Spatial Impact of Industrial Structure Upgrading and Corporate Social Responsibility on Carbon Emissions: Evidence from China

Abstract

Identifying the spatial attributes of economic, social and environmental development is a prerequisite for China to raise the quality of development. Based on the parallel data of 30 Chinese provinces from 2010 to 2019, this study uses a spatial econometric model to explore the spatial impacts of corporate social responsibility (CSR) and industrial structure upgrading on carbon emissions. The regulating effect of CSR during industrial structure adjustment for carbon emission reduction was also analyzed. It was found that regional carbon emissions were reduced due to CSR and industrial structure adjustment, and the former was beneficial during industrial structure adjustment for carbon emission reduction. The carbon emissions in neighboring areas have also been suppressed to some extent as a result of industrial structure adjustment. However, CSR can encourage some “free riding” behaviors, due to the economic externalities, which emit more carbon into the surroundings. Additionally, carbon emissions show different responses to CSR in various regions. Therefore, strategies must enhance overall social responsibility and formulate different policies in various regions to promote CSR as an influential factor in curbing carbon emissions. Moreover, spatial governance should consider the comparative advantages of different regions, form complementary advantages and fully enhance the cooperation between CSR and industrial structure upgrading on carbon emission reduction.

1. Introduction

China’s gross domestic product (GDP) in 2020 was over USD14 trillion, ranking second in the world, and its carbon emission reached 1.016 × 10¹⁰ tons, ranking first in the world. From the perspective of “integrated sustainability”, sustainable development’s environmental and economic functions interact considerably, meaning that economic growth and carbon emissions have not yet decoupled [1,2]. This relationship especially matters to China, which has entered a stage of high-quality development while still maintaining medium–high growth. Urbanization and industrialization require increasing amounts of energy. Thus, with economic growth, the total carbon emissions will increase further in the coming years. As a result, China will face long-term pressure to curb carbon emissions [3,4]. In September 2020, China officially put forward the goal of “dual carbon” at the 75th session of the United Nations, according to which the goals are to reach a carbon peak by 2030 and carbon neutrality by 2060. Achieving the “dual carbon” goal as scheduled requires completing the industrial layout in advance and forming a green industrial structure with a low-carbon mode of production [5].

As the carriers of regional economic behavior, enterprises play a critical role in regional economic development by affecting its scale and structure [6,7]. It is thought-provoking that the interaction between enterprises and regional economies is changing due to the awakening of CSR awareness. Traditionally, enterprises have played the role of a simple “economic man”. A pertinent difference between enterprises and regions is the difference in development goals. The survival and development of enterprises depend on obtaining more profits, whereas regional development relies on promoting the coordinated development of regions. Driven by their interests, enterprises accelerate the growth of the surrounding regions through trickle-down and network effects, while transferring the environmental pollution within the region to the surrounding areas [8,9]. As enterprise competition and efficiency change, the sense of CSR makes enterprises gradually transform into “social people” alongside increasingly prominent resource and environmental problems [10,11]. Driven by low-carbon production and consumption, a low-carbon economy is gradually coming into being [12,13].

CSR is the responsibility that enterprises bear to society, the environment, communities, customers, and employees when making profits and meeting legal responsibilities [14,15]. Based on that concept, enterprises should abandon the traditional idea of putting profits first and emphasize the social and environmental effects of their production and operations. In the current complex business ecosystem, the historic money model fails to evaluate companies’ business activities because it cannot comprehensively evaluate enterprise value.

The CSR report, especially the nonfinancial information in it, has become the critical path for the market supervision of enterprises. In the 21st century, low-carbon economies have become increasingly important as global climate change intensifies. CSR reports of many multinational enterprises have shifted their focus from employee and community responsibility to environmental protection and carbon emissions [16,17]. The present study proposes many questions. Does fulfilling CSR and industrial structure adjustment always result in lower carbon emissions? Is there a spatial effect? What role does CSR play in industrial structure adjustment during the efforts of carbon emission reduction? Do the result vary in different regions? To answer the above questions, this study used the parallel data from 30 Chinese provinces from 2010 to 2019 to explore the spatial and regulatory impacts of CSR and industrial structure adjustment on carbon intensity through a spatial Durbin model (SDM).

The purpose of this study is to explore the practical path of carbon emission reduction from the perspectives of industry and enterprise. To this end, industrial structure upgrading and CSR were incorporated as the core explanatory variables into the SDM, and those variables’ direct effects and spillover effects on regional carbon emissions across the country were explored. Subsequently, the key role of enterprises in the scale and structure of regional economic development was taken into account. The interaction terms of industrial structure upgrading and CSR were added into the benchmark model to study the moderating effect of CSR on industrial structure upgrading to promote carbon emission reduction. Due to differences in geographical location and resource endowment, there were obvious differences in regional development in China’s Eastern, Central and Western regions. Therefore, the heterogeneous effects of industrial structure upgrading and CSR on carbon emissions was also one of the focuses of this work. Through this study, we hope to provide valuable theoretical research insights for the economic community, while providing policy recommendations for carbon emission reduction in China or other countries and regions in practice.

This study makes the following contributions: (1) The demonstration and competition of CSR and industrial structure adjustment were considered. Based on the spatial perspective, a spatial econometric model was constructed to explore the direct and spatial spillover effects of industrial structure upgrading and CSR on regional carbon emissions. Additionally, CSR’s positive regulatory role in upgrading industrial structure to curb CE was analyzed. (2) The important roles of industries and enterprises in regional development were considered. This study matched the micro-data of enterprises with the macro-data of industrial structure upgrading and carbon emission, and explored the effective path of carbon emission reduction from the perspective of industry and enterprise. (3) There were obvious differences in industrial development between Eastern, Central, and Western China. The research also analyzed the heterogeneous responses of carbon emissions in various areas to CSR and industrial structure upgrading.

The second part is a literature review of this field. Then, in the third part, the design intention, process, methods, and models are introduced. The empirical study is reported and interpreted in the fourth part. The final part is the conclusion, in which some practical and theoretical suggestions are offered.

2. Literature Review

Industrial structure upgrading implies that tertiary industry has increased more than secondary industry [18,19]. Because China has entered a stage of high-quality development, upgrading the industrial structure has become an internal driver and an important part of modernizing economy. Many scholars believe that to cut the increasing carbon emissions, industrial structure upgrading is the preferable approach [20,21,22,23]. Yet, many researchers dispute this viewpoint. For example, Yang et al., (2018) [24] studied how aging influences carbon emissions and industrial structure. They found an N-shaped correlation between carbon intensity and industrial structure. Additionally, Wang et al., (2019) [25] and Zheng et al., (2020) [26] showed that the industrial influence on carbon footprints is variable and substantially differs in various regions. Li et al., (2021) [27] confirmed that there is an obvious threshold for the effect of industrial structure upgrading on carbon emissions. Yang et al., (2022) [28] found that the effect of industrial structure evolution on carbon emissions showed a V-shaped trend of first acceleration and then inhibition. In addition, Casler et al., (1998) [29] analyzed the carbon emission structure of the United States and concluded that alternative energy sources are primarily responsible for emission reduction. Similarly, Schipper et al., (2001) [30] investigated the evolution of manufacturing carbon footprints in 13 member countries of the International Energy Agency (IEA) and claimed that the change in carbon footprint relies mainly on the energy consumption structure. They also found that industrial structure had little influence on CO₂ emission trends and carbon emission intensity. Various studies have focused on the connection between industrial structure adjustment and carbon footprints, but few researchers concentrate on the spatial effect of industrial structure adjustment on carbon emission reduction. That research gap is also a focus of this paper.

Sheldon (1930) [31], a British scholar, first proposed the concept of CSR. He emphasized that enterprises should consider social and environmental goals other than financial profit. On this basis, Elkington (1998) [32] used the three pillars of sustainable development to divide CSR practices into economic, environmental, and social aspects. CSR is a widely used concept in enterprise management, corporate governance, and finance [33,34,35,36]. In recent years, some scholars have also introduced this concept into the field of macro research. For example, Pivato et al., (2007) [37] investigated organic food consumers and found that CSR affects consumer behavior. Škare et al., (2014) [38] investigated the influence of the governments’ social responsibility support policies and found that enterprises’ social responsibility performance is essential in enhancing economic growth. Shahzad et al., (2020) [39] took the manufacturing industry in Pakistan as an example to study how CSR influences environmental sustainability and green innovation, arguing that all dimensions of CSR are environmentally friendly and beneficial for sustainable development. Li et al., (2022) [40] explored the problem of CSR from the three levels of basic, senior, and super-social responsibility, and compared it with the third distribution system, showing that CSR is consistent with common prosperity.

In the 21st century, with the intensification of the climate crisis and the rise of low-carbon development, CSR reports have gradually shifted their focus to content related to environmental protection and carbon emissions [41,42,43]. Intuitively, CSR always helps to reduce carbon emissions [44,45]. However, studies by Fukuda et al., (2020) [46] show that when carbon emission reduction is cost-inefficient and environmental damage is serious, CSR will mislead companies to emit more carbon. However, there are few reports on the correlation between carbon footprints and CSR, especially for quantitative studies.

In conclusion, existing research has ignored the spatial effect of industrial structure adjustment on carbon reduction. Quantitative studies on the relationship between CSR and carbon emissions are also rare. Therefore, this research aimed to explore the spatial impacts of CSR and the industrial structure adjustment on carbon reduction, and the regulatory effects of CSR during industrial structure adjustment for carbon emission reduction. The parallel data of 30 Chinese provinces from 2010 to 2019 were used to construct an SDM, which was then divided into three geographical regions to measure the heterogeneous responses of regional carbon footprints to CSR and industrial structure upgrading.

3. Model and Data

3.1. Variables

3.1.1. Dependent Variable

Carbon Emission (CE)

The data on CO₂ emission per capita were leveraged to illustrate regional carbon emissions. The data was taken from China’s Carbon Emission Accounts Database (CEADs) for two reasons. The first reason is the rapid changes in China, including the coal combustion efficiency differences, energy structure, and other uncertainties. In accordance with the International Panel on Climate Change (IPCC) accounting method, CEADs recalculate the emissions by measuring carbon emission factors, which can reflect China’s actual carbon emissions in a more systematic and reasonable way. The second reason is that the journal Scientific Data published the carbon emission measurement results. The basic data of the research was adopted by the Third Initial Information Bulletin on Climate Change of the People’s Republic of China, cited many times by world-renowned environmental economists John Hansen and Frank Green.

3.1.2. Independent Variables

• Industrial structure upgrading (IS)

Following Chen et al., (2015) [47] and Qiu et al., (2023) [48], the research considers the proportion of tertiary industry to secondary industry as the proxy index to reflect the evolution of the industrial infrastructure.

• Corporate Social Responsibility (CSR)

Hexun’s CSR assessment covers most listed companies in China, and is recognized by many scholars for its comprehensive evaluation and timely data updates. Based on the CSR of listed companies (Hexun.com, accessed on 1 May 2022), this research divided listed companies into regions and calculated the average social responsibility index of the studied provinces. It was found that the average level of CSR fulfillment of regional enterprises increased as the index increased.

3.1.3. Control Variables

Carbon emissions are influenced by numerous elements. This research chose the control variables below in light of Salari et al., (2021) [49] and Wang et al., (2022) [50].

• Gross Domestic Product (GDP)

The GDP per capita represents the economic development in a specific region;

• Energy structure (ES)

The regional energy structure is measured by coal’s share of the overall energy consumption;

• Urbanization level (UL)

The regional urban population percentage was employed to represent the urbanization level;

• Trade openness (TRA)

The regional trade openness is illustrated by the proportion of total import and export trade;

• Technology innovation (TEC)

The regional technology innovation was represented by the number of patents per 10,000 people in a region;

• Foreign Direct Investment (FDI)

The overall foreign capital utilized by these provinces represents the foreign direct investment.

3.2. Data

Considering data availability, the research explored how CSR and industrial structure affected carbon footprints with the use of the parallel data of 30 Chinese provinces from 2010 to 2019, excluding Taiwan, Xizang, Macao, and Hong Kong. The China Energy Statistical Yearbook provided the energy structure data and CEADs provided the carbon emission data. The data on CSR were from Hexun. The National Bureau of Statistics and the China Statistical Yearbook was the source of the remaining data. A linear interpolation method was used to estimate the single missing data. The logarithm of non-ratio variables were taken to reduce heteroscedasticity.

Table 1. Descriptive statistics of variables.

Variables	Obs	Mean	var	Min	Max
CE	300	9.7518	8.6673	2.7719	48.6144
IS	300	1.1444	0.6558	0.4991	5.1543
CSR	300	21.9433	4.5123	10.2600	40.6000
GDP	300	1.5527	0.4590	0.2715	2.7986
ES	300	40.0297	14.9049	1.2138	68.6766
UL	300	57.7297	12.6071	33.8100	89.6000
TRA	300	28.4048	32.5810	1.2270	160.1185
TEC	300	1.7172	1.0719	−0.7573	4.1133
FDI	300	6.5053	1.3807	3.1566	9.8798

shows the descriptive data of variables.

3.3. Model

3.3.1. Spatial Autocorrelation

The global Moran’s I index was used to examine the spatial autocorrelation; it lays a foundation for constructing a spatial metrology model, as shown in Equation (1).

$$I=\frac{{\displaystyle \sum _{i=1}^n{\displaystyle \sum _{j=1}^n{w}_{ij}\left(Y_i-\overline{Y}\right)\left(Y_j-\overline{Y}\right)}}}{S^2{\displaystyle \sum _{i=1}^n{\displaystyle \sum _{j=1}^n{w}_{ij}}}}j$$

(1)

where $I$ is Moran’s I index, and $w_{ij}$ refers to the weight matrix. $Y_i$ and $Y_j$ are the values of indicators in region $i$ and region $j$, respectively, and $\overline{Y}$ and $S$ are the mean and variance of $Y$, respectively. $I$ ranges within [−1, 1]. If the calculation result is closer to 0, the spatial correlation is less obvious; if it is closer to −1, the negative correlation is more obvious; if closer to 1, the positive correlation is more obvious.

3.3.2. Spatial Durbin Model (SDM)

In accordance with the spatial correlation of dependent and independent variables, the SDM is representative and comprehensive. Therefore, the following SDM was constructed:

$$C{E}_{it}=\delta {\displaystyle \sum _{j=1}^n{W}_m}C{E}_{jt}+{\alpha }_{1}I{S}_{it}+{\alpha }_{2}CS{R}_{it}+{\alpha }_{3}CONT{R}_{it}+{\beta }_1{\displaystyle \sum _{j=1}^n{W}_m}I{S}_{jt}+{\beta }_2{\displaystyle \sum _{j=1}^n{W}_m}CS{R}_{jt}+{\beta }_3{\displaystyle \sum _{j=1}^n{W}_m}CONT{R}_{jt}+{\mu }_i+{\upsilon }_t+{\epsilon }_{it}$$

(2)

where $CE$ refers to the carbon emission intensity, $IS$ is the industrial structure upgrading index, $CSR$ is the CSR score, $CONTR$ is the control variable, $\delta$, ${\alpha }_1$, ${\alpha }_2$, ${\alpha }_3$, ${\beta }_1$, ${\beta }_2$, and ${\beta }_3$ are the parameters of the model, $W_m$ is the space weight matrix, ${\epsilon }_{it}$ is the random error, and ${\mu }_i,{\upsilon }_t$ show the effect of the individual. This paper uses the spatial weight matrix of geographical distance ($W_1$) and economic distance ($W_2$) for empirical analysis, and uses the adjacency matrix ($W_3$) for the robustness test, according to Equations (3) and (4):

$$W_1=\left\{\begin{array}{c}1/d_{ij},i\ne j\\ 0,i\equiv j\end{array}\right.$$

(3)

$$W_2=1/\left({\overline{y}}_i-{\overline{y}}_j\right),i\ne j$$

(4)

where, $d_{ij}$ is the Euclidean distance between the provincial capital cities $i$ and $j$, respectively, and ${\overline{y}}_i$ and ${\overline{y}}_j$ are the GDP per capita of provinces $i$ and $j$, respectively.

4. Empirical Analysis

4.1. Spatial Autocorrelation Test

To record the spatial autocorrelation, the global Moran’s I indices of carbon emissions (CE) of Chinese provinces from 2010 to 2019 were calculated. It was found that all studied regions exceeded 0 during the study period, which had a statistical significance of 1%. An obvious spatial autocorrelation was thus indicated between domestic carbon footprints. Therefore, in the empirical research, the spatial econometric model was applied. The spatial autocorrelation test is illustrated in

Table 2. Moran’s I of carbon emissions.

Year	Moran’s I
2010	0.074 ***
2011	0.068 ***
2012	0.072 ***
2013	0.065 ***
2014	0.061 ***
2015	0.058 ***
2016	0.059 ***
2017	0.052 ***
2018	0.054 ***
2019	0.054 ***

Note: *** represent the statistical significance at 1% levels.

.

4.2. Empirical Analysis

4.2.1. Baseline Regression Analysis

Before the regression analysis, a series of tests determined the specific spatial econometric model. The reliability of the SDM was affirmed by the outcomes of the LM test, and a constant effect model was selected using the Hausman test. Regarding the outcomes of the Wald and LR tests, the SDM failed to decrease to a spatial error model (SEM) and a spatial lag model (SLR). Thus, to enhance the robustness of empirical results, this research used a time- and individual-fixed dual-fixed model and adopted robust standard error to correct heteroscedasticity. The data of the regression analysis are in

Table 3. Regression analysis.

	(1)	(2)
	W1	W2
IS	−2.6928 **	−1.2103 *
	(1.1813)	(1.1914)
CSR	−0.0204 *	−0.0528 *
	(0.0533)	(0.0519)
GDP	−8.9262 ***	−7.6502 ***
	(2.1467)	(1.8944)
ES	0.1310 ***	0.1420 ***
	(0.0360)	(0.0364)
UL	0.5437 ***	0.5215 ***
	(0.1614)	(0.1791)
TRA	−0.0477 **	−0.0206
	(0.0239)	(0.0256)
TEC	−0.5426 *	−0.3908
	(0.7121)	(0.7406)
FDI	−1.4113 *	−0.4752
	(0.7543)	(0.6948)
Observations	300	300
R-squared	0.4335	0.4234

Note: *, ** and *** represent the statistical significance at 10%, 5% and 1% levels, respectively.

.

In Table 3, columns (1) and (2) represent how industrial structure and CSR influence carbon footprints, respectively, when adopting the economic and geographical distance spatial weight matrix. It was found that the upgrading of industrial structure was markedly positive on regional carbon footprints. Limiting the development of secondary industry reduced high-polluting industries with excessive energy requirements while promoting the growth of tertiary industry. Hence, regional carbon emissions were reduced through industrial structure adjustment. The CSR score negatively correlated with the regional carbon emissions since CSR fulfillment is environmentally friendly according to standards like carbon emission reduction, environmental protection, and energy savings.

Regarding the control variables, advanced economic growth requires more local investment in environmental protection, so carbon emissions are relatively low. According to Our World in Data, an increase in the energy structure index markedly heightens carbon emissions, coinciding with the evidence in China, where coal is a major energy source and accounted for 69% of overall carbon emissions in China in 2020. Under urbanization, the economic scale effect, agglomeration effect, and urban residents’ energy consumption characteristics are highly influential on carbon footprints, which coincides with the conclusions of Zhang et al., (2015) [51] and Wang et al., (2021) [52]. Similar to Rehman et al., (2022) [53], regional carbon emission negatively correlates with trade openness. Carbon emissions can be inhibited through technology innovation because technology promotes the use and economic value of carbon energy, reduces the development and use cost of emerging energy, and provides support for optimizing the regional energy framework. FDI inhibits carbon emissions through the technology effect, industrial structure optimization, and income effect, which is consistent with Paramatia et al., (2021) [54].

4.2.2. Spatial Effect Analysis

According to the regression analysis of the SDM, the spatial effect outcomes of each variable are listed in

Table 4. Spatial effect analysis.

	(1)	(2)
	W1	W2
W *IS	−1.6849 **	−1.2103
	(0.3488)	(1.1914)
W *CSR	0.8362 **	0.0528
	(0.4027)	(0.0519)
W *GDP	−3.9080 ***	−7.6502 ***
	(1.9428)	(1.8944)
W *ES	0.3016	0.1420 ***
	(0.2584)	(0.0364)
W *UL	3.5571 ***	0.5215 ***
	(1.1030)	(0.1791)
W *TRA	−0.3777 **	−0.0206
	(0.1587)	(0.0256)
W *TEC	−11.5298 **	−0.3908
	(5.6749)	(0.7406)
W *FDI	−1.6748 ***	−0.4752
	(0.1215)	(0.6948)
Observations	300	300
R-squared	0.4335	0.4234

Note: ** and *** represent the statistical significance at 5%, and 1% levels, respectively.

.

According to Table 4, the spatial effect of CSR was positive, whereas that of industrial structure upgrading was negative. Control variables, including FDI, trade openness, technological innovation, and economic situations, had negative spatial effects, whereas energy structure and urbanization levels had positive spatial effects. For a deeper understanding of the spatial effects of variables, further partial differential decomposition of the SDM is required to investigate the indirect and direct effects of variables.

Table 5. Results of spatial effect decomposition.

	W1			W2
	(1)	(2)	(3)	(4)	(5)	(6)
	Direct	Indirect	Total	Direct	Indirect	Total
IS	−0.2442 **	−1.1081 *	−1.3523 **	−1.2799	0.4291	−0.8508
CSR	−0.0362	0.6187 **	0.5825 *	−0.0606	0.2382 **	0.1776
GDP	−0.8025 ***	−2.5220 ***	−3.3245 ***	−7.3347 ***	−4.3227	−11.6574 **
ES	0.1267 ***	0.1809	0.3075	0.1333 ***	0.2514 **	0.3847 ***
UL	0.4708 ***	2.3081 **	2.7789 ***	0.5288 ***	−0.3748	0.1540
TRA	−0.0388 *	−0.2452 **	−0.2840 **	−0.0210	0.0604	0.0394
TEC	−0.7942	8.3379 **	7.5437 *	−0.4528	1.6601	1.2073
FDI	−0.1125	−1.1111 **	−1.2236 **	−0.5454	0.9969	0.4515

Note: *, ** and *** represent the statistical significance at 10%, 5%, and 1% levels, respectively.

shows the spatial effect decomposition.

According to Table 5, CSR and industrial structure adjustment had negative direct effects, indicating their environmental contribution. There was a negative spatial spillover effect on industrial structure upgrading, mainly because it was in line with low-carbon development, which is the current general trend in China. Because of the strong economic externality, high cost, and poor return of CSR, it showed a positive spatial spillover effect. Therefore, competitive enterprises in neighboring areas were more inclined to “free ride”, thus reducing their social responsibility, leading to an increase in carbon emissions in a particular region.

The direct effects of control variables on carbon emissions conformed to the baseline regression results, but their spatial spillover effects were markedly different. Economic growth, trade openness, and FDI had negative spatial spillover effects, indicating that those factors might damage the environment in neighboring regions. The spatial spillover effects of urbanization, energy structure, and technological innovation were positive, which suggests that they also exacerbated carbon emissions in neighboring regions.

4.2.3. Moderate Effect Analysis

This section considers the moderating role of CSR in optimizing carbon footprints via industrial structure adjustment. To this end, the interactive terms of CSR and industrial structure were introduced in the benchmark model. This was done to avoid multicollinearity between the interaction and the original variable and to increase the economic significance of the interaction variable. This approach follows Smith et al., (1979) [55] and Balli et al., (2013) [56] for centralized processing of interaction term variables before regression. The model is as follows:

$$\begin{array}{ll}C{E}_{it}& =\delta {\displaystyle \sum _{j=1}^n{W}_m}C{E}_{jt}+{\alpha }_{1}I{S}_{it}+{\alpha }_{2}CS{R}_{it}+{\alpha }_{3}I{S}_{it}\ast CS{R}_{it}+{\alpha }_{4}CONT{R}_{it}+{\beta }_1{\displaystyle \sum _{j=1}^n{W}_m}I{S}_{jt}\\ & +{\beta }_2{\displaystyle \sum _{j=1}^n{W}_m}CS{R}_{jt}+{\beta }_3{\displaystyle \sum _{j=1}^n{W}_m}I{S}_{jt}\ast CS{R}_{jt}+{\beta }_4{\displaystyle \sum _{j=1}^n{W}_m}CONT{R}_{jt}+{\mu }_i+{\upsilon }_t+{\epsilon }_{it}\end{array}$$

(5)

Table 6. Moderate effect analysis.

	W1		W2
	(1)	(2)	(3)	(4)
	Direct	Indirect	Direct	Indirect
IS	−3.0076 **	−9.9710	−2.7439 *	−1.9263
	(1.6119)	(7.3062)	(1.6758)	(2.7747)
CSR	−0.0770 *	0.5846 *	−0.1508 **	0.0267
	(0.0752)	(0.3748)	(0.0768)	(0.1612)
IS *CSR	0.0431 **	0.0221 *	0.0771 **	0.1481 *
	(0.0604)	(0.2780)	(0.0640)	(0.1093)
controls	YES	YES	YES	YES
Observations	300	300	300	300
R-squared	0.3324	0.3324	0.2114	0.2114

Note: *, ** represent the statistical significance at 10%, 5% levels, respectively.

shows the regression results, including interaction items.

Based on Table 6, the model estimations with interactive terms had negative direct effects and positive spillovers of CSR and negative direct effects and spillovers of industrial structure adjustments. It was found that the estimated interaction coefficient was positive with statistical significance, indicating the positive moderating effect of CSR on optimizing carbon footprints through upgrading the industrial structure. The result implies that CSR was positively correlated with the carbon inhibition of industrial structure adjustment. Currently, the key point is to enhance the social and environmental effects and transform the mechanism of economic growth. Enterprises are both economic and social organizations, whose survival and development depend on a certain social environment. Enterprises are responsible for stabilizing society, keeping an ecological balance, and achieving sustainable growth. By conforming to the industrial structure upgrading, enterprises actively assuming social responsibility can greatly cut carbon emissions.

4.3. Robustness Test

The previous analysis adopted a static SDM model, ignoring the lag term of explained variables. From a practical viewpoint, however, the current carbon emission depends on its amount in the previous stage, accepting “time inertia”. Based on that analysis, this research followed Wu et al., (2022) [57] and Liu et al., (2022) [58] to develop a dynamic SDM for the robustness test and added the first-order lag term of the dependent variable as the independent variable. This study also replaced the weight matrix for the robustness test with the classic spatial adjacency matrix. According to the empirical outcomes in

Table 7. Robustness test.

	Add the Lag Term		Replace the Space Matrix
	W1	W2	W3
IS	−1.8314 ***	−1.3709 *	−1.2963 *
	(0.6934)	(0.7045)	(1.0143)
CSR	−0.1053 ***	−0.1074 ***	−0.0573 **
	(0.0325)	(0.0316)	(0.0504)
GDP	−2.5124 *	−1.3501	−5.1092 **
	(1.2868)	(1.1498)	(2.1466)
ES	0.0074	0.0390	0.1440 ***
	(0.0240)	(0.0238)	(0.0347)
UL	0.0714	0.1315	0.4890 ***
	(0.1015)	(0.1152)	(0.1567)
TRA	−0.0020	0.0091	−0.0440 *
	(0.0146)	(0.0154)	(0.0229)
TEC	−1.1912 ***	−0.9568 **	−0.7959
	(0.4479)	(0.4532)	(0.6760)
FDI	−0.1838	−0.2425	−1.1056
	(0.4603)	(0.4091)	(0.6834)
L.CE	0.8334 ***	0.8566 ***
	(0.0350)	(0.0352)
Observations	270	270	300
R-squared	0.3321	0.3822	0.2208

Note: *, ** and *** represent the statistical significance at 10%, 5%, and 1% levels, respectively. One stage lag of the dependent variable resulted in the loss of 30 samples.

, after adding the lag term and replacing the space matrix, the coefficient significance and symbols of key variables conformed to the previous empirical analysis results, which confirmed the robustness of the above-mentioned empirical study.

4.4. Heterogeneity Analysis

Carbon emissions in different regions have idiosyncratic responses to the industrial structure and CSR because of regions’ unique resources, geography, and economy. Thus, in accordance with various geographical locations, this research divided the selected provinces into three areas, namely, Eastern, Central, and Western, to explore the heterogeneity.

Table 8. Heterogeneity analysis.

	East		Central		West
	W1	W2	W1	W2	W1	W2
IS	−0.7032 **	−0.4556	−0.5840 *	−3.6065	−7.4238 ***	−6.8543 ***
	(0.5700)	(0.3970)	(3.8721)	(2.6138)	(1.5072)	(1.4382)
CSR	−0.0469 **	−0.0356	−0.1957	−0.4265 **	0.1132 *	0.0637
	(0.0231)	(0.0225)	(0.1917)	(0.1911)	(0.0662)	(0.0515)
controls	YES	YES	YES	YES	YES	YES
Observations	110	110	80	80	110	110
R-squared	0.4015	0.4001	0.3139	0.3091	0.2701	0.2913

Note: *, ** and *** represent the statistical significance at 10%, 5%, and 1% levels, respectively.

gives the experimental results.

According to Table 8, carbon emissions in the three regions were obviously suppressed via industrial structure adjustments, which is similar to the results of the full sample regression. CSR in Eastern China had an influence on regional carbon footprints that was similar to that of the full sample analysis. As a comprehensive transportation hub and a vital base for high-tech industry, raw materials, energy, manufacturing, and grain production, the influence of CSR in Central China is higher than that in the other two regions. As the study of Zhang et al., (2022) [59] shows, spatial compactness and spatial stability in Western China are low. CSR in Western China on regional carbon emissions is insignificant. Due to the small total GDP in that area, the business scale is limited and more dispersed than that in the other two areas.

5. Conclusions and Recommendations

Identifying the spatial attributes of economic, social, and environmental development is a prerequisite for China to raise the quality of its development. Industry is the foundation of development, and enterprise is the carrier of development; both are closely related to the economy, society, and environment [60,61]. Industrial structure upgrading is China’s main path to transforming its economic growth mode and development model in the new development stage [62,63]. Enterprise is the carrier of regional economic behavior; the development of enterprise not only affects the scale and structure of regional economies, but is also responsible for regional social stability and environmental optimization [64,65]. Under China’s strategic goals of reaching a carbon peak and carbon neutrality, the green and low-carbon development of industries and enterprises is very important [66,67]. This is different from the existing research on the low-carbon development of industry or enterprise alone. Based on the unique perspective of industry and enterprise, this paper discusses the effect of industrial structure upgrading and CSR on carbon emissions, and the regulatory effect of CSR. Additionally, existing research has shown that geospatial factors must be considered in economic management research [68,69,70]. Industrial development, enterprise behavior, and pollution emissions all have strong spatial attributes, but few studies have paid attention to that. From a spatial perspective, this study explored the spatial spillover effects of industrial structure and CSR on regional carbon emissions based on the SDM.

There are three main conclusions in this paper: (1) Regional carbon emissions can be curbed by improving of CSR and industrial structures, and the former can positively regulate carbon emissions thanks to the latter. Therefore, higher CSR promotes industrial structure adjustment, thus decreasing regional carbon emissions. (2) Industrial structure adjustment shows a negative spatial spillover effect, which can lead to a carbon emission decrease in neighboring areas. However, neighboring areas might have “free rider” behavior due to the high cost, lack of short-term benefits, and obvious economic externalities of CSR fulfillment. Companies in the surrounding areas reduce their social responsibilities, increasing regional carbon emissions. (3) Carbon emissions show various responses to CSR depending on the studied region. For example, CSR fulfillment in Western China has little influence on regional carbon emissions due to the relatively low GDP, low number of enterprises, and scattered locations in that region. However, in the central and eastern areas, CSR can effectively achieve carbon emission reduction due to their superior geographical location and good business environment.

After an in-depth analysis, two proposals are made. (1) Policymakers should improve relevant regulations and standards on CSR to build a systematic and objective evaluation system for it. Decision-makers should also guide social capital to participate in CSR investment, improving the overall level of social responsibility. (2) The government should consider how CSR and industrial structures affect regional heterogeneity of carbon emission reduction to fully enhance the synergistic effects of dual incentives. In this regard, improvement of spatial governance promotes the comparative advantages of different regions and forms complementary advantages.

This study had some limitations. First, industrial structure upgrading is a complex transformation process, including advances, rationalization, and ecological aspects, among others. Industrial structure adjustment has a more profound definition and evaluation method, which cannot be measured simply by the proportion of the output value of tertiary industry to that of secondary industry, as in this paper. Future research should extend the measurement index of industrial upgrading. Additionally, CSR is the most important social responsibility of enterprises. Future studies might try to measure how regional carbon emissions affect corporate environmental responsibility.