Spatio-Temporal Characteristics of Industrial Carbon Emission Efficiency and Their Impacts from Digital Economy at Chinese Prefecture-Level Cities

Abstract

In the pursuit of China’s dual carbon goals, identifying spatio-temporal changes in industrial carbon emission efficiency and their influencing factors in cities at different stages of development is the key to effective formulation of countermeasures to promote the low-carbon transformation of Chinese national industry and achieve high-quality economic development. In this study, we used balanced panel data of 270 Chinese cities from 2005 to 2020 as a research object: (1) to show spatio-temporal evolution patterns in urban industrial carbon emission efficiency; (2) to analyze the aggregation characteristics of industrial carbon emission efficiency in Chinese cities using Global Moran’s I statistics; and (3) to use the hierarchical regression model for panel data to assess the non-linear impact of the digital economy on the industrial carbon emission efficiency of cities. The results show the following: (1) the industrial carbon emission efficiency of Chinese cities exhibited an upward trend from 2005 to 2020, with a spatial distribution pattern of high in the south and low in the north; (2) China’s urban industrial carbon emission efficiency is characterized by significant spatial autocorrelation, with increasing and stabilizing correlation, and a relatively fixed pattern of spatial agglomeration; (3) there is a significant inverted-U-shaped relationship between the digital economy and the industrial carbon emission efficiency of cities. The digital economy increases carbon emissions and inhibits industrial carbon emission efficiency in the early stages of development but inhibits carbon emissions and promotes industrial carbon emission efficiency in mature developmental stages. Therefore, cities at all levels should reduce pollution and carbon emissions from high-energy-consuming and high-polluting enterprises, gradually reduce carbon-intensive industries, and accelerate the digital transformation and upgrading of enterprises. Western, central, and eastern regions especially should seek to promote the sharing of innovation resources, strengthen exchanges and interactions relating to scientific and technological innovation, and jointly explore coordinated development routes for the digital economy.

1. Introduction

With the intensification of global climate change and environmental pollution problems, a consensus has emerged in the international community regarding the need for a reduction in carbon emissions and the development of a low-carbon economy worldwide. Carbon peaking and carbon neutrality are two important measures in the global response to climate change; together, these also constitute an important development goal put forward by China [1]. Industry, as an important pillar of economic development, is also one of the main sources of carbon emissions [2]. The Second China Digital Carbon Neutrality Summit Forum pointed out that industrial green and low-carbon transformation is essential for integrating development and achieving emission reductions and that energy transformation and digital twinning will act as a powerful stimulant to China’s “dual carbon” goal of carbon peaking and carbon neutrality. As an important engine for China’s green and low-carbon industrial transformation, the digital economy should be fully exploited for its potential to reduce carbon emissions. The stimulating power of the digital economy should also be used to achieve high-quality economic development and accelerate the achievement of digital and low-carbon transformation in the industrial field [3,4]. As a new economic form, the digital economy involves digital industrialization, industrial digitization, and other evolutionary modes and is characterized by constantly improving levels of economic and social digitization, networking, and intelligence; as such, the digital economy may now be understood as a central means by which China will achieve its national goals [5,6]. In 2022, the State Council issued the “14th Five-Year Plan for the Development of the Digital Economy”, which anticipated that the added value generated by the core industries of the digital economy would account for 10% of GDP by 2025. Against this backdrop, the digital economy is also driving profound changes in the modes of production and governance, as well as the personal lifestyles of individuals; moreover, it has become a key force in reorganizing ecological resources and rebuilding economic structure.

Industry is the main source of carbon emissions in China, with industrial carbon emissions accounting for more than 70% of China’s total, and it is also the pillar industry of the national economy, accounting for 32% of the total [7]. Therefore, improving industrial carbon efficiency is a key path to achieving high-quality development of China’s economy and realizing the green and low-carbon transformation of industry. Domestic and international studies have been conducted on the spatial and temporal evolution characteristics [8], spatial effects [9,10], and influencing factors [11] of industrial carbon emission efficiency. Scholars have already provided different explanations of the factors affecting industrial carbon efficiency from the perspectives of urban development level [12], economic growth [13], industrial structure [14], energy structure [15], and technological innovation [16]. Wang et al. [17] applied a non-radial model to estimate the carbon emission efficiency of Chinese provinces and explained the reasons for the low carbon efficiency in terms of industrial structure, technological level, and management capacity. Ding et al. [18] studied the dynamics of carbon efficiency in 30 provinces in China from the perspective of regional carbon efficiency evolution. However, these provincial-scale carbon emission efficiency studies have not explored the spatial heterogeneity and evolution of industrial carbon emission efficiency at the city scale and lack the impact of a new economic form, the digital economy, on it, which is urgently needed to expand the study of the digital economy on key industries (industrial sectors) and at the city scale.

China’s industrial structural system is undergoing major changes, with the central and western regions being the main energy resource bases, the eastern regions undergoing capacity upgrading and transformation, and a large amount of high-energy-consuming and high-pollution shifting to the central and western regions, which has led to drastic changes in the carbon efficiency of China’s industry. A previous study found that there is a significant correlation between industrial carbon emissions and the spatial distribution in China [19]. The overall gap in the spatial distribution of the industrial carbon emissions showed an expanding trend with its evolution to the east [20]. Current studies have focused on describing the distribution of carbon emissions and carbon emission intensity in China’s provinces and cities, and few studies have explored the development of industrial carbon efficiency in China’s prefecture-level cities [21,22]. As a result, the spatial distribution pattern of industrial carbon efficiency is ambiguous, and the law of evolution characteristics is unclear, which cannot provide detailed basic information support for the green and low-carbon development of the industrial sector.

Currently, the impact of the digital economy on carbon emissions, carbon intensity, and per capita carbon emissions has been reported by many scholars [23,24]. This is because the development of a digital economy breaks the constraints of geographic distance on economic activities, strengthens regional connectivity, promotes the synergistic development of technological innovation and the common transformation of industrial structure in various regions, and influences the cooperation and innovation of neighboring regions through spatial spillover effects [25]. Previous studies on carbon emission efficiency have mostly utilized econometric models and spatial econometric models to emphasize the spatial effect of the digital economy on carbon emission efficiency [9,10,26], but traditional econometric models simplify the linear relationship between the variables and the dependent variable. Another study found a potential non-linear relationship between the digital economy and industrial carbon emission efficiency. Based on the global panel data of 190 countries from 2005 to 2016. Li et al. [27] found an inverted U-shaped, non-linear relationship between CO2 emissions and the digital economy, which supports the environmental Kuznets curve (EKC) hypothesis. Cheng et al. [28] confirmed that there is a significant inverted U-shaped relationship between digital economy and carbon intensity by adding a quadratic term and conducting a U-test. This does not allow for the identification of the specific impacts of variables on changes in carbon efficiency at different quartile stages and does not provide targeted guidance for specific stages of urban development.

In short, the question of how the development of the digital economy affects industrial carbon emission efficiency has yet to be answered. Specifically, is there a non-linear relationship between the two phenomena? Answering the above question is of potentially great significance for advancing the digital power of China and for the Chinese achievement of carbon peaking and carbon neutrality. In this study, then, to make up for the shortcomings of previous research, we used balanced panel data of 270 Chinese cities from 2005 to 2020 as a research object: (1) to show the spatio-temporal evolution pattern of urban industrial carbon emission efficiency; (2) to analyze the aggregation characteristics of industrial carbon emission efficiency in Chinese cities using Global Moran’s I statistics; and (3) to use the hierarchical regression model for panel data to explore the non-linear impact of the digital economy on the industrial carbon emission efficiency of cities.

2. Materials and Methods

2.1. BAM-DEA Model

Carbon efficiency is an evaluation based on input and output elements, and its influence is related to the selected inputs and outputs. The main algorithms are data envelopment analysis (DEA) and stochastic frontier analysis (SFA). Compared with the latter, DEA does not need to set specific function forms and is more suitable for multi-input and multi-output decision-making units (DMU) efficiency measurement [26]. Therefore, DEA models are more widely used. Compared with the traditional DEA and derivative models, the Bounded Adjusted Measure-Data Envelopment Analysis (BAM-DEA) model has the advantages of non-radiality, non-angle, additivity, higher discrimination, etc. The BAM-DEA model is utilized to calculate industrial carbon emission efficiency. The design of input-output indicators takes into account the mutual influence effect between indicators, and the results are more scientific and accurate [29].

$$\min 1-\frac{1}{m+q}\left({\sum }_{i=1}^m\frac{s_i^{-}}{L_{ik}^{-}}+{\sum }_{r=1}^q\frac{s_r^{+}}{U_{rk}^{+}}\right)$$

(1)

$$\begin{array}{ccc}s.t.& & \hspace{1em}\hspace{1em}X\lambda +s^{-}=x_k\\ & & \hspace{1em}\hspace{1em}Y\lambda -s^{+}=y_k\\ & & e\lambda =1\\ & & \hspace{1em}\hspace{1em}\lambda ,s^{-},s^{+}\ge 0\\ & & \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}{L}_{ik}^{-}=x_{ik}-\min \left(x_i\right)\\ & & \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}{L}_{ik}^{-}=x_{ik}-\min \left(x_i\right)\\ & & \hspace{1em}\hspace{1em}\hspace{1em}\hspace{1em}{U}_{rk}^{+}=\max \left(y_r\right)-y_{rk}\end{array}$$

where $L_{ik}^{-}$ is the difference between the value of the ith input indicator of decision unit K and the minimum value of that indicator, and $U_{rk}^{+}$ is the difference between the maximum value of the rth output indicator and the value of the rth indicator of decision unit K.

2.2. The Moran’s I

2.2.1. The Global Moran’s I

The global Moran’s I statistic was used to analyze the overall spatial distribution of geographic elements, to determine whether all geographic elements were clustered or not, and to examine whether there was an obvious distribution pattern of geographic elements in space [30]. The global Moran’s I is calculated by the following formula:

$$\mathrm{I}=\frac{n\sum _{i=1}^n \sum _{i=1}^n H_{ij}\left(x_i-\overline{x}\right)\left(x_j-\overline{x}\right)}{\sum _{i=1}^n \sum _{i=1}^n H_{ij}\sum _{i=1}^n {\left(x_i-\overline{x}\right)}^2}$$

(2)

where I is the global Moran’s index; $x_i$ and $x_j$ denote the industrial carbon emission efficiency of elements i and j, respectively; $H_{ij}$ is the spatial weight matrix of elements i and j; and $\overline{x}$ denotes the average value of the industrial carbon emission efficiency of all elements.

2.2.2. Local Moran’s I

Local Moran’s I is used to analyze the local characteristic differences existing in the distribution of spatial objects, reflecting the spatial heterogeneity and instability within the local area, which can be mainly analyzed by LISA aggregation map or Moran’s scatter plot [31]. The formula is as follows:

$$I_i=\frac{\left(x_i-\overline{x}\right)}{{\displaystyle {\sum }_{i=1}^n} {\left(x_i-\overline{x}\right)}^2}{\sum }_{j=1}^n H_{ij}\left(x_i-\overline{x}\right)$$

(3)

where $I_i$ denotes the Local Moran’s I index of a city, and its relationship with the global Moran’s I index is $I=\frac{1}{n}{\sum }_{i=1}^n I_i$.

2.3. The Panel Quantile Model

In this study, a panel quantile regression model was used to quantify the extent to which the main influencing factors affected the industrial carbon emission efficiency of cities at different stages of development. When the data distribution is non-normal, or the regression coefficients fluctuate greatly at different quartiles, quantile regression can capture the degree of influence of the explanatory variables at different quartile levels; in addition, regression results are less susceptible to the influence of outliers and more comprehensively encompass individual heterogeneity [32]. The expression is as follows:

$${\mathrm{Quant}}_{Y_i}\left({\theta }_k\mid X_{it}\right)=\beta \left({\theta }_k\right)X_{it}+{\alpha }_i \left(i=1,2,\cdots N;t=1,2,\cdots T\right)$$

(4)

where $Y_i$ is the city industrial carbon emission efficiency; ${\mathrm{Quant}}_{Y_i}\left({\theta }_k\mid X_{ii}\right)$ represents the city industrial carbon emission efficiency at the θth quantile; $X_{it}$ is the vector of exogenous variables for city i in year t; ${\theta }_k$ is denoted as the kth quantile level; $\beta \left({\theta }_k\right)$ represents the vector of estimated coefficients; i is the city studied; t is the year of the study; and ${\alpha }_i$ is the unobserved individual effect.

2.4. Variable Description

1. Explained variable: industrial carbon emission efficiency. Industrial carbon emission efficiency was measured by applying the BAM-DEA model by including industrial input and output factors in the analytical framework. Among these, the input factors were industrial capital stock, industrial labor force, industrial land use, industrial energy input, and industrial water input. Constant price industrial fixed asset investment was obtained by deflating the difference between total fixed asset investment and real estate investment according to the fixed asset investment price index, and then the perpetual inventory method was used to estimate the industrial capital stock for 270 prefecture-level cities.The industrial labor force is measured by the difference between the number of employees in the secondary industry and the number of employees in the construction industry. Industrial land use is selected to characterize the area of industrial land. Industrial energy input is proxied by the industrial electricity consumption of each prefecture-level city. Industrial water input is obtained by using the total urban water supply minus residential water consumption estimates. For the output factors, the deflated industrial gross domestic product of each prefecture-level city was taken as the desired output, and the city’s carbon emission was taken as the non-desired output. Carbon emissions were obtained by summing the products of the cities’ industrial natural gas consumption, industrial liquefied petroleum gas consumption, and industrial electricity consumption, with their respective carbon emission factors (

Table 1. Indicator system for the industrial carbon emission efficiency.

Variable	Dimension	Measurement Method
Input Variables	Industrial capital stock	Total investment in fixed assets
		Real estate investment
	Industrial labor force	Employees in secondary industry
		Employees in the construction industry
	Industrial Land Use	Industrial Land Area
	Industrial Energy Inputs	Industrial Electricity Consumption
	Industrial water inputs	Total urban water supply
		Residential Water Consumption
Desired outputs	Gross industrial product by prefecture-level
Undesired outputs	carbon emission

).

2. Explanatory variable: digital economy. In this study, we measured the level of the urban digital economy using the evaluation system and measurement method of Zhao et al. [33] and constructed a digital economy development index with four dimensions, namely, digital infrastructure, digital industry, digital innovation, and digital application. Digital infrastructure was expressed by the number of cell phone subscribers and internet users per 100 people; digital industrial base was expressed by per capita telecommunication business revenue; digital innovation capacity was expressed by the ratio of computer and software employees to the number of employees and the number of patents related to the digital economy per 10,000 people; digital application was expressed by the Digital Inclusive Finance Index. In order to avoid bias caused by subjective factors, the entropy method was used to determine the index weights.

3. Control variables: relevant variables were introduced to control the accuracy of results relating to the impact of the digital economy on industrial carbon emission efficiency. Per capita GDP indicates the level of regional economic development, and economic growth brings carbon emissions [34]; Population size as an engine of economic growth but also causes energy consumption, which brings carbon emissions [35]; Urbanization represents the stage of urban development, and the expansion of the urban scale promotes the expansion of the urban industrial system and also promotes the transformation and upgrading of enterprises [12]. Increased size of industrial enterprises promotes economic development but also brings about environmental pollution, which negatively affects urban economic high-quality development [25]. Government intervention impacts regional economic development, industrial structure adjustment, and the level of technological innovation [36]. Science and technology input helps to improve the level and efficiency of industrial scientific and technological progress and suppresses the intensity of industrial carbon emissions [16]. Foreign direct investment promotes industrial development and also provides a channel for highly polluting and energy-consuming enterprises to transfer to regions with less stringent environmental regulations [37].

2.5. Data Sources

For this study, considering the principles of data authenticity and availability, we collected and organized panel data from 270 cities in China in 2005, 2010, 2015, and 2020 so as to explore the impact of the digital economy on industrial carbon emission efficiency. The Digital Inclusive Finance Index is from the Digital Finance Research Center of Peking University, and the rest of the data is from the China Statistical Yearbook, the China Energy Statistical Yearbook, prefecture-level Municipal Statistical Bulletin, carbon emission Accounts and Datasets, National Research Network Database and Wind database, etc. After matching the city’s industrial carbon efficiency with other economic, environmental, and innovation indicators to determine the stability and balance of the panel data, 270 cities were finally selected as the sample cities for the study. In order to eliminate unit differences among the data and unify the scale, all the data were standardized. The descriptive statistics of relevant variables are shown in

Table 2. Results of descriptive statistics for variables.

Variable Type	Variable	Symbol	Sample Size	Mean	Standard Deviation	Min	Max	Variance	Skewness	Kurtosis
Explained variable	industrial carbon emission efficiency	ICEE	1080	0.65	0.12	0.48	1	0.01	1.14	3.94
Explanatory variable	Digital economy	Dig	1080	0.09	0.08	0.01	0.83	0.01	3.85	24.94
Control variables	Per capita GDP	PGDP	1080	417,242	31,695	2300	207,163	0	1.49	5.64
	Population size	Popd	1080	11,538	25,966	0	206,065	0.87	0.82	2.59
	Urbanization	Ur	1080	0.51	0.18	0.03	1	0.03	0.22	2.82
	Industrial enterprise size	Ien	1080	418.24	234.08	1	833	0.86	1.81	7.21
	Government intervention	Gov	1080	0.17	0.10	0.04	0.85	0.01	2.07	9.85
	Science and technology input	Sti	1080	0.19	0.04	0.02	0.50	0	0.35	5.56
	Foreign direct investment	Fdi	1080	0.02	0.02	0	0.13	0	1.89	7.30

.

3. Results

3.1. Spatio-Temporal Distribution of Industrial Carbon Emission Efficiency in Chinese Cities

In order to visualize the spatio-temporal variability of industrial carbon emission efficiency of Chinese cities, the industrial carbon emission efficiency of Chinese cities was classified into three levels based on efficiency values, as follows: low efficiency (0–0.4), medium efficiency (0.41–0.7); and high efficiency (0.71–1). A zero value indicated no data (Figure 1). The industrial carbon emission efficiency of Chinese cities showed an upward trend from 2005 to 2020, with urban industrial carbon emission efficiency increasing from 0.59 in 2005 to 0.73 in 2020. It may be that since China’s accession to the WTO in 2001, the process of industrialization in China has rapidly advanced so that, in recent years, the industrial structure has been optimized and upgraded, and economic development has shifted to high-quality development, with a consequent rise in urban industrial carbon emission efficiency. The number of cities with high industrial carbon emission efficiency in the eastern region increased significantly between 2005 and 2020, mainly due to the transfer of heavy industry, industrial restructuring, a rising proportion of new industries, and the gradual clean-up of their energy consumption structure [38]. The spatial distribution pattern of urban industrial carbon emission efficiency from 2005 to 2020 showed higher values in the south and lower values in the north. Urban industrial carbon emission efficiency was mostly at a medium level in 2005 and 2010, with high-efficiency cities concentrated in Shaanxi, Ningxia, and Gansu. By 2020, the numbers of cities with medium and high carbon efficiencies were almost equal. There were 28 high-efficiency cities in 2010, and 88 such cities in 2015, and 153 in 2020. The lower industrial carbon emission efficiency of cities in northern China may be explained as follows: northern Chinese provinces are rich in coal and petrochemical energy and are characterized by high levels of traditional energy extraction and consumption; their urban development has thus been highly dependent on carbon-intensive industries, resulting in lower industrial carbon emission efficiency. In contrast, the Shandong Peninsula city cluster, the Yangtze River Delta region, and the Pearl River Delta region are all characterized by strong scientific and technological innovation and industrial integration development capabilities, high levels of industrial integration, and extensive energy-saving and consumption-reduction efforts, resulting in high levels of industrial carbon emission efficiency; as a result, these regions are at the forefront of China’s economic transformation and development today. We also found that cities with high industrial carbon emission efficiency are surrounded by circles of medium and low-efficiency cities. This may be due to the pursuit of industrial upgrading and transformation in developed regions, as well as the relocation of high-pollution and high-energy-consumption industries and the transfer of industries to economically underdeveloped regions, leading to low industrial carbon emission efficiency. In northeast China, cities with high industrial carbon emission efficiency are concentrated in provincial capitals and their neighboring cities, such as Changchun, Baicheng, Heihe, Suihua, and Mudanjiang. Industries in northeastern cities are mainly heavy industries, which are characterized by high consumption of resources and energy and low energy-utilization efficiency. This explains why the industrial carbon emission efficiency of most cities in northeastern China is still at a relatively low level.

3.2. Aggregation Characteristics of Industrial Carbon Emission Efficiency in Chinese Cities

In this paper, the Global Moran’s I statistic and LISA distribution were used to test the spatial correlation of urban industrial carbon emission efficiency and to characterize its spatial agglomeration (Figure 2). We found that the spatial autocorrelation of industrial carbon emission efficiency in Chinese cities is significant, that this correlation is increasing and stabilizing, and that the pattern of spatial agglomeration is relatively fixed. The Global Moran’s I of industrial carbon emission efficiency of 270 Chinese cities is greater than zero and passes the significance test at the 95% level. In addition, cities with similar industrial carbon emission efficiency are spatially clustered (Figure 1).

The center of gravity of high-high agglomeration cities tends to move southward with time, and the number of cities increases, while the center of gravity of low-low agglomeration cities moves northward, and the number of cities decreases. Since 2005, the center of gravity of high-high agglomeration cities has moved from Gansu to Shaanxi, and the new high-high concentration cities are Baoshan in Yunnan, Lianyungang, and Huaian, in Jiangsu, and Guangyuan, Mianyang, and Dazhou, in Sichuan. Tianshui has remained a high-high agglomeration city. However, high-high concentration has disappeared in the resource-oriented city of Heihe in northeast China. The heavy industry system, which is highly dependent on resource consumption, has long been the economic pillar of the northeastern region; consequently, it is difficult for enterprises in this region to transform and upgrade, and improvements in industrial carbon emission efficiency have been slow as a result. The industrial transfer from the eastern region to the central and western regions has been mainly dominated by high-consumption and high-emission industries, and the improvement of energy utilization and carbon reduction technology in the northwestern region has a lag relative to that of the districts and counties in the developed regions, which makes it difficult to break through the bottleneck of the city’s industrial carbon emission efficiency for a certain period of time [32]. Therefore, the center of gravity of cities with high concentration has shifted to economically developed cities and regions due to improved environmental regulations and increased efforts towards corporate pollution control, which prompt cities to accelerate their elimination of backward production capacity and high-pollution industries, thus reducing carbon emissions caused by resource and energy consumption, and improving carbon emission efficiency.

Cities with low-low agglomeration were distributed in the Shandong Peninsula Urban Agglomeration, Yangtze River Delta Region, and Beijing-Tianjin-Hebei Region in the early stages of the study period but were concentrated in Shanxi, Inner Mongolia, Liaoning, and other provinces by the end of the study period. We may note that, on the one hand, the Shandong Peninsula Urban Agglomeration has cultivated and developed strategic emerging industries with a focus on industrial structure adjustment and optimization, thus promoting reductions in energy usage and consumption. In addition, the Yangtze River Delta region has made full use of its industrial base and port conditions to achieve rapid development of advanced manufacturing industries, high levels of industrial agglomeration and research and development (R&D) investment, a strong capacity for scientific and technological innovation, and extensive industrial integration and development, all of which have increased the industrial carbon emission efficiency of cities in this region [39]. Similarly, in the Beijing-Tianjin-Hebei region, adjustments in industrial structure and environmental pollution controls have served to effectively control carbon emission levels, and urban industrial carbon emission efficiency has been improved [40]. On the other hand, Shanxi and Inner Mongolia are provinces characterized by high levels of traditional energy extraction and consumption; in areas like these, coal-based energy utilization is relatively crude, carrying the transfer of high-energy-consuming enterprises from Beijing. As a result, urban development has been highly dependent on carbon-intensive industries, and the enhancement of urban industrial carbon emission efficiency has been slow in these areas.

China’s central and southern regions have developed urban economies, relatively clean energy consumption structures, and fewer carbon-intensive manufacturing industries, so the application of innovative green technologies by enterprises has been more common, and higher industrial carbon emission efficiency has been attained. The eastern region plays a leading role in achieving China’s “carbon peak, carbon neutral” energy saving and emission reduction goals. The central and western regions must now seize the opportunity to actively promote industrial intelligence and low-carbon development, improve urban industrial carbon emission efficiency, and work to promote high-quality economic development and sustainable development in China. It should be noted that there are problems such as industrial homogeneity and inefficient resource utilization in various regions; however, these may serve to strengthen inter-regional exchanges and cooperation in terms of technology and talents, reduce the phenomenon of inter-regional transfer of polluting industries, and achieve coordinated development of the digital economy, while at the same time rationally and efficiently configuring digital economy resources according to the actual economic growth situation of each region, as well as reducing the inter-regional development gaps, and better exploiting the carbon emission reduction effect of the digital economy.

3.3. Basic Regression Analysis (BRA)

Table 3. Benchmark regression results of the impact of the digital economy on ICEE.

		(1)	(2)
Variable	Symbol	ICEE	ICEE
Digital economy	Dig	0.201 ***	0.135 **
		(4.712)	(2.545)
Urbanization	Ur		−0.165 ***
			(−4.040)
ln(Per capita GDP)	lnPGDP		0.0389 ***
			(4.542)
ln(Population size)	lnPopd		0.0106 ***
			(5.917)
ln(Industrial enterprise size)	InIen		0.0026
			(0.791)
Government intervention	Gov		0.427 ***
			(10.020)
Science and technology input	Sti		0.388 ***
			(4.813)
Foreign direct investment	Fdi		0.0125
			(0.0725)
Constant		0.635 ***	0.0881 *
		(122.1)	(1.128)
Fixed effects		Control	Control
Sample size		1080	1080
R2		0.401	0.524

Note: ***, **, and * indicate that the model results are significant at the 1%, 5%, and 10% confidence levels, and the values in parentheses are t-test values.

presents basic estimation results showing how the digital economy affects urban industrial carbon emission efficiency. Figure 3 shows the distribution of regression elasticity coefficients of factors affecting industrial carbon emission efficiency in cities. It can be seen that the influence of socio-economic factors on industrial carbon emission efficiency involves stage differences. Among these, column (1) shows results without adding any control variables, while column (2) shows estimation results with the addition of the remaining city economic indicators. As can be seen in Table 2, in comparison with column (1) without control variables, the explanatory power of the model rises in column (2) after city economic characteristics and government intervention variables are added, although the regression coefficients of the digital economy decrease slightly. After controlling for other factors, we may state that the digital economy has a positive effect on improving urban industrial carbon emission efficiency, which is significant at the 5% confidence level.

The regression coefficients of the digital economy become larger with the increase in quartiles, and the positive effect of the digital economy gradually strengthens, with the greatest negative effect in the lowest quartile. This shows that there is a significant inverted-U-shaped relationship between the digital economy and urban industrial carbon emission efficiency so that the digital economy increases carbon emissions and inhibits industrial carbon emission efficiency in the early stages of development but inhibits carbon emissions and increases industrial carbon emission efficiency in the mature stage of development. This finding is similar to the results of previous studies. The environmental Kuznets curve shows that when the economic growth level is low, people have little demand for environmental quality. Facing the pressure of economic assessment, local governments pay more attention to the economy and neglect environmental protection, adopting the development mode of “pollute first, treat later”.

When the level of economic growth increases and per capita GDP rises, people’s demand for high-pollution, high-energy consumption products decreases while demand for more environmentally friendly products rises. In order to adapt to changes in consumer demand, enterprises adapt to the development trend of the digital economy and promote industrial technological innovation, including industrial digital transformation and upgrading, thereby reducing the intensity of their industrial carbon emissions and improving industrial carbon emission efficiency [5].

The regression coefficient of population size is positive, so the regression coefficient increases and then decreases with the increase in quartiles, with a positive effect on the enhancement of urban industrial carbon emission efficiency in the middle quartile and more obvious inhibitory effects in the low and high quartiles. Urbanization has a significant negative effect on industrial carbon emission efficiency; the regression coefficient decreases with the increase in quartiles, and the inhibitory effect of urbanization is more critical in the high quartile. Urbanization drives the development of urban industrialization, resulting in the generation of jobs that attract people and resources, with a resulting increase in population density; this is conducive to the effective allocation of energy resources and the achievement of higher profits. The level of urban industrialization then rises, urban industrial carbon emission declines and industrial carbon emission efficiency is thereby improved [41].

Per capita GDP, government intervention, and science and technology input all have a positive effect on urban industrial carbon emission efficiency at the 1% significance level at different quantiles. The regression coefficient of science and technology input increases with the increase in quartiles and the positive effect is gradually strengthened. In the stage of high-quality development, new green energy technology replaces the traditional high-pollution and high-energy-consumption production mode, the industrial structure is upgraded, urban industrial carbon emission efficiency reaches a high level, and the industrial development mode is more environmentally friendly. Science and technology input can provide practical methods that can help industry achieve low-carbon and zero-carbon development at a technical level and efficiently empower industrial green and low-carbon transformation, resulting in lower carbon emissions and improved industrial carbon emission efficiency [42]. The improvement of industrial carbon emission efficiency, in turn, promotes scientific and technological innovation and upgrading in the industry.

The regression coefficient of foreign direct investment is 0.0125. This does not pass the significance test and indicates that foreign direct investment may have a certain positive effect on urban industrial carbon emission efficiency, but this linear relationship was not confirmed by the statistical model. The regression coefficient of industrial enterprise size is 0.0026. This also fails the significance test. It can be seen that the regression coefficient keeps changing with the increase in quartiles. This indicates that there is an inverted-U-shaped relationship between industrial enterprise size and urban industrial carbon emission efficiency. In the early stages of development, the rapid development of high-value-added industries requires more energy and resource consumption, leading to an increase in industrial carbon emissions and inhibiting industrial carbon emission efficiency. As industrial enterprises continue to expand, to become transformed and upgraded, a more advanced industrial structure emerges, which can aid the reduction of carbon emissions and improve industrial carbon emission efficiency. In short, industrial transformation and upgrading is the key to achieving carbon emission reduction and improving industrial carbon emission efficiency in both high- and low-scoring cities.

4. Conclusions

In this study, we used balanced panel data from 270 Chinese cities from 2005 to 2020 as a research object: (1) to show the spatio-temporal evolution pattern of urban industrial carbon emission efficiency; (2) to analyze the aggregation characteristics of industrial carbon emission efficiency in Chinese cities using the Global Moran’s I statistic; and (3) to use the hierarchical regression model for panel data to explore the non-linear impact of the digital economy on the industrial carbon emission efficiency of cities. The main conclusions are as follows:

(1) From 2005 to 2020, the industrial carbon emission efficiency of Chinese cities exhibited an upward trend, with a spatial distribution pattern of high in the south and low in the north. Urban industrial carbon emission efficiency was mostly at a medium level in 2005 and 2010. On the whole, China’s urban industrial carbon emission efficiency increased significantly from 2005 to 2020, and urban industrialization shifted from high-speed development to high-quality development.

(2) The spatial autocorrelation of industrial carbon emission efficiency in Chinese cities is significant; the correlation is increasing and stabilizing, and the pattern of spatial agglomeration is relatively fixed. The center of gravity of high-high agglomeration cities has moved from Gansu to Shaanxi, and the number of such cities has increased. At the beginning of the study period, low-low agglomeration cities were distributed in the Shandong Peninsula urban agglomeration, Yangtze River Delta region, and Beijing-Tianjin-Hebei region, but were concentrated in Shanxi, Inner Mongolia, Liaoning, and other provinces by the end of the study period. In addition, the total number of low-low agglomeration cities declined.

(3) There is a significant inverted-U-shaped relationship between the digital economy and urban industrial carbon emission efficiency. The digital economy increases carbon emissions and inhibits industrial carbon emission efficiency at the early stage of development and inhibits carbon emissions at the mature stage of development. Urbanization has a significant negative effect on industrial carbon emission efficiency; per capita GDP, government intervention, and science and technology input also have significant negative effects. The positive effects of per capita GDP, government intervention, and science and technology input on urban industrial carbon emission efficiency are significant at the 1% level in different quartiles, and the inverse-U-shaped relationship between industrial enterprise size and urban industrial carbon emission efficiency is observed, indicating that industrial transformation and upgrading is the key to realize carbon emission reduction and improve industrial carbon emission efficiency in cities in the high and low quartiles.

As a new type of economic form, exploring the non-linear impact of the digital economy on industrial carbon emission efficiency in Chinese cities can help clarify the relationship between the digital economy and industrial carbon emission efficiency and provide suggestions for further promoting the development strategy of the digital economy and for the transformation and upgrading of traditional industries. It can be extended to other industry sectors, such as the fast-growing carbon emissions from transportation and services, to explore their emission reduction factors and promote green and sustainable development. For other developing countries similar to China, which is mainly industrial and has rapid development and large carbon emissions, with reference to the results of the study on the non-linear impact of the digital economy on industrial carbon emission efficiency of Chinese cities, we can carry out emission reduction measures in response to the carbon emission reduction effect of the digital economy, and promote the high-quality development of the economy.

5. Discussion

This study first measured the industrial carbon emission efficiency of Chinese cities over many years based on the BAM-DEA efficiency methodology. The design of input-output indicators in this methodology takes into account the mutual influence effect between indicators, incorporates industrial input-output elements into the analytical framework, and uses carbon emissions as a non-desired output. Integrating multiple elemental dimensions, it is more objective, accurate, and robust to analyze its impact on carbon efficiency in the industrial sector than to directly analyze the digital economy on carbon emissions and carbon intensity [23,24].

It has been found that the development of the digital economy significantly improves carbon emission efficiency [9,43]. Another study found a potential non-linear relationship between the digital economy and industrial carbon emission efficiency but only argued for an inverted U-shaped relationship between the digital economy and carbon emission intensity by adding the squared term of the digital economy [28]. This fails to identify the specific changing effects of factors on carbon efficiency at different quartile stages and cannot provide targeted guidance for specific stages of urban development. This study evaluates the non-linear impact of the digital economy on the industrial carbon emission efficiency from different development stages of the city based on a panel quantile regression model, where the regression coefficients of the digital economy are not the same in different quartiles. The development of the digital economy in different quartiles directly affects the industrial carbon emission efficiency. In the early stage of the development of the digital economy, the digital industry increased significantly, and the carbon emission increased with it, which showed an inhibitory effect on the industrial carbon emission efficiency, and the industrial carbon emission efficiency increased in the early stage of the development of the digital economy. In the mature stage of development, it promotes the transformation and upgrading of the manufacturing industry, optimizes the allocation of factor resources, reduces industrial carbon emissions, and improves industrial carbon emission efficiency. The results of previous studies have also shown that the digital economy indirectly increases per capita carbon emissions by promoting economic growth, industrial structure upgrading, and financial development [44]. When digitalization develops to a certain extent, the digital economy is beneficial to promote technological transformation and upgrading and strengthen the creation of green technology, which changes the energy consumption structure and thus promotes energy conservation and emission reduction [45].

The main contribution of this paper is to provide more theoretical and empirical support for the impact of the digital economy on industrial carbon emission efficiency in Chinese cities and to enrich the research results in this area. Based on the findings of this paper, the following policy recommendations are proposed:

1. There is a significant inverted U-shaped relationship between the digital economy and urban industrial carbon emission efficiency. In the early stage of the development of the digital economy, the local government pays more attention to economic development and neglects environmental protection, and the industrial carbon emission efficiency is lower. During the period of rapid economic development, the public will increase their preference for green industries and demand for carbon reduction, and the digital economy will increase the industrial carbon emission efficiency. Therefore, it is necessary to speed up the application and promotion of the digital economy, an emerging economic form, to help the low-carbon transformation of urban industry, to rationally and optimally lay out the urban industrial system, and to guide the green transformation of the industrial structure, especially in urban agglomerations such as the Yangtze River Delta and the Pearl River Delta, which have advanced industrial levels. It is also necessary to consider the spatial spillover effect of the digital economy, make use of the advantages of network dissemination and information sharing of the digital economy, enhance its spatial contribution to the industrial carbon emission efficiency, strengthen the exchange and cooperation mechanism between the south-eastern region and the inland northwestern region, and jointly explore the coordinated development route of the digital economy.

2. Government intervention can improve urban industrial carbon emission efficiency, and the government should formulate effective environmental regulation strategies to reduce the proportion of high-pollution and high-energy-consumption industries in inefficient district cities. The European Commission published the European Great Deal in 2019 to realize the full potential of the digital transformation to stop climate change and promote stable and sustainable economic growth in Europe by transitioning to clean energy and a circular economy. The Build Back Better Act bill adopted by the United States in 2021 supports the development of the wind power, photovoltaic, and new energy automobile industries and promotes the development of green industries to meet carbon reduction targets. Therefore, the government’s consideration of policies for relevant industrial sectors can actively guide foreign investment to invest more in energy-saving and environmental protection and other industries and enhance the positive effect of technology investment and economic growth on reducing carbon emissions.

3. Enhanced science and technology inputs can significantly improve urban industrial carbon emission efficiency. With a developed urban economy, a relatively clean energy consumption structure, and a less carbon-intensive manufacturing industry, the application of innovative green technologies by enterprises is more prevalent, the industrial carbon emission efficiency is higher, and the eastern region plays an important leading role in realizing China’s energy saving and emission reduction goals of “carbon peaking and carbon neutrality”. The central and western regions also need to seize the opportunity to actively promote industrial intelligence and low-carbon development, develop clean energy sources such as photovoltaic and wind power, and improve urban industrial carbon emission efficiency so as to jointly promote China’s high-quality economic development and sustainable development.

4. There is an inverted U-shaped relationship between industrial enterprise size and urban industrial carbon emission efficiency. Different cities have different characteristics of industrial development and different priorities for measures. For economically backward regions, reducing highly polluting and energy-consuming enterprises and upgrading technology can effectively improve urban industrial carbon emission efficiency. For economically developed areas, the traditional industries are large and perfect, and the industries are in the stage of transformation and upgrading; adjusting the industrial structure is the key to improving the efficiency of urban industrial carbon emission. At the same time, interregional exchanges and cooperation in terms of technology and human resources should be strengthened in order to reduce the phenomenon of interregional transfer of polluting industries and to explore coordinated development routes.

This study has made some progress on how the digital economy affects industrial carbon emission efficiency. An inverted U-shaped relationship was found between industrial enterprise size and urban industrial carbon emission efficiency. The regression coefficients of the digital economy become larger with the increase of quartiles, and the positive effect of the digital economy gradually strengthens, with a negative effect at the lower quartiles. However, there are still some limitations:

1. Based on the principle of data availability, the urban panel data in this study do not cover all the cities in China, and the digital economy and industrial carbon emission efficiency indexes calculated in this study are not reasonable enough. In addition, more factors could be considered in the future, such as environmental regulations.

2. For the study of the non-linear relationship between industrial enterprise size and urban industrial carbon emission efficiency, this paper did not find a clear key threshold interval. The threshold effect can be considered in future studies to determine the threshold value, making the non-linear relationship between the two more specific and precise.

3. In this paper, multiple control variables are selected to accurately determine the influence of industrial carbon emission efficiency on urban industrial carbon emission efficiency. Various factors are complicated and may interact with each other. However, this study did not capture exactly. Machine learning includes excellent generalizability, a fast training speed, and robustness against falling into locally optimal solutions [46]; in the future, machine learning can be used to explore the influence mechanism of industrial carbon emission efficiency and its sub-indexes on industrial carbon emission efficiency.