Based on panel data from 2000 to 2017 in 29 Chinese provinces, this paper analyzes the impact of industrial structure upgrading on carbon emissions by constructing a spatial panel model and a panel threshold model. The results show that (1) there is a significant spatial correlation between carbon emissions in Chinese provinces, and the carbon emissions of a province are affected by the carbon emissions of surrounding provinces; (2) in China, carbon emissions have a significant time lag feature, and current carbon emissions are largely affected by previous carbon emissions; (3) industrial structure upgrading can effectively promote carbon emission reductions in local areas, and the impact of industrial structure upgrading on carbon emissions has a significant threshold effect. With continued economic development, the promotion effect of industrial structure upgrading on carbon emission reductions will decrease slightly, but this carbon emission reduction effect is still significant. (4) In addition, there is a clear difference between the impact of energy consumption intensity and population size on carbon emissions in short and long terms. In the short term, the increase in energy consumption intensity and the expansion of population size not only increase the carbon emissions of a local area but also increase the carbon emissions of neighboring areas. In the long term, the impact of energy consumption intensity and population size on carbon emissions of neighboring areas will be weakened, but the promotion impact on carbon emissions in local areas will be strengthened.

  • This paper studies the impact of industrial structure upgrading on carbon emissions in detail.

  • This paper uses a static spatial panel model and a dynamic spatial panel model for analysis.

  • This paper uses economic development as a threshold variable to determine whether the impact direction and degree of industrial structure upgrading on carbon emissions will change with economic development.

Global warming has become one of the most serious problems in the international community, and carbon emissions resulting from human activity are known as the main cause of global warming (Mundaca & Markandya 2016). It is an indisputable fact that global warming has affected the survival and development of human society (Yang et al. 2015). As the world's second-largest economy, China has achieved remarkable development; however, China now consumes more energy and emits more carbon on an annual basis. To a certain extent, the rapid development of China's economy has actually come at the cost of the environment (Zhang et al. 2021).

At present, promoting low-carbon development has become an important task in the development of various countries. To reduce carbon emissions, various countries have formulated diverse regulatory and trading measures in their sustainability policies, such as carbon emissions trading systems, carbon allowances and trading, and carbon taxes (Ghosh et al. 2020). In 2020, President Xi Jinping delivered an important speech at the 75th United Nations General Assembly. In his speech, President Xi mentioned that ‘China will increase its national independent contribution, adopt more powerful policies and measures, strive to reach the peak of carbon dioxide emissions by 2030, and strive to achieve carbon neutrality by 2060.’ China is currently in a stage of rapid economic development, and the increase in demand for resources has directly led to an increase in carbon emissions. Therefore, China faces huge challenges in the reduction of carbon emissions.

Related studies have discussed and analyzed the forecasting problems, emissions patterns, and driving factors of carbon emission (Rapauch et al. 2007; Gregg et al. 2008; Du et al. 2015). Carbon emissions resulting from the production process are mainly determined by economic growth, technical efficiency, and industrial structure (Adom et al. 2012). Adjusting the industrial structure, improving energy efficiency and structure, and promoting technological progress are three major ways to improve energy conservation and ensure emissions reductions in China (Foxon 2011), among which industrial structure adjustments are considered to be particularly important (Zhang et al. 2019).

Scholars have conducted research on the relationship between the industrial structure and carbon emissions. Research methods mainly include the index decomposition method (Hoekstra & Jeroen 2003; Li & Wei 2015) and the input–output analysis method and the structural analysis method (Xiang & Sha 2013; Wu et al. 2018). With the development of econometrics, some scholars have begun to use econometric analysis to study the relationship between industrial structure upgrading and carbon emissions. Zhou et al. (2013) analyzed the relationship between industrial structure adjustments and carbon dioxide emissions by using provincial panel data from 1995 to 2009. Zhang & Ren (2011) discussed the relationship between industrial structure adjustment and carbon dioxide emissions in Shandong Province by using the time-series analysis method, and the results showed that adjusting the industrial structure lowered carbon dioxide emissions.

The research methods adopted previously by most scholars rarely took spatial correlation into account. According to the ‘First Law of Geography’, everything is related to everything else, but near things are more related than distant things (Tobler 1970). In China, the economic ties of various regions are becoming gradually strengthened, and economic activities are always carried out in certain temporal and spatial dimensions. These economic phenomena not only show a temporal correlation but also a spatial correlation; therefore, there may be some errors in existing research results. It is necessary to account for temporal and spatial correlations in-depth analysis and to implement relevant policies according to the research results to promote carbon emission reductions. In addition, some studies show that there is U-shaped or inverted U-shaped curve relationship between economic development and environmental pollution (Gill et al. 2018). In view of this, whether the impact of industrial structure upgrading on carbon emissions will change is also one of the areas that needs to be further studied.

To fill the above-mentioned research gaps, this paper takes 29 provinces in China as the research subject, selects carbon emissions as the dependent variable, and uses industrial structure upgrading as the independent variable. To improve the accuracy of estimation, some control variables including economic development, energy consumption intensity, urbanization rate, population size, international trade, foreign direct investment, and technological progress are introduced. Through the establishment of a spatial panel model and a panel threshold model, the impact of industrial structure upgrading on carbon emissions is analyzed. Compared with the existing research, the contributions of this paper mainly include the following three aspects:

  1. This paper focuses on the impact of industrial structure upgrading on carbon emissions, which can enrich the research content of carbon emissions.

  2. In terms of the research methods, in addition to taking the spatial correlation into account and establishing a spatial panel model for analysis, this paper also considers both spatial correlation and time lag, and constructs a dynamic spatial panel model for further analysis, which can greatly improve the accuracy of the research results.

  3. In order to determine whether the impact direction and degree of industrial structure upgrading on carbon emissions change with economic development, this paper uses economic development as a threshold variable and establishes a threshold panel model, which can broaden the relevant research content.

Currently, industrial structure upgrading is understood to be an effective way for China to transform its mode of economic development. China is implementing a supply-side structural reform plan; one of the main goals is to optimize the industrial structure. In this context, carbon emission reduction and the development of low-carbon economy are key tasks of China's sustainable development plans. Therefore, it is necessary to study the impact of industrial structure upgrading on carbon emissions and give full play to the potential for carbon emissions reduction. This research is of great significance to carbon emission reduction.

Data

This paper selects 29 provinces in China as its study areas. Due to the unavailability of data, Tibet, Hainan, and Taiwan are not included. We analyze panel data of those provinces from 2000 to 2017 and the main analytical purpose is to determine the impact of industrial structure upgrading on carbon emissions.

Research shows that economic development, technological progress, international trade, urbanization, population size, energy consumption, and other variables have significant impacts on carbon emissions (Özbuğday & Erbas 2015; Kang et al. 2016; Li et al. 2019). Therefore, to improve the accuracy of the estimation results, in addition to selecting the upgrading of industrial structure as an explanatory variable, this study also introduces economic development, energy consumption intensity, urbanization rate, technological progress, foreign direct investment, international trade, and population size as control variables.

Referring to descriptions of datasets in relevant research (Pekkan et al. 2021), all variables are explained in Table 1. The relevant data comes from the China Statistical Yearbook, China Energy Statistical Yearbook, and the Wind database.

Table 1

List of variables in the regression models

VariablesSymbolExplanationUnit
Carbon emission cp Carbon emission Million tons 
Industrial structure upgrading cg Ratio of the output value of tertiary industry to the output value of secondary industry Percentage 
Energy consumption intensity ec Ratio of total energy consumption to GDP Thousands of tons/100 million yuan 
Economic development ag Per capita GDP 1,000 yuan/person 
Urbanization ur Ratio of urbanization Percentage 
Technological progress te Number of patent grants per year Hundred pieces 
International trade tr Total import and export 100 million yuan 
Population size pe Permanent population Million persons 
Foreign direct investment fd Actual utilization of foreign direct investment 100 million yuan 
VariablesSymbolExplanationUnit
Carbon emission cp Carbon emission Million tons 
Industrial structure upgrading cg Ratio of the output value of tertiary industry to the output value of secondary industry Percentage 
Energy consumption intensity ec Ratio of total energy consumption to GDP Thousands of tons/100 million yuan 
Economic development ag Per capita GDP 1,000 yuan/person 
Urbanization ur Ratio of urbanization Percentage 
Technological progress te Number of patent grants per year Hundred pieces 
International trade tr Total import and export 100 million yuan 
Population size pe Permanent population Million persons 
Foreign direct investment fd Actual utilization of foreign direct investment 100 million yuan 
On the calculation of carbon emissions, the calculation formula is as follows:
(1)
cp is the carbon emissions, Ei refers to the consumption of energy i (i = 1,2,…,8), Ai refers to the carbon emission coefficient of the ith type of energy, and i is denoted as a type of energy, and Ci represents the standard conversion factor for the ith energy.

Referring to the related literature (Chen & Hu 2020; Miao et al. 2020; Zhao et al. 2020), the carbon emission coefficients and standard conversion coefficients of each type of energy are shown in Table 2.

Table 2

Carbon emission coefficient and standard conversion coefficient of each energy

EnergyCarbon emission coefficientStandard conversion coefficient
Coal 0.7559 (kg/kgce) 0.7143 (kgce/kg) 
Coke 0.8550 (kg/kgce) 0.9714 (kgce/kg) 
Crude oil 0.5857 (kg/kgce) 1.4286 (kgce/kg) 
Gasoline 0.5538 (kg/kgce) 1.4714 (kgce/kg) 
Kerosene 0.5714 (kg/kgce) 1.4714 (kgce/kg) 
Diesel 0.5921 (kg/kgce) 1.4571 (kgce/kg) 
Fuel oil 0.6185 (kg/kgce) 1.4286 (kgce/kg) 
Natural gas 0.4483 (kg/kgce) 1.3300 (kgce/m3
EnergyCarbon emission coefficientStandard conversion coefficient
Coal 0.7559 (kg/kgce) 0.7143 (kgce/kg) 
Coke 0.8550 (kg/kgce) 0.9714 (kgce/kg) 
Crude oil 0.5857 (kg/kgce) 1.4286 (kgce/kg) 
Gasoline 0.5538 (kg/kgce) 1.4714 (kgce/kg) 
Kerosene 0.5714 (kg/kgce) 1.4714 (kgce/kg) 
Diesel 0.5921 (kg/kgce) 1.4571 (kgce/kg) 
Fuel oil 0.6185 (kg/kgce) 1.4286 (kgce/kg) 
Natural gas 0.4483 (kg/kgce) 1.3300 (kgce/m3

The statistical description results of all variables are shown in Table 3.

Table 3

Statistical description of all variables

VariablesUnitMinMaxMean
cp Million tons 3.8536 396.5577 88.2605 
cg Percentage 49.7053 423.6677 97.4952 
ec Thousands of tons/100 million yuan 2.5461 45.2403 12.7929 
ag 1,000 yuan/person 2.7590 128.9941 31.2950 
ur Percentage 23.2000 91.0400 49.8308 
te Hundred pieces 0.7500 3,326.5200 230.3584 
tr 100 million yuan 13.2231 68,168.8000 5,788.5200 
pe Million persons 5.1700 111.6900 45.2211 
fd 100 million yuan 0.9930 2,257.3220 365.4256 
VariablesUnitMinMaxMean
cp Million tons 3.8536 396.5577 88.2605 
cg Percentage 49.7053 423.6677 97.4952 
ec Thousands of tons/100 million yuan 2.5461 45.2403 12.7929 
ag 1,000 yuan/person 2.7590 128.9941 31.2950 
ur Percentage 23.2000 91.0400 49.8308 
te Hundred pieces 0.7500 3,326.5200 230.3584 
tr 100 million yuan 13.2231 68,168.8000 5,788.5200 
pe Million persons 5.1700 111.6900 45.2211 
fd 100 million yuan 0.9930 2,257.3220 365.4256 

Methodology

The STIRPAT model is often used to study the impact of human economic activities on the environment, and the standard form is as follows:
(2)
where I represents environmental impact, P represents population, A refers to affluence, and T stands for technology.
After taking natural logarithms, a convenient linear specification is gained:
(3)
where I, P, A, T are the same as in Equation (2); a is a constant term. et is a random error term. b, c, and d are the elasticity coefficients, that is, the percentage of change is in the environmental factor (I) when the target factors (P, A, T) change 1% under the condition that other factors are constant.
Referring to the STIRPAT model, the static spatial panel model is firstly established for analysis. Currently, the spatial econometric model mainly includes the spatial lag model (SAR), spatial error model (SEM), and spatial Durbin model (SDM). Among them, the SDM model not only introduces the spatial lag of the explained variable but also the explanatory variables, so we first chose the SDM model as an example. The specific type of the spatial panel model will be determined by subsequent Lagrange multiplier (LM) tests. All variables are processed logarithmically to reduce heteroscedasticity.
(4)
where W is the spatial weight matrix. This paper uses the geographic distance matrix for analysis; Wlncp is the spatial lag term of the dependent variable, and its main function is to estimate the degree of spatial correlation, r1 is the coefficient of the corresponding spatial lag term; b, q1, q2, q3, q4, q5, q6, and q7 are estimated constant regression parameters; Wlncg is the spatial lag term of the independent variable, and m1 is the corresponding influence coefficient; Wlnec, Wlnag, Wlnur, Wlnte, Wlntr, Wlnpe, and Wlnfd are the spatial lag independent variables; h1, h2, h3, h4, h5, h6, and h7 are constant regression parameters; ekt is a random error term.
According to related research (Wang et al. 2021), the process by which carbon emissions take place is dynamic and constantly changing; hence, we establish a dynamic spatial panel model for further research based on the static spatial panel model and take the time dependence of carbon emissions into account. Furthermore, we take the dynamic SDM as an example, and the specific form of the dynamic spatial panel model will be determined by subsequent related tests. The form of the dynamic SDM is shown in Equation (5), where L.lncp represents the first-order lag item of carbon emissions, and the meanings of the other symbols are the same as those in Equation (4).
(5)
In addition, to examine the nonlinear correlation between industrial structure upgrading and carbon emissions, this paper also constructs the threshold panel model that was put forward by Hansen (1999) for further analysis. Using economic development (per capita GDP) as the threshold variable, this paper takes the double-threshold panel model as an example for description, as is shown in Equation (6). The number of threshold values will be determined by the threshold test results in the subsequent empirical process.
(6)
I(.) is an indicator function, f1 and f2 are the estimated threshold values, and lnag is the threshold variable. cp, cg, ec, ag, ur, te, tr, pe, and fd are the same as in Equation (4).

This section mainly demonstrates and analyzes the relevant test and empirical results. To avoid spurious regression, the unit root test was firstly carried out; next, to preliminarily verify the existence of spatial correlation, the Moran's I index was measured, and its significance was also tested; finally, the estimation results of the static spatial panel model, the dynamic spatial panel model, and panel threshold model were analyzed.

Unit root test

In the panel model, spurious regression will lead to errors in the estimation results. To avoid this situation, we refer to relevant studies (Harris & Tzavalis 1999; Choi 2001; Levin et al. 2002; Im & Shin 2003) and adopt three commonly used unit root test methods to test the stationarity of the data.

Table 4 displays the results of the stationarity test. The first-order difference of each variable's logarithmic form has passed the significance test at the 1% level, which implies that all variables are first-order single integer variables. Therefore, we construct related models and continue with the following analysis.

Table 4

The stationarity test of each variable

MethodD.lncpD.lncgD.lnecD.lnagD.lnurD.lnteD.lntrD.lnpeD.lnfd
IPS − 9.1103 (0.000) − 5.8782 (0.000) − 9.0555 (0.000) − 4.2957 (0.000) − 9.7658 (0.000) − 8.4326 (0.000) − 8.6966 (0.000) − 8.2721 (0.000) − 8.7362 (0.000) 
Fisher -PP − 19.2402 (0.000) − 8.2751 (0.000) − 22.338 (0.000) − 6.014 (0.000) − 21.2258 (0.000) − 17.4416 (0.000) − 17.4621 (0.000) − 13.7487 (0.000) − 16.9311 (0.000) 
LLC − 8.8957 (0.000) − 3.3820 (0.0004) − 7.8861 (0.000) − 2.7517 (0.003) − 26.862 (0.000) − 5.8134 (0.000) − 8.2723 (0.000) − 6.4334 (0.000) − 8.0902 (0.000) 
MethodD.lncpD.lncgD.lnecD.lnagD.lnurD.lnteD.lntrD.lnpeD.lnfd
IPS − 9.1103 (0.000) − 5.8782 (0.000) − 9.0555 (0.000) − 4.2957 (0.000) − 9.7658 (0.000) − 8.4326 (0.000) − 8.6966 (0.000) − 8.2721 (0.000) − 8.7362 (0.000) 
Fisher -PP − 19.2402 (0.000) − 8.2751 (0.000) − 22.338 (0.000) − 6.014 (0.000) − 21.2258 (0.000) − 17.4416 (0.000) − 17.4621 (0.000) − 13.7487 (0.000) − 16.9311 (0.000) 
LLC − 8.8957 (0.000) − 3.3820 (0.0004) − 7.8861 (0.000) − 2.7517 (0.003) − 26.862 (0.000) − 5.8134 (0.000) − 8.2723 (0.000) − 6.4334 (0.000) − 8.0902 (0.000) 

Note: P-values are in parentheses.

Moran's I test

The values of Moran's I can be seen in Table 5. Except for several particular years, almost all of the Moran's I values are significantly positive, which indicate a significant positive spatial correlation between the carbon emissions of different provinces in China.

Table 5

The test results of Moran's I

YearMoran's IP-valueYearMoran's IP-value
2000 0.074 0.004 2009 0.052 0.022 
2001 0.081 0.002 2010 0.053 0.020 
2002 0.074 0.003 2011 0.046 0.030 
2003 0.065 0.007 2012 0.035 0.063 
2004 0.062 0.009 2013 0.029 0.089 
2005 0.065 0.007 2014 0.028 0.094 
2006 0.060 0.012 2015 0.029 0.088 
2007 0.056 0.016 2016 0.026 0.111 
2008 0.054 0.019 2017 0.026 0.108 
YearMoran's IP-valueYearMoran's IP-value
2000 0.074 0.004 2009 0.052 0.022 
2001 0.081 0.002 2010 0.053 0.020 
2002 0.074 0.003 2011 0.046 0.030 
2003 0.065 0.007 2012 0.035 0.063 
2004 0.062 0.009 2013 0.029 0.089 
2005 0.065 0.007 2014 0.028 0.094 
2006 0.060 0.012 2015 0.029 0.088 
2007 0.056 0.016 2016 0.026 0.111 
2008 0.054 0.019 2017 0.026 0.108 

Regression results of the static spatial model

With recent developments in econometrics, the spatial econometric models mainly include SAR, SEM, and SDM. In terms of model selection, the LM test can be used to determine whether to select the SAR model or the SEM model. If the two spatial models are appropriate, the SDM should be constructed.

The test results are shown in Table 6. Whether it is the LM or Robust LM tests, the SAR passed the 1% significance level test, while the SEM did not pass the significance test, so the SAR model is adopted, and the model form is shown in Equation (7). γ1 is the spatial autocorrelation coefficient, and bi (i = 1, 2 … 8) is the coefficient value of each variable.
(7)
Table 6

The spatial correlation of the model test

Test methodStatistical valueP-value
LM test (spatial error) 0.168 0.682 
Robust LM test (spatial error) 0.641 0.423 
LM test (spatial lag) 27.198 0.000 
Robust LM test(spatial lag) 27.670 0.000 
Test methodStatistical valueP-value
LM test (spatial error) 0.168 0.682 
Robust LM test (spatial error) 0.641 0.423 
LM test (spatial lag) 27.198 0.000 
Robust LM test(spatial lag) 27.670 0.000 

Table 7 demonstrates the estimation results of the SAR model. According to the regression coefficients, the coefficient of industrial structure upgrading (lncg) is significantly negative, but the coefficient value is relatively small. As for other control variables, the coefficients of energy consumption intensity (lnec), economic development (lnag), urbanization (lnur), and population size (lnpe) are significantly positive, and the coefficients of technological progress (lnte) and foreign direct investment (lnfd) are significantly negative. Among these control variables, the regression coefficients of energy consumption intensity, economic development, and population size are higher than those of other control variables.

Table 7

The regression results of spatial panel model

VariablesCoefficientsEffect decomposition
Direct effectsIndirect effectsTotal effects
lncg −0.0947*** (0.001) −0.0937*** (0.001) 0.0076 (0.117) −0.0861*** (0.001) 
lnec 1.1117*** (0.000) 1.1111*** (0.000) −0.0896* (0.063) 1.0215*** (0.000) 
lnag 1.1562*** (0.000) 1.1596*** (0.000) −0.0942* (0.067) 1.0655*** (0.000) 
lnur 0.2307*** (0.000) 0.2289*** (0.000) −0.0182* (0.095) 0.2107*** (0.000) 
lnte −0.0473*** (0.001) −0.0484*** (0.001) 0.0040 (0.125) −0.0444*** (0.001) 
lntr −0.0216 (0.175) −0.0210 (0.185) 0.0014 (0.338) −0.0195 (0.189) 
lnpe 0.8473*** (0.000) 0.8470*** (0.000) −0.0698* (0.083) 0.7772*** (0.000) 
lnfd −0.0215** (0.012) −0.0219** (0.009) 0.0018 (0.138) −0.0202** (0.009) 
VariablesCoefficientsEffect decomposition
Direct effectsIndirect effectsTotal effects
lncg −0.0947*** (0.001) −0.0937*** (0.001) 0.0076 (0.117) −0.0861*** (0.001) 
lnec 1.1117*** (0.000) 1.1111*** (0.000) −0.0896* (0.063) 1.0215*** (0.000) 
lnag 1.1562*** (0.000) 1.1596*** (0.000) −0.0942* (0.067) 1.0655*** (0.000) 
lnur 0.2307*** (0.000) 0.2289*** (0.000) −0.0182* (0.095) 0.2107*** (0.000) 
lnte −0.0473*** (0.001) −0.0484*** (0.001) 0.0040 (0.125) −0.0444*** (0.001) 
lntr −0.0216 (0.175) −0.0210 (0.185) 0.0014 (0.338) −0.0195 (0.189) 
lnpe 0.8473*** (0.000) 0.8470*** (0.000) −0.0698* (0.083) 0.7772*** (0.000) 
lnfd −0.0215** (0.012) −0.0219** (0.009) 0.0018 (0.138) −0.0202** (0.009) 

Note: ***, **, and * show the significance levels of 1, 5, and 10%, respectively, and P-values are in parentheses.

When conducting analysis using the spatial econometric model, regression coefficients can partially explain the influence of each variable, but they cannot explain the marginal effects of each variable well. Therefore, it is necessary to decompose the effect to accurately understand the spatial variation mechanism of each variable. In the spatial panel model, the total effect can be decomposed into direct and indirect effects. The direct effect refers to the average impact of the independent variable on the dependent variable in the local areas, and the indirect effect refers to the impact of the independent variable on the dependent variable in the neighboring area. The partial differential method can be used to make the correct measurement.

According to the effect decomposition in Table 7, the total effect coefficient of industrial structure upgrading is significantly negative, the upgrading of industrial structure can effectively reduce carbon emissions. Specifically, the direct effect coefficient of industrial structure upgrading is significantly negative, but its indirect effect coefficient did not pass the 10% significance level test. This implies that industrial structure upgrading can directly reduce local carbon emissions but has no significant effect on the carbon emissions of neighboring areas. Energy consumption is one of the main sources of carbon emissions. According to the results of the effect decomposition, the direct effect of energy consumption intensity is particularly high, indicating that an increase in energy consumption will directly increase local carbon emissions. Furthermore, economic development and population expansion will also greatly increase the local carbon emissions.

Similar to the effect decomposition of industrial structure upgrading, the direct effect coefficients of technological progress and foreign direct investment are significantly negative, while the indirect coefficients did not pass the significance test, indicating that technological progress and an increase in foreign direct investment will promote local carbon emissions reduction. However, in terms of specific values, compared with the energy consumption intensity, economic development level, and population size, the promotion effect of technological progress and foreign direct investment on carbon emission reduction is very limited.

Regression results of the dynamic spatial panel model

The static spatial panel model only includes the spatial lag effect of carbon emissions. However, the change in carbon emissions is actually a dynamic process of continuous accumulation, and carbon emissions are affected by some unmeasurable factors. Therefore, considering the possible time lag of carbon emissions, we use the first-order time lag of carbon emissions as an independent variable to express the comprehensive influence of other unmeasurable variables on carbon emissions. We further construct a dynamic spatial panel model for analysis. According to the above LM test results, here, the dynamic SAR rather than the dynamic SDM will be established; the specific model form is shown in Equation (8).
(8)

The main difference between the dynamic and the static spatial panel models is that the dynamic spatial panel model can effectively distinguish the short- and long-term effects simultaneously.

From the estimated results of the regression coefficients in Table 8, the first-order time lag coefficient of carbon emissions (L.lncp) is 0.7077, which passed the significance test at the 1% level, implying that carbon emissions in the early stage will have a significant impact on later carbon emissions. Furthermore, higher carbon emissions in the previous period will directly lead to an increase in carbon emissions in the current period. This also proves the effectiveness of the dynamic spatial panel model.

Table 8

Regression results and effect decomposition of the dynamic spatial panel model

VariablesCoefficientsShort-term effects
Long-term effects
Direct effectsIndirect effectsTotal effectsDirect effectsIndirect effectsTotal effects
L1.lncp 0.7077*** (0.000) – – – – – − 
lncg −0.0157 (0.371) −0.0140 (0.412) −0.0080 (0.448) −0.0220 (0.417) −0.0474 (0.412) −0.0039 (0.891) −0.0513 (0.455) 
lnec 0.3933*** (0.000) 0.3975*** (0.000) 0.2390*** (0.000) 0.6365*** (0.000) 1.3499*** (0.000) 0.1684 (0.722) 1.5183*** (0.001) 
lnag 0.3616*** (0.000) 0.3659*** (0.000) 0.2204*** (0.001) 0.5864*** (0.000) 1.2426*** (0.000) 0.1578 (0.720) 1.4004*** (0.002) 
lnur −0.0592 (0.207) −0.0585 (0.213) −0.0355 (0.259) −0.0940 (0.220) −0.1987 (0.213) −0.0276 (0.783) −0.2263 (0.272) 
lnte −0.0110 (0.252) −0.0116 (0.227) −0.0068 (0.265) −0.0185 (0.231) −0.0394 (0.227) −0.0041 (0.804) −0.0435 (0.262) 
lntr −0.0052 (0.612) −0.0056 (0.557) −0.0035 (0.573) −0.0092 (0.558) −0.0191 (0.557) −0.0032 (0.830) −0.0223 (0.581) 
lnpe 0.1098* (0.054) 0.1139** (0.040) 0.0684* (0.082) 0.1823** (0.045) 0.3869** (0.040) 0.0470 (0.754) 0.4339* (0.087) 
lnfd −0.0032 (0.554) −0.0028 (0.589) −0.0017 (0.613) −0.0046** (0.593) −0.0097 (0.588) −0.0015 (0.864) −0.0112 (0.615) 
VariablesCoefficientsShort-term effects
Long-term effects
Direct effectsIndirect effectsTotal effectsDirect effectsIndirect effectsTotal effects
L1.lncp 0.7077*** (0.000) – – – – – − 
lncg −0.0157 (0.371) −0.0140 (0.412) −0.0080 (0.448) −0.0220 (0.417) −0.0474 (0.412) −0.0039 (0.891) −0.0513 (0.455) 
lnec 0.3933*** (0.000) 0.3975*** (0.000) 0.2390*** (0.000) 0.6365*** (0.000) 1.3499*** (0.000) 0.1684 (0.722) 1.5183*** (0.001) 
lnag 0.3616*** (0.000) 0.3659*** (0.000) 0.2204*** (0.001) 0.5864*** (0.000) 1.2426*** (0.000) 0.1578 (0.720) 1.4004*** (0.002) 
lnur −0.0592 (0.207) −0.0585 (0.213) −0.0355 (0.259) −0.0940 (0.220) −0.1987 (0.213) −0.0276 (0.783) −0.2263 (0.272) 
lnte −0.0110 (0.252) −0.0116 (0.227) −0.0068 (0.265) −0.0185 (0.231) −0.0394 (0.227) −0.0041 (0.804) −0.0435 (0.262) 
lntr −0.0052 (0.612) −0.0056 (0.557) −0.0035 (0.573) −0.0092 (0.558) −0.0191 (0.557) −0.0032 (0.830) −0.0223 (0.581) 
lnpe 0.1098* (0.054) 0.1139** (0.040) 0.0684* (0.082) 0.1823** (0.045) 0.3869** (0.040) 0.0470 (0.754) 0.4339* (0.087) 
lnfd −0.0032 (0.554) −0.0028 (0.589) −0.0017 (0.613) −0.0046** (0.593) −0.0097 (0.588) −0.0015 (0.864) −0.0112 (0.615) 

Note: ***, **, and * show the significance levels of 1, 5, and 10%, respectively, and P-values are in parentheses.

Regarding other control variables, in the short term, direct and indirect effect coefficients of energy consumption intensity, economic development level, and population size are all significantly positive, indicating that an increase in energy consumption intensity, economic development, and population expansion will not only increase local carbon emissions but also promote carbon emissions in neighboring areas. In the long term, the direct effect coefficients of the three variables are significantly positive, and the values are much larger than the short-term direct effect coefficients. However, their indirect effect coefficients did not pass the significance test, which shows that in the long run, the impact of energy consumption intensity, economic development level, and population size on carbon emissions of neighboring areas will be weakened, while the promotion impact on the local carbon emissions will be enhanced.

Regression results of the threshold model

Before performing the threshold model regression, the threshold effect was tested. Table 9 displays the threshold effect test results, which shows that the single- and double-threshold models passed the significant test at the 10% level. Therefore, the threshold effect indeed exists, meaning that it is appropriate to establish the threshold model for further analysis.

Table 9

Threshold effect test results

ModelF-valueP-valueCritical value
1%5%10%
Single-threshold model 41.12 0.0500 57.1590 41.1126 35.1635 
Double-threshold model 47.59 0.0067 43.6325 34.0890 29.5894 
Triple-threshold model 30.66 0.4200 91.6662 71.2863 60.6673 
ModelF-valueP-valueCritical value
1%5%10%
Single-threshold model 41.12 0.0500 57.1590 41.1126 35.1635 
Double-threshold model 47.59 0.0067 43.6325 34.0890 29.5894 
Triple-threshold model 30.66 0.4200 91.6662 71.2863 60.6673 

According to the threshold effect test results, strong evidence concludes that industrial structural upgrading has a double-threshold effect on carbon emissions when dealing with economic development as a threshold variable, so the double-threshold model should be established for analysis. The specific threshold values are presented in Table 10, and the two threshold values are 2.5854 and 3.5834.

Table 10

The values of thresholds and the confidence intervals

ModelThreshold estimators95% Confidence intervals
Single-threshold model 2.5854 [2.5592, 2.5854] 
Double-threshold model 3.5834 [3.5550, 3.5944] 
ModelThreshold estimators95% Confidence intervals
Single-threshold model 2.5854 [2.5592, 2.5854] 
Double-threshold model 3.5834 [3.5550, 3.5944] 

According to relevant mathematical knowledge, when lnag is 2.5854, ag = elnag = e2.5854 = 13.269 (1,000 yuan/person), that is, the per capita GDP is 13,269 RMB; similarly, when lnag is 3.5834, the corresponding ag is 35.996 (1,000 yuan/person), that is, the per capita GDP is 35,996 RMB.

Table 11 displays the threshold regression results. It can be seen that in different stages of economic development, the impact degree of industrial structure upgrading on carbon emissions is different. According to the specific regression coefficients of industrial structural upgrading (lncg), when the per capita GDP is less than 13,269 RMB, that is lnag < 2.5854, the carbon emissions will be reduced by 0.4327% with a 1% increase in industrial structure upgrading; when the per capita GDP is between 13,269 and 35,996 RMB, that is 2.5854 < lnag ≤ 3.5834, the coefficient of industrial structure upgrading is −0.3819; when the per capita GDP is higher than 35,996 RMB, that is lnag > 3.5834, the regression coefficient of industrial structure upgrading is only −0.3433. Therefore, the upgrading of industrial structure is indeed conducive to reducing carbon emissions, which is consistent with the estimation results of the spatial panel model. However, the carbon emission reduction effect of industrial structure upgrading will obviously be weakened by economic development.

Table 11

The regression results of threshold model

VariablesCoef.Std. Err.tP > |t|
lncg (lnag < 2.5854) −0.4327 0.0449 −9.63 0.000 
lncg (2.5854 <lnag ≤ 3.5834) −0.3819 0.0466 −8.20 0.000 
lncg (lnag > 3.5834) −0.3433 0.0462 −7.43 0.000 
lnec 0.3457 0.0566 6.11 0.000 
lnur 0.9008 0.0988 9.12 0.000 
lnte 0.0985 0.0236 4.18 0.000 
lntr 0.1444 0.0233 6.19 0.000 
lnpe 0.7882 0.1376 5.73 0.000 
lnfd − 0.0262 0.0146 −1.79 0.074 
VariablesCoef.Std. Err.tP > |t|
lncg (lnag < 2.5854) −0.4327 0.0449 −9.63 0.000 
lncg (2.5854 <lnag ≤ 3.5834) −0.3819 0.0466 −8.20 0.000 
lncg (lnag > 3.5834) −0.3433 0.0462 −7.43 0.000 
lnec 0.3457 0.0566 6.11 0.000 
lnur 0.9008 0.0988 9.12 0.000 
lnte 0.0985 0.0236 4.18 0.000 
lntr 0.1444 0.0233 6.19 0.000 
lnpe 0.7882 0.1376 5.73 0.000 
lnfd − 0.0262 0.0146 −1.79 0.074 

Everything is related to everything else (Tobler 1970), especially in China, economic ties of various regions are becoming gradually strengthened, so we choose the spatial measurement method for analysis. This paper firstly established a static spatial model to determine the existence of spatial effects. Based on the static spatial panel model, we also establish the dynamic spatial model to test the time lag of carbon emissions. In addition, most scholars who examined the relationship between economic development and environmental pollution found that there was a U-shaped or inverted U-shaped nonlinear relationship between the two variables (Du et al. 2018). Thus, to test whether the impact direction and degree of industrial structure upgrading on carbon emissions will change with economic development, this paper has constructed the threshold panel model for further analysis.

Consistent with the conclusions of most studies, the estimation results of the spatial panel model show that industrial structure upgrading can indeed promote carbon emissions reductions (Chen et al. 2019; Zhou & Li, 2020). However, the empirical results also show that the promotion effect is very limited in this paper. Furthermore, there are several other important findings. Firstly, this study discovers that the carbon emissions of the current period is greatly affected by those of the previous period; secondly, the impact of industrial structure upgrading on carbon emissions has a threshold effect – as with the development of the economy, the carbon emission reduction effect from industrial structure upgrading will be weakened; finally, regarding the impact of other variables on carbon emissions, this paper also has some new discoveries. In the short term, the increase in energy consumption intensity and population expansion not only affects local carbon emissions but also affects the carbon emissions of neighboring areas. However, in the long run, the impact of energy consumption intensity and population size on local carbon emissions will deepen, and the impact on carbon emissions in neighboring areas will gradually weaken.

The industrial structure can be simply understood as the proportion of primary, secondary, and tertiary industries in the national economic structure. Industrial structure upgrading refers to a process in which the industrial structure gradually evolves from a low- to high-level stage. The main manifestations are that the proportion of the primary and secondary industries will gradually decline, and the tertiary industry will gradually become more dominant. Eventually, the tertiary industry will be larger than the secondary industry and the secondary industry will be larger than the primary industry.

In most cases, the total energy consumption and the energy consumption structure required for the development of the three major industries are significantly different. In the primary industry, agriculture – which occupies a dominant position – does not have a high demand for energy, and its total carbon emissions are relatively low. The secondary industry includes most high-energy consumption sectors and is characterized by high pollution; the secondary industry – which is dominated by the heavy industry – has a large demand for energy consumption, thereby increasing the reliance on coal and oil, which greatly increases carbon emissions. The development of the secondary industry comes at the cost of much energy consumption (Zhang et al. 2018; Zhu et al. 2021). The tertiary industry mainly includes two major sectors: circulation and services, which are less dependent on energy and produce less carbon emission.

In this paper, the ratio of the tertiary to secondary industry output values was used to measure industrial structure upgrading. The larger the ratio, the larger the tertiary industry in the national economic structure. Most sub-industries in the tertiary industries have lower demands for energy consumption, but most industries in the secondary industry need to consume more energy. Carbon emissions are mainly attributed to fuel combustion, and the secondary industry is the main source of carbon emissions (Talukdar & Meisner 2001; Karen et al. 2006). With the industrial structure upgrading, the secondary industry will gradually decline, and the tertiary industry will grow larger. Therefore, the total energy consumption in society will decrease, and carbon emissions will decrease accordingly.

Except for industrial structure upgrading, technological progress is also beneficial to promoting carbon emission reductions. In all walks of life, the application of advanced technology can help improve the efficiency of energy utilization, thereby reducing the input of high-energy consumption production factors, such as coal and oil, and reduce carbon emissions. Furthermore, the increase in foreign direct investments is conducive to promoting the development of science and technology, so to a certain extent, it can also help reduce carbon emissions.

Conclusions

This paper studies the influence of industrial structure upgrading on carbon emissions in China. By considering the spatial correlation among carbon emissions, the spatial panel model is selected with reference to the STIRPAT model. Here, we analyze the specific impact of industrial structure upgrading on carbon emissions. Then, the double-threshold panel model is established to investigate the threshold effect. According to the estimation results of the models, we come to the following conclusions.

Upgrading the industrial structure is helpful in reducing carbon emissions, but this carbon emission reduction effect is limited. The effect coefficients of industrial structure upgrading are relatively low. Furthermore, the specific degrees of influence differ in different stages of economic development. With economic development, the carbon emission reduction effect of industrial structure upgrading will be weakened.

Whether in the short or long term, the increase in energy consumption intensity and population expansion will directly lead to a sharp increase in carbon emissions. Moreover, according to the estimation results of the dynamic spatial panel model, in the long term, the impact of this promotion of carbon emissions will be strengthened.

Scientific and technological progress is an important step in reducing carbon emissions. The development of low-carbon technologies is beneficial in improving energy utilization, thus reducing the total amount of carbon emissions. Furthermore, with the progress of science and technology, an increased number of new products can be used in the production of high-energy consumption industries, thereby promoting carbon emission reduction.

Policy recommendations

Based on the above research results, we propose the following suggestions.

China should actively promote the process of industrial structure upgrading. In China, the upgrading of industrial structure can effectively promote carbon emission reduction. Therefore, various regions should actively promote the process of industrial structure upgrading. Preferential policies, such as taxation and subsidies, can be formulated to encourage the development of modern service and high-tech industries and other tertiary industries, including information, electronic equipment manufacturing, and other industries. Additionally, it is necessary to change the industrial structure of industrial zones and to introduce relevant laws and regulations to restrict the excessive development of high-energy consumption secondary industries, especially the excessive growth of energy- and carbon-intensive industries, such as steel, non-ferrous metals, coal, electricity, petroleum, petrochemical, chemical, building materials, and other industries. Furthermore, it is essential to improve the exit mechanisms of overcapacity industries, guide the rational layout and orderly transfer of industries, and promote the rational development of the industrial structure by optimizing the allocation of production factors and the flow of resources between industries.

The key to promote industrial structure upgrading is to improve independent innovation capabilities and strengthen technological innovation. The government should encourage the establishment of research and development centers, improve the research and development capabilities of enterprises, and promote the formation of independent intellectual property rights through the formulation of fiscal and taxation policies, national procurement policies, and the strict enforcement of intellectual property laws.

China should accelerate technological innovation, promote the recycling of industrial solid waste, and adjust energy structures. The regression results have clearly demonstrated that technological progress is an important way to reduce carbon emissions, but the energy consumption intensity will significantly increase carbon emissions. This influence will be strengthened in the long term. Therefore, government departments and enterprises should make full use of contemporary information technology, promote the deep integration of informatization and industrialization, accelerate the upgrading and transformation of traditional industries, develop low-carbon energy, and improve the efficiency of energy utilization.

All relevant data are included in the paper or its Supplementary Information.

Adom
P. K.
,
Bekoe
W.
,
Franklin
A. M.
&
Botchway
E.
2012
Carbon dioxide emissions, economic growth, industrial structure, and technical efficiency: empirical evidence from Ghana, Senegal, and Morocco on the causal dynamics
.
Energy
47
(
1
),
314
325
.
https://doi.org/10.1016/j.energy.2012.09.025
.
Chen
L.
&
Hu
W. T.
2020
Research on the synergy effect of financial development, technological progress and carbon emissions – based on the VAR analysis of China's 30 provincial carbon emissions from 2005 to 2017
.
Study & Exploration
6
,
117
124
(In Chinese)
.
Chen
X.
, , Chen, Y. E. & Chang, C.
2019
The effects of environmental regulation and industrial structure on carbon dioxide emission: a non-linear investigation
.
Environ. Sci. Pollut. Res.
26
,
30252
30267
.
https://doi.org/10.1007/s11356–019–06150–6.
Choi
I.
2001
Unit root tests for panel data
.
Journal of International Money and Finance
20
(
2
),
249
272
.
https://doi.org/10.1016/S0261-5606(00)00048-6
.
Du
Q.
,
Wang
N.
&
Che
L.
2015
Forecasting China's per capita carbon emissions under a new three-step economic development strategy
.
Journal of Resources and Ecology
6
(
5
),
318
323
.
https://doi.org/10.5814/j.issn.1674-764x.2015.05.005
.
Du
G.
,
Liu
S.
,
Lei
N.
&
Huang
Y.
2018
A test of environmental Kuznets curve for haze pollution in China: evidence from the penal data of 27 capital cities
.
Journal of Cleaner Production
205
,
821
827
.
https://doi.org/10.1016/j.jclepro.2018.08.330
.
Foxon
T. J.
2011
A coevolutionary framework for analysing a transition to a sustainable low carbon economy
.
Ecological Economics
70
(
12
),
2258
2267
.
https://doi.org/10.1016/j.ecolecon.2011.07.014
.
Ghosh
P.
,
Jha
A.
&
Sharma
R.
2020
Managing carbon footprint for a sustainable supply chain: a systematic literature review
.
Modern Supply Chain Research and Applications
2
(
3
),
123
141
.
https://doi.org/10.1108/MSCRA-06-2020-0016
.
Gill
A. R.
,
Viswanathan
K. K.
&
Hassan
S.
2018
The Environmental Kuznets Curve (EKC) and the environmental problem of the day
.
Renewable and Sustainable Energy Reviews
81
,
1636
1642
.
https://doi.org/10.1016/j.rser.2017.05.247
.
Gregg
J. S.
,
Andres
R. J.
&
Marland
G.
2008
China: the emissions patterns of the world leader in CO2 emissions from fossil fuel consumption and cement production
.
Geophysical Research Letters
35
(
8
),
L08806-1
L08806-5
.
doi:10.1029/2007GL032887
.
Hansen
B. E.
1999
Threshold effects in non-dynamic panels: estimation, testing and inference
.
Journal of Econometrics
2
,
345
368
.
https://doi.org/10.1016/S0304–4076(99)00025–1
.
Harris
R. D. F.
&
Tzavalis
E.
1999
Inference for unit roots in dynamic panels where the time dimension is fixed
.
Journal of Econometrics
91
(
2
),
201
226
.
https://doi.org/10.1016/S0304-4076(98)00076-1
.
Hoekstra
R.
&
Jeroen
C. J. M. v. d. B.
2003
Comparing structural decomposition analysis and index
.
Energy Economics
25
(
1
),
39
64
.
https://doi.org/10.1016/S0140-9883(02)00059-2
.
Im
K. S.
&
Shin
Y.-C.
2003
Testing for unit roots in heterogeneous panels
.
Journal of Econometrics
115
(
1
),
53
74
.
https://doi.org/10.1016/S0304-4076(03)00092-7
.
Kang
Y. Q.
,
Zhao
T.
&
Wu
P.
2016
Impacts of energy-related CO2 emissions in China: a spatial panel data technique
.
Natural Hazards
81
(
1
),
405
421
.
https://doi.org/10.1007/s11069-015-2087-x
.
Karen
F.-H.
,
Jefferson
G. H.
,
Ma
J. K.
&
Xu
J. Y.
2006
Technology development and energy productivity in China
.
Energy Economics
28
(
5
),
690
705
.
https://doi.org/10.1016/j.eneco.2006.05.006
.
Levin
A.
,
Lin
C. F.
&
Chu
C.-S. J.
2002
Unit root tests in panel data: asymptotic and finite-sample properties
.
Journal of Econometrics
108
(
1
),
1
24
.
https://doi.org/10.1016/S0304-4076(01)00098-7
.
Li
H. N.
&
Wei
Y. M.
2015
Is it possible for China to reduce its total CO2 emissions?
Energy
83
,
438
446
.
https://doi.org/10.1016/j.energy.2015.02.041
.
Li
L.
,
Hong
X. F.
&
Peng
K.
2019
A spatial panel analysis of carbon emissions, economic growth and high-technology industry in China
.
Structural Change and Economic Dynamics
49
(
C
),
83
92
.
https://doi.org/10.1016/j.strueco.2018.09.010
.
Miao
J. J.
,
Hua
C.
&
Feng
J. C.
2020
Upgrading effect of industrial collaborative agglomeration and carbon emission: an empirical analysis based on spatial econometric model
.
Ecological Economy
36
(
2
),
28
33
.
(In Chinese)
.
Mundaca
L.
&
Markandya
A.
2016
Assessing regional progress towards a ‘Green energy economy’
.
Applied Energy
179
(
C
),
1372
1394
.
https://doi.org/10.1016/j.apenergy.2015.10.098
.
Pekkan
O. I.
,
Senyel
K. M. A.
,
Cabuk
S. N.
,
Aksoy
T.
,
Yilmazel
B.
,
Kucukpehlivan
T.
,
Dabanli
A.
,
Cabuk
A.
&
Cetin
M.
2021
Assessing the effects of wind farms on soil organic carbon
.
Environmental Science and Pollution Research
28
(
14
),
18216
18233
.
https://link.springer.com/article/10.1007/s11356-020-11777-x
.
Raupach
M. R.
,
Marland
G.
,
Ciais
P.
,
Le Quéré
C.
,
Canadell
J. G.
,
Klepper
G.
&
Field
C. B.
2007
Global and regional drivers of accelerating CO2 emissions
.
Proceedings of the National Academy of Sciences of the USA
104
(
24
),
10288
10293
.
doi:10.1073/pnas.0700609104
.
Talukdar
D.
&
Meisner
C. M.
2001
Does the private sector help or hurt the environment? Evidence from carbon dioxide pollution in developing countries
.
World Development
29
(
5
),
827
840
.
https://doi.org/10.1016/S0305-750X(01)00008-0
.
Wang
W. Z.
,
Liu
L. C.
,
Liao
H.
&
Wei
Y. M.
2021
Impacts of urbanization on carbon emissions: an empirical analysis from OECD countries
.
Energy Policy
151
,
112171
.
https://doi.org/10.1016/j.enpol.2021.112171
.
Wu
Y. L.
,
Xiong
Y. L.
,
Tian
X.
&
Shi
F.
2018
Decoding the carbonization mode of the south coastal economic zone in China from the perspective of a dynamic industrial structure
.
Journal of Cleaner Production
199
,
518
528
.
https://doi.org/10.1016/j.jclepro.2018.07.139
.
Yang
Y. Y.
,
Zhao
T.
,
Wang
Y. N.
&
Shi
Z. H.
2015
Research on impacts of population-related factors on carbon emissions in Beijing from 1984 to 2012
.
Environmental Impact Assessment Review
55
,
45
53
.
https://doi.org/10.1016/j.eiar.2015.06.007
.
Zhang
X. Q.
&
Ren
J. L.
2011
The relationship between carbon dioxide emissions and industrial structure adjustment for Shandong Province
.
Energy Procedia
5
,
1121
1125
.
https://doi.org/10.1016/j.egypro.2011.03.197
.
Zhang
P. D.
,
Yuan
H. M.
,
Bai
F. L.
&
Shi
F.
2018
How do carbon dioxide emissions respond to industrial structural transitions? Empirical results from the northeastern provinces of China
.
Structural Change and Economic Dynamics
47
,
145
154
.
https://doi.org/10.1016/j.strueco.2018.08.005
.
Zhang
S.
,
Li
H.
,
Zhang
Q.
,
Tian
X.
&
Shi
F.
2019
Uncovering the impacts of industrial transformation on low-carbon development in the Yangtze River Delta
.
Resources, Conservation & Recycling
150
,
104442
.
https://doi.org/10.1016/j.resconrec.2019.104442
.
Zhang
M.
,
Sun
R.
&
Wang
W.
2021
Study on the effect of public participation on air pollution control based on China's Provincial level data
.
Environment, Development and Sustainability
,
1
14
.
https://doi.org/10.1007/s10668-020-01186-y
.
Zhao
M. X.
,
Lu
L. H.
,
Zhang
C.
&
Luo
H.
2020
The dynamic relationship among energy consumption, economic growth and carbon emissions in China
.
Research of Environmental Sciences
,
1
20
.
https://doi.org/10.13198/j.issn.1001-6929.2020.12.20
.
Zhou
A.
&
Li
J.
2020
The nonlinear impact of industrial restructuring on economic growth and carbon dioxide emissions: a panel threshold regression approach
.
Environ Sci Pollut. Res.
27
,
14108
14123
.
https://doi.org/10.1007/s11356–020–07778–5.
Zhou
X. Y.
,
Zhang
J.
&
Li
J. P.
2013
Industrial structural transformation and carbon dioxide emissions in China
.
Energy Policy
57
,
43
51
.
https://doi.org/10.1016/j.enpol.2012.07.017.
Zhu
Q. Y.
,
Li
X. C.
,
Li
F.
&
Sun
J. S.
2021
Analyzing the sustainability of China's industrial sectors: a data-driven approach with total energy consumption constraint
.
Ecological Indicators
122
,
107235
.
https://doi.org/10.1016/j.ecolind.2020.107235
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY 4.0), which permits copying, adaptation and redistribution, provided the original work is properly cited (http://creativecommons.org/licenses/by/4.0/).