The Impact of Multidimensional Regional Integration on Low-Carbon Development: Empirical Evidence from the Yangtze River Delta

Abstract

Amid the deep integration of China’s “dual-carbon” goals with regional coordinated development strategies, this study develops a multidimensional analytical framework of regional integration based on panel data from 41 prefecture-level cities in the Yangtze River Delta urban agglomeration from 2009 to 2023. The framework encompasses five dimensions: urban–rural integration, innovation coordination, infrastructure connectivity, ecological co-governance, and public service sharing. Using structural equation modeling (SEM), the study empirically investigates the mechanisms and pathways through which regional integration shapes low-carbon development. The results indicate that different dimensions exert differentiated impacts: urban–rural integration and infrastructure connectivity significantly promote low-carbon development, whereas public service sharing has an adverse effect due to a phenomenon known as “carbon lock-in”. By contrast, the impact of innovation coordination and ecological co-governance is not statistically significant. Moreover, substantial regional heterogeneity exists: Jiangsu Province demonstrates the leading performance in the manifest development level; Zhejiang Province shows strong systemic capacity level, but limited conversion into manifest outcomes. At the same time, most cities in Anhui Province lag in both aspects. Coordination analysis further identifies four typical development patterns: dual-high, system-driven, performance-dominant, and dual-low. Drawing on these findings, this study proposes policy recommendations across four dimensions—regional coordination, low-carbon pathway optimization, targeted empowerment, and collaborative governance—to facilitate the green and low-carbon transition of the Yangtze River Delta urban agglomeration.

1. Introduction

Amid growing global efforts in climate governance and the Sustainable Development Goals (SDGs) framework, advancing regional low-carbon transitions has become a crucial way to meet worldwide temperature control targets. As the world’s largest carbon emitter, China’s pursuit of carbon peaking and carbon neutrality—known as the “dual-carbon” goals—is not only vital for fulfilling its commitments under the Paris Agreement but also key to promoting high-quality economic growth and building an ecological civilization [1,2]. Currently, China contributes around 30% of global carbon emissions. Facing the dual challenges of maintaining economic growth and reducing emissions, achieving the “dual-carbon” goals requires coordinated, systemic policies that combine regional efforts (IPCC, 2023) [3].

The Yangtze River Delta urban area, one of China’s most dynamic and innovative regions, produces about 24% of the country’s GDP on less than 4% of its land, while accounting for roughly 15% of total carbon emissions (CEADs, 2023). This makes it a prime example for studying the relationship between regional coordination and low-carbon transitions. Regional integration offers a promising route for lowering carbon intensity by breaking down administrative barriers, improving resource distribution, and encouraging coordination between technology and policy [4]. Unlike the market-driven methods of Europe and the United States, the Yangtze River Delta adopts a unique approach—“government-led, policy-coordinated, and infrastructure-driven”—which provides a distinctive framework for regional low-carbon management within the context of national strategic goals.

Although existing studies have provided key insights into the relationship between regional integration and low-carbon transition, several areas still require further development. First, prior research has mainly focused on single dimensions (such as innovation-driven or environmental governance). At the same time, relatively little attention has been paid to an integrated framework covering multiple dimensions, including urban–rural integration, innovation coordination, infrastructure connectivity, ecological co-governance, and shared public services [5]. Second, even when multidimensional perspectives are considered, analyses often remain conceptual, lacking empirical studies capable of identifying interaction effects. This makes it difficult to fully understand whether different subsystems create synergistic benefits or offsetting mechanisms [6]. Third, most existing studies are centered on the Chinese context, with limited focus on cross-regional or cross-national comparisons [7]. Since regional integration and low-carbon transition are global challenges, this limitation is especially significant. Therefore, building a comprehensive framework that incorporates multidimensional factors, interaction mechanisms, and international comparisons remains an important goal for future research.

Against this background, this study focuses on three interrelated research questions:

First, how can a comprehensive and multidimensional evaluation framework for regional integration and low-carbon development be built to accurately reflect the unique developmental features of the Yangtze River Delta?

Second, how do the various dimensions of regional integration—such as urban–rural integration, innovation coordination, infrastructure connectivity, ecological governance, and public service sharing—impact low-carbon development?

Finally, based on the empirical findings, what specific policy tools and governance mechanisms can be suggested to improve the regional integration process and thereby promote more effective low-carbon development?

To address these questions, this study uses panel data from 41 cities in the Yangtze River Delta from 2009 to 2023 and follows a research process of “framework construction–mechanism testing–policy application.” A structural equation model (SEM) is developed to systematically analyze the multidimensional impacts of regional integration on low-carbon development. It aims to empirically uncover differentiated policy effects and provide both theoretical and practical insights for promoting high-quality urban agglomeration development through low-carbon strategies.

2. Literature Review

The relationship between regional integration and low-carbon development has long been a significant topic in economic geography and environmental economics. Classical theories of regional integration (Viner, 1950; Balassa, 1961) [8,9] suggest that regional cooperation can improve overall welfare through market expansion, factor mobility, and policy coordination. However, the effectiveness of integration varies across regions, depending on institutional quality, governance capacity, and technological spillovers (Badinger, 2005; Crescenzi et al., 2018) [10,11]. From an environmental economics perspective, the Environmental Kuznets Curve (EKC) hypothesis proposes an inverted U-shaped relationship between economic growth and environmental performance (Grossman & Krueger, 1995) [12]. Later studies have also indicated that regional collaboration and institutional innovation can speed up the transition to a low-carbon economy by fostering the spread of clean technologies and green investments (Costantini & Crespi, 2013; Stern, 2017) [13,14].

At the empirical level, scholars have studied the relationship between regional integration and low-carbon development from various perspectives. Some studies focus on economic integration and industrial specialization, arguing that interregional trade cooperation and technological diffusion can boost energy efficiency and lower carbon emissions (Xiao et al., 2022; Wang et al., 2024) [15,16]. Others highlight the importance of infrastructure connectivity and spatial flows, noting that improvements in transportation networks, information and communication systems, and energy transmission infrastructure are crucial for achieving green integration (Xie et al., 2017; Li et al., 2024) [17,18]. Additionally, international institutional cooperation—such as the European Union’s Emissions Trading System (EU ETS), green investment funds, and cross-border energy markets—has demonstrated the low-carbon benefits of deep regional integration in practice (UNEP, 2022; OECD, 2023) [19,20]. While there is growing focus on urban–rural integration, public service sharing, and ecological co-governance, most existing studies are case-based or focus on a single aspect, lacking comprehensive quantitative validation and mechanism analysis.

Overall, existing research has made significant progress in uncovering the connection between regional integration and low-carbon development, establishing a strong foundation for further exploration. However, there is still room for improvement. First, most studies focus on only one aspect—such as economic integration, transport connectivity, or ecological cooperation—while comprehensive multidimensional analytical frameworks remain limited. Second, methodologically, many studies rely on static linear models or single-layer regressions, which identify general trends but fail to capture the complex interactions and structural relationships among regional factors, as well as their dynamic differences across development stages. Third, at the theoretical level, the link between regional integration studies and the fields of environmental economics and sustainable transition remains underdeveloped. This emphasizes the need for a unified analytical framework to deepen understanding of the “regional coordination–institutional innovation–low-carbon transition” mechanism.

Building on these insights, this paper aims to contribute both theoretically and empirically. Specifically, it seeks to (1) create a multidimensional analytical framework of regional integration covering economic coordination, innovation cooperation, infrastructure connectivity, urban–rural integration, and ecological co-governance, thereby capturing the multiple aspects of regional coordination; (2) use a structural equation modeling (SEM) approach to empirically analyze how different dimensions of integration influence low-carbon development, focusing on interaction effects and regional differences; and (3) suggest targeted policy recommendations and governance mechanisms based on the classification of cities in the Yangtze River Delta. Overall, this study broadens the current research on the link between regional integration and low-carbon development, offering new empirical evidence for building an analytical framework of regional collaborative governance and low-carbon transition.

3. Mechanism of Action and Research Hypotheses

3.1. Theoretical Framework and Synergistic Effects of Integration in the Yangtze River Delta Urban Agglomeration

According to the Master Plan for the Integrated Development of the Yangtze River Delta Region, regional integration in the Yangtze River Delta (YRD) can be systematically broken down into five key areas: urban–rural integration, technological innovation, infrastructure connectivity, ecological co-governance, and public service sharing. To strengthen the theoretical basis, this study frames these five areas within the concepts of classical regional integration theory (Balassa, 1961) [9], collaborative governance theory (Emerson et al., 2012) [21], and sustainability transition theory (Geels, 2002) [22].

Specifically, urban–rural integration illustrates how spatial factors come together and promote inclusive growth; innovation collaboration demonstrates the processes of technological progress and knowledge sharing; infrastructure connectivity highlights the linking of material and information networks; ecological co-governance involves managing environmental issues across jurisdictions and addressing ecological externalities; and public service sharing focuses on social fairness and inclusiveness during low-carbon transitions.

Although these five dimensions are interconnected, they are believed to correspond to separate collaborative processes across the economic, technological, spatial, ecological, and social fields, collectively forming a multidimensional coordination system for regional low-carbon development [23]. Among these, urban–rural integration promotes the two-way flow of production factors through land and household registration reforms, encouraging agricultural modernization and improving emission reduction efficiency [24,25]; technological innovation integration, supported by platforms such as the G60 Science and Innovation Corridor, speeds up the research and spread of green technologies like renewable energy and carbon capture [26]; infrastructure integration increases resource allocation efficiency through better transportation and energy networks [27]; ecological integration advances joint pollution control via cross-regional ecological compensation and unified environmental standards [28]; and public service integration boosts social resilience and inclusiveness in the low-carbon transition through shared access to healthcare, education, and other public services.

Given the advantages of structural equation modeling (SEM) in addressing multicollinearity and capturing complex causal relationships among latent variables, this study uses SEM to systematically analyze the overall effects of these five integration dimensions on regional low-carbon development.

3.2. Urban–Rural Integration and Low-Carbon Development

Urban–rural integration encourages low-carbon development by reducing the income gap between urban and rural areas, increasing per capita GDP, and promoting urbanization [29]. First, narrowing income disparity boosts consumption upgrades, leading to higher demand for green products and increased low-carbon investments in rural regions [30]. Second, higher levels of economic development establish the financial foundation for research, development, and large-scale deployment of clean technologies, supporting a shift toward a low-carbon industrial structure [31]. Third, the urbanization process improves energy resource allocation through agglomeration effects and economies of scale, enhancing energy efficiency and speeding up the spread of green technologies across urban and rural areas [32]. To effectively understand these mechanisms, this study uses the urbanization rate of the permanent population, the urban–rural income gap, and per capita GDP as key indicators. The urbanization rate indicates the degree of factor integration; the income gap reflects urban–rural disparity; and per capita GDP represents the economic base for low-carbon transition. Therefore, we propose the following hypothesis:

H1. 

Urban–rural integration significantly promotes low-carbon development.

3.3. Innovation Coordination and Low-Carbon Development

Innovation coordination promotes the low-carbon transition through R&D investment, patent efficiency, and the expansion of high-tech industries [33]. First, R&D investment forms the basis for developing low-carbon technologies like clean energy and carbon capture [34]. Second, effective patent output speeds up the spread of these technologies, supporting the clean transformation of energy systems [35]. Third, the growth of high-tech industries lowers regional carbon intensity by replacing traditional high-emission sectors and creating positive technological spillovers [36]. Additionally, innovation coordination not only fosters technological progress but also encourages industrial upgrading, which helps decouple economic growth from carbon emissions [37]. Consequently, R&D investment intensity, patent authorization rate, and the share of high-tech industry output serve as observed variables. These indicators collectively represent the green innovation coordination chain from technology development to market adoption. Based on this, we propose the following hypothesis:

H2. 

Coordinating innovation has a substantial positive impact on low-carbon development.

3.4. Infrastructure Connectivity and Low-Carbon Development

Infrastructure connectivity promotes low-carbon development through transportation networks, internet access, and information exchange. First, improved transportation infrastructure reduces energy use and carbon emissions in logistics and freight transport [38]. Second, widespread internet access supports the development of intelligent energy systems, enabling better energy management and demand regulation [39]. Third, increased information exchange speeds up the sharing of low-carbon knowledge and the adoption of clean technologies, encouraging behavioral shifts toward low-carbon lifestyles [40]. Additionally, by boosting total factor productivity and making resource allocation more efficient, infrastructure connectivity provides a solid foundation for the transition to a low-carbon economy. This study uses highway density, Internet penetration rate, and per capita postal volume as measurement indicators. Highway density indicates the physical connectivity of transportation networks. Meanwhile, the Internet penetration rate measures digital information connectivity, and per capita postal volume reflects the intensity of information exchange and logistics activity. Together, these indicators represent the physical and digital infrastructure essential for low-carbon development. Therefore, we propose the following hypothesis:

H3. 

Infrastructure connectivity greatly supports low-carbon development.

3.5. Ecological Co-Governance and Low-Carbon Development

Ecological co-governance promotes low-carbon development by improving air quality, water quality, and urban green spaces. First, controlling air pollution reduces reliance on carbon-intensive mitigation measures [41]. Second, enhancing water quality governance decreases energy use related to pollution treatment [42]. Third, urban greening boosts carbon sequestration capacity [43]. Additionally, ecological co-governance creates synergistic benefits by encouraging industrial restructuring and technological innovation through regulatory pressures and collaboration. This generates a positive feedback loop where environmental protection and low-carbon goals strengthen each other. To measure these effects, this study uses PM10 concentration (an inverse indicator), the surface water quality compliance rate, and the green coverage ratio of built-up areas. These indicators reflect regional efforts in air, water, and terrestrial ecosystems, demonstrating joint actions in pollution control and ecological restoration. Therefore, we propose the following hypothesis:

H4. 

Ecological co-governance greatly promotes low-carbon development.

3.6. Service Sharing and Low-Carbon Development

Service sharing influences low-carbon development through fiscal support, healthcare services, and educational access. First, government fiscal spending promotes green infrastructure and sustainable urban growth. Second, enhanced medical services indirectly support a low-carbon transition by boosting human capital and labor productivity [44]. Third, increased educational access raises public environmental awareness and encourages the adoption of low-carbon behaviors [45]. However, the energy demands of expanding public service facilities—such as hospitals and schools—may temporarily raise carbon emissions, creating a “carbon lock-in” effect during the transition [46]. Consequently, the overall impact of public service sharing on low-carbon development reflects the net balance of its positive effects and potential lock-in drawbacks. To measure this net effect, the study uses per capita public fiscal spending, hospital beds per 1000 people, and the gross enrollment ratio in upper secondary education, representing government investment, healthcare provision, and human capital development, respectively.

Since SEM is used to simultaneously examine the combined effects of all five dimensions on low-carbon development, the analysis concentrates on the net influence of public service sharing. The investigation of its dynamic, stage-dependent impacts (that is, nonlinear relationships) is reserved for future research. Based on this, the following hypothesis is proposed:

H5. 

Public service sharing greatly influences low-carbon development, with its direction shaped by the balance between positive impacts and negative “carbon lock-in” effects.

4. Construct Measurement and Model Specification

4.1. Construct Measurement

The core variables in this study—low-carbon development, urban–rural integration, innovation coordination, infrastructure connectivity, ecological co-governance, and service sharing—are defined as latent variables in the empirical analysis. Since they are complex and abstract concepts, they cannot be directly observed and must be operationalized through multiple observable indicators.

Based on existing literature [47,48,49] and the policy orientations of the National New-Type Urbanization Plan (2021–2035) and the Yangtze River Delta Regional Integration Development Plan, this study selects representative observable indicators for each latent variable, taking into account the actual development context of the Yangtze River Delta region. To assess low-carbon development, three indicators—energy intensity, forest coverage, and the share of non-fossil energy consumption—are used based on theoretical reasoning and data availability. This indicator framework follows the “energy–ecology–structure” logic often used in prior studies and offers a relatively comprehensive reflection of energy efficiency, ecological absorption capacity, and energy structure optimization.

It should be noted that natural endowments may partly affect forest coverage. At the same time, the estimate of the non-fossil energy consumption share is based on provincial energy structures due to the lack of city-level data, which may cause some measurement errors. To ensure measurement accuracy, the estimation method and robustness tests are described in detail later in the paper. Because city-level data are limited, several potential indicators—such as per-capita carbon emissions, green patents, and renewable energy capacity—were not included in the current analysis. Future research will expand the indicator system as data availability improves, increasing the comprehensiveness and explanatory power of low-carbon development measurement.

Subsequently, Structural Equation Modeling (SEM) is used. Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA) are performed to assess the reliability, validity, and factor structure of the measurement model, ensuring that the selected indicators accurately represent the corresponding theoretical constructs. The final multidimensional measurement system for regional integration and low-carbon development is shown in

Table 1. Variable definitions.

Latent Variable	Manifest Indicator	Symbol	Definition/Calculation	Indicator Attribute
Low-Carbon Development	Energy Intensity (Tons of Standard Coal per 10,000 CNY)	Y1	Total energy consumption/GDP	−
	Forest Coverage Ratio (%)	Y2	Forest area/Land area	+
	Share of Non-fossil Energy Consumption (%)	Y3	Non-fossil energy consumption/Total energy consumption	+
Urban–Rural Integration	Urban–rural Income Gap (Times)	X1	Per capita disposable income of urban residents/Per capita disposable income of rural residents	−
	Per Capita GDP (CNY per Capita)	X2	GDP/Number of permanent residents	+
	Urbanization Rate of Permanent Population (%)	X3	Urban population/Permanent resident population	+
Innovation coordination	R&D Investment Intensity (%)	X4	Total R&D investment of the whole society/GDP	+
	Patent Authorization Rate (%)	X5	Number of authorized patents/Number of patent applications	+
	Share of High-Tech Industry Output (%)	X6	Output value of high-tech industries/Gross output value of industrial enterprises above designated size	+
Infrastructure Connectivity	Highway Density (km/100 km2)	X7	Total highway mileage/Land area of the region	+
	Internet Penetration Rate (%)	X8	Number of internet users/Resident population	+
	Per Capita Volume of Postal (CNY per Capita)	X9	Total postal and telecommunications business volume/Number of permanent residents	+
Ecological Co-Governance	Concentration of PM10 (μg/m3)	X10	Annual average concentration of inhalable particulate matter	−
	Surface Water Quality Compliance Rate (%)	X11	Number of surface water monitoring sections meeting Class III or above water quality standards/Total number of monitoring sections	+
	Green Coverage Ratio of Built-up Areas (%)	X12	Green coverage area in built-up areas/Built-up area	+
Shared Services	Per Capita Public Fiscal Expenditure (CNY per Capita)	X13	General public budget expenditure of the region/Number of permanent residents in the region	+
	Number of Hospital Beds per 1000 People (Beds per Thousand People)	X14	(Number of hospital beds in medical and health institutions/Total population) × 1000	+
	Gross Enrollment Ratio in Upper Secondary Education (%)	X15	(Number of students in regular senior high schools + Number of students in adult senior high schools + Number of students in secondary vocational schools)/Total population in the eligible age group for senior high school education	+

Note: “−” denotes a negative indicator (where higher values indicate poorer performance), and “+” denotes a positive indicator (where higher values indicate better performance).

.

4.2. Data Sources

This study uses panel data from 41 prefecture-level cities in the Yangtze River Delta urban area, covering the period 2009–2023. The primary sources include the Shanghai Statistical Yearbook, Jiangsu Statistical Yearbook, Zhejiang Statistical Yearbook, and Anhui Statistical Yearbook, as well as municipal statistical yearbooks, statistical bulletins, and the official websites of local bureaus of statistics. When data are missing for certain cities, the gaps are filled using the geometric mean and temporal interpolation based on nearby years, ensuring data continuity and reliability.

Additionally, because official statistical and energy yearbooks offer limited or unpublished data on the share of non-fossil fuel energy consumption, this study uses an estimation method. Specifically: (i) total energy consumption for each prefecture-level city is calculated by multiplying its energy intensity per unit of GDP by its total GDP; (ii) non-fossil energy consumption is estimated by multiplying total electricity consumption by the provincial share of non-fossil energy in electricity generation (for Shanghai, Jiangsu, Zhejiang, and Anhui); and (iii) the share of non-fossil energy consumption is determined by dividing the estimated non-fossil energy use by total energy consumption. This approach offers a reasonable proxy for a key indicator of low-carbon development when direct statistical data are unavailable.

It should be noted that this estimation method assumes that the provincial energy structure is evenly distributed across cities, which could introduce potential measurement bias. To test the robustness of the main results against this assumption, a sensitivity analysis was performed by removing the indicator “share of non-fossil energy consumption” from the low-carbon development index. Only energy intensity and forest coverage rate were kept as observed variables, and the structural equation model was re-estimated accordingly.

As reported in Appendix A 

Table A1. Robustness test results after excluding the indicator “proportion of non-fossil energy consumption”.

Variable	Excluding “Proportion of Non-Fossil Energy Consumption”		Baseline SEM Model
	Unstandardized Coefficient	Significance	Unstandardized Coefficient	Significance
Urban–Rural Integration	3.714 ***	Significant	1.973 ***	Significant
Innovation Coordination	−1.616	Not Significant	−0.126	Not Significant
Infrastructure Connectivity	−2.164 **	Significant	2.032 ***	Significant
Ecological Co-Governance	0.711	Not Significant	−0.358	Not Significant
Shared Services	4.260 ***	Significant	−2.677 ***	Significant

Note: *** and ** denote significance at the 1% and 5% levels, respectively.

, the positive effect of urban–rural integration remains statistically significant and consistent in direction, confirming its robustness. The coefficients of innovation coordination and ecological co-governance remain statistically insignificant, while infrastructure connectivity and public service sharing show sign reversals, indicating some sensitivity to indicator specification. Overall, despite minor variations in coefficient direction, the main relationships identified in the baseline model remain statistically robust and theoretically consistent.

Moreover, the “top-down” downscaling approach, which estimates city-level energy consumption from provincial data, has been systematically developed and validated in the fields of energy and environmental economics as an effective solution to the problem of data unavailability. By creating more advanced allocation indicators, previous studies have shown the reliability and accuracy of this method in overcoming the lack of city-level data [50,51]. The simplified estimation logic used in this study aligns with this methodological framework, further improving the rationality and credibility of our measurement approach.

4.3. Structural Equation Modeling (SEM)

Structural Equation Modeling (SEM) is an advanced statistical technique capable of effectively addressing issues related to latent variables that cannot be directly observed. Compared with traditional statistical methods, SEM has the advantage of analyzing the interactions among multiple independent variables and their effects on multiple dependent variables simultaneously. In this study, SEM is used to examine the impact pathways and effect strengths of multidimensional regional integration on low-carbon development. The specific formulation of the model is shown in Equations (1)–(3).

$$\eta =\beta \eta +\Gamma \xi +\zeta$$

(1)

$$X={\Lambda }_X\xi +\delta$$

(2)

$$Y={\Lambda }_Y\eta +\epsilon$$

(3)

where $\eta$ denotes endogenous latent variables, $\xi$ denotes exogenous latent variables, and $\beta$ and $\Gamma$ are path coefficients. $X$ represents the observed indicators of exogenous latent variables, while $Y$ represents the observed indicators of endogenous latent variables. ${\Lambda }_X$ and ${\Lambda }_Y$ are the factor loadings of observed indicators on the corresponding latent variables, and $\zeta$, $\delta$, and $\epsilon$ are the error terms.

4.4. Initial Structural Equation Model Specification

Building on the theoretical framework and research hypotheses outlined above, this study develops a structural equation model (SEM) to examine the relationships between multidimensional regional integration and low-carbon development. The model specifies direct paths from five latent variables—urban–rural integration, innovation coordination, infrastructure connectivity, ecological co-governance, and service sharing—to the endogenous latent variable representing low-carbon development, while also allowing correlations among the exogenous latent variables.

Following established methodologies and prior empirical studies [47,52,53], the initial model is estimated using AMOS 29, with the path structure shown in Figure 1. In this specification, each of the five exogenous latent variables is connected to the endogenous construct of low-carbon development through directed paths, allowing for the simultaneous testing of hypotheses H1–H5. This initial specification forms the basis for future evaluation and refinement of the model.

It is important to note that this study defines the causal direction as “regional integration drives low-carbon development.” This assumption is based on three primary considerations. First, from a theoretical perspective, the mechanisms discussed in Section 3 show that factors like urban–rural integration and infrastructure connectivity affect energy use and carbon emissions through proactive channels such as factor mobility and technological diffusion. Second, from a policy perspective, the Yangtze River Delta Regional Integration Development Plan—a national strategic document—clearly states that regional integration is a necessary prerequisite and key tool for promoting green and low-carbon transformation. Third, from an empirical and practical viewpoint, the large-scale push for regional integration in the Yangtze River Delta occurred before the systemic phase of low-carbon transition in the time series, supporting the logical and temporal validity of this causal direction.

5. Empirical Analysis

5.1. Descriptive Statistics

This study utilizes SPSS 27 to perform descriptive statistics on all observed variables based on panel data from 41 cities in the Yangtze River Delta from 2009 to 2023. The analysis includes key statistical measures such as mean (M), median (Mdn), minimum (Min), maximum (Max), standard deviation (SD), skewness (Skew), and kurtosis (Kurt). The detailed results are shown in

Table 2. Descriptive statistics.

Variable Name	M	Mdn	Min	Max	SD	Skew	Kurt	N
Energy Intensity	0.613	0.557	0.104	2.180	0.282	1.520	3.800	615
Forest Coverage Ratio	35.583	27.330	6.200	83.250	19.579	0.799	−0.517	615
Share of Non-fossil Energy Consumption	1.390	0.793	0.017	12.522	1.632	2.587	8.915	615
Urban–Rural Income Gap	2.172	2.102	0.231	3.606	0.378	0.648	1.612	615
Per Capita GDP	74,517.670	66,463.000	7288.000	206,300.000	41,597.080	0.713	−0.115	615
Urbanization Rate of Permanent Population	61.327	61.840	29.100	89.600	12.668	−0.102	−0.172	615
R&D Investment Intensity	1.991	2.010	0.100	4.400	0.883	0.053	−0.531	615
Patent Authorization Rate	70.137	70.306	24.196	107.509	16.889	−0.251	−0.642	615
Share of High-Tech Industry Output	36.800	35.600	4.900	73.200	11.859	0.267	0.134	615
Highway Density	0.044	0.038	0.007	0.368	0.028	4.035	34.065	615
Internet Penetration Rate	29.436	28.280	1.420	83.750	15.931	0.238	−0.768	615
Per Capita Volume of Postal	0.215	0.147	0.013	1.475	0.220	2.686	7.855	615
Concentration of PM10	74.032	73.000	0.102	143.000	20.904	0.241	−0.112	615
Surface Water Quality Compliance Rate	75.777	83.300	0.000	100.000	25.593	−0.999	0.102	615
Green Coverage Ratio of Built-up Areas	42.057	42.480	21.740	77.780	3.794	0.301	18.258	615
Per Capita Public Fiscal Expenditure	10,672.479	9734.367	1576.791	38,748.558	5694.497	1.426	3.366	615
Number of Hospital Beds per 1000 People	5.087	5.060	1.635	8.794	1.393	0.050	−0.684	615
Gross Enrollment Ratio in Upper Secondary Education	97.704	99.000	55.200	136.250	8.779	−0.486	6.020	615

.

In terms of distributional characteristics, the variables show marked heterogeneity, indicating significant differences in development levels and structural makeup across the Yangtze River Delta region. The average energy intensity is 0.613, with a standard deviation of 0.282, and its distribution is right-skewed (Skew = 1.520). This suggests that although most cities have relatively high energy efficiency, a few with extremely low efficiency pull down the overall regional average, resulting in notable disparities in energy consumption. The share of non-fossil energy use averages only 1.390% with a standard deviation of 1.632 and a strong right skew (Skew = 2.587). This implies that clean energy adoption remains limited overall and is mainly driven by a few leading cities, highlighting the challenge of achieving a balanced, regional energy transition.

In the economic and social aspects, per capita GDP averages 74,517.67 RMB with a significant standard deviation of 41,597.08, showing a right-skewed distribution. This pattern highlights significant disparities in economic development, as a few highly developed cities significantly raise the regional average. The urban–rural income ratio has an average of 2.172 and a standard deviation of 0.378, indicating a slightly right-skewed distribution. This suggests that while most cities maintain the income gap within a manageable range, a small number still face notable urban–rural inequalities. R&D intensity shows relatively slight variation (SD = 0.883%), whereas patent authorization rates display much greater variability (SD = 16.889%). This contrast implies that differences in innovation output efficiency across cities are much larger than those in innovation input, pointing to inefficiencies in converting R&D investments into concrete technological results.

For infrastructure and public services, indicators such as internet penetration (SD = 15.931%), the proportion of surface water monitoring sections meeting water quality standards (SD = 25.593%), and PM10 concentration (SD = 20.904 µg/m³) show significant variability, emphasizing uneven progress across cities in digital infrastructure, water quality management, and air quality control. In contrast, public service indicators like hospital beds per thousand residents (SD = 1.393) and senior secondary education enrollment rates (SD = 8.779%) display much less variation. This indicates that basic public services have achieved relatively high and equitable coverage throughout the region.

5.2. Reliability and Validity Analysis

This study first conducted a multicollinearity test for the five dimensions of regional integration—urban–rural integration, innovation coordination, infrastructure connectivity, ecological co-governance, and public service sharing—using SPSS 27. The results are shown in

Table 3. Multicollinearity diagnostics.

Latent Variable	VIF	Tolerance
Urban–Rural Integration	5.088	0.197
Innovation Coordination	3.834	0.261
Infrastructure Connectivity	3.017	0.331
Ecological Co-Governance	1.618	0.618
Shared Services	3.776	0.265

. The analysis indicates that all variance inflation factors (VIF) are below 10, and all tolerance values exceed 0.1, suggesting that there is no serious multicollinearity among the variables. Therefore, the model structure demonstrates good discriminant validity and stability.

Based on this, to further assess the reliability and validity of the measurement model, the study used SPSS 27 to conduct analyses, and the results are summarized in

Table 4. Reliability and validity analysis.

Parameter	Overall Scale	Low-Carbon Development	Urban–Rural Integration	Innovation Coordination	Infrastructure Connectivity	Ecological Co-Governance	Shared Service
Number of Items	18	3	3	3	3	3	3
Cronbach’s Alpha	0.958	0.805	0.934	0.682	0.771	0.654	0.868
KMOMeasure	0.966	0.611	0.743	0.534	0.617	0.627	0.623
Bartlett’s Test (Approx. χ2)	11,786.882 ***	876.353 ***	1633.615 ***	589.723 ***	581.556 ***	280.587 ***	1382.852 ***

Note: *** indicates significance at the 1% level.

.

For reliability, the overall scale achieved a Cronbach’s Alpha of 0.958, indicating excellent internal consistency and confirming that the items reliably measure the intended constructs. All dimensions surpassed the acceptable threshold (α > 0.6). Among them, urban–rural integration (α = 0.934) and service sharing (α = 0.868) showed strong reliability, reflecting high consistency among their respective indicators. Innovation coordination (α = 0.682) and ecological co-governance (α = 0.654) reported somewhat lower but still acceptable values, likely due to the inherent complexity of these constructs and the influence of external contextual factors on their indicators.

For validity, the overall Kaiser–Meyer–Olkin (KMO) value was 0.966, and Bartlett’s test of sphericity was significant (p < 0.001), confirming that the dataset is highly suitable for factor analysis. At the dimension level, urban–rural integration (KMO = 0.743) and infrastructure connectivity (KMO = 0.617) showed good sampling adequacy. Although innovation coordination (KMO = 0.534) and ecological co-governance (KMO = 0.627) had lower values, both surpassed the minimum criterion (KMO > 0.5), indicating that their factor structures remain understandable and capable of capturing the complexity of regional innovation and ecological collaboration.

Overall, the measurement scale shows high reliability and sufficient construct validity, meeting established psychometric standards and supporting its use in later SEM analysis.

5.3. Testing the Pathways of the Yangtze River Delta Urban Agglomeration Integration on Low-Carbon Development

To verify the validity of the theoretical model linking regional integration and low-carbon development in the Yangtze River Delta urban agglomeration, this study performed a structural equation modeling (SEM) goodness-of-fit test. As shown in

Table 5. Results of structural equation model fit tests.

Fit Type	Fit Index	Model Fit Evaluation		Reference Standard	Judgment
		Initial Model	Modifies Model
Absolute Fit	GFI	0.869	0.946	≥0.9 (good), ≥0.8 (acceptable)	Good Fit
	AGFI	0.813	0.911	≥0.9 (good), ≥0.8 (acceptable)	Good Fit
	RMSEA	0.091	0.057	<0.08	Good Fit
	CMIN/DF	6.107	2.966	<3	Good Fit
Parsimony Fit	PCFI	0.743	0.662	≥0.5	Good Fit
	PNFI	0.736	0.656	≥0.5	Good Fit
	PGFI	0.610	0.570	≥0.5	Good Fit
Incremental Fit	IFI	0.948	0.983	≥0.9 (good), ≥0.8 (acceptable)	Good Fit
	CFI	0.948	0.983	≥0.9 (good), ≥0.8 (acceptable)	Good Fit
	TLI	0.934	0.974	≥0.9 (good), ≥0.8 (acceptable)	Good Fit
	NFI	0.939	0.974	≥0.9 (good), ≥0.8 (acceptable)	Good Fit

, the model was adjusted after the initial estimation based on theoretical considerations and statistical diagnostics. The revision process followed the principle of “theory first, statistical improvement second,” and was guided by two main criteria:

(1)

In the measurement model, the standardized factor loadings of each observed variable on its corresponding latent construct needed to be above 0.6 (with a minimum acceptable level of 0.5). Meanwhile, their squared multiple correlations (SMC) were expected to be ideally greater than 0.36 and not less than 0.25. These thresholds ensure that the latent constructs have enough association and explanatory power for their observed indicators.

(2)

Referring to the Modification Indices (MI) output from AMOS, correlations between specific measurement error terms were only allowed when they were theoretically justified, to reduce residual correlations and improve the overall model fit.

After model modification, all fit indices reached acceptable or higher levels. Among the absolute fit indices, the Goodness-of-Fit Index (GFI = 0.946), Adjusted Goodness-of-Fit Index (AGFI = 0.911), Root Mean Square Error of Approximation (RMSEA = 0.057), and CMIN/DF (2.966) collectively show that the model structure fits well with the observed data. The parsimony fit indices (PCFI, PNFI, and PGFI) are all above 0.5, indicating that the model maintains simplicity without over-parameterization. The incremental fit indices (IFI = 0.923, CFI = 0.918, TLI = 0.904, and NFI = 0.901) are near or surpass the usual cutoff of 0.90, further demonstrating significant improvement compared to the baseline model.

Overall, the SEM shows strong performance across various fit criteria, confirming the validity of the proposed theoretical framework. The model thus offers a reliable foundation for future hypothesis testing and path analysis, supporting a systematic examination of how urban–rural integration, innovation coordination, infrastructure connectivity, ecological co-governance, and service sharing influence low-carbon development.

Based on the revised model (Figure 2) and hypothesis testing results (

Table 6. Path coefficients and hypothesis testing results.

Path			Unstandardized Coefficient	S.E.	T-Value	Hypothesis	Significance
Low-Carbon Development	<---	Urban–Rural Integration	1.973 ***	0.57	3.463	H1	Significant
Low-Carbon Development	<---	Innovation Coordination	−0.126	0.231	−0.544	H2	Not Significant
Low-Carbon Development	<---	Infrastructure Connectivity	2.032 ***	0.542	3.751	H3	Significant
Low-Carbon Development	<---	Ecological Co-Governance	−0.358	0.352	−1.017	H4	Not Significant
Low-Carbon Development	<---	Shared Services	−2.677 ***	0.805	−3.327	H5	Significant

Note: *** indicates significance at the 1% level.

), urban–rural integration, infrastructure connectivity, and service sharing have significant effects on low-carbon development, with unstandardized path coefficients of 1.973 (T = 3.463, p < 0.001), 2.032 (T = 3.751, p < 0.001), and −2.677 (T = −3.327, p < 0.001), respectively. Therefore, hypotheses H1, H3, and H5 are supported. Specifically, infrastructure connectivity has the strongest positive effect (B = 2.032), indicating that a one-unit increase in this area results in a notable 2.032-unit improvement in low-carbon development, underscoring the crucial role of infrastructure in advancing the region’s green transition. Urban–rural integration also demonstrates a steady positive influence (B = 1.973), suggesting it effectively promotes low-carbon development by encouraging factor integration and structural optimization.

In contrast, service sharing has a significant adverse effect (B = −2.677), indicating that each one-unit increase in this domain may decrease low-carbon development by 2.677 units. This confirms the presence of a “carbon lock-in” effect during the expansion of public service facilities, where the energy-intensive construction and operation of healthcare and educational infrastructure add to short-term carbon burdens. This result supports the theoretical expectation of H5.

On the other hand, innovation coordination (B = −0.126, T = −0.544) and ecological co-governance (B = −0.358, T = −1.017) do not show statistically significant effects on low-carbon development, so H2 and H4 are not supported. The limited impact of innovation coordination may stem from the long cycle of green technology development—from R&D to commercialization—and the still-developing regional innovation system. The insignificant role of ecological co-governance is probably due to inconsistent enforcement of environmental regulations, underdeveloped ecological compensation mechanisms, and weak governance coordination across the region. This is especially evident in end-of-pipe issues such as air pollution (standardized factor loading = 0.733), where their ongoing challenges continue to hinder broader ecological progress.

Furthermore, the model shows strong correlations among the five dimensions (all correlation coefficients (r) > 0.90, p < 0.001), indicating that the dimensions of regional integration are highly interconnected. This finding not only confirms the independent contributions of urban–rural integration, infrastructure connectivity, and service sharing to low-carbon development but also suggests that innovation coordination and ecological co-governance have yet to fulfill their expected roles in the region’s current low-carbon transition stage.

5.4. Robustness Test

To verify the reliability of the baseline results obtained from the structural equation model (SEM), this study conducts a cross-validation using a panel fixed-effects model. To ensure consistency with the SEM measurement system, all variables used in this test are constructed based on the SEM measurement model’s results. Specifically, standardized regression weights estimated by the SEM are applied to the normalized indicators to calculate composite indices for “low-carbon development” and the five dimensions of regional integration. Using these indices, the following two-way fixed-effects panel model is specified:

$$\begin{array}{ll}{LowCarbon}_{it}=& {\beta }_0+{\beta }_1{UrbanRural}_{it}+{\beta }_2{Infrastructure}_{it}+{\beta }_4{Ecology}_{it}\\ & +{\beta }_5{PublicaService}_{it}+{\mu }_i+{\lambda }_t+{\epsilon }_{it}\end{array}$$

(4)

where i and t denote city and year, respectively; LowCarbon represents the composite index of low-carbon development; UrbanRural, Innovation, Infrastructure, Ecology, and PublicService denote the composite scores of urban–rural integration, innovation coordination, infrastructure connectivity, ecological co-governance, and service sharing; ${\mu }_i$ and ${\lambda }_t$ capture city and time fixed effects; and ${\epsilon }_{it}$ is the random error term.

The robustness test results are shown in

Table 7. Robustness test results based on the panel fixed-effects model.

Variable	Panel Fixed-Effects Model		Baseline SEM Model
	Unstandardized Coefficient	Significance	Unstandardized Coefficient	Significance
Urban–Rural Integration	0.313 ***	Significant	1.973 ***	Significant
Innovation Coordination	0.077	Not Significant	−0.126	Not Significant
Infrastructure Connectivity	0.128 **	Significant	2.032 ***	Significant
Ecological Co-Governance	0.094	Not Significant	−0.358	Not Significant
Shared Services	0.210 ***	Significant	−2.677 ***	Significant

Note: *** and ** denote significance at the 1% and 5% levels, respectively.

. Overall, the estimates from the panel fixed-effects model largely align with those of the baseline SEM. Specifically, urban–rural integration shows a significant positive effect in both models, further confirming its robustness in promoting low-carbon development. Conversely, innovation coordination and ecological co-governance remain statistically insignificant, indicating that their direct impacts on low-carbon development are not yet apparent at this stage. Notably, the two approaches produce different results for infrastructure connectivity and service sharing, suggesting that their mechanisms of influence may be more complex or sensitive to model specification.

Specifically, the coefficient of “public service sharing” shows opposite signs in the two models: it is negative in the SEM but becomes significantly positive in the fixed-effects model. This difference likely arises from the distinct mechanisms each approach captures. The SEM highlights long-term structural relationships, where expanding public services and infrastructure often involve carbon-intensive construction processes, leading to a “carbon lock-in effect.” Conversely, the fixed-effects model focuses on short-term within-city variations, where improvements in public services can enhance human capital and social welfare, thereby promoting low-carbon transformation in the near term. This discrepancy underscores the complexity of the mechanism and suggests that the impacts of public service sharing can differ depending on the time scale.

Taken together, the main conclusion—that different aspects of regional integration have varying effects on low-carbon development—remains strong.

5.5. Discussion

The empirical results of this study generally align with recent development trends in the Yangtze River Delta. From 2009 to 2023, the positive impact of urban–rural integration on low-carbon development has become more apparent, mainly reflecting China’s progress in reducing the urban–rural income gap, enhancing rural energy consumption structures, and promoting new urbanization. The positive influence of infrastructure connectivity also highlights the region’s rapid progress in high-speed rail, intelligent transportation, and digital infrastructure, which are typically linked to a decline in regional carbon intensity. This finding aligns with existing studies that stress the important role of transportation and information infrastructure in lowering energy use and carbon emissions [39,40]. In contrast, innovation coordination and ecological co-governance do not show significant direct effects at this stage, indicating that their impacts may involve time lags and require a more extended period to become noticeable.

Compared to previous studies, this paper’s conclusions show some similarities, provide additional evidence to related research, and reveal specific differences. On one hand, the finding that infrastructure connectivity promotes low-carbon development generally aligns with earlier research emphasizing the positive impact of transportation networks, information infrastructure, and knowledge diffusion in improving energy efficiency and reducing emissions [39,40]. On the other hand, the insignificant direct effect of innovation coordination differs from existing studies, which typically highlight the importance of R&D investment, green patent output, and high-tech industries in supporting low-carbon transitions [33,34,35,36,37]. A likely explanation is that green technological innovation in the Yangtze River Delta is still at the intermediate stage of the “R&D–commercialization–application” process, with a noticeable lag in turning innovation results into measurable effects. Regarding ecological co-governance, the absence of a significant direct effect in our findings matches studies suggesting that the benefits of environmental management and ecosystem restoration often take longer to become evident [41,42,43]. Cross-jurisdictional ecological coordination may take even longer to become effective than governance within a single administrative area, primarily due to the greater need for institutional alignment and policy integration, which can lead to a delayed impact. Additionally, public service sharing may involve complex mechanisms, as previous research has shown its relevance to human capital development and green productivity improvements [44]. This may indicate that its effects could vary depending on the time frame considered.

Overall, the findings of this study seem to align with long-term trends of low-carbon development in the Yangtze River Delta and are generally consistent with and complementary to existing domestic and international research. These observations offer supporting evidence for the plausibility of the conclusions and provide valuable insights for future cross-regional and cross-country comparative studies.

6. Measurement and Analysis of Low-Carbon Development Levels in the Yangtze River Delta Urban Agglomeration

6.1. Construction of a Comprehensive Evaluation Model for Low-Carbon Development

Based on the empirical results of the structural equation model (SEM) (

Table 8. The standardized path coefficients and normalized coefficients of each variable.

Path			Standardized Path Coefficient	Normalized Coefficient
Low-Carbon Development	<---	Urban–Rural Integration	1.751 ***	1.867
Low-Carbon Development	<---	Innovation Coordination	−0.107	−0.114
Low-Carbon Development	<---	Infrastructure Connectivity	1.185 ***	1.263
Low-Carbon Development	<---	Ecological Co-Governance	−0.213	−0.227
Low-Carbon Development	<---	Shared Services	−1.678 ***	−1.789
Energy Intensity (Y1)	<---	Low-Carbon Development	0.939 ***	0.398
Forest Coverage Ratio (Y2)	<---	Low-Carbon Development	0.522 ***	0.221
Share of Non-fossil Energy Consumption (Y3)	<---	Low-Carbon Development	0.896 ***	0.380
Urban–rural Income Gap (X1)	<---	Urban–Rural Integration	0.847***	0.311
Per Capita GDP (X2)	<---	Urban–Rural Integration	0.977 ***	0.358
Urbanization Rate of Permanent Population (X3)	<---	Urban–Rural Integration	0.903 ***	0.331
R&D Investment Intensity (X4)	<---	Innovation Coordination	0.903 ***	0.449
Patent Authorization Rate (X5)	<---	Innovation Coordination	0.266 ***	0.132
Share of High-Tech Industry Output (X6)	<---	Innovation Coordination	0.843 ***	0.419
Highway Density (X7)	<---	Infrastructure Connectivity	0.701 ***	0.314
Internet Penetration Rate (X8)	<---	Infrastructure Connectivity	0.929 ***	0.416
Per Capita Volume of Postal (X9)	<---	Infrastructure Connectivity	0.603 ***	0.270
Concentration of PM10 (X10)	<---	Ecological Co-Governance	0.733***	0.385
Surface Water Quality Compliance Rate (X11)	<---	Ecological Co-Governance	0.578 ***	0.303
Green Coverage Ratio of Built-up Areas (X12)	<---	Ecological Co-Governance	0.594 ***	0.312
Per Capita Public Fiscal Expenditure (X13)	<---	Shared Services	0.980 ***	0.388
Number of Hospital Beds per 1000 People (X14)	<---	Shared Services	0.931 ***	0.369
Gross Enrollment Ratio in Upper Secondary Education (X15)	<---	Shared Services	0.615 ***	0.243

Note: *** indicates significance at the 1% level.

), this study uses both standardized and normalized path coefficients to develop a multidimensional weighted evaluation system for measuring the low-carbon development levels of cities in the Yangtze River Delta. The normalized regression coefficients are derived from the standardized path coefficients of the observed variables to their latent variables, along with the path coefficients among latent variables reported in Table 8 [54]. This transformation converts path effects with different units into comparable weights. The proposed system aims to overcome the limitations of traditional single-indicator measurements and provide a comprehensive view of the actual outcomes and internal driving forces of urban low-carbon transition from both performance-based and systemic perspectives.

The manifest development level indicates the direct results of low-carbon development. It is calculated from the weighted standardized factor loadings of three observed variables—energy intensity, forest coverage, and the proportion of non-fossil energy consumption. Essentially, it represents the composite score of the endogenous latent variable (η) in the SEM.

The systemic capacity level represents the regional systemic capacity that influences the low-carbon transition. Its calculation is based on the total effects, including both direct and indirect impacts, of the five integration dimensions on low-carbon development. Specifically, it is obtained by multiplying the standardized factor loadings of observed variables by the normalized path coefficients of their corresponding latent variables related to low-carbon development, and then summing these values. This measure indicates the overall force of multidimensional regional integration driving low-carbon development through complex mechanisms.

The calculation formula is as follows:

Manifest Development Level:

$$\eta =0.398Y_1+0.221Y_2+0.380Y_3$$

(5)

Systemic Capacity Level:

$$\begin{array}{c}\mathsf{\zeta }=0.273\times (0.311X_1+0.358X_2+0.331X_3)-0.114\times (0.449X_4+0.132+\\ 0.419X_6)+1.263\times (0.314X_7+0.416X_8+0.270X_9)-0.227\times (0.385X_{10}+\\ 0.303X_{11}+0.312X_{12})-1.789\times (0.388X_{13}+0.369X_{14}+0.615X_{15})\end{array}$$

(6)

6.2. Analysis of Low-Carbon Development Levels in the Yangtze River Delta Urban Agglomeration

Based on the empirical results of the structural equation model (SEM) (Table 8), this study evaluates the low-carbon development of 41 cities in the Yangtze River Delta in 2023 from two perspectives: the manifest development level and the systemic capacity level. Accordingly, a comprehensive ranking of urban low-carbon development is created (

Table 9. Measurement results of low-carbon development levels in 41 prefecture-level cities of the Yangtze River Delta, 2023.

City	η	Rank	ζ	Rank	City	η	Rank	ζ	Rank
Shanghai	0.969	14	1.190	3	Quzhou	0.761	33	1.055	9
Nanjing	0.984	6	0.604	38	Zhoushan	0.877	28	1.038	11
Wuxi	0.999	1	0.811	27	Taizhou (Zhejiang)	0.925	24	0.715	34
Xuzhou	0.980	9	0.470	39	Lishui	0.751	34	1.060	8
Changzhou	0.999	1	1.009	15	Hefei	0.987	4	1.124	4
Suzhou (Jiangsu)	0.981	7	1.014	14	Huaibei	0.900	25	0.831	24
Nantong	0.957	17	0.993	16	Bozhou	0.749	35	0.753	31
Lianyungang	0.968	15	0.984	17	Suzhou (Anhui)	0.977	12	0.785	30
Huaian	0.843	31	0.918	20	Bengbu	0.995	2	0.802	28
Yancheng	0.945	20	1.083	6	Fuyang	0.978	11	0.819	26
Yangzhou	0.999	1	0.704	35	Huainan	0.893	27	0.658	37
Zhenjiang	0.936	22	0.735	33	Chuzhou	0.994	3	0.746	32
Taizhou (Jiangsu)	0.999	1	1.034	12	Luan	0.894	26	1.028	13
Suqian	0.999	1	0.703	36	Maanshan	0.947	19	0.880	22
Hangzhou	0.949	18	1.077	7	Wuhu	0.935	23	0.970	18
Ningbo	0.739	36	0.953	19	Xuancheng	0.985	5	0.796	29
Wenzhou	0.974	13	1.043	10	Tongling	0.871	29	0.326	41
Jiaxing	0.980	8	1.469	1	Cizhou	0.979	10	0.877	23
Huzhou	0.866	30	1.099	5	Anqing	0.966	16	0.828	25
Shaoxing	0.726	37	1.342	2	Huangshan	0.941	21	0.368	40
Jinhua	0.778	32	0.888	21

). This analysis aims to reveal the different patterns and representative development models of regional low-carbon transition, providing empirical support for the targeted implementation of the “dual-carbon” strategy at the regional level.

Based on the measurement results in Table 9, the Yangtze River Delta (YRD) shows significant differences in low-carbon development. Jiangsu Province demonstrates overall strength in specific performance metrics, with seven of its cities ranked among the top ten. Wuxi, Changzhou, Taizhou, and Suqian jointly hold the first position (η = 0.999), reflecting their notable achievements in industrial emission reduction and green industry development. Zhejiang Province, on the other hand, excels in systemic momentum, with three cities ranked among the top five at the system level. Jiaxing (ζ = 1.469) and Shaoxing (ζ = 1.342) stand out because of proactive policies and innovation strategies. However, cities like Hangzhou and Jiaxing reveal a gap where systemic drivers have not been effectively converted into explicit performance, indicating possible bottlenecks in turning innovation capacity and systemic integration into measurable emission reductions. Shanghai, with a system level of ζ = 1.190, leads the region, highlighting its comprehensive advantages in economic low-carbon transition and ecological infrastructure as the regional core city.

Most cities in Anhui Province lag, with Bozhou and Suzhou performing poorly in both areas. Tongling (η = 0.871, ζ = 0.326), as a typical example of traditional industrial cities, faces dual challenges: reliance on energy-intensive industries and limited green innovation capacity within the measurement framework. Huangshan (η = 0.941, ζ = 0.368) exemplifies the common dilemma of ecologically advantaged regions: its relatively high explicit performance mainly depends on natural resources, but systemic momentum remains weak. This suggests that the transformation of ecological resources into green industries and sustainable drivers is not yet fully reflected in the current framework.

Furthermore, based on the coordination between explicit and systemic levels, four distinct development patterns can be identified, emphasizing the multi-path and complex nature of the low-carbon transition in the YRD.

(1)

Dual-high cities (e.g., Changzhou, Yancheng): Both explicit performance and systemic momentum are at advanced levels. This indicates a positive interaction among policy support, technological innovation, and emission reduction outcomes, highlighting the beneficial role of institutional design and market mechanisms in advancing coordinated low-carbon development. Consequently, they serve as regional benchmarks for collaborative green growth and provide replicable models for other cities.

(2)

System-driven cities (e.g., Jiaxing, Shaoxing): These cities show strong systemic momentum but relatively lag in explicit performance, indicating that their solid policy and innovation bases have not yet fully translated into measurable emission reductions during the observation period. This “conversion deficit” may arise from barriers in commercializing green technologies, limited industrial chain coordination, or a lack of market incentives, similar to the “green patent paradox” seen in some European innovation regions.

(3)

Performance-dominant cities (e.g., Wuxi, Suqian): These cities exhibit strong explicit performance but have relatively weak systemic momentum, indicating that their achievements may depend more on short-term administrative measures or external investments rather than on institutionalized and endogenous low-carbon capacity. The long-term viability of this model remains uncertain.

(4)

Dual-low cities (e.g., Tongling, Huangshan): These cities perform poorly in both explicit performance and systemic momentum, highlighting common challenges faced by traditional industrial cities and ecological function zones during the transition. They are hindered by structural bottlenecks such as a reliance on a single industry (e.g., Tongling’s dependence on metallurgy) and the limited conversion of ecological value into economic growth. Breakthroughs in these cities will require industrial restructuring and the promotion of eco-industrialization.

Overall, the classification shows that the low-carbon transition of YRD cities does not follow a single pathway but is influenced by the combined effects of policy effectiveness, industrial foundations, technological innovation capacity, and institutional coordination. Different types of cities display significant differences in transitional stages, driving forces, and development bottlenecks, providing empirical evidence for creating tailored regional emission-reduction policies.

It should be emphasized that conclusions regarding classification and path differentiation should be interpreted with caution. The absence of statistical significance for specific dimensions (e.g., innovation coordination and ecological co-governance) does not imply that these factors are structurally unimportant. Instead, this outcome might reflect methodological constraints and indicator limitations. For example, green innovation and ecological governance generally involve long-term and delayed processes, which are challenging to capture with the current indicator system fully. Therefore, the findings of this study are representative of the current stage, given existing data conditions, rather than dismissing the long-term significance of these factors.

7. Conclusions and Policy Recommendations

7.1. Conclusions

Based on an empirical study using panel data from 41 prefecture-level cities in the Yangtze River Delta urban cluster from 2009 to 2023, this research draws four main conclusions:

First, low-carbon development in the Yangtze River Delta is characterized by significant multidimensional heterogeneity and regional imbalance. Descriptive statistics and comprehensive assessments show substantial differences among cities in key indicators such as energy efficiency, clean energy adoption, economic growth, and innovation momentum. This systemic variation highlights the need for tailored and targeted regional low-carbon policies.

Second, different aspects of regional integration have varied effects on low-carbon development. Urban–rural integration and infrastructure connectivity show clear positive impacts, with infrastructure connectivity having the most substantial marginal effect. Conversely, public service sharing demonstrates a significant negative impact, suggesting a “carbon lock-in” caused by the growth of public service infrastructure. Meanwhile, the impacts of innovation coordination and ecological co-governance are not statistically significant, reflecting structural issues like long technology cycle times and limited coordination in regional environmental management.

Third, the study highlights significant regional differences and diverse development patterns in the Yangtze River Delta’s low-carbon transition. Jiangsu Province leads in overall development level, Zhejiang Province shows a stronger systemic capacity, while most cities in Anhui Province lag in both areas. Additionally, four typical development patterns—dual-high, system-driven, performance-dominant, and dual-low—identified through coordination analysis reveal systematic differences among cities regarding transition stages, driving forces, and development challenges.

Fourth, the multidimensional regional integration framework and the dual-dimensional explicit–systematic measurement model developed in this study provide new theoretical insights and methodological tools for understanding collaborative emission reduction mechanisms in metropolitan clusters. The empirical findings indicate that a successful low-carbon transition requires systemic coordination across various dimensions—including institutional design, technological innovation, infrastructure development, and environmental governance; however, relying solely on individual policy instruments or isolated dimensions is unlikely to produce significant breakthroughs.

7.2. Policy Recommendations

Drawing on the empirical results of structural equation modeling (SEM) and the spatio-temporal differentiation features of low-carbon development in the Yangtze River Delta urban cluster, this study offers policy suggestions across four areas—regional coordination, optimization of low-carbon pathways, targeted empowerment, and collaborative governance. These suggestions aim to solve the imbalances, path dependence, and institutional barriers faced in promoting low-carbon development within regional integration efforts.

First, develop specialized regional coordination mechanisms to address developmental imbalances. For cities with energy intensity significantly above the regional average (e.g., Tongling, Huangshan), an industrial energy efficiency benchmarking system should be established. This can be accomplished by requiring the phased removal of outdated production capacities and conducting clean production audits to accelerate low-carbon industrial restructuring. For cities with forest coverage below 30% (e.g., Suzhou [Jiangsu], Wuxi), dedicated ecological compensation funds should be created. These funds, supported by fiscal transfers, can finance afforestation initiatives and pilot programs in carbon sink trading, thereby boosting ecological carbon sequestration capacity. Building on the Yangtze River Delta Ecological Green Integration Demonstration Zone, a cross-provincial platform for factor mobility can be set up. Through this platform, market-based trading of key resources, such as energy consumption rights and carbon emission rights, can be encouraged, along with implementing a horizontal compensation mechanism where “high-carbon cities pay and low-carbon cities benefit”. Price signals can then guide optimal resource allocation and help reduce conflicts over emission reduction responsibilities caused by regional development disparities.

Secondly, enhance low-carbon transition pathways in key sectors and optimize infrastructure development. In urban–rural fringe areas, priority should be given to integrated green infrastructure projects, focusing on deploying distributed photovoltaic systems and smart microgrids. Mandatory technical specifications should require that renewable energy make up at least 40% of the supporting energy systems for newly built public service facilities, such as healthcare and education (based on annual electricity generation), to reduce carbon lock-in effects from public service expansion at the source. At the same time, promote the electrification of regional transport infrastructure by establishing clear timelines and technical standards for building charging networks at motorway service areas. Additionally, develop a unified regional energy efficiency standard system for industrial products to support the coordinated low-carbon development of regional infrastructure.

Thirdly, implement a targeted empowerment strategy for the “four categories of cities” to boost systemic development momentum. Support should be tailored to each city type’s specific traits. For dual-high cities (e.g., Changzhou, Yancheng), focus on advancing green technology industrialization by harnessing existing industrial bases to create application centers for hydrogen energy storage and carbon capture technologies. For system-driven cities (e.g., Jiaxing, Shaoxing), enhance green financial backing by establishing risk compensation funds to offer interest-subsidised loans for the new energy equipment supply chain, easing funding barriers for tech upgrades. For dual-low cities (e.g., Tongling, Huangshan), introduce ecological value conversion projects centered on forest carbon sinks, while strictly restricting high-energy-use projects. For performance-oriented cities (e.g., Wuxi, Suqian), strengthen industrial coupling by developing an assessment system for clean energy linkage in key sectors, promoting a shift from isolated emission cuts to coordinated systemic efforts.

Fourthly, establish a multi-tiered collaborative governance and performance evaluation system to provide institutional safeguards. First, set up a Joint Conference on Low-Carbon Development for the Three Provinces and One Municipality in the Yangtze River Delta. This conference will be responsible for coordinating cross-regional policies and standards, designing benefit compensation mechanisms, and arbitrating policy conflicts, thereby forming an institutional framework for safeguards. Second, develop a digital twin platform for urban low-carbon development. By integrating multi-source heterogeneous data—such as energy consumption, traffic flow, and ecological remote sensing—a city-level carbon metabolism simulation system can be built to enable dynamic monitoring of emission reduction progress, multi-dimensional simulation of policy effects, and intelligent recommendations for optimization pathways. Finally, incorporate key indicators, including the full life-cycle carbon emission intensity of public buildings and industry–energy interdependencies, into local governments’ green development performance assessment systems. This quantitative constraint approach will help drive the implementation of low-carbon development objectives.

Taken together, policy coordination across these four dimensions can create a dual-engine model driven by institutional frameworks and technological innovation. This model is expected to systematically address challenges like regional development fragmentation, path dependence, and fragmented governance, thereby supporting the effective transition of low-carbon development in the Yangtze River Delta urban agglomeration from planning to implementation.

In addition, it is important to highlight that the low-carbon transition experience and policy suggestions of the Yangtze River Delta (YRD) urban area also hold international significance. Compared to other integrated regions like the European Union (EU) and North America, the YRD’s efforts in cross-jurisdictional institutional coordination, regional carbon market development, and green infrastructure connectivity can serve as valuable references. For instance, although the EU has built a relatively mature carbon emissions trading system, it still encounters challenges in multilevel governance coordination; the cross-provincial horizontal compensation mechanisms of the YRD may provide valuable insights for designing institutions. North America excels in clean energy technology innovation and market development. Meanwhile, the YRD’s unique “four-type city” empowerment strategy offers inspiration for implementing precise low-carbon governance in urban areas with diverse structures. However, due to the substantial differences in development stages, governance structures, and energy systems across regions, it is essential to adapt these policy recommendations locally and contextually rather than simply copying them.

8. Research Limitations and Future Prospects

This study has several limitations that also suggest directions for future research. First, regarding research scope, the analysis concentrates on the Yangtze River Delta urban agglomeration. While this region is highly representative, the applicability of the findings to other areas—such as the Beijing–Tianjin–Hebei region or the Guangdong–Hong Kong–Macao Greater Bay Area—needs further validation. Future studies could expand the analytical framework to include urban agglomerations at different development stages or within diverse institutional contexts, and even conduct cross-national comparisons to improve the universality of the conclusions. Second, in terms of methodology, this paper uses structural equation modeling (SEM) for empirical analysis. Some standardized path coefficients seem relatively high, possibly due to strong correlations among latent variables or limited discriminant validity of specific indicators. Although current data constraints limit further model revisions, the overall model fit is acceptable, and the conclusions remain theoretically sound. Future research could broaden the measurement indicators (e.g., institutional governance and policy enforcement variables), explore hierarchical structural equation modeling (HSEM), employ dynamic panel models, or conduct additional robustness checks to enhance the reliability and explanatory power of the findings. Furthermore, potential issues like endogeneity, spatial autocorrelation, and placebo testing were not explicitly addressed in this study. Although a robustness check using a panel fixed-effects model partly addresses endogeneity concerns, future work could apply spatial econometric techniques (such as spatial Durbin or spatial error models), instrumental variable (IV) methods, or placebo tests to identify better the causal mechanisms by which regional integration impacts low-carbon development, thereby improving the robustness and generalizability of the conclusions.