The Evolution of Urban Environmental Governance Networks: Evidence from China

Abstract

Urban environmental issues are fundamental to ecological civilization development, with key stakeholders such as governments, enterprises, social organizations, and community residents playing crucial roles in the governance process. From a network governance perspective, this study innovatively applied social network analysis (SNA) to policy co-occurrence networks, analyzing over 2300 policy documents related to China’s urban environmental governance from 2017 to 2023 to investigate evolutionary trends. The key findings indicate the following. (1) The comparative analysis across two periods reveals that China’s urban environmental governance network structure has stabilized, with both network density and centralization indices showing an upward trend. (2) The degree of centrality and betweenness centrality shows that government agencies remain the core entities within the network, while the role and influence of business enterprises have steadily increased. (3) The participation of social organizations in governance continues to increase, but community public participation in governance is insufficient. Therefore, China’s ecological environment governance network has formed a network structure with the government as the leader, enterprises as the key role, and social organizations providing effective support.

1. Introduction

With the rapid acceleration of urbanization, cities have emerged as the primary source of environmental challenges. The global environmental challenges have been increasingly severe, and among them, climate change and pollution governance are two of the most critical issues that need to be addressed urgently [1,2]. A high-quality urban environment can ensure ecological balance during expansion, enhance citizens’ quality of life, mitigate the urban heat island effect, and effectively promote green economic transitions. Efficient, precise, and sustainable urban environmental governance helps build healthy, livable cities, enhancing their resilience against climate change, air pollution, and other environmental challenges. This governance provides a solid foundation for dual-carbon development and aligns closely with global sustainable development goals.

However, China, the world’s largest developing country, faces unique challenges compared to developed Western countries, due to its rapid pace, large scale, significant urban–rural disparities, and swift development of urban clusters and metropolitan areas. These challenges include a large urban population, immense pressure on resources and ecological environments, and biodiversity conservation. This exerts distinctive environmental governance pressures and raises our core research question: How do China’s urbanization characteristics shape stakeholder governance relationships and structure urban environmental governance networks?

This paper investigates urban environmental governance issues in China, utilizing social network analysis (SNA) to examine the changes and development trends in governance network relationships. Traditional SNA applications in natural environmental issues often encounter limitations such as small samples, high costs of interviews and surveys, and insufficient regional representation. To address these challenges, this study innovatively selected a large sample of policy texts nationwide as data sources, aiming to effectively mitigate these constraints. Additionally, the study incorporated expert evaluations into the selection of textual analysis indicators based on machine filtering, enhancing the alignment of thematic keywords and dimensional indicators and reducing the influence of researchers’ subjective factors. This study examined whether policy co-occurrence networks effectively capture environmental governance relationships in urban systems, using large-scale sample network relationship data as empirical validation.

The current research emphasizes the importance of identifying diverse governance actors and constructing network governance mechanisms essential for green development [3]. Consequently, scholars have undertaken a series of studies on governance networks concerning ecological and environmental issues. First, systematic argumentation has provided theoretical support, highlighting the application and theoretical value of network governance in addressing environmental challenges [4,5]. Second, based on network model construction, researchers have explored the development of governance systems for natural resources such as fisheries and mining [6,7]. Third, model analysis and governance evaluation have been conducted using social network analysis (SNA) to assess biodiversity conservation networks [8,9]. In addition, researchers have conducted urban environmental governance studies on multiple fronts, including digital transformation driving corporate energy conservation and emission reduction, the effects of urban population changes on biodiversity, and sector-specific policies boosting ecosystem service functions [10,11,12]. These efforts are geared towards enhancing overall urban environmental quality [12,13], offering scientific evidence for global environmental policies and actions. This enables global environmental improvement and technological innovation, advancing high-quality coordinated development of environment, economy, and society. Studies have established the theoretical value of governance network research and used social network analysis (SNA) applications in biodiversity conservation and environmental policy evaluation. Furthermore, significant gaps persist, particularly regarding network evolution analysis and multi-level collaborative governance. Against China’s urban development backdrop, this study investigated stakeholder role dynamics within multi-tiered governance networks.

Urban environmental governance constitutes a complex system where stakeholders continuously interact, forming a core–periphery structure. Within the peripheral governance domain, four stakeholder categories exhibit tight interlinkages. Governments exert constraints and controls over industrial organizations and social groups through policy formulation and implementation. Industrial organizations influence government behavior via sectoral development initiatives—including tax contributions and appeals—to secure policy support. Social groups (e.g., NGOs, public opinion, media) establish communication channels with governments, coordinating demands for policy adjustments. Community residents, as agents of human activity, collectively impact core objectives. Yet given public environmental goods attributes and China’s government-led governance, residents predominantly function through demand articulation.

During the network governance process of urban environments, the demands of major stakeholders vary. The basic governance network framework, illustrated in Figure 1, focuses on three primary governance objectives: ecosystem services, biodiversity conservation, and urban economic development. The four main categories of stakeholders, through their distinct actions, form a web of interactions encompassing needs, behaviors, feedback, and relationships [14], influencing their decisions and judgments.

Building on this foundation, conducting research on urban environmental governance in China holds significant theoretical and practical value. First, China’s rapid urbanization, large population base, and diverse resource demands provide a wealth of research material. This contributes to refining related theoretical studies and exploring governance models suitable for different stages of socioeconomic development. Additionally, this research provides scientific evidence for urban environmental policymaking, promotes sustainable urban development, and offers insights and references for other developing countries. This research has tangible implications for global environmental governance and the construction of ecological civilization.

2. Materials and Methods

2.1. Subjects and Design

Accompanying China’s rapid economic growth brought by its reform and opening-up policies, urban environmental incidents such as smog, water pollution, and climate anomalies have garnered close attention from society. Over the last decade, the Chinese government has introduced significant national policies to address ecological and environmental governance issues. Notable milestones include the 18th National Congress of the Communist Party of China in 2012, which incorporated ecological civilization construction into the overall layout of building socialism with Chinese characteristics. In 2015, the Central Committee and the State Council issued the “Opinions on Accelerating Ecological Civilization Construction” and the “Overall Plan for the Reform of the Ecological Civilization System,” comprehensively deploying ecological civilization reforms. In September 2017, the “Overall Plan for Establishing a National Park System” was issued, followed by the “Guiding Opinions on Establishing a Nature Reserve System with National Parks as the Main Body” in 2019. On 22 September 2020, at the 75th session of the United Nations General Assembly, China officially announced its dual carbon goals. The ecological environment, with its characteristics of a typical public good, has led to increasing demands and awareness from governments, enterprises, related organizations, and communities, gradually forming a networked, diversified, and centralized governance structure in China.

The major environmental policy reforms in China have transformed grassroots governance and regional economic trajectories. The official data from 2024 China Ecological Environment Status Bulletin confirm striking progress. Urban PM_2.5 concentrations had plummeted 34.8% from 2015 baselines, while surface water quality compliance had surged 23.3 percentage points to 84.9% since 2012. Concurrently, carbon intensity fell 34.4% as biodiversity initiatives culminated in China’s pivotal leadership of the Kunming–Montreal Global Biodiversity Framework, cementing green development as China’s core growth paradigm.

This paper examines the formulation and execution of environmental and ecological policies, focusing on their effectiveness and representativeness in China’s environmental governance. Utilizing large-scale policy text analysis and social network analysis [15], this study explored the structure and trends of governance networks. Significant national policies have clearer and more profound implications compared to general laws and regulations. Therefore, the key policy documents are used as control elements [16]. Given the vast expanse of China and the lag in policy implementation, the sample study was divided into different phases to illustrate the evolving trends of China’s governance network. The research followed these basic steps: (1) collection and cleansing of policy texts; (2) identification of social networks and selection of stakeholder indicators; (3) analysis of urban environment governance networks; and (4) analysis of network evolution trends.

2.2. Data and Tool Analysis

In this study, China was selected as the overall research area for environmental governance networks, focusing on stakeholders such as government departments, business enterprises, social organizations, and the public. Interaction data were collected from publicly available information. Government departments’ data were mainly sourced from the Peking University Law Database, which contains extensive legal and policy documents from the establishment of the People’s Republic of China. This database provides comprehensive and accurate collections, utilizing advanced search technology to filter and match relevant information, ensuring the accuracy and relevance of the data collected. Policy text collection was carried out using keywords like “urban,” “environment,” “ecology,” and “nature protection,” gathering all central and local level documents from early 2015 to the end of 2023. Classification was mainly based on central governments at all levels and environmental governance authorities. The types were classified in accordance with the categories on the PKU law website. Additional relevant annual policy texts were collected from the State Council, the Ministry of Ecology and Environment, and various provincial government websites. Python web crawlers were used to scrape the data, which were then filtered and deduplicated, excluding texts with low relevance, industry-specific regulations, group rules, internal party regulations, and judicial interpretations. Ultimately, over 4700 policy texts were included. The basic information is summarized in the following table (

Table 1. Number of policy text types.

	Central Regulations (Articles)	Type	Local Regulations (Articles)
Law	60	Local Regulations	230
Administrative Regulations	190	Normative Documents	848
Departmental Regulations	409	Local Work Documents	3021
Total	4758

).

To avoid subjective errors in the autonomous selection of keyword quantities inherent in traditional methods, preliminary expert interviews were conducted before extracting thematic keywords from relationship data using Python 3.13. Nineteen experts were selected to score stakeholder associations via rating tables, establishing a relationship matrix for network and principal component analysis. Based on this, the most representative and widely recognized conceptual keywords were selected to form the relationship matrix for text analysis. Subsequently, a Chinese stop-word list was created based on similar studies to exclude irrelevant and meaningless stop words from the text. Utilizing the built-in Jieba module in Python, the selected text data were further cleaned, filtered, segmented, and keywords extracted. Thirdly, Python coding was employed to extract governance structure relationship matrices from policy texts, converting frequencies into rates to eliminate disparities in word counts across texts. Ucinet 6.0 was then used to analyze the transformed co-occurrence matrices through social network analysis, converting co-occurrence matrices into similarity matrices to reflect the consistency of relationships expressed by the selected texts and keywords, thereby validating the initial expert interview ratings. Finally, network analysis and feature studies were conducted based on the urban environment governance network status and parameter analysis results at different periods, aiming to uncover intrinsic governance mechanisms and external influencing factors.

2.3. Selection of Network Elements and Phase Comparison

In traditional network analysis, bidirectional relational data acquired through expert interviews or questionnaires among stakeholder groups often suffer from constrained sample sizes, prohibitive research costs, and limited regional representativeness. This study employed large-scale textual data to circumvent these limitations. Consequently, our methodology constructed a text-derived keyword repository for quantitative relationship extraction, where entities associated in policy documents were treated as unidirectional projected relationships. Given the Chinese government’s dominant role in governing public environmental goods, its policy stances serve as proxies for governance orientations and social dynamics, with textual keywords expressing operational pathways and expectations toward associated actors [17]. While unidirectional data collection proves more feasible for large-scale network research and reveals latent associations often obscured in bidirectional frameworks [18], it concurrently introduces challenges, including lower relationship reliability, reduced information accuracy, and incomplete structural representation.

Therefore, we employed social network analysis to study the trends in the relationship network of urban environment governance in China, reflecting its governance mechanism and structural system. Based on the recent publication nodes of major environmental policies in China and the impact of significant events in recent years, the study ultimately divided the periods into two phases: from the beginning of 2017 to the end of 2019, and from the beginning of 2021 to the end of 2023. This phase comparison was used to measure the social networks of stakeholders involved in urban ecological environment participatory governance during different periods, focusing on four dimensions: government departments, enterprises, social organizations, and the public. Indicator keywords are extracted and abbreviated, and the interactions between actors are displayed with lines, reflecting changes in governance trends and evaluating policy implementation effectiveness. The basic reference information is as follows.

As an example of stakeholders of government, we considered the data associated with policy regulations and announcements issued by individual departments as interaction behaviors between two participants in advancing the same project. This data collection method is representative and has been proven effective by many researchers, reducing the limitations of traditional SNA studies that often face high research costs, limited sample sizes, and homogeneity of survey groups [19]. Most of the research subjects were departmental units. We assumed that the communication interactions between participants were unidirectional, primarily reflected in the publication of official texts showing their current work and future plans, making the network a directed network.

Networks with high density and centralization facilitate better communication and interaction, leading to improved ecological governance characteristics and demonstrating the actual impact of regional policies [20]. Finally, to identify the most critical stakeholders in the regional environmental governance network, we calculated the betweenness centrality and degree of centrality of each stakeholder. This was to identify bridge entities that link other stakeholders and promote governance effectiveness aside from the core roles, thus recognizing their intermediary roles in the governance network.

Additionally, domestic and international research has shown that as a type of public good, the governance mechanisms of ecological environments are trending toward the steady development of diversified governance networks. In China, the construction of ecological civilization has made multi-stakeholder governance a critical evaluation indicator. This study attempted to incorporate network evolution trends over different periods to evaluate the effectiveness of policy guidance in ecological civilization construction. Consequently, understanding among governance participants about the impact of stakeholder relationships on natural resource governance effectiveness was established. In our study, we hypothesized that the stability of governance networks can reflect the level of natural environment governance. This hypothesis was analyzed and verified through a series of tests.

2.4. Research Hypotheses

In this study, environmental governance objectives under multi-stakeholder participatory governance aim to meet the interests of all parties. The research involved matching texts through keyword screening from each party and using keyword frequency and search indices to display relationship strength. Relevant data were encoded and cleaned, selecting key time nodes (important text publications) to form a research database. Network analysis was then conducted to evaluate trends in network changes and examine the trends in relationship indicators and their connection with regional attribute indicators [21].

We used Ucinet 6.0 to measure network density and centrality and netdraw module of it for visualizing the network analysis. Data collection and organization were supported by programming through various Python packages. Network density indicates the level of interaction among participants in the governance network. For instance, a density of 1 means all parties in the governance network can effectively communicate and provide feedback on natural governance, whereas a density of 0 indicates no participant collaboration in natural environment governance. Therefore, network density reflects the overall degree of coordination among participants, with the number of connections indicating the flow of information within the governance network. This can provide valuable insights for decision-making, with the hypothesis being that higher or lower density would impact the effectiveness of governance network construction. Centrality, with a core value of 1, means the core participants in an individual network are concentrated in a dominant group, and other stakeholders are linked to various parties only through the dominant group. A centrality value of 0 indicates evenly distributed relationships among participants, resulting in a loosely connected network lacking a dominant party to guide governance strategies, making policy implementation and model optimization more challenging.

Based on the above discussion, we propose the following hypotheses. First, given the public good nature of natural environments and resources, the connectivity and centrality of government, which naturally assumes primary responsibilities, in the network could significantly influence the effectiveness of regional environmental governance practices. Additionally, we map the relationships among stakeholders within the ecological governance network from different perspectives, measuring the network density and centrality values, and visualize the differences in stakeholder participation in governance across different regions [22,23]. Considering the current situation in China, our second hypothesis is that urban environment governance networks exhibit higher centralization, indicating enhanced coordination awareness under governance networks. The core–periphery structure has become more pronounced and centrality tends to increase, thereby clarifying the influence, practice, and implementation of ecological environment policies within the network.

3. Results

Before analyzing the collected policy texts and annual reports of listed companies, it was essential to identify the keyword components of their network co-occurrence matrices. Traditional text analysis typically involves segmenting texts using the Jieba module in Python and applying the LDA model in a reserved word library to cluster thematic words, thereby extracting the optimal themes and keywords for different periods. However, since this study selected two types of text materials—policy texts and listed company reports—to eliminate subjective biases of different stakeholders, we innovatively conducted expert surveys to collect relevant evaluation data after text segmentation. Experts rated the data through questionnaires, and the data were then subjected to principal component analysis using Ucinet software to determine the indicator keywords for each stakeholder dimension.

3.1. Network Identification and Indicator Selection

To mitigate limitations inherent in large-scale unidirectional network research and enhance analytical accuracy, this study engaged 19 domain experts representing key stakeholder groups during preliminary investigations. The cohort comprised: five government administrators (G1–G5) from urban park/protected area management agencies; five enterprise representatives (E1–E5), predominantly from manufacturing sectors; four social organization members (T1–T4), spanning associations, universities, media, and research institutes; and five community participants (S1–S5), including residents, tourists, and netizens. The group was selected from Guizhou Province, designated among China’s inaugural National Ecological Civilization Pilot Zones in 2016, and participants demonstrated heightened ecological awareness and established collaborative networks conducive to rigorous research engagement. Following comprehensive study briefings, respondents addressed 22 targeted questions corresponding to 22 stakeholder keywords extracted from the clustered textual repository (

Table 2. Examples of expert questions.

Question (22 Items)		Reserved Thesaurus
1	Do you think the keyword “xx” reflects the role positioning or interest needs of stakeholders such as the government, enterprises, social organizations, and communities in environmental governance?	Department, Central, Policy, Government, City, Provincial, Enterprise, Industry, Sector, Productivity, University, College, Research, social organization, Official organization, Association, Group, People, Community, Public, Citizens, Community
2	If the keyword “xx” can represent the above, please rate its importance in governance (1–5 points, a higher score means more important)

and

Table 3. Corresponding research indicators.

	Name	Indicator	Name	Indicator
1	Department	Dep.	College	Col.
2	Central	Cen.	University	Univ.
3	Policy	Pol.	Research	Res.
4	Government	Gov.	Academic	Acad.
5	City	Cit.	Association	Asso.
6	Provincial	Prov.	Official Organization	Org.
7	Enterprise	Ent.	People	Peop.
8	Industry	Sec.	Group	Grp.
9	Sector	Ind.	Public	Pub.
10	Productivity	Pro.	Citizen	Citz.
11	Social Organization	NGO	Community	Com.

).

Based on the expert ratings, keywords with a high proportion of negative responses (greater than 50%) were first excluded. After removing five terms—city, provincial committee, productivity, group, and populace—a 17 × 17 relationship matrix of expert ratings was established. This matrix was then subjected to principal component analysis using the Factor Analysis program within the Scaling module of Ucinet software. The visualization network was drawn using the built-in Netdraw module.

Figure 2 shows that the eigenvalues of the first four factors are all greater than 1, with a cumulative percentage of 85.9%. This indicates that the selected indicators are well-clustered into four dimensions. The cumulative percentage for the first two dimensions reaches 72.7%, demonstrating the representativeness and real-world reflection of the research design, which analyzes governance networks through text analysis from the perspectives of different stakeholders, particularly government and enterprise. Furthermore, it indicates that social and public sectors in China’s urban environment governance have attained a certain level of participation and expert consensus, although their influence remains limited. Lastly, based on the factor loading matrix table and the visual network from expert ratings, the keywords “association” and “group” were found to lack significant explanatory meaning. Therefore, they were excluded from subsequent research, finalizing a keyword library of 15 thematic keywords for use in Python programming to advance the subsequent research on governance network trend evolution.

3.2. Analysis of Urban Environment Governance Network Evolution Trends

Building on the 15 keywords selected through expert ratings, we also considered the central target dimension beyond the four stakeholder categories, specifically incorporating the sustainable development dimension of urban environment governance. During the initial text collection and sorting process, although keywords were used for collection, differences in volume, direction, and type of texts were also considered in the analysis. Thus, approximately 10 additional keywords were included in the urban environment dimension, ultimately constructing a specialized lexicon of 25 phrases from policy texts.

Based on the selected research periods, the texts were classified, resulting in 1096 urban environment governance policy texts from 2017 to 2019 and 1215 policy texts from 2021 to 2023. The exclusion of 2020 data stemmed from a dual rationale: first, China’s stringent pandemic containment measures significantly disrupted industry operations; second, policy document issuance volume and thematic priorities diverged markedly from other years, and this would create substantive analytical discontinuities for this text-based study. All texts were read using Python’s built-in modules. After obtaining the keyword frequency table, it was converted into a frequency table to eliminate the influence of different text content. Consequently, a 25 × 25 keyword co-occurrence matrix and a similarity matrix were obtained. Ucinet software was used to explore and visualize the trends in urban environment governance networks during these two periods.

We first selected 25 optimal keywords as network nodes, using their co-occurrence frequencies in different texts related to urban environment governance policies as weights. These weights were used to calculate the weighted network density, reflecting the collaboration effectiveness within the urban environment governance network through co-occurrence relationships. In network analysis, network density measures the ratio of actual relationships to the maximum number of potential relationships within the network. However, a large sample can cause the density value to decrease due to frequent connections. To eliminate the interference caused by a large sample, the frequency co-occurrence matrix was standardized before conducting the network analysis.

3.2.1. Overall Network Analysis

In an urban environment governance network, stakeholders are considered key nodes. Once indicators and dimensions were defined, the interactions and behaviors among stakeholders form the edges of the network were analyzed. Measuring network density reflected the fluency and connectivity of information and behaviors within the urban environment governance network [24,25]. Given the public good nature of ecological environments, governmental governance orientations substantially steer behavioral strategies across societal groups in China’s urban environmental governance. This study’s policy text analysis specifically traced the dynamic diffusion pathways of governance information—rather than assessing interaction depth or participation levels among stakeholders.

Crucially, unidirectional network construction facilitates large-scale relational data collection and analytical tractability, allowing focused structural analysis of urban environmental governance networks to elucidate latent transmission patterns and emerging environmental governance trajectories. Consequently, higher overall network density indicates closer ties among stakeholders, stronger group dependency, and better alignment with governance objectives. We used the Ucinet program to measure the network density of governance networks for two periods. The density values were found to be 0.096 and 0.101 (

Table 4. Network index.

Index	2017–2019	2021–2023
Network Density	0.096	0.101
Network Centralization Index (%)	1.15%	5.77%

), showing an overall upward trend.

3.2.2. Individual Network Characteristics

In the network, the degree of centrality of each node includes both in-degree and out-degree. In-degree represents the quantity of information resources and decision-making actions flowing into the stakeholder during policy implementation. Out-degree reflects the resources and actions the stakeholder outputs in the network. The difference between in-degree and out-degree can indicate the stakeholder’s level of interest in the network.

Figure 3 shows the degree of centrality (standardized in-degree and out-degree) of nodes across different periods. Compared to the first period, both in-degree and out-degree values increased in the second period, indicating more active connectivity among nodes and enhanced interactivity in the governance network, which strengthens collaboration. Based on the differences between in-degree and out-degree, nodes can be categorized into three types, as follows.

(1) Dominant Participants: These have a much higher out-degree than in-degree, setting the development goals for urban environment governance. (2) Balanced Participants: These have relatively equal in-degree and out-degree, playing a role in adjusting and optimizing governance. (3) Receptive Participants: These have a higher in-degree than out-degree, acting more passively and following the dominant participants’ policies. Overall, the governance network is mainly led by government and enterprises, with clear governance objectives. The public acts as an adjuster of governance goals, while social organizations in China play a more passive role, relying on dominant participants for organization, research, and coordinated management.

Additionally, we measured the betweenness centrality and closeness centrality of the network structure at different periods (Figure 4). Higher betweenness centrality for a node indicates a stronger intermediary role of the stakeholder in urban environment governance, reflecting greater control over decisions and operations within the network. Higher closeness centrality means that a node receives core information faster and over shorter distances, indicating effective internal governance.

Our calculations showed that most nodes had a closeness centrality close to 1, indicating tight connections among nodes in large-sample analysis. Different variance values reflect the activity level of stakeholders in the governance network. Moreover, the network centralization index increased from 1.15% to 5.77% (Table 4), indicating a continuous rise in network centrality and the strengthening dominance of core nodes. This suggests that the importance placed on urban environment governance by various sectors of society is increasing. However, the current distribution remains relatively dispersed, and the public nature of environmental good dictates the need for comprehensive government guidance and management.

4. Discussion

In China, there is a strong expectation for the government to fulfill its functional responsibilities, and most relevant land is not privately owned. As a public good, the governance of ecological environments usually falls largely under government management. Other stakeholders such as businesses, social organizations, and community residents engage in governance through limited time and financial contributions. However, these expenditures and actions may not fully reflect the actual demands of stakeholders. Social network analysis can accurately capture the objective expectations of various stakeholders regarding governance effectiveness and the future development of participation forms.

Combined with the network structure diagrams in Figure 5, the evolutionary trend and policy implementation effectiveness can be observed. The analysis of policy texts collected from different periods reveals that policy directions and target groups vary across different development stages. By constructing keyword co-occurrence matrices and drawing co-occurrence networks for policy texts from different stages, we can observe changes in the strength of relationships among nodes, as well as their positions and roles within the network.

4.1. Development of Urban Environment Governance Networks

Urban environment governance is a multifaceted and complex issue involving various dimensions of urbanization. Different stakeholders have distinct needs regarding industrial development, transportation planning, resource utilization, pollution control, policy constraints, and awareness cultivation. Regional collaboration is crucial in managing public environmental goods. Therefore, urban environment governance should consider the needs of all parties, encouraging cooperation among governments, businesses, social organizations, and the public to ensure a stable and rational governance network. This collaboration aims to achieve sustainable development in urban environment governance, enhancing the efficiency of public resource utilization and the effectiveness of ecological civilization development.

4.1.1. Governance Network Construction and Initial Exploration (2017–2019)

As China’s urbanization steadily progresses and public environmental awareness continues to rise, the Chinese government has increasingly emphasized urban environment governance issues alongside economic development. Since the 18th National Congress of the Communist Party of China incorporated ecological civilization construction into national strategic development, building a beautiful China has become a vision for the entire society, with public awareness and participation in governance gradually strengthening. The formal implementation of the “Environmental Protection Tax Law of the People’s Republic of China” in 2018 marked an increasing policy constraint on enterprises, enriching governance methods through market-based adjustments to demand.

Network structure and parameter calculations reveal that during this period, the development of ecology and environment was a primary focus for all parties. A substantial proportion of information transmission concentrated on this issue. Government departments, as the dominant role in the network, held a core functional position, whereas departments had limited specific functions due to the highly specialized nature of environmental governance and the differentiated responsibilities among departments.

Additionally, enterprises, the main forces in the resource and environmental industries, have taken on significant responsibilities in environmental governance, integrating it into their daily operations under government guidance and constraints. With increasing national emphasis on environmental issues, universities and research institutes have become key organizations providing scientific decision-making and technical support. However, NGOs had a relatively minor role in the network during this period, with limited attention to related issues. Public and community awareness of environmental problems has started to rise due to earlier urban environmental issues, but active participation remains insufficient, resulting in decentralized and marginalized positions within the network. Overall, during this period, the centralization trend of China’s urban environment governance network was low. However, this does not mean low centralization equates to more balanced stakeholder interests. The data sources in this study primarily came from government documents, which have a guiding and mandatory influence on environmental governance issues. The lower centralization mainly stems from the government’s initial understanding of the roles of other stakeholders in governance participation. Therefore, the overall network was still in an initial and constantly improving construction stage during this period.

4.1.2. Governance Network Development and Mid-Term Interaction (2021–2023)

As the largest developing country, China has placed significant emphasis on urban development and environmental governance. Even in 2020, the year most affected by the pandemic, the Chinese government made a global commitment to the “dual carbon goals,” earning widespread national and international recognition. During this period, the country introduced important policy documents such as the “14th Five-Year Plan for Ecological Environmental Protection” and the “Action Plan for Carbon Peaking Before 2030,” further clarifying governance paths and targets.

The network structure and parameter measurements indicated that the government remained the dominant entity in the network during this period. However, the network positions of businesses, social organizations, and community public stakeholders were strengthened. The economic recovery post-pandemic increased businesses’ development needs, combined with new ecological constraints, enhancing the central role of enterprises in the network. Social organizations significantly increased their involvement, mainly through scientific research, policy analysis, and social feedback. Public communities showed a shift towards organized, collective participation. This was partly due to ongoing pandemic control measures in 2021 and the pressing need to improve communication efficiency and reduce governance costs for environmental issues as public concerns.

Finally, the significant increase in network centralization is primarily due to the clear urban environment governance goals under the dual-carbon development requirements during this stage. Data calculations show that the overall network focuses on the coordination of pollution control and urban development, addressing environmental issues such as climate, air, traffic, and ecology in cities, which are fundamentally governance and development issues under the dual carbon goals. Combining the changes in network centralization indices with the actual situation of environmental governance indicated that the governance network is still in the stage of improvement and development. The variation in goals influences stakeholder participation, achieving a moderate level of centralization rather than pursuing low centralization. Excessively low centralization means the network lacks core nodes, and the removal of any node or stakeholder would not affect the operation of the governance network. However, this does not align with China’s status as a developing country, nor does it meet the practical needs of urban environment governance as a public good. The governance networks are primarily driven by governmental and corporate entities, while experiencing limited engagement from the public and social organizations, may face long-term challenges in legitimacy, effectiveness, and sustainability. Concurrently, public participation remains constrained by economic contexts, educational structures, and social norms. This study posits that urban environmental governance frameworks should be calibrated according to national and regional developmental stages through stakeholder resource allocation. Furthermore, analytical tools must delineate ecological governance relationships to co-create inclusive frameworks, thereby refining decision-making systems and relational pathways to achieve effective institutional equilibrium [26].

By establishing a governance framework, selecting indicators, constructing networks, and calculating indices, we innovatively utilized large-scale policy text analysis to explore the evolution of urban environment governance networks across different periods. The related results provided good validation and explanation, avoiding issues of high research costs, limited survey groups, and insufficient representativeness and typicality in environmental problem studies under the lens of network governance.

However, this study still has certain limitations and areas for improvement. First, the relationship composition patterns in social networks may be influenced by regional attribute differences. Exploring these influences from different perspectives can reveal their impact differences. Although policy texts are representative of governance, they fail to reflect the structural differences between governance networks from various perspectives effectively. Second, significant economic and social attribute differences between regions may lead to varying policy execution and practice intensities. The impact of attribute variable differences between regions with different resource endowments on the effectiveness of governance network operations was not addressed in this study. These issues were worth exploring in future research.

4.2. Hypothesis Testing

In our study, we proposed two hypotheses: first, the central role of the government in the urban environment governance network; second, the increasing prominence of other stakeholders such as businesses over time, enhancing overall network stability and centralization. The results for the relevant indicators showed that the standardized indices for the government dimension (Gov; Dep; Cen; Pol) generally declined, yet the standardized out-degree, except for the Cen indicator, remained above 0.4. The difference between out-degree and in-degree was a large positive value, indicating a dominant position in the network. This demonstrates that following the establishment of the fundamental national policy for ecological civilization reform in 2015 and the formal establishment of the first national ecological civilization experimental zones in 2016, policy constraints have been strengthening, with increasingly clear planning goals for collaborative governance among various stakeholders.

However, due to the lag in policy discussion, introduction, and implementation, the government played an absolute dominant role in the early stages of urban governance networks, crucial in goal setting, policy constraints, project guidance, and social cultivation. Hence, the relevant indices were higher in the first period than in the second. Nevertheless, the network centralization level increased from 1.15% to 5.77% between the two periods, reflecting the stabilization of the initially dispersed governance network structure due to policy advancement and effectiveness. Other stakeholders’ roles also strengthened, improving their positions in the network. This continued emphasis on collaborative governance among all societal sectors in urban environment governance is increasingly apparent. Moreover, the increase in network density from 0.096 to 0.101 indicates that the ability and means of different stakeholders to obtain critical information in the network have improved, contributing to network stability and goal alignment. Thus, the first hypothesis is well supported.

The second hypothesis was also well validated, with the network centralization index significantly increasing from 1.15% to 5.77%. The results from the network measurement indices indicate that while the government and enterprises retain the dominant roles in the governance network, the positions of social organizations and community public have significantly improved, with their proportion in most indices increasing. Additionally, the network structure indices show that the urban environment governance goals of different stakeholders have become clearer and the collaborative concepts in the governance process have been solidified. For example, the betweenness centrality of Pol. and Gove. significantly increased, while the betweenness centrality of Peop., Pub., and UEn. decreased considerably. This suggests that post-pandemic, the general public has become more reliant on the government for managing public environmental goods, with a declining willingness to participate actively. The economic downturn has also led the government to incorporate urban environmental issues into core development agendas, placing greater emphasis on pollution problems during development.

5. Conclusions

China’s rapid economic growth and urbanization have generated significant environmental governance challenges. In response, the central government has consistently strengthened environmental protections and assumed global leadership on climate change. The welfare losses and impacts stemming from environmental degradation have elevated urban environmental governance as a critical priority across all sectors. Effective governance improvements will enhance environmental quality, stimulate economic development, and foster green industrial systems. For local governments and officials, ecological civilization represents both a political imperative and a viable sustainable development pathway, with the central question being whether the Chinese government plays a decisive role in environmental governance [27].

This study substantiates these focal concerns through granular network metrics and extensive textual evidence. The evolving structural trends necessitate coordinated alignment of governmental, market, and societal interests to enhance inclusivity and resilience within China’s urban environmental governance frameworks. Concrete implementation pathways warranting exploration encompass institutionalized citizen participation mechanisms, capacity-building initiatives for social organizations, and multisectoral coordination guided by ecosystem service assessments and ecological asset valuation techniques.

The results indicate that after years of construction and development, China’s urban environment governance network has formed a structure led by the government, with businesses as key players and social organizations providing effective support. Public awareness of participation in governance is still in the cultivation stage, often expressed through organized forms within the network, but individual willingness to participate remains somewhat lacking. As the governance network continues to develop, the roles and positions of business stakeholders have become increasingly prominent, undertaking more rights and responsibilities, aligning with government expectations of achieving efficient allocation of public environmental resources through market mechanism.

However, it is worth noting that while the roles of other stakeholders are on the rise, both social organizations and public communities still act as passive recipients. They primarily follow the directives of leading entities, and their awareness and willingness to actively participate in urban environment governance remain insufficient. This phenomenon raises questions about whether it might affect the successful achievement of the government’s environmental governance goals, which aim to build a collaborative governance system involving the government, businesses, social organizations, and the public. Future research could explore these related issues.

Natural resources and ecological environment governance networks have consistently been a focal point for researchers. This study comprehensively reviewed policy texts formulated and implemented over two periods, combining large-scale text analysis with social network analysis to explore and identify the operational modes and areas for improvement in urban environment governance networks in China. The related hypotheses were thoroughly validated, enriching research methods in this field. These results will assist researchers in effectively substituting research objects, samples, and tools in related studies, thereby enhancing the representativeness and generalizability of research outcomes. Moreover, this study extended the application of social network analysis tools by integrating machine-based selection and expert ratings in the indicator selection process, better aligning indicators and dimensions and laying a solid foundation for subsequent research.

Finally, this analysis relied solely on policy texts, constituting a unidirectional network methodology that inherently constrains the assessment of stakeholder relationship depth and intensity. Consequently, future research must integrate complementary datasets such as corporate disclosures from listed enterprises. Such multi-source evidence would enable comparative and hybrid analytical strategies capable of overcoming the limitations intrinsic to large-scale unidirectional network paradigms. Future research directions include identifying commonalities and individual characteristics of environmental governance networks from different stakeholder perspectives, examining the impact of major environmental policies’ exogenous shocks on governance networks, and analyzing regional attribute differences based on varying resource endowments.