade activities, because the import of consumer goods could replace the domestic production and thereby reduce the domestic carbon emissions. Moreover, China has attached more importance to import the resources-intensive products in order to reduce the
Table 5 Decomposition of the total effect at the national level. Variable
Direct Effect
Indirect Effect
Total Effect
lnxm
0.0897*** (2.58) 0.418*** (2.85) 0.0192 (0.30) −0.762*** (-6.01) 0.339*** (4.64)
0.0820* (1.92) 0.379** (2.15) 0.0112 (0.19) −0.673*** (-4.36) 0.315** (2.34)
0.172** (2.32) 0.797*** (2.60) 0.0304 (0.25) −1.436*** (-6.81) 0.654*** (3.30)
0.112*** (2.84) 0.414*** (2.79) 0.0113 (0.17) −0.785*** (-6.65) 0.335*** (5.15)
0.105** (1.96) 0.385** (2.08) 0.00358 (0.06) −0.710*** (-4.09) 0.315** (2.48)
0.217** (2.44) 0.799** (2.52) 0.0149 (0.12) −1.495*** (-6.71) 0.650*** (3.57)
0.0247 (0.63) 0.398*** (2.75) 0.0384 (0.62) −0.756*** (-5.52) 0.329*** (4.06)
0.0223 (0.56) 0.353** (2.10) 0.0283 (0.51) −0.654*** (-4.18) 0.301** (2.22)
0.0470 (0.60) 0.751** (2.54) 0.0667 (0.57) −1.410*** (-6.23) 0.630*** (3.05)
lnind Model 1
lntec lnenergy lnpeo lnimp lnind
Model 2
lntec lnenergy lnpeo lnexp lnind
Model 3
lntec lnenergy lnpeo
Note: ***, ** and * means significant level of 1%, 5% and 10%, respectively.
domestic resource exploitation. For example, the import share of minerals (including metal and non-metallic minerals, oil, natural gas and other combustible organic minerals) rose from 10.9% to 23.4% over the period of 2000 to 2014.4 While the resource-
4
Data source: China Statistical Yearbooks.
Please cite this article in press as: Zhang, L., et al., How does foreign trade influence China’s carbon productivity? Based on panel spatial lag model analysis. Struct. Change Econ. Dyn. (2018), https://doi.org/10.1016/j.strueco.2018.08.008
G Model
ARTICLE IN PRESS
STRECO-743; No. of Pages 9
L. Zhang et al. / Structural Change and Economic Dynamics xxx (2018) xxx–xxx
7
Table 6 Decomposition of the total effect at the regional level. Region
Model
Variable
Direct Effect
Indirect Effect
Total Effect
Eastern region
Model 4 Model 5 Model 6
lnxm lnimp lnexp
−0.151*** −0.118*** −0.113***
−0.0412*** −0.0397*** −0.0286***
−0.192*** −0.157*** −0.141***
Central region
Model 7 Model 8 Model 9
lnxm lnimp lnexp
−0.0291 0.0353 −0.0722**
−0.0203 0.0258 −0.0503**
−0.0494 0.0611 −0.122**
Western region
Model 10 Model 11 Model 12
lnxm lnimp lnexp
0.136*** 0.123*** 0.0830***
0.0850*** 0.0857*** 0.0523**
0.221*** 0.209*** 0.135***
Note: ***, ** and * means significant level of 1%, 5% and 10%, respectively.
intensive products, especially energy-related and metal-related products, is the category with the great proportion of the embodied carbon. Its imports will avoid the carbon emissions generated by the exploitation. Therefore, the import has strong effects on improving carbon productivity. On the other hand, the environmental standards between China and the developed countries still have certain gaps. When carrying out export trade, Chinese enterprises have to pay high costs to reach the strict environmental standards in the developed countries through improving their own cleaner production technology. Thus part of the enterprises may export to the countries or regions with lower environmental standards. This kind of export does not accelerate the environmental technology to improve carbon productivity. Meanwhile, China’s export structure is dominated by the low-end manufacturing and labor-intensive products that aggravate resource and energy consumption, resulting in the massive carbon emissions (Liu et al., 2016). Consequently, the export presents weak influence on the improvement of carbon productivity. Besides, adjacent areas may stimulate the improvement of carbon productivity of the whole region through two ways. First, as China’s foreign trade keeps growing, economic ties and factors flows between regions have been reinforced. Capital, labor and other factors in the developed regions will overflow to the neighboring less-developed regions, so as to drive the economic efficiency and promote carbon productivity in the adjacent areas. Second, low-carbon technology will spill over to the surrounding regions along with the interprovincial flow of talents. Enterprises with advanced low-carbon technology will bring competitive pressure to the local enterprises in the surrounding areas and stimulate them to improve cleaner production technology. Thus, carbon productivity in adjacent regions could be enhanced. From the results of control variables, the direct and indirect effects of industrial structure and population size on carbon productivity are significantly positive. It demonstrates that the higher the proportion of the tertiary industry is, the more carbon productivity tends to raise. Besides, population growth is featured by the improvement of human capital, instead of the simple increase in amount, which is benefit to enhance the productivity. Regarding the variable of energy consumption structure, the direct and indirect effects are significantly negative. This is consistent with China’s current situation that coal consumption occupies a huge proportion in the total energy consumption, which is adverse to the improvement of carbon productivity. As for the variable of technological progress, both direct and indirect effects are positive but not significant, which may be related to China’s long-term extensive economic growth pattern. China’s traditional economy ¨ is characterized by high energy consumption, high input and high pollution¨, resulting in the growth of total economic output along with the rapid increase of carbon emissions. Therefore, the growth rate of carbon emission may exceed the contribution of technologi¨ ¨ cal progress to carbon emission reduction, resulting in the failure or
¨inefficiencyof ¨ technological progress in the promotion of regional carbon productivity.
4.3.1.2. Effect decomposition at the regional level. Since China’s opening up is carried out gradually from the coastal to the inland regions, thus regional differences are obvious in terms of economic level, opening degree and technology level in the eastern, central and western regions. Such difference is also reflected in the impact of foreign trade toward carbon productivity. Consequently, it is necessary to decompose the total effect based on the three regions in order to understand the spatial differences. In this section, 30 provinces in China are divided into the eastern, central and western regions.5 Table 6 shows the effect decomposition of spatial lag models in the eastern, central and western regions. The effects of foreign trade on carbon productivity have obvious differences in the three regions. The direct, indirect and total effect of foreign trade on carbon productivity in the eastern region are negative. In the case of the central region, the three effects of export on carbon productivity are significant negative. Import and export-import have no prominent effects. On the contrary, the three types of effects of foreign trade in the western region are significantly positive on carbon productivity. The reasons of such differences may be as follows: although the eastern and central regions used to have comparative advantages in the rapid economy and high degree of trade openness, due to China’s economic slowdown, the cost of labor and raw materials has been dramatically increased. The mode of economic growth driven by foreign trade volume has confronted challenges. Thus improving carbon productivity through foreign trade has been losing the advantages in the eastern and central regions. Moreover, the middle and low-end processing trade takes a dominating position in the eastern and central regions, making the industrial structure dominated by processing and manufacturing. Such extensive development mode leads to huge energy consumption and CO2 emissions. For instance, the eastern provinces of Shandong, Jiangsu, Guangdong and Zhejiang are the largest net exporters of carbon emissions in China, with large share of heavy, energy-consuming products (metal and non-metal products and equipment, etc.) (Liu et al., 2016; Feng et al., 2012). Over relying on processing trade restricts the transformation to tertiary industry, sinking into the
5 According to the No. 33 (2000) document of China, we divide 30 provinces in China into the eastern, central and western regions based on the territorial division involved in The Great Western Development Strategy. The eleven provinces in eastern region are Beijing, Tianjin, Hebei, Shandong, Liaoning, Jiangsu, Shanghai, Zhejiang, Guangdong, Fujian, and Hainan. The central provinces include eight provinces, namely Heilongjiang, Jilin, Shanxi, Henan, Anhui, Hubei, Hunan and Jiangxi. As for western region, there are total eleven provinces, including Inner Mongolia, Ningxia, Gansu, Shaanxi, Qinghai, Sichuan, Chongqing, Guizhou, Guangxi, Yunnan and Xinjiang.
Please cite this article in press as: Zhang, L., et al., How does foreign trade influence China’s carbon productivity? Based on panel spatial lag model analysis. Struct. Change Econ. Dyn. (2018), https://doi.org/10.1016/j.strueco.2018.08.008
G Model STRECO-743; No. of Pages 9
ARTICLE IN PRESS L. Zhang et al. / Structural Change and Economic Dynamics xxx (2018) xxx–xxx
8
carbon-intensive economy. Hence, foreign trade in the eastern and central regions inhibits the improvement of carbon productivity. In the western region, foreign trade has been always weak in the history due to the landlocked location and slow economy. In recent years, with the advancement of The Great Western Development Strategy, the regional investment and development conditions have been improved, which contributes to the accumulation of forthcoming strength for economic development in the western region. Furthermore, “The Belt and Road Initiative” put forward in 2013 has invigorated the development of western region such as the construction of regional international port area, bonded zone and railway, so as to greatly increase the trade activities in the western region. In the past 15 years, the value of foreign trade in western region has raised nearly 17 times, higher than the national average. Foreign trade can stimulate regional comparative advantages and promote resource allocation effectively as well as industrial layout. The rapid growth of foreign trade in the western region has promoted the migration of low-end processing and labor-intensive industries from the eastern and central regions to the western region where the cheap labor and other resources remain abundant. Although such transfer of carbon-intensive industries has led to the substantial increase of carbon emissions in the western region, its growth rate is less than the contribution of foreign trade to economic growth and technology spillover embodied in foreign trade to carbon emissions reduction. Compared with the eastern and central regions, the impact of foreign trade on carbon productivity in the western region performs a late-developing advantage and gradually emerges. Thus, foreign trade exerts a positive effect upon carbon productivity in the western region. The large international events can occasionally influence the regional environmental performance, such as the “APEC blue”, 6 due to the joint prevention and control of neighbor provinces (Lin et al.,2017). However, the essential way to achieve low-carbon development is still industrial upgrading in the long term. Therefore, the opportunity of “The Belt and Road Initiative” should be seized to adjust industrial structure and generate mutual benefits among the countries along the “Belt and Road” in terms of economic development and carbon productivity. 5. Conclusions and policy recommendations 5.1. Conclusions This research measured China’s carbon productivity and analyzed the effect of foreign trade on carbon productivity through applying spatial lag model with the data of 30 provinces in China from 2000 to 2014. The main findings can be summarized as follows: (1) China’s regional carbon productivity had a strong spatial spillover effect and it was not only affected by the factors within the region but also affected by the carbon productivity in the adjacent areas. (2) At the national level, foreign trade helped to improve China’s carbon productivity and import showed more effect than export on the enhancement of carbon productivity. (3) At the regional level, foreign trade in the eastern region manifested prominent negative effects (either direct, indirect effect or total effect) on carbon productivity, while export-import and import in the central region had no significant effects (the afore-
6 During the Asia-Pacific Economic Cooperation (APEC) Summit in 2014, the environment quality of Beijing was improved significantly owing to the pollution control measures jointly conducted by Beijing City, Hebei province and Tianjin City, which was regard as “APEC Blue”.
said three) on carbon productivity. The aforesaid three effects of export were prominently negative. Generally, foreign trade had restrained the increase of carbon productivity in the eastern and central regions. In the western region, the three effects of foreign trade on carbon productivity were significantly positive and the total effect was bigger than that of eastern and middle region. It has been in the stage of enhancing carbon productivity.
5.2. Policy recommendations First, attention should be given to the role of foreign trade, especially import trade, in order to further enhance carbon productivity. China has been over-relying on export trade for years, however, the overseas market demand has been shrinking. This greatly challenges the “export oriented” trade mode. Thus, the driving force of economic growth should emphasize import trade simultaneously in order to realize its positive effects on carbon productivity. Second, the quality and efficiency of foreign trade should be paid much attention as well as the optimization of trade structure. The trade structure need to be transformed to reduce the export of carbon-intensive products with low value-added and to increase the import of capital goods and high-tech products. This can also promote the technology spillover effect within the import trade to drive the domestic technology innovation, especially the cleaner production technology to reduce carbon emissions. Once the trade upgrading and knowledge economy is under way, low energy intensity and low carbon would emerge accordingly (Fouquet, 2017). Third, the customized measures should be taken according to the various conditions of different regions. In the eastern and central regions, the dividends from the increase of trade volume have gradually disappeared. Therefore, the new trade growth engine should be sought. The key is to enhance technology, so as to radically improve the quality of the trade products and furthermore showing positive effect of foreign trade upon carbon productivity. For the western region, the foreign trade starts late and is still at the low level. Thus, we are expecting to make full use of the late-developing advantages of trade and then actively strengthen cooperation with the eastern and central regions as well as introducing advanced management and production technology. Meanwhile, the old path of “treatment after pollution” happened in the eastern and central regions should be avoided. Thereby the western region can realize the sustainability of foreign trade upon enhancement of carbon productivity. Fourth, since China’s carbon productivity has revealed positive spatial dependence, provinces should coordinate with each other and eliminate inter-regional barriers to promote the marketoriented distribution of technology, human capital and other production factors. In this way, the spillover effect of the areas with high-carbon productivity can radiate and stimulate the neighboring areas. Furthermore, the empirical results show that technological progress did not significantly increase carbon productivity. Therefore, technological absorptive capacity should be strengthened while increasing enterprises’ R&D investment, so that the positive effect of technology on carbon productivity can be released. Finally, the limitation still exists in this research. Although we have discussed the influence of the adjustment of industrial structure on carbon productivity, the regression models did not calculate the factor of industrial structure. In fact, trade can promote the changes of industrial structure and subsequently act on carbon productivity. Therefore, the indirect effects of industrial transformation and upgrading on carbon productivity deserve more attention in the future research.
Please cite this article in press as: Zhang, L., et al., How does foreign trade influence China’s carbon productivity? Based on panel spatial lag model analysis. Struct. Change Econ. Dyn. (2018), https://doi.org/10.1016/j.strueco.2018.08.008
G Model STRECO-743; No. of Pages 9
ARTICLE IN PRESS L. Zhang et al. / Structural Change and Economic Dynamics xxx (2018) xxx–xxx
Acknowledgments This research was supported by National Natural Science Foundation of China (No.71873016 and No.71804012) and the Fundamental Research Funds for the Central Universities (No.2015ZCQ-JG-02 and No.2017JC04). We are indebted to the anonymous reviewers and editor. References Almodóvar, P., Saiz-Briones, J., Silverman, B.S., 2014. Learning through foreign market participation: the relative benefits of exporting, importing, and foreign direct investment. J. Technol. Transf. 39 (6), 915–944. Anselin, L., Bera, A., 1998. Spatial dependence in linear regression models with an introduction to spatial econometrics. In: Ullah, A., Giles, D.E.A. (Eds.), Handbook of Applied Economic Statistics. Marcel Dekker, New York, pp. 237–289. Anselin, L., Griffith, D.A., 2010. Do spatial effects really matter in regression analysis? Pap. Reg. Sci. 65 (1), 11–34. Antweiler, W., Copeland, B.R., Taylor, M.S., 2001. Is free trade good for the environment? Am. Econ. Rev. 91 (4), 877–908. Beinhocker, E., Oppenheim, J., Irons, B., et al., 2008. The Carbon Productivity Challenge: Curbing Climate Change and Sustaining Economic Growth. Mckinsey Global Institute. Bucci, A., 2015. Product proliferation, population, and economic growth. J. Hum. Cap. 9 (2), 170–197. Casey, G., Galor, O., 2017. Is faster economic growth compatible with reductions in carbon emissions? The role of diminished population growth. Environ. Res. Lett. 12 (1), 1–30. Chen, G., Hou, F., Chang, K., 2018. Regional decomposition analysis of electric carbon productivity from the perspective of production and consumption in China. Environ. Sci. Pollut. Res. 25 (2), 1508–1518. Coe, D.T., Helpman, E., 1993. International R&D Spillovers. National Bureau of Economic Research (NBER), Working Papers Series, No: 4444. Feng, K., Siu, Y.L., Guan, D., et al., 2012. Analyzing drivers of regional carbon dioxide emissions for China. J. Indust. Ecol. 16 (4), 600–611. Fouquet, R., 2017. Make low-carbon energy an integral part of the knowledge economy. Nature 551 (7682). Gillard, R., Lock, K., 2017. Blowing policy bubbles: rethinking emissions targets and low-carbon energy policies in the U.K. J. Environ. Policy Plan. 19 (6), 638–653. Grossman, G.M., Helpman, E., 1990. Trade, Knowledge Spillovers, and Growth. National Bureau of Economic Research (NBER), Working Papers Series, No: 3485. Grossman, G.M., Krueger, A.B., 1994. Environmental Impacts of a North American Free Trade Agreement. MIT Press, Massachusetts, USA. Guo, J.Y., Liu, J.Q., 2011. Empirical research on the relationship of export trade and carbon emissions. International Conference on Management Science and Engineering (18th), September 13–15. Guo, W., Sun, T., Dai, H., 2016. Effect of population structure change on carbon emission in China. Sustainability 8 (3), 225. He, J.K., Su, M.S., 2011. Carbon productivity analysis to address global climate change. Chin. J. Popul. Resour. Environ. 9 (1), 9–15. Hu, X., Liu, C., 2016. Carbon productivity: a case study in the Australian construction industry. J. Clean. Prod. 112, 2354–2362. IPCC, 2007. Intergovernmental Panel on Climate Change (IPCC) Climate Change 2007: The Fourth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge. Kaya, Y., Yokobori, K., 1997. Environment, Energy, and Economy: Strategies for Sustainability. United Nations University Press, Tokyo, Japan.
9
Knight, K.W., Schor, J.B., 2014. Economic growth and climate change: a cross-national analysis of territorial and consumption-based carbon emissions in high-income countries. Sustainability 6 (6), 3722–3731. Korves, N., Martinez-Zarzoso, I., Voicu, A.M., 2011. Is free trade good or bad for the environment? New empirical evidence. In: Blanco, J.A., Kheradmand, H. (Eds.), Climate Change-Socioeconomic Effects. InTech, Rijeka, Croatia, pp. 1–30. Li, W., Wang, W., Wang, Y., et al., 2018. Historical growth in total factor carbon productivity of the Chinese industry - A comprehensive analysis. J. Clean. Prod. 170, 471–485. Liang, L., Hu, X., Tivendale, L., et al., 2017. The log mean divisia index based carbon productivity in the Australian construction industry. Constr. Econ. Build. 17 (3), 68–84. Lin, H., Tao, L., Fang, F., et al., 2017. Mortality benefits of vigorous air quality improvement interventions during the periods of APEC Blue and Parade Blue in Beijing, China. Environ. Pollut. 220, 222–227. Liu, Z., Davis, S.J., Feng, K., et al., 2016. Targeted opportunities to address the climate-trade dilemma in China. Nat. Clim. Change 6 (2), 201–206. Long, R.Y., Shao, T.X., Chen, H., 2016. Spatial econometric analysis of China’s province-level industrial carbon productivity and its influencing factors. Appl. Energy 166, 210–219. Lu, Z., Yang, Y., Wang, J., 2014. Factor decomposition of carbon productivity chang in China’s main industries: based on the laspeyres decomposition method. Energy Procedia 61, 1893–1896. Meng, M., Niu, D., 2012. Three-dimensional decomposition models for carbon productivity. Energy 46 (1), 179–187. Naranpanawa, A., 2011. Does trade openness promote carbon emissions? Empirical evidence from Sri Lanka. Empir. Econ. Lett. 10, 973–986. Ohlan, R., 2015. The impact of population density, energy consumption, economic growth and trade openness on CO2 emissions in India. Nat. Hazards 79 (2), 1–20. Pan, J.H., Zhang, L.F., 2011. Research on the regional variation of carbon productivity in China. Chin. Ind. Econ. 5, 47–57 (In Chinese). Poon, J.P.H., Casas, I., He, C., 2006. The impact of energy, transport, and trade on air pollution in China. Eurasian Geogr. Econ. 47 (5), 568–584. Qiao, L., Cai, J., Wang, H., et al., 2014. PM2.5 constituents and hospital emergency-room visits in Shanghai, China. Environ. Sci. Technol. 48 (17), 10406–10414. Revesz, R.L., 1992. Rehabilitating interstate competition: rethinking the “race-to-the-bottom”. Rationale for federal environmental regulation. N. Y. Univ. Law Rev. 67 (6), 1210. Rivera-Batiz, L.A., Romer, P.M., 1991. Economic integration and endogenous growth. Q. J. Econ. 106 (2), 531–555. Sandrini, M., Censor, N., 2009. Does trade openness improve environmental quality? J. Environ. Econ. Manage. 58 (3), 346–363. Sasaki, H., 2017. Population growth and trade patterns in semi-endogenous growth economies. Struct. Change Econ. Dyn. 41, 1–12. Seck, A., 2012. International technology diffusion and economic growth: explaining the spillover benefits to developing countries. Struct. Change Econ. Dyn. 23 (4), 437–451. Wang, Y., Chen, W., Kang, Y., et al., 2018. Spatial correlation of factors affecting CO2 emission at provincial level in China: A geographically weighted regression approach. J. Clean. Prod. 184, 929–937. Zhang, C., Wang, J., 2014. Decomposition on the fluctuation of China’s regional carbon productivity growth. Chin. Popul. Resour. Environ. 24 (10), 41–47 (In Chinese). Zhao, Z., Yan, Y., 2014. Consumption-based carbon emissions and international carbon leakage: an analysis based on the WIOD database. In: Xue, J., Zhao, Z. (Eds.), Anaual Report on China’s Low-carbon Economic Development (2014). Social Sciences Academic Press, Beijing, China, pp. 262–272 (In Chinese). Zhao, X., Zhang, J., 2016. The impact of foreign trade on carbon productivity: an empirical study based on 1992–2011 panel data from 88 countries. Int. Bus. 1, 28–39 (In Chinese).
Please cite this article in press as: Zhang, L., et al., How does foreign trade influence China’s carbon productivity? Based on panel spatial lag model analysis. Struct. Change Econ. Dyn. (2018), https://doi.org/10.1016/j.strueco.2018.08.008