Frontiers in Business, Economics and Management 
ISSN: 2766-824X | Vol. 19, No. 1, 2025 

 

27 

Spatial spillover effects of carbon trading pilot policies 
on regional energy consumption structure 
optimization 

Yuanchun Yu a, Xin Chen b, * 

School of Management, Sichuan University of Science & Engineering, Sichuan 644000, China 

a yyc735654@163.com, b 19882446082@163.com 

 

Abstract: Using panel data covering 30 provinces and cities in China between 2009 and 2019, this study provides an in-depth 
analysis of the spatial spillover effects of carbon trading pilot regions in promoting the optimization of energy consumption 

structure. Through the test of the spatial econometric model, it is found that the carbon trading pilot policy exhibits obvio us 

spatial positive autocorrelation, which positively promotes the optimization of the energy consumption structure of the 

neighboring non-pilot regions. This positive impact also has a time lag characteristic, i.e., its effect on the optimization of the 

energy consumption structure of the neighboring regions gradually appears in the two years after the implementation of the 

policy. From a geographical perspective, the central region is the only region that shows a significant positive spillover effect. 

Keywords: Carbon trading pilot policy; Energy consumption structure optimization; Spatial spillover effect. 

 

1. Introduction 

With the deepening connection between provinces and 

cities in China in terms of economic development, industrial 

structure adjustment and political system formulation, 

scholars have begun to analyze the interconnection of China's 

regional carbon trading pilot policies from the spatial 

dimension. 

In the early stage of the implementation of carbon trading 

policy, some scholars [1] have explored the spatial 

agglomeration phenomenon of carbon emissions of the policy, 

and found that the invisible competition among enterprises 

selling carbon emission rights in the market transaction leads 

to the price of carbon emission rights to be constantly low, 

which triggers the concentration of regional carbon emissions, 

and in the long run, this phenomenon of agglomeration will 

pose a greater threat to the regional ecological environment. 

While Xue Ling and other scholars [2] constructed a spatial 

economic model to explore the dynamic impact between 

carbon emissions trading and enterprise location choice 

behavior, pointing out that the tendency of regional carbon 

policy will change the spatial layout of enterprises, which will 

help narrow the gap between regions. Scholar Mao Chunmiao 

et al [3] studied the spatial spillover effect of carbon trading 

policy on the level of high-quality development from the 

perspective of manufacturing industry, and found that the 

high-quality development of the manufacturing industry in 

the pilot region is not only affected by the local carbon trading 

policy, but also affected by the ripple effect of the 

implementation of carbon trading policy in other provinces 

and cities. Scholar Yu [4], on the other hand, explored the 

spatial spillover effect of carbon trading policy from the 

perspective of enterprise labor demand, and found that the 

policy has a positive spillover effect of about 10% on the 

employment market of enterprises in neighboring provinces 

and cities in the pilot region, and this positive effect is 

gradually increasing. Scholar Zhang [5] used the SCD model 

to study the spatial spillover effect of China's carbon market 

on renewable energy, and found that although the 

implementation of the carbon market inhibited the 

development of renewable energy in the pilot region, it 

promoted the development of renewable energy in the 

neighboring regions, and the spatial spillover effect 

outweighed the direct inhibition effect. 

Since President Xi Jinping announced at the 75th United 

Nations General Assembly on September 22, 2020 that China 

would peak its carbon dioxide emissions by 2030 and achieve 

carbon neutrality by 2060, the issue of carbon emissions has 

been of great concern to academics. The research of Li Zhiguo 

and other scholars [6] shows that the carbon emissions trading 

mechanism has a significant policy spillover effect on carbon 

emissions, but the indirect effect only accounts for 27% of the 

total effect, indicating that the carbon trading policy has a 

poor channel in promoting the spatial spillover effect of 

emission reduction, and it still poses a threat of growth to 

carbon emissions in neighboring regions. The research of Guo 

Li et al [7] shows again that carbon emissions trading not only 

improves the local carbon emissions efficiency, but also has a 

positive role in promoting the carbon emissions efficiency of 

neighboring regions. Scholars Yang Jianping et al [8] 

similarly argued that the spatial spillover effect of the carbon 

trading pilot policy significantly inhibits carbon emissions in 

neighboring regions, and the long-term effect is more 

significant. Wuzheni et al [9] studied the spatial spillover 

effect of the carbon trading pilot policy from the perspective 

of carbon emissions transfer responsibility, and found that the 

implementation of the policy obviously prompted the pilot 

cities to transfer the implied carbon emissions responsibility 

to non-pilot cities. In addition, other scholars have proposed 

that the carbon emissions trading policy has a significant 

spatial spillover effect on energy utilization efficiency [10]. 

At present, most of the assessments on the optimization 

effect of carbon trading pilot policies on energy consumption 

structure focus on the pilot regions of the policies, and few 

studies consider the correlation of energy consumption 

structure between adjacent regions, especially the spatial 

spillover effect of environmental regulations on the 

surrounding areas. This may lead to biased policy assessment 



 

28 

results. Therefore, this study will start from the spatial 

spillover effects of carbon trading pilot policies on energy 

consumption structure optimization, to provide more 

theoretical and empirical support for the study of carbon 

trading pilot policies and energy consumption structure. 

2. Analysis of Spatial Autocorrelation 
Test Results 

Waldo Tobler put forward the first law of geography in 

1970, pointing out that geographic things or their attributes 

have certain spatial correlation. Some scholars have found 

after research that the optimization of energy consumption 

structure in various regions of China presents significant 

spatial autocorrelation. From the theoretical point of view, the 

carbon trading pilot policy has the same potential to promote 

the optimization of the energy consumption structure of 

neighboring regions. Under the current background of the 

country actively advocating the concept of green 

environmental protection and low carbon, the spatial 

correlation of energy consumption structure should become 

an important consideration in the formulation of low carbon 

policies in China. 

2.1. Global Spatial Autocorrelation Test 

In order to examine whether there is a global spatial 

correlation between energy consumption in 30 provinces and 

cities in China, the neighboring spatial weight matrix (W1) 

and the geographic distance weight matrix (W2) are used as 

the basis, respectively. In this paper, the spatial correlation 

between different provinces and cities is firstly examined by 

applying Moran's I index, which is the first step of the 

regression calculation of spatial modeling. The Moran's I 

statistic can be expressed as: 

I =
𝑛 ∑ ∑ 𝜔𝑖,𝑗

𝑛
𝑗=1 𝑧𝑖𝑧𝑗

𝑛
𝑖=1

𝑆0 ∑ 𝑧𝑖
2𝑛

𝑖=1

 

Where 𝑧𝑖 is the deviation of the attribute of element i from 
its mean (𝑥𝑖 − �̅�) , 𝜔𝑖,𝑗  is the spatial weight between 

elements i and j, 𝑛 is equal to the total number of elements, 
and 𝑆0 is the aggregation of all spatial weights: 

𝑆0 = ∑ ∑𝜔𝑖,𝑗

𝑛

𝑗=1

𝑛

𝑗=1

 

The statistical 𝑧𝐼 score is calculated in the following form: 

𝑧𝐼 =
𝐼 − 𝐸[𝐼]

√𝑉[𝐼]
 

Among:  

𝐸[𝐼] = − 1 (𝑛 − 1)⁄  

𝑉[𝐼] = 𝐸[𝐼2] − 𝐸[𝐼]2 

The Moran index lies between (-1, 1), the larger the 

absolute value value indicates the stronger the correlation, 

when the value tends to 1, it indicates that there is a strong 

spatial positive correlation between the explained variables 

and the explanatory variables, if the value tends to 0, it 

indicates that there is no spatial correlation between the 

explained variables and the explanatory variables. 

Able3-1 shows the results of the spatial autocorrelation test 

of energy consumption structure indicators for 30 provinces 

and cities in China under two different spatial weight matrices 

during the period from 2009 to 2019. Specifically, in the 

context of the spatial neighbor matrix (labeled W1), the global 

Moran indices of energy consumption structure lie in the 

range of 0.182 to 0.313, and these index values all show 

positive values at the 5% significance level. In contrast, in the 

context of the geographic distance weight matrix (labeled 

W2), the global Moran index lies in the range of 0.066 to 

0.180, and most of the index values are significant at the 10% 

significance level. Together, these results indicate that there is 

a significant positive correlation between the energy 

consumption structures of the provinces and cities in China. 

 

Table 1. Values of the Moran Index of Energy Consumption 
Structure Indicators, 2009-2019 

Year W1  W2  

 I p-value* I p-value* 

2009 0.236 0.015*** 0.118 0.072* 

2010 0.182 0.036** 0.066 0.159 

2011 0.296 0.004*** 0.133 0.055* 

2012 0.313 0.003*** 0.180 0.020** 

2013 0.296 0.004*** 0.156 0.034** 

2014 0.286 0.005*** 0.143 0.045** 

2015 0.243 0.013** 0.104 0.091* 

2016 0.261 0.009*** 0.130 0.057* 

2017 0.248 0.011** 0.109 0.084* 

2018 0.264 0.008*** 0.108 0.085* 

2019 0.261 0.009*** 0.105 0.090* 

 

(Note: ***, **, * indicate that the variable is statistically 

significant at the 1%, 5%, and 10% levels, respectively) 

2.2. Local Spatial Autocorrelation Test 

In the analysis process, although the global Moran index 

can reveal the overall characteristics of the explanatory and 

interpreted variables, it may fail to adequately reflect the 

specific degree of spatial correlation among provinces and 

cities. To compensate for this limitation, the local Moran's 

index is used to detect outliers as well as the specific extent 

and location of the clustering of high and low values. In this 

paper, based on the spatial adjacency matrix (W1) and the 

geographic distance weight matrix (W2), local Moran scatter 

plots of the energy consumption structure of 30 provinces and 

cities in China in 2015 and 2018 are plotted to explore the 

spatial autocorrelation of each region in greater depth. 

As shown in Figures1 and 2, whether in 2015 or 2018, the 

fitted regression line mostly crosses the first and third 

quadrants, presenting a clustered distribution of “high - high, 

low - low”. Specifically, the provinces and cities located in 

the first quadrant indicate that the degree of optimization of 

energy consumption structure indicators is high, and the 

degree of optimization of the surrounding areas is also 

relatively high; while the provinces and cities located in the 

third quadrant indicate that the degree of optimization of 

energy consumption structure indicators is low, and the 

degree of optimization of the surrounding areas is also 

relatively low. This distribution feature is clearly shown by 

the Moran scatter plot in the first and third quadrants, which 

further confirms the significant spatial agglomeration and 

positive autocorrelation of the optimization level of China's 

energy consumption structure. 

 



 

29 

 

 
Figure 1. Moran Scatter Plot of Energy Consumption Structure Indicators for 2015 and 2018 based on the W1 Matrix under the 

 



 

30 

 

 
Figure 2. Moran Scatter Plot of Energy Consumption Structure Indicators for 2015 and 2018 based on the W2 Matrix under the 

3. Spatial Econometric Model 
Construction 

By analyzing the global and local Moran indices, we find 

that the carbon trading pilot policy and the energy 

consumption structure indicators exhibit significant spatial 

dependence at both the aggregate and local levels, which 

prompts us to consider constructing a spatial econometric 

model to test the spatial spillover effect between these two 

variables. Given the existence of three basic spatial 

econometric models to choose from, we adopt the LM test to 

determine the type of spatial relationship between carbon 

trading pilot policies and energy consumption structure 

optimization indicators. 

The LM test bases its judgment on the results of LM-Log 

and LM-Error. When both are not significant, it indicates that 

there is no spatial relationship between the two variables, and 

an ordinary regression model should be used; if LM-Log is 

significant and LM-Error is not, a spatial lag model (SLM) is 

more appropriate; on the contrary, if LM-Log is not 

significant and LM-Error is significant, then a spatial error 

model (SEM) should be chosen; and if both are significant, 

then a further Wald and LR tests to make a decision. 

In this paper, two matrices, W1 and W2, were utilized and 

the LM test results in Table 2 were referred to. The results 

show that the coefficients of LM-error, Robust LM-lag, and 

LM-error of the energy consumption structure optimization 

index are all positive, and their P-values are less than 0.05, 

passing the 5% significance test. This indicates that the spatial 



 

31 

effect of the carbon trading pilot policy on the optimization of 

energy consumption structure contains both the spatial error 

effect and the lag effect, so we choose the spatial Durbin 

model (SDM) as the optimal spatial measurement model. 

 

Table 2. LM test results 

Test  LM-lag Robust LM-lag LM-error Robust LM-error 

energy consumption 

structure 

W1 
53.821*** 

(0.000) 

6.537** 

(0.011) 

108.213*** 

(0.000) 

60.929*** 

(0.000) 

W2 
85.647*** 30.281*** 56.456*** 1.090 

(0.000) (0.000) (0.000) (0.296) 

To ensure the robustness and exact selection of the model, 

we further employ the Wald and LR tests to verify whether 

the spatial Durbin model (SDM) reduces to a spatial error 

model or a spatial autoregressive model. Specifically, if the 

Wald-lag test result is not significant, it indicates that the 

SDM can be reduced to a spatial autoregressive model; if the 

Wald-error test result is not significant, it implies that the 

SDM can be reduced to a spatial error model; similarly, if the 

LR-SDM-SEM test result is not significant, the SDM will be 

reduced to a spatial error model; if the LR-SDM-SAR test is 

not significant, the SDM will reduce to a spatial 

autoregressive model. 

We conducted Wald and LR tests using both W1 and W2 

matrices and summarized the results in Table 3. In the context 

of the W1 matrix, the coefficients of Wald-lag, Wald-error, 

LR-SDM-SEM, and LR-SDM-SAR are all significant at the 

1% significance level, which means that we rejected the 

original hypothesis, i.e., the SDM model is not degraded, and 

passed the robustness test. 

 

Table 3. Wald, LR test results 

Test  Wald-lag Wald-error LR-SDM-SEM LR-SDM-SAR 

Energy consumption 

structure 

W1 
40.60*** 

(0.0000) 

44.30*** 

(0.0000) 

41.68*** 

(0.0000) 

38.48*** 

(0.0000) 

W2 
64.66*** 

(0.0000) 

55.52*** 

(0.0000) 

52.53*** 

(0.0000) 

60.68*** 

(0.0000) 

4. Analysis of Spatial Econometric 
Model Results 

Before the model estimation, it is also necessary to carry 

out the Hausman test to determine whether the model applies 

fixed effects or random effects. According to Table 4, it can 

be seen that the Hausman statistic of the index of the degree 

of optimization of energy consumption structure under the 

two spatial matrices is 33.76 and 34.16, with a P-value of 

0.0000, which is less than 0.01, and the test result is 

significant at the 1% level, i.e., the original hypothesis of the 

random effect model is rejected, and a fixed effect model is 

chosen to do the subsequent empirical analysis. 

 

Table 4. Space Hausman test results 

VARIABLES W1 W2 

Chi2 statistic 33.76*** 34.16*** 

P (0.0000) (0.0000) 

Observations 330 330 

self-selection effect fixed effect fixed effect 

4.1. Spatial Spillover Analysis of The Effect of 
Carbon Trading Pilot Policy on 

Optimizing Regional Energy Consumption 

Structure 

In this paper, the level of industrialization (INL), the share 

of industrial value added in GDP is used. For government 

intervention (GC), the share of local general budget 

expenditures in GDP is selected. The level of external 

development (O) is measured by the amount of regional 

foreign-invested enterprise investment, which is converted to 

RMB using the current year's exchange rate, and is 

characterized by its share of regional GDP. The level of 

economic development (GDP), expressed in logarithmic 

terms as per capita GDP at constant 2008 prices. The level of 

urbanization (U) is the share of urban population in the total 

population of both regions. 

Tables 5 and 6 show the regression results of spatial 

econometric models for the indicators of the degree of 

optimization of energy consumption structure in detail, while 

Tables 7-5 and 7-6 present the estimation results of the four 

models, SEM (spatial error model), SAR (spatial 

autoregressive model), SDM (spatial Durbin model), and 

dynamic SDM, respectively, under the neighboring spatial 

weighting matrix (W1) and the geographic distance weighting 

matrix (W2). Results. Comparing the data in these tables, we 

can derive the following analysis: 

The coefficients of the dummy variable policytreat are all 

significantly positive at the 1% statistical level in the SEM, 

SAR and SDM models, but in the dynamic SDM model, the 

coefficient is significantly negative at the 1% level. This 

indicates that the carbon trading pilot policy has played a 

significant role in promoting the optimization of energy 

consumption structure in pilot provinces and cities, while the 

spillover effect on adjacent non-pilot provinces and cities has 

a certain lag. However, in general, the policy still has a certain 

positive impact on the optimization of energy consumption 

structure in adjacent non-pilot provinces and cities. This may 

be since the pilot regions have fulfilled their emission 

reduction tasks by adjusting their resource consumption 

structure and other measures, which has brought pressure and 

challenges to the neighboring non-pilot regions, and at the 

same time provided experience and reference, prompting the 

neighboring regions to take corresponding measures to 

optimize their resource consumption structure and thus 

reduce GHG emissions. 

 



 

32 

Table 5. Estimation results of SEM, SAR, SDM and dynamic SDM for energy consumption structure under W1 

Variable SEM SAR SDM  Dynamic state SDM  

 Main Main Main Wx Main Wx 

L. policytreat     
0.8173*** 

(4.68) 
 

policytreat 
0.0068*** 

(3.46) 

0.0067*** 

(3.48) 

0.0068*** 

(3.54) 

0.0102** 

(2.55) 

0.0131*** 

(3.47) 

-0.0128* 

(-1.76) 

Gdp 
0.0011 

(0.75) 

0.0010 

(0.68) 

-0.0008 

(-0.54) 

-0.0011 

(-0.43) 

0.0064** 

(2.30) 

0.0028 

(0.61) 

O 
-0.0003* 

(-1.94) 

-0.0003* 

(-1.93) 

-0.0004** 

(-2.38) 

-0.0009** 

(-2.31) 

0.0003** 

(2.18) 

0.0002 

(0.63) 

U 
-0.0454*** 

(-2.70) 

-0.0455*** 

(-2.73) 

-0.0414** 

(-2.51) 

0.0331 

(0.88) 

-0.0360*** 

(-5.56) 

0.04328** 

(2.49) 

Gc 
-0.0406** 

(-2.43) 

-0.0429*** 

(-2.57) 

-0.0463*** 

(-2.77) 

-0.0189 

(-0.58) 

-0.0098 

(-0.62) 

-0.0540 

(-1.45) 

Inl 
0.0342*** 

(3.16) 

0.0326*** 

(3.36) 

0.0164* 

(1.65) 

0.0969*** 

(5.73) 

-0.0039 

(-1.46) 

-0.0228*** 

(-2.84) 

lambda 
0.08 

(0.86) 

0.1579** 

(2.02) 

0.09 

(1.12) 
 

0.0314 

(0.31) 
 

sigma2 
0.0001*** 

(12.83) 

0.0001*** 

(12.8) 

0.0001*** 

(12.83) 
 

0.0001*** 

(13.47) 
 

Province Yes Yes Yes Yes Yes Yes 

Year Yes Yes Yes Yes Yes Yes 

Observations 330 330 330 330 330 330 

R-squared 0.7263 0.7349 0.7197  0.6255  

 

Table 6. Estimation results of SEM, SAR, SDM and dynamic SDM of energy consumption structure under W2 

Variable SEM SAR SDM  Dynamic state SDM  

 Main Main Main Wx Main Wx 

L. policytreat     
-0.4956*** 

(-3.29) 
 

policytreat 
0.0089*** 

(4.61) 

0.0073*** 

(3.79) 

0.0072*** 

(3.95) 

0.0148*** 

(3.18) 

0.0076*** 

(4.04) 

0.0173*** 

(3.61) 

Gdp 
0.0011 

(0.71) 

0.0008 

(0.54) 

0.0020 

(1.40) 

0.0005 

(0.16) 

0.0010 

(0.74) 

-0.0000 

(-0.01) 

O 
-0.0004*** 

(-2.76) 

-0.0004** 

(-2.38) 

-0.0004*** 

(-2.68) 

-0.0006 

(-1.31) 

-0.0004*** 

(-2.85) 

-0.0009* 

(-1.68) 

U 
-0.0534*** 

(-3.23) 

-0.0457*** 

(-2.71) 

-0.0346** 

(-2.21) 

-0.3255*** 

(-6.68) 

-0.0374** 

(-2.24) 

-0.3343*** 

(-6.59) 

Gc 
-0.0543*** 

(-3.11) 

-0.0421** 

(-2.51) 

-0.0549*** 

(-3.43) 

-0.0916*** 

(-2.64) 

-0.0639*** 

(-3.78) 

-0.1262*** 

(-3.39) 

inl 
0.0501*** 

(5.41) 

0.0396*** 

(4.11) 

0.0295*** 

(3.03) 

0.0293*** 

(2.73) 

0.0400*** 

(3.49) 

0.0412*** 

(3.21) 

lambda 
0.3254*** 

(-2.96) 

-0.024 

(-0.25) 

-0.2562** 

(-2.52) 
 0.1423(1.25)  

sigma2 
0.0001*** 

(12.68) 

0.0001*** 

(12.87) 

0.0001*** 

(12.69) 
 

0.0001*** 

(13.4) 
 

Province Yes Yes Yes Yes Yes Yes 

Year Yes Yes Yes Yes Yes Yes 

Observations 330 330 330 330 330 330 

R-squared 0.7258 0.7233 0.7870  0.7467  

Observing the regression results of other control variables 

on the indicators of the degree of optimization of energy 

consumption structure, we can find that, on the one hand, the 

urbanization level, the degree of government intervention and 

the level of opening up to the outside world of the pilot 

regions have negative spatial spillover effects on the 

optimization of the energy consumption structure of non-pilot 

regions. Specifically, the higher level of urbanization, the 

stronger degree of government intervention, and the higher 

level of opening up to the outside world in the pilot regions 

may inhibit the optimization of energy consumption structure 

in the surrounding non-pilot regions. This may be due to the 

fact that the urbanization process in the pilot provinces and 

cities intensifies the demand for energy resources and creates 

resource competition with neighboring regions; the spillover 

effects of government policies may lead to difficulties in 

policy implementation in neighboring regions; and the effects 

of technology introduction and diffusion brought about by 

opening up to the outside world may be limited by the 

absorption capacity and innovation capacity in neighboring 

regions. On the other hand, the industrialization level of the 

pilot regions has a significant positive spillover effect on the 

optimization of the energy consumption structure of the non-

pilot regions. As the industrialization level of the pilot regions 

increases, their industrial chain extends to the surrounding 

non-pilot regions, promoting the optimization and upgrading 



 

33 

of the industrial structure in the adjacent regions, and thus 

promoting the optimization of the energy consumption 

structure. 

In summary, the carbon trading pilot policy has an impact 

on the optimization of the energy consumption structure in 

both the pilot and adjacent non-pilot regions, but there are 

differences in the way and extent of the impact. Meanwhile, 

other control variables also have significant spatial spillover 

effects on the optimization of energy consumption structure. 

In order to more clearly articulate the specific impact of the 

dummy variables in the pilot regions on the neighboring non-

pilot regions, we need to consider the direct and spatial 

spillover effects of the dummy variables on the explanatory 

variables in an integrated manner. Therefore, we further 

decompose the effects of policy implementation on the 

explanatory variables. Among them, the direct effect 

measures the direct impact of the carbon trading pilot policy 

on the optimization of energy consumption structure in the 

pilot regions; the indirect effect reveals the facilitating effect 

of the carbon trading pilot policy on the optimization of 

energy consumption structure in the non-pilot regions through 

the spatial spillover mechanism; and the total effect integrally 

reflects the overall impact of the carbon trading pilot policy 

on all provinces and municipalities in terms of optimization 

of energy consumption structure. 

Table 7 shows in detail the impacts of the direct, indirect 

and total effects of the carbon trading pilot policy on the 

optimization of energy consumption structure under the two 

matrix frameworks of W1 and W2. By analyzing the table, the 

following conclusion can be drawn: at the 1% statistical 

significance level, the direct, indirect and total effects of 

carbon trading policies on the optimization of energy 

consumption structure are all significantly positive. This 

means that the carbon trading pilot policy not only positively 

promotes the optimization of energy consumption structure in 

all regions, but also positively promotes the optimization of 

energy consumption structure in the surrounding neighboring 

regions through spatial spillover effects. 

 

Table 7. Effect decomposition of the impact of carbon trading pilot 
policy on energy consumption structure 

Test  LR_Direct LR_Indirect LR_Total 

energy 

consumption 

structure 

W1 
0.0071*** 

(3.62) 

0.0116*** 

(2.65) 

0.0187*** 

(3.79) 

W2 
0.0067*** 

(3.52) 

0.0106*** 

(2.65) 

0.0173*** 

(4.38) 

4.2. Analysis of Spatial Spillover Effect 
Regression Results Based on Regional 

Heterogeneity 

In order to deeply study the spatial effects of the impact of 

carbon trading pilot policies on the optimization of energy 

consumption structure in different regions, this paper 

conducted a spatial econometric model test based on the 

neighboring spatial weight matrix (W1) and geographic 

distance weight matrix (W2), using the spatial Durbin model 

with time and individual double fixed effects for samples in 

the eastern, central and western regions during the period of 

2009-2019, and the regression results are detailed in Tables 8. 

From the regional level, there are significant differences in 

the impact of carbon trading pilot policies on the optimization 

of energy consumption structure. Specifically, the regression 

coefficients of the dummy variables in the central region are 

significantly positive at the 1% level, indicating that the 

carbon trading pilot policy in the central region effectively 

promotes the optimization of energy consumption structure in 

the region. In contrast, the regression coefficients of the 

dummy variables in the eastern and western regions do not 

pass the significance test, indicating that the carbon trading 

pilot policies in these regions do not have a significant direct 

impact on the optimization of energy consumption structure. 

In the test results of the spatial econometric model under the 

W1 matrix, the coefficient of the dummy variable in the 

western region is negative, while it turns positive under the 

W2 matrix, which may be due to the influence of different 

spatial weight matrices on the direction and intensity of the 

spatial spillover effect. 

In order to further analyze the relationship between the 

spatial effects of the carbon trading pilot policy on the 

optimization of the energy consumption structure in the 

region and adjacent regions, the direct and indirect effects of 

the variables under the SDM model are decomposed and 

tested in Tables 9. The results show that the carbon trading 

pilot policy in the central region has a significant positive 

effect on the optimization of its own energy consumption 

structure (the coefficient is 0.0165, significant at the 1% level), 

and has a significant positive spillover effect on the 

optimization of the energy consumption structure in the 

neighboring regions of the region (significant at the 5% level). 

However, neither the direct nor indirect effects are significant 

for the eastern and western regions. 

For the eastern region, although its industry, energy 

restructuring and economic development are in a leading 

position, providing a good economic environment and 

institutional foundation for the implementation of the carbon 

trading pilot policy, the direct effect of the carbon trading pilot 

policy on the optimization of energy consumption structure is 

not obvious under the existing energy consumption 

constraints as well as the policy conditions because the 

existing energy consumption structure is already close to the 

optimal state. At the same time, the indirect effect of the 

policy on the optimization of the energy consumption 

structure of the neighboring non-pilot provinces and cities is 

also not significant enough. 

In contrast, in the western region, due to the relatively 

single industrial structure, the dominant industries are energy- 

or resource-consuming traditional industries with high energy 

consumption, high pollution, and low efficiency, and the level 

of economic development is relatively low, and the research 

and development and application of clean energy and low-

carbon technologies are also relatively lagging. Therefore, the 

effect of the carbon trading pilot policy on the optimization of 

the energy consumption structure in the western region is not 

significant, and the pilot provinces and cities have not been 

able to better drive the adjustment and optimization of the 

energy consumption structure of the neighboring non-pilot 

provinces and cities. 

 

 

 

 

 



 

34 

Table 8. Model regression results of SDM with spatial effect based on regional heterogeneity 

 W1   W2   

 East Middle West East Middle West 

policytreat 
0.0014 

(0.63) 

0.0202*** 

(6.34) 

-0.0032 

(-0.63) 

0.0012 

(0.55) 

0.0223*** 

(5.43) 

0.003 

(0.56) 

lambda 
-0.1893* 

(-1.70) 

-0.3819*** 

(-3.58) 

-0.362** 

(-2.30) 

-0.2658** 

(-2.04) 

-0.3817*** 

(-2.64) 

-0.2828** 

(-2.50) 

sigma2 
0.0001*** 

(7.68) 

0.0000*** 

(6.30) 

0.0001*** 

(7.59) 

0.0001*** 

(7.70) 

0.0001*** 

(6.69) 

0.0001*** 

(7.71) 

Control Yes Yes Yes Yes Yes Yes 

Province Yes Yes Yes Yes Yes Yes 

Year Yes Yes Yes Yes Yes Yes 

Observations 121 88 121 121 88 121 

R-squared 0.5631 0.8123 0.6754 0.0693 0.8278 0.7801 

 

Table 9. Spatial Durbin model regression decomposition results based on regional heterogeneity 

 East  Middle  West  

 LR_Direct LR_Indirect LR_Direct LR_Indirect LR_Direct LR_Indirect 

W1 
0.0013 

(0.57) 

0.0022 

(0.64) 

0.0165*** 

(5.70) 

0.0171** 

(2.36) 

-0.0010 

(-0.21) 

-0.0195 

(-1.33) 

W2 
0.0011 

(0.47) 

0.0026 

(0.64) 

0.0186*** 

(5.20) 

0.0229** 

(2.51) 

0.0032 

(0.61) 

0.0001 

(0.00) 

5. Main Findings and 
Recommendations for 
Countermeasures 

5.1. Main Research Conclusions 

Based on referring to and sorting out many domestic and 

international literature, this paper explores the spatial 

spillover effect of the carbon trading pilot policy by using 

spatial econometric modeling based on the panel data of 30 

provinces and municipalities (except Tibet) in China from 

2009 to 2019. The main conclusions are as follows: through 

the spatial econometric model test, it is found that the carbon 

trading pilot policy has spatial positive autocorrelation on the 

optimization of the energy consumption structure in the 

region, which is characterized by the spatial distribution of 

“high-high, low-low”, and at the same time, it has a 

significant role in promoting the optimization of the energy 

consumption structure of the neighboring non-pilot regions. 

From the national level, the carbon trading pilot policy has a 

positive spatial spillover effect, and from the sub-regional 

perspective, only the carbon trading pilot policy in the central 

region has a significant positive spillover effect on the energy 

consumption structure of the neighboring non-pilot regions. 

5.2. Countermeasure Suggestions 

Based on the optimization impact and spatial spillover 

effect of carbon trading pilot policy on energy consumption 

structure, this paper puts forward the following suggestions: 

Strengthen the implementation of carbon trading policy in 

the central region. Given that the carbon trading pilot policy 

in the central region has a significant positive impact on the 

optimization of energy consumption structure, it is 

recommended to further increase the implementation of the 

policy in the region, and to promote the continuous 

optimization of energy consumption structure by means of 

perfecting the market mechanism, strengthening regulation 

and incentives. 

Explore the new path of carbon trading policy in the eastern 

region. Although the direct impact of the carbon trading pilot 

policy in the eastern region is not significant, its economic 

environment and institutional foundation are good. Therefore, 

it is recommended to explore new policy paths suitable for the 

characteristics of the eastern region, such as combining 

industrial upgrading, technological innovation and green 

transformation strategies to seek new breakthroughs in 

optimizing the energy consumption structure. 

Increase policy support and guidance for the western region. 

Due to the single industrial structure and relatively low level 

of economic development in the western region, the impact of 

the carbon trading pilot policy on the optimization of its 

energy consumption structure is limited. Therefore, it is 

recommended to increase policy support and guidance for the 

western region, and through financial support, technical 

assistance and talent cultivation measures, help it to enhance 

the R&D and application capacity of clean energy and low-

carbon technologies, and promote the transformation and 

upgrading of the energy consumption structure. 

Strengthen inter-regional policy coordination and 

cooperation. Considering the spatial spillover effect of carbon 

trading pilot policies, it is recommended to strengthen inter-

regional policy coordination and cooperation, establish cross-

regional policy synergy mechanisms, and jointly promote the 

optimization and upgrading of energy consumption structure. 

Maximize the effect of the policies through information 

sharing, experience exchange and joint actions. 

Improve the construction and supervision of the carbon 

trading market. In order to give full play to the role of carbon 

trading policies in the optimization of energy consumption 

structure, it is recommended to further improve the 

construction and supervision mechanism of the carbon 

trading market. This includes strengthening the construction 

of market infrastructure, enhancing market transparency, 

strengthening supervision and risk prevention, etc., to ensure 

the healthy and orderly development of the carbon trading 

market. 

Promote the diversification of energy consumption 

structure. In the process of optimizing the energy 

consumption structure, attention should be paid to promoting 

the diversified development of energy consumption. Through 



 

35 

the development of renewable energy, improving energy 

utilization efficiency, promoting clean energy and other ways, 

reduce the degree of dependence on traditional energy sources, 

and improve the stability and sustainability of the energy 

consumption structure. 

Acknowledgement 

This research was Supported by The Innovation Fund of 

Postgraduate, Sichuan University of Science & Engineering 

(Y2023037). 

References 

[1] Yanyan Zhang and Xiangqin Xu, Spatial Agglomeration 
analysis of carbon emission trading J Machinery 
Manufacturing. 2011. 49(03): p. 74-77. 

[2] Ling Xue, et al., The impact of carbon market integration on 
the spatial distribution of heterogeneous firms J Population, 
Resources and Environment in China. 2018. 28(08): p. 1-11. 

[3] Chunmiao Mao and Yingliang Huang, Carbon trading policies 
and high-quality development in manufacturing: spatial 
spillovers J Energy and the Environment. 2023(05): p. 68-72. 

[4] Yu, D.-J. and J. Li, Evaluating the employment effect of 
China’s carbon emission trading policy: Based on the 
perspective of spatial spillover. Journal of Cleaner Production, 
2021. 292: p. 126052. 

[5] Zhang, M., et al., Impacts of carbon emission trading schemes 
on the development of renewable energy in China: Spatial 
spillover and mediation paths. Sustainable Production and 
Consumption, 2022. 32: p. 306-317. 

[6] Zhiguo Li and Jie Wang, Spatial Emission Reduction effects of 
carbon emission Trading in China: quasi-natural experiments 
and policy spillovers J Population, Resources and Environment 
in China. 2021. 31(01): p. 26-36. 

[7] Li Guo and Xinli Liang, The Impact of carbon emission trading 
policies on carbon emission efficiency: Analysis of J price 
theory and practice based on spatial multi-period differential 
model. 2023(09): p. 146-150+209. 

[8] Jianping Yang and Juan Yang, Carbon Reduction and spatial 
spillover effects of low-carbon pilot policies. 2023(01): p. 147-
154. 

[9] Zhenni Wu, Yingkai Yin, and Ming Jin, Research on 
Interregional Implied Carbon Emission Transfer 
Responsibility under the spillover effect of China's Carbon 
Trading Pilot Policy J Statistics and Information Forum. 2024. 
39(07): p. 82-96. 

[10] Yue Bi, Carbon Emission trading System and Energy use 
efficiency and their spatial spillover effects. 2023.