Unveiling the Catalytic Role of Digital Trade in China’s Carbon Emission Reduction under the Dual Carbon Policy

Abstract

Digital trade (DT), a key component of today’s digital economy, is pivotal in attaining “carbon neutrality and carbon peaking”, essential for low-carbon and high-quality growth. This study delves into the intermediary role of carbon emissions (CE) reduction in DT, analyzing both production and consumption angles, and examines the moderating influences of CE in DT through industrial agglomeration and low-carbon pilot policy. The research employs spatial panel and system GMM models for an empirical investigation. On the production side, the scale and technological effects on CE outweigh the structural impact on emissions. In terms of consumption, the mediating role of urban residents’ consumption upgrading is to enhance the effect of DT on reducing CE by promoting consumption upgrading, whereas the mediating role of rural residents’ consumption upgrading is to promote consumption upgrading but weaken the effect of DT on reducing CE. Regarding regulatory influences, the factor of industrial agglomeration tends to diminish the impact of DT on reducing CE; thus, industrial agglomeration does not amplify the reduction effect of DT on CE. Low-carbon pilot policy(pol)s can enhance the CE reduction effect of DT, showing stronger CE reduction effects in provinces participating in low-carbon pilot programs.

1. Introduction

Given the global imperative to address climate change, reducing carbon emissions (CE) has become a pivotal goal internationally. As a significant contributor to global emissions, accounting for nearly 30% of the total, the role of China in this sphere is particularly crucial [1]. The Chinese government has adopted a comprehensive “dual carbon” strategy, which targets achieving a carbon peak by 2030 and carbon neutrality by 2060 [2]. In this scenario, the impact and mechanisms through which digital technology can facilitate China’s transition to a low-carbon economy and refine its economic framework are explored. The exploration not only holds theoretical importance and enriches the policy implications of the “dual carbon” approach but also provides crucial insights for devising effective CE reduction strategies in other developing countries. The advancements in the digital economy and the fusion of digital technologies are instrumental in supporting the objectives of CE reduction.

The CE reduction mechanism from the perspective of trade has always been a prominent area of research, and the existing research is mainly based on the theory of sustainable development of trade [3,4] and the theory of trade benefits [5,6]. Digital trade (DT) is an outcome of both the digital revolution and globalization, and trade activities using digital technologies cover the digital purchase, sale, and delivery of goods and services. With the wide application of digital technology, digital technology has gradually sunk into the trade domain, promoting the steady growth of DT in China and around the world. Per United Nations Trade Statistics data, the scale of China’s service trade that can be digitally delivered in 2020 has reached USD 294.7 billion, an increase of 13.9 percentage points over the early 13th Five-Year Plan period, accounting for 44.5% of the total service trade. In the context of digitization, technological innovation provides the technical basis for DT and also brings digital transformation to traditional trade, but the essence of its trade has not changed. Digital technologies affect CE by automating and optimizing production processes to reduce energy consumption, efficiently managing and deploying renewable energy, and optimizing logistics and distribution route planning [7]. As the advancement of green technology, industrial restructuring, and the improvement of DT level lead to the reduction in CE, this form of trade combined with digital technology promotes the growth of a low-carbon economy [8]. A quintessential example in China is Alibaba, a leading force in the e-commerce and technology sectors, playing a pivotal role in the country’s landscape. By harnessing cutting-edge digital technologies, Alibaba has streamlined operational processes and caused substantial environmental benefits. Its digitalized logistics infrastructure utilizes intelligent routing systems to optimize delivery routes, thereby mitigating CE in transportation. Furthermore, Alibaba actively promotes eco-friendly consumption, advocates for the purchase of environmentally sustainable products, and facilitates the production and distribution of low-carbon goods. This instance underscores how DT and technological innovations can propel traditional industries toward a low-carbon transition, thereby making noteworthy contributions to global efforts in CE reduction. Hence, other nations and regions can glean insights from and adopt analogous strategies, especially amid endeavors toward tackling low-carbon economic development and addressing global climate change. So does DT help reduce CE? Firstly, it helps to broaden the relevant theories of trade and sustainable development; secondly, it helps to explore and implement CE reduction measures or policies of DT to fulfill the needs of green economic development and the sustainability of developing countries.

As a new economic growth point, DT is also becoming an important part of the global trading system. But what about the correlation between DT growth and its impact on CE? The analysis of its influence path has great research value. CE mainly come from industrial production and human consumption. In economic activities, human consumption behavior is the main driving force of production activities, and these production activities constitute the primary source of CE. From existing research, there is research on CE reduction from the perspective of industrial production, such as the use of fossil fuels in industrial production [9], CE from industrial intermediate links [6,10], and steel production [11], and research on CE reduction from the perspective of consumption, such as household natural gas consumption [12] and household-scale energy consumption [13]. Some research measures the CE intensity of consumers and producers through the CE responsibility sharing method [14], but few research studies consider production and consumption as a necessary link in the economic development system to study the CE reduction mechanism. In the production link, regional-scale output level [15], industrial structure [16], and technological progress [17] are important factors affecting carbon emission reduction. In the consumption demand, consumption upgrades naturally occur with rising incomes and changing consumption patterns [18]. And consumption upgrades will feed back into production, which will indirectly affect CE. Therefore, this research includes both production and consumption sides in the impact path of DT on CE reduction.

Moreover, might other factors influence the CE reduction effect of DT? Alfred Marshall (1890) first mentioned the related concept of industrial agglomeration in his book Principles of Economics; that is, similar or related firms tend to concentrate in specific areas, which brings economic benefits, including the specialization of the labor market, sharing of knowledge and technology, and convenience of supply chain. This agglomeration later became known as the “Marshall externality”. Since then, the “externality” of industrial agglomeration has been the research focus of industrial development. Among this research, investigations into how industrial concentration affects CE reduction from the standpoint of the low-carbon economy have yielded significant findings [19,20,21], but the research results are still controversial. Different from existing research, this research explores the impact of different forms of industrial agglomeration on CE and examines how the agglomeration of high-tech industries, producer service industries, and their combined synergistic effects regulate the influence of DT on CE reduction. Moreover, the low-carbon pilot policy aims to decrease greenhouse gas emissions and foster sustainable social, economic, and environmental development, aligning with China’s “dual carbon” objectives. China launched its carbon trading pilot policy in 2010, with implementation commencing after 2012. Research has shown that low-carbon pilot policies can markedly enhance the efficiency of CE reduction, with a positive impact extending to the surrounding area through spatial spillover effects [22,23]. Therefore, it is important to explore whether DT has a regulatory effect on CE reduction under the low-carbon pilot policy and if there is a correlation between low-carbon pilot policies and digital technology in reducing CE. This is a very significant problem amid ongoing investigations into how DT affects CE reduction.

In conclusion, scholars’ research on CE reduction at the DT level has achieved certain results. These research studies mainly discuss the following questions: (1) Is China’s DT level conducive to CE reduction? What is the precise operational mechanism? (2) Do various types of industrial agglomeration and low-carbon pilot policies moderate the influence of DT on CE reduction? Based on the existing research and the above issues, this research aims to thoroughly investigate China’s DT level’s CE reduction mechanism by combining the analysis framework of both China’s production and consumption sides.

The contributions of this research are outlined below: (1) It develops a metric system for assessing regional DT levels focusing on the aspects of digital innovation, factor endowment, infrastructure, innovation environment, and trade potential, performs measurements using the entropy method, and provides a reference for measuring the level of DT in other regions. (2) Analyzing the mechanism of DT in CE reduction from the production-side and consumption-side perspectives is beneficial for revealing the mediating role of different types of production and consumption variables in the impact of DT on CE reduction. In addition, the regulation effect of CE level on CE reduction is also discussed from different types of industrial agglomeration and low-carbon pilot policies. (3) Analyzing the effect of China’s DT level on CE reduction, in conjunction with the regulatory influence of industrial agglomeration and low-carbon pilot policies, can elucidate the role of DT in cutting CE. This provides valuable insights and policy recommendations for developing countries aiming to foster high-quality green economic growth through DT.

2. Mechanism and Research Hypothesis

2.1. DT and CE Reduction

This research explores the carbon reduction mechanism of DT based on mediating and regulating operations. With reference to the general equilibrium model of trade and environmental policy or environmental protection [24], this research discusses the mediating effect of DT on CE reduction from the production side and consumption side. With reference to the new trade theory proposed by Krugman (1979), this research discusses the regulating effect of DT on CE reduction considering industrial agglomeration and low-carbon pilot policy.

2.1.1. Production-Side Influence Mechanism

Considering the production side, the impact of DT on CE reduction mainly includes the scale effect, structure effect, and technology effect.

Research on traditional trade shows that the expansion of the economic or trade production scale usually leads to an increase in CE [25]. Although DT is essentially the same as traditional trade, it is different from traditional trade in participant involvement, transaction methods, geographical scope, cost and efficiency, personalization, data analysis, and policy orientation. Therefore, DT and traditional trade are different in terms of the scale effects of CE [26]. During the influence of DT on CE reduction, the scale effect is generated through two paths. First, DT relies on digital technology and an efficient logistics system to effectively broad sales reach and capacity, improve the efficiency of DT, and then expand the scale of trade production, resulting in an increase in CE [27]. Secondly, e-commerce, as an important part of DT, has given rise to the demand for digital service trade and digital goods, which reduce physical logistics and transportation demand, reduce resource waste, expand the scale of trade, but do not increase CE, and decrease reliance on high-carbon energy [28].

Based on the above analysis, this research proposes Hypothesis 1.

(H1): 

There is a scale effect in the impact of DT on CE reduction, which plays an intermediary role in CE and may have a positive or negative impact.

Structural effects of DT on CE reduction. DT will enhance the optimization and modernization of traditional industries’ structure and drive the digital transformation of traditional productive industries through a positive impact so that the original high-carbon business model will gradually transform into a low-carbon business model [29]. Firstly, DT necessitates numerous data centers to facilitate online communication and data storage. To lower CE, numerous data centers often operate on renewable energy, aiding in curbing the CE footprint of the power grid. Moreover, DT is also driving data centers to become more energy efficient, reducing energy waste by adopting more efficient cooling systems and server management technologies [30]. Thirdly, DT drives digital transformation in traditional industries, such as supply chain optimization and the use of digital technology and Internet of Things technology to optimize transportation and inventory. Fourthly, DT is driving the popularity of telecommuting and digital production in productive services, and enterprises are increasingly turning to cloud computing, virtual collaboration tools, and online project management to support remote working and global team collaboration, which will reduce employee commuting needs and business travel, thereby reducing transport-related CE [31].

Based on the above analysis, this research proposes research Hypothesis 2.

(H2): 

In the process of the influence of DT on CE reduction, the structural effect is positive.

Technological effects of DT on CE reduction. There are significant differences between DT and traditional trade in the technical effects of CE reduction. DT significantly reduces CE by reducing physical transport needs and optimizing supply chain management. In China, the adoption of an emissions trading scheme can effectively reduce overall CE, initially through reducing coal consumption and optimizing the energy structure, and alternatively by relying on green technology innovation [32]. DT also contributes indirectly to carbon reduction by driving technological innovation and adoption. Some research studies have shown that institutional opening significantly inhibits China’s CE, and this effect has heterogeneity and spatial spillover effects. Then, they further found that institutional opening actually achieves CE reduction through technological progress. In addition, there is technology trade in DT, which helps to reduce energy consumption and carbon emissions by promoting more energy-efficient technologies and products, such as energy-saving equipment and renewable energy technologies, in the global market [33].

Based on the above analysis, this research proposes research Hypothesis 3.

(H3): 

In the process of the impact of DT on CE reduction, the technology effect is positive.

2.1.2. Heterogeneous Side

Different regions exhibit differences in terms of economic development levels, infrastructure, energy structures, and policy environments, which lead to variations in the impact of DT on CE across different regions. High-income regions typically have more mature digital economy infrastructure and stricter environmental regulations. This likely makes DT’s impact on CE more significant in these areas, as it can reduce CE by improving logistics efficiency and reducing unnecessary transportation [28]. Regional differences in logistics, digital communication, and transportation infrastructure mean that developed regions may more easily harness the advantages of DT, such as using advanced logistics systems to reduce CE, while underdeveloped regions may be unable to fully utilize the carbon reduction potential of DT due to their inadequate infrastructure [34]. In areas with a higher prevalence of renewable energy, the energy consumption driven by DT has a relatively small impact on CE. However, in regions with high fossil fuel usage, digital-trade-induced energy consumption can have a significant negative impact on CE. Different regions have varying policy support for CE and digital economy development [35]. Some regions may implement incentives to promote low-carbon development, thus allowing DT to play a more positive role in reducing CE [36]. Therefore, due to differences in economy, infrastructure, energy structures, and policy environments, the influence of DT on CE exhibits regional heterogeneity. DT may play a pivotal role in reducing CE, while in others, this impact may be relatively weak.

Based on the above analysis, this research proposes research Hypothesis 4.

(H4): 

DT exhibits regional heterogeneity in its impact on CE, with different regions showing varying impacts due to regional differences.

2.1.3. Consumption-Side Influence Mechanism

The consumption effect of DT on CE reduction. DT promotes consumer awareness and acceptance of environmentally friendly and low-carbon products by providing wider information and more efficient access to markets. This consumption upgrade is not only reflected in the preference for efficient energy and renewable energy products but also in the pursuit of energy conservation and low-carbon lifestyles. For example, digital platforms make it easier for consumers to access information about a product’s carbon footprint and environmental attributes, enabling them to make more environmentally responsible purchasing decisions [37]. Second, DT promotes a shift in consumption patterns. With the popularity of e-commerce and online services, consumers are increasingly opting for digital products and services, which typically have lower CE. For example, digital entertainment and online education reduce the need for physical transportation, thus lowering CE [38]. In addition, DT also promotes the participation of consumers in a sustainable and circular economy and promotes the upgrading of green consumption [39]. Through online platforms, consumers can more easily buy used goods, participate in the sharing economy, or choose products made from recycled and recycled materials. These actions not only reduce resource consumption but also help reduce waste and CE. Consumer demand for low-carbon products and services has in turn promoted green innovation in production and supply chain management [40]. In order to meet market demand, enterprises are increasingly adopting energy-saving technologies and sustainable production methods, which further promote the transition of the entire economic system to low carbon.

Based on the above analysis, this research proposes research Hypothesis 5.

(H5): 

In the process of the influence of DT on CE reduction, the effect of consumption upgrading is positive.

2.2. The Regulatory Role of Industrial Agglomeration and Low-Carbon Pilot Policy

2.2.1. The Regulatory Role of Industrial Agglomeration

Industrial agglomeration occurs where numerous related or similar industries and service providers cluster within a specific geographical area. This clustering is not merely a quantitative increase but reflects significant interaction and synergistic effects among industries, often referred to as “specialized agglomeration” [41]. Due to the inherent characteristics of different industries, there are notable differences in their CE levels. For example, specialized agglomerations in high-tech and green technology sectors, such as Silicon Valley, typically exhibit lower CE due to their focus on high-tech industries, showcasing reduced CE traits [40]. Moreover, specialized agglomeration often fosters technological innovation, including the advancement of clean and low-carbon technologies. These innovations can lead to more environmentally friendly production methods and more efficient resource use, thereby helping to reduce overall CE [42]. Under the policy framework promoting coordinated regional economic development in China, the spatial layout of industries tends to cluster [43]. This agglomeration not only stimulates regional economic growth but also enhances environmental protection to some extent. This suggests that industrial agglomeration may exert a positive moderating influence in the impact of DT on CE.

Based on the above analysis, this research proposes research Hypothesis 6.

(H6): 

In the process where DT impacts CE, industrial agglomeration has a positive moderating effect.

2.2.2. The Regulatory Role of Low-Carbon Pilot Policies

China is an active implementer of pilot low-carbon policy, and since late 2010, the Chinese government launched pilot low-carbon initiatives in various provinces and cities, including building low-carbon cities and promoting clean energy and energy efficiency improvements. China has also issued the “Carbon Market Management Measures” and established a CE trading market to encourage companies to reduce CE. Compared with environmental regulation tools, low-carbon pilot policy exerts a substantial influence on controlling CE. For example, Huo et al. [42] through the Dual Difference-in-Differences (Dual DID) method, confirmed that the first and second batches of low-carbon pilot policies can significantly reduce the level of urban CE, but the policy shows short-term impact and lacks significant long-term effects. Considering regional heterogeneity, Wang et al. [18] segmented the region into resource-dependent and non-resource-dependent cities for research and found that the CE reduction effect on non-resource-based cities was more significant. Therefore, this research believes that low-carbon pilot policies contribute to the CE of DT.

Based on the above analysis, this research proposes research Hypothesis 7.

(H7): 

Companies from provinces that participated in the pilot low-carbon policies program experienced a greater reduction in carbon emissions, due to digital trading, than companies in other provinces.

3. Research Methodology

3.1. Measurement Model

3.1.1. Dynamic Panel Model

The research employs the Generalized Method of Moments (GMM) for empirical analysis. Utilizing the GMM approach allows for better capture of the dynamic features of the data and the relationships between variables, while effectively addressing endogeneity issues. The inclusion of lagged dependent variables as independent variables is not intended to control for endogeneity but rather to capture the temporal evolution of CE. According to the research by Akbar, Poletti-Hughes et al. (2016) [44], the inclusion of lagged dependent variables helps to describe the dynamic changes in the dependent variable over time. Given the potential endogeneity problems introduced by lagged dependent variables, Ordinary Least Squares (OLS) may produce biased and inconsistent estimates. Therefore, this research uses the system GMM method for estimation. The system GMM method can use the lagged values of endogenous variables as instruments, thus addressing endogeneity issues and providing consistent estimates. CE may have a certain path-dependent effect, so this research constructs a dynamic panel model with lagging CE.

$$\ln \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_1\ln \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}-1}+{\mathsf{\beta }}_2\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$$

(1)

In the formula, $\mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}}$ represents the carbon emissions of the i province in the t year; $\ln \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}}$ represents logarithmic treatment of carbon emissions; $\ln \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}-1}$ represents first-order logarithmic treatment of lagging carbon emissions, capturing the path-dependent effect; $\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}$ represents the level of digital trade in year t of the i province; ${\mathrm{X}}_{\mathrm{i}\mathrm{t}}$ represents the control variable; ${\mathsf{\mu }}_{\mathrm{i}}$ represents the location fixed effect; ${\mathsf{\lambda }}_{\mathrm{t}}$ represents time fixed effect; and ${\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}$ represents random interference term.

3.1.2. Mediation Model

As in the above theoretical analysis, DT has a certain impact on the production side and the consumption side of CE reduction, this research constructs the mediation effect model of the production side and sets it as Formula (2):

$$\begin{array}{cc}& \mathrm{O}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_1\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}\\ & \mathrm{I}{\mathrm{G}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_1\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}\\ & \mathrm{T}{\mathrm{C}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_1\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}\\ & \mathrm{l}\mathrm{n} \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_1\mathrm{l}\mathrm{n} \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}-1}+{\mathsf{\beta }}_2\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3\mathrm{O}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4\mathrm{I}{\mathrm{G}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_5\mathrm{T}{\mathrm{C}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_6{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}\end{array}$$

(2)

In the formula, OT, IG, and TC represent mediating variables on the production side, and they represent regional output level, advancement of industrial structure, and technological progress, respectively. These variables are posited to mediate the relationship between DT and CE, offering a pathway through which DT can influence CE on the production side. To assess the existence of mediation effects on the production side, this research employs the Sobel test for verification [45].

The mediation effect model on the consumption side is set as Formula (3):

$$\begin{array}{cc}& \mathrm{R}{\mathrm{U}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_1\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}\\ & \mathrm{C}{\mathrm{U}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_1\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}\\ & \mathrm{l}\mathrm{n} \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_1\mathrm{l}\mathrm{n} \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}-1}+{\mathsf{\beta }}_2\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3\mathrm{R}{\mathrm{U}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_4\mathrm{R}{\mathrm{C}\mathrm{U}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_5{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}\end{array}$$

(3)

In the formula, RU and CU represent mediating variables on the production side, and they represent rural consumer upgrading and urban consumer upgrading, respectively. To assess the existence of mediation effects on the consumption side, this research employs the Sobel test for verification.

3.1.3. Moderation Model

This research further analyzes the regulatory role of various forms of industrial agglomeration and low-carbon pilot policy in the impact mechanism of DT on CE. To construct a regulatory effect model, see Formula (4), which includes various industrial agglomeration variables, virtual variables of low-carbon pilot policies, interaction terms between DT and various industrial agglomeration, and virtual variables of DT and low-carbon pilot policy.

$$\begin{array}{rr}& \mathrm{l}\mathrm{n} \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_1\mathrm{l}\mathrm{n} \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}-1}+{\mathsf{\beta }}_2\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}\times \mathrm{agg}+{\mathsf{\beta }}_4\mathrm{a}\mathrm{g}\mathrm{g}+{\mathsf{\beta }}_5{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}\\ & \mathrm{l}\mathrm{n} \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}}={\mathsf{\beta }}_1\mathrm{l}\mathrm{n} \mathrm{C}{\mathrm{E}}_{\mathrm{i}\mathrm{t}-1}+{\mathsf{\beta }}_2\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\beta }}_3\mathrm{D}{\mathrm{T}}_{\mathrm{i}\mathrm{t}}\times \mathrm{pol}+{\mathsf{\beta }}_4\mathrm{pol}+{\mathsf{\beta }}_5{\mathrm{X}}_{\mathrm{i}\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\lambda }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i}\mathrm{t}}\end{array}$$

(4)

In the formula, agg represents industrial agglomeration, it is mainly divided into high-tech industry agglomeration, productive service industry agglomeration, and industrial synergy agglomeration; pol represents low-carbon pilot policy.

3.2. Variable Selected

3.2.1. Explanatory Variables

This research adopts the IPCC (the Intergovernmental Panel on Climate Change) method for calculating CE, using precise emission factors and energy consumption data to assess direct emissions from various sources. Unlike carbon footprint assessments, which include indirect emissions, the IPCC method comprehensively addresses both direct and wider emissions, making it ideal for policy development and international compliance [46]. Moreover, the consistently updated and peer-reviewed emission factors and calorific values ensure accuracy and reliability. Thus, the IPCC approach enhances the ability to monitor and compare emissions across different times and regions, providing a strong basis for policy analysis [47].

Explained variable CE from energy consumption. Since China’s CE from energy consumption account for a relatively high proportion, it is more representative to choose the CE from energy usage. Consulting Wu et al.’s research [48], energy consumption CE are calculated using the formulas in the IPCC Guidelines:

$${\mathrm{C}\mathrm{E}}_{\mathrm{i}}={\sum }_{\mathrm{j}=1}^{10} {\mathrm{M}}_{\mathrm{i}\mathrm{j}\mathrm{t}}{\mathrm{Q}}_{\mathrm{j}}{\mathrm{C}}_{\mathrm{j}}$$

(5)

In the formula, ${\mathrm{C}\mathrm{E}}_{\mathrm{i}}$ represents CE of the i province, ${\mathrm{M}}_{\mathrm{i}\mathrm{j}\mathrm{t}}$ represents the j type of the i province (j = 1,2, …, 10, respectively, represents the consumption of coal, coke, crude oil, gasoline, kerosene, diesel, fuel, liquefied petroleum gas, natural gas, and electricity) energy in t year, ${\mathrm{Q}}_{\mathrm{j}}$ represents the net calorific value of the j energy, and ${\mathrm{C}}_{\mathrm{j}}$ represents the CO₂ emission factor of the j energy. The data are derived from the annual China Energy Statistical Yearbook and the IPCC Guidelines.

3.2.2. Primary Explanatory Variables

Digital trade level (DT). It can gauge a region’s maturity and development in DT, and its measurement usually takes into account multiple indicators, which should be comprehensively considered and evaluated in combination with the specific situation of the region. Currently, the measurement of the level of DT is mainly divided into two aspects: One is the analysis of DT statistics, such as reflecting the level of DT from the regional digital trade volume, the types of DT products, and the growth trend of DT [49,50], but the statistical data are more concentrated on the national and industrial levels. The second is to develop the index of DT level and put forward a new index or index system for measuring DT level [51]. Based on this, this research focuses on the construction of a new DT level measurement index system, starting from China’s provincial data, following the principles of objectivity, effectiveness, hierarchy, comprehensiveness, and scientific index system construction, and using existing studies to build a comprehensive index system of DT level. It covers 5 first-level indicators, including digital innovation, infrastructure environment, technological innovation environment, DT capability, and trade potential, and 25 s-level subdivision indicators (see

Table 1. DT level measurement indicator system.

Primary Indicators	Secondary Indicators	Tertiary Indicators	Unit	Direction of Indicators	Indicator Symbols
Digital Innovation	Innovation outputs	Publication of scientific and technical papers	article	+	x1
		Number of patent applications granted	case	+	x2
	Innovative inputs	Research and experimental development investment intensity	%	+	x3
	Innovation incubation	Number of technology business incubators	unit	+	x4
Infrastructure environment	Network infrastructure	Internet broadband access port	ten thousand	+	x5
		Number of websites owned by enterprises	ten thousand	+	x6
		Length of long-distance fiber-optic cable routes	kilometer	+	x7
		Number of domain names		+	x8
		Mobile Internet user	ten thousand household	+	x9
	Logistics and transport environment	Traffic Operational Capacity Index	%	+	x10
		Ownership of road-operating goods vehicles	ten thousand vehicles	+	x11
		Logistics and transport practitioners	ten thousand people	+	x12
Technological innovation environment	Scientific and technological input	Research and experimental development expenditure of industrial enterprises above designated size	ten million CNY	+	x13
		Full-time equivalent of research and experimental development personnel in industrial enterprises above designated size	people/year	+	x14
	Scientific and technical outputs	Domestic patent applications received	term (in a mathematical formula)	+	x15
		Technology market turnover	billion	+	x16
Digital trade capacity	Digital industrialization trade	Total telecommunication services	billion	+	x17
		Revenue from software operations	billion	+	x18
		Income from information technology services	billion	+	x19
	Industrial digital trade	Number of enterprises with e-commerce trading activities	unit	+	x20
		E-commerce sales	billion	+	x21
		E-commerce purchases	billion	+	x22
Trade potential	Level of openness	Trade openness	%	+	x23
	Residential consumer potential	Consumption expenditure per inhabitant	billion	+	x24
		Total retail sales of consumer goods	billion	+	x25
		per capita GDP	CNY/year	+	x26

). The entropy method was used to measure DT levels for 30 Chinese provinces (excluding Tibet, Hong Kong, Macao, and Taiwan due to data limitations).

Table 2. Partial sample of DT levels in 30 provinces of China.

Province	2013 DT Level	2022 DT Level	Mean DT Level 2013–2022	Province	2013 DT Level	2022 DT Level	Mean DT Level 2013–2022
Beijing	0.270	0.614	0.425	Henan	0.123	0.183	0.171
Hebei	0.106	0.153	0.149	Hunan	0.133	0.182	0.141
Tianjin	0.104	0.161	0.133	Guangdong	0.329	0.820	0.558
Zhejiang	0.189	0.445	0.305	Guangxi	0.074	0.126	0.078
Inner Mongolia	0.068	0.085	0.081	Hubei	0.101	0.238	0.149
Shanghai	0.232	0.462	0.310	Hainan	0.031	0.132	0.061
Fujian	0.051	0.227	0.171	Chongqing	0.067	0.170	0.112
Jiangxi	0.073	0.114	0.109	Yunnan	0.069	0.074	0.088
Jilin	0.109	0.057	0.072	Guizhou	0.033	0.092	0.062
Anhui	0.110	0.204	0.172	Sichuan	0.113	0.224	0.200
Heilongjiang	0.076	0.102	0.087	Shanxi	0.097	0.163	0.140
Jiangsu	0.321	0.628	0.453	Gansu	0.032	0.080	0.059
Liaoning	0.118	0.162	0.140	Qinghai	0.074	0.031	0.036
Fujian	0.100	0.227	0.171	Ningxia	0.033	0.066	0.044
Shandong	0.213	0.466	0.307	Xinjiang	0.061	0.067	0.048

Evaluations based on relevant data by entropy weight method.

displays DT levels for 2013 and 2022, in conjunction with the average DT levels from 2013 to 2022. DT Level 2013/2022: DT levels for each province in 2013 and 2022. Mean DT Level 2013–2022: Average DT level across the 2013–2022 period for each province.

3.2.3. Control Variables

According to pertinent literature [10,52], this research identifies the following control variables: Industrial Structure (INS); Urbanization Level (UL); Degree of Openness (OP); Research and Experimental Development Intensity (RD); Innovation Level (INO); Environmental Regulation Intensity (ER); and Economic Development Level (EC). Indicator descriptions can be found in Table 3.

3.2.4. Mediator Variables

(1)

Production-side variables

Main selection of production side variables: Regional Output Level (OT); Advancement of Industrial Structure (IG); Technological Progress (TC).

(2)

Consumption-side variables

Main selection of consumption-side variables: Rural Consumer Upgrading (RU); Urban Consumer Upgrading (CU).

The above indicator descriptions are provided in Table 3.

3.2.5. Moderator Variables

(1)

Industrial agglomeration variables

This research employs location entropy to gauge the industrial agglomeration index. Referring to Yang et al.’s research [53], the formula for calculating the high-tech industry agglomeration index is as follows:

$$\mathrm{h}\mathrm{a}{\mathrm{g}\mathrm{g}}_{\mathrm{i}}={\mathrm{M}\mathrm{a}\mathrm{x}}_{\mathrm{j}} \left({\mathrm{S}}_{\mathrm{j}\mathrm{i}}/{\mathrm{S}}_{\mathrm{i}}\right)$$

(6)

In the formula, $\mathrm{h}\mathrm{a}{\mathrm{g}\mathrm{g}}_{\mathrm{i}}$ represents the degree of agglomeration within high-tech industries. By calculating the regional concentration degree of each high-tech industry in a province and comparing it with the concentration degree of all high-tech industries in the country, the maximum value is selected to represent the high-tech industry specialization concentration index of the province at this time point. ${\mathrm{S}}_{\mathrm{j}\mathrm{i}}$ represents the proportion of the number of employments in j industry in i province to the number of employments in high-tech industry;${\mathrm{S}}_{\mathrm{i}}$ represents the proportion of j industry employment in high-tech industry in China. For the calculation of high-tech industry agglomeration, high-tech industry subdivision industry is selected. Based on the categorization of high-tech industries in the High-tech Industry Statistical Yearbook, data from five high-tech industries are chosen, including pharmaceutical manufacturing, spacecraft and equipment manufacturing, electronic and communication equipment manufacturing, computer and office equipment manufacturing, medical equipment, and instrumentation manufacturing.

With reference to Huang et al.’s research [54], the formula for calculating the productive service industry agglomeration index is as follows:

$$\mathrm{p}\mathrm{a}{\mathrm{g}\mathrm{g}}_{\mathrm{i}}={\mathrm{L}}_{\mathrm{j}\mathrm{i}}/{\mathrm{L}}_{\mathrm{i}}$$

(7)

In the formula, ${\mathrm{L}}_{\mathrm{j}\mathrm{i}}$ represents the proportion of employment in productive services j in province i relative to regional employment.${\mathrm{L}}_{\mathrm{i}}$ represents the proportion of productive services in China’s employment.

With reference to He et al.’s research [55], the industrial synergy agglomeration of high-tech industry and producer service industry is calculated as follows:

$$\mathrm{c}\mathrm{a}\mathrm{g}{\mathrm{g}}_{\mathrm{i}}=1-\left|\frac{\mathrm{h}\mathrm{a}\mathrm{g}{\mathrm{g}}_{\mathrm{i}}-\mathrm{p}\mathrm{a}\mathrm{g}{\mathrm{g}}_{\mathrm{i}}}{\mathrm{h}\mathrm{a}\mathrm{g}{\mathrm{g}}_{\mathrm{i}}+\mathrm{p}\mathrm{a}\mathrm{g}{\mathrm{g}}_{\mathrm{i}}}\right|+\mathrm{h}\mathrm{a}\mathrm{g}{\mathrm{g}}_{\mathrm{i}}+\mathrm{p}\mathrm{a}\mathrm{g}{\mathrm{g}}_{\mathrm{i}}$$

(8)

In the formula, $\mathrm{c}\mathrm{a}\mathrm{g}{\mathrm{g}}_{\mathrm{i}}$ represents the degree of industrial synergistic agglomeration between the high-tech industry and the production service industry and considers the synergistic effect between the two different industries.

(2)

Low-carbon pilot policy variable

Low-carbon pilot policy is the dummy variable of this research. The pilot low-carbon policy was launched in 2010 at the earliest, and the sample period of this research is from 2013 to 2021 (Shenzhen is a special zone, not included in Guangzhou Province). The dummy variable of the pilot provinces is set at 1; otherwise, it remains at 0. For example, if the low-carbon pilot policy in Shanghai started in 2012, the dummy variable of Shanghai after 2013 is 1; Ningxia started the pilot in 2017, so the dummy variable is 0 from 2013 to 2016 and 1 from 2017 and beyond.

3.2.6. Descriptive Statistics

In this research, panel data from 30 provinces in China (excluding Tibet, Hong Kong, Macao, and Taiwan) spanning from 2013 to 2021 are utilized. The primary sources include the EPS database, China Energy Statistical Yearbook, China Statistical Yearbook, and China Trade and Foreign Economic Statistical Yearbook, among others. Data in this paper were collected manually.

Table 3 presents the definitions and descriptions of each variable. The mean of LnCE is 10.463 with a standard deviation of 0.761, the minimum value is 8.616, and the maximum value is 12.274, showing significant differences in carbon emissions across provinces. The mean of DT is 0.164 with a standard deviation of 0.137, the minimum value is 0.019, and the maximum value is 0.820, indicating a large disparity in digitalization levels among provinces. Control variables such as ER, LnEC, and HR also exhibit variability to varying degrees. The descriptive statistics for mediating variables (LnOT, IG, TC, RU, CU) show moderate to significant variability in other investments, industrial agglomeration, technological capabilities, and urban–rural development levels across provinces. The descriptive statistics for moderating variables (hagg, pagg, cagg, pol) reveal significant differences and variability in population, industrial, commercial agglomeration, and policy factors among provinces.

Table 4 displays the descriptive statistics for each variable. It is worth noting that throughout the analysis, all available observations were utilized, and no data were excluded due to unavailability.

From the significant correlations between LnCE and various variables, it can be seen that LnCE is significantly positively correlated with DT. This indicates that provinces with higher DT levels have higher CE, possibly because the initial stages of digital trade require substantial energy and resource inputs, leading to increased CE. Additionally, it may reflect that high levels of digital transformation bring about increased economic activities, resulting in higher CE. LnCE is significantly negatively correlated with INS, indicating that provinces with better infrastructure have relatively lower carbon emissions. This is because improved infrastructure helps enhance energy efficiency and reduce waste, thereby lowering CE. LnCE is significantly positively correlated with LnOT, suggesting that provinces with higher other investments have higher carbon emissions. This is because these investments often involve substantial industrial and infrastructure construction, which typically leads to increased carbon emissions. LnCE is significantly negatively correlated with IG, indicating that provinces with higher levels of industrial agglomeration have relatively lower CE. This is because industrial agglomeration helps optimize resource allocation and improve production efficiency, thereby reducing CE. LnCE is significantly positively correlated with TC, indicating that provinces with higher technical capabilities also have higher CE. This is because high technical capabilities are often accompanied by extensive industrial activities and energy consumption, thus increasing CE.

From the significant correlations between DT and various variables, it can be seen that DT is significantly positively correlated with INS, indicating that provinces with better infrastructure have higher levels of digital transformation. This is because improved infrastructure provides the necessary support and conditions for digital transformation. DT is significantly positively correlated with UL, indicating that provinces with higher urbanization levels also have higher levels of digital transformation. This is because urbanization provides more resources and market demand, promoting digital transformation. DT is significantly positively correlated with OP, indicating that provinces with higher degrees of openness have higher levels of digital transformation. This is because openness facilitates the exchange of technology and knowledge, thereby promoting digital transformation. DT is significantly positively correlated with hagg, indicating that provinces with higher levels of high-tech industry agglomeration have higher levels of digital trade. This is because high-tech industry agglomeration helps drive technological innovation and diffusion, promoting digital transformation.

From the significant correlations between other variables, it can be seen that INS is significantly positively correlated with UL, indicating that provinces with higher levels of urbanization have better infrastructure. This is because the urbanization process requires substantial infrastructure construction. INS is significantly positively correlated with OP, indicating that provinces with higher degrees of openness have better infrastructure. This is because openness facilitates the inflow of capital and technology, promoting infrastructure construction. UL is significantly positively correlated with RD, indicating that provinces with higher levels of urbanization also have higher R&D investment. This may be because urbanized areas have more resources and market demand, promoting R&D activities. ER is significantly negatively correlated with DT, indicating that provinces with higher environmental regulation intensity have lower levels of digital transformation. This is because stringent environmental regulations may limit the construction of high-energy-consuming digital equipment and infrastructure. INS is significantly positively correlated with OP, indicating that provinces with higher degrees of openness have better infrastructure. This is because openness facilitates the inflow of capital and technology, promoting infrastructure construction. INS is significantly positively correlated with RD, indicating that provinces with better infrastructure also have higher R&D investment. This is because a good infrastructure environment helps attract R&D investment. INS is significantly positively correlated with LnEC, indicating that provinces with better infrastructure have higher energy consumption. This is because infrastructure construction and maintenance require substantial energy. INS is significantly positively correlated with IG, indicating that provinces with better infrastructure also have higher levels of industrial agglomeration. This is because good infrastructure can attract and support more industrial agglomeration. UL is significantly positively correlated with RD, indicating that provinces with higher levels of urbanization also have higher R&D investment. This is because urbanized areas have more resources and market demand, promoting R&D activities. UL is significantly positively correlated with INO, indicating that provinces with higher levels of urbanization have stronger innovation capabilities. This is because urbanization provides more resources, talent, and markets, promoting innovation activities. OP is significantly positively correlated with RD, indicating that provinces with higher degrees of openness have higher R&D investment. This is because openness facilitates technology exchange and cooperation, increasing R&D investment. RD is significantly positively correlated with INO, indicating that provinces with higher R&D investment have stronger innovation capabilities. This is because R&D investment directly drives technological innovation and the conversion of research results. RD is significantly positively correlated with LnEC, indicating that provinces with higher R&D investment also have higher energy consumption. This is because R&D activities require substantial energy support. RD is significantly positively correlated with HR, indicating that provinces with higher R&D investment also have higher human resource levels. This is because R&D activities require high-quality human resources. RD is significantly positively correlated with IG, indicating that provinces with higher R&D investment have higher levels of industrial agglomeration. This is because industrial agglomeration can share R&D resources and promote R&D activities. RD is significantly positively correlated with TC, indicating that provinces with higher R&D investment have stronger technical capabilities. This is because R&D investment directly enhances technical levels. INO is significantly positively correlated with LnEC, indicating that provinces with stronger innovation capabilities have higher energy consumption. This is because innovation activities require substantial energy support. INO is significantly positively correlated with HR, indicating that provinces with stronger innovation capabilities have higher human resource levels. This is because innovation activities require high-quality human resources. ER is significantly negatively correlated with DT, indicating that provinces with higher environmental regulation intensity have lower levels of DT. This is because stringent environmental regulations may limit the construction of high-energy-consuming digital equipment and infrastructure. LnEC is significantly positively correlated with HR, indicating that provinces with higher energy consumption have higher human resource levels. This is because regions with high energy consumption are usually accompanied by high levels of industrial and economic activity, attracting more human resources. IG is significantly positively correlated with UL, indicating that provinces with higher levels of urbanization have higher levels of industrial agglomeration. This is because urbanization provides more resources and markets, promoting industrial agglomeration. IG is significantly positively correlated with OP, indicating that provinces with higher degrees of openness have higher levels of industrial agglomeration. This is because openness attracts more industries and investments, promoting industrial agglomeration. TC is significantly positively correlated with RD, indicating that provinces with higher technical capabilities also have higher R&D investment. This is because enhancing technical capabilities requires substantial R&D investment. TC is significantly positively correlated with INO, indicating that provinces with higher technical capabilities also have stronger innovation capabilities. This is because enhancing technical capabilities is accompanied by increased innovation activities. CU is significantly positively correlated with RU, indicating that provinces with higher levels of urban development also have higher levels of rural development. This is because urban development drives the economic and infrastructure development of rural areas. hagg is significantly positively correlated with DT, indicating that provinces with higher levels of high-tech industry agglomeration have higher levels of digital transformation. This is because high-tech industry agglomeration helps drive technological innovation and diffusion, promoting DT development. hagg is significantly positively correlated with OP, indicating that provinces with higher degrees of openness have higher levels of high-tech industry agglomeration. This is because openness attracts more high-tech industries and investments. pagg is significantly positively correlated with UL, indicating that provinces with higher levels of urbanization have higher levels of productive service industry agglomeration. This is because urbanization provides more resources and markets, promoting the agglomeration of productive service industries. cagg is significantly positively correlated with DT, indicating that provinces with higher levels of comprehensive agglomeration have higher levels of DT. This is because comprehensive agglomeration promotes technological innovation and diffusion, driving DT development. cagg is significantly positively correlated with IG, indicating that provinces with higher levels of comprehensive agglomeration have higher levels of industrial agglomeration. This is because comprehensive agglomeration attracts more industries and investments, promoting industrial agglomeration. pol is significantly negatively correlated with LnCE, indicating that provinces with higher participation in low-carbon policies have lower carbon emissions. This is because the implementation of low-carbon policies effectively reduces carbon emissions.

Moreover, the correlation coefficients for the variables in Table 5 are all generally below 0.50, indicating that there is no multicollinearity among the variables in this research. Through the above economic interpretations of significant correlations, a better understanding of the relationships between various variables and their economic significance can be achieved. These interpretations help support the hypotheses and conclusions in the research and provide a basis for further policy recommendations.

Table 3. Symbols and definitions of variables.

Variable Name	Indicator Symbol	Definition
Carbon Emissions	CE	Calculated according to the formulas listed in the IPCC Guidelines
Digital Trade	DT	Calculated using the entropy weight method
Industrial Structure	INS	The ratio of tertiary industry added value to secondary industry added value
Urbanization Level	UL	Urban population proportion
Degree of Openness	OP	The ratio of total imports and exports to GDP
Research and Experimental Development Intensity	RD	Internal expenditure on research and experimental development as a percentage of GDP
Innovation Level	INO	Logarithm of the count of domestic patent applications accepted
Environmental Regulation Intensity	ER	Investment completed in industrial pollution control as a percentage of the added value of the secondary industry
Economic Development Level	EC	Per capita GDP
Human Capital Level	HR	The proportion of university students to total population
Regional Output Level	OT	Fixed capital stock (based on Zhang Jun’s perpetual inventory method [56], with the base year as 2000 and a depreciation rate of 9.6%, the unit of this variable is ten thousand CNY)
Advancement of Industrial Structure	IG	The proportion of tertiary industry’s added value to the secondary industry
Technological Progress	TC	Domestic patents granted per 10,000 people
Rural Consumer Upgrading	RU	The proportion of four types of rural consumer spending to total consumer spending
Urban Consumer Upgrading	CU	The proportion of four types of urban consumer spending to total consumer spending
High-tech Industry Agglomeration	hagg	Calculated according to the above formula
Productive Service Industry Agglomeration	pagg	Calculated according to the above formula
Industrial Synergy Agglomeration	cagg	Calculated according to the above formula
Low-carbon Pilot Policy	pol	Set to 1 for provinces that are pilot areas in the current and subsequent years, otherwise 0.

Table 3. Symbols and definitions of variables.

Variable Name	Indicator Symbol	Definition
Carbon Emissions	CE	Calculated according to the formulas listed in the IPCC Guidelines
Digital Trade	DT	Calculated using the entropy weight method
Industrial Structure	INS	The ratio of tertiary industry added value to secondary industry added value
Urbanization Level	UL	Urban population proportion
Degree of Openness	OP	The ratio of total imports and exports to GDP
Research and Experimental Development Intensity	RD	Internal expenditure on research and experimental development as a percentage of GDP
Innovation Level	INO	Logarithm of the count of domestic patent applications accepted
Environmental Regulation Intensity	ER	Investment completed in industrial pollution control as a percentage of the added value of the secondary industry
Economic Development Level	EC	Per capita GDP
Human Capital Level	HR	The proportion of university students to total population
Regional Output Level	OT	Fixed capital stock (based on Zhang Jun’s perpetual inventory method [56], with the base year as 2000 and a depreciation rate of 9.6%, the unit of this variable is ten thousand CNY)
Advancement of Industrial Structure	IG	The proportion of tertiary industry’s added value to the secondary industry
Technological Progress	TC	Domestic patents granted per 10,000 people
Rural Consumer Upgrading	RU	The proportion of four types of rural consumer spending to total consumer spending
Urban Consumer Upgrading	CU	The proportion of four types of urban consumer spending to total consumer spending
High-tech Industry Agglomeration	hagg	Calculated according to the above formula
Productive Service Industry Agglomeration	pagg	Calculated according to the above formula
Industrial Synergy Agglomeration	cagg	Calculated according to the above formula
Low-carbon Pilot Policy	pol	Set to 1 for provinces that are pilot areas in the current and subsequent years, otherwise 0.

Table 4. Descriptive statistics for each variable.

Variables	Observations	Mean	Standard Deviation	Min	Max	VIF
LnCE	300	10.463	0.761	8.616	12.274
DT	300	0.164	0.137	0.019	0.820	3.41
INS	300	1.332	0.713	0.572	5.297	4.80
UL	300	0.614	0.114	0.379	0.896	0.95
OP	300	0.255	0.262	0.008	1.342	2.00
RD	300	0.018	0.012	0.004	0.066	4.03
INO	300	9.819	1.333	6.254	12.515	4.67
ER	300	0.003	0.003	0.000	0.025	1.12
LnEC	300	9.334	0.466	8.647	10.808	1.67
HR	300	0.022	0.006	0.009	0.044	3.96
LnOT	300	11.025	0.735	8.843	12.431	3.48
IG	300	1.332	0.713	0.572	5.297	1.79
TC	300	8.473	13.350	0.050	103.469	1.19
RU	300	0.393	0.057	0.193	0.584	2.51
CU	300	0.406	0.048	0.246	0.572	2.10
hagg	300	0.763	0.776	0.045	3.667	3.61
pagg	300	1.269	1.436	0.421	7.509	3.94
cagg	300	2.664	1.768	0.845	9.307	5.16
pol	300	0.757	0.430	0.000	1.000	2.17

Note: The VIF test shows that the average VIF for all variables is about 2.80, and all individual VIF values are less than 10, so there is no multicollinearity problem.

Table 4. Descriptive statistics for each variable.

Variables	Observations	Mean	Standard Deviation	Min	Max	VIF
LnCE	300	10.463	0.761	8.616	12.274
DT	300	0.164	0.137	0.019	0.820	3.41
INS	300	1.332	0.713	0.572	5.297	4.80
UL	300	0.614	0.114	0.379	0.896	0.95
OP	300	0.255	0.262	0.008	1.342	2.00
RD	300	0.018	0.012	0.004	0.066	4.03
INO	300	9.819	1.333	6.254	12.515	4.67
ER	300	0.003	0.003	0.000	0.025	1.12
LnEC	300	9.334	0.466	8.647	10.808	1.67
HR	300	0.022	0.006	0.009	0.044	3.96
LnOT	300	11.025	0.735	8.843	12.431	3.48
IG	300	1.332	0.713	0.572	5.297	1.79
TC	300	8.473	13.350	0.050	103.469	1.19
RU	300	0.393	0.057	0.193	0.584	2.51
CU	300	0.406	0.048	0.246	0.572	2.10
hagg	300	0.763	0.776	0.045	3.667	3.61
pagg	300	1.269	1.436	0.421	7.509	3.94
cagg	300	2.664	1.768	0.845	9.307	5.16
pol	300	0.757	0.430	0.000	1.000	2.17

Note: The VIF test shows that the average VIF for all variables is about 2.80, and all individual VIF values are less than 10, so there is no multicollinearity problem.

Table 5. Correlation test.

	LnCE	DT	INS	UL	OP	RD	INO	ER	LnEC	HR	LnOT	IG	TC	RU	CU	hagg	pagg	cagg	pol
LnCE	1.000
DT	0.31 **	1.00
INS	−0.28	0.32 *	1.00
UL	0.01	0.41 ***	0.36 **	1.00
OP	−0.04	0.37 **	0.41 **	0.18	1.000
RD	−0.02	0.13	0.24	0.36 **	0.47 ***	1.00
INO	0.29	0.24	0.17	0.48 ***	0.17	0.21	1.00
ER	0.10	−0.34 *	−0.17	−0.16	−0.22	−0.26	0.42 ***	1.00
LnEC	−0.08	0.01	0.32 *	0.23	0.16	0.47 ***	0.47 ***	−0.14	1.00
HR	−0.05	0.11	0.28	0.13	0.21	0.47 ***	0.32 *	−0.26	0.29	1.00
LnOT	0.31 **	0.23	−0.14	0.24	0.31 *	0.28	0.31 *	−0.27	0.13	0.10	1.00
IG	−0.28	0.34 *	0.47 ***	0.46 ***	0.46 ***	0.41 ***	0.18	−0.18	0.50 ***	0.28	−0.13	1.00
TC	0.30 *	0.39 **	0.01	0.39 **	0.48 ***	0.48 ***	0.38 **	−0.25	0.37 *	0.04	0.13	0.01	1.00
RU	0.19	−0.17	−0.03	0.01	−0.32 *	−0.15	−0.17	0.28	−0.14	0.12	0.05	−0.03	−0.12	1.00
CU	0.14	−0.32 *	−0.29	−0.27	−0.37 **	−0.28	−0.31 *	0.28	−0.30	−0.11	−0.12	−0.29	−0.23	0.42 **	1.00
hagg	0.11	0.46 ***	0.13	0.20	0.48 ***	0.41 ***	0.32 *	−0.29	0.48 ***	0.18	0.11	0.13	0.47 ***	−0.29	−0.21	1.00
pagg	−0.16	0.27	0.12	0.36 **	0.210	0.44 **	0.17	−0.04	0.20	0.16	−0.15	0.41 **	0.06	−0.06	−0.11	0.26	1.00
cagg	−0.09	0.42 **	0.37 **	0.22	0.45 **	0.12	0.33 *	−0.21	0.19	0.25	0.16	0.39 **	0.39 **	−0.21	−0.23	0.48 ***	0.38 **	1.00
pol	−0.31 *	0.22	0.21	0.31 *	0.31 *	0.28	0.15	−0.27	0.27	0.08	0.04	0.203	0.21	−0.15	−0.24	0.28	0.21	0.28	1.00

Note: ***, **, * respectively indicate significance levels above 1%, 5%, and 10%.

Table 5. Correlation test.

	LnCE	DT	INS	UL	OP	RD	INO	ER	LnEC	HR	LnOT	IG	TC	RU	CU	hagg	pagg	cagg	pol
LnCE	1.000
DT	0.31 **	1.00
INS	−0.28	0.32 *	1.00
UL	0.01	0.41 ***	0.36 **	1.00
OP	−0.04	0.37 **	0.41 **	0.18	1.000
RD	−0.02	0.13	0.24	0.36 **	0.47 ***	1.00
INO	0.29	0.24	0.17	0.48 ***	0.17	0.21	1.00
ER	0.10	−0.34 *	−0.17	−0.16	−0.22	−0.26	0.42 ***	1.00
LnEC	−0.08	0.01	0.32 *	0.23	0.16	0.47 ***	0.47 ***	−0.14	1.00
HR	−0.05	0.11	0.28	0.13	0.21	0.47 ***	0.32 *	−0.26	0.29	1.00
LnOT	0.31 **	0.23	−0.14	0.24	0.31 *	0.28	0.31 *	−0.27	0.13	0.10	1.00
IG	−0.28	0.34 *	0.47 ***	0.46 ***	0.46 ***	0.41 ***	0.18	−0.18	0.50 ***	0.28	−0.13	1.00
TC	0.30 *	0.39 **	0.01	0.39 **	0.48 ***	0.48 ***	0.38 **	−0.25	0.37 *	0.04	0.13	0.01	1.00
RU	0.19	−0.17	−0.03	0.01	−0.32 *	−0.15	−0.17	0.28	−0.14	0.12	0.05	−0.03	−0.12	1.00
CU	0.14	−0.32 *	−0.29	−0.27	−0.37 **	−0.28	−0.31 *	0.28	−0.30	−0.11	−0.12	−0.29	−0.23	0.42 **	1.00
hagg	0.11	0.46 ***	0.13	0.20	0.48 ***	0.41 ***	0.32 *	−0.29	0.48 ***	0.18	0.11	0.13	0.47 ***	−0.29	−0.21	1.00
pagg	−0.16	0.27	0.12	0.36 **	0.210	0.44 **	0.17	−0.04	0.20	0.16	−0.15	0.41 **	0.06	−0.06	−0.11	0.26	1.00
cagg	−0.09	0.42 **	0.37 **	0.22	0.45 **	0.12	0.33 *	−0.21	0.19	0.25	0.16	0.39 **	0.39 **	−0.21	−0.23	0.48 ***	0.38 **	1.00
pol	−0.31 *	0.22	0.21	0.31 *	0.31 *	0.28	0.15	−0.27	0.27	0.08	0.04	0.203	0.21	−0.15	−0.24	0.28	0.21	0.28	1.00

Note: ***, **, * respectively indicate significance levels above 1%, 5%, and 10%.

4. Empirical Results

4.1. Dynamic Panel Model Regression Results

This research adopts the systematic GMM method to conduct empirical research, which can effectively solve the endogenous problem of DT level. According to Formula (1), the measurement results are shown in

Table 6. Model calculation results.

Variables	GMM	GMM	GMM
l.lnCE	1.081 ***	1.011 ***	1.018 ***
	(0.029)	(0.023)	(0.014)
DT	−0.061 *	−0.411 *	−0.091 **
	(0.033)	(0.429)	(0.039)
Control variables	Yes	Yes	Yes
Time effect	No	Yes	Yes
Regional effect	Yes	No	Yes
AR(1)	0.003	0.001	0.001
AR(2)	0.598	0.877	0.932
Sargan	0.133	0.169	0.101
Hansen	0.237	0.877	0.193
Obs	300	300	300

Note: ***, **, and * are significant above 1%, 5%, and 10%, respectively. Standard error in parentheses. AR(1), AR(2), Sargan test, and Hansen test correspond to p-values; the same below.

. For further comparative analysis, this research mainly reports the outcomes of the double fixed-effect measurement in conjunction with the GMM results for fixed regional and time effects, as well as the double fixed system.

The AR(1) values below 0.1 and AR(2) values above 0.1 indicate significant first-order autocorrelation without second-order autocorrelation. The p-values of the Hansen and Sargan tests exceed 0.1, validating the effectiveness of the instrumental variables and confirming the overidentification tests are passed. These indicators demonstrate that the system GMM results are reliable.

From column (3), it is observed that the coefficient of l.lnCE is positive and significant. This indicates a “snowball effect” in China’s CE within the sample period, where CE from one year led to higher emissions in the following year, suggesting that China’s CE are accumulating annually rather than abruptly increasing. Additionally, the coefficients for DT levels are consistently negative and achieve statistical significance at least at the 10% level, underscoring that overall, DT level has a significant inhibitory effect on CE. DT facilitates the optimization of industrial structures and drives technological innovation, thereby aiding in the reduction of CE. Through mechanisms such as digital technology, digital supply chains, data trade, digital trading platforms, and digital content trading, further development of regional DT levels can be promoted, thereby positively impacting regional CE reduction.

In contrast, European discussions on DT and CE provide a distinctive perspective. European scholars and policymakers are actively exploring avenues similar to those in China, particularly through initiatives like the European Green Deal, which aims for Europe to achieve climate neutrality by 2050 [57,58,59]. Discussions within European academic and policy circles often emphasize the integration of DT technologies with stringent environmental regulations as a means to expedite the transition toward a sustainable economy [60,61]. A comparative analysis of these European methods with findings from China reveals that while both regions recognize the potential of digital technologies to reduce CE, Europe tends to integrate these technological strategies within more comprehensive regulatory frameworks. European academics argue that such integration significantly enhances the efficacy of DT in reducing carbon footprints, especially when combined with incentives for green technology and stricter CE standards [62]. This comparative analysis not only broadens the understanding of global strategies toward carbon neutrality but also highlights the importance of aligning technological advancement with environmental sustainability goals through coherent policies. These discussions illuminate potential areas for China to strengthen its policy framework to better align with global best practices, potentially fostering a more collaborative international effort to combat climate change.

4.2. Heterogeneity Analysis

Regional economic development in China varies. Given the superior economic development of the eastern region compared to the central and western regions in China, an analysis is conducted to examine whether the disparity in regional economic development affects the influence of DT on CE, and the results are shown in

Table 7. Heterogeneity test results.

Variables	Eastern	Central	Western
l.lnCE	1.013 ***	1.129 ***	0.789 ***
	(−0.021)	(0.112)	(0.880)
DT	−0.429 **	−0.352 *	−0.189 *
	(0.210)	(0.608)	(0.123)
Control variables	Yes	Yes	Yes
Time effect	Yes	Yes	Yes
Regional effect	Yes	Yes	Yes
AR(1)	0.019	0.016	0.030
AR(2)	0.238	0.183	0.322
Sargan	0.678	0.621	0.180
Hansen	0.081	0.070	0.071
Obs	300	300	300

Note: ***, **, * denote significance at the 1 per cent, 5 per cent and 10 per cent levels, respectively.

.

Based on the outcomes of the regional heterogeneity analysis, the influence of DT level on CE in eastern, central, and western China is negative and statistically significant, suggesting that higher DT levels foster CE reduction. The coefficient value showed a gradually decreasing trend from the east to the west, and the significance level in the east was higher, indicating that the effect of DT level on CE reduction in the east was more obvious than that in the central and western regions. This observation is consistent with the H4 proposed in this research. Analysis of the gradual decline from the east to the west is mainly as follows: ① The central and western regions may trail the eastern regions in infrastructure and technology, including Internet coverage, logistics system efficiency and openness, and the adoption of digital technologies. These disparities constrain the potential of DT to enhance energy efficiency and decrease CE. ② The central and western regions exhibit lower economic development levels compared to the eastern regions, and the industrial structure is more dependent on traditional manufacturing and heavy industry, which have relatively high CE intensity. The penetration and impact of DT in these industries is likely to be limited. ③ The central and western regions may be less aggressive and strict in the development and enforcement of environmental policies than the eastern regions, which may slow the role of DT in driving CE reduction. ④ The eastern regions mainly have sufficient government financial funds, while the central and western regions have relatively inadequate financial resources and investment, which limits the investment in DT and related low-carbon technologies.

5. Further Discussion

5.1. Mediation Effect

To examine the mediating role of production-side and consumption-side variables in the impact of DT level on CE, the measurement is based on Formulas (2) and (3), and the Sobel test was conducted to verify whether there was a mediating role. The results are shown in

Table 8. Calculation results of mediation effects on the production side.

Variables	OT (1)	IG (2)	TC (3)	lnCE (4)
l.lnCE				0.953 ***
				(0.151)
DT	3.102 **	0.379 *	7.923 ***	−0.192 *
	(0.522)	(0.252)	(0.073)	(0.401)
OT				−0.059 *
				(0.167)
IG				0.041 **
				(0.105)
TC				−0.003 ***
				(0.005)
Control variables	Yes	Yes	Yes	Yes
Time effect	Yes	Yes	Yes	Yes
Regional effect	Yes	Yes	Yes	Yes
AR(1)	0.012	0.061	0.022	0.002
AR(2)	0.188	0.120	0.229	0.407
Sargan	0.123	0.834	0.200	0.298
Hansen	0.251	0.267	0.214	0.177
Sobel test Z	10.123 **	12.001 *	20.321 ***	18.279 **
	(0.043)	(0.054)	(0.001)	(0.013)
Obs	300	300	300	300

Note: ***, **, * denote significance at the 1 per cent, 5 per cent and 10 per cent levels, respectively. The Sobel test has a p value in parentheses. The same below.

.

Columns (1) to (3) introduce the impact of DT level on production-side mediating variables (OT, IG, TC), and column 4 introduces the regression results of production-side mediating variables, explanatory variables, and explained variables (lnCE). The coefficient values of DT are positive and statistically significant, indicating that DT level facilitates regional output improvement, industrial structure upgrading, and technological progress. It is evident from column 4 that the OT coefficient value is negative and statistically significant, the IG coefficient value is positive and statistically significant, and the TC coefficient value is negative and statistically significant, suggesting an intermediary effect of the scale effect, structure effect, and technology effect on production-side variables, and the promoting effect of scale effect and technology effect on CE reduction surpasses that of the structure effect on CE, which supports H1 and H2. Moreover, under the influence of mediating variables on the production side, the DT coefficient value is still negative and statistically significant, indicating that DT level still promotes CE reduction under the influence of the production side.

The Z-values of the Sobel test in columns (1) to (4) are 10.123, 12.001, 20.321, and 18.279, respectively, indicating the robustness of the mediating effect of the OT, IG, and TC variables. Column 4 shows a significant negative coefficient for OT, indicating that a scale effect indeed exists and that an expansion in scale mediates a reduction in CE. This supports H1, which posits that DT promotes a reduction in CE by increasing production scale. However, the coefficient for the advancement of IG is positive and significant, suggesting that while the structural effect acts as a mediator between DT and CE, it paradoxically increases CE, contrary to the expectations set forth in H2, which predicted that structural effects would facilitate a reduction in CE. Therefore, H2 does not receive full support from this part of the analysis. Conversely, the coefficient for TC is negative and significant, lending support to H3. This indicates that the technological effect, acting as a mediator, does reduce CE, aligning with the expectation that DT fosters CE reduction through technological advancement.

The results of the mediation effect on the consumption side are shown in

Table 9. Calculation results of mediation effects on the consumption side.

Variables	RU (1)	CU (2)	lnCE (3)
l.lnCE			1.021 ***
			(0.006)
DT	0.101 *	0.104 ***	−0.024 *
	(0.055)	(0.024)	(0.009)
RU			0.029 *
			(0.220)
CU			−0.249 **
			(0.347)
Control variables	Yes	Yes	Yes
Time effect	Yes	Yes	Yes
Regional effect	Yes	Yes	Yes
AR(1)	0.032	0.043	0.001
AR(2)	0.172	0.912	0.857
Sargan	0.463	0.342	0.369
Hansen	0.300	0.276	0.279
Sobel test Z	28.832 ***	13.450 ***	14.330 ***
	(0.005)	(0.002)	(0.001)
Obs	300	300	300

Note: ***, **, * denote significance at the 1 per cent, 5 per cent and 10 per cent levels, respectively.

. Columns (1) and (2) introduce the influence of consumption-side mediating variables (RU, CU) on explanatory variables (DT), and column (4) introduces the regression results of consumption-side mediating variables, explanatory variables, and explained variables (lnCE).

It is evident from columns (1) and (2) that the DT coefficient value is positive and significant, indicating that the DT level promotes consumption upgrading among rural and urban residents. The Z-values of the Sobel test in columns (1) to (3) are 28.832, 13.450, and 14.330, respectively, indicating the robustness of the mediating effect of RU and CU variables. In Column (3), the coefficient value of the consumption upgrading variable (RU) of urban residents is positive and statistically significant, while the coefficient value of the consumption upgrading variable (CU) of rural residents is negative and statistically significant, indicating that the consumption-side variable has an intermediary role, and the intermediary role of RU is to enhance the CE reduction effect of DT by promoting consumption upgrading. The mediation role of CU is to promote consumption upgrading to weaken the CE reduction effect of DT. Moreover, DT values in column (3) exceed DT values in column (3) of Table 5, indicating that consumption upgrade variables composed of RU and CU reflect that consumption upgrade weakens the impact of DT on CE reduction, rather than enhancing the CE reduction of DT. This is contrary to hypothesis 5, indicating that hypothesis 5 is not valid.

5.2. Moderation Effect

Table 10. Calculation results of the regulatory effects of industrial agglomeration and low-carbon pilot policy.

Variables	(1)	(2)	(3)	(4)
l.lnCE	1.011 ***	1.015 ***	1.024 ***	1.013 ***
	(0.012)	(0.012)	(0.028)	(0.011)
DT	−0.141 *	−0.176 *	−0.621 **	0.247 *
	(0.132)	(0.242)	(0.381)	(0.783)
hagg	−0.062 **
	(0.032)
hagg × DT	0.139 *
	(0.108)
pagg		−0.009 *
		(0.012)
pagg × DT		0.055 *
		(0.042)
cagg			−0.030 **
			(0.012)
cagg × DT			0.124 **
			(0.066)
pol				0.141 ***
				(0.124)
pol × DT				−0.412 **
				(0.923)
Control variables	Yes	Yes	Yes	Yes
Time effect	Yes	Yes	Yes	Yes
Regional effect	Yes	Yes	Yes	Yes
AR(1)	0.012	0.001	0.002	0.002
AR(2)	0.884	0.988	0.787	0.903
Sargan	0.254	0.165	0.531	0.414
Hansen	0.232	0.292	0.222	0.277
Obs	300	300	300	300

Note: ***, **, * denote significance at the 1 per cent, 5 per cent and 10 per cent levels, respectively.

assesses the regulatory influence of industrial agglomeration and low-carbon pilot policy on the impact of DT level on CE, the measurement is conducted using Formula (4). Columns (1) to (3) are the regulatory effects of industrial agglomeration, and column (4) is the regulatory effects of low-carbon pilot policy.

From column (1) to column (3), it is evident that the coefficients of hagg, pagg, and cagg, in conjunction with their interaction terms with DT, are all negative and statistically significant. This indicates that these industrial agglomerations exert a negative moderating effect on the impact of DT on CE. However, upon including the industrial agglomeration factors and interaction terms, the coefficient value of CE increases and remains negative. This suggests that industrial agglomeration factors exhibit a trend of weakening the effect of DT on reducing CE. Therefore, this result contradicts H6, indicating that industrial agglomeration does not amplify the reduction effect of DT on CE.

From column (4), the coefficient of the interaction term between policy dummy variables and DT level is negative and statistically significant, indicating that the low-carbon pilot policy can enhance the CE reduction effect of DT level, thus verifying H7. Additionally, after incorporating the interaction terms and policy dummy variables, the coefficient value of DT changes from negative to positive and becomes statistically significant. This suggests that even under the influence of the low-carbon pilot policy, DT has a more significant impact on reducing CE for companies from provinces involved in the pilot CE reduction program. Therefore, DT exhibits a stronger CE reduction effect in provinces participating in the low-carbon pilot CE reduction program.

6. Summary, Policy Implications, and Limitations

Based on the empirical analysis, the following conclusions are drawn: (1) The level of DT significantly reduces CE, and the CE effect of regional heterogeneity demonstrates a declining trend from the east to the west. (2) From a production standpoint, the mediating role of urban residents’ consumption upgrading is to enhance the effect of DT on reducing CE by promoting consumption upgrading. Conversely, the mediating role of rural residents’ consumption upgrading is to promote consumption upgrading but weaken the effect of DT on reducing CE. (3) Regarding regulatory impact, the factor of industrial agglomeration tends to diminish the impact of DT on reducing CE; therefore, industrial agglomeration does not amplify the reduction effect of DT on CE. Conversely, low-carbon pilot policy can enhance the CE reduction effect of DT, demonstrating stronger CE reduction effects in provinces that participate in these low-carbon pilot programs.

Based on the aforementioned conclusions, this research offers significant policy implications for developing countries: (1) Developing countries should actively advance DT development to enhance economic efficiency and mitigate CE. The impact of DT on CE reduction varies across different regions, and DT development strategies should be tailored to local conditions and adapted to the specific circumstances of different regions. The medium level of DT is more sensitive to the impact of CE, and DT development at this stage should be concentrated on and streamlined. Simultaneously, it is imperative to carry out the digital transformation of traditional trade forms, including digital technology trade, digital supply chain, data trade, DT platform, and digital content trade, to reduce CE caused by industrial structure upgrading as much as possible. (2) Encourage enterprises to expand production scale and reduce CE per unit product through the scale effect. Enhance the research, development, and deployment of innovative technologies, boost production and energy efficiency, and lower CE. Optimize industrial structure, foster green industries and low-carbon economy, and lower the percentage of high-CE sectors. (3) Promoting the upgrading of consumption among urban and rural residents. The government should support urban residents in adopting more digital platforms for consumption, thereby improving the convenience of online shopping and digital services. Additionally, increased investment in innovation and technology adoption in urban areas is essential, particularly in sectors such as smart homes, electric vehicles, and energy-efficient devices. For rural residents, it is crucial to provide digital literacy education and training to help them better understand and use digital technologies. Furthermore, improving Internet coverage and access speeds in rural areas will ensure that rural residents have easy access to digital platforms. (4) Strengthening the management and optimization of industrial agglomeration. While promoting industrial agglomeration, the government should emphasize achieving environmental protection goals. Corresponding policies and regulations should be formulated to ensure that the industrial agglomeration process does not lead to environmental deterioration. Encouraging the formation of green industrial clusters and introducing clean energy and environmental protection technologies can help mitigate the negative impact of clustered industries on the environment. (5) Implementing and expanding low-carbon pilot policy. Low-carbon pilot projects should be promoted in more provinces and regions. Through policy incentives and technical support, local governments and enterprises should be encouraged to participate in low-carbon transformation. Ensuring the effective implementation of low-carbon pilot policy requires the establishment of robust supervision and evaluation mechanisms to ensure that all policies and measures are properly executed.

The primary constraints of this research include the following: (1) It depends largely on data sourced from the China Energy Statistical Yearbook and IPCC guidelines, potentially restricting the applicability of its findings to varying datasets or regions. (2) The research recognizes the regional disparities in how DT affects CE but may not encompass all the socio-economic, cultural, and policy variances that could alter these outcomes. (3) While the research concentrates on the direct CE resulting from energy use for clarity, it might not consider the indirect CE associated with DT, such as CE from manufacturing digital equipment, server farms, and other related infrastructure, which also significantly contribute to the CE footprint.