Digital Economy, Energy Structure Transformation, and Regional Carbon Dioxide Emissions

Abstract

For China to reach its “dual carbon” aim, the digital economy presents both opportunities and obstacles. This paper examines the potential impact of digital economy development on regional carbon dioxide emissions, concluding that while the direct impact on regional carbon dioxide emissions through industrial structure upgrading and technological progress is unclear, the indirect impact through promoting energy structure transformation is both positive and clear. This research experimentally validates the results of a theoretical analysis using panel data from 30 Chinese provinces from 2011 to 2019 and the STIRPAT model and discovers an inverted U-shaped association between digital economy development and regional carbon dioxide emissions. According to the transmission path test, the digital economy primarily achieves its carbon dioxide emission reduction effect by promoting energy structure transformation, while the carbon dioxide emission reduction effect from upgrading industrial institutions and technological progress is insignificant. By region, the expansion of the digital economy in eastern China has a large carbon dioxide emission reduction effect but not in central and western China. The policy implication of this paper is that curbing new carbon dioxide emissions caused by digital infrastructure construction using the digital economy to accelerate energy structure transformation and accelerating digital economy development in central and western China can be effective policy options for the Chinese government to achieve the goal of carbon dioxide emission reduction.

1. Introduction

Carbon dioxide emissions are a global issue that everyone is dealing with. Massive emissions of greenhouse gases, primarily carbon dioxide, are endangering not only sea level rise and extreme weather but also agricultural production and human life. Although China’s carbon dioxide emissions per capita are low, the total amount is significant. China’s carbon dioxide emissions increased from 8.15 billion tons to 9.90 billion tons between 2010 and 2020, according to the World Energy Statistical Yearbook 2021, making China the world’s greatest carbon emitter. As a major responsible country, China pledged at the UN General Assembly’s 75th session in 2020 to “increase its nationally determined contributions, adopt more effective policies and measures, strive to peak carbon dioxide emissions by 2030, and strive to achieve carbon neutrality by 2060.” The Chinese Government’s 2022 Work Report emphasized the need to “orderly promote carbon peak and carbon neutrality, promote the energy revolution and low-carbon energy transformation, promote research and development and application of green and low-carbon technologies, and build a green manufacturing and service system.” This raises the question of how we move forward. How can we attain green development, a goal shared by all humanity? The digital economy appears to be a key mechanism and starting point for China to achieve this goal.

In recent years, digital technologies such as big data, cloud computing, and artificial intelligence have been increasingly integrated into different commercial and social domains. The subsequent wave of the digital economy has become a massive force in transforming the traditional economic model, restructuring the global value chain, and promoting the economy’s green transformation. According to the 2022 White Paper on China’s Digital Economy Development, China’s digital economy is thriving, with its size increasing from 2.6 trillion yuan in 2005 to 45.5 trillion yuan now. Its GDP share has increased from 14.2 percent in 2005 to 39.8 percent today. The Chinese government places a high value on the new economic drivers and benefits inspired by the digital economy. The Chinese government specifically proposed to “accelerate digital development and build a digital China” in its 14th Five-Year Plan for National Economic and Social Development and Outline of Long-Term Goals for 2035, published in 2021. The Chinese government’s 14th Five-Year Plan for Digital Economy Development, issued in 2022, emphasizes the need to “accelerate the deep integration of digital economy and real economy, as well as promote green development in the process of industrial digital transformation”.

Is China’s digital economy capable of encouraging green development? Will it aid China’s objectives of carbon peaking and carbon neutrality? Has it contributed to China’s carbon dioxide emission reduction? These issues have emerged as a major source of concern in the field of economic theory. In order to answer the above issues, this article will analyze how the digital economy influences carbon dioxide emissions and investigate the mechanism and effect of the digital economy on China’s carbon dioxide emissions. This paper’s key contributions and academic merits are as follows: (1) unlike most studies, this paper examines the relationship between the digital economy and carbon dioxide emissions in the short and long term, discusses its linear and nonlinear effects, and provides new empirical evidence for policies related to the implementation of low-carbon controls during the digital economy’s pre-development period. (2) The chosen topic offers a fresh perspective. The digital economy, energy structure upgrading, and carbon dioxide emissions are combined into a unified theoretical framework to investigate the impact of the digital economy on carbon dioxide under the mediating role of energy structure upgrading as well as to clarify the operational implementation path of policy recommendations. (3) Given the significant disparities in inter-provincial economic development and resource endowments in China, this work investigates regional heterogeneity and develops a multi-perspective robustness test to increase the trustworthiness of research findings. This paper’s research has important theoretical and practical implications for understanding how the digital economy enables green and sustainable development, for understanding China’s realistic strategies and effectiveness in achieving the goal of carbon peaking and carbon neutrality, and for understanding how China can use differentiated regional digital economy promotion policies to achieve the goal of carbon peaking and carbon neutrality.

2. Literature Review

The digital economy refers to a series of economic activities based on digital technology, which includes using digital platforms as the main medium and digital empowerment infrastructure as an important support [1]. The current articles on the topic of the digital economy mainly pay attention to its effects on the economy and society. On the one hand, the digital economy has a significant impact on high-quality development [2], financial inclusion [3,4], global value chain [5,6] industrial structure upgrading [7,8], etc. On the other hand, the digital economy plays a favorable role in innovation and development, efficiency, and cost reduction by influencing the behavior of micro-enterprises [9,10,11]. From the existing research, there is less literature on the digital economy and carbon dioxide emissions. Barrett M et al. summarized three types of methods to reduce carbon dioxide emissions in their research of economic development and energy consumption, including the use of Information and Communications Technology (ICT) [12]. Robert analyzed the positive and negative relationship between ICT and the environment based on OECD data and made recommendations for international organizations and individuals to participate in this issue [13]. Popescu G’s study found that ICT dematerialization and efficiency improvements can reduce carbon dioxide emissions [14]. With the promotion and application of digital technology in the economy and society, digitalization, networking, and intelligence are developing rapidly. Some scholars have started to attach importance to the effect of the digital economy on the environment. Li [15] believes that the digital economy can unleash the driving force of innovation and can lower the emission of sulfur dioxide. Sun [16] and Deng [17] empirically determined that the developing digital economy can remarkably suppress the emission of environmental pollutants from multiple dimensions and that the upgrading of industrial structure is an important mechanism to promote improvement in environmental quality.

However, articles on the influence of the digital economy on carbon dioxide emissions are in the primary stage, and the few research results in this area are still controversial. On the theoretical side, the paths of the digital economy to achieve the carbon neutrality goal are mainly listed, and thoughts on the coordinated digital transformation of industries and related policy directions are proposed [18,19]. Most scholars hold a promising attitude toward the carbon dioxide emission reduction effect of the digital economy. The SMARTer2030 report released by the Global e-Sustainability Initiative shows that the world is expected to reduce global emissions by 12 GtCO₂e by 2030 through ICT solutions and to promote sustainable economic growth. Empirically, it is mainly based on panel data. Moreover, tests find that the development of the digital economy can remarkably lower carbon dioxide emissions through various mathematical models and mechanism analyses [20,21]. Some papers have also examined the spatial spillover effects of the digital economy affecting carbon dioxide emissions by using spatial models [22,23].

On the other hand, there are some different conclusions. Khan [24] studied the relationship between ICT, economic growth, financial development, and environmental quality in a sample of emerging economies and found that the moderating effects of ICT stimulated carbon dioxide emissions. Zhou [25] adopted an input–output approach to develop a specific carbon analysis framework, and the result was that ICT is far from environmentally friendly, with implied carbon impacts tens of times greater than direct impacts. Nizam [26] used time series data for Pakistan from 1975–2017 to study the long-term relationship between ICT, energy demand, and carbon dioxide emissions and found that the application of ICT had an obvious effect of increasing carbon dioxide emissions.

Throughout the literature, most of the papers recognize the carbon dioxide emission reduction capability of the digital economy but do not investigate this impact dynamically. There is a lack of research that explores the specific impact mechanisms from both direct and indirect effects. This paper attempted to build a multi-perspective analysis framework with energy structure transformation, technological progress, and industrial structure upgrading as paths and selected China’s provincial panel data as research samples to study the dynamic impact and mechanisms of the digital economy on regional carbon dioxide emissions. Therefore, we summarized the dynamic relationship, influence mechanisms, and regional heterogeneity of the two, which provides more ideas and inspirations for the formulation of digital economy policies and the realization of carbon peaking and carbon neutrality goals.

3. Theoretical and Mechanistic Analysis

3.1. How the Digital Economy as a Whole Affects Local Carbon Dioxide Emissions

The digital economy is developed mainly by the deep integration of a series of digital technologies such as big data, cloud computing, artificial intelligence, and the Internet of Things with the real economy. With modern information networks as an important carrier, digital technology has become a key factor of production in the era of the digital economy through efficient processing of knowledge and information, and it has the advantages of cross-time and space information dissemination, data creation, and reducing transaction costs. From a macro perspective, the digital economy has given China a strong impetus for new economic growth, will lead to high-quality development, and will significantly reduce China’s carbon dioxide emissions with strong technical and financial backing. The government will have more policy leeway to encourage industrial enterprises to manage their energy demand with the help of digital technology. Accurate energy macro supply and demand analysis will also be more possible, and this will improve the implementation path to achieve carbon peaking and carbon neutrality. From a macro perspective, as the digital economy grows, the industrial structure can be upgraded and optimized, and resource allocation will become much more effective. New business models appear as the time demands, industrial integration and innovation continue, and the constant appearance of new energy-saving goods will continually lower the demand for fossil fuels and lower carbon dioxide emissions. Digital finance lowers enterprise financing limitations, decreases resource mismatches, and significantly boosts production efficiency at the micro level. The digital transformation of businesses also makes it possible to create flexible manufacturing, which dramatically reduces costs and boosts production efficiency. In conclusion, there are many opportunities for energy conservation and carbon reduction in the digital economy.

However, the dual nature of the digital economy and digital technology in terms of energy conservation and carbon reduction cannot be overlooked. Building digital infrastructure is a crucial first step in the growth of the digital economy. Carbon dioxide emissions have risen sharply in recent years as a result of China’s phenomenal growth in digital infrastructure. The carbon dioxide emissions of data centers and 5G facilities, will climb by 103% and 321%, respectively, by 2035, according to the report “Decarbonization Road of China’s Digital Infrastructure”. China’s efforts to reach the aim of carbon peaking and carbon neutrality would undoubtedly face new obstacles due to the increase in carbon dioxide emissions brought on by the digital infrastructure. Additionally, improper or abusive use of digital technology will raise costs and waste resources, which is counterproductive to carbon reduction. This is especially true when the digital economy is just getting started and requires a lot of energy and resource extraction, which will result in higher carbon dioxide emissions. The development of digital infrastructure and the use of digital technologies will only have a beneficial impact on carbon reduction once the production of the digital economy is steady and on a sizeable scale.

The above analysis supports Hypothesis 1, which is proposed:

H1. 

The growth of the digital economy and local carbon dioxide emissions are correlated in an inverted U pattern, with the digital economy first promoting and later inhibiting local carbon dioxide emissions.

3.2. The Direct Impact of Regional Carbon Dioxide Emissions from the Digital Economy

The modernization of the industrial structure and advancements in technology are actually the essence and primary manifestation of the development of the digital economy. From the standpoint of industrial structure, technological adoption and industrial integration are responsible for the optimization effect of the digital economy on industrial structure. On the one hand, the labor- and resource-intensive industrial structure has gradually given way to a knowledge- and environment-intensive industrial structure with the development of a new digital industrial system. Big data, cloud computing, artificial intelligence, and traditional production factors work together to circulate resources within and between industries, efficiently distribute and use resources, and give businesses’ industrial development and green transformation a technical boost. On the other hand, conventional sectors are transforming as a result of the power of digital technology, and new business models such as the “sharing economy” and “digital payment” are emerging. The typical two-way integration of modern manufacturing and services, particularly manufacturing services, deepens the complexity of the enterprise value chain, helps to improve the division of enterprises in the value chain, and helps enterprises get rid of high fixed costs and low-end locking risk. Global value chain embedding can also encourage the growth of low-end manufacturing. In other words, the improvement and optimization of the industrial structure against the backdrop of the digital economy are anticipated to open up new opportunities for energy efficiency, emission reduction, and the growth of a green economy.

From the standpoint of technological progress, the digital economy, which is based on the Internet platform, improves the efficiency of information distribution, collects more innovative resources, and can optimize the allocation of production elements, thereby affecting regional carbon dioxide emissions. To begin with, new analysis technologies such as artificial intelligence can alter decision making, and developing computing technologies such as cloud computing will give birth to new big data applications in a variety of industries. Data mining and data crawler technologies fully utilize digital information resources, eliminate unnecessary intermediate links, substantially simplify the decision-making operation process, and aid in the reconstruction and optimization of the industrial chain structure. Big data may also assist organizations in obtaining more and more important information, precisely understanding market demand, improving their ability to make scientific judgments and foresee dangers, and assisting enterprises in reducing costs and increasing efficiency, saving energy, and reducing emissions. Second, digital technology has the potential to open up new resource areas and methods of use, accomplish resource recycling on a larger scale and at a higher level, and improve the optimization of the industrial chain ecosystem. The widespread use of digital technology may not only improve corporate operational efficiency but also break down information barriers and build an innovative knowledge exchange system, allowing businesses to save costs, increase efficiency, conserve energy, and reduce emissions. Finally, digital technology can assist enterprises in establishing super-large-scale innovation and scientific research platforms, deepening exchanges and cooperation through technology integration and information sharing, and providing development space for more innovation activities, thereby assisting enterprises in lowering costs and increasing efficiency, saving energy, and lowering emissions.

Simultaneously, as previously said, the digital economy and digital technology have a dual nature in terms of energy conservation and emission reduction. Figure 1 depicts the dual mechanism of the direct impact of the digital economy on regional carbon dioxide emissions. As a result, the direct influence of the digital economy on regional carbon dioxide emissions via industrial structure optimization and technical innovation is both positive and negative.

In conclusion, Hypothesis 2 is proposed:

H2. 

The direct influence of the digital economy on regional carbon dioxide emissions is mostly represented in the upgrading of industrial structure and technological progress, which have duality and an ambiguous direct effect.

3.3. The Indirect Impact of Regional Carbon Dioxide Emissions from the Digital Economy

The key factor influencing carbon dioxide emissions globally is the change in energy structure. The key for China, in particular, is to change the coal-based energy structure because China’s energy endowment characteristics and coal’s low-price advantage have long made coal the main body of China’s energy, and coal has brought the greatest problems of carbon dioxide emissions and environmental pollution. The transformation of energy structure is the conversion of fossil energy to non-fossil energy; the thorough transformation and adjustment of all connections of energy production, transportation, conversion, and storage; the increase in renewable energy proportion; and the construction of a new energy system. The digital economy offers opportunities for energy structure transformation. First, as energy collection efficiency and online interconnections improve, energy supply links become more intensive, data-driven, and refined, providing safe and dependable technical support for energy production and operation. Data crawler and digital twin technology can assist firms in real-time monitoring and flexible production, which is beneficial to enhancing energy transportation and distribution efficiency. Through the application of coordinate control systems, computer-aided simulation, mining robots, and other technologies and equipment, the combined application of digital technology and modern coal mining technology has brought about a revolution in the coal industry, improved the limiting factors of mining, realized the accurate scheduling of energy, improved production efficiency, reduced waste of resources, and achieved the effects of energy conservation and emission reduction. Second, the strong development and application of new energy and clean energy have reduced reliance on coal, oil, and other fuels to some extent. In terms of technically exploitable hydropower resources, China can presently replace 1.143 billion tons of raw coal annually. With a ubiquitous energy network built with digital technology, the demand side can realize integrated utilization of various energy sources, the information asymmetry at both ends of supply and demand can be alleviated, the distortion of energy structure and the low efficiency of resource allocation can be improved, and carbon dioxide emissions can be reduced. Finally, the development of the digital economy has profoundly integrated conventional and new energy technologies, given rise to new forms of business such as new energy vehicles and smart grids, and pushed the development of energy systems in the direction of clean, efficient, and low-carbon development. The use of digital technology, in particular, has hastened the substitution of clean energy, lowered the cost of solar and wind power generation, and facilitated the realization of the “Hydrogen China” strategy. By gathering information through precise monitoring, digital technologies such as the Internet of Things, intelligent decision making, and remote sensing data are playing an increasingly crucial role in the capture, utilization, and storage of carbon dioxide emissions. As a result, the expansion of the digital economy can not only increase energy structure transformation but significantly cut carbon dioxide emissions caused by energy structure transformation.

In conclusion, Hypothesis 3 is proposed:

H3. 

The growth of the digital economy can stimulate energy structure change, which helps to minimize regional carbon dioxide emissions.

4. Model Construction and Data Description

4.1. Measurement Model

4.1.1. STIRPAT Model

Referring to Dietz and Rosa [27], the STIRPAT model is able to quantify the non-proportional impact of human factors on the environment [28]. The general form is:

$$I=a{P}^{{\alpha }_1}{A}^{{\alpha }_2}{T}^{{\alpha }_3}e$$

(1)

Taking logarithms on both sides gives:

$$lnI=lna+{\alpha }_{1}lnP+{\alpha }_{2}lnA+{\alpha }_{3}lnT+lne$$

(2)

Since the model allows for appropriate splitting, the STIRPAT model is refined in this paper with reference to the factors that have an impact on carbon dioxide emissions.

$$lnce={\alpha }_0+{\alpha }_{1}lnd{e}_{it}+{\alpha }_{2}lnpgd{p}_{it}+{\alpha }_{3}lnpop{u}_{it}+{\alpha }_{4}lnur{b}_{it}+{\alpha }_{5}lnfd{i}_{it}+{\alpha }_{6}lne{r}_{it}+{\alpha }_{6}lngo{v}_{it}+{\epsilon }_{it}$$

(3)

In Equation (3), $lnce$, $lnd{e}_{it}$, $lnpgd{p}_{it}$, $lnpop{u}_{it}$, $lnur{b}_{it}$, $lnfd{i}_{it}$, $lne{r}_{it}$, and $lngo{v}_{it}$ denote carbon intensity, digital economy, GDP per capita, population, urbanization level, openness, environmental regulation, and government behavior, respectively. ${\epsilon }_{it}$ are the error terms, ${\alpha }_0$ is the constant term, and ${\alpha }_1$ to ${\alpha }_6$ are the coefficients.

To further verify whether there is a non-linear relationship between the effects of the digital economy on carbon dioxide emissions, the quadratic form of the core explanatory variables is added.

$$\begin{array}{ll}lnce={\alpha }_0+& {\alpha }_{11}lnd{e}_{it}+{\alpha }_{12}\ln {\left(d{e}_{it}\right)}^2+{\alpha }_{2}lnpgd{p}_{it}+{\alpha }_{3}lnpop{u}_{it}+{\alpha }_{4}lnur{b}_{it}+{\alpha }_{5}lnfd{i}_{it}+{\alpha }_{6}lne{r}_{it}\\ & +{\alpha }_{6}lngo{v}_{it}+{\epsilon }_{it}\end{array}$$

(4)

4.1.2. Mediating Effect Model

Drawing on the mediating effects test [29] approach, the following model was set up.

$$lnc{e}_{it}=a+{\delta }_{1}lnd{e}_{it}+{\gamma }_{1}contro{l}_{it}+{\epsilon }_{it}$$

(5)

$$lnc{e}_{it}=b+{\delta }_{2}lnene{r}_{it}+{\gamma }_{2}contro{l}_{it}+{\omega }_{it}$$

(6)

$$lnene{r}_{it}=c+{\delta }_{3}d{e}_{it}+{\gamma }_{3}contro{l}_{it}+{\theta }_{it}$$

(7)

$$lnc{e}_{it}=d+{\delta }_{4}d{e}_{it}+{\delta }_{5}lnene{r}_{it}+{\gamma }_{4}contro{l}_{it}+{\mu }_{it}$$

(8)

$lnene{r}_{it}$ are the mediating variables and $lnd{e}_{it}$ are the core explanatory variables. ${\epsilon }_{it}$, ${\omega }_{it}$, ${\theta }_{it}$, and ${\mu }_{it}$ are the random disturbance terms, and $contro{l}_{it}$ are the control variables. Models 5 and 6 examine the direct effects of the two independent variables on carbon intensity. Model 7 examines whether there is a correlation between the independent variable and the mediating variable of energy structure. Model 8 puts the two variables into the regression model, focusing on the significance and direction of ${\delta }_4$ and ${\delta }_5$.

4.2. Variable Selection

4.2.1. Explained Variable: Carbon Dioxide Emissions

Referring to Chen’s [30] measurement method, the top-down accounting of carbon dioxide emissions is based on the IPCC National Greenhouse Gas Guidelines (hereinafter referred to as the Guidelines), which has strong data availability and is one of the most widely used methods for calculating carbon dioxide. On the basis of the China Energy Statistics Yearbook and some scholars’ practice, a total of eight categories of energy consumption were selected: raw coal, coke, crude oil, gasoline, kerosene, diesel, fuel oil, and natural gas [31,32]. The specific calculation formula is as follows.

$$C{E}_{ij}=C{O}_{ij}\times NC{V}_i\times C{C}_i\times O_{ij}\times 44/12$$

(9)

$C{E}_{ij}$ denotes the carbon dioxide emissions from fossil fuel i in sector j. $C{O}_{ij}$ denotes the consumption of fossil fuel i in sector j. $NC{V}_i$ denotes the average low-level heat output of fossil fuel i, known as the net heating value. $C{C}_i$ is the carbon content. $O_{ij}$ denotes the carbon oxidation factor. In addition, 44/12 is the molecular weight ratio of carbon dioxide to carbon.

Table 1. Carbon dioxide emission factors for various types of fuels (Data source: China Energy Statistics Yearbook and IPCC National Greenhouse Gas Guidelines).

Fuel Type	Low-Level Heat Generation (KJ/kg)	Carbon Content (kgC/GJ)	Oxidation Rate	Carbon Dioxide Emission Factor (tC/t)
Raw Coal	20,908	25.8	1	0.5394
Coke	28,435	29.2	1	0.8303
Crude Oil	41,816	20.0	1	0.8363
Gasoline	43,070	18.9	1	0.8140
Kerosene	43,070	20.2	1	0.8399
Diesel	42,652	21.1	1	0.8616
Fuel Oil	41,816	15.3	1	0.8823
Natural Gas	38,931 (kJ/m3)	19.5	1	0.4478

shows the carbon dioxide emission factor data for each major fuel.

The carbon dioxide emission intensity (ce) is expressed by the ratio of the calculated total carbon dioxide emissions of each province to the actual GDP of each province.

4.2.2. Core Explanatory Variables

Since the research on the development of the digital economy is still in its infancy, its measurement standards are not unified. Moreover, the main purpose is to reflect the real digital economy development as much as possible by constructing a relevant index system. This paper mainly refers to the method of Zhao Tao et al. to establish indicators from three aspects: digital infrastructure construction, digital factor inputs and outputs, and digital transaction development. The specific indicators include the number of Internet broadband access users per 100 people, the proportion of computer service and software industry employees to urban unit employees, the total amount of telecommunication services per capita, the number of cell phone users per 100 people, and the financial inclusion index [33]. The final data were calculated by the entropy method.

4.2.3. Mediating Variable

The mediating variable was energy structure (lnener), which was expressed as the ratio of coal consumption to total energy consumption in each region.

4.2.4. Control Variables

The control variables were: (1) the level of economic development, expressed by the GDP per capita of each province (lnpgdp); (2) the level of population, expressed by the total population of each province at the end of the year (lnpopu); (3) the level of urbanization (lnurb), expressed by the ratio of the urban population to the total regional population; (4) the level of openness to the outside world (lnfdi), measured by the ratio of the amount of foreign direct investment in each province; (5) environmental regulation (lner), measured by using the emissions of three wastes in each province and calculating the comprehensive index of environmental regulation using the entropy method; (6) technological progress (lntech), expressed by using the number of patent applications granted in each province or city; (7) industrial structure upgrading (lnindu), measured by the ratio of the added value of tertiary industry to the added value of secondary industry in each province; and (8) governmental behavior (lngov), referring to the marketization index in China’s Marketization Index by Provinces 2021 compiled by Wang et al. [34].

4.3. Data Sources and Descriptive Statistics

Considering the availability of data, this research selected data from 30 provinces (municipalities and autonomous regions) in China from 2011–2019 to construct a panel model, and Tibet, Hong Kong, Macao, and Taiwan were excluded since they are lacking a lot of data. The reference coefficients of various energy-discounted standard coals were obtained from the China Energy Statistical Yearbook. One of the indicators of the digital economy, the digital inclusive finance index, was selected from the Digital Finance Research Center of Peking University (Guo et al. [35]). The measured data of other digital economy indicators were from the China City Statistical Yearbook, China Electronic Information Industry Statistical Yearbook, and China Science and Technology Statistical Yearbook. The rest of the data were derived from the China Statistical Yearbook, China Environment Yearbook, China Energy Statistical Yearbook, and China Science and Technology Statistical Yearbook. The indicators related to GDP were deflated according to the constant price in 2011.

Table 2. Descriptive statistics of each variable.

		(1)	(2)	(3)	(4)	(5)
Variables	Meaning	N	Average Value	Standard Deviation	Minimum Value	Maximum Value
lnde	Digital Economy	270	−1.680	0.695	−3.228	−1 × 10−7
slnde	Digital Economy Squared	270	3.304	2.327	1 × 10−14	10.42
lnpce	Carbon Intensity	270	2.237	0.567	1.322	3.826
lnpgdp	GDP Per Capita	270	10.82	0.435	9.706	11.83
lnpopu	Population	270	8.204	0.735	6.342	9.352
lnurb	Urbanization Level	270	−0.572	0.201	−1.051	−0.110
lnindu	Industry Structure	270	0.167	0.380	−0.640	1.655
lnfdi	Openness	270	−4.344	1.126	−9.227	−2.582
lnener	Energy Mix	270	−1.046	0.558	−4.411	−0.376
lntech	Technological Advances	270	10.00	1.433	6.219	13.18
lner	Environmental Regulation	270	−1.743	2.794	−16.12	0.950
lngov	Government Actions	270	1.836	0.309	0.846	2.303

is the result of the descriptive statistics of each variable.

Since the level of urbanization, the degree of opening up to the outside world, and the energy structure are measured by a ratio and the result is less than 1, the mean value is negative after taking the logarithm. Environmental regulation and the digital economy were calculated by calculating the composite index, which is small and less than 1 and also shows negative values after taking the logarithm. The standard deviation showed that, overall, the selected sample data were less discrete, and the mean distribution was more representative. From the control variables, it can be seen that there are great differences in the level of economic development between regions.

5. Analysis of Empirical Results

5.1. Regression Results of the STRIPAT Model

First, to determine whether there is multicollinearity among the variables, the variance inflation factor (VIF) of each variable was calculated. The result reflects the mean value of VIF, which is 6.40. This is less than 10, indicating that there is no serious multicollinearity problem, and the regression can be carried out in the next step. According to the Hausman test results, it is suitable to apply a fixed-effects model. Additionally, to further avoid bias due to different years, this paper used a double fixed-effects model.

In

Table 3. STRIPAT regression results (Note: ***, **, and * indicate passing the 1%, 5%, and 10% significance levels, respectively, as in the following table).

Variables	Model 1 lnce	Model 2 lnce
lnde	−0.1436 *** (−4.0284)	−0.4335 *** (−2.8084)
slnde		−0.0622 * (−1.9299)
lnpgdp	−1.4982 *** (−6.8244)	−1.4354 *** (−6.5057)
lnpopu	0.3575 (0.9179)	0.2884 (0.7417)
lnurb	1.0923 *** (4.6435)	1.0422 *** (4.4299)
lnfdi	−0.0000 (−0.0018)	0.0016 (0.1306)
lner	0.0131 * (1.9306)	0.0151 ** (2.2081)
lngov	0.1855 * (1.7406)	0.1779 * (1.6784)
_cons	13.9247 *** (3.6161)	13.5454 *** (3.5342)
N	270	270
adj.R2	0.835	0.837

, Model 1 shows the linear estimation results of the impact of the digital economy on carbon dioxide emissions in each province. The preliminary judgment is that the digital economy has a remarkable negative correlation with regional carbon dioxide emissions. Adding the quadratic term of the digital economy to the model mainly examines whether there is a nonlinear relationship. From Model 2, it can be seen that the primary term coefficient of the linear term of the digital economy is negatively correlated with regional carbon dioxide emission intensity at a 1% significance level, and the coefficient of the second term is negative at a 10% significance level, which is a preliminary judgment that there is an inverted U-shaped relationship between the two. The results were further tested by the UTEST function, which verified Hypothesis 1.

In terms of control variables, GDP per capita, urbanization level, environmental regulation, and government behavior were statistically significant, and all were consistent with the actual situation. Among them, GDP per capita was significantly negative, which implies that the dependence of China’s GDP growth on carbon dioxide emissions is gradually decreasing. A series of policies adopted by the government for energy conservation and pollution reduction has played a positive role, and the rugged development model is gradually transforming. Additionally, with the increase in income level, residents have more choices, and they are more inclined to choose green and healthy lifestyles and practice the vision of green development. The urbanization level is significantly positive, implying that the development of urbanization fostered the trend toward carbon dioxide emission. With the influx of the population to cities, housing and transportation pressure are increasing, which brings more energy consumption for living and production, thus increasing carbon dioxide emissions. Environmental regulation is positive, indicating that the more serious environmental pollution, the greater the intensity of carbon dioxide emissions. The environmental regulation index in this paper was obtained from the three main pollutants by the entropy method. The burning of fossil fuels such as coal increases carbon dioxide emissions while producing pollutants such as sulfur dioxide and soot. Governmental behavior was represented by the marketization index, which was significantly positive. This means that the higher the marketization, the more carbon dioxide emissions are generated. The possible reason is that in the process of market-oriented reform, some enterprises overly pursue the growth of efficiency and profit and give up the social benefits. Their crude and highly energy-consuming development patterns add to the burden on the environment and increase carbon dioxide emissions.

5.2. Transmission Mechanism Test

Table 4. Results of the conduction mechanism test (*** and ** indicate passing the 1% and 5% significance levels, respectively).

Variables	(1) lnce	(2) lnce	(3) lnce	(4) lnener	(5) lnce
lnde	−0.1227 *** (−3.4761)	−0.1505 *** (−4.1796)	−0.1476 *** (−4.0991)	−0.1942 ** (−2.5796)
lnener	0.1074 *** (3.4890)				0.1256 *** (4.0397)
lntech		0.0329 (1.2696)
lnindu			0.0556 (0.8105)
_cons	12.7673 *** (3.3837)	14.3089 *** (3.7095)	12.4910 *** (2.9458)	10.7722 (1.3239)	13.4627 *** (3.4882)
Control Variables	Yes	Yes	Yes	Yes	Yes
N	270	270	270	270	270
adj.R2	0.843	0.835	0.835	0.459	0.835

is the result of the transmission mechanism test of the above hypothesis. The digital economy has an impact on regional carbon dioxide emission intensity through energy mix transformation and upgrading (lnener). Column 1 adds energy structure variables to the baseline regression, and the effects of digital economy variables and energy structure variables on carbon dioxide emission intensity are significant at the 1% level. Columns 2 and 3 add technological progress variables and industrial structure variables, respectively, based on the baseline regression. From the table, it can be seen that in a statistical sense, the variable of technological progress and industrial structure upgrading do not have an impact on carbon dioxide emissions. Therefore, Hypothesis 2 is also verified.

The results in column 4 show that there is a significant negative relationship between the digital economy and energy structure. The stronger the level of digital economy development, the less coal-based energy consumption, and the more transformative and upgradable the energy structure. Column 5 shows that the effect of energy structure on regional carbon dioxide emission intensity is significantly positive at the 1% level, i.e., the more coal-based energy consumption, the greater the carbon dioxide emission. On the contrary, the transformation and upgrading of energy structure, developing more new energy sources, and adopting more green and clean energy sources can lead to a reduction in regional carbon dioxide emissions. Therefore Hypothesis 3 holds.

5.3. Heterogeneity Analysis

There is regional heterogeneity in the level of digital economy development, and there are also differences in its effect on carbon dioxide emission reduction. In this paper, we divided the regions into eastern and midwestern regions and compared the results with the baseline regression. The eastern region, with its advantageous geographical location and relatively complete digital technology facilities, is the pioneer in the digital economy era to implement the goal of carbon peaking and carbon neutrality.

As shown in

Table 5. Heterogeneity analysis regression results (*** indicates passing the 1% significance level).

Variables	(1) National lnce	(2) Eastern Region lnce	(3) Midwest Region lnce
lnde	−0.1436 *** (−4.0284)	−0.2159 *** (−2.6711)	−0.0236 (−0.5088)
_cons	13.9247 *** (3.6161)	15.5965 *** (4.7821)	1.7375 (0.2843)
Control Variables	Yes	Yes	Yes
N	270	99	171
adj.R2	0.835	0.939	0.852

, the regression results of the eastern region are consistent with the results of the benchmark regression for the whole sample. The coefficient of lnde is negative at the 1% significance level, which means that the carbon dioxide emission reduction effect of the digital economy was significant in the eastern region; the regression results of the central and western regions showed that the estimated coefficient of lnde was not significant, which indicates that the development of the digital economy does not significantly reduce the carbon dioxide emission level in the central and western regions.

5.4. Robustness Test

(1)

Replacement of core explanatory variables: The independent variable in this paper is the level of digital economy development, and since there is a time lag in the effect of digital economy development on carbon dioxide emissions, the regression was conducted using digital economy data with a lag of one period. The results of Model 1 in Table 6 indicate that the coefficient of the level of digital economy development with a lag of one period is −0.0976 and is significantly negative at the 1% level, which verifies the robustness of the baseline regression findings.

(2)

Substitution of explanatory variables: The China Carbon Accounting Database (CEADs) is a powerful database covering a multi-scale carbon accounting inventory that is collected, verified, and compiled under the joint support of the National Natural Science Foundation of China and the International Cooperation Project of the Ministry of Science and Technology. Its carbon dioxide emission data has strong credibility. In this section, the provincial carbon dioxide emission inventories in CEADs from 2001–2019 were selected to construct the data on the carbon dioxide emission intensity of each province to replace the explanatory variables. The results of Model 2 showed that the estimated coefficient of variance was negative at least at the 5% significance level, which is consistent with the baseline regression results. This indicates that the development of the digital economy still significantly reduces regional carbon dioxide emissions.

(3)

Adjustment of the sample period: Here, the robustness test was conducted using the shortened sample period method. The digital economy cannot be divorced from the development of the Internet, and the broadband China strategy is carried out to provide more digital infrastructure for digital economic activities. Its development goal is divided into two phases: to initially build a next-generation national information infrastructure that meets the needs of social development by 2015 and to significantly narrow the gap between the level of broadband network infrastructure development and that of developed countries by 2020. Based on this, the sample period was narrowed to 2015–2019 for re-estimation. The regression results are shown in Model 3, and the regression coefficient of carbon dioxide emission intensity is apparently negative at the 1% level, which again verifies the conclusion of this paper.

Table 6. Robustness regression results (***, **, and * indicate passing the 1%, 5%, and 10% significance levels, respectively).

Variables	The Explanatory Variables Are Lagged by One Period	Substitution of Core Explanatory Variables	Adjust the Sample Period to 2015–2019
	Model 1	Model 2	Model 3
lnde		−0.0830 ** (−2.4825)	−0.1001 *** (−2.6826)
L.lnde	−0.0976 *** (−2.7311)
Cons	12.5399 *** (3.1069)	6.2628 * (1.7334)	6.4145 (1.0124)
Control Variables	Yes	Yes	Yes
Time Fixed Effects	Yes	Yes	Yes
Provincial Fixed Effects	Yes	Yes	Yes
N	240	270	150
adj.R2	0.988	0.833	0.611

Table 6. Robustness regression results (***, **, and * indicate passing the 1%, 5%, and 10% significance levels, respectively).

Variables	The Explanatory Variables Are Lagged by One Period	Substitution of Core Explanatory Variables	Adjust the Sample Period to 2015–2019
	Model 1	Model 2	Model 3
lnde		−0.0830 ** (−2.4825)	−0.1001 *** (−2.6826)
L.lnde	−0.0976 *** (−2.7311)
Cons	12.5399 *** (3.1069)	6.2628 * (1.7334)	6.4145 (1.0124)
Control Variables	Yes	Yes	Yes
Time Fixed Effects	Yes	Yes	Yes
Provincial Fixed Effects	Yes	Yes	Yes
N	240	270	150
adj.R2	0.988	0.833	0.611

5.5. Further Discussion of the Empirical Results

The empirical findings show an inverted U-shaped relationship between digital economy development and regional carbon dioxide emissions, which is consistent with theoretical expectations (see theoretical Hypothesis 1) and also with the view proposed by scholars such as Wang et al. (2023) [36] and Li et al. (2022) [37] that digital economy development is not always conducive to carbon dioxide emission reduction. When combined with reality, early-stage digital economy infrastructure investment, such as the construction and operation of data centers and 5G base stations, would raise demand for energy and carbon dioxide emissions. As a result, a certain carbon increase effect is unavoidable throughout this time period. However, as the digital economy develops further, not only will the production efficiency of traditional sectors improve but so will the efficiency of energy use. As a result, the inverted U-shaped relationship between digital economy development and regional carbon dioxide emissions is likewise rational from a realistic standpoint.

According to the transmission mechanism test results, the growth of the digital economy has a significant impact on carbon dioxide emission reduction, but it has less of an impact on carbon dioxide emission reduction when technological advancement and upgrading of industrial structures are encouraged. There should be no doubt about the former point, given that the digital economy provides a technical foundation for the creation and exploitation of new energy, injects powerful digital power into the transformation of energy structure, and can have a large carbon dioxide emission reduction effect. However, the carbon dioxide emission reduction effect of the digital economy in fostering technical progress and industrial structure upgrading was negligible in this paper, which does not match prior research. This is due to research (such as Gao et al. (2020) [38] and Yang et al. (2023) [39]) demonstrating that technical progress and industrial structure upgrading have specific carbon dioxide emission reduction effects. In fact, careful consideration will reveal that the conclusion of this paper does not directly contradict their conclusion. First and foremost, the development of the digital economy can promote technological progress, but whether such technological progress is necessarily green or energy-saving is unknown. Second, while the expansion of the digital economy can encourage the upgrading of industrial structures, this is a very long evolutionary process that will not be completed overnight.

The results of heterogeneity analysis show that the carbon dioxide emission reduction effect of digital economy development is significant in the eastern region of China but not in the central and western regions. A reasonable explanation for this result is that the development of the digital economy in eastern China has been deep enough, so it can form a significant carbon dioxide emission reduction effect. However, in the central and western regions, the construction of digital infrastructure is relatively backward, the application of digital technology is relatively small, and the development of a digital economy is still in its initial stages (see Wang et al. (2021) [36], Yan et al. (2022) [40], etc. for relevant research).

6. Conclusions and Recommendations

This paper analyzed the impact of the digital economy on regional carbon dioxide emissions from the perspective of influence mechanisms and regional heterogeneity. It incorporated the digital economy, energy structure, and carbon dioxide emissions into the same dimension for analysis, employed the STRIPAT model to investigate the relationship between digital economy development and carbon dioxide emission reduction, and examined the impact of the digital economy on regional carbon dioxide emissions from the perspective of influence mechanisms and regional heterogeneity. The findings suggest that there is an inverted U-shaped link between the expansion of the digital economy and regional carbon dioxide emissions and that the digital economy will first increase and subsequently decrease regional carbon dioxide emissions. The robustness test revealed that the total carbon dioxide emission reduction effect of the digital economy remains after adding the lag time of explanatory variables, substituting the explained variables, and shortening the sample period. Second, while the digital economy has a major inhibitory effect on regional carbon dioxide emissions by driving energy structure transformation, the effect of industrial structure upgrading and technological progress on carbon dioxide emission reduction was negligible. Third, there is clear geographical variation in the influence of the digital economy on regional carbon dioxide emissions. Based on the foregoing conclusions, this study makes the following policy suggestions. (1) In the early stages of digital economy development, focus on low-carbon management and control, improve the market mechanism and incentive mechanism for the digital infrastructure industry to use renewable energy, and strengthen the green orientation of digital infrastructure building. (2) Actively promote energy structure transformation and accelerate the construction of a new energy structure based on new energy as the primary source and coal, oil, and natural gas as auxiliary sources. (3) Increase tax and financial support for the center and western areas, accelerate digital economy development in these regions, and build a nationwide coordinated carbon dioxide emission reduction pattern.

The paper could be improved by finding and using a more accurate measurement method, which can reduce the impact of measurement error on the results of empirical analysis, so it is undoubtedly a direction worthy of further research.