Spatiotemporal Dynamics and Structural Drivers of Urban Inclusive Green Development in Coastal China

Abstract

In China’s rapidly urbanizing coastal areas, inclusive green development (IGD) has become an important way to achieve a reduction in economic development disparities, environmental sustainability, and social equity. This study investigates the spatiotemporal dynamics and structural drivers of IGD across 54 coastal cities within three marine economic zones (MEZs) using a hybrid analytical framework that integrates evaluation techniques, inequality decomposition, spatial factor detection, and spatial econometrics. The result shows that a distinctive “four-pillar” spatial structure has emerged, centered on the Shandong Peninsula, Yangtze River Delta (YRD), West Coast of the Taiwan Strait, and Pearl River Delta (PRD). Spatial autocorrelation has intensified since 2020, indicating the cumulative effect of China’s post-2020 regional integration policies and digital infrastructure investments, which accelerated resource flows between cities. Spatial econometric analysis further reveals that economic development and equitable public service provision are the most influential drivers, while public investment in R&D and digital transformation exhibit significant cross-city spillover effects. The findings highlight the importance of regionally adaptive and digitally integrated strategies to promote inclusive and sustainable urban development in coastal economies. Therefore, efforts should be intensified to strengthen the role of core cities as diffusion engines for neighboring areas, with a strategic focus on regional digital transformation and R&D investment, to advance inclusive and sustainable development in coastal economies.

1. Introduction

Over the past four decades, China’s rapid industrialization and urbanization have fueled remarkable economic growth and significantly improved living standards nationwide. However, these developments have also led to severe environmental degradation, rising carbon emissions, and widening socioeconomic disparities, particularly in rapidly expanding urban and coastal areas [1,2]. In response, the Chinese government has gradually reoriented its national development agenda to prioritize ecological sustainability, social equity, and quality-oriented growth. Since the 18th National Congress of the Communist Party of China, the concept of an “ecological civilization” has been elevated to a central policy priority, institutionalized through national strategic plans such as the 14th Five-Year Plan (2021–2025) and the Dual Carbon Goals framework. These documents outline concrete goals for green transformation, common prosperity, and cross-sectoral governance mechanisms aimed at ensuring more inclusive and sustainable urban futures [3].

In this context, inclusive green development (IGD), a paradigm that integrates environmental sustainability, social inclusion, and high-quality economic development, has emerged as a key objective in China’s coastal urban regions [4]. As global-facing hubs and growth engines, coastal cities face unique challenges: managing land and resource constraints, adapting to climate risks, and meeting rising public expectations for equitable services and livable environments [5]. Analyzing the spatial and temporal dynamics of IGD in these cities provides helpful information regarding how policy interventions, institutional capabilities, and localized drivers interact across regions. Moreover, such research not only helps identify internal development imbalances within the coastal belt but also provides critical lessons for inland cities seeking to achieve more equitable and sustainable growth trajectories [6].

Current research on IGD can be broadly categorized into three major streams: theoretical foundations, spatial–temporal dynamics, and driving factors. Inclusive green development has emerged as a research hotspot in the field of sustainable development recently. It integrates the focus of the green development of the ecological environment and the pursuit of social equity in inclusive development, emphasizing the synergistic progress of economic growth, social inclusion, and ecological protection. Theoretically, it is rooted in the theories of sustainable development, inclusive growth, and environmental justice. The theory of sustainable development provides it with the underlying logic of “intergenerational equity” and “ecological carrying capacity.” The theory of inclusive growth highlights the attention to vulnerable groups and regional balance in the development process. The theory of environmental justice further explains the fairness of different groups in environmental resource allocation and environmental risk bearing. In the evolution of research, early scholars mostly focused on a single dimension of green development or inclusive development. Many studies have built different ways to analyze these issues, often based on the United Nations Sustainable Development Goals (SDGs). They focus on how protecting the environment, promoting social inclusion, and supporting economic growth are all connected [7]. Some studies broaden this perspective by incorporating the concepts of intergenerational equity and spatial justice into IGD assessments [8]. Some contend that IGD should be perceived as a consequence of economic co-benefits arising from synergistic interactions among economic, social, and ecological subsystems [9,10]. With regard to performance evaluation, two main approaches dominate the literature: index-based level measurement and efficiency analysis. The former typically employs composite indicators across three dimensions—economic vitality, environmental quality, and social equity—to analyze spatiotemporal patterns at national, regional, and sectoral levels [1,4,10]. In contrast, the latter adopts production theoretic frameworks such as the Super Epsilon-Based Measure (Super-EBM) model, incorporating undesirable outputs into assessments of IGD [2,11,12]. Recent research has started using models that look at how things change over time and how they spread between regions, like in the work by Ren and others [13].

The second area of research focuses on spatial–temporal dynamic analysis. The spatial–temporal dynamic analysis of regional development is the key to understanding the heterogeneity of inclusive green development. In terms of research methods, tools such as spatial econometric models, panel data analysis, and geographic information systems are widely used. From a national perspective, there are studies on the relationship between digital transformation and innovation efficiency against the background of “dual carbon” and explorations of the spatial spillover effect of green financial policies [14,15]. These studies reveal the interaction mechanism between macro policies and regional development. However, as the most economically active and open region in China, the uniqueness of coastal cities has not been fully explored. At the city scale, some studies focus on the green development dynamics of a single city or urban agglomeration. For example, Huang analyzed the mitigation effect of the digital economy on urban carbon emissions in China [16]. Although not directly targeting coastal cities, it provides a methodological reference for the study of green development mechanisms at the city level. However, research on the spatial–temporal dynamics of inclusive green development specifically for coastal cities in China is still relatively scarce. Existing achievements are mostly scattered in subtopics, such as the regional economy and the environmental economy, lacking systematic integration and in-depth analysis of the characteristics of coastal cities.

The third area of research mainly looked at what causes IGD and how it works. Driving factors are central to explaining the differences in inclusive green development. Existing research mainly focuses on dimensions such as economic structure, policy system, technological innovation, and social participation. The interaction between the digital economy and green innovation is an important direction. Chen studied the relationship between the digital economy, green innovation, and high-quality economic development [17]. Xu explored the relationship between the digital economy, green innovation efficiency, and new quality productive forces based on provincial panel data [18]. These studies indicate that the digital economy promotes green development through paths such as technological empowerment and industrial integration. However, research on the coupling mechanism between the industrial structure of coastal cities and inclusive green development is still insufficient. In terms of policy systems, Xu and Liu verified the promoting effect of green financial policies from the perspectives of inclusive green growth and corporate carbon emissions, respectively [18,19]. Dai took G7 economies as a sample to analyze the enabling effect of green innovation on sustainable development [20]. Some studies also focus on the impact of R&D investment and energy efficiency on carbon emissions [21]. Also, empirical findings suggest that government infrastructure investment, environmental taxation, and public service accessibility are positively associated with IGD, while foreign direct investment may inhibit inclusive green outcomes under weak regulatory environments [22,23,24]. Institutional quality, particularly anti-corruption efforts, administrative efficiency, and transparent governance, has been identified as a consistent enabler of IGD, especially in emerging economies [25,26]. An expanding range of studies focuses on the effects of digital technologies, covering digital infrastructure, smart governance, and financial inclusion. For instance, recent studies find that digital connectivity and e-government applications improve the delivery of social services and environmental management, thereby enhancing urban IGD [27,28]. Digital financial inclusion really helps boost green innovation and pushes the economy to upgrade in more sustainable ways. It offers two strong paths for more comprehensive growth [29,30].

Despite increasing scholarly attention, three critical gaps remain in the literature on IGD. Most studies tend to focus on bigger areas like entire countries or provinces, but they often overlook differences within cities. This is especially true in coastal regions, where local differences can be quite major. Intra-regional disparities primarily stem from heterogeneous resource endowments and uneven policy implementation speeds, as evidenced by the southern MEZ’s intra-zone Gini (0.28) surpassing its inter-zone value (0.21), revealing stark contrasts between PRD core cities (e.g., Guangzhou) and peripheral areas (e.g., Maoming). Kernel density plots further validate this polarization, with bimodal distributions in the southern MEZ confirming a dualistic structure of high and low-level clusters. These findings align with fiscal decentralization theories, underscoring the spatial inequality driven by uneven resource allocation and policy diffusion. Secondly, indicator systems are often restricted to macro-level dimensions of economy, society, and environment, which limits their ability to capture the multi-layered and structural attributes of IGD. Thirdly, methodological approaches have largely relied on descriptive evaluation techniques, with insufficient exploration of spatial heterogeneity, temporal dynamics, and underlying drivers.

To address these limitations, this study contributes in three ways. We focus explicitly on coastal cities that are China’s strategic growth frontiers and key ecological corridors to assess IGD disparities and their policy relevance for inland regions. We refine and expand the evaluation system by integrating indicators of economic growth potential, social welfare sharing, and green production practices, thereby providing a more comprehensive measure of IGD. Methodologically, we adopt a hybrid framework that combines vertical–horizontal scatter degree analysis, kernel density estimation (KDE), Dagum’s Gini decomposition, exploratory spatiotemporal data analysis, GeoDetector, and the Spatial Durbin Model (SDM) to jointly analyze IGD’s spatiotemporal patterns and driving mechanisms. By using this comprehensive approach, we can identify the causes of regional differences and understand how to share resources or provide support. It helps us devise real-world ideas for how to promote fair and more balanced growth in China’s coastal areas. At the same time, coastal IGD exhibits a high sensitivity to marine–terrestrial interactions, necessitating the analysis of spatial clustering and temporal evolution to identify policy intervention windows.

2. Theoretical Analysis and Research Hypotheses

Based on the theoretical frameworks of land–sea coordination and institutional lock-in effects, the spatial differentiation of inclusive green development (IGD) in coastal cities is fundamentally rooted in the interplay of regional resource endowments, policy gradients, and historical trajectories. The development of marine economic corridors relies heavily on port–hinterland synergy, while China’s three major Marine Economic Zones (MEZs) exhibit significant institutional disparities: The eastern MEZ leverages world-class port clusters to establish industry–innovation synergy hubs. The northern MEZ suffers from efficiency constraints due to Bohai Bay’s icy period. The southern MEZ faces a scarcity of deep-water port infrastructure.

Concurrently, national policy pilots prioritize implementation in the eastern MEZ, generating institutional first-mover advantages. In contrast, the northern MEZ struggles with path dependency in traditional heavy industries, while the southern MEZ experiences diminished policy effectiveness due to cross-provincial administrative fragmentation. Furthermore, transformational inertia persists in the northern MEZ from legacy industries, and the southern MEZ’s fragmented administrative divisions suppress economies of scale.

Thus, the Hypothesis I is proposed: the compound heterogeneity of institution and geography will lead to the hierarchical structure of “core–edge” in IGD. The eastern MEZ forms the development pole core by virtue of the triple synergy of policy–geography–industry, while the northern and southern MEZs show gradient attenuation due to endowment constraints and institutional lag.

Based on the policy-driven theory and the evolution law of the Environmental Kuznets Curve (EKC), the green transformation of coastal cities in China will show a non-linear acceleration feature under policy intervention. National strategies such as the “dual carbon target” reconstruct the development priority of coastal areas through mandatory constraints and incentive mechanisms. Firstly, the top-level design of carbon neutrality forces the upgrading of high-emission industries. For example, the steel and petrochemical industries in the northern MEZ are facing the pressure of clean technology substitution, which stimulates the demand for green innovation. Secondly, there is a critical juncture in green investment. When per capita GDP crosses the threshold, the marginal benefits of environmental governance investment will be significantly improved, prompting green elements to jump from “auxiliary indicators” to “core drivers.” Finally, marine ecology has certain particularities. Coastal cities need to simultaneously address land-based pollution and marine ecological degradation, which strengthens the irreplaceability of green governance and promotes the weight of ecological indicators in the IGD system.

Thus, Hypothesis II is proposed: the green ecological dimension will break through the linear evolution path, and the explanatory power of IGD will achieve leap-forward growth from the marginal supporting elements to the core driving force of the system.

Based on the theory of spatial interaction and the principle of environmental externality, the digital process and pollution control practice of coastal cities have the natural attribute of cross-regional transmission. The openness of the marine economic system determines that its environmental governance and technological innovation must break through administrative boundaries. Digital infrastructure such as coastal smart ports and the marine Internet of Things form an interconnected technical network so that innovative management models can spread to surrounding cities through industrial chain association and technical imitation. Through collaborative governance, the negative externalities of pollution flow are internalized, giving rise to the cross-regional cooperation paradigm of “joint monitoring–cost sharing.” In addition, the national marine ecological compensation pilot promotes the institutionalization of the digital pollution control experience through the fiscal transfer payment mechanism, forming a gradient spillover path of “innovation in core cities–adaptation in hinterland cities.” This path is not only reflected in the direct technological diffusion in geographically adjacent areas but also enables cross-scale environmental governance efficiency through the institutional coordination of marine economic zones, providing a new path to break the “environmental trap of administrative regions.”

Therefore, Hypothesis III is proposed: the interaction between digitalization and pollution control has a significant positive spatial spillover effect.

3. Materials and Methods

3.1. Study Area

The focus is on China’s coastal cities, encompassing 54 municipalities (Figure 1). These cities are administered by 11 coastal provinces, autonomous regions, and municipalities directly under the central government: Liaoning, Hebei, Tianjin, Shandong, Jiangsu, Shanghai, Zhejiang, Fujian, Guangdong, Guangxi, and Hainan. According to the 14th Five-Year Plan for the Development of the Marine Economy, these cities are grouped into three major marine economic zones (MEZs). The northern MEZ consists of 17 cities in Liaoning, Hebei, Tianjin, and Shandong. The eastern MEZ comprises 11 cities in Jiangsu, Shanghai, and Zhejiang. The southern MEZ includes 26 cities in Fujian, Guangdong, Guangxi, and Hainan. The three major marine economic zones contribute approximately 85% of the country’s marine gross domestic product. Taking Shandong as an example: its “14th Five-Year Plan” for the marine economy proposes building “world-class marine ports and a modern marine industrial system,” and by 2025, the comprehensive competitiveness of its marine economy needs to enter the medium and high end of the global value chain. The annual container throughput of the Shanghai and Ningbo-Zhoushan port clusters in the Yangtze River Delta accounts for nearly 40% of the national total, making them the core hubs for global marine shipping and trade. These regions form the core areas of China’s marine economy and play a pivotal role in both national development strategies and international cooperation.

3.2. Data Sources

The research data are derived from the “2013–2023 China Urban Statistical Yearbook,” the “China Urban Construction Statistical Yearbook,” statistical yearbooks of 11 coastal provinces (autonomous regions and municipalities directly under the Central Government), as well as public statistical data released by the natural resources (marine) and ecological environment administrative departments of each coastal city. For some annual data that are missing, this study uses the linear interpolation method to complete the data. The data on R&D investment and expenditure for some years are missing. The average of the data from the previous year and the following year is used to replace the missing data. These interpolation data account for approximately one percent in total, and their impact on the potential results is negligible.

3.3. Evaluation Indicator System

The concept of IGD has been most widely framed by the United Nations and international organizations. The meaning of IGD is economic growth that improves human well-being and social equity while reducing environmental risks and ecological degradation [31,32]. Building on this definition, scholars generally converge on a tripartite framework of economic development, social equity, and environmental sustainability as the three fundamental pillars of IGD [33,34]. The design of the indicator system in this study is based on the three-dimensional framework of the United Nations Sustainable Development Goals (SDGs) and integrates the theory of spatial justice, the theory of green growth, the theory of capacity development, and the theory of innovation diffusion. It covers the cross-regional fairness of resource allocation, the relationship between ecological efficiency and economic development, the development capacity of people, and how technology spillover drives regional inclusive development. The selection of indicators takes into account operability, policy relevance, spatial interactivity, and frontiers so as to build a scientific indicator system from theoretical drive to data verification.

Based on this, this study establishes a three-dimensional indicator system, comprising economic development (E), social inclusion (S), and green ecology (G). Each dimension is further subdivided into three sub-dimensions. Specifically, economic development includes development level, development impetus, and growth potential; social inclusion covers achievement sharing, equal opportunity, and public services; while green ecology addresses resource endowment, green production, and ecological governance. Each sub-dimension has been assigned corresponding measurable indicators. The system consists of 9 indicator layers, and 24 specific indicators have been established (

Table 1. Evaluation indicator system for IGD.

Dimension	Primary Indicator	Secondary Indicator	Expression	Basis for Index Selection
Economic Development (E)	Development Level (E1)	E11 GDP per capita	GDP/Permanent residents	GDP per capita represents an individual’s development level, public budget revenue per capita reflects the fiscal situation of the region, and advanced industrial structure reflects the development level at the industrial level.
		E12 Public budget revenue per capita	General public budget revenue/Permanent residents
		E13 Advanced industrial structure	Industrial structure level index
	Development Momentum (E2)	E21 Total factor productivity	DEA–Malmquist’s method	High total factor productivity indicates high production efficiency, high social labor productivity indicates high efficiency, and the higher the fixed asset investment growth rate, the stronger the economic vitality. High E21, E22, and E23 all represent a good development momentum.
		E22 Social labor productivity	GDP/Number of employed persons
		E23 Fixed asset investment growth rate	Fixed asset investment growth over previous year
	Growth Potential (E3)	E31 Market potential	Ishengoma and Shao, 2025 [28]	The market potential and S & Technology expenditure proportion reflect the development potential of this region from both economic and technological aspects.
		E32 S & Technology expenditure proportion	Public finance science and technology expenditure/GDP
Social Inclusion (S)	Shared Results (S1)	S11 Average years of education	Xue and Zhang, 2022 [29]	The increase in the average years of education, the decline in the Engel coefficient, and the rise in the social insurance participation rate will all promote the development of the sharing level towards a more balanced and inclusive direction.
		S12 Engel’s coefficient	Engel coefficient
		S13 Social insurance participation rate	Medical insurance, social insurance participation rate
	Equal Opportunity (S2)	S21 Theil’s index	Chen and Zhang, 2024 [30]	The reduction in the registered urban unemployment rate, the weakening of the Theil index, and the increase in the urban road area per capita will help create fairer and more inclusive development conditions for equal opportunities.
		S22 Urban registered unemployment rate	Urban registered unemployment rate
		S23 Urban road area per capita	Urban road area/Permanent residents
	Public Services (S3)	S31 Hospital beds per 1000 people	Hospital beds/Permanent residents × 1000	An increase in hospital beds per 1000 people, a rise in the number of books in public libraries per 100 people, and an optimization of the student–teacher ratio in regular primary and secondary schools can all enhance the quality and accessibility of public service supply.
		S32 Public library collections per 100 people	Public library collections/Permanent residents × 100
		S33 Student–Teacher Ratio in Regular Primary and Secondary Schools	Number of Students Enrolled in Primary and Secondary Schools/Number of Full-time Teachers
Green Ecology (G)	Resource Endowment (G1)	G11 Park Green Space Area Per Capita	Park Green Space Area/Permanent Resident Population	The expansion of park green space area per capita, the increase in the green coverage rate in built-up areas, and the sufficiency of water resources per capita will enhance the ecological support capacity and sustainability of regional resource endowments.
		G12 Green Coverage Rate in Built-up Areas	Green Coverage Rate in Built-up Areas
		G13 Water Resources Per Capita	Total Water Resources/Permanent Resident Population
	Green Production (G2)	G21 Energy Consumption Per Unit of GDP	Total Energy Consumption/GDP	The reduction in energy consumption per unit of GDP and the decrease in carbon emissions per unit of GDP are the core manifestations and key indicators of the improvement in the level of green production.
		G22 Carbon Emissions Per Unit of GDP	Total Carbon Emissions/GDP
	Ecological Governance (G3)	G31 Intensity of Pollution Control Investment	Environmental Protection Expenditure/General Public Budget Expenditure	The increased intensity of pollution control investment and the improvement of the pollution control index are important supports and direct reflections of the enhanced effectiveness of ecological governance.
		G32 Pollution Control Index	Escap, 2013 [31]

).

3.4. Methodology

3.4.1. Vertical and Horizontal Scatter Degree Method

To better reveal the dynamic temporal changes in IGD in coastal cities across multiple periods, this study adopts the vertical and horizontal scatter degree method to measure the constructed indicator system. The method overcomes the shortcomings of the traditional cross-sectional data evaluation method regarding multi-period dynamics. This method relies entirely on the data itself to conduct objective horizontal measurement and ranking. The steps are as follows:

(1)

Set up the comprehensive evaluation function

$${\mathrm{y}}_{\mathrm{i}}=({\mathrm{t}}_{\mathrm{k}})=\sum _{\mathrm{j}=1}^{\mathrm{m}}{\mathrm{w}}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{j}}\left({\mathrm{t}}_{\mathrm{k}}\right) (\mathrm{i}=1,2,\cdots ,\mathrm{n};\mathrm{j}=1,2,\cdots ,\mathrm{m};\mathrm{k}=1,2,\cdots ,\mathrm{T})$$

(1)

${\mathrm{x}}_{\mathrm{i}\mathrm{j}}\left({\mathrm{t}}_{\mathrm{k}}\right)$ represents the value of the j indicator for the i evaluation object at time ${\mathrm{t}}_{\mathrm{k}}$, ${\mathrm{w}}_{\mathrm{j}}$ represents the weight of the $\mathrm{j}$ indicator, and ${\mathrm{y}}_{\mathrm{i}}\left({\mathrm{t}}_{\mathrm{k}}\right)$ represents the comprehensive score of the i evaluation object at time ${\mathrm{t}}_{\mathrm{k}}$.

(2)

Determine the indicator weights

Let the sum of squared deviations ${\mathsf{\sigma }}^2={\sum }_{\mathrm{k}=1}^{\mathrm{T}}{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\left|{\mathrm{y}}_{\mathrm{t}}\right({\mathrm{t}}_{\mathrm{k}})-\overline{\mathrm{y}}|}^2$ of ${\mathrm{y}}_{\mathrm{i}}\left({\mathrm{t}}_{\mathrm{k}}\right)$ reach its maximum value. After dimensionless processing of the original data, the formula is as follows:

$$\overline{\mathrm{y}}={\displaystyle \frac{1}{\mathrm{T}}}\sum _{\mathrm{k}=1}^{\mathrm{T}}[{\displaystyle \frac{1}{\mathrm{n}}}\sum _{\mathrm{i}=1}^{\mathrm{n}}\sum _{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{w}}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{i}\mathrm{j}}\left({\mathrm{t}}_{\mathrm{k}}\right)]=0$$

(2)

Therefore, Equation (3) holds as follows:

$${\mathsf{\sigma }}^2=\sum _{\mathrm{k}=1}^{\mathrm{T}}\sum _{\mathrm{i}=1}^{\mathrm{n}}{{[\mathrm{y}}_{\mathrm{i}}\left({\mathrm{t}}_{\mathrm{k}}\right)]}^2=\sum _{\mathrm{k}=1}^{\mathrm{T}}\left({\mathrm{W}}^{\mathrm{T}}{\mathrm{H}}_{\mathrm{k}}\mathrm{W}\right)={\mathrm{W}}^{\mathrm{T}}\sum _{\mathrm{k}=1}^{\mathrm{T}}{\mathrm{H}}_{\mathrm{k}}\mathrm{W}={\mathrm{W}}^{\mathrm{T}}\mathrm{H}\mathrm{W}$$

(3)

where $\mathrm{W}={({\mathrm{w}}_1,{\mathrm{w}}_2,\cdots ,{\mathrm{w}}_{\mathrm{m}})}^{\mathrm{T}}$, $\mathrm{H}={\sum }_{\mathrm{k}=1}^{\mathrm{T}}{\mathrm{H}}_{\mathrm{k}}$ is a m × n symmetric matrix, and ${\mathrm{H}}_{\mathrm{k}}={\mathrm{X}}_{\mathrm{k}}^{\mathrm{T}}{\mathrm{X}}_{\mathrm{k}}(\mathrm{k}=1.2,\cdots ,\mathrm{T})$.

$${\mathrm{X}}_{\mathrm{k}}=\left[\begin{array}{ccc}{\mathrm{x}}_{11}\left({\mathrm{t}}_{\mathrm{k}}\right)& \cdots & {\mathrm{x}}_{1\mathrm{m}}\left({\mathrm{t}}_{\mathrm{k}}\right)\\ ⋮& \ddots & ⋮\\ {\mathrm{x}}_{\mathrm{n}1}\left({\mathrm{t}}_{\mathrm{k}}\right)& \cdots & {\mathrm{x}}_{\mathrm{n}\mathrm{m}}\left({\mathrm{t}}_{\mathrm{k}}\right)\end{array}\right]$$

(4)

If we restrict $\mathrm{W}{\mathrm{W}}^{\mathrm{T}}=1$, then the eigenvector W corresponding to the maximum eigenvalue of matrix H is the weight coefficient, at which point ${\mathsf{\sigma }}^2$ reaches its maximum value.

Finally, the normalized eigenvector becomes the indicator weights. Combining the dimensionless indicator data and their weights, the comprehensive index of IGD for city i in year k can be obtained.

3.4.2. Kernel Density Estimation

Kernel density estimation (KDE) is a non-parametric estimation method for analyzing spatial distribution inequality. It comprehensively reveals the dynamic evolution patterns by examining characteristics such as the location, shape, and extensibility of the distribution curve of the research object. The IGD level, calculated via the longitudinal and cross-sectional-level analysis method, is treated as a random variable in this study. Kernel density function plots are employed to visually illustrate the evolution trends of IGD and its dimensions. The formula is as follows:

$$\mathrm{f}\left(\mathrm{x}\right)={\displaystyle \frac{1}{\mathrm{n}\mathrm{h}}}\sum _{\mathrm{i}=1}^{\mathrm{n}}\mathrm{K}\left({\displaystyle \frac{{\mathrm{x}}_{\mathrm{i}}-\mathrm{x}}{\mathrm{h}}}\right)$$

(5)

$$\mathrm{K}(\mathrm{t})={\displaystyle \frac{1}{\sqrt{2\mathsf{\pi }}}}{\mathrm{e}}^{{\displaystyle \frac{{\mathrm{t}}^2}{2}}}$$

(6)

where $\mathrm{f}\left(\mathrm{x}\right)$ is the density function of random variable $\mathrm{x}$, $\mathrm{n}$ is the total number of samples, ${\mathrm{x}}_{\mathrm{i}}$ is the sample observation value, h is the bandwidth, $\mathrm{K}\left(\mathrm{t}\right)$ is the kernel function, which is a weighting function or smoothing transformation function. This study adopts the Gaussian kernel function.

3.4.3. Dagum’s Gini Coefficient

The Dagum Gini coefficient is an improvement on the traditional Gini coefficient. It cannot only measure the overall difference Gini coefficient and decompose it into intra-regional differences (Gw), inter-regional differences (Gb), and transvariation intensity (Gt), but also dynamically examine changes in the contribution rates of these three types of differences. The Dagum Gini coefficient and its decomposition method are used to measure the spatial disparities in IGD levels across coastal regions. The formula is as follows:

$$\mathrm{f}\left(\mathrm{x}\right)={\displaystyle \frac{1}{\mathrm{n}\mathrm{h}}}\sum _{\mathrm{i}=1}^{\mathrm{n}}\mathrm{K}\left({\displaystyle \frac{{\mathrm{x}}_{\mathrm{i}}-\mathrm{x}}{\mathrm{h}}}\right)$$

(7)

where ${\mathrm{y}}_{\mathrm{j}\mathrm{i}}$(${\mathrm{y}}_{\mathrm{h}\mathrm{r}}$) is the IGD level of a city in region $\mathrm{j}\left(\mathrm{h}\right)$, $\overline{\mathrm{Y}}$ represents the average IGD level of all cities, $\mathrm{n}$ represents the number of coastal cities (54), k represents the number of regions (3), and ${\mathrm{n}}_{\mathrm{j}}$(${\mathrm{n}}_{\mathrm{h}}$) represents the number of cities in region $\mathrm{j}\left(\mathrm{h}\right)$.

3.4.4. Exploratory Spatial Analysis

Exploratory spatial analysis is a collection of spatial data analysis methods and techniques used to reveal the spatial distribution patterns of geographical phenomena. It can effectively explain the dependency and heterogeneity characteristics of geospatial data. To investigate the spatial distribution patterns of IGD, global and local autocorrelation are used to examine the global and local spatial correlation characteristics at the city-level scale. This analysis further reveals the agglomeration and dispersion characteristics of IGD. The most commonly used measurement indicator is Moran’s Index (Moran’s I), and the formula is as follows:

$$\mathrm{M}\mathrm{o}\mathrm{l}\mathrm{a}\mathrm{n}’\mathrm{s} \mathrm{I}={\displaystyle \frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{W}}_{\mathrm{i}\mathrm{j}}({\mathrm{D}}_{\mathrm{i}}-\overline{\mathrm{D}})({\mathrm{D}}_{\mathrm{j}}-\overline{\mathrm{D}})}{{\mathrm{S}}^2{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\mathrm{W}}_{\mathrm{i}\mathrm{j}}}}$$

(8)

where ${\mathrm{S}}^2=1/\mathrm{n}{\sum }_{\mathrm{i}=1}^{\mathrm{n}}({\mathrm{D}}_{\mathrm{i}}-\overline{\mathrm{D}})$ represents the variance of IGD level; ${\mathrm{W}}_{\mathrm{i}\mathrm{j}}$ represents the spatial weight matrix, and an adjacency-based spatial weight is adopted for this matrix. Moran’s I ranges from [−1,1], reflecting the overall spatial correlation characteristics of development levels, with values closer to 1 indicating a stronger positive spatial correlation, and conversely, a stronger negative correlation. Additionally, this study introduces the LISA statistic to study the spatial clustering patterns between cities, with the calculation formula as follows:

$$\mathrm{L}\mathrm{I}\mathrm{S}\mathrm{A}={\displaystyle \frac{{\mathrm{D}}_{\mathrm{i}}-\overline{\mathrm{D}}}{{\mathrm{S}}^2}}\sum _{\mathrm{j}=1}^{\mathrm{n}}\left[{\mathrm{w}}_{\mathrm{i}\mathrm{j}}\right({\mathrm{D}}_{\mathrm{j}}-\overline{\mathrm{D}}\left)\right]$$

(9)

3.4.5. Geographic Detector

The geographic detector is an analytical method used to explore the spatial heterogeneity distribution of geospatial elements and their influencing factors. The single-factor detection of geographic detectors can identify influencing factors and measure the magnitude of their forces, while also testing the stratified heterogeneity of single variables. Two-factor interaction detection can reveal the interaction between multiple influencing factors and the dependent variable. By testing the coupling of the spatial distribution of two variables, it can detect the possible causal relationship between them and has significant advantages in the study of the driving mechanism of complex geographical factors. Geographic detectors are employed to perform single-factor analysis and two-factor interaction analysis on the IGD of the three major Marine Economic Circles (MECs). The aim of this approach is to dissect the spatial stratified heterogeneity of green development in the spatiotemporal context and explore the interaction between various internal factors. Its calculation formula is

$$\mathrm{q}=1-{\displaystyle \frac{{\sum }_{\mathrm{h}=1}^{\mathrm{L}}{\mathrm{N}}_{\mathrm{h}}{\mathsf{\sigma }}_{\mathrm{h}}^2}{\mathrm{N}{\mathsf{\sigma }}^2}}$$

(10)

In the formula, h represents the stratification of the variable Y or the factor X; N represents the number of units in the HTH layer and the entire region; ${\mathsf{\sigma }}_{\mathrm{h}}^2$ and ${\mathsf{\sigma }}^2$ represents the variances of the y-values of the h-th layer and the entire region. The value range of q is [0,1]. The larger the value of q, the stronger the explanatory power of factor X for Y.

Interactive detection aims to identify the interactions among different influencing factors, determine whether their combined effect enhances or weakens the explanatory power for the driving factors of green development, and whether they independently affect green development.

3.4.6. Spatial Durbin’s Model

Tobler put forward the view that although all phenomena are interrelated, the connection between geographically adjacent entities is more pronounced [35]. Based on this basic geographical principle, spatial econometrics combines computer science, operational research, and statistical techniques to analyze regional cross-sectional and panel data. In the study of the spatial spillover effects of urban carbon emissions, if we only focus on the impact of regional resource endowments, economic structures, and development levels while ignoring the spatial correlation between cities and the ecological welfare or economic factors of neighboring cities, the use of traditional panel data methods will lead to biased estimation results. Therefore, it is necessary to comprehensively consider spatial correlation factors, introduce spatial econometric methods to re-examine the spillover effects of urban carbon emissions, and conduct an in-depth exploration of their mechanisms.

Elhorst summarized three classic spatial econometric models: the Spatial Lag Model (SLM), the Spatial Error Model (SEM), and the Spatial Durbin Model (SDM) [36]. Their basic expressions are as follows:

$$y_{i,t}={\alpha }_{i,t}+\rho W{y}_{i,t}+x_{i,t}{\beta }_{i,t}+{\epsilon }_{i,t}$$

(11)

$$y_{i,t}={\alpha }_{i,t}+x_{i,t}{\beta }_{i,t}+u_{i,t},u_{i,t}=\lambda W{u}_{i,t}+v_{i,t}$$

(12)

$$y_{i,t}={\alpha }_{i,t}+\rho W{y}_{i,t}+W{x}_{i,t}{\gamma }_{i,t}+x_{i,t}{\beta }_{i,t}+{\epsilon }_{i,t}$$

(13)

Among them, $y_{i,t}$ is the explanatory variable; W is the spatial weight matrix; $x_{i,t}$ is the explanatory variable; ${\epsilon }_{i,t}$ and $v_{i,t}$ are the random error terms with mean 0.$\rho$ and $\gamma$are the spatial lag parameters, whose values are significantly different from 0 or cannot be used to select the type of spatial econometric model. When $\gamma$ = 0 and $\rho$ ≠ 0, the SDM degenerates to a SLM; when $\gamma$ + $\rho$ = 0, the SDM simplifies to a spatial error model.

Following the modeling principle of “from general to specific”, model selection begins with the SDM. The SDM is as follows:

$$Y_{it}=\rho \sum _{j=1}^N{W}_{ij}{Y}_{ij}+\beta X_{it}+\sum _{j=1}^N\theta W_{ij}{X}_{ji}+{\mu }_i+{\varphi }_t+{\epsilon }_{it}$$

(14)

Y denotes the dependent variable, which is the IGD level of prefecture i in year t. W is an element in the spatial weight matrix of order N × N, which denotes the proximity relationship between prefecture i and j. In this study, the N is 54, which represents the 54 coastal prefectural-level cities in China. Two methods are selected to measure the spatial matrix, with one being the adjacency matrix. Second, the reciprocal of the shortest distance between the centers of each prefecture-level city is used as the weight matrix. $\rho$ is the spatial lag regression coefficient, representing the direction and degree of influence of the IGD level of adjacent regions on the green development level of this region and is an unknown parameter vector. When it is 0, the SDM can be transformed into a SLM. ${\varphi }_t$ represents the time fixation effect; ${\epsilon }_{it}$ represents the spatial autocorrelation error term; ${\mu }_i$ represents the individual fixed effect; $X_{it}$ represents the various influencing factors of City j in year t; $W_{ij}{Y}_{ij}$ represents the interaction influence of the city $X_{jt}$ on $X_{it}$ adjacent to city i.

The Technology–Organization–Environment (TOE) framework is a theoretical model proposed by Tomatzky and Fleischer in 1990, which is used to analyze the influencing factors of enterprise technological innovation and technology adoption. The framework categorizes factors influencing enterprise technology application into three levels: technology, organization, and environment. Based on existing research, this study selects six influencing factors, including the proportion of fiscal research and development investment, the level of innovation performance, the level of human capital, the degree of digital transformation, the superiority of the market environment, and the market potential index. X1 represents the proportion of fiscal investment in scientific research, reflecting the level of science and technology. X2 represents the level of innovation performance, reflecting the innovation capabilities of each city. X3 represents the level of human capital and reflects the level of the labor force. X4 represents the degree of digital transformation and reflects the level of technological application in the region. X5 indicates the superiority of the market environment, reflecting the degree of superiority or inferiority of the comprehensive development conditions of the market. X6 represents the market potential index, reflecting the potential growth value and development space of the market.

At the same time, the consistency between the research method and the research hypothesis is further explained. Methodological alignment with hypotheses: H1 validation: Dagum’s Gini decomposition quantifies inter/intra-MEZ disparities. H2 validation: GeoDetector analyzes driver explanatory power (q-value) dynamics. H3 validation: The Spatial Durbin Model (SDM) with the Wald test identifies spillovers.

4. Results

4.1. Temporal Evolution Characteristics

4.1.1. Overall Trend Analysis

Based on the VHSDM, the IGD levels of the three MEZs from 2012 to 2022 were calculated (Figure 2). The level of IGD in coastal regions maintains an overall growth trend, increasing from 0.624 in 2012 to 0.865 in 2022, with an average annual increase of 3.87%. The eastern MEZ had the highest development level and fastest growth rate, with an increase of 0.375. This also reflects that the degree of variation among cities in this region is the smallest, which indicates relatively balanced development. This is followed by the southern MEZs, where both the development level and growth rate are in the middle of the eastern and northern MEZs, but its internal cities have the largest variance. Hypothesis H1 was verified. The reason for this is that a few high-level or regional central cities, such as Shenzhen, Guangzhou, and Shanghai, have a strong advantage in the aggregation of factor resources. Its comprehensive development level has improved rapidly. Meanwhile, its capacity to radiate and drive the development of surrounding cities is still not strong enough, which has further widened the development gap between regions. The northern MEZs have the lowest level of development and the slowest growth rate, with an increase of only 0.152 during the observation period, gradually distancing itself from the eastern and southern Marine Economic Circles.

4.1.2. Dimensional Difference Analysis

We compare the structural changes in the level of IGD in coastal cities (Figure 3). Overall, the distribution between the three dimensions of economic development, social inclusion, and green ecology is relatively balanced, indicating a high degree of synergy within IGD. According to the change in the dimension with the highest share, three phases can be divided: the social dimension led development from 2012 to 2014, the economic dimension led from 2015 to 2019, and the environmental dimension led from 2020 to 2022. This reveals that there is a shift in the priority focus of the integrated development of coastal cities in different periods. During the observation period, the increase in the environmental quality dimension is far ahead, with an average annual growth rate of 6.8%. The average annual growth rates for the social well-being and economic growth dimensions are 2.9% and 2.4%, respectively. This reflects that coastal cities pay more attention to environmental protection as they take the lead in realizing economic and social development. Comparing the changes in the extremes between different dimensions, the fluctuating downward trend indicates that the coastal areas are continuing to make up for the shortcomings of the ecological environment in high-quality development. Coastal areas will continue to realize a higher degree of synergistic development.

4.1.3. Kernel Density Analysis

KDE is used to further characterize the dynamic evolution of the level of IGD in the three major marine economic zones (Figure 4), specifically as follows: (1) In terms of distribution location, the kernel density curves all show an overall shift to the right. It indicates that the level of IGD in coastal areas has been improved during the study period, with the eastern region moving the longest distance and improving its level most significantly. (2) In terms of distribution pattern, the height of the main peak of the kernel density curves becomes lower and wider. This indicates that the difference in the level of IGD in coastal areas is gradually expanding. Also, the fastest decline in the peak is in the southern region, which indicates that the differentiation of this region is very obvious. This further contributes to the expansion of geographic differentiation in the overall level of development. Additionally, regarding distribution extensiveness, all kernel density curves exhibit a significant right trailing phenomenon, with the length of the tail gradually increasing over time. Although most of the coastal cities are still concentrated in the low-level state, the number of cities entering the high-level state tends to increase. (3) In terms of the polarization characteristics, the kernel density curve in the early stage is dominated by a “single peak.” Then in the middle and later stages, the “multi-peak” characteristics gradually increase, with the curves in the middle and high-level positions having a protrusion. In the middle and late stages, the characteristic of “multiple peaks” gradually increases, and the curves at the middle and high levels show signs of protrusion. This indicates that the IGD level of the coastal cities has been multipolarized.

4.2. Spatial Distribution Characteristics

4.2.1. Spatial Pattern Analysis

Arcgis was used to present the characteristics of the spatial distribution of IGD in coastal cities in 2012, 2017, and 2022 (Figure 5). In terms of trend, except for Dandong, all 52 coastal cities realized growth: 25 cities increased more than 0.2, and the number of high-level cities (>1) increased from 4 in 2012 to 12 in 2022. The cities with the largest increases are all located in the eastern or southern marine economic zones, with the top five in order being Guangzhou, Hangzhou, Xiamen, Shanghai, and Ningbo, all of which are large cities at the sub-provincial level and above. In contrast, cities with smaller increases are mainly located in the Northern Marine Economic Circle, especially coastal cities in Liaoning, such as Huludao and Jinzhou, as well as non-Pearl River Delta areas in Guangdong Province, such as Maoming, Jieyang, and Shanwei. Most of them are small and medium-sized cities. Spatially, the three major Marine Economic Circles have each formed high-level agglomerations of IGD, which in 2012 corresponded to the three economically developed regions of the Liaodong Peninsula, the YRD, and the PRD. By 2017, the scope of high-level development areas in the YRD and PRD regions had further expanded. The Shandong Peninsula in the northern region has rapidly emerged. On the other hand, the Liaodong Peninsula has remained relatively unchanged. By 2022, the original three high-level regions continued to improve and expand. Also, a high-level development region will emerge on the west coast of the Fujian Strait. As a result, the spatial pattern of the Shandong Peninsula, YRD, West Coast of the Taiwan Strait, and PRD will be formed in China’s coastal area from north to south.

4.2.2. Spatial Correlation Analysis

(1)

Global Moran’s I

This index reflects the overall correlation characteristics of IGD in coastal cities (

Table 2. Global Moran’s I for IGD.

Year	Moran’s I	Z	p
2012	0.316	3.192	0.004
2013	0.331	3.463	0.003
2014	0.345	3.594	0.002
2015	0.359	3.725	0.001
2016	0.373	3.856	0.001
2017	0.387	3.987	0.001
2018	0.401	4.119	0.001
2019	0.415	4.250	0.001
2020	0.342	3.518	0.002
2021	0.443	4.512	0.001
2022	0.457	4.643	0.001

). The results show that the global Moran’s I is greater than 0.3. The statistic Z value is greater than 2.58, which passes the significance test at a 99% confidence level. It indicates that there are significant positive spatial correlation characteristics of IGD in China’s coastal cities. The reason might be that there are multiple city clusters or economic belts in China’s coastal areas, with frequent flows of factors such as capital, talent, and technology within them, and a deepening degree of integrated development; on the other hand, in the context of the construction of an ecological civilization, the coastal areas are exploring a sound cross-regional collaborative governance mechanism for the ecological environment to prevent and resolve the problem of the cross-border transfer of pollution. In terms of the trend of change, the global Moran’s I is fluctuating before 2020 and continues to grow steadily after 2020. The trend of spatial agglomeration is further enhanced. The finding suggests that as the comprehensive strength of China’s coastal regions continues to improve, it promotes industrial collaboration and knowledge spillover between neighboring cities, which in turn exacerbates the spatial agglomeration characteristics of IGD.

(2)

Local Moran’s I

The local Moran’s I is mainly used to analyze the degree of spatial clustering of spatial unit attributes with similar values to the surrounding units. It can fully reveal the correlation characteristics of variable observations outside a number of spatial ranges with different spaces. According to the four quadrants in the Moran’s I scatterplot, it is divided into high-high (H-H), low-high (L-H), high-low (H-L), and low-low (L-L) (Figure 6). Each circle denotes a city.

The level of IGD of coastal cities presents three characteristics: (1) Significant spatial dependence: Consistent with the global autocorrelation results, the more geospatially neighboring cities there are, the more consistent their level of IGD is. For example, H-H-type cities are mainly concentrated in the YRD with Shanghai as the core neighboring city and the PRD with Guangzhou and Shenzhen as the core neighboring cities. The spatial linkage effect between cities is significant. L-L-type cities are mainly distributed in the Bohai Rim City Circle, the Beibu Gulf City Circle, and the junction areas of southern Fujian and eastern Guangdong. In general, the coastal areas have formed a “high middle and two sides.” The distribution pattern of IGD in coastal areas is “high in the middle and low on both sides.” (2) Strong spatial heterogeneity: The number of cities distributed by the four types of spatial agglomeration varies greatly. During the observation period, nearly half of the cities are distributed in the L-L type. The comprehensive development of these cities and their surrounding areas is on the low side, which belongs to the “depression” of IGD. This is followed by the H-H type, where the number of distributed cities continues to grow with time. These regional center cities constitute the “highland” of IGD; the L-H and H-L types have the smallest number of cities and show a declining trend over time. The former are mainly located in the southeast and far away from regional center cities, making it difficult for them to receive high-level regional development; the latter are mostly near the center city of the northern region, which may have a certain siphoning effect on the surrounding areas. (3) Higher spatial stability: From 2012 to 2022, the number of cities included in the four types of spatial agglomeration did change much in general. About 1/5 of the cities changed types and most cities maintained their type, reflecting the spatial “cohort effect” of IGD in coastal cities.

4.3. Driving Mechanisms of IGD

4.3.1. GeoDetector Two-Factor Detection Analysis

Figure 7 presents the results of the interaction probes of the drivers in 2012 and 2022. Among them, indicators G1–G3 measure the contribution of green ecology to inclusive green development; G1 is resource endowment, G2 is green production, and G3 is ecological governance. As can be seen from Figure 7, the explanatory power of resource endowment and ecological governance has declined, indicating that inclusive development no longer simply depends on resources but realizes inclusive green development through green production and technological progress. It is worth noting that the explanatory power of ecological governance has declined, which may be due to the one-size-fits-all phenomenon in the implementation of China’s ecological governance policies, which has inhibited inclusive development to a certain extent. The explanatory power of green production has increased significantly, which is the core element that promotes inclusive green development. In 2022, green development became a top priority rather than just a side issue, both worldwide and in China. This shift was pushed forward by stronger interests from markets and society, which made all the parts of the green transition more influential. Back in 2012, China was facing serious environmental problems, especially heavy pollution. At that time, there was an urgent need to tackle these issues. Policies were strong, and the focus on ecological governance made a big difference in addressing the problems that were clear and pressing. Hypothesis H2 was verified.

Specifically, in 2012, the interaction between the social inclusion dimension of equal opportunity and the economic dimension of development level had the highest explanatory power for green development, reaching 0.88. The result suggests that equal social opportunity is a foundational support for green development. This can solve the resource mismatch and make green factors flow to the efficient field. It also enhances governance effectiveness. In 2022, the level of development of the economic dimension and economic growth potential, the level of development of the economic dimension and ecological governance, and the level of development of the economic dimension and equal opportunity of the social inclusion dimension all reached 0.94. The implication is that economic development is the main driver influencing the IGD of the three marine economic zones (Figure 8).

4.3.2. Spatial Model Estimation Results

Before estimating the spatial model, we use LM_Lag, RLM_Lag (spatial lag robustness test) and LM_Error and RLM_Error (spatial error robustness test) combined to select the specific form of the spatial model. The test results indicate that the linear model based on no spatial effects rejects all the original hypotheses except RLM_error and thus can accept both the SAR model and the SEM. However, the Wald test and the LR test significantly reject the hypothesis that the SDM can be degraded into an SLM and a spatial error model, and thus this study prioritizes the use of the SDM for estimation.

In this study, the SDM is estimated using the greatest likelihood estimation method. Since the Hausman test indicates that the fixed-effects model is more suitable for the generation of real data, this study adopts the SDM with double fixation in time and space for estimation. The Hausman test results of the spatial panel model are shown in

Table 3. Hausman’s test.

Test Method	Statistic	p-Value
Hausman	564.42	0.0000
Comparison of double fixation with individual fixation	18.86	0.0008
Dual fixed vs. time fixed	686.47	0.0000

.

The SDM is employed in this research to decompose spatial effects. It identifies the impact of each explanatory variable on the local area, as well as the “spillover effect” exerted on the explained variables of neighboring regions. Avoiding misguidance from a single result and providing a more specific basis for policy formulation are the advantages of this approach. The results of the effect decomposition are shown in

Table 4. Decomposition of spatial effects based on inverse geographic matrices.

Variables	Direct Effect	Spatial Spillover Effect	Total Effect
X1	0.0004 ***	0.0021 ***	0.0026 ***
X2	−5.39 × 10−7 ***	−1.47 × 10−6	−2.01 × 10−6
X3	0.0628 ***	0.3831 ***	0.4459 ***
X4	0.0069	−0.0014	0.0055
X5	−0.0029	0.2074 ***	0.2045 ***
X6	0.0565 ***	0.0565 *** − 0.1189	−0.0624

Notation: *** indicates significance at the level of 1%.

. The results show that under the spatiotemporal double-fixed model, it is still X1, X4, and X5 that are significant and the sign is unchanged, which is in line with the conclusion above. Also, in the indirect effect, X1 and X4 are still significantly positive, which indicates that among the influencing factors studied in this study, the ratio of financial investment in scientific research and the degree of digital transformation have significant spatial spillover effects. Hypothesis H3 was verified. In contrast, the impacts of the level of innovation performance, the level of human capital, the superiority of the market environment, and the market potential index are more concentrated within cities. Specifically, these four factors mainly exert intra-city impacts. The spatial spillover effects of these factors between cities are not significant.

The significant spatial spillover effect of the ratio of financial research investment and the degree of digital transformation between cities indicates that the correlation between the two is not limited to a single city but presents a synergistic linkage across administrative boundaries.

5. Conclusions and Discussion

5.1. Conclusions

This study examined the spatiotemporal evolution and driving factors of inclusive green development (IGD) across coastal cities in China between 2012 and 2022. By applying a hybrid analytical framework that combined VHSDM evaluation, Dagum’s Gini decomposition, GeoDetector analysis and spatial econometric modeling, several key findings were obtained.

Firstly, the level of IGD in coastal China improved steadily during the study period, increasing from 0.624 in 2012 to 0.865 in 2022. However, regional differences remain significant. The eastern MEZ achieved the highest development, while the northern MEZ consistently lagged behind. This is consistent with earlier research that emphasizes persistent north–south disparities in China’s green transition.

Secondly, IGD followed distinct stages of evolution. The focus shifted from social inclusion (2012–2014) to economic development (2015–2019) and to environmental quality (2020–2022). The connection between this shift and national strategies (the “dual carbon” goals) as well as local policy adjustments is strong. Similarly to recent studies, the rising weight of environmental indicators demonstrates that ecological governance has become a central policy priority. However, the reduced contribution of social inclusion reveals an uneven distribution of green development benefits.

Thirdly, spatial clustering has grown stronger since 2020. Four high-performing poles have emerged: the Shandong Peninsula, the YRD, the west coast of the Taiwan Strait, and the PRD. Strong positive spatial dependence suggests that neighboring cities benefit from shared resources, industrial collaboration, and knowledge spillovers. These results align with findings on urban clustering effects in green innovation and digital transformation. Importantly, fiscal investment in R&D and digital transformation were identified as the main drivers with significant spillover effects across city boundaries.

5.2. Discussion

5.2.1. Implications

The results suggest important policy implications, as overall IGD progress has not prevented uneven regional trends. High-performing poles in the YRD and PRD continue to accumulate advantages in infrastructure, innovation, and governance, while many cities in the northern MEZ show only limited progress. The difference indicates that the benefits of IGD are unevenly distributed and may intensify regional inequality if not addressed. For lagging cities, the risk is not simply slower growth but the possibility of becoming locked into a low-level development pattern due to weak absorptive capacity and limited access to shared resources. It implies that spatial polarization is not an incidental byproduct of IGD but a structural issue that demands attention in regional planning and policy design.

Another implication concerns the internal balance of IGD dimensions. The strong rise in environmental indicators shows that ecological governance has made notable progress under the dual carbon strategy. However, this growth has been accompanied by a weaker role for social inclusion compared with the earlier period. In practice, economic and environmental goals are being achieved more visibly, while improvements in equity are less significant and uneven across regions. If this tendency persists, the integrative goal of IGD may be retarded: sustainability gains would not be matched by corresponding advances in inclusiveness. The imbalance could also reduce the perceived fairness of green transition policies, especially in cities or groups that benefit less directly. Therefore, ecological improvements combined with broad social benefits are critical for maintaining both the credibility and long-term effectiveness of IGD.

5.2.2. Policy Recommendations

The empirical results suggest several directions for policy arrangements. The first priority is to reduce the widening disparities among coastal regions by strengthening the role of hub cities as engines of diffusion. Cities such as Shanghai, Guangzhou, and Qingdao already benefit from strong innovation systems, digital infrastructure, and access to capital. They should be encouraged to share these advantages with surrounding cities through joint industrial projects, regional innovation platforms, and inter-city transport and digital networks. Policies that guide the transfer of resources and expertise outward from these centers can help smaller cities avoid being left behind.

A second recommendation is to treat digital transformation and R&D investment as regional rather than purely local strategies. The analysis shows that these factors generate strong spillover effects across administrative boundaries. Coordinated investment in cross-regional digital infrastructure, shared data platforms, and collaborative R&D projects could therefore yield higher overall returns than fragmented, city-level initiatives. Cooperation of this kind would also reduce resource duplication, enabling lagging regions to benefit from the momentum of stronger neighbors.

Thirdly, attention should return to the social inclusion dimension of IGD. The weakening role of equity-related indicators indicates that economic and environmental progress is not always accompanied by improvements in access to education, healthcare, or public services. Closing these gaps could be achieved by three key measures: expanding fiscal transfers to underperforming areas, enhancing the provision of basic services, and developing inclusive financial tools—such as green bonds and targeted subsidies. Another effective approach is to link ecological programs with social welfare initiatives. The linkage would make IGD more tangible to citizens, in turn strengthening both its legitimacy and effectiveness.

Finally, environmental governance should be further harmonized across regions. While progress has been made, uneven enforcement still exists, especially in smaller coastal cities. Stronger mechanisms for cross-regional monitoring, information sharing, and joint enforcement would help ensure that environmental improvements are not concentrated only in more developed areas. By combining regional coordination, social inclusiveness, and consistent ecological regulation, coastal China can move toward a more balanced and sustainable pattern of IGD.

5.2.3. Limitations and Future Research

While integrating green production metrics, micro-level data on corporate green innovation were unavailable, potentially underestimating market-driven IGD drivers. Furthermore, analyzing only coastal cities omits inland–coastal interactions, limiting systemic policy insights. Therefore, future research should incorporate enterprise-level green innovation data to quantify bottom-up IGD drivers. We advise developing trans-regional IGD networks to model resource flows between coastal and inland cities.