How the Digital Economy Shapes Green and Low-Carbon Development in the Yangtze River Economic Belt

Abstract

Faced with increasingly severe resource shortages and environmental pressures, exploring the impact of the digital economy on green and low-carbon development and its potential mechanisms is of great significance. Drawing on a comprehensive panel dataset spanning the decade from 2014 to 2023, this study examines 11 provincial administrative regions situated within the Yangtze River Economic Belt in China, systematically examining the effects and underlying pathways of the digital economy on green and low-carbon development. We construct an evaluation index system for the digital economy and green and low-carbon development, and use a two-way fixed effects model, a moderating effect model, and a threshold regression model for empirical analysis. Empirical results show that the digital economy significantly promotes green and low-carbon development, and this conclusion remains robust after a series of robustness tests. Mechanism analysis indicates that green technology innovation plays a significant moderating role, amplifying the environmental benefits of the digital economy; industrial structure upgrading exhibits a double threshold effect, with the promoting effect of the digital economy on green and low-carbon development increasing as the threshold is exceeded. Heterogeneity analysis shows that the ecological effects of the digital economy are significant in the midstream and southwest cluster and in areas with high factor allocation efficiency. We conclude that optimizing the environment for digital economic development, emphasizing innovation in digital green technologies, and implementing differentiated regional and structural policies can achieve a coordinated advancement of digital transformation and green and low-carbon development, providing valuable empirical evidence and policy implications for regional sustainable development.

1. Introduction

Climate change and ecological constraints have become core challenges to global sustainable development, forcing economic systems to seek deeper coordination between efficiency improvement and green development. Recently, China has increasingly shaped the trajectory of global climate efforts. Since the strategic shift in 2012, China has significantly increased its investment in ecological civilization construction, elevating it to a core national growth strategy, and green and low-carbon transformation has been systematically promoted. During the 75th UN General Assembly, China’s pledge to reach a carbon peak by 2030 and realize carbon neutrality by 2060 was officially articulated by President Xi Jinping [1,2], marking a profound shift in China’s development model from factor-driven expansion to quality-oriented and structurally optimized development. Against this backdrop, the Yangtze River Economic Belt (YREB) has become a particularly crucial paradigm for testing the compatibility of economic growth and low-carbon transformation. Spanning the eastern, central, and western reaches of the country, the YREB stands as a premier hub of economic vitality and population density. It functions not only as a core engine for China’s macro-economy but also as a fundamental safeguard for regional ecological integrity [3]. Despite the region’s rapid economic growth (Figure 1a), this region is responsible for over thirty percent of China’s overall carbon output (Figure 1b). Furthermore, significant differences exist among provinces in terms of industrial structure, energy efficiency, and green transformation capabilities [4,5], leading to a clear structural contradiction between high-quality development goals and environmental constraints.

At the same time, the digital economy, as a potential core motivator for innovation, is profoundly reshaping factor allocation, production and consumption patterns, governance, and regional cooperation mechanisms, and is seen as a vital route to coordinating economic growth and ecological construction [6]. From a theoretical perspective, the digital economy deeply empowers green and low-carbon development through multiple channels such as resource allocation optimization, technological innovation stimulation, and industrial structure upgrading. Specifically, relying on information and communication technology, the digital economy achieves deep integration of production factors through data-driven, platform-supported, and intelligent applications, optimizing resource utilization efficiency while reducing energy consumption intensity, and effectively underpinning efforts to enhance the ecological environment [7]. On the one hand, digital platforms and intelligent management systems enable live tracking and flexible tuning across the whole cycle—from how goods are made and shared to how they are eventually used, significantly suppressing resource waste and ineffective investment; on the other hand, the digital economy essentially reshapes how businesses are structured and run, which, to some extent, replaces the traditional high-energy-consuming model and drives the industrial structure to evolve towards green and high-value-added directions [8]. Through the above multi-dimensional paths, the digital economy provides a systematic mechanism for promoting regional green and low-carbon transformation.

However, whether and how the digital economy can effectively promote green and low-carbon development in vast, geographically complex regions like the YREB, with its fragile ecological environment, remains a subject of in-depth research. Whether the environmental benefits brought about by digitalization will change disproportionately as structural optimization goes through different stages remains highly uncertain, as does whether its impact exhibits systematic differences across different sub-regions. Answering these questions is crucial not only for determining feasible paths for the ecological transition of the YREB but also for providing important policy references for other developing countries facing similar trade-offs between growth and the environment.

The primary contribution of this study is its granular analysis of how the digital economy drives green and low-carbon progress in the Yangtze River Basin. First, grounded in the synergistic effects between resource allocation and environmental benefits, we develop a multi-dimensional digital economic development index and a green development evaluation system to illustrate the integrated effects and intrinsic links between digital growth and green transition. Second, moving beyond the limitations of previous studies that focused solely on mediating effects, we position green technology innovation as a moderating variable to reveal how it shapes the digital economy’s influence. Furthermore, we employ a threshold effect model to uncover the non-linear relationship between the digital economy and green development across different stages of industrial upgrading, capturing the dynamic characteristics of this link. Given the current resource scarcity and developmental pressures, our findings provide a solid rationale for promoting green development in the YREB and offer targeted policy recommendations for coordinating digital and environmental transformations.

2. Literature Review

2.1. Theoretical Basis

This paper uses the theories of production factor allocation, green technology bias, and Kuznets’ industrial structure evolution as analytical frameworks to explain the inherent logic of the digital economy’s role in the green and low-carbon development of the YREB. The production factor allocation theory emphasizes the decisive impact exerted by the allocation structure of factors through different fields and uses on economic performance [9]. In the YREB, where factors are highly concentrated and regional differences are significant, as data assets are increasingly woven into the fabric of traditional production, the digital economy changes the way factors are allocated and the efficiency of collaboration, enabling the resource allocation process to simultaneously possess the characteristics of efficiency improvement and structural optimization, thus providing the foundational impetus for decarbonized and sustainable growth [10]. By examining the orientation of technological advancement, the green technology bias theory elucidates why the digital economy’s environmental consequences may vary across different contexts. This theory argues that the environmental consequences of technological innovation depend on its bias type; under given institutional and market conditions, technological progress may strengthen or alleviate constraints on resource consumption and pollution emissions. The digital economy, by influencing information acquisition, R&D decisions, and technology diffusion processes, changes the relative benefit structure of technology selection, thereby causing technological progress to exhibit differentiated evolutionary characteristics in different directions [11]. Viewed through a long-term lens, the impact of digitalization remains contingent upon the specific phase of industrial structural evolution. Following Kuznets’ logic on industrial evolution, economic maturation entails a structural migration away from resource-heavy, low-yield sectors toward those defined by high value added, technological sophistication, and service integration, a process that profoundly impacts resource utilization and environmental performance [12]. Depending on the specific phase of industrial evolution, the modalities through which digitalization permeates the real economy and its impact on resource allocation and technology application differ significantly. These three theories, from different perspectives such as factor allocation, technological progress, and structural evolution, provide necessary theoretical support for analyzing the multifaceted impact of the digital economy on green and low-carbon development.

2.2. Empirical Research and Research Gaps

Green and low-carbon development is essentially how we get to a sustainable future. Extensive research has been conducted on energy [13], ecological [14], and economic sustainability [15], but a unified standard for regional green and low-carbon indicators remains lacking, with most focusing on carbon emissions [16], carbon emission efficiency [17], and sustainable development efficiency [18]. Promoting environmental sustainability is essentially a profound transformation, with its core being the achievement of economic growth alongside low-carbon development, underpinned by structural industrial upgrading and the catalytic effects of sustainable technological breakthroughs [19,20]. In light of China’s specific circumstances, constructing a comprehensive green and low-carbon development indicator system will provide a more comprehensive assessment of the current development status [21].

Tapscott first proposed the theoretical framework of the digital economy in the 1990s [22], which suggests the digital economy has evolved from a network economy into a multi-factor integrated economic form and has become a global research focus. In China, the localization of the digital economy encompasses various activities such as digital infrastructure construction [23], digital industrialization [24], industrial digitization [25], and digital applications [26]. Measurement methods mainly include the industrial value-added method [27], the proxy variable method [28], and the multi-dimensional comprehensive indicator method [29], with multi-dimensional comprehensive measurement gradually becoming the mainstream. The digital economy, by promoting the digitalization, decarbonization, and intelligentization of economic activities, has emerged as a key force for accelerating green and sustainable growth [30].

Existing studies have delved into the theory of factor allocation, emphasizing the importance of the rational distribution of factors for optimizing economic structure and improving development efficiency. This theory provides policymakers with an important analytical framework for developing and sharing resources and knowledge at the regional level and promoting coordinated development. However, current scholarly attention has primarily centered upon the aggregate consequences of factor reallocations for economic development, with a primary emphasis on output expansion and structural optimization [9]. Nevertheless, the ecological implications of how digitalization reconfigures factor allocation remain under-explored, particularly lacking a systematic examination of its green transformation effects against increasingly tight resource and environmental constraints. This lacuna requires further scrutiny.

Amidst the burgeoning wave of digitalization, academia has gradually expanded its research scope to encompass the digital economy landscape, exploring its potential role in reconciling the tension between economic expansion and ecological limits. From a direct effect perspective, the digital economy promotes the flow of both conventional and data-driven production factors, strengthens the coordinated allocation of different factors in the green field, and improves the regional ecological environment [31]. Simultaneously, using digital tools and smart tech makes energy use much more efficient and cuts the carbon cost of manufacturing. It also fixes inefficiencies in how goods are made and moved, which leads to a real drop in pollution [32]. Furthermore, the digital economy, by elevating informational symmetry and curtailing transactional friction, and promoting the efficient flow of factors of production, reduces resource misallocation and increases per capita income, creating favorable conditions for the spread of green production methods and low-carbon consumption patterns [7]. At the level of its mechanisms, relevant research mainly analyzes these mechanisms from the perspectives of technological innovation and industrial restructuring. In one respect, the digital economy, by improving information access conditions and reducing R&D costs, can incentivize green technology innovation activities, and foster the adoption and proliferation of low-carbon innovations alongside eco-efficient manufacturing solutions, thereby empowering the entire economy to shift toward a low-carbon model more effectively [33]. In another respect, digital transformation accelerates the upgrading of industrial structures from factor-heavy modalities to an innovation-oriented and service-oriented industrial paradigm, curtailing the dominance of carbon-intensive sectors while fostering the expansion of eco-friendly startups and premium-value domains [34]. Furthermore, some studies have begun to focus on the non-linear characteristics and regional differences in the impact of the digital economy on green and low-carbon development. Some scholars have pointed out that the environmental dividends of the digital economy are highly contingent on a region’s current stage of development, industrial structure levels, or institutional environments, exhibiting threshold effects or diminishing marginal returns [34,35]. At the same time, owing to pronounced spatial disparities in financial footings, natural assets, and institutional frameworks, the digital economy’s role in catalyzing eco-friendly transitions exhibits marked regional variations.

In summary, while the existing literature furnishes a robust conceptual and empirical backdrop for understanding the digital economy and green and low-carbon development, several critical research gaps persist. First, most studies prioritize the economic dimensions of factor allocation, often overlooking the environmental externalities inherent in digitally driven factor reallocation. We aim to fill this knowledge gap by integrating environmental performance into the analytical framework, thereby offering a more comprehensive perspective on how the digital economy influences green and low-carbon development. Second, although technological innovation has been extensively discussed, it is frequently analyzed in isolation, failing to fully capture its interaction mechanisms with the digital economy. To address this limitation, we introduce green technology innovation as a moderator to examine how it either strengthens or attenuates these impacts. Third, there is a noticeable scarcity of literature exploring the non-linear dynamics between the digital economy and environmental outcomes across different stages of industrial structure upgrading. To fill this gap, this study investigates the non-linear relationship between these two variables and further accounts for regional heterogeneity.

The distinctiveness of this work lies in several aspects: First, it constructs a comprehensive indicator system for both the digital economy and green and low-carbon development, systematically characterizing the theoretical logic of the interaction between the digital economy and green and low-carbon development within a comprehensive framework. Second, it explains the transmission path of the digital economy’s influence on green and low-carbon development, focusing on the effects of industrial-scale technological innovation. Finally, it deeply analyzes the non-linear characteristics of the threshold effect in the YREB and the reasons for regional differences in heterogeneity.

3. Research Hypothesis

3.1. The Direct Impact of the Digital Economy on Green and Low-Carbon Development

From the perspective of production factor allocation theory, the digital economy, relying on data, algorithms, and platforms, integrates fragmented production factors to optimize factor structure and reallocate resources, promoting green and low-carbon development.

For individuals, the digital economy directly boosts income growth by expanding employment opportunities and improving factor participation [36]. The development of the platform economy, digital services, and flexible employment lowers labor market entry barriers, improves labor factor allocation efficiency, and enables individuals to obtain income through diversified channels. Simultaneously, increased information transparency helps reduce information asymmetry in income acquisition and enhances the match between labor compensation and production contribution, thereby increasing personal income while restraining the proliferation of energy-intensive and carbon-heavy consumption patterns [7].

At the enterprise level, the digital economy embeds data elements into production and management processes, optimizing the allocative precision and synergy of conventional productive assets alongside auxiliary elements, helping to reduce operating costs and improve revenue structure. It drives enterprises to shift from extensive expansion to intensive management, reducing high-energy-consuming and high-emission processes, and providing conditions for enterprises to achieve stable development under green constraints [37].

At the regional level, the digital economy promotes the digitalization of government services, and achieves the spatial reallocation of productive elements and industrial synergy, driving the reallocation of income opportunities across a wider geographical area. It helps drive regional employment expansion and increases residents’ income, and indirectly reduces dependence on high-carbon production and high-energy-consuming growth paths by raising overall income levels and upgrading consumption patterns, thereby promoting green and low-carbon development in the region [38].

Based on the above analysis, Hypothesis 1 is proposed.

Hypothesis 1.

The digital economy has a positive impact on green and low-carbon development.

3.2. The Impact Mechanism of the Digital Economy on Green and Low-Carbon Development

Anchored in green technology bias theory, the environmental effects of technological progress depend on its direction. Green tech acts as a vital filter, shaping the environmental footprint of the digital shift by guiding technological progress toward cleaner and more efficient production modes.

The digital economy empowers the advancement of green inventions by enabling enterprises to reduce R&D costs and accelerate knowledge diffusion. However, the environmental benefits of the digital economy are highly contingent upon a region’s green innovation capacity. In other words, green innovation modulates the strength of the link between digital growth and green transition [39]. When green technology innovation levels are high, digital technologies are more easily integrated with energy conservation, emissions reduction, and cleaner production, synergizing improvements in resource allocation efficiency with pollution reduction targets, thereby magnifying the digital economy’s empowerment effect on sustainable development.

Conversely, in regions with weaker green technology innovation capabilities, the optimization impetus of the digital sphere on factor allocation performance may manifest more as output expansion and economies of scale, with relatively limited effects on curbing energy consumption and carbon emissions, and potentially even exacerbating environmental pressures in the short term [33]. Therefore, green technology innovation plays a crucial moderating role by influencing the intensity and direction of the interaction between the digital economy and green and low-carbon development. Thus, green technology innovation is not only a key transmission path for the digital economy’s impact on green and low-carbon development but also significantly amplifies the positive relationship between the two through its moderating effect.

Therefore, rather than serving primarily as a mediating variable that explains the transmission mechanism [40,41], green technology innovation functions as a moderating factor that conditions the strength and direction of how the digital economy interacts with green development.

Thus, Hypothesis 2 is proposed.

Hypothesis 2.

Green technology innovation plays a significant positive moderating role in the process of the digital economy promoting green and low-carbon development.

The spatial effects of the digital economy are not a one-dimensional promotional effect, but rather an intertwining of collaboration and competition, exhibiting heterogeneity and complexity. Beyond its localized influence, the progression of the digital sphere’s contribution to decarbonization pathways is substantially moderated by the regional industrial configuration [42]. Specifically, the industrial structure advancement index, as a threshold variable, determines how digitalization manifests: where the industrial fabric remains rudimentary, the impetus provided by the digital realm tends to be constrained; when the industrial structure exceeds a certain threshold level, its promoting effect significantly strengthens.

Based on our analysis, we propose Hypothesis 3.

Hypothesis 3.

The green dividends of a digitalized economy exhibit a non-linear threshold effect. Once the industrial framework surpasses a critical threshold, the digital economy’s facilitative influence undergoes a substantial intensification.

Figure 2 shows the theoretical framework and mechanism described above.

4. Materials and Methods

4.1. Study Area and Data Sources

Serving as a pivotal economic artery for China (Figure 3), the YREB encompasses 11 administrative divisions, accounting for roughly one-fifth of the nation’s total landmass and hosting more than 40% of its inhabitants. It constitutes a natural testing ground characterized by highly coordinated development strategies yet it manifests marked internal disparities across various growth phases and factor endowments. This coexistence of “strategic unity” and “provincial heterogeneity” provides a highly representative analytical framework for examining the effectiveness of regional coordination policies, the economic impact of environmental regulations, and growth convergence mechanisms under multi-level governance from the perspectives of new economic geography and institutional evolution.

Considering data availability and maturity, panel data from 2014 to 2023 were selected. Our dataset was primarily sourced from the China Statistical Yearbook, supplemented by regional yearbooks from various provinces and cities. The empirical analysis incorporated green patent metrics from the CNRDS database, which were supplemented by digital financial inclusion data officially released by Peking University. Carbon emission coefficients were sourced from the Global Atmospheric Research Emissions Database (EDGAR). Missing values were supplemented using linear interpolation and mean substitution methods.

4.2. Model Construction

4.2.1. Entropy Weight–Topsis Method

For a robust assessment of digital economic progression, this study employs an Entropy–Topsis approach, where the entropy weight method determines indicator significance and Topsis derives the finalized composite scores. Rooted in information theory, the entropy weighting technique quantifies the relative significance of metrics by evaluating the degree of dispersion and disorder within the observed values, thus eliminating the influence of subjective factors. The Topsis method is a classic multi-indicator decision analysis method which ranks alternatives by evaluating their geometric proximity to the theoretically superior point and their divergence from the least desirable baseline. First, the data is preprocessed and standardized to ensure all indicators show a positive trend and to eliminate the influence of dimensions. Second, the entropy method is used to calculate the information entropy, information utility value, and weight of each indicator for each year to objectively reflect the information content and degree of variation in the indicators. Finally, these objective weights are subsequently applied to the standardized indicator set to formulate a weighted evaluation matrix, enabling the estimation of annual comprehensive scores for digital economic advancement across regions.

4.2.2. Benchmark Regression Model

Considering how digitalization is reshaping green growth in so many different ways, in order to clarify its impact mechanism, referring to the research of Breuer and Dehaan [43], a two-way fixed effects model is specified. This model effectively addresses potential heteroscedasticity and serial correlation, thereby minimizing the estimation bias typically induced by unobserved and omitted variables.

$${\mathrm{GLCD}}_{\mathrm{i},\mathrm{t}}={\mathsf{\alpha }}_0+{\mathsf{\alpha }}_1{\mathrm{D}\mathrm{E}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\alpha }}_{\mathrm{j}}\sum {\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(1)

Within the baseline model, the subscripts i and t signify the province and year, while $\mathsf{\alpha }$ represents the estimated coefficients. ${\mathrm{G}\mathrm{L}\mathrm{C}\mathrm{D}}_{\mathrm{i},\mathrm{t}}$ is the dependent variable, representing the level of green and low-carbon development. ${\mathrm{D}\mathrm{E}}_{\mathrm{i},\mathrm{t}}$ is the independent variable, representing the level of digital economic development. $\sum {\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}}_{\mathrm{i},\mathrm{t}}$ represents the control variables. ${\mathsf{\mu }}_{\mathrm{i}}$ and ${\mathsf{\delta }}_{\mathrm{t}}$ are incorporated as province-specific and year-specific fixed effects, with ${\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$ representing the idiosyncratic error term.

4.2.3. Moderation Effect Model

Considering that green technology innovation may have a moderating effect on the role of the digital economy in promoting green and low-carbon development, this paper constructs the following model to test this mechanism.

$${\mathrm{GLCD}}_{\mathrm{i},\mathrm{t}}={\mathsf{\beta }}_0+{\mathsf{\beta }}_1{\mathrm{D}\mathrm{E}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_2{\mathrm{L}\mathrm{n}\mathrm{G}\mathrm{T}\mathrm{I}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\beta }}_3{(\mathrm{D}\mathrm{E}}_{\mathrm{i},\mathrm{t}}\times {\mathrm{L}\mathrm{n}\mathrm{G}\mathrm{T}\mathrm{I}}_{\mathrm{i},\mathrm{t}})+{\mathsf{\beta }}_{\mathrm{k}}\sum {\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\mu }}_{\mathrm{i}}+{\mathsf{\delta }}_{\mathrm{t}}+{\mathsf{\epsilon }}_{\mathrm{i},\mathrm{t}}$$

(2)

$\mathrm{L}\mathrm{n}{\mathrm{G}\mathrm{T}\mathrm{I}}_{\mathrm{i},\mathrm{t}}$ represents green technology innovation; ${\mathrm{D}\mathrm{E}}_{\mathrm{i},\mathrm{t}}\times {\mathrm{L}\mathrm{n}\mathrm{G}\mathrm{T}\mathrm{I}}_{\mathrm{i},\mathrm{t}}$ represents the interaction term between the digital economy and green technology innovation. If both ${\mathsf{\beta }}_1$ and ${\mathsf{\beta }}_3$ coefficients are significant, it indicates that the contribution of the digital economy to green and low-carbon development is moderated by green technology innovation. Specifically, coefficients with the same sign reflect the synergistic enhancement logic of green technology innovation on the above relationship; coefficients with opposite signs suggest that it plays a counterbalancing or negative moderating role in this.

4.2.4. Threshold Effect Model

Recognizing that whether digitalization leaves a mark on green growth is contingent upon the evolution of industrial structures, a panel threshold model was constructed using Hansen’s model [44] to verify this mechanism.

$${\mathrm{G}\mathrm{L}\mathrm{C}\mathrm{D}}_{\mathrm{i},\mathrm{t}}={\mathsf{\gamma }}_0+{\mathsf{\gamma }}_1{\mathrm{D}\mathrm{E}}_{\mathrm{i},\mathrm{t}}\times \mathrm{I}\left({\mathrm{U}\mathrm{I}\mathrm{S}}_{\mathrm{i},\mathrm{t}}\le {\mathsf{\omega }}_1\right)+{\mathsf{\gamma }}_2{\mathrm{D}\mathrm{E}}_{\mathrm{i},\mathrm{t}}\times \mathrm{I}\left({\mathsf{\omega }}_1\le {\mathrm{U}\mathrm{I}\mathrm{S}}_{\mathrm{i},\mathrm{t}}\right)+{\mathsf{\gamma }}_{\mathrm{j}}\sum {\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}}_{\mathrm{i},\mathrm{t}}+{\mathsf{\eta }}_{\mathrm{i},\mathrm{t}}$$

(3)

In this model, UIS serves as the threshold variable and ${\mathsf{\omega }}_1$ as the threshold value, I( ) denotes the indicator function, and $\sum {\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}}_{\mathrm{i},\mathrm{t}}$ is the control variable group, consistent with the previous one.

4.3. Variable Definition

4.3.1. Explained Variable

The explained variable is Green Low-Carbon Development (GLCD). To capture the multi-dimensional nature of sustainable transformation, we constructed a four-dimensional evaluation index system for GLCD, which comprises seven specific indicators including pollution reduction, carbon reduction, greening expansion, and economic efficiency (

Table 1. Indicator System for digital economy and Green Low-Carbon Development.

System Layer	Criterion Layer	Criterion Layer Weights	Indicator Layer/Specific Unit	Indicator Layer Weights
Digital Economy Evaluation Index System	Digital Infrastructure	0.2615	Mobile phone penetration rate/units (per 100 people)−1	0.0645
			Internet penetration rate/%	0.0590
			Fiber optic cable line length/104 km	0.0880
			Internet broadband access rate/%	0.0500
	Digital Industrialization	0.4522	Percentage of employees in information transmission, software and information services/%	0.2031
			Total telecommunications business volume per capita/104 ¥ (per person)−1	0.2491
	Industrial Digitalization	0.2153	Percentage of enterprises with e-commerce transaction activities/%	0.1851
			Total e-commerce transaction volume/104 ¥	0.0302
	Digital Applications	0.0710	Digital government/number	0.0198
			Digital Inclusive Finance Index	0.0512
Green and Low-Carbon Development Evaluation Index System	Economic Efficiency	0.3921	GDP per capita/104 ¥	0.3921
	Greening Expansion	0.2743	Green coverage rate of built-up areas/%	0.1168
			Per capita park green space area/m2 (per person)−1	0.1575
	Pollution Reduction	0.2099	Nitrogen oxide emissions/104 t	0.0145
			Sulfur dioxide emissions/104 t	0.1101
			Total industrial wastewater discharge/104 t	0.0853
	Carbon Reduction	0.1237	Total carbon dioxide emissions/104 t	0.1237

Note: Please see Section 4.1 for data source information; indicator weights are derived via the entropy weight method.

).

The selection of these dimensions is theoretically grounded in the synergy framework of pollution reduction and carbon mitigation, which is a central concept in China’s ecological development strategy. Specifically: (1) pollution and carbon reduction represent the intensity of environmental governance and climate mitigation [45]; (2) greening expansion reflects carbon sink capacity and ecological resilience [46]; and (3) economic efficiency (denoted by GDP per capita) represents the improvement in development quality and the decoupling of economic growth from resource consumption [47].

To maintain methodological rigor and ensure the reliability of the results, we employed an objective weighting method for aggregation and further verified the structural robustness of this framework in the robustness checks (Section 5.2).

4.3.2. Explanatory Variables

This study, based on the definition of digital economy by the China Academy of Information and Communications Technology (CAICT), adopts an indicator system consisting of four levels: digital infrastructure, digital industry development, digital innovation capability, and digital applications [48]. The reasons for choosing this system are as follows: First, CAICT proposed four frameworks for digital economy: digital industrialization, industrial digitization, digital governance, and data valorization. Second, digital government affairs adopt the number of local government websites from the “Research Report on the Development of China’s Digital Economy” published by CAICT. Digital governance serves as a vital proxy for how digital technologies benefit society, underscoring their facilitative role in advancing the digital economy. Furthermore, by integrating digital financial inclusion data officially released by Peking University, this study establishes a multi-dimensional framework for evaluating the digital economy.

Based on existing research [26], a criteria layer is constructed from four aspects: (1) Digital infrastructure, reflecting the popularization and coverage of next-generation digital foundations such as mobile communication, internet access, and optical fiber networks. (2) Digital industrialization, measuring the development scale, employment structure, and output efficiency of information transmission, software, and information technology services. (3) Industrial digitization, reflecting the breadth of e-commerce application and transaction scale in enterprises, and measuring the synergistic convergence of legacy industries with advanced digital applications. (4) Digital applications: This section covers the practical application level in areas such as digital government and digital finance, and assesses the penetration of the digital economy in social services and governance.

To bolster the index’s reliability, we used Entropy Weight–Topsis to obtain fair weights and work out the overall score (Table 1). By minimizing subjective bias, this method—already a standard in the field—ensures that the measurement results are both credible and comparable. Furthermore, Section 5.2 provides robustness checks to confirm the index’s stability.

4.3.3. Moderating Variables

Based on a theoretical analysis framework, this study introduced green technology innovation as a moderator to clarify the boundary conditions under which the digital economy fosters regional GLCD. Given the objectivity, traceability, and consistency of examination standards for patent data, this study adopted the patent measurement method [49] as the core measurement method.

Specifically, referring to Deng’s method [50], green technology innovation was quantified by the aggregate number of green invention and utility model patents (both applications and grants), was incremented by one to handle zero-value observations, then took the natural logarithm, denoted as LnGTI, to ensure data continuity.

4.3.4. Threshold Variable

Drawing on the methods widely adopted by Zhang [51], the ratio of tertiary sector output to secondary sector output is used to measure industrial upgrading (UIS). This indicator effectively captures the direction of upgrading and transformation between industries: an increasing ratio indicates that the economic structure is evolving towards a service-oriented direction, and the industrial structure is becoming increasingly sophisticated.

4.3.5. Control Variables

To ensure a comprehensive analysis and mitigate potential estimation bias, we incorporated the following set of control variables into the model: Social Consumption Level (SCI), defined as total retail sales of consumer goods divided by regional GDP; Human Capital Level (HUMAN), defined as the higher education enrollment divided by the year-end resident population in each region; Financial Development Level (FD), using Zhang’s method [52], defined as the total year-end loans of financial institutions divided by the city’s GDP; Environmental Regulation Level (ER), referring to the method of Hao et al. [53], calculated as industrial pollution control investment divided by industrial added value; and Transportation Infrastructure Level (TRAN), measured by highway mileage, following Shen et al. [54] and Yang et al. [55]; highway mileage was used as a proxy, as it reflects the primary mode of regional factor mobility and freight transport. Although it does not capture railways and new energy facilities, it provides a consistent and comparable indicator and is closely related to transportation-related carbon emissions.

We can find a comprehensive breakdown of the summary statistics for our primary variables (

Table 2. Descriptive statistics.

Variables	Sample Size	Mean	Standard Deviation	Min	Max
GLCD	110	0.715	0.102	0.387	0.910
DE	110	0.512	0.143	0.127	0.756
SCI	110	0.428	0.043	0.320	0.504
HUMAN	110	0.022	0.005	0.012	0.034
FD	110	1.504	0.376	0.803	2.631
ER	110	0.078	0.038	0.018	0.174
TRAN	110	11.996	0.854	9.466	12.943
LnGTI	110	9.314	0.943	7.265	11.460
UIS	110	1.358	0.438	0.704	3.058

). Specifically, the GLCD index averaged 0.715, with a dispersion ranging from 0.387 to 0.910 (SD = 0.102). This spread highlights noticeable disparities and shifts in green development levels among the sampled cities. For the digital economy (DE), the mean value was 0.512, reflecting the general development status across the region. The distributional patterns observed here align well with the prior literature. Furthermore, all control variables exhibit reasonable distributions, and their details are omitted here for brevity.

Before analysis, to rule out multicollinearity, we calculated the Variance Inflation Factor (VIF) for all explanatory variables. All variables had VIF values below 10, and the mean VIF was 4.22. Following Mubeen et al. [56], there was no multicollinearity problem.

5. Empirical Analysis

Both the DE index and the GLCD index exhibited a consistent upward trajectory over the sampled period (Figure 4). This upward trend underscores that DE advancements have substantially bolstered GLCD.

From a regional perspective, Shanghai leads in both indicators, demonstrating its leading role at the forefront of regional development. The western regions such as Sichuan, Yunnan, and Guizhou, although starting from a lower point, had gradually approached or partially surpassed the level of eastern provinces in one of their indicators by 2023, reflecting a trend of coordinated regional development.

5.1. Benchmark Regression

To select a suitable panel model, this paper conducted the Hausman test (

Table 3. Hausman test.

Variables	Fe	Re	Difference	S.E.
DE	0.376	0.402	−0.026	-
SCI	−0.080	−0.105	0.025	-
HUMAN	0.517	4.271	−3.754	1.578
FD	0.056	0.048	0.008	0.006
ER	−1.931	0.122	−2.053	0.531
TRAN	−0.113	0.009	−0.121	0.056

). The results showed that ${\chi }^2\left(6\right)=15.95, p=0.0014 < 0.05$, rejecting the null hypothesis that “random effects models are consistent and effective”. Consequently, the fixed effects approach is preferred for this analysis. To account for potential biases from unobserved individual and time-specific factors, we adopt a two-way fixed effects framework as our primary empirical strategy.

The first two columns present the benchmark estimates derived from a dual fixed effects approach (

Table 4. Benchmark regression and moderating effect test.

Variables	Benchmark Regression		Moderating Effect
	(1) No Control Variables	(2) Add Control Variables	(3) No Interactive Items	(4) Add Interactive Items
DE	0.226 *** (2.62)	0.239 *** (2.91)	0.248 *** (2.99)	0.221 *** (2.75)
LnGTI			0.023 (0.91)	0.034 (1.41)
DE × LnGTI				0.081 *** (2.79)
SCI		−0.148 (−1.06)	−0.117 (−0.82)	−0.061 (−0.44)
HUMAN		4.494 (1.44)	3.798 (1.18)	6.681 ** (2.04)
FD		0.098 *** (2.69)	0.104 *** (2.80)	0.108 *** (3.03)
ER		−1.508 ** (−2.27)	−1.693 ** (−2.43)	−2.373 *** (−3.33)
TRAN		−0.113 ** (−2.03)	−0.117 ** (−2.09)	−0.109 ** (−2.03)
Constant	0.560 *** (13.53)	0.640 *** (8.06)	0.595 *** (7.36)	0.400 *** (3.80)
Year FE	Yes	Yes	Yes	Yes
Province FE	Yes	Yes	Yes	Yes
N	110	110	110	110
R2	0.072	0.264	0.271	0.335

Note: *** and ** indicate significance at the 1% and 5% statistical levels, respectively. Robust t-statistics are reported in parentheses, and yes indicates that the model has been conditionally controlled. R² is the within R². The same applies below.

). The DE showed a significant positive impact in all regressions, with coefficients stabilizing between 0.226 and 0.239, suggesting that DE expansion serves as a powerful catalyst for GLCD. This positive impact remained stable after including control variables; specifically, for every 1% increase in the DE index, GLCD significantly increased by 0.239%, which serves to reinforce the reliability of our primary results. Therefore, Hypothesis 1 holds.

From the perspective of factor allocation, the above empirical results further demonstrate that the promoting effect of the DE on GLCD does not simply stem from technological substitution or scale expansion, but rather from a structural improvement achieved by reshaping factor allocation methods. The digital economy, with data serving as a core production factor, strengthens the coordinated allocation among capital, labor, and technology, reduces factor mismatch, and directs more limited resources to high-efficiency, low-emission production and consumption sectors.

Regarding control variables, the FD variable exhibited a highly significant positive coefficient, indicating that a well-developed financial system positively supports GLCD, reflecting the allocation effect of capital factors concentrating in the green sector under digital conditions. The coefficient for HUMAN is positive, but it fails to reach statistical significance. The coefficients for ER and TRAN are both significantly negative at the 5% level, indirectly revealing that under conditions where factor allocation is not yet fully optimized, traditional input expansion may constrain green and low-carbon goals, reflecting potential efficiency losses in the current environmental governance model and the short-term tension between transportation network expansion and green goals. The coefficient for the SCI is negative and insignificant, indicating that its promoting effect has not yet manifested.

Overall, the digital economy provides endogenous impetus for green and low-carbon development by improving factor allocation efficiency and optimizing factor combination structure.

5.2. Robustness Test

To ensure the reasonableness of the conclusions, several sensitivity checks are conducted:

(1)

To test the robustness of the constructed indices, alternative specifications of the key variables were employed. For the digital economy, we considered digital infrastructure as the most direct reflection of environmental changes, and used digital infrastructure as the core proxy variable to replace the composite index, with the results in column (1) in

Table 5. Robustness test.

Variables	(1)	(2)	(3)	(4)	(5)	(6)	(7)
DE	0.205 ** (2.22)	0.334 *** (4.16)	0.267 *** (3.01)	0.236 *** (2.98)	0.253 ** (2.19)		0.941 *** (3.76)
IV						0.016 *** (4.21)
Control	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Constant	2.391 *** (3.45)	1.878 *** (2.94)	1.796 ** (2.54)	1.892 *** (2.93)	1.908 *** (2.84)	−1.832 * (2.06)	1.022 * (1.68)
Year FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes
Province FE	Yes	Yes	Yes	Yes	Yes	Yes	Yes
N	110	110	110	110	110	110	110
R2	0.432	0.501	0.296	0.268	0.234	-	-

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% statistical levels, respectively. Columns (1) through (6) report t-statistics in parentheses, while column (7) provides z-values. The two-stage least squares (2SLS) estimates are detailed in columns (6) and (7); notably, R² is omitted for the 2SLS models as it lacks a conventional interpretation in this context.

. Regarding the GLCD index, considering that the ecological dividend of the digital economy is not limited to directly improving productivity, we re-estimated the model by excluding the ‘economic efficiency’ dimension; an alternative index was reconstructed based solely on the remaining three pillars, the results of which are shown in Table 5, column (2). Under these alternative settings, the results are consistent with the benchmark, indicating that the indicator system and conclusions were reliable.

(2)

The GLCD indicator was constructed by replacing the total CO₂ with CO₂ intensity to eliminate the impact of differences in the size of different cities; the results are shown in Table 5, column (3).

(3)

All continuous variables were subjected to 1% quantile shrinkage to reduce the interference of extreme values on the estimation results; the results are detailed in column (4).

(4)

The DE index was re-estimated through the entropy weight approach to verify whether the core conclusions under different measurement methods are consistent; the results are detailed in column (5).

To mitigate potential endogenous biases within our framework, this study adopts the methodology of Xia [57] and Miao [58]. Specifically, in our instrumental variable (IV) estimation, we use a composite interaction term—combining regional postal service volume with the national internet penetration rate—as a proxy for the digital economy. The regional postal infrastructure reflects the potential conditions for digital economic development, while the number of internet users nationwide represents the exogenous impact of digital technology adoption. This interaction term satisfies the correlation requirement and ensures exogeneity by controlling for regional and time fixed effects.

Table 5 (columns 6–7) details the 2SLS estimation results. Specifically, the first-stage output in column (6) revealed that our IV is significant at the 1% level, confirming its strong predictive power for the digital economy. As presented in column (7), the DE yielded a substantial estimated effect of 0.941, which remained highly significant (p < 0.01). To ensure the validity of our IV approach, we perform several diagnostic checks. The Kleibergen–Paap rk LM test yields a statistic of 10.007, which is highly significant (p < 0.01), thereby ruling out the possibility of under-identification. Furthermore, the Kleibergen–Paap rk Wald F-statistic (17.071) exceeded the critical values for weak instruments, confirming that our chosen instrumental variable is both relevant and sufficiently strong.

Overall, the empirical evidence remains highly stable across various sensitivity checks, further validating the reliability of the baseline conclusion: the DE significantly promotes GLCD in the YREB.

5.3. Moderation Effect Analysis

The findings presented in Table 4 (columns 3 and 4) illustrate the interaction analysis regarding the role of green technological innovation. To mitigate multicollinearity, the core variables were centered, and the standard error clustering was performed at the province level. This paper uses a stepwise regression method for testing. As shown in column (3), without the interaction term, the estimated effect of the DE on GLCD stood at 0.248, passing the 1% significance test. This finding implies that DE growth acts as a strong catalyst for regional GLCD. Once the interaction between green tech innovation and the DE was introduced in column (4), the coefficient of the DE remained significant (0.221), indicating that its promoting effect has a certain robustness. Concurrently, the interaction term yields a coefficient of 0.081, suggesting that green technological innovation significantly amplifies the DE and is shown to play a constructive role in accelerating GLCD. Hypothesis 2 is verified.

5.4. Further Analysis

5.4.1. Non-Linear Impact of DE on GLCD: A Threshold Analysis of Industrial Upgrading

The fixed effects threshold results in

Table 6. Threshold effect test.

Threshold Variables	Threshold Effect	F Value	Threshold
UIS	Single threshold	62.08 *** (0.002)	1.0363
	Double threshold	36.39 ** (0.016)	1.0363, 1.0224
	Triple threshold	3.64 (0.652)	0.9404

Note: ** and *** indicate significance at the 1% and 5% statistical levels.

revealed that the DE’s impact on GLCD is not a simple straight line; instead, it shows a clear non-linear dynamic. The testing process confirmed that both the single and double threshold models were statistically significant, picking up pivotal tipping points at 1.0363 and 1.0224, whereas the triple threshold failed to reach statistical significance. This evidence collectively points toward a clear double threshold framework.

The threshold regression results (

Table 7. Threshold estimates and confidence intervals.

	GLCD
UIS < 1.0224	0.0827 (1.27)
1.0224 ≤ UIS < 1.0363	0.1447 *** (2.67)
UIS ≥ 1.0363	0.2894 *** (8.46)
Control	Yes
Constant	3.0143 *** (5.17)
Year FE	Yes
Province FE	Yes
N	110
R2	0.937

Note: *** indicate significance at the 1% statistical levels.

) revealed that the DE’s influence on GLCD follows a non-linear path, characterized by distinct stage-specific variations tied to industrial upgrading. Specifically, the marginal contribution of the digital economy remains negligible as long as the industrial structure advancement index remained below the 1.0224 threshold. However, once the index entered the 1.0224–1.0363 range, the corresponding impact of the digital economy was notably amplified, yielding a significantly positive coefficient of 0.1447. When it exceeds 1.0363, its promoting effect further increases significantly, with a coefficient reaching 0.2894, approximately twice that of the middle range.

This indicates that at a more advanced stage of industrial structural upgrading, the DE acts as a more potent driver for advancing GLCD, while its role is not yet apparent at a stage where the industrial structure is still dominated by traditional industries. Therefore, an advanced industrial structure serves as a critical prerequisite for the DE to manifest its environmental dividends, thus verifying Hypothesis 3.

5.4.2. Heterogeneity Analysis

We have just seen how the digital economy can boost green and low-carbon development. But that is only half the story. It is just as important to look at the gaps in how resources are actually used across different areas, and how these differences change the way the DE affects carbon emissions. Because of this, the next part of this paper will dive into the variations in how the digital economy shapes green growth from two specific angles.

Regional Heterogeneity Analysis. Specifically, this paper examined three distinct regions within the Yangtze River Economic Belt: the Yangtze River Delta (YRD), the provinces in the middle reaches of the Yangtze River, and the southwest cluster (Chongqing, Sichuan, and Guizhou). Yunnan does not really fit the mold. Since it is all about tourism and green mandates, it is hard to group it.

Table 8. Heterogeneity analysis.

Variables	Region			Factor Allocation Efficiency
	(1) Yangtze River Delta Region	(2) Provinces in the Middle Reaches of the Yangtze River	(3) Southwest Cluster	(4) High	(5) Low
DE	−0.061 (−0.21)	0.572 *** (4.17)	0.407 ** (2.74)	0.361 ** (2.43)	0.006 (0.04)
Control	Yes	Yes	Yes	Yes	Yes
Constant	12.231 *** (2.94)	0.687 (1.10)	2.588 (1.57)	2.405 (1.54)	2.104 ** (2.73)
Year FE	Yes	Yes	Yes	Yes	Yes
Province FE	Yes	Yes	Yes	Yes	Yes
N	40	30	30	60	50
R2	0.530	0.660	0.807	0.395	0.264

Note: ** and *** indicate significance at the 1% and 5% statistical levels.

presents these empirical findings across columns (1)–(3).

In the provinces in the middle reaches of the Yangtze River and the southwest cluster, the DE coefficients were strongly positive, suggesting that DE growth served as a powerful catalyst for GLCD in these regions. The primary reason is that these areas are still industrializing, and their economies are heavily built on traditional sectors. By bringing in digital tools, these industries can streamline their production and boost energy efficiency. This shift not only drives green innovation but also leads to a much stronger impact on cutting emissions.

In contrast, in the YRD, the digital economy’s impact leaned toward the negative side, but the finding was not statistically robust. One plausible explanation is that both the digital economy and green transformation in this region have reached a relatively advanced stage, leading to diminishing marginal effects of digital technologies. This echoes Li’s point [59] that the digital economy’s green benefits tend to taper off as a region gets more developed. Additionally, the rebound effect of the digital economy—manifested in an increased production scale, higher energy demand, and the construction of digital infrastructure—may partially offset its environmental benefits, resulting in a weak or even negative net effect in the short term.

These results indicate that the digital economy does not automatically benefit the environment everywhere. Instead, its green impact really depends on how far along a region is in its development. Such a pattern supports the notion that the digital economy exhibits stage-dependent effects, with stronger promotion effects in less developed regions and diminishing or ambiguous effects in highly developed regions. Consequently, the influence of the digital economy on green growth exhibits a ‘catching-up’ dynamic; this effect is particularly noticeable in less-digitized regions that possess significant room for industrial upgrading.

Heterogeneity analysis of factor allocation efficiency. Among the results in Table 8 (columns 4 and 5), there was a clear distinction. For regions with better resource distribution, the impact of the digital economy was both positive and significant, providing strong evidence that digital growth substantially fosters a greener and more low-carbon trajectory. Essentially, high allocation efficiency acts as a bridge, allowing digital and traditional assets to integrate better. This synergy makes production more efficient and strengthens the ‘green’ impact of digital growth. On the flip side, for regions where resource distribution is less efficient, the impact of the digital economy remains positive but does not reach a significant level. Conversely, in another group, the coefficient of the digital economy was positive but not significant. This shows that when there are obstacles in factor flow efficiency and allocation mode, the digital economy fails to fully leverage its technological strengths, leading to a restricted impact on green growth. To unlock its environmental potential, it is essential to first enhance the efficiency of resource allocation, which serves as a critical foundation for driving the digital economy.

6. Research Findings and Policy Recommendations

To promote sustainable development, we must explore the path of regional green and low-carbon development driven by the digital economy. This paper systematically explored the impact of the digital economy on the green and low-carbon development of the YREB and its mechanism of action from the perspective of resource-side and environmental benefits. The following are the research results. First, we found that the digital economy really helps regions go green and stay low-carbon. The reason is that the digital economy changes how we use our resources. By fixing and improving the way things are allocated, it shifts the whole growth model toward a much cleaner and greener path [37]. Second, based on the idea of green tech bias, the digital economy works much better for the planet in places that are already good at green innovation. Basically, if a region has a strong green tech foundation, it can squeeze out more environmental benefits from its digital growth; that is, technological innovation, as an “amplifier”, effectively strengthens the marginal contribution of the digital economy to empowering green transformation [41]. Third, it shows that the digital economy does not always help green growth in the same way. It is not a straight line. Instead, it really depends on where a region is with its industry—like if they are still in heavy factory mode or moving toward services [42]. This positive effect gets even stronger as a region’s industry keeps upgrading. Basically, the more advanced the local economy becomes, the more it benefits from going digital, which means that under the high-level industrial structure, the digital economy can release stronger green explosive power. Finally, differences in geographical endowments and factor allocation efficiency led to significant regional heterogeneity. Spatially, the midstream and southwestern clusters benefited the most due to their rapid growth, while the marginal effect in the YRD tended to weaken. In terms of factors, while optimal resource distribution provided a strong impetus for the green transition of the digital economy, inefficient allocation acted as a structural bottleneck.

Research results indicate that the digital economy promotes diversified pathways to green development by enhancing factor mobility and industrial coupling. However, potential limitations in areas such as environmental governance and transportation infrastructure still require attention. To further narrow regional gaps in green development, future efforts should focus on deepening the understanding of resource flows and exploring how to cultivate a sustainable digital–green ecosystem in the long term.

Fostering a ‘win–win’ scenario for the economy and the environment is vital for the YREB. Therefore, the following policies are recommended.

First off, we need to keep improving the environment for the digital economy. That way, it can really start to help green development all over the world. It is not just about boosting one country’s economy; it is a global tool that helps everyone reach the UN’s sustainability goals [60]. On the policy side, building 5G networks is great, but we should not stop there. We really need to fix how data is bought and sold in the market. We need a fair and smart way to move data around so it is used where it is needed most, breaking down data barriers between regions and industries, and improving the cross-border efficiency and cross-regional flow of digital production factors. By enhancing the effectiveness of digital governance, reducing transaction costs and optimizing resource allocation, the digital economy can generate significant efficiency benefits. At the same time, we must be vigilant against the widening of the global “digital divide”, preventing an excessive concentration of digital resources in a few developed regions, thereby weakening the positive spillover effect of the digital economy on global carbon neutrality goals and ensuring that the benefits of green transformation reach more countries.

Second, we need to make sure digital growth and green tech move forward together. The research is clear: when green tech improves, it supercharges the digital economy’s ability to help both the planet and the economy. For one thing, it really cuts down on carbon, makes energy use way more efficient, and helps businesses shift toward a ‘waste-less’ circular model. On the other hand, it provides a “technological nervous system” for environmental governance (such as AI-based smart grids), enabling precise emissions tracking and climate governance. Therefore, innovation incentive mechanisms should be strengthened through various policy tools such as fiscal support, intellectual property protection, and green finance. Also, we really need to advance the role that digital tech plays in saving energy and cutting pollution. Whether it is cleaner manufacturing or better environmental management, we must blend them together. This is how we get the most ‘green value’ out of our digital tools.

Finally, given the threshold effect of industrial structure and regional heterogeneity, differentiated regional and structural policies should be implemented. For places where the digital economy is already well-established, we need to stop trying to just grow. Instead, we should focus on improving and let green innovation lead the way, avoiding a reliance on digital economy expansion that leads to diminishing ecological effects. For regions with advanced industrial structures, the focus should be on developing high-end digital service industries and green innovation industries to continuously release the incremental effects of green development. For regions changing their industrial landscape, the goal is to help legacy industries adopt digital and become greener at the same time. We also need to push for market reforms that help labor, money, and data move to where they are most needed. In places where resources are not yet being used well, fixing these institutional gaps will finally let the digital economy’s green potential shine through.

7. Limitations

While this study offers new insights, it is not without its limits—mostly due to data gaps. For one, our way of measuring the digital economy could be more thorough. Future work should probably include “digital value” to build a more rounded picture of digital growth. Also, with green finance becoming a huge deal globally, it is worth exploring how digital tools can actually support it. Lastly, since this paper mostly sticks to linear and threshold effects, the next step would be to look into spatial spillovers and how different regions interact. This would really help push the boundaries of what we know about digital–green synergy.