Exploring the Impacts of the Digital Economy on Carbon Emissions: Lessons from 268 Cities in China

Abstract

Based on the panel data from 268 prefecture-level cities in China from 2011–2020, this study explores the impact of the digital economy on China’s carbon emissions and its mechanisms. The results reveal that the digital economy has a significant urban carbon emission reduction effect, and the robustness test results confirm the reliability of this conclusion. The heterogeneity analysis indicates that regional and city endowment influences this effect, with the effect being relatively stronger in the eastern region and high-grade cities, whereas the effect is not notable in the central and western regions and low-grade cities. In addition, digital economy development in the central region and non-resource cities can reduce carbon emissions, although its impact on peripheral and resource cities remains uncertain. Further mediation effect tests show that the urban carbon emission reduction effect occurs through energy consumption reduction, industrial structure upgrading, and green technology innovation. This study contributes to a deeper understanding of the relationship between the digital economy and carbon emissions, which is significant for formulating digital economy policies to reduce carbon emissions.

1. Introduction

Since the initiation of China’s reform and opening-up policies, its economy grew to become the second largest in the world and was accompanied by qualitative improvements in the living standards of its population. However, escalating carbon emissions resulting from continuous industrialization created climate and environmental challenges that are not conducive to sustained economic and social development. In response, the Chinese government implemented a series of measures aimed at reducing emissions and set forth strategic objectives for achieving peak carbon by 2030 and carbon neutrality by 2060 [1].

China currently finds itself in a critical period of transitioning from an “industrial” to a “digital” economy, with the latter serving as a pivotal catalyst for driving high-quality economic development [2]. This transition entails adjusting the energy structure by adopting digital technologies, facilitating the transformation of the industrial chain toward reduced energy consumption and increased value, and nurturing the emergence of innovative businesses and industries that contribute to the green and low-carbon development of digital technology [3]. Moreover, digital economy integration can effectively mitigate the adverse environmental impacts associated with economic growth, infusing environmental considerations into economic activities and assisting China in attaining its “dual-carbon” goal [4]. The significance of the digital economy in the process of carbon emissions reduction has become increasingly prominent. In the current context of China’s advancement in digital economy development and its dual-carbon goal, what role does the digital economy play in the process of carbon emissions? Does the effect of carbon emissions on digital economy development differ by region or type of city? By what pathways does digital economy development influence carbon emissions? Clarifying these questions can deepen the theoretical understanding of the relationship between the digital economy and carbon emissions, identify the carbon reduction pathways of the digital economy, and promote progress in the digital economy and carbon emission reduction. The results can provide policy insights into China’s efforts to achieve its dual-carbon targets.

The deep integration of digital technology with the economy and society, as well as the role of digital technology in achieving the dual-carbon goal, has garnered increasing attention within the academic community. Scholars emphasized the significant environmental improvement effect of the digital economy and its sustained impetus for enhancing carbon emissions performance [5,6,7]. The digital economy has spatial spillover effects on carbon emissions, indicating that a city’s digital economy also affects carbon emissions in neighboring areas [8,9,10]. The digital economy influences carbon emissions reduction, manifesting in both direct and indirect modalities. Notably, the direct impact is considerably more pronounced than the indirect impact [11].

Research on the indirect impact pathways of the digital economy on carbon emission reduction primarily focuses on three aspects. First, in terms of technology, telecommunications infrastructure facilitates the diffusion and advancement of green technologies, thereby enhancing green technological innovation [12]. Through novel production techniques, digital technologies directly impact energy efficiency, resulting in reduced energy consumption per unit and consequently lowering carbon intensity [13,14,15]. Xu et al. [16] highlighted that digital economy development fosters biased technological progress, enhances resource allocation efficiency, and reduces environmental pollution and carbon emissions, particularly in remote areas. Second, the industries associated with the digital economy are predominantly low-carbon industries, the development of which upgrades the industrial structure, consequently enhancing the overall carbon emission efficiency of the economic system. Industrial structure upgrades mediate the relationship between the digital economy and carbon emissions [17].

Furthermore, digital technology facilitates the transformation of consumer preferences toward environmentally friendly choices, promoting the adoption of a green and low-carbon lifestyle among individuals [18]. Third, the digital economy significantly improves energy utilization efficiency, with the promotional effect on energy efficiency intensifying as the level of economic development increases [19]. Energy intensity and the scale of energy consumption are intermediary mechanisms by which digital economy development improves carbon emission performance [3]. Zhou and Li [20] demonstrated that the application of digital technology enhances resource integration capabilities, aids in scientific decision-making, and enables effective environmental oversight, thus providing crucial support for green production practices in enterprises.

Simultaneously, it is imperative to acknowledge that the digital economy alone does not deliver a unilateral emission reduction effect [21]. The digital economy’s impact on carbon emissions exhibits a rebound effect, resulting in negative environmental externalities and a continuous increase in carbon emissions [22]. Hamdi et al. [23] noted that the rapid development of digital infrastructure led to accelerated growth in the information and communication technology sector and related industries. However, this growth resulted in increased electricity and heat generation, which may not fully harness the green emissions reduction potential of the digital economy. Ghobakhloo [24] emphasized that data characterized by both green and technological attributes, when utilized as core production factors, positively contribute to energy savings and emissions reductions. However, a sole focus on increasing productivity can be detrimental to efforts to reduce carbon emissions. Moreover, the role of digital technological advancements in carbon intensity is uncertain [25,26]. The progress of digital technology represents a “double-edged sword” that leads to improved energy utilization efficiency but can also result in excessive energy consumption, thereby negatively impacting the achievement of China’s carbon emissions reduction targets.

Compared with existing research, this study offers the following potential marginal contributions: First, prior studies tended to employ relatively singular indicators to measure the level of digital economy development, failing to comprehensively reflect the diverse characteristics of the digital economy. This study assesses the level of digital economy development using a comprehensive index system encompassing four key dimensions: digital infrastructure, digital industry development, digital innovation capability, and digitally inclusive finance. This expanded framework for measurement and research provides a broader understanding of the digital economy, reflects its characteristics more fully, and enhances the accuracy and reliability of the empirical analyses. Second, exploring the heterogeneous characteristics of the digital economy’s carbon emission reduction effects from multiple perspectives, such as geographical location, city classification, administrative position, and resource endowment, provides an empirical basis for formulating targeted carbon emission reduction policies. Third, while the current research primarily examines the impact of digital economy development on carbon emissions from a single perspective, such as technological progress, industrial development, and energy efficiency, studies exploring the multiple pathways through which digital economy development affects carbon emissions are scarce. This study includes factors such as energy consumption intensity, industrial structure upgrading, and green technological innovation within a unified analytical framework, enriching research on the mechanisms by which the digital economy influences carbon emissions. Based on the impact of the digital economy on urban carbon emissions reduction, this study provides new insights for promoting digital economy development and advancing carbon emissions reduction, thus enriching the government’s policy toolkit to foster carbon emissions reduction. This is of significant importance and of value for China’s efforts to achieve its dual-carbon targets.

The remainder of this paper is organized as follows. The second section presents the research hypotheses and theoretical analysis. The third section discusses the methodology and data. The fourth and fifth sections provide an analysis of the empirical results and the mechanisms of influence, respectively. The final section offers the conclusions and policy implications. Based on the aforementioned main research content, Figure 1 depicts the technical roadmap for this study.

2. Research Hypothesis and Theoretical Analysis

2.1. Digital Economy, Energy Consumption, and Urban Carbon Emissions

The digital economy plays a crucial role in optimizing the structure of traditional energy consumption and improving energy-use efficiency, thereby reducing urban carbon emissions. This impact is evident at the individual, enterprise, and governmental levels.

First, at the individual level, the digital economy, characterized by advancements in big data and artificial intelligence, has permeated various aspects of individual production and daily life [27]. This integration breaks the limitations of time and space in knowledge acquisition; enhances access to resources, education, and culture; facilitates the sharing of high-quality resources; and contributes to the widespread adoption of green and low-carbon concepts and behaviors [20]. Individuals adapt to environmental and business model changes through hands-on learning or reeducation, leading to an upgrade in human capital [28]. By leveraging advanced technology, knowledge, and experience, individuals accumulate human capital, which replaces traditional factors, optimizes factor endowments, and enhances technological advancement within enterprises. This, in turn, promotes energy conservation and improves energy use efficiency, resulting in urban carbon reduction.

Second, at the enterprise level, firms can apply digital technology to adjust their industrial layout and improve production processes and organizational forms, thereby enhancing innovation efficiency [29]. Digital technologies, such as big data, cloud computing, and 5G, can be integrated into all stages of energy production and consumption. This integration facilitates the accurate measurement of carbon footprints, guides the efficient allocation of energy factors, promotes the improvement of green total factor productivity, reduces pollution control costs through digital means, and ultimately reduces carbon emissions in urban areas [30].

Third, at the government level, the digital economy provides technical support for optimizing the government’s environmental regulatory model. Digitally operating the carbon emissions trading market improves the government’s level of informatization and enables intelligent operations [31]. Additionally, digital media help address information asymmetry between the government and society. It creates new opportunities and channels for the public to access environmental information and develops environmental protection concepts, thereby fostering synergistic governance among the government, enterprises, and the public in the field of environmental protection to reduce urban pollution emissions.

Based on the analyses provided, we propose the following hypothesis:

H1. 

The digital economy can reduce urban carbon emissions through energy consumption optimization.

2.2. Digital Economy, Industrial Structure Upgrading, and Urban Carbon Emissions

The digital economy has significantly transformed the combination of factors and structure of inputs in the economic system. It introduced new production factors and drove profound changes in traditional production methods and industrial structures, thereby promoting the development of industries toward intelligence and sustainability [32]. First, the digital economy facilitates interconnection between upstream and downstream enterprises in the manufacturing industry chain, simplifying production, circulation, distribution, and exchange processes. This effectively reduces the production and operations costs of enterprises [33]. Second, the digital economy promotes industrial innovation to enhance energy efficiency. The digital transformation of enterprises has enabled breakthroughs in traditional manufacturing technologies and the production of complex structural components. This enables more refined processing and production, fosters scale effects that drive regional innovation and contributes to addressing challenges, such as the “necklace problem” [34]. Third, the digital economy has penetrated all aspects of production, operation, and sales in traditional industries through the deployment of digital technologies, services, and information. This integration enhances an industry’s overall efficiency [34]. Furthermore, the digital economy leverages existing resources to empower traditional industries, drive product and business model innovation, create value-added opportunities, and revitalize dormant resources [32]. Consequently, the digital economy promotes upgrading the industrial structure through four key aspects: cost savings, economies of scale, efficiency improvement, and innovation empowerment. Industrial structure, as a major contributor to carbon emissions, has a strong and enduring relationship with carbon emissions [35].

Upgrading the industrial structure can facilitate the flow of production factors to high-productivity sectors, ensuring a more rational allocation of resources among industrial sectors. Upgrading also fosters the coordinated development of various industries, mitigates fluctuations caused by resource mismatches, enhances regional resource utilization efficiency, reduces energy consumption per unit of product, and decreases carbon emissions intensity [36]. Industrial structure upgrading is characterized by the development of a more service-oriented economy. The gradual shift in development focus from primary and secondary industries to tertiary industries can challenge the current dominance of coal-based energy consumption, leading to reduced fossil energy consumption and, subsequently, lower urban carbon emission intensity. Based on this analysis, we propose the following hypothesis:

H2. 

The digital economy can reduce urban carbon emissions by upgrading industrial structures.

2.3. Digital Economy, Green Technology Innovation, and Urban Carbon Emissions

The digital economy, which is characterized by the utilization of data and information as essential production factors, can drive green technological innovation by reducing research costs, facilitating regional cooperation, and improving the market environment [37]. First, the application of digital information networks enhances regional connectivity and plays a catalytic role in the R&D of green technologies. By reducing information asymmetry, overcoming financing constraints, and stimulating the willingness to innovate green technologies, the digital economy promotes green technological innovation [37]. Second, the extensive use of big data, the internet, and other digital technologies in enterprises creates a foundation for the efficient dissemination and allocation of knowledge among innovation factors. This facilitates the effective sharing of knowledge and information in green technology innovation activities, enhances information-sharing capacity, and collectively strengthens regional green technology innovation capacity [13]. Furthermore, the integration of digital and energy development technologies can enhance macroeconomic programmability and improve the market environment. This provides market players with easier access to innovation resources within value networks, thus amplifying the spillover effect of green technological innovation.

Green technological innovation serves as a fundamental driver for achieving green and low-carbon development and effectively addressing environmental pollution challenges. First, green technological innovation contributes to the dual-carbon goal. By incorporating existing green technologies into efficient information networks, existing and potential resources can be harnessed to generate high composite value and promote green and low-carbon urban transformations through scale and synergistic effects [38]. Green, low-carbon technologies prioritize environmental sustainability. Their effective utilization can accelerate the environmental transformation of highly polluting industries and promote cleaner and more efficient production practices within enterprises. Consequently, the contribution of non-green industry output to the overall economic output decreases, leading to a reduction in carbon emissions intensity within cities [38]. Based on the above analysis, we propose the following hypothesis:

H3. 

The digital economy enables urban carbon reduction through green technology innovation.

3. Methodology and Data

3.1. Model

To explore the relationship between the digital economy and urban carbon reduction, Model (1) was constructed as follows:

$$C{I}_{it}={\alpha }_0+{\alpha }_{1}D{E}_{it}+{\displaystyle \sum {\alpha }_{2}Contro{l}_{it}}+u_i+v_t+{\epsilon }_{it}$$

(1)

where the subscripts $i$ and $t$ represent cities and years, respectively, $C{I}_{it}$ is the carbon intensity of cities, $D{E}_{it}$ reflects the level of development of the digital economy, ${\alpha }_0$ is the intercept term, ${\alpha }_1$ represents the coefficient of influence of digital economic development, $Contro{l}_{it}$ represents the relevant control variables, $u_i$ and $v_t$ are location and time fixed effects, respectively, and ${\epsilon }_{it}$ is a randomized disturbance term.

3.2. Variables

3.2.1. Explanatory Variables

Carbon emissions intensity ($CI$) is expressed as the proportion of total carbon emissions ($C{O}_2$) to regional GDP ($GDP$) [39]. Total carbon emissions use carbon emission coefficients to discount the carbon emissions generated when utilizing energy and sum to measure the total carbon emissions at the city level. In Equation (2), $E_h$ is the $h$ class energy consumption, and $h$ is the class energy carbon emission factor. Furthermore, measuring the carbon intensity of cities in terms of total urban carbon emissions as a share of real GDP is consistent with the core explanatory variable quantiles for the reasonableness of the study results.

$${CO}_2={\displaystyle \sum E_h}\cdot C_h$$

(2)

$$CI=C{O}_2/GDP$$

(3)

3.2.2. Principal Explanatory Variables

According to Zhang et al. [3], and considering available data, there is currently no uniform standard for measuring the digital economy development index ($DE$) at the city level. This study constructs a digital economy development indicator system using four parameters: digital infrastructure, industry development, innovation capacity, and financial inclusion (

Table 1. Digital economy development level index construction.

Digital Economy Development Level Index	Primary Index	Secondary Index	Measurement	Stats
	Digital infrastructure	Internet penetration	Number of broadband Internet access users per 100 people (households)	+
		Mobile phone penetration	Number of mobile phones per 100 people (households)	+
	Digital industry development	Information industry foundation	Number of employees in information transmission, computer services, and software (10,000)	+
		Telecommunication industry foundation	Total telecommunications services per capita (Yuan)	+
	Digital innovation capability	Foundation of digital innovation	Science and technology expenditure (ten thousand yuan)	+
		Digital talent support	Number of institutions of higher learning (number)	+
	Digital universal financial	Digital Financial Inclusion Index	Digital financial inclusion coverage (−)	+

). To reduce bias in the index measurement, we use a relatively objective entropy value method to measure the urban digital economy development index.

3.2.3. Control Variables

To control the interference of other factors on urban carbon emission intensity to minimize estimation bias, we follow Yu et al. [15], Chen et al. [38], and Guo et al. [36] and add the following control variables. (1) Level of financial development ($FDE$), measured using the balance of loans from financial institutions as a share of GDP at the end of the year. (2) Degree of utilization of foreign capital ($FDI$), measured using the share of industrial output value of foreign-invested enterprises in the GDP of each city. (3) The urbanization rate ($UBR$), measured using the proportion of the resident urban population to the regional population. (4) Degree of fiscal decentralization ($FID$), measured as the ratio of local budget expenditure to local budget revenue. (5) Capital formation rate ($FAI$), measured using gross investment in fixed assets as a share of the GDP.

3.2.4. Intermediary Variable

(1) Energy consumption intensity ($ECI$): As in Lin and Ma [40], this study uses electricity consumption per unit of GDP in each city. (2) Industrial structure upgrading ($UIS$): Drawing on Liu et al. [37], we adopt the proportion of tertiary industry output value to secondary industry output value. Upgrading industrial structure is an important part of China’s low-carbon city development. (3) Green technology innovation ($GTI$): Following Du et al. [41], we adopt the number of green invention patent applications per 10,000 people as a measure, which corresponds to the comprehensive capacity of a city’s green technology output.

3.3. Data Description

The sample period for this study is 2011–2020. Data are obtained from the China Energy Statistical Yearbook, China Industrial Statistical Yearbook, China Urban Statistical Yearbook, China Environmental Statistical Yearbook, China Patent Database, EPS Database, CSMAR Database, and Wind Database. Some missing values are filled in using the average annual growth rate method, and the remaining observations with severe missing data were excluded, yielding 2680 samples after processing. The descriptive statistics for each variable are shown in

Table 2. Descriptive statistics.

Variable	Obs.	Mean	Std.	Min.	Max.
CI	2680	0.038	0.036	0.002	0.434
DE	2680	0.066	0.072	0.008	0.650
FDE	2680	1.022	0.590	0.310	3.418
FDI	2680	0.111	0.147	0.000	0.716
UBR	2680	0.569	0.145	0.223	1.000
FID	2680	2.801	1.690	0.649	19.745
FAI	2680	0.912	0.487	0.010	5.666
ECI	2680	0.090	0.068	0.015	0.937
UIS	2680	1.016	0.567	0.114	5.348
GTI	2680	5.011	1.648	0.000	10.252

.

3.4. Comprehensive Digital Economy Level Measurement

To remove the differences in the outline of the indicators, the original indicators were standardized. Given that the selected indicators are all positive, the specific calculation method is:

Where $Y_{ij}$ denotes the standardized indicator value; $i$ denotes the year; $j$ denotes the jth indicator; $X_{ij}$ denotes the original value of the $j$th indicator in year $i$; and max ($X_i$) and min ($X_j$) are the maximum and minimum values of the jth indicator for all years, respectively.

Positive indicator processing formula:

$$Y_{ij}={\displaystyle \frac{max(X_j)-X_{ij}}{max(X_j)-min(X_j)}}$$

(4)

Calculate the share of the value of the jth indicator in year i:

$$N_{ij}={\displaystyle \frac{Y_{ij}}{{\displaystyle \sum _{i=1}^m{Y}_{ij}}}}$$

(5)

Calculate the information entropy:

$$e_j=-{\displaystyle \frac{1}{lnm}}{\displaystyle \sum _{i=1}^m(N_{ij}\times ln{N}_{ij})},0\le e_j\le 1$$

(6)

Calculate the information entropy redundancy:

$$d_j=1-e_j$$

(7)

Calculate indicator weights:

$$w_j={\displaystyle \frac{d_j}{{\displaystyle \sum _{j=1}^m{d}_j}}}$$

(8)

Calculate the composite score for a single indicator:

$$S_{ij}=w_j\times Y_{ij}$$

(9)

Calculate the composite indicator score for year i:

$$S_i={\displaystyle \sum _j^n{S}_{ij}}$$

(10)

3.5. Dynamic Evolution Analysis

To determine the distribution pattern and evolutionary trend of the digital economy and carbon emissions intensity in different periods, this study applies the kernel density estimation method to examine their dynamic evolution. The results are shown in Figure 2 and Figure 3.

Figure 2 shows the kernel density curve of China’s digital economy. A tendency for the kernel density curve to gradually shift to the right during the sample period is observed. This result indicates that the level of development of China’s digital economy continued to improve from 2011 to 2020. Based on the analysis of the kurtosis of China’s digital economy for each year, the gap in the digital economy between regions became smaller, and the distribution tended to converge. Figure 3 shows the kernel density curve of China’s carbon emission intensity, which shows that the kernel density curve of China’s carbon emission intensity shifted slightly to the left during the sample period. From 2011 to 2020, China made progress in reducing its urban carbon emissions in the context of the digital economy. However, the shape of the curve suggests that the gap between carbon emission intensities of different regions shows a U-shaped development trend, which first narrows and then widens.

4. Empirical Results

4.1. Benchmark Regression Results

Table 3. Benchmark regression results.

Variable	(1) CI	(2) CI
DE	−0.095 *** (−4.613)	−0.068 *** (−3.274)
FDE	-	0.007 *** (2.766)
FDI	-	0.012 * (1.720)
UBR	-	0.024 * (1.951)
FID	-	0.005 *** (3.085)
FAI	-	−0.005 *** (−3.269)
Constant terms	0.108 *** (9.683)	0.053 *** (2.890)
Control variables	NO	YES
Year fixed	YES	YES
City fixed	YES	YES
Sample size	2680	2680

Note: Numbers in parentheses represent t-values; “*” and “***” indicate significant at the 10% and 1% levels, respectively.

reports the direct impact of digital economy development on urban carbon intensity. Columns (1) and (2) show the regression validation of urban carbon emission intensity without and with control variables, respectively, and the coefficients’ positivity, negativity, and significance remain unchanged, which indicates that the digital economy effectively reduces urban carbon emission intensity. The digital economy can promote industrial structure leapfrogging, promote the transformation of the energy consumption structure to low-carbon and green, promote green technological innovation, effectively reduce energy consumption, and curb urban carbon emissions.

With respect to the control variables, the coefficient of the impact of the financial development level is significantly positive at the 1% confidence level. This may be because the crude financial system, which focuses on the scale mechanism, has gradually become the main “contributor” to incremental carbon emissions, leading to resource and environmental problems. The regression coefficient of foreign capital utilization is positive and significant at the 10% level. This may be because opening up to the outside world facilitates the relocation of high-pollution and high-expenditure firms to areas with less stringent environmental regulations, which complicates the process of reducing the resource dependence of incoming firms. The coefficient of influence of the urbanization level is significantly positive. Accelerating urbanization is an important factor affecting carbon emissions in the region, and the resulting population buildup exacerbates the carbon emissions problem. Meanwhile, the degree of fiscal decentralization and intensity of urban carbon emissions are positively correlated. Fiscal decentralization endows local governments with greater autonomy, and under the existing incentive system, they focus more on economic development than on improving environmental quality, leading to high energy consumption and carbon emissions. Additionally, the fixed asset investment variable is significantly negative at the 1% confidence level, likely because fixed asset investment plays an important role in the industrial development process under digital empowerment. This can contribute to the formation of production-scale effects and thus effectively curb the total amount of carbon emissions generated per unit of investment.

4.2. Robustness Analysis

4.2.1. Core Explanatory Variables Lagged One Period

Given that the impact of the digital economy on urban carbon emission intensity has a lag, we lag the core explanatory variables by one period in the regression equation to effectively reduce model estimation bias. In Column (1) of

Table 4. Robustness test.

Variable	(1) Core Explanatory Variables Lagged One Period	(2) Adjustment of Sample Period	(3) Bilateral 1% Retraction	(4) Dynamic GMM Panels	(5) Excluding Municipalities	(6) Instrumental Variable Approach
DE	−0.073 *** (−3.325)	−0.085 *** (−3.467)	−0.062 *** (−3.435)	−0.052 *** (−2.60)	−0.061 *** (−2.678)	−0.312 *** (−3.468)
Constant terms	0.058 *** (3.399)	0.069 *** (3.141)	0.037 *** (2.697)	−0.019 *** (−3.19)	0.032 *** (3.344)	0.224 *** (4.634)
AR(1)	-	-	-	0.045	-	-
AR(2)	-	-	-	0.352	-	-
Hansen test	-	-	-	0.309	-	-
Phase I F-value	-	-	-	-	-	122.041
Control variables	YES	YES	YES	YES	YES	YES
Year fixed	YES	YES	YES	YES	YES	YES
City fixed	YES	YES	YES	YES	YES	YES
Sample size	2412	2144	2680	2412	2640	2680

Note: “***” indicate significant at the 1% level.

, the coefficient of the impact of the digital economy is significantly negative at the 1% level, indicating that digital economy development can promote urban decarbonization, which makes the reliability of previous research findings debatable.

4.2.2. Adjusting the Sample Period

Considering that the double pressure of the growth rate shift and structural adjustment faced by China after 2012 affected the accuracy of the model estimation, we excluded samples before 2013 and performed the regression test. The results in Column (2) of Table 4 indicate that digital economic development is beneficial for realizing urban carbon emission reduction, which is consistent with the results of the benchmark regression, demonstrating its robustness.

4.2.3. Bilateral 1% Retraction

To avoid the effects of extreme values and outliers, we applied a two-sided 1% shrinkage to all continuous variables. The results are shown in Column (3) of Table 4, where the coefficients of the core explanatory variables remain significantly negative, indicating that the results of the benchmark regression in this study are robust.

4.2.4. Dynamic GMM Panels

Continuing the analytical line of previous studies and considering the endogeneity of UNIL, we adopt the method of Arellano and Bond [42], taking the lagged term of urban carbon emission intensity as an endogenous variable and using the GMM method to test the validity of the model setting. The regression results are shown in Column (4) of Table 4. The p-value for the AR(1) test was <0.1, while the AR(2) test was not statistically significant. The perturbation terms in the GMM estimation were shown to have a first-order serial correlation, and there was no second- or higher-order autocorrelation. Combined with the Hansen test results, the p-value was >0.1, which passes the over-identification constraint test and indicates that the GMM instrumental variables were valid. After the dynamic GMM panel test, the regression coefficients remain significantly negative, further supporting the rationality of the underlying model conclusions.

4.2.5. Excluding Municipalities

Given the favorable policies of municipalities and their special geographic locations, studying municipalities may lead to a greater enabling effect on the digital economy. Therefore, we exclude Beijing, Shanghai, Chongqing, and Tianjin from additional empirical regression tests. The coefficient of the digital economy’s impact was significantly negative at the 1% level, as shown in Column (5) of Table 4, further supporting the accuracy of the baseline regression findings.

4.2.6. Instrumental Variable Approach

Other unobserved factors may also influence urban carbon intensity. To mitigate this possible endogeneity problem in the model, we use the number of post offices per million people in each city in 1984 as the instrumental variable. However, because this variable is cross-sectional, it cannot be used in panel data for model estimation. According to Zhang et al. [21], the number of internet users in the entire country in the previous year is used to estimate the model with its respective constructed interaction term. Cities with historically high post office penetration are well-positioned for rapid growth in the digital economy. However, the correlation with the current carbon emissions situation was not significant, and the conditions of correlation and exogeneity were satisfied for the selection of instrumental variables. The F-statistic is 122.041, which is significantly greater than the empirical value of 10 and corresponds to a p-value of 0.001. The weak instrumental variable test indicates better exogenous conditions; therefore, the selected instrumental variables are more appropriate, as they meet the expected conditions. The results in Column (6) of Table 4 show that, controlling for possible endogeneity, the urban carbon abatement effect of the digital economy remains significant. Furthermore, the regression coefficients became larger in absolute value after performing the instrumental variable tests. This suggests an error in the model estimation caused by overlooking endogeneity. The instrumental variable regression analysis results confirm the robustness of the main results.

4.3. Heterogeneity Test

4.3.1. Regional Heterogeneity

As shown in Columns (1) and (2) of

Table 5. Regional heterogeneity.

Variable	(1) East	(2) Midwest	(3) High-Level Cities	(4) Low-Level Cities
DE	−0.101 *** (−4.896)	−0.020 (−0.482)	−0.045 *** (−3.898)	−0.034 (−0.402)
Constant terms	0.062 ** (2.172)	0.053 *** (3.007)	0.013 (1.239)	0.076 *** (4.854)
Control variables	YES	YES	YES	YES
Year fixed	YES	YES	YES	YES
City fixed	YES	YES	YES	YES
Sample size	1090	1590	1130	1550

Note: “**” and “***” indicate significant at the 5% and 1% levels, respectively.

, the coefficients of the impact of the core explanatory variables are significantly negative at the 1% level for the eastern region. First, digital economy development in the Eastern region could significantly reduce the intensity of carbon emissions. This may be because this region leads the country in economic development, technological development, and the degree of integration and development of the digital economy with traditional industries. These factors enhance the region’s ability to integrate resources and undergo innovation. It can effectively utilize the potential of data elements, promote the rapid development of digital industries, and facilitate the intensive transformation of industrial production methods. It has comparative advantages in terms of financial resources, scientific and technological talent, and green technological innovation, which are conducive to unleashing the pollution control effects embedded in the digital economy. Second, digital economy development in the Central and Western regions failed to significantly reduce carbon emissions. The probable reasons for this are the lagging economic development of the central and western regions, the relative lag in the construction of digital infrastructure, and a high dependence on traditional resources, which are susceptible to high pollution and energy consumption. At the same time, the region has relatively low labor costs and is home to many labor-intensive and highly polluting industries. Initially, the energy consumption brought about by digital economy development and the role generated by digital empowerment offset each other, which is not conducive to releasing the carbon emission reduction dividend of the digital economy, resulting in the carbon emission reduction effect of digital economy development in the Central and Western regions not being significant.

The classification of city grades is based on the 2023 China City Business Attractiveness Ranking as the division standard. High-grade cities include four categories: first-, new-first-, second-, and third-tier cities, whereas other cities are classified as low-grade. From the regression results in Columns (3) and (4) of Table 5, the higher-ranking cities have a significant negative impact on carbon emissions intensity at the 1% level, whereas lower-ranking cities have a negative but not significant impact coefficient. A probable reason is that high-level cities have significant advantages in terms of commercial resource agglomeration, urban hub characteristics, resident activity levels, lifestyle diversity, and future malleability. The concentration of commercial resources facilitates synergistic effects and dissemination of innovation among enterprises, promoting the application of clean energy and efficient production technologies, thereby effectively reducing carbon emissions. The hub feature of cities optimizes the flow of resources through intelligent transportation and logistics systems, thereby reducing energy consumption. Concurrently, the increased activity levels of urban residents decrease their reliance on traditional high-carbon activities and foster the development of green consumption patterns and a sharing economy. Moreover, the diversity of lifestyles supported by the digital economy provides residents with low-carbon lifestyle options such as remote work and online education, further reducing carbon emissions. The future malleability of cities highlights the potential of digital technology for monitoring and managing carbon emissions and providing technical support for the green transformation and sustainable development of industries.

4.3.2. Urban Endowment Heterogeneity

In this study, we refer to Ding et al. [2] and divide the sample into central and peripheral cities. Central cities include municipalities, sub-provincial cities, and provincial capitals, while the remaining prefecture-level cities are considered peripheral cities. Moreover, based on the results in Columns (1) and (2) of

Table 6. Urban endowment heterogeneity.

Variable	(1) Central Cities	(2) Outlying Cities	(3) Resource-Based Cities	(4) Non-Resource-Based Cities
DE	−0.082 *** (−3.095)	−0.027 (−0.771)	0.036 (0.554)	−0.090 *** (−4.091)
Constant terms	0.080 ** (2.330)	0.043 *** (4.078)	0.039 *** (2.868)	0.077 *** (3.421)
Control variables	YES	YES	YES	YES
Year fixed	YES	YES	YES	YES
City fixed	YES	YES	YES	YES
Sample size	360	2320	1090	1590

Note: “**” and “***” indicate significant at the 5% and 1% levels, respectively.

, the regression coefficients of the digital economy in the central city are significantly negative at the 1% level, and the impact coefficients of the peripheral cities are negative but not significant. First, digital economy development in central cities began earlier, and the digital infrastructure was relatively perfect. It can provide transformation and upgrading paths for traditional industries, accelerate the process of green technological innovation of enterprises, and improve the security capacity of renewable energy supply, thus producing a strong carbon reduction effect. Second, central cities have a “high degree of primacy” and a serious polarization phenomenon, which will siphon off many types of production in the early stages of development. The absorption of resources and talents far exceeds the trickle-down effect, resulting in the slow development of neighboring cities, which will deprive peripheral cities of their green development capacity; therefore, the carbon emission reduction effect of its digital economy is not notable.

According to the National Sustainable Development Plan for Resource-Based Cities (2013–2020), the research sample is divided into resource- and non-resource-based cities. Based on the empirical results in Columns (3) and (4) of Table 6, the carbon emissions reduction effect of the digital economy on non-resource cities is notable; however, the impact on resource cities remains uncertain. One possible reason is that resource-based cities are tasked with supplying basic resources and important raw materials, and their industries are mainly concentrated in the low- and medium-end production and processing sectors, with a high degree of dependence on initial resources and a weak screening mechanism for foreign investment. Forming a development mode dominated by resource-based industries is convenient but results in the excessive consumption of resources, environmental pollution, and other problems. This results in path dependence and a lock-in effect, prolonging the transformation of enterprises. In contrast, non-resource-based cities rely more on innovation-driven green economic development models, generally have higher levels of science and technology innovation, and capitalize on the opportunities of the digital economy. This can accelerate the R&D of green and low-carbon technologies, promote the upgrading of regional industrial structures, and further reduce regional carbon emissions.

5. Mechanism Analysis

Based on the previous theoretical analyses, digital economy development may empower urban carbon emission reduction through three paths: energy consumption reduction, industrial structure upgrading, and green technology innovation. Therefore, this study builds Models (11), (12), and (13), drawing on the mediation mechanism test by Zhao et al. [43]:

$$EC{I}_{it}={\varphi }_0+{\varphi }_{1}D{E}_{it}+{\displaystyle \sum {\varphi }_{2}Contro{l}_{it}}+u_i+v_t+{\epsilon }_{it}$$

(11)

$$UI{S}_{it}={\phi }_0+{\phi }_{1}D{E}_{it}+{\displaystyle \sum {\phi }_{2}Contro{l}_{it}}+u_i+v_t+{\epsilon }_{it}$$

(12)

$$GT{I}_{it}={\chi }_0+{\chi }_{1}D{E}_{it}+{\displaystyle \sum {\chi }_{2}Contro{l}_{it}}+u_i+v_t+{\epsilon }_{it}$$

(13)

In Models (11)–(13), $EC{I}_{it}$, $UI{S}_{it}$, and $GT{I}_{it}$ are the mechanism variables energy consumption intensity, industrial structure upgrading, and green technology innovation, respectively, and the remaining variables are defined as in Model (1). Models (11)–(13) explore the relationships between the core explanatory and mechanism variables. If the coefficients of the core explanatory variables are significant and have the same sign as expected, then digital economy development can significantly enhance the carbon emission reduction capacity of cities through these three paths. Moreover, energy consumption intensity, industrial structure upgrading, and green technology innovation can significantly reduce urban carbon emission intensity. This indicates that digital economy development can significantly enhance the carbon emission reduction capacity of cities through these three paths.

5.1. Energy Consumption Intensity Channel Analysis

Column (1) of

Table 7. Intermediary mechanism test.

Variable	(1) ECI	(2) UIS	(3) GTI
DE	−0.122 *** (−3.754)	0.713 * (1.767)	1.250 *** (3.196)
Constant term	0.052 (1.419)	3.242 *** (9.649)	7.501 *** (20.760)
Control variables	YES	YES	YES
Year fixed	YES	YES	YES
City fixed	YES	YES	YES
Sample size	2680	2680	2680

Note: “*” and “***” indicate significant at the 10% and 1% levels, respectively.

shows the regression results of energy consumption for the reduction-effect mechanism. The coefficient of influence of the digital economy is significantly negative at the 1% level, indicating that digital economy development can help realize a reduction in energy consumption, which can effectively improve a city’s carbon emission reduction capacity. The digital economy, as a new model of economic development with its deep embeddedness and applied innovation in the fields of energy, resources, and environment, lays the foundation for scientific production decisions. This improved the efficiency of production and consumption of existing mechanisms and energy patterns and facilitated the shift from experience- to data-driven production activities of enterprises. Injecting new kinetic energy into the promotion of a low-carbon transformation of the economy can effectively control carbon emissions from the manufacturing industry, thus contributing to improved carbon emission efficiency. The above analyses show that the digital economy can achieve urban carbon reduction through the mechanism of energy consumption promotion and reduction; thus, H1 was tested effectively.

5.2. Industrial Structure Upgrading Channel Analysis

Column (2) of Table 7 shows the regression results for the effect mechanism of industrial structure upgrades. The coefficient of influence of the digital economy is significantly positive at the 10% level, indicating that digital economy development is conducive to upgrading the industrial structure, which can effectively improve the capacity for urban carbon emission reduction. The digital economy enables industrial structure upgrading, which can deepen the integration and development of digital economy industries with traditional industries and provide enterprises with resources and motivation for innovation. Industrial transformation and upgrading may be driven by four factors: breaking bottlenecks in the innovation chain, improving the quality of the manufacturing chain, optimizing the efficiency of the supply chain, and expanding the space of the service chain [43]. The characteristics of the industrial structure are the intrinsic driving forces for the high-quality development of China’s economy and can fundamentally determine the distribution patterns of energy consumption and pollution emissions. Upgrading the industrial structure can promote scale operation and specialized division of labor in economic activities, enhance the efficiency of resource factor utilization, and promote the development of a green economy with cost savings and optimal resource allocation. The analysis above demonstrates that the digital economy can realize urban carbon emission reduction through the mechanism of industrial structure upgrading, and H2 is effectively verified.

5.3. Green Technology Innovation Channels Analysis

Column (3) of Table 7 presents the regression results for the mechanism of the effect of green technological innovation. The coefficient of influence of the digital economy is significantly positive at the 1% level, indicating that digital economy development is beneficial for promoting green technological innovation, which can effectively improve a city’s carbon emission reduction capacity. Digitalization is becoming the dominant force driving innovation and transformation, and the competitive advantages brought about by innovation are encouraging firms to transform their organizational management and business models, which will motivate them to increase their investment in R&D in green technologies. The application of digital information networks strengthens regional connectivity, improves the docking efficiency of urban innovation subjects and innovation factors, and supports the idea that different innovation subjects can participate simultaneously in green innovation activities in different spaces. Green innovation is an important driving force for green sustainable development and an effective path for achieving the strategic goal of carbon neutrality. Green technological innovation can enhance the advanced and environmentally friendly nature of technology and equipment, promote the transformation of traditional industries to low-carbon and intelligent industries, improve the efficiency of traditional energy sources such as coal, reduce energy consumption and carbon emissions per unit of output, and empower the low-carbon development of cities. Simultaneously, it facilitates the application and popularization of clean energy, such as solar, wind, and water energy, which is an effective way to solve urban diseases. The above analysis shows that the digital economy can realize urban carbon emissions reduction through green technological innovation, and H3 is effectively verified.

6. Conclusions and Policy Implications

The digital economy is a significant driver of enhanced quality and efficiency of economic development. Developing strategies to leverage the digital economy to promote carbon emission reduction under the dual-carbon goal is a crucial long-term plan. This study empirically examines the impact of the digital economy on urban carbon emission intensity and explores the underlying mechanisms using panel data from 268 prefecture-level cities in China from 2011 to 2020.

The main conclusions drawn from the results are as follows. (1) The digital economy empowers the low-carbon transition of cities, as evidenced by multiple robustness tests. Its development improves the government’s capacity for low-carbon governance, accelerates the low-carbon transformation of industrial enterprises, and cultivates green lifestyles among residents. These factors provide opportunities for energy conservation and emission reduction. (2) We find regional heterogeneity in the effects of the digital economy on urban carbon emission reduction. The impact is significant in the eastern region and high-grade cities, while it is less pronounced in the central and western regions and low-grade cities. (3) The digital economy prompted extensive changes in the economy and society, guiding the gradual shift of resource-based industries toward high technology, high value-added, and low energy consumption. (4) Digital economic development empowers urban carbon reduction through green technology innovation. Innovation encourages efficient innovation factor allocation and accelerates the diffusion of innovation. This enables traditional industries to achieve decarbonization, cleanliness, and efficiency, thus effectively promoting energy conservation and emission reduction.

Based on the aforementioned conclusions, the following policy recommendations are proposed. Governments should base their strategies on the distinct characteristics of regional development and tailor their governance approaches to local conditions. Each city should establish digital industries with distinctive development features and leverage its own resource endowments to achieve coordinated digital economy development. Eastern regions and high-tier cities, as frontiers of digital economy development, should take on the crucial task of breaking through core and key technologies and contributing to continuous advancements in the digital economy and digital technology innovation. The central and western regions and low-tier cities should significantly increase their digital infrastructure, promote the digital transformation of traditional industries, and increase the cultivation of digital talent. Central cities should utilize their radiating and driving effects to foster better development of the digital economy in peripheral cities, thereby aiding carbon emission reduction. Non-resource-based cities should adhere to an innovation-driven green economic development model that uses the digital economy to reduce carbon emissions more effectively. Resource-based cities should seek to transform their economic development models, explore clean and efficient low-carbon production forms, and promote the digitalization and low-carbon transformation of industries.

This study argues that strengthening the construction of digital infrastructure and creating an environment is conducive to digital economy development. This involves investing in new types of infrastructure, such as the industrial internet and industrial big data centers, promoting the construction of digital platforms, and improving and innovating synergistic mechanisms for regional pollution control. Policymakers should aim for comprehensive digital transformation of the regional economy with a focus on applying digital technology in low-carbon areas to accelerate the realization of green and low-carbon transformation. Moreover, policymakers should aim to develop governance strategies according to regional development differences and leverage the role of information technology in shaping competition. This includes promoting interregional cooperation and exchange, enhancing the radiation-driven role of high-value-efficiency enterprises, and fostering local conditions for governance.

Policymakers should aim to enrich innovative energy utilization and tap into the emission reduction potential of the digital economy. This entails accelerating the integration of digital technologies with energy activities, such as power transmission, improving the accuracy of carbon emission monitoring and management, enhancing energy efficiency, creating an intelligent energy resource allocation platform, and cultivating digital consumption patterns to promote energy conservation and emission reduction. This also involves optimizing the energy consumption structure, accelerating the low-carbon transformation of cities, accelerating the transformation and upgrading of traditional industries, and establishing a long-term mechanism for low-carbon emission reduction. This includes encouraging the development of tertiary industries, increasing the proportion of high-tech industries and modern service industries in the industrial structure, and promoting the shift of China’s industrial structure to the middle and high ends. Emphasis should first be placed on quality and the formation of a new industrial system to avoid falling into a quantity-first and overcapacity expansion cycle. The application of new models and modes of operation should be promoted to guide the healthy development of these new modes.

We contend that enhancing the level of green technological innovation and steadily promoting a comprehensive low-carbon transformation will be beneficial. This involves building a robust policy-support system for green technological innovation and accelerating the integration and innovation of digital and green technologies. The focus should be on the roles of big data, cloud computing, and other digital intelligence technologies in promoting the development of key green and low-carbon technologies. Enhancing enterprises’ ability to share information and identify opportunities, guiding the pooling of innovative factors, grasping the direction of technological progress, improving production technology, and promoting green economic development. Policymakers can apply these policy implications to harness the potential of the digital economy to drive carbon emissions reduction and facilitate the transition toward a sustainable and low-carbon future.

We note that this study focuses on the impact of the digital economy on carbon emissions but pays insufficient attention to its spatial impact effects, which should be addressed in subsequent research, including the spatial spillover effects of the digital economy on carbon emissions. Compared with previous studies at the provincial level in China, this study, while refining the research scale to the city level, is limited by the continuity and availability of data and could not conduct a county-level analysis. Future research should conduct surveys to obtain more granular, detailed, and updated data for related studies. This study also discusses the impact of the digital economy on carbon emissions at the macro level. With the gradual improvement in industry classification standards for the digital economy, it is particularly important to objectively measure the effects of carbon emissions on different digital economy industries, which warrant further exploration in future research.