The Carbon Emission Reduction Effect of the Digital Economy: Mechanism Reconstruction Based on the Suppression Effect—A Case Study of the Pearl River Delta Urban Agglomeration

Abstract

With the continuous expansion of the digital economy, its share in China’s overall economy has been steadily increasing. Against the backdrop of the national “dual-carbon” goals, an important question arises: how does the digital economy contribute to carbon reduction? This study selects panel data from nine cities in the Pearl River Delta (PRD) urban agglomeration between 2011 and 2023. The development level of the digital economy is measured using the entropy weight method and an index system. A two-way fixed effects model and a mediation effect model are then employed to empirically examine the relationship and mechanisms between the digital economy and urban carbon emissions. The main findings are as follows: (1) the development of the digital economy exerts a significant negative regulatory effect on carbon emissions, which remains robust after a series of tests; (2) heterogeneity analysis reveals that the inhibitory effect of the digital economy on carbon emissions is more evident in economically advanced cities, and the development level of metropolitan areas significantly influences this relationship; (3) mechanism analysis indicates that stronger environmental regulation significantly enhances the carbon reduction effect of the digital economy; and (4) the scale of e-commerce in the PRD plays a “suppression effect”, offsetting the original carbon-increasing effect of the digital economy and emerging as the key factor underlying its net carbon-reducing impact. Based on these results, the paper provides policy recommendations to better leverage the digital economy in supporting regional carbon reduction.

1. Introduction

With the acceleration of global industrialization and the continuous improvement in living standards, the problem of global warming caused by carbon emissions has become increasingly serious. Countries and international organizations around the world are actively seeking solutions to environmental challenges. As the world’s largest developing country, China faces tremendous pressure from the rapid growth of carbon emissions. In response, President Xi Jinping announced at the 75th session of the United Nations General Assembly that China will strive to peak carbon emissions by 2030 and achieve carbon neutrality by 2060, forming the “dual-carbon goals”.

Meanwhile, the concept of “new quality productive forces” introduced in 2023 highlights the role of data and the digital economy in reshaping production and driving green development. The digital economy has become a critical force in achieving carbon reduction, industrial upgrading, and sustainable growth [1]. As an essential component of new quality productive forces, data elements play a vital role in optimizing resource allocation and enhancing the flow of other production factors. The digital economy, with data as its core production element, has gradually become a critical driving force in fostering the development of these new productive forces.

According to the China Academy of Information and Communications Technology (CAICT) Report on the Development of China’s Digital Economy (2024), the scale of China’s digital economy reached 53.9 trillion yuan in 2023, accounting for 42.8% of GDP, with both scale and proportion continuously increasing. Meanwhile, the White Paper on Digital Carbon Neutrality highlights the important role that the digital economy can play in addressing global warming. Digital technologies help build low-carbon, safe, and efficient energy systems. They also promote industrial restructuring and provide digitalized, networked, and intelligent tools for advancing green development. The digital economy is thus becoming an important force in driving new quality productive forces and achieving the dual-carbon strategy.

Therefore, clarifying the relationship between the digital economy and urban carbon emissions, understanding its mechanisms of action and development pathways, and exploring its intrinsic spatial-geographical linkages are of great theoretical and practical significance. Such work will not only inform the formulation of urban carbon reduction policies but also help balance the relationship between economic growth and green low-carbon development.

2. Literature Review

The concept of the digital economy was first introduced in 1996. The digital economy was described as a new mode of social development driven by information and communication technologies. It can reshape the structure of economic development and alter modes of social production [2]. Although the digital economy has received increasing attention over time, there is still no unified definition. Current mainstream international studies typically define it from four perspectives: economic sectors, economic activities, economic forms, and technological characteristics [3].

In China, the most widely recognized definition of the digital economy is drawn from the G20 Digital Economy Development and Cooperation Initiative, which describes it as economic activities that use digital knowledge and information as key production factors, modern information networks as critical carriers, and effective application of information and communication technologies to improve efficiency and optimize economic structure. Furthermore, the 14th Five-Year Plan for National Economic and Social Development and the Long-Range Objectives Through the Year 2035 emphasizes that China’s digital economy development requires the coordinated promotion of both industrial digitalization and digital industrialization [4]. However, most current research does not effectively distinguish between these two dimensions, but rather analyzes the digital economy as a generalized and holistic concept.

Regarding the relationship between the digital economy and carbon emissions, academic perspectives are highly divided. On the one hand, many scholars argue that the development of the digital economy can significantly reduce carbon emissions and mitigate tensions between economic growth and environmental challenges. For example, the digital economy can transform the production methods and organizational structures of traditional industries, accelerate industrial integration, and foster new industries and business models, thereby indirectly reducing carbon emissions through industrial restructuring [5]. The development of digital technology may also significantly reduce urban carbon emission intensity in China, and this effect tends to strengthen over time [6]. On the other hand, some studies suggest that the digital economy has a “carbon-increasing effect”. Since its development relies on infrastructure construction, it inevitably results in additional emissions. Moreover, the expansion of the digital economy stimulates domestic consumption and enlarges the overall economic scale, which unavoidably increases social carbon emissions—emissions that cannot be fully offset by the reductions brought about by digital technologies [7]. Some scholars integrate the above two perspectives and argue that the relationship between the digital economy and carbon emissions follows an inverted U-shaped curve. This is because the development of the digital economy initially enhances the efficiency of traditional energy-intensive industries, thereby increasing their energy demand and carbon emissions. Only through continuous digital transformation and restructuring of traditional industries can the digital economy ultimately exert its inhibitory effect on carbon emissions [8]. Based on the existing literature on the relationship between the digital economy and carbon emissions, most current studies are conducted at the national or provincial level and lack an in-depth exploration of intra-urban agglomeration differences. In addition, there has been insufficient explanation of the phenomenon whereby the digital economy contributes to emission reductions in some regions while potentially increasing emissions in others.

In terms of mechanisms, most existing studies agree that the digital economy affects urban carbon emissions by adjusting industrial structure, reducing energy intensity, and promoting technological innovation. For instance, these variables were used to analyze the mechanism by which the digital economy impacts urban carbon emissions [9,10]. Some studies also take industrial agglomeration as a mediating variable, suggesting that the promotion of industrial agglomeration driven by the digital economy may hinder the reduction in urban carbon emission levels [11]. While such mediators shed light on the relationship between the digital economy and emissions, they fail to capture regional heterogeneity.

The Pearl River Delta (PRD) urban agglomeration, one of China’s largest and most dynamic economic regions, includes Guangzhou, Shenzhen, Zhuhai, Foshan, Huizhou, Dongguan, Zhongshan, Jiangmen, and Zhaoqing. The PRD’s economy is highly diverse, with deep integration between the digital industry, manufacturing, and services. Accounting for less than 0.9% of the national land area, it generates over 9% of China’s GDP [12]. Within the process of industrial digitalization, e-commerce—particularly cross-border and B2B e-commerce—has emerged as a core application scenario, with its scale in 2023 representing one-fifth of the national total. This makes e-commerce an unavoidable topic when studying the PRD digital economy [13]. Meanwhile, the PRD is also a pioneer region for China’s low-carbon transition. Guangdong Province was among the earliest pilot regions for the national carbon market and launched a regional carbon-inclusive mechanism in 2025. The region is also characterized by strong inter-city economic linkages, such as the Guangzhou–Foshan integration and Shenzhen–Dongguan industrial cooperation [14]. At the policy level, the PRD is simultaneously subject to multiple overlapping policy frameworks, including the “dual-zone construction”, “carbon peaking action plan”, and “digital government reform”. Therefore, focusing on specific industries within the PRD urban agglomeration and considering the micro-mechanisms shaped by coordinated digital economy planning at the city-area level addresses a gap in previous studies. The unique regional characteristics of the PRD justify further exploration of how the digital economy influences carbon emissions in this context.

In summary, although previous studies have examined the carbon reduction potential of the digital economy, several research gaps remain. First, the existing literature rarely focuses on heterogeneity within urban agglomerations, particularly in economically diverse city areas such as the PRD. Second, as a region where multiple policy initiatives converge, the PRD has not been thoroughly studied with respect to the interaction between environmental regulation and digital economy development. Third, the role of e-commerce in shaping the relationship between the digital economy and carbon emissions remains unclear, making it difficult to propose targeted policy recommendations for regions such as the PRD, where e-commerce plays a central role.

Building upon the identified gaps in the literature, this study seeks to answer the following research questions:

RQ1: To what extent does the development of the digital economy influence urban carbon emissions within economically diverse city areas such as the PRD?

RQ2: How does environmental regulation moderate the relationship between the digital economy and carbon emissions in the PRD region?

RQ3: Does the expansion of e-commerce play a suppression effect by offsetting the potential carbon-increasing impact of the digital economy?

Therefore, this paper selects nine PRD cities as its research sample and employs a two-way fixed-effects model to empirically analyze the relationship between the digital economy and carbon emissions. It further examines the moderating role of environmental regulation and the mediating role of e-commerce. The potential contributions of this study include: (1) constructing a comprehensive index system of digital economy development—covering digital infrastructure, digital industry scale, and digital innovation capacity—and applying the entropy weight method to measure each city’s digital economy development level; (2) testing environmental regulation intensity as a moderating variable to assess how policy enhances or weakens the relationship between the digital economy and carbon emissions; and (3) selecting e-commerce scale as a characteristic mediating variable of the PRD to uncover its transmission mechanism.

3. Theoretical Framework and Hypotheses

3.1. Mechanisms Through Which the Digital Economy Affects Regional Carbon Emissions

The inhibitory effect of the digital economy on carbon emissions can be analyzed from micro, meso, and macro perspectives.

At the micro level, the digital economy—comprising technologies such as the Internet of Things (IoT), cloud computing, and big data—enhances the mobility of production factors by overcoming temporal and spatial constraints, thereby reducing resource losses during circulation and improving resource utilization efficiency. This, in turn, contributes to emission reductions. Furthermore, artificial intelligence and big data enable more accurate predictions of consumer behavior and actual demand, allowing enterprises to avoid misallocation of resources and lower emissions. By building digital platforms, firms can obtain more precise data on both supply and demand, better matching them and reducing mismatches of resources. This improves total factor productivity and suppresses carbon emissions [15].

At the meso level, the development of the digital economy accelerates the transformation and upgrading of traditional industries. Because data serves as the key production factor in the digital economy, its promotion of industrial upgrading is inherently linked to green and clean production. The process of industrial restructuring reallocates resources from low-efficiency to high-efficiency sectors, improving energy efficiency and lowering emissions [16].

At the macro level, the digital economy facilitates the establishment of carbon markets, accelerates the deployment of carbon trading pilots, and enables governments to monitor energy market price fluctuations more effectively. This allows for better regulation and control of total carbon emissions, thus achieving emission reduction goals [17].

To address RQ1, we propose the following hypothesis:

Hypothesis H1:

In the PRD urban agglomeration, the digital economy exerts a significant inhibitory effect on carbon emissions.

3.2. The Moderating Role of Environmental Regulation

Environmental regulation refers to government policies and regulations that limit excessive pollution and energy consumption during production, thereby achieving goals such as environmental protection, carbon reduction, and sustainable development. When properly designed, environmental regulation can guide enterprises to redesign production modes in line with green development principles, thereby positively moderating the relationship between the digital economy and carbon emissions.

Specifically, on one hand, stricter environmental regulations impose higher compliance costs on high-carbon industries, which enhances the “cost-reducing” advantage of digital technologies. To comply, enterprises increasingly rely on IoT and other digital tools to monitor emissions in real time, optimize production processes, and adjust strategies to mitigate risks of non-compliance [18]. The stronger the regulation, the greater the incentive for firms to apply digital technologies to emission reduction, strengthening the inhibitory role of the digital economy. On the other hand, governments can use fiscal incentives such as tax reductions and subsidies to channel resources toward digital and green industries. In this context, firms demand advanced abatement technologies, while government support for digital environmental R&D lowers innovation barriers, enhancing firms’ efficiency in emission reduction.

To address RQ2, we propose the following hypothesis:

Hypothesis H2:

Environmental regulation positively moderates the impact of the digital economy on carbon emissions in the PRD urban agglomeration.

3.3. The Mediating Role of E-Commerce

3.3.1. Impact of the Digital Economy on E-Commerce

The digital economy is closely associated with e-commerce, as its development fosters the expansion of e-commerce. On one hand, digital technologies reshape e-commerce models by breaking the temporal and spatial constraints of traditional trade, shortening transaction processes, and reducing time costs. On the other hand, the emergence of third-party payment platforms, driven by the digital economy, has not only facilitated secure and convenient transactions but also lowered transaction costs. Additionally, the rapid diffusion and replication of data enhance economies of scale in e-commerce development [19].

3.3.2. Impact of E-Commerce on Carbon Emissions

E-commerce also plays an important role in reducing carbon emissions. First, it reduces multi-tier distribution and unnecessary logistics links in traditional trade, improving supply chain efficiency. Through IoT and big data-enabled smart scheduling, it reduces energy consumption in warehousing and transportation [20]. Second, e-commerce promotes the circulation of green products, such as new energy goods and low-carbon consumer goods, expanding their market share and compelling the production side to shift toward low-carbon and green outputs.

However, prior studies have also noted that the early expansion of the digital economy may increase emissions due to infrastructure construction and logistics growth. Such “scale effects” can interact with the emission reduction effects of e-commerce, potentially resulting in a suppression effect, where e-commerce offsets or conceals the carbon-increasing impact of initial digital economy growth. This possibility aligns with the rebound and substitution effects in environmental economics [21].

To address RQ3, we propose the following hypothesis:

Hypothesis H3:

E-commerce serves as a transmission mechanism in the causal chain between the digital economy and carbon emissions in the PRD urban agglomeration.

• H3a: The development of the digital economy reduces carbon emission intensity by expanding the scale of e-commerce in the PRD urban agglomeration.
• H3b: E-commerce suppresses the original carbon-increasing effect of the digital economy, thereby creating an apparent net reduction in carbon emissions in the PRD urban agglomeration.

4. Methodology

4.1. Model Construction

4.1.1. Baseline Regression Model

To examine the relationship between the digital economy and urban carbon emissions, this study adopts a two-way fixed-effects model as the baseline regression framework. The model is specified as follows:

$${\mathrm{ce}}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{dige}}_{\mathrm{it}}+\sum {\mathsf{\beta }}_{\mathrm{k}}\mathrm{C}{\mathrm{V}}_{\mathrm{itk}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\sigma }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(1)

To further investigate the moderating role of environmental regulation, the model incorporates the environmental regulation variable:

$${\mathrm{ce}}_{\mathrm{it}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{dige}}_{\mathrm{it}}+{\mathsf{\alpha }}_2{\mathrm{regu}}_{\mathrm{it}}+{\mathsf{\alpha }}_3{\mathrm{dige}}_{\mathrm{it}}\times {\mathrm{regu}}_{\mathrm{it}}+\sum {\mathsf{\beta }}_{\mathrm{k}}\mathrm{C}{\mathrm{V}}_{\mathrm{itk}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\sigma }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(2)

${\mathrm{ce}}_{\mathrm{it}}$ represents the carbon emissions of region i in year t. ${\mathrm{dige}}_{\mathrm{it}}$ denotes the level of digital economy development in region i in year t; ${\mathrm{regu}}_{\mathrm{it}}$ indicates the intensity of environmental regulation in the PRD. ${\mathrm{dige}}_{\mathrm{it}}\times {\mathrm{regu}}_{\mathrm{it}}$ is the interaction term used to test whether environmental regulation strengthens the emission reduction effect of the digital economy. ${\mathrm{CV}}_{\mathrm{it}}$ is a set of control variables. ${\mathsf{\mu }}_{\mathrm{i}}$, ${\mathsf{\sigma }}_{\mathrm{t}}$, and ${\mathsf{\epsilon }}_{\mathrm{it}}$ represent individual effects, time effects, and random error terms, respectively.

4.1.2. Mediation Effect Model

Based on the hypotheses, the digital economy may influence carbon emissions through its impact on the scale of e-commerce, suggesting the existence of a mediating effect. To test this, a mediation effect model is constructed as follows:

$${\mathrm{M}}_{\mathrm{it}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{dige}}_{\mathrm{it}}+\sum {\mathsf{\beta }}_{\mathrm{k}}\mathrm{C}{\mathrm{V}}_{\mathrm{itk}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\sigma }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(3)

$${\mathrm{ce}}_{\mathrm{it}}={\mathsf{\pi }}_0+{\mathsf{\pi }}_1{\mathrm{dige}}_{\mathrm{it}}+{\mathsf{\pi }}_2{\mathrm{M}}_{\mathrm{it}}+\sum {\mathsf{\beta }}_{\mathrm{k}}\mathrm{C}{\mathrm{V}}_{\mathrm{itk}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\sigma }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{it}}$$

(4)

${\mathrm{M}}_{\mathrm{it}}$ represents the mediating variable, while other variables retain the same definitions as in Equation (1).

4.2. Variable Description

4.2.1. Dependent Variable

The dependent variable in this study is urban carbon emissions (ce). The estimation of city-level carbon emissions generally follows the emission factor method:

$$\mathrm{C}\mathrm{a}\mathrm{r}\mathrm{b}\mathrm{o}\mathrm{n} \mathrm{e}\mathrm{m}\mathrm{i}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n}\mathrm{s}=\mathrm{A}\mathrm{c}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{i}\mathrm{t}\mathrm{y} \mathrm{d}\mathrm{a}\mathrm{t}\mathrm{a}\times \mathrm{E}\mathrm{m}\mathrm{i}\mathrm{s}\mathrm{s}\mathrm{i}\mathrm{o}\mathrm{n} \mathrm{f}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{o}\mathrm{r}$$

(5)

Activity data include all production activities that generate CO₂ emissions. This study defines such activities as consumption of liquefied petroleum gas, natural gas, thermal energy, and transportation. The total carbon emissions are obtained by aggregating emissions from these activities [22].

4.2.2. Core Independent Variable

The core independent variable is the level of digital economy development (dige), measured through an index system approach. As no consensus exists on how to measure the digital economy, and different methods emphasize different aspects, this study selects three primary dimensions: digital infrastructure, digital industry scale, and digital innovation capacity [23]. Digital infrastructure is measured by mobile phone penetration rate and Internet penetration rate [24]. Digital industry scale is measured by the development of telecommunications, smart logistics, and digital finance [25]. Digital innovation capacity is measured by human innovation capacity and economic innovation capacity [26]. The selection is based on the “Four Digitalizations” framework proposed by the China Academy of Information and Communications Technology (CAICT) for the digital economy, which includes industrial digitalization, digital industrialization, digital governance, and data value activation. Therefore, the construction of digital economy indicators is built upon the above four dimensions. On this basis, the Digital Inclusive Finance Index released by the Digital Finance Research Center of Peking University is incorporated to form a comprehensive evaluation system for the digital economy.

The entropy weight method is employed to calculate weights for each secondary indicator to minimize subjectivity and avoid missing information. The weights are presented in

Table 1. Index system and weights for measuring the level of digital economy development.

Primary Indicator	Secondary Indicator	Description (Unit)	Direction	Weight
Digital Infrastructure	Mobile phone penetration	Number of mobile subscribers (10,000 households)	+	0.140
	Internet penetration	Number of Internet users (10,000 households)	+	0.139
Digital Industry Scale	Telecommunications development	Telecom business revenue (10,000 yuan)	+	0.141
	Smart logistics development	Express delivery revenue (10,000 yuan)	+	0.148
	Digital finance development	Digital Inclusive Finance Index	+	0.136
Digital Innovation Capacity	Human innovation base	Number of employees in IT and software (10,000 people)	+	0.150
	Economic innovation base	Expenditure on science and technology (10,000 yuan)	+	0.147

Note: “+”refers that the indicator has a positive impact to dige.

.

4.2.3. Control Variables

• Population density (den): Calculated as the ratio of total urban population at year-end to administrative area [27]. Population scale changes may directly increase energy consumption and indirectly affect emissions through consumption patterns.
• Economic development level (lnpGDP): Measured by per capita GDP (log-transformed) [28]. Following the Environmental Kuznets Curve (EKC), environmental quality first deteriorates and then improves with rising income.
• Urbanization rate (urb): Measured as the proportion of urban population to total population. Higher urbanization rates are associated with high energy consumption and increased emissions [29].
• Openness (open): Measured as the ratio of utilized foreign direct investment (converted into RMB at annual average exchange rate) to regional GDP. Openness affects economic vitality, industrial structure, and emissions.
• Industrialization level (lnind): Measured by the number of industrial enterprises (log-transformed) [30]. Industrial structure (resource-intensive vs. technology-intensive) significantly influences carbon emissions.

4.2.4. Moderating Variable

Environmental regulation (regu) can influence the relationship between the digital economy and carbon emissions through cost constraints and innovation incentives. The intensity of environmental regulation is measured using a text-mining approach applied to government work reports, extracting the frequency of environment-related keywords. Text-based measures have been increasingly adopted in recent studies as they better reflect policy emphasis and local variations compared to traditional single indicators such as pollution treatment investment [31].

4.2.5. Mediating Variable

The mediating variable is the scale of e-commerce (ec). The digital economy reduces costs and fosters economies of scale in e-commerce, which in turn helps reduce carbon emissions through more efficient logistics and by promoting green consumption. This study measures e-commerce scale using regional e-commerce transaction volume [32]. This indicator directly reflects the overall level of regional e-commerce activity and has been widely used in previous studies to capture the expansion of the digital economy on the consumer side and its potential impact on carbon emissions.

The definitions and units of variables can be found in

Table 2. Key variables: Definitions, units and indicators used in the study.

Variable Type	Definition (Variable Name)	Unit	Indicator
Dependent variable	Carbon emissions (ce)	Mt	CO2 emissions
Core independent variable	Digital economy level (dige)	-	Index value
Control variables	Population density (den)	10,000 people/km2	Population/Area
	Economic development level (lnpGDP)	-	GDP/Population
	Urbanization rate (urb)	-	Urban population/Total population
	Openness (open)	-	FDI/GDP
	Industrialization level (lnind)	-	Number of industrial firms
Mediating variable	E-commerce scale (ec)	billion yuan	Transaction volume
Moderating variable	Environmental regulation (regu)	-	Keyword frequency ratio

.

4.2.6. Data Sources and Descriptive Statistics

The sample period covers 2011–2023 for nine cities in the PRD urban agglomeration, yielding a total of 117 observations. Data were collected from the China City Statistical Yearbook and the National Bureau of Statistics, supplemented with information from Markdata. Carbon emission factors and oxidation factors were obtained from the 2006 IPCC Guidelines for National Greenhouse Gas Inventories. The Digital Inclusive Finance Index was sourced from Ant Financial and Peking University [33]. Descriptive statistics of the variables are presented in

Table 3. Descriptive statistics of variables.

Variables	Obs.	Mean	Std. Dev.	Min	Max
ce	117	3.136	1.481	1.477	7.178
dige	117	0.256	0.207	0.0178	0.886
den	117	855.795	557.229	276	3005
lnpGDP	117	11.446	0.436	10.424	12.223
urb	117	83.140	14.295	53.09	99.800
open	117	0.0263	0.0176	0.00239	0.102
lnind	117	8.153	0.767	6.795	9.614
ec	117	3633.130	5479.366	55.403	26,117.040
regu	117	0.00729	0.00208	0.00381	0.0131

.

5. Results and Analysis

5.1. Baseline Regression Analysis of the Impact of the Digital Economy on Carbon Emissions

To ensure the robustness of the empirical analysis, this study gradually incorporates control variables into the regression model. The results are presented in

Table 4. Baseline regression results: Impact of the digital economy on carbon emissions (core variables).

	ce (1)	ce (2)	ce (3)	ce (4)	ce (5)	ce (6)
dige	−1.494 *** (−4.15)	−2.191 *** (−4.16)	−2.153 *** (−4.11)	−2.153 *** (−4.11)	−2.169 *** (−4.12)	−2.163 *** (−4.08)
den	-	0.000322 ** (2.13)	0.000316 ** (2.10)	0.000316 ** (2.10)	0.000314 ** (2.06)	0.000326 ** (2.09)
Observations	117	117	117	117	117	117
R2	0.985	0.985	0.985	0.985	0.985	0.985
Time Fixed Effects	Y	Y	Y	Y	Y	Y
Prefecture-Level City Fixed Effects	Y	Y	Y	Y	Y	Y

Note: t-statistics in parentheses. * p < 0.1, ** p < 0.05, *** p < 0.01. “Y” refers to “Yes”; the same applies to all subsequent tables.

and

Table 5. Baseline regression results: Impact of control variables on carbon emissions.

	ce (1)	ce (2)	ce (3)	ce (4)
lnpGDP	0.114 (0.50)	0.114 (0.50)	0.124 (0.53)	0.139 (0.59)
urb	-	0.0543 *** (14.61)	0.0544 *** (14.61)	0.0565 *** (6.88)
open	-	-	0.619 (0.36)	0.473 (0.28)
lnind	-	-	-	−0.545 (−0.31)
Constant	2.524 (1.00)	−2.519 (−1.12)	−2.654 (−1.15)	−2.545 (−1.07)
Observations	117	117	117	117
R2	0.985	0.985	0.985	0.985
Time Fixed Effects	Y	Y	Y	Y
Prefecture-Level City Fixed Effects	Y	Y	Y	Y

.

Analysis of the Core Independent Variable: As shown in the first row of Table 4, after sequentially adding control variables, the coefficient of the core explanatory variable—the level of digital economy development—remains negative and statistically significant at the 1% level. This indicates that the development of the digital economy effectively reduces carbon emissions in the PRD urban agglomeration. For example, in regression model ce (6) in Table 4, after including all control variables, the coefficient of the digital economy variable is −2.163, suggesting that a 1% increase in the level of digital economy development leads to a 2.163% decrease in carbon emissions, significant at the 1% level. This finding confirms Hypothesis H1 and is consistent with studies such as [5,6], which emphasize that digitalization can enhance energy efficiency and reduce carbon intensity through technological upgrading.

Analysis of Control Variables (Table 5):

• Population density: The coefficients are consistently positive and significant at the 5% level, indicating that a 1% increase in urban population density raises carbon emissions by approximately 0.000326%. Higher density increases daily life and transportation demands, thereby raising emissions.
• Economic development level: Per capita GDP positively influences urban emissions, with a 1% increase in GDP per capita associated with a 0.139% increase in emissions, though the effect is not significant.
• Urbanization rate: The coefficient is significantly positive at the 1% level. A 1% increase in urbanization raises carbon emissions by 0.0565%, suggesting that infrastructure expansion and population inflows during urbanization elevate emissions.
• Openness: A 1% increase in openness raises emissions by 0.473%, though the effect is not statistically significant. This may be linked to increased foreign investment and industrial activity.
• Industrialization: Interestingly, higher industrialization levels are associated with lower emissions, though insignificantly. This may be because PRD industries are largely technology-intensive, with digital technologies accelerating industrial upgrading and emission reductions. This observation echoes the view of [9] that advanced industrial upgrading can decouple growth from emissions.

5.2. Robustness Tests

To further validate the impact of the digital economy on carbon emissions in the PRD and ensure reliability of results, three robustness checks are conducted.

5.2.1. Replacing the Dependent Variable

Carbon emissions per capita are used to replace total emissions as the dependent variable [34]. As shown in

Table 6. Robustness tests: Digital economy consistently reduces carbon emissions across alternative specifications.

	Replace the Dependent Var. (1)	Shorten the Sample Period (2)	Lag 1 Period (3)	Lag 2 Periods (4)	Lag 3 Periods (5)
dige	−0.00436 *** (−3.80)	−1.991 *** (−3.59)	−1.722 ** (−2.56)	−1.560 ** (−2.24)	−1.477 * (−1.82)
Constant	0.00205 (0.29)	−0.932 (0.741)	−1.389 (−0.55)	−0.928 (−0.34)	−2.516 (−1.18)
Observations	117	99	108	99	90
R2	0.956	0.989	0.986	0.988	0.991
Control Variables	Y	Y	Y	Y	Y
Time Fixed Effects	Y	Y	Y	Y	Y
Prefecture-Level City Fixed Effects	Y	Y	Y	Y	Y

(column 1), the coefficient of the digital economy remains significantly negative at the 1% level, confirming robustness.

5.2.2. Shortening the Sample Period

Since China’s digital economy began to grow rapidly after 2013, the sample is restricted to 2013–2023. As shown in Table 6 (column 2), the coefficient is −1.991 and significant at the 1% level, consistent with baseline findings.

5.2.3. Lag Effect Tests

To address potential endogeneity from reverse causality or omitted variables, the digital economy variable is lagged by 1–3 periods [35]. As shown in Table 6 (columns 3–5), the coefficients remain significantly negative, though the magnitude decreases with longer lags, suggesting that the inhibitory effect of the digital economy weakens over time but remains robust.

5.3. Heterogeneity Analysis

5.3.1. Heterogeneity by Economic Development Level

Although the PRD overall has a high level of economic development, there are internal disparities among cities. To test for heterogeneity, cities are divided into high- and low-economic-development groups based on the median GDP per capita (2011–2023). Results in

Table 7. Heterogeneity analysis: Digital economy reduces emissions in high-development cities; GFZ cluster benefits most.

	High-Development Cities (1)	Low-Development Cities (2)	GFZ (3)	SHD (4)	ZZJ (5)
dige	−1.156 ** (−2.33)	4.038 ** (2.28)	−1.069 * (−1.76)	−0.379 (−0.48)	2.758 (1.18)
Constant	29.701 *** (9.2)	−0.482 (−0.19)	−0.611 (−0.15)	31.667 *** (12.45)	−63.520 *** (−3.01)
Observations	52	65	39	39	39
R2	0.997	0.924	0.998	0.989	0.949
Control Variables	Y	Y	Y	Y	Y
Time Fixed Effects	Y	Y	Y	Y	Y
Prefecture-Level City Fixed Effects	Y	Y	Y	Y	Y

Note: GFZ refers to Guangzhou–Foshan–Zhaoqing, SHD refers to Shenzhen–Huizhou–Dongguan, and ZZJ refers to Zhuhai–Zhongshan–Jiangmen.

(columns 1–2) show that in high-development cities, the coefficient of the digital economy is significantly negative at the 5% level, confirming its emission-reducing role. In contrast, in low-development cities, the coefficient is significantly positive at the 5% level, implying that digital economy expansion in less-developed areas may initially increase emissions. This may result from reliance on energy-intensive growth strategies and high infrastructure costs in early stages of digital economy development.

5.3.2. Heterogeneity by Metropolitan Areas

According to the Guangdong Metropolitan Area Spatial Planning Coordination Guidelines, the PRD is divided into three metropolitan areas: “Guangzhou–Foshan–Zhaoqing (GFZ)”, “Shenzhen–Huizhou–Dongguan (SHD)” and “Zhuhai–Zhongshan–Jiangmen (ZZJ)”. Regression results in Table 7 (columns 3–5) show that only the GFZ metropolitan area exhibits a significantly negative coefficient for the digital economy, whereas results for SHD and ZZJ are insignificant. The likely explanation is that GFZ, as one of the earliest integrated metropolitan areas, benefits from greater synergy in infrastructure, industrial division, and environmental policies, which amplifies the emission reduction effect of the digital economy. By contrast, SHD and ZZJ remain in earlier development stages, where expansion is prioritized over green integration, and weaker inter-city coordination reduces the effect.

5.4. Mechanism Effect Tests

The baseline regressions and robustness checks confirm that the digital economy has a significant overall inhibitory effect on carbon emissions. To further examine the transmission mechanisms, both moderating and mediating variables are introduced. The results are presented in

Table 8. Mechanism effect tests for moderating effect of environmental regulation and mediating role of e-commerce scale.

	ce (1)	ec (2)	ce (3)
dige	−1.474 *** (−2.90)	45,828.910 *** (16.01)	1.899 ** (2.09)
regu	−21.319 * (−1.66)	-	-
ec	-	-	−0.0000886 *** (−5.20)
interact	206.494 *** (3.63)	-	-
Constant	1.237 (0.48)	25780.360 ** (2.32)	−0.611 (−0.15)
Observations	117	117	117
R2	0.988	0.969	0.989
Control Variables	Y	Y	Y
Time Fixed Effects	Y	Y	Y
Prefecture-Level City Fixed Effects	Y	Y	Y

.

5.4.1. Moderating Effect of Environmental Regulation

Environmental regulation can reshape costs and stimulate innovation. As shown in Table 8 (column 1), the coefficient of the digital economy is −1.474, significant at the 1% level, consistent with earlier findings. Environmental regulation alone also has a negative coefficient of −21.319, significant at the 10% level, suggesting that stronger regulation tends to reduce emissions, though with less stability. Importantly, the interaction term between the digital economy and regulation is 206.494, significantly positive at the 1% level, indicating that stronger environmental regulation amplifies the emission reduction effect of the digital economy. This finding supports Hypothesis H2. This finding aligns with the Porter hypothesis, which argues that well-designed environmental regulation can induce technological innovation and offset compliance costs. It is also consistent with studies such as [36], which highlight that regulation intensity can strengthen the low-carbon transformation enabled by digitalization. However, our results extend this literature by showing that when regulation intensity is higher, the digital economy’s emission-reducing role becomes significantly stronger in the PRD urban agglomeration, a city-area context often overlooked in prior research.

5.4.2. Mediating Role of E-Commerce Scale

To test the transmission role of e-commerce, two regressions are conducted and shown in Table 8 (column 2–3).

• In pathway a (X→M), the digital economy significantly promotes e-commerce scale (coefficient = 45,828.91, p < 0.01).
• In pathway b (M→Y), e-commerce scale significantly reduces emissions (coefficient = −0.0000886, p < 0.01).
• When including the mediating variable, the direct effect of the digital economy on emissions shifts from significantly negative (c = −2.163) to significantly positive (c’ = 1.899), while the indirect effect (a × b ≈ −4.06) offsets the positive direct effect.

The transmission mechanism is illustrated in Figure 1. The digital economy (X) significantly increases the scale of e-commerce (M), which in turn significantly reduces carbon emissions (Y). Notably, when the mediating variable is introduced, the direct effect of the digital economy on emissions changes sign—from negative to positive—indicating that while digital development tends to increase emissions, the strong carbon reduction effect of e-commerce offsets and suppresses this tendency. Since the indirect and direct effects have opposite signs, a suppression effect exists [21]. For the PRD, the development of the digital economy may increase carbon emissions to a certain extent. However, the carbon reduction effect of e-commerce is more significant, which suppresses the actual carbon increase caused by the digital economy itself, ultimately resulting in the manifestation that the digital economy can inhibit carbon emissions.

The verification of the aforementioned suppression effect challenges the conclusion of “the inherently low-carbon nature of the digital economy” obtained in most previous studies. Combined with the regional characteristics of the PRD urban agglomeration, there may be the following reasons for such a conclusion. First, in core cities of the PRD, such as Dongguan and Foshan, the main body of the digital economy lies in the intelligentization of the manufacturing industry, which incurs relatively high direct carbon costs. The consumption of industrial Internet platforms and intelligent production equipment has caused digital technology itself to intensify carbon emissions in traditional industrial bases, which also explains why the direct effect coefficient c’ is positive. At the same time, Guangzhou and Shenzhen are two national-level e-commerce demonstration cities in the PRD region, and their e-commerce penetration rates and logistics efficiency are significantly higher than the national average. On this basis, e-commerce platforms can generate a huge scale of carbon reduction effects through supply chain optimization and logistics network sharing, thereby offsetting the original carbon costs generated by the digital economy. Therefore, Guangzhou and Shenzhen, as the actual e-commerce centers in the PRD, can better benefit from carbon reduction; while other manufacturing cities need to undertake the construction of data centers and bear the cost of increased carbon emissions, they also share the overall carbon reduction achievements through the e-commerce network. As a result, a net inhibitory effect is manifested at the macro level of the urban agglomeration scale. Thus, Hypothesis H3b is verified.

6. Conclusions and Policy Recommendations

This study investigates how the digital economy influences carbon emissions within the PRD urban agglomeration under China’s “dual-carbon” strategy. Compared with existing studies that mainly focus on national or provincial levels [8], this study provides new evidence from intra-urban agglomerations, revealing spatial heterogeneity within the PRD. Moreover, by identifying the suppression effect of e-commerce and the moderating role of environmental regulation, this paper expands the theoretical framework of the digital economy–carbon emissions nexus. The findings provide important implications for both policy and practices. The results give valuable insights for local governments seeking to design region-specific digital economy strategies and low-carbon transition policies. They also provide practical guidance for enterprises in accelerating digital transformation, optimizing resource use and reducing emissions through technological innovation. By clarifying the mechanisms through which the digital economy contributes to emission reduction, this study supports the realization of sustainable, inclusive and low-carbon development.

By analyzing panel data for nine cities in the region from 2011 to 2023, the main conclusions are as follows: (1) Using a two-way fixed effects model, it is found that the development of the digital economy in the PRD exerts a significant negative moderating effect on carbon emissions, and this conclusion remains valid after a series of robustness tests. (2) Heterogeneity tests reveal that within the PRD, the digital economy inhibits carbon emissions in economically more advanced areas, whereas in relatively less developed areas, the development of the digital economy instead promotes carbon emissions. Moreover, the level of development of metropolitan areas in the PRD has a substantial impact on the relationship between the digital economy and carbon emissions. (3) Mechanism effect tests indicate that the intensity of environmental regulation in the PRD significantly enhances the negative moderating effect of the digital economy on carbon emissions; that is, the stronger the environmental regulation, the stronger the inhibitory effect of the digital economy on carbon emissions. (4) The scale of e-commerce plays a “suppression effect” in the relationship between the digital economy and carbon emissions in the PRD, and it is the core factor underlying the observed inhibitory effect of the digital economy on carbon emissions in this region.

At the same time, the heterogeneity analysis offers nuanced insights for policy design. Our findings show that in high-development cities (e.g., Guangzhou–Foshan–Zhaoqing metropolitan area), the digital economy can effectively reduce carbon emissions due to stronger infrastructure, industrial upgrading, and regulatory capacity. In contrast, in lower-development cities, the early-stage expansion of the digital economy tends to increase emissions, reflecting infrastructure buildout, energy mix constraints, and less efficient logistics. Moreover, metropolitan coordination matters: the GFZ area benefits from integrated governance and infrastructure, while the SHD and ZZJ areas show weaker spillover absorption.

Based on the above conclusions, this paper proposes the following recommendations.

First, promote differentiated digital economy development and strengthen regional coordination. The inhibitory effect of the digital economy on carbon emissions varies with the level of economic development. Therefore, policymakers should design targeted measures that match each city’s development stage. These measures should promote sustainable economic growth and help narrow the regional digital divide. Tailor digital economy strategies to local development levels: in advanced cities, accelerate the transition to green digital infrastructure (e.g., energy-efficient data centers and low-carbon cloud computing); in less-developed areas, combine digital expansion with early carbon risk assessment and capacity building. Enhance cross-city cooperation in the PRD to reduce coordination costs and improve shared infrastructure for green digitalization.

Second, adopt adaptive environmental regulation to guide low-carbon digital transformation. In advanced cities, stricter environmental standards and refined carbon trading mechanisms can amplify the carbon reduction effect of the digital economy. In less-developed cities, avoid overly rigid regulation that may hinder growth; instead, provide gradual regulation combined with incentives (e.g., green finance and technology subsidies) to support digital manufacturing and cleaner logistics.

Third, leverage e-commerce as a key mediator for carbon reduction. Within the urban agglomeration, give full play to the leading roles of Shenzhen and Guangzhou, use their e-commerce resources and agglomeration effects to radiate to surrounding cities, and strengthen coordination within the agglomeration in e-commerce development. Strengthen e-commerce infrastructure and logistics electrification, particularly in lower-development areas, to offset the potential carbon-increasing effects of digital expansion. By advancing “digital economy + e-commerce” in parallel, truly unlock the digital economy’s low-carbon potential.

This study also makes several contributions to the global academic discussion on the digital economy and sustainability. While existing international research—particularly in developed economies—has primarily emphasized the carbon reduction potential of digital technologies and green innovation, these findings reveal that in rapidly developing and economically heterogeneous urban agglomerations, such as the PRD, the expansion of e-commerce can exert a suppression effect that offsets the short-term carbon-increasing impact of early-stage digital economy growth. This insight adds nuance to the global understanding of how digitalization interacts with carbon emissions in emerging economies. Furthermore, by confirming the moderating role of environmental regulation, this study provides new empirical evidence that supports and extends the Porter Hypothesis, which posits that appropriate regulation can stimulate green innovation and improve environmental performance. The framework integrating the suppression effect into the digital economy–carbon emissions nexus offers a novel theoretical perspective that could be applied in other rapidly digitizing regions worldwide.