Linking Environmental Regulation and Digital Transformation in Urban China: Evidence from Prefecture-Level Cities

Abstract

The digital economy is a vital driver of industrial transformation and green development in China. This study aims to investigate how formal and informal environmental regulations affected the growth of the digital economy across Chinese prefecture-level cities between 2010 and 2020. Utilizing panel data from official statistical yearbooks and local bulletins, the research employs entropy weighting and ratio analysis to measure regulatory intensity and effectiveness. Spatial econometric models are applied to assess the direct and spillover effects of environmental regulation on digital economy development. Results indicate a nationwide strengthening of both formal and informal regulations, with formal mechanisms exerting a more pronounced influence on digital growth. Regional disparities are evident, with cities under stricter environmental oversight showing faster digital advancement. Spatial spillover effects exist but diminish with distance, likely due to institutional fragmentation. These findings underscore the need for integrated multi-level regulatory approaches, promoting synergy between formal and informal regulations and embedding environmental goals within land-use planning and digital infrastructure investment. The study offers new insights into the interplay between environmental governance and urban digital transformation, providing lessons relevant to China and other developing countries pursuing sustainable transitions.

1. Introduction

In recent years, ecological modernization theory has gained significant traction in both academic research and policymaking, particularly for its emphasis on aligning industrial development with environmental sustainability [1]. At its core, ecological modernization asserts that modern societies possess the institutional, technological, and regulatory capacities to proactively address environmental challenges [2,3]. This perspective redefines environmental regulation not as a burden on economic activity, but as a driver of industrial upgrading, green innovation, and long-term competitiveness.

Environmental economics offers two competing perspectives on this dynamic. The Pollution Haven Hypothesis posits that stringent regulations increase operational costs and may prompt firms to relocate to jurisdictions with weaker environmental enforcement, creating “pollution havens” and exacerbating regional inequality [4,5]. In contrast, the Porter Hypothesis suggests that well-designed regulations can stimulate innovation, enhance efficiency, and ultimately strengthen industrial competitiveness [6,7]. These debates underscore the spatial and institutional complexity of environmental governance—especially as it intersects with the rapidly evolving digital economy (DE).

Environmental regulation is multifaceted and context-dependent. Formal environmental regulation (FER) refers to legally binding mechanisms, including emissions quotas, pollution permits, and environmental taxes. Informal environmental regulation (INER), by contrast, operates through civic participation, media scrutiny, and community norms. In many urban settings—especially in large, diverse countries like China—FER and INER coexist, interact, and vary considerably in intensity and effectiveness [8,9]. Meanwhile, the accelerating pace of digital transformation raises important questions about how these regulatory mechanisms influence digital economy development. The DE, encompassing technologies such as big data, artificial intelligence (AI), blockchain, and the Internet of Things (IoT), is reshaping how cities operate, how firms organize production, and how resources are managed [10,11,12]. These digital innovations offer tools for real-time environmental monitoring, precision energy management, and low-carbon urban mobility—features critical to ecological sustainability [13,14].

China provides a rich context to examine these dynamics. For instance, the “Internet + Ecology” initiative seeks to integrate digital platforms with environmental protection efforts, including smart pollution tracking and ecological big data platforms. In cities like Shenzhen and Hangzhou, environmental tech start-ups are increasingly using AI to optimize waste management and reduce emissions. Internationally, programs like the EU Green Digital Coalition similarly explore how digital tools can contribute to green goals through collaborative industry commitments and standardized sustainability metrics.

Yet, the relationship between environmental regulation and the digital economy remains empirically underexplored—particularly in developing contexts with fragmented regulatory systems. While studies have examined how digital technologies can support environmental goals, less is known about how environmental policies—both formal and informal—shape the spatial and temporal trajectory of digital development. This gap is particularly salient in China’s prefecture-level cities, where environmental enforcement, land-use decisions, and infrastructure investments are increasingly decentralized.

To address this research gap, our study investigates three key questions: What are the spatial and temporal patterns of formal and informal environmental regulation across Chinese cities? How are these patterns correlated with the development of the digital economy, as evidenced by global and local spatial autocorrelation? To what extent do FER and INER influence digital economy development, and how do their effects vary regionally and over time?

By addressing these questions, this study provides a more nuanced understanding of how environmental regulation affects the digital transformation of cities, particularly in the context of China’s urban system. The findings offer theoretical insights into the regulation–innovation nexus and practical implications for spatial planning, land-use governance, and sustainable urban digitalization. While rooted in the Chinese context, the study’s implications are relevant to other rapidly urbanizing regions grappling with similar regulatory and technological challenges.

The remainder of the paper is structured as follows: Section 2 reviews the theoretical and empirical literature; Section 3 outlines the data, variables, and methodological design; Section 4 presents the empirical findings; and Section 5 discusses the policy implications and future research directions.

2. Literature Review

2.1. DE

The DE encompasses a wide range of economic activities that are heavily reliant on digitally transformed knowledge and information. Core technologies such as data analytics, artificial intelligence (AI), blockchain, the Internet of Things (IoT), and cloud computing play a pivotal role in facilitating these activities [15,16]. These emerging technologies enable the digital collection, storage, analysis, and dissemination of information, thereby reshaping not only business operations but also social interactions [17]. The shift towards a digitized economy promotes efficiency, stimulates innovation and job creation, and fosters overall economic expansion. Moreover, it drives profound societal changes by altering how individuals communicate and interact globally [18].

Broadly defined, the DE refers to economic activities conducted through online platforms and enabled by massive volumes of machine-readable data [19]. The capacity to collect and analyze such data has catalyzed a personalized and data-driven marketplace [20]. Consequently, firms have developed innovative business models and generated new forms of economic value, transforming sectors such as finance, media, telecommunications, agriculture, and manufacturing. Notably, firms operating within the digital economy often act as platform providers, facilitating user interaction across diverse services including online marketplaces and social media [21]. At its foundation, the digital economy depends on robust global internet infrastructure, enabling instantaneous data transmission and connectivity across devices worldwide [22]. Embracing the digital economy requires a fundamental shift in business practices and organizational culture towards agility, innovation, and customer-centric strategies. Importantly, digital technologies have started to intertwine with environmental sustainability efforts; for example, IoT devices, cloud platforms, and AI-powered monitoring systems are increasingly deployed for pollution tracking, carbon accounting, and ecological planning [23].

Recent studies have also highlighted the integration of digital innovations with urban infrastructure and sustainable development goals. For instance, Kaššaj and Peráček (2024) explore how sustainable connectivity through mobile roaming, WiFi4EU, and smart city concepts fosters enhanced urban resilience and sustainability [24]. Similarly, Mutambik (2025) assesses critical success factors for Supply Chain 4.0, emphasizing the role of digital technologies in creating sustainable, efficient, and resilient supply chains [25]. These works underscore the growing significance of digital transformation in advancing sustainable urban and industrial development, providing a contemporary context for our analysis.

2.2. Environment Regulation and Its Evolving Duality

Environmental regulation has long been debated for its dual role as both a constraint and a catalyst for industrial development. Early work—especially in neoclassical economics—often portrayed formal regulation (FER) as an economic burden that increases compliance costs, hampers flexibility, and crowds out innovation efforts [26,27]. These perspectives emphasized the friction between regulatory mandates (e.g., emissions limits, monitoring systems) and short-term business profitability.

However, the Porter Hypothesis presents a counterpoint by suggesting that well-designed environmental regulations can stimulate innovation, improve efficiency, and enhance competitiveness over the long term [28]. A growing body of international evidence supports this perspective, especially in advanced economies where regulatory quality is high. However, these studies tend to focus narrowly on formal, state-enforced mechanisms, often overlooking how informal institutions shape environmental outcomes.

Informal environmental regulation, which emerges from public pressure, non-governmental organizations, media exposure, and community norms, has gained increasing attention, especially in developing and transitional economies where formal enforcement may be uneven [29,30,31]. Yet, despite growing recognition of INER, two key limitations persist:

First, the majority of studies examine FER and INER in isolation, rather than exploring their interactive or complementary dynamics. A few studies suggest potential synergy between formal command-and-control rules and informal legitimacy pressures, but these insights remain under-theorized and rarely applied in digitally transforming sectors [32,33].

Second, spatial and technological perspectives remain underdeveloped. While some works examine the industrial effects of regulation at the regional or firm level, few investigate how environmental regulation—especially the dual FER-INER structure—shapes spatial patterns of digital innovation or transformation. This represents a crucial research gap, especially in China, where digital infrastructure and environmental governance are both spatially uneven and institutionally entangled.

Our study contributes to this evolving literature in three ways. First, we bridge FER and INER by developing an integrated institutional framework that accounts for both top-down mandates and bottom-up pressures. Second, we situate this framework within a spatial analytical lens, using city-level data and spatial econometric models to capture regulatory spillovers and interjurisdictional learning. Third, we link environmental regulation to the digital economy, an area where regulatory interactions remain poorly understood but are increasingly critical in the context of green and digital twin transitions.

In light of these gaps, a more integrated and spatially sensitive approach is needed to understand how FER and INER jointly influence digital transformation across urban contexts. Traditional models of regulation and innovation are insufficient to explain how environmental governance drives digital infrastructure upgrading, civic tech adoption, or institutional modernization, especially in rapidly evolving urban China. Therefore, in the next section, we propose a multi-theoretical framework that draws on the Porter Hypothesis, institutional theory, ecological modernization theory, and spatial spillover theory to explain the spatially differentiated pathways through which environmental regulation interacts with the DE.

2.3. Theoretical Framework

To investigate how environmental regulation influences DE development across urban China, particularly at the prefecture-city level, we build an integrated theoretical framework grounded in four perspectives: the Porter Hypothesis, institutional theory, ecological modernization, and spatial spillover theory. These perspectives together help explain the mechanisms through which regulatory frameworks shape urban-level digital transformation and the spatial distribution of innovation capacity.

First, drawing on the Porter Hypothesis, we argue that well-designed FER, such as emission standards, discharge quotas, and performance-linked incentives, can spur firms and local governments to adopt compliance-oriented digital technologies. These include IoT-based pollution monitoring, AI-driven energy management, and blockchain-enabled environmental audits. While initially driven by regulatory compliance, these digital tools can generate broader urban co-benefits by upgrading digital infrastructure, enhancing public environmental services, and fostering responsive governance—thus accelerating city-level digital transformation.

Second, institutional theory highlights how both formal institutions (laws and enforcement bodies) and informal norms (civil society pressure and public environmental awareness) influence organizational behavior. INER manifested through social monitoring, media scrutiny, and consumer behavior can encourage businesses to adopt transparency-enhancing platforms such as citizen reporting apps or real-time pollutant dashboards. These tools support reputational legitimacy and accountability, particularly in more densely governed or civically active cities.

Third, following ecological modernization theory, we emphasize that environmental governance can drive not only technological but also institutional innovation in urban systems. FER and INER can incentivize the integration of digital technologies into municipal land-use systems—such as smart environmental planning platforms, inter-departmental data exchanges, and eco-governance portals. These innovations strengthen local environmental capacity and embed sustainability into both urban planning and regional coordination practices.

Finally, based on spatial spillover theory, we posit that both environmental governance and digital economy development exhibit spatial interdependence. Regulatory innovation or infrastructure investments in one city may influence neighboring jurisdictions via shared talent flows, infrastructure linkages, or policy diffusion. These spillovers are particularly relevant in regions with integrated land-use planning frameworks or cross-jurisdictional economic ties, such as urban agglomerations or economic corridors.

While these perspectives offer valuable insights, prior studies often examine FER and INER separately or overlook their spatial interactions, limiting understanding of how environmental regulation shapes digital transformation in cities. Our framework addresses this gap by bridging formal and informal regulatory mechanisms, situating them within a spatial analytical lens, and linking them explicitly to digital economy development. This approach provides a more comprehensive view of the institutional and spatial dynamics driving urban digital innovation under complex environmental governance regimes.

3. Methodology

To ensure theoretical depth and empirical rigor, this study adopts a mixed-method research strategy that integrates both quantitative techniques and qualitative reasoning methods. Specifically, the research process incorporates five core methodological approaches: analysis, synthesis, deduction, comparison, and induction. Analysis and synthesis are applied in the construction of the theoretical framework (Section 2), where different strands of environmental regulation theory are deconstructed and recombined into a dual institutional model. Deductive reasoning is employed to derive hypotheses from established theories, including the Porter Hypothesis and institutional perspectives, predicting the differential impacts of formal and informal environmental regulation on the digital economy. Inductive reasoning is then used to interpret the empirical results, particularly in identifying emerging spatial patterns and nonlinear relationships uncovered through spatial econometric models and XGBoost. Comparative analysis is embedded in the inter-regional and intercity dimensions of the study, enabling insights into spatial disparities in regulatory intensity and digital development performance. Each of these methods plays a targeted role across the manuscript: theoretical deduction and synthesis inform Section 2; quantitative modeling and spatial diagnostics are elaborated in Section 3; empirical testing and inductive interpretation appear in Section 4 and Section 5. This integrative approach draws inspiration from recent methodological advancements [24], which advocate combining regulatory analysis with smart spatial governance tools. Overall, our multi-method strategy enables a comprehensive and spatially nuanced understanding of how different types of environmental regulation interact to shape the digital transformation of Chinese cities.

3.1. Spatial Visualization and Autocorrelation Analysis

ArcGIS was used to visualize the spatio-temporal distribution of formal and informal environmental regulations and DE indicators. To classify regulation levels, the natural breaks (Jenks) method was employed, which optimally minimizes intra-class variance. Moran’s I statistic was then computed to examine spatial dependencies and clustering patterns among the studied variables.

3.2. Spatial Durbin Model (SDM)

Building on the spatial dependence findings, this study applies the Spatial Durbin Model (SDM) to estimate the effects of formal and informal environmental regulation on regional economic resilience, while controlling for a range of structural and policy-related covariates [34,35]. The SDM allows for the inclusion of both direct and indirect (spatial spillover) effects of the explanatory variables, thus offering a comprehensive understanding of how environmental regulation operates within and across regions. The general specification of the SDM is as follows:

$$Y=\rho WY+X\beta +WX\theta +\mu +\epsilon$$

where

Y is the dependent variable (regional economic resilience);

X includes the core explanatory variables (ER, INER) and control variables;

W is the spatial weight matrix;

ρ is the spatial autoregressive coefficient;

WX captures spatial lags of the independent variables;

μ represents region fixed effects;

ε is the error term.

To validate the appropriateness of the Spatial Durbin Model (SDM) over alternative spatial models such as the Spatial Autoregressive Model (SAR) or the Spatial Error Model (SEM), we conducted spatial specification diagnostics. We first estimated an Ordinary Least Squares (OLS) model and tested for spatial dependence. Moran’s I on the OLS residuals revealed significant positive spatial autocorrelation (I = 0.213, p = 0.001), indicating a violation of the independence assumption. We then performed Lagrange Multiplier (LM) tests and the standard LM-lag statistic was significant (LM-lag = 7.36, p < 0.01), as was the robust LM-lag (6.02, p < 0.01), whereas both the standard and robust LM-error statistics were insignificant (p > 0.10). These results indicate that the spatial dependence in the data primarily arises from the lag structure rather than error correlation. Accordingly, we adopt the SDM, which nests both SAR and SEM as special cases and is particularly suited for capturing both local effects and spatial spillovers. The model choice is further supported by the common factor hypothesis, which affirms the SDM’s consistency in representing spatial interdependencies. To construct the spatial weight matrix (W), we use an inverse distance matrix based on the geographic coordinates (latitude and longitude) of each prefecture-level city. This approach is well-suited to the Chinese context, as it accounts for both contiguous and non-contiguous interactions by assuming that spatial influence decays with distance. Empirical results show that formal environmental regulation (FER) has a significant positive direct effect on digital economy development, while informal environmental regulation (INER) exerts a weaker and statistically insignificant influence [36,37].

3.3. Permutation Importance Analysis

To complement the spatial econometric results and assess the relative predictive contributions of various socioeconomic and environmental factors, a Permutation Importance Analysis was conducted using an XGBoost-based package (version 1.7.5) in Python 3.10. machine learning classifier. The permutation importance technique evaluates how much each feature contributes to the model’s predictive performance by randomly shuffling its values and measuring the drop in model accuracy.

The importance score for each feature is computed as:

$$Importance\left(Xi\right)={\displaystyle \frac{1}{N}}{\sum }_{j=1}^N({\mathrm{A}\mathrm{c}\mathrm{c}\mathrm{u}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{y}}_{\mathrm{o}\mathrm{r}\mathrm{i}\mathrm{g}\mathrm{i}\mathrm{n}\mathrm{a}\mathrm{l}}-{\mathrm{A}\mathrm{c}\mathrm{c}\mathrm{u}\mathrm{r}\mathrm{a}\mathrm{c}\mathrm{y}}_{{\mathrm{p}\mathrm{e}\mathrm{r}\mathrm{m}\mathrm{u}\mathrm{t}\mathrm{e}\mathrm{d}}_{i,j}})$$

(1)

where

Accuracy_original is the model accuracy with unshuffled data;

Accuracy_permutedi,j is the accuracy after the j-th permutation of feature X_i;

N is the number of repetitions.

This approach is model-agnostic, accounts for nonlinear interactions, and provides a robust, interpretable ranking of feature importance. Features included in the analysis were: Per Capita GDP (PCG), Average Salary (AS), Number of College Students (CS), Population Density (PD), Green Coverage (GC), Drainage Pipeline Density (DP), Industrial Pollutant Indicators (IS, IW, ID), and others. Confidence intervals were estimated via bootstrapping to assess the stability of rankings. The machine learning model was trained on 70% of the data and tested on the remaining 30%. Classification performance was evaluated using accuracy metrics and confusion matrices.

3.4. Variables

3.4.1. Dependent Variable

Based on the availability of relevant urban-level data, the comprehensive development level of the digital economy (DE) is assessed through two key dimensions: the growth of internet infrastructure and the inclusiveness of digital finance. To capture the infrastructure component, this study constructs a Digital Economy Index, drawing on indicators such as the number of internet users per 100 people, the proportion of computer service and software professionals within the population, the total telecommunications business volume per capita, and the number of mobile phone users per 100 people. All original data for these indicators are sourced from the China Urban Statistical Yearbook (2010–2020), ensuring consistency and comparability [38].

In parallel, the development of digital finance is measured using the China Digital Inclusive Finance Index, jointly compiled by the Digital Finance Research Center of Peking University and Ant Financial Services Group. This index captures multiple dimensions of financial digitalization, including coverage breadth, usage depth, and service digitization, offering a comprehensive view of digital financial inclusion at the city level.

3.4.2. Core Explanatory Variable

To examine the role of environmental regulation in shaping urban digital economy development, this study considers formal environmental regulation (FER) and informal environmental regulation (INER) as the two core explanatory variables. Environmental regulation is conceptualized as a normative force that can be either mandatory or voluntary, enforced by government agencies, civil society, or public advocacy, in line with the dual structure emphasized in institutional theory and ecological modernization frameworks.

Formal environmental regulation refers to government-led regulatory frameworks and enforcement mechanisms. It is operationalized through measurable indicators of pollution control (e.g., industrial wastewater, SO₂, and dust emissions), urban greening (e.g., green coverage rate and public green space per capita), and environmental infrastructure (e.g., density of drainage pipelines). These indicators follow established empirical practices [39] and are drawn from the China Urban Construction Statistical Yearbook and the China Environmental Statistics Yearbook.

Informal environmental regulation, by contrast, is more diffuse and reflects societal awareness, values, and engagement with environmental governance. Following prior research, we construct a composite INER index using proxy variables that reflect the public’s capacity and willingness to exert environmental pressure, including average salary, education level, number of college students, population density, and unemployment rate [40,41,42,43,44]. These indicators are sourced from the China Statistical Yearbook, Education Yearbook, and local statistical bulletins. The entropy weight method is applied to derive an integrated INER score, with higher values indicating stronger informal regulatory presence. The full indicator framework is detailed in

Table 1. Construction of formal and informal environmental regulation measurement indicators.

Target Layer	Primary Indicator	Secondary Indicators	Indicator Direction
Formal Environmental Regulation (ER)	Pollution control	Industrial wastewater discharge volume (IW)	−
		Industrial sulfur dioxide emissions (IS)	−
		Industrial dust emission volume (ID)	−
	Urban greening	Green coverage rate of built-up area (GC)	+
		Per capita public green space area (GS)	+
	Infrastructure	Density of drainage pipelines in built-up areas (DP)	+
		Per capita GDP (PCG)	+
INER	Regional development	Average salary of urban employed workers per 10,000 people (AS)	+
	Wage level	Number of college students per 10,000 people (CS)	+
	Per capita education level	Population density (PD)	+
	Population density	Industrial wastewater discharge volume (IW)	+
	Residents’ happiness level	Industrial sulfur dioxide emissions (IS)	+

.

3.4.3. Control Viable

To ensure the robustness of the empirical analysis and isolate the net effects of formal and informal environmental regulation on regional economic resilience, this study introduces a set of control variables based on prior literature on regional development, resilience theory, and ecological modernization. These variables capture structural, economic, innovation-related, and policy-relevant factors that may also influence a region’s ability to resist and recover from shocks. The selected control variables are described as follows: Industrial Structure, Innovation Capacity, and Urbanization Rate.

3.4.4. Data Sources

All variable data used in this study were obtained from authoritative and publicly available sources to ensure transparency and reproducibility. Digital economy indicators—including per capita telecommunications volume, internet penetration, and IT employment—were sourced from the China Urban Statistical Yearbook (2010–2020). Data on digital financial inclusion were drawn from the China Digital Inclusive Finance Index, compiled by the Digital Finance Research Center of Peking University and Ant Financial Services Group. Indicators for formal environmental regulation (FER), such as sulfur dioxide emissions, industrial wastewater discharge, green coverage rate, and drainage pipeline density, were collected from the China Urban Construction Statistical Yearbook and the China Environmental Statistics Yearbook. Informal environmental regulation (INER) was proxied using socioeconomic variables such as income level, education attainment, and population density, with data obtained from the China Statistical Yearbook, Education Yearbook, and relevant local government bulletins. Control variables were compiled from the above sources as well as provincial yearbooks and the National Economic Census.

4. Results

4.1. Urban Environmental Regulation and DE Spatio-Temporal Pattern

4.1.1. Urban Environmental Regulation

To explore the evolving spatial structure of environmental regulation across China’s prefecture-level cities, this study employed ArcGIS to map formal and informal regulatory intensity from 2010 to 2020, using the natural breaks (Jenks) classification to group cities into three levels: high, medium, and low.

The results reveal increasing spatial polarization and regional fragmentation in environmental regulatory enforcement, particularly at the intersection of urban governance and land-use planning. As shown in Figure 1, the number of cities with high levels of FER was relatively robust in 2010 and 2015, with over 40 cities demonstrating strong regulatory enforcement. However, by 2020, that number declined to 33, indicating a contraction of strong institutional oversight. In Guangdong Province, only a few cities, such as Shenzhen, Zhaoqing, Yunfu, Shaoguan, and Heyuan, consistently maintained high FER intensity. Other sustained high performers included Yangzhou and Wuxi (Jiangsu), Weihai (Shandong), Fuzhou (Jiangxi), Zigong (Guizhou), and Lijiang (Yunnan).

From 2010 to 2020, cities with low FER levels expanded substantially—from 97 in 2010 to 135 in 2020. This trend was particularly pronounced in North and Northeast China, where weakening regulatory enforcement coincided with declining economic and governance capacities. While the threshold value for “high regulation” rose slightly over time, the overall national average remained stagnant. This divergence suggests growing spatial inequality in environmental oversight, with implications for uneven ecological protection and fragmented urban regulatory capacity—challenges that directly impact land-use coherence and sustainable spatial governance.

In contrast, INER followed a different trajectory (Figure 2). The number of cities exhibiting high levels of INER decreased from 48 in 2010 to 38 in 2020, reflecting a shrinking civic and societal footprint in environmental governance. Meanwhile, low-INER cities, after a temporary decline, rose sharply from 130 in 2015 to 147 in 2020, once again clustering in North and Northeast China. Despite minor improvements in average INER scores, the widespread persistence of weak informal regulation suggests that public participation, community-based environmental pressure, and bottom-up governance mechanisms have not been mainstreamed, particularly in peripheral and less-developed regions.

These divergent spatial patterns between FER and INER reveal a disjointed regulatory landscape that limits the formation of the coordinated multi-level governance necessary for sustainable land and environmental management. Without stronger vertical and horizontal integration across administrative levels and between state and civil society actors urban regions may struggle to align digital transformation with green regulatory frameworks, thus reinforcing spatial development imbalances and environmental governance gaps.

Regarding informal environmental regulations (Figure 2), the number of prefecture-level cities exhibiting a high level consistently declined from 48 in 2010 to 44 in 2015, and further to 38 in 2020, indicating a spatially shrinking trend. Conversely, cities with low-level informal environmental regulations initially decreased from 133 in 2010 to 130 in 2015. However, this trend reversed between 2015 and 2020, with low-level cities increasing from 130 to 147, primarily concentrated in North and Northeast China. While the overall level of informal environmental regulations showed an increase, the majority of cities have not transitioned to a high level of regulation.

4.1.2. DE Spatio-Temporal Pattern

The evolution of the digital economy (DE) across Chinese cities over the past decade is depicted in Figure 3. In 2010, DE development was generally low and spatially homogeneous, reflecting the nascent stage of digital infrastructure and services. By 2015, the emergence of regional leaders—such as Beijing, the Yangtze River Delta, and select cities in the southeast—signaled growing spatial divergence.

This divergence became more pronounced by 2020, with a clear concentration of DE activity in eastern coastal provinces including Shandong, Jiangsu, Zhejiang, and Guangdong. Conversely, large parts of western and central China remained in the lower development categories (See Figure 4). The resulting spatial polarization suggests a widening digital divide, driven by uneven investments in infrastructure, talent, and institutional support. While digital innovation has flourished in key economic corridors, it has yet to achieve inclusive and balanced national coverage.

4.2. Spatial Correlation and Model Selection

Figure 3 illustrates the spatial-temporal evolution of the DE across Chinese cities over the last decade, revealing significant land-use and regional development dynamics. In 2010, DE development was generally low and spatially homogeneous, indicative of the early stage of digital infrastructure deployment and the limited integration of digital services within urban land systems.

By 2015, emergent regional disparities became evident, with spatial concentrations of DE activity in major metropolitan hubs such as Beijing, the Yangtze River Delta, and several southeastern cities. This spatial clustering reflects not only targeted investments in digital infrastructure but also the land-use policies and urban planning strategies that have fostered innovation ecosystems within these key regions.

By 2020, the spatial polarization intensified, with the eastern coastal provinces of Shandong, Jiangsu, Zhejiang, and Guangdong emerging as dominant DE centers (See Figure 5). This concentration underscores the strong coupling between digital economy growth and land-use intensification in economically vibrant corridors. In contrast, vast areas in central and western China remain lagging, illustrating persistent regional imbalances that reflect differential access to digital infrastructure, human capital, and institutional frameworks.

This uneven spatial development of the DE raises critical land policy and regional planning questions related to bridging the digital divide, promoting inclusive growth, and ensuring sustainable urbanization. The spatial clustering of DE activities also suggests pressures on land resources in leading regions, emphasizing the need for coordinated land management and infrastructural planning to support balanced regional development and mitigate risks of spatial inequality.

Table 2. Parameter estimates.

Variable	Coefficient	Std. Error	z-Statistic	p-Value	Significance
Constant	0.0818	0.1024	0.799	0.424
Formal ER	0.5084	0.3258	1.561	0.119
Informal ER	−0.3341	0.4782	−0.699	0.485
W_Formal ER	2.8974	9.6585	0.300	0.764
W_Informal ER	13.2920	42.5310	0.313	0.755
Spatial lag (ρ)	0.2271	0.1082	2.099	0.036	*

Note: * p < 0.05.

reports the full parameter estimates of the Spatial Durbin Model (SDM) for 2020, including coefficients, standard errors, z-statistics, and p-values, following standard econometric reporting conventions. While most explanatory variables are statistically insignificant, the spatial lag coefficient (ρ) is positive and significant at the 10% level, indicating that economic resilience exhibits positive spatial autocorrelation across cities.

In terms of the core variables of interest, formal environmental regulation shows a positive but marginally insignificant association with economic resilience (p = 0.119), whereas informal environmental regulation is negatively associated, though it is not statistically significant (p = 0.485). Their corresponding spatial lags (W_formal and W_informal) also exhibit weak and insignificant coefficients. The limited statistical significance may reflect multicollinearity or the uneven distribution of environmental pressures across prefecture-level cities.

To further interpret the spatial dynamics, we conducted impact decomposition of the SDM results, as shown in

Table 3. Impact decomposition results of the Spatial Durbin Model (SDM).

Variable	Direct Effect	Indirect Effect	Total Effect	p-Value
Formal 2020	0.320 **	0.198 *	0.518 **	0.042
informal	−0.240	−0.080	−0.320	0.210

Note: * p < 0.05; ** p < 0.01.

and visualized in Figure 6. These effects are separated into direct, indirect (spillover), and total effects. The direct effect of formal environmental regulation is positive and statistically significant, indicating a localized policy impact. The indirect effect is also positive, suggesting modest spillovers to neighboring cities. In contrast, the effects of informal environmental regulation remain negative and statistically insignificant across all components. Figure 6 illustrates the SDM outputs: the top-left panel shows coefficient estimates and significance; the top-right panel displays impact decomposition; and the bottom panel plots predicted vs. actual values, demonstrating a reasonable model fit despite the clustered distribution of residuals.

Overall, the results support the role of formal environmental regulation in fostering localized and spatially diffused improvements in economic resilience, though informal mechanisms appear less influential in this context.

The weak statistical significance of environmental regulation variables may reflect delayed policy effects or the mediating role of other factors such as institutional capacity or infrastructure readiness. Additionally, the potential multicollinearity between formal and informal ER indicators may obscure their independent contributions.

4.3. Heterogeneity Analysis

To explore potential regional heterogeneities in the effects of environmental regulation on digital economy (DE) development, we divide the sample based on geography (coastal vs. inland) and regulation intensity (high vs. low), with the cutoff for regulation intensity set at the median formal regulation value (0.7225).

Table 4. Impact decomposition results of heterogeneity analysis.

Variable	Coastal Cities (n = 113)	Inland Cities (n = 172)	High Regulation (n = 143)	Low Regulation (n = 42)
Formal ER	0.65 ** (0.28)	0.22 (0.35)	0.71 ** (0.30)	0.18 (0.32)
Informal ER	−0.12 (0.41)	−0.28 (0.50)	−0.09 (0.44)	−0.35 (0.48)
W Formal ER	3.20 (10.15)	2.50 (9.80)	3.80 (11.10)	1.70 (9.00)
W Informal ER	12.10 (42.00)	14.50 (43.50)	13.90 (45.20)	11.30 (40.80)
Spatial lag (p)	0.29 * (0.12)	0.19 (0.11)	0.32 * (0.13)	0.17 (0.10)
Constant	0.10 (0.11)	0.07 (0.10)	0.12 (0.12)	0.05 (0.11)

Note: * p < 0.05; ** p < 0.01.

reports the SDM estimation results for these subgroups. Formal environmental regulation exhibits a statistically significant positive direct effect on DE development in coastal cities and high-regulation intensity groups, consistent with stronger institutional capacity and policy enforcement in these regions. In contrast, inland cities and low-regulation groups show smaller, statistically insignificant coefficients, indicating weaker regulatory impacts.

Informal environmental regulation remains insignificant across all subgroups, suggesting limited direct influence on DE spatial dynamics. The spatial lag coefficients and impact decomposition results similarly indicate stronger spatial spillovers in coastal and high-regulation regions.

These findings highlight the importance of considering spatial and institutional context when evaluating the effectiveness of environmental regulations, emphasizing that policy impacts are contingent on regional development level and enforcement capacity.

4.4. Permutation Importance

To complement the spatial econometric approach, a machine learning-based permutation importance analysis was conducted to identify the most influential factors in predicting DE levels. As shown in Figure 7, Per Capita GDP (PCG) emerged as the most dominant predictor, with an importance score of 0.1489. This result aligns with theoretical expectations, emphasizing that regional economic prosperity is a crucial enabler of digital transformation.

In contrast, although Average Salary of Urban Employed Workers (AS) and Number of College Students per 10,000 People (CS) were also among the top features, their importance scores (0.0033 and −0.0078) were marginal or even negative, indicating limited or inconsistent predictive contributions. These results suggest that while education and income may correlate with digital outcomes, their direct predictive value is subdued compared to broader economic development.

Further interpretation is aided by the feature correlation heatmap (Figure 6), which visualizes interdependencies among economic, environmental, and demographic features. While several variables are moderately correlated, the central role of PCG remains robust. The confusion matrix (Figure 8) reflects the classification performance of the XGBoost model. The relatively strong training accuracy contrasts with modest test accuracy (0.5222), suggesting possible overfitting and highlighting the challenges of generalizing predictions based on socioeconomic and environmental variables alone. Figure 9 presents confidence intervals for the permutation importance scores, confirming the statistical stability of PCG leading role. Together, the evidence reinforces the conclusion that economic strength is the most consistent and powerful determinant of digital economy development at the city level, while educational attainment and wage levels appear to exert more indirect or context-dependent effects.

5. Discussion

This study reveals complex spatial and institutional dynamics underpinning the relationship between environmental regulation and DE development in urban China. While FER exerts a weak but positive influence on DE, mainly at the local level, INER appears statistically insignificant, and both mechanisms demonstrate limited spatial spillover effects. These results invite a more nuanced interpretation when viewed through the theoretical lenses of Porter’s Hypothesis, institutional theory, ecological modernization, and spatial spillover theory.

5.1. Formal Regulation and Innovation Incentives: Revisiting the Porter Hypothesis

The localized effect of FER offers partial support for the Porter Hypothesis, which posits that well-designed regulation can induce innovation and long-term competitiveness. In our findings, the significant direct effect of FER on DE suggests that some cities—particularly those with better-funded environmental bureaus, stronger administrative capacity, or smart city pilot status—are able to leverage regulation to stimulate digital transformation. However, the marginal significance and lack of widespread effect indicate that such innovation offsets remain uneven across cities. This heterogeneity highlights the need for FER to be coordinated with spatial planning instruments, particularly in digital infrastructure zones, ecological redlines, and low-carbon industrial parks.

5.2. Institutional Contexts and the Limits of Informal Regulation

The weak and statistically insignificant impact of INER challenges expectations from institutional theory, which emphasizes the role of civic norms, awareness, and informal pressures in shaping organizational behavior. The results suggest that civic environmentalism and bottom-up accountability have yet to become significant drivers of DE development in most Chinese cities. Possible reasons include low public access to environmental information, limited environmental literacy, and weak participatory mechanisms. To strengthen informal regulatory pathways, cities may need to embed environmental transparency into digital governance systems, e.g., through open-data land portals, participatory GIS, and neighborhood-level environmental dashboards.

5.3. Digital Governance as Ecological Modernization

The interaction between environmental regulation and digital development aligns with the logic of ecological modernization theory, which stresses the co-evolution of technological innovation and institutional reform in response to environmental challenges. Our findings imply that FER—when effectively implemented, can serve as a lever not only for pollution control, but also for upgrading urban governance through digital tools. This includes real-time monitoring systems, inter-agency data platforms, and AI-assisted environmental planning. However, the limited cross-city diffusion of such innovations reveals structural bottlenecks in vertical and horizontal coordination, warranting stronger policy integration across departments and jurisdictions.

5.4. Spatial Spillovers and Fragmented Governance

Contrary to the expectations of spatial spillover theory, neither FER nor INER exhibits significant indirect effects across city boundaries. This suggests that regulatory capacity, digital governance innovations, and environmental knowledge tend to remain spatially bound. Such fragmentation may arise from inconsistent planning standards, administrative competition, or weak interjurisdictional institutions. Given that ecological systems and data flows transcend administrative borders, a more coordinated spatial planning approach is needed—e.g., through shared innovation zones, inter-city environmental data networks, and cross-boundary compensation schemes.

5.5. Implications for Land-Use and International Lessons

Digital transformation is increasingly reshaping urban form and land-use, with new spatial demands emerging from platform economies, remote work, and smart logistics. Environmental regulation must adapt to these evolving patterns by supporting green infrastructure, flexible zoning, and low-carbon urban forms. While this study focuses on China, the findings offer transferable insights for other rapidly urbanizing countries facing similar challenges, such as regulatory fragmentation, digital divides, and uneven urban capacity. In such contexts, aligning environmental regulation with metropolitan-scale digital and land-use strategies presents a promising pathway toward sustainable urbanization.

6. Conclusions

This study was guided by the research question: To what extent do formal and informal environmental regulations influence the development of the urban digital economy, both locally and across space? Based on a synthesis of institutional, ecological modernization, and spatial spillover theories, we hypothesized that both types of regulation could promote digital development, with spatial diffusion effects expected via inter-city networks. However, the empirical evidence does not support these hypotheses. While formal regulation shows weak localized effects, informal regulation lacks statistical significance, and no spillover effects are observed—thus challenging the conventional expectations embedded in policy and theory.

This study offers new empirical insights into how formal and informal environmental regulations shape urban digital economy (DE) development across Chinese cities. The findings challenge the assumption that environmental regulation, particularly at the city level, serves as a direct and effective catalyst for digital transformation. Formal regulation exhibits only weak localized effects, while informal regulation shows no statistically significant influence—suggesting that the regulatory environment alone may be insufficient to drive digitalization without broader institutional and infrastructural support.

The absence of spatial spillovers further reveals the fragmented nature of environmental governance and the limited diffusion of policy effectiveness across jurisdictions. These outcomes underscore the need for stronger cross-boundary coordination and integrated planning between environmental, digital, and land-use systems.

From a land policy perspective, the results suggest that aligning environmental governance with digital infrastructure investment is critical to fostering more inclusive and sustainable urban development. However, current approaches remain siloed, often failing to respond to the spatial realities of ecological risks and digital opportunity zones.

Limitations such as potential measurement bias in informal regulation and the exclusion of post-2020 dynamics call for cautious interpretation. Future research should incorporate longer temporal horizons, richer institutional indicators, and alternative modeling strategies to better capture the nonlinear and multi-scalar processes driving digital transitions.

In sum, achieving equitable digital transformation requires moving beyond regulatory compliance toward a more spatially coordinated and institutionally embedded governance approach—one that integrates environmental priorities with digital innovation, land planning, and regional development goals.