Heterogeneity Analysis of Factors Influencing Carbon Emissions in the Yangtze River Basin: The Impact of National High-Quality Economic Development

Abstract

Evaluating the relationship between dynamic carbon emission intensity (CEI) and high-quality economic development (HQED) provides crucial insights for advancing national strategies focused on ecological preservation and sustainable high-quality development. This study employed an integrated analytical framework that combines the entropy-weight TOPSIS model, the coupling coordination degree (CCD) model, the spatial autocorrelation, and a two-way fixed effects model to examine the spatiotemporal patterns and influencing factors of carbon emissions in the Yangtze River Basin from 2010 to 2022. The results indicated that: (1) Temporal analysis revealed a consistent annual decline in CEI levels, coupled with steady improvements in HQED. The coordination between these two systems was reflected in the estimated CCD, and it showed an upward trend, with the lower reaches experiencing the most rapid progress in coordination. (2) Spatial analysis revealed a polycentric development pattern, with Shanghai serving as the central core, and other metropolises such as Nanjing and Hangzhou acting as secondary cores. The high–high agglomeration area has been progressively expanding each year. (3) Analysis of influencing factors revealed that their impacts diminished in the following order: human capital, economic development, urbanization, green innovation, government support, industrial structure, and openness. Each of these influencing factors demonstrated distinct spatiotemporal heterogeneity, varying in their impact across different regions and time periods. The study finally provided recommendations, emphasizing the need for coordinated development strategies in the YREB, taking regional dynamics into account, and promoting green economic transformations while ensuring ecological and environmental sustainability.

1. Introduction

The ongoing emission of greenhouse gases driven by human activities has generated multifaceted climate risks, posing a systemic threat to global sustainable development by intensifying societal vulnerabilities [1]. In response to this pressing challenge, China has committed to a binding target under the Paris Agreement, aiming to reduce carbon emissions by 60–65% relative to 2005 levels by 2030 [2]. The Yangtze River Economic Belt (YREB), a key engine of economic growth and a major coal and petroleum production hub in China, faces a profound structural dilemma. Its reliance on a high-pollution energy system perpetuates the resource curse, with persistent inefficiencies in energy consumption exacerbating environmental degradation [3]. At the same time, the region grapples with mounting challenges arising from unsustainable resource extraction practices [4]. Furthermore, the prevailing economic growth model in the YREB increasingly undermines urban regions’ capacity to withstand ecological and environmental pressures, threatening the long-term sustainability of urban development [5].

In September 2019, China launched a major national initiative aimed at enhancing ecological conservation while driving sustainable economic development in the YREB [6]. By optimizing geographic and ecological resources, the YREB has established a model of sustainable social and economic growth characterized by regional innovation and general synergy [7]. This progress is not only intended to elevate living standards but also to ensure a harmonious coexistence with the natural environment [7]. Therefore, exploring the intrinsic relationship between carbon emission intensity (CEI) and high-quality economic development (HQED) becomes crucial for the YREB. Promoting the coordinated development of CEI and HQED is not only a strategic necessity for advancing ecological civilization but also a fundamental aspect of ensuring regional sustainability. These efforts are essential to align economic growth with environmental preservation, paving the way for a more balanced and sustainable future for the YREB.

This study focused on the YREB agglomerations to explore the synergistic development between carbon emission intensity (CEI) and high-quality economic development (HQED). Faced with the dual pressures of tightening resource constraints and fragile ecosystems, a panel dataset encompassing 110 cities in the YREB, spanning from 2010 to 2022, was collected and pre-processed. Methodologically, this study introduces an integrated analytical framework, which systematically combines multiple research approaches. For instance, the entropy-weight TOPSIS model provided objective measurements of system development levels, while the coupled coordination degree (CCD) model quantified the synergistic relationship between the two systems of CEI and HQED. The kernel density estimation and spatial autocorrelation revealed the spatiotemporal evolution patterns, while the two-way fixed effects model with the heterogeneity analysis identified the driving mechanisms. This multifaceted approach facilitated a comprehensive investigation, from system measurement and relationship assessment to pattern recognition and causal inference.

This study might contribute to the field by establishing a comprehensive analytical framework for assessing ecological-economic coordination. Empirically, it provided novel insights into the spatiotemporal heterogeneity of driving factors across the YREB, moving beyond average effects to provide region-specific policy implications. However, the study did have some limitations. First, the existing carbon emission indicator system primarily focused on direct emissions from energy consumption and indirect emissions from electricity and heat production, overlooking supply chain emissions and those related to land-use changes. This oversight might lead to an underestimation of the full systemic impact of urban carbon footprints. Second, there was a potential risk of interpretation bias in the selection of driving factors, influenced by subjective judgment, which may compromise the robustness of the research findings.

To address the above limitations, researchers should explore and pursue several directions in the future. First, developing a comprehensive carbon accounting framework that can integrate both energy-related and non-energy-related emissions would provide a more complete understanding of carbon dynamics. Second, incorporating the influence of external factors, such as global economic trends and policy shifts, would strengthen the explanatory power of the analysis. Third, future studies could explore the advanced analytical techniques, such as geographically weighted regression models or spatial Durbin models, to examine more comprehensively the mechanisms through which HQED influences the CEI level.

The structure of this study was organized as follows: Section 2 reviewed the relevant literature on global carbon emissions and HQED. Section 3 described the study area, data sources, and research methods. Section 4 presented the temporal and spatial evolution characteristics of CEI, HQED, and their coupling coordination. Section 5 reported the empirical research results, including the influencing factors and robustness tests. Section 6 discussed the findings in relation to existing theories and studies. Section 7 concluded with policy recommendations and research limitations. Figure 1 illustrates the framework and technical roadmap of the research.

2. Literature Review

Amid the escalating global climate crisis, research on carbon reduction strategies has become a critical focal point, underscoring the urgent need to advance studies on low-emission technological solutions. As a pivotal player in the global carbon governance framework, China has significantly expanded its carbon emission research across various dimensions. This includes the analysis of spatial and temporal variations in carbon emissions [8,9], the quantification of underlying drivers through exponential decomposition and regression models [10,11], and the measurement and optimization of carbon emission efficiency across different regions and sectors [12,13]. In addition, some studies have examined the bidirectional relationship between carbon emissions and land-use transformations [14], while others have explored hidden carbon flows between industries using input-output analysis [15]. Furthermore, several reports have evaluated carbon emission levels in relation to national reduction targets, assessing the feasibility of achieving carbon neutrality [16].

Researchers have also identified significant spatial disparities in carbon emissions across China, with high-emission zones concentrated in eastern coastal urban agglomerations, contrasting sharply with the lower emissions observed in the western and inland regions [17,18]. However, the specific reasons behind these spatial differences in carbon emissions remain underexplored. This study focuses on the entire YREB region, conducting a spatiotemporal analysis of carbon emissions. By doing so, it aims to better identify and clarify the heterogeneity in emissions distribution and its influencing factors or drivers through the spatial nesting characteristics of the YREB’s economic-ecological complex system. Furthermore, key drivers of emissions have been identified, such as technological advancements [19], economic development [20], and the industrial scale [21], all of which collectively shape the dynamics of carbon output across the whole country. Moreover, the assessment of CEI reflects the coordinated development of both economic and ecological systems, with its application spanning key sectors such as industrial processes [22], power systems [23], and agricultural activities [24]. The research framework in this area exhibits three notable characteristics. First, in terms of content, CEI assessment has developed into a comprehensive logical progression, encompassing its measurement, evolution, and attribution. This includes CEI measurement techniques [25], the resolution and evaluation of spatial-temporal evolutionary patterns [26,27], and the identification of underlying driving mechanisms [28,29]. Second, at the spatial scale, CEI studies have primarily focused on global [30,31], national [32,33], and provincial carbon emission patterns [34,35]. Additionally, with respect to methodology, carbon emission quantification studies have primarily relied on the IPCC inventory methodology [36,37] and nighttime lighting data [38]. Finally, the decomposition of driving factors has often utilized the LMDI (Logarithmic Mean Divisia Index) decomposition model [39,40], while various models, such as the SDM (Spatial Durbin Model) framework [41] and the STIRPAT (Stochastic Impacts by Regression on Population, Affluence, and Technology) model [42] have been employed for influencing factor analysis and future scenario projections. The literature mentioned above has made significant progress in the measurement of carbon emissions, the identification of driving factors, and the assessment of efficiency. However, most studies continue to focus on single-system analysis, and others are confined to the macro scale of countries and provinces. There remains a lack of in-depth exploration into the complex coupling and coordination relationship between CEI and HQED at the urban level, as well as their spatiotemporal heterogeneity.

In October 2017, the Chinese government introduced the national strategy of high-quality development, marking a critical shift in socioeconomic development from rapid growth to a governance model centered on quality. This strategic framework aims to enhance both the developmental quality and efficiency through the coordinated progress of five core dimensions, i.e., economy, society, environment, innovation, and security [43]. Simultaneously, it is crucial to address the entrenched link between economic growth and fossil energy consumption in traditional development models, while establishing a new mechanism that fosters the harmonious coexistence of innovation-driven progress and ecological protection [44]. In the studies related to economic growth, scholars have traditionally relied on a single economic indicator to assess the quality of growth. However, this approach has notable limitations, such as insufficient spatial and temporal sensitivity, and limited value for decision-making [45]. Composite metrics, by contrast, can streamline the decision-making process by synthesizing complex, multidimensional information. Methodologically, other researchers have developed indices for economic growth quality, primarily derived from the integration of multiple indicators, such as urbanization, economic agglomeration [46], environmental quality and energy consumption [47]. However, the theoretical development and expansion of economic growth quality remain somewhat constrained compared to the broader concept of high-quality development. The comprehensive evaluation of HQED has been employed to assess developmental quality through multidimensional frameworks, surpassing singular economic growth metrics [48]. Its core principles encompass innovation, coordination, sustainability (greenness), openness, and inclusivity, with its essence lying in driving a structural leap in the economic and social systems through systemic institutional reforms [49,50]. This transition is reflected not only in the continuous expansion of economic output metrics but also in the transformative reconfiguration of modern development paradigms.

The studies on CEI (Carbon Emission Intensity) and HQED (High-Quality Economic Development) have now evolved into three primary strands. The first thread focuses on the interaction between carbon emissions and economic expansion. Rapid economic and social progress is inherently linked to environmental costs, particularly those stemming from high energy consumption and pollution. Building on the environmental Kuznets curve (EKC) framework, scholars introduced the carbon emission Kuznets curve hypothesis, with substantial empirical evidence supporting its validity [51]. However, various methodologies employed to test the carbon emission Kuznets curve have not yielded a unified conclusion. Lau et al. [52] proposed an inverted U-shaped relationship, while Alshehry and Belloumi [53] contested this hypothesis, arguing that carbon emissions do not follow the EKC framework. Moreover, Moomaw and Unruh [54] argued that the correlations between carbon emissions and income, as well as their isolated trends, are unreliable predictors of future behavioral patterns. The second key thread explores the systematic investigation of the interdependencies between carbon emissions and green development, with research confirming significant negative correlation mechanisms [55]. To operationalize a synergistic development pathway, systematic interventions are necessary, including the transition to clean energy and the adoption of cleaner production technologies. These measures enhance energy efficiency and facilitate compliance with carbon budgets within ecological capacity limits [56]. While carbon emission trading mechanisms effectively promote urban green transitions, they may inadvertently create barriers to technological diffusion due to inter-regional resource competition, potentially hindering coordinated development among neighboring cities [57]. Meanwhile, green finance has the potential to maximize the effectiveness of emission reductions by aligning closely with regional factor endowments [58]. The third area of focus is the frequent exploration of the co-evolutionary relationship between carbon emissions and social economic development. Several scholars have examined the coupling relationship between carbon emissions and economic growth from various perspectives, including energy consumption [59] and the carbon emission efficiency [60]. Other researchers have delved deeper into the interactive dynamics between carbon emissions and economic development, considering factors such as the new-type urbanization [61], the control of pollutants and pollution [62], and the digital economy [63]. Additionally, some studies have investigated the spatial and temporal evolution patterns of the coupling coordination degree (CCD) between economic growth and carbon emissions, employing integrated approaches that combine coupled coordination models with geographically weighted regression analysis [64].

While the above studies provided some valuable insights, several methodological and conceptual gaps persisted. First, existing studies using CCD models often lack systematic indicator weighting methods. They typically relied on equal weighting or subjective assignment, which may fail to fully capture the complex interdependencies within economic-environmental systems. Second, many reports employing entropy-based methods, such as entropy-weight TOPSIS, focus primarily on single-system evaluations without integrating spatial analysis and causal inference methods. Third, limited research has systematically combined entropy-weight TOPSIS, CCD, spatial autocorrelation, and fixed effects within a unified analytical framework. This study adopts an integrated methodology, applying entropy-weight TOPSIS to objectively determine weights in both the CEI and HQED systems, thereby addressing potential subjective bias in indicator weighting. By combining CCD with spatial autocorrelation, the study reveals both the degree of coordination and its spatially dependent distribution patterns. Additionally, the use of two-way fixed effects models allows for the identification of potential causal drivers and the spatial geographical heterogeneity of carbon emissions.

3. Materials and Methods

3.1. Study Area

The YREB, a crucial nexus connecting the national dual-carbon strategy with regional coordinated development initiatives (Figure 2), plays an essential role in fostering institutional innovations that emphasize ecological sustainability and green growth in China. Its evolutionary path serves as a significant research model for other regions in China. The study focused on three primary economic regions or zones in the YREB: the upper reaches, renowned for ecological barrier conservation; the middle reaches, defined by urban agglomeration development; and the lower reaches, a hub for innovation-driven leadership. This gradient of geographical and functional distinctions provided a unique perspective for further exploring the spatial coupling mechanism between CEI and HQED. Figure 2 illustrates the geographical boundaries of the region under investigation.

3.2. Data Sources and Processing

The dataset, covering the period from 2010 to 2022, primarily relied on available data and social development planning in China. The year 2010 marked the commencement of China’s 12th Five-Year Plan, which introduced significant policy shifts toward low-carbon development and the construction of an ecological civilization. The endpoint of 2022 represented the most recent year for which comprehensive city-level data were available for all required indicators at the time of data collection. This panel dataset provided sufficient temporal depth for robust spatial and temporal analyses, whereas it ensured the data quality and consistency across all indicators too.

3.2.1. CEI Variables

The CEI effectively captured the impact of economic scale interference, offering a more precise measure of the decoupling degree between economic growth and carbon emissions. This aligned with the need to analyze the non-linear relationship within the EKC framework. Due to data availability constraints in this study, it was important to note that the measurement of CEI variables primarily focused on the direct energy-based emissions, with the data sources retrieved from the China Carbon Accounting Database (CEDAs) [65]. While this approach addressed the core of urban carbon emissions, it did not account for carbon emissions from industrial processes, land-use changes, or supply chains, which might lead to an underestimation of total urban carbon footprints. This data was adopted to calculate the CEI variables and other relevant data for each city.

3.2.2. HQED Variables

The selection of HQED indicators was theoretically grounded in China’s national development paradigm shift, which transcended mere GDP growth to emphasize a multidimensional and sustainable approach. This framework, as outlined in national strategies, prioritized five core dimensions, i.e., innovation, coordination, greenness, openness, and sharing. The innovation dimension captured the natural transition from factor-driven to innovation-driven growth, which was essential for long-term competitiveness. Coordination focused on addressing internal structural imbalances, such as urban–rural disparities and industrial upgrading, which were crucial for stable and inclusive development. Greenness integrated environmental costs into economic systems, reflecting the imperative of ecological civilization. Openness recognized the role of global integration in facilitating the technology transfer and market competition. Lastly, sharing ensured that the benefits of development were broadly distributed, promoting social welfare and stability.

Drawing from previous reports [66,67], this study established a comprehensive evaluation framework covering the five key aspects of HQED in the YREB, i.e., innovation, coordination, greenness, openness, and sharing. The complete indicator system was shown and detailed in

Table 1. The evaluation index system of HQED.

Primary Indicators	Secondary Indicators	Description	Unit	Effect	Weight
Innovation	Education level expenditure	Education expenditure to the general expenditure of government finance	%	+	0.062
	Science and technology expenditure intensity	Science and technology spending to total government expenditure	%	+	0.010
	Human capital stock	Number of students enrolled in standard higher education institutions	persons/104 people	+	0.018
Coordination	Urban–rural income disparity	Urban to rural per capita disposable income	%	−	0.035
	Urban–rural consumption disparity	Urban to rural per capita consumption spending	%	−	0.026
	Consumption rate	Total retail sales of consumer goods to GDP	%	+	0.033
	Industrial structure upgrading	Value-added of the tertiary sector of GDP	%	+	0.026
	Employee structure	Share of the tertiary sector in the total workforce	%	+	0.056
Greenness	Atmospheric pollution intensity	SO2 emissions from industry per unit of GDP	tons/104 yuan	−	0.003
	Wastewater discharge intensity	Unit GDP industrial wastewater discharge	tons/104 yuan	−	0.013
	Built-up green coverage rate	Green coverage in built-up areas	%	+	0.020
	Waste treatment capacity	Domestic waste disposal rate	%	+	0.005
	Sewage treatment capacity	Centralized sewage treatment rate	%	+	0.019
Openness	Foreign capital dependence	FDI is actually utilized in GDP	%	+	0.284
	Trade openness	Total trade in goods of GDP	%	+	0.031
Sharing	Road area per capita	Total road area per capita	m2	+	0.120
	Green Park area	Green Park area per capita	m2	+	0.059
	Public cultural infrastructure	Public library collections per 104 people	item/104 people	+	0.030
	Medical resource sharing	Health facilities per 104 residents	units/104 people	+	0.151

. Notably, the relatively high weight of “Foreign capital dependence” (0.284) reflected its significant spatial-temporal variability across cities and its substantial informational contribution to the openness dimension. Similarly, the weight of “Medical resource sharing” (0.151) highlighted considerable regional disparities, making it a key differentiator in the sharing dimension. In contrast, indicators with lower weights, such as “Atmospheric pollution intensity” (0.003) and “Waste treatment capacity” (0.005), showed less variability across the focused cities and reported periods, contributing less information to the overall evaluation.

This was primarily attributed to the data-driven nature of the entropy weight method, which assigned higher weights to the indicators exhibiting greater variability across cities and over time as they provided more discriminatory power in the evaluation process. Conversely, indicators with very low weights typically showed minimal variations across spatial and temporal dimensions, thus contributing limited informational value to the composite index system.

3.2.3. Influencing Factors

From the dual perspectives of endogenous supply and exogenous drive, this study comprehensively considered multiple actors—including government, enterprises, and society, while it also integrated relevant research findings with the public data available. Seven core indicators were selected to systematically analyze the driving effects of the CCD variables on the CEI and HQED variables: RGDP [68], HM [69], IS [70], GI [71], GOV [72], UR [73] and OPEN [74].

Table 2. The influencing factors and explanatory variables of CCD.

Influence Level	Definition	Symbol	Variable Connotation	VIF Value
Internal supply	Economic development level	RGDP	Real GDP per capita	4.629
	Human capital	HM	General higher education students to the total population	1.720
	Industrial structure	IS	Industrial structure advanced index	1.702
	Technology	TEC	Technology innovation and application	1.657
External drivers	Green innovation	GI	Green patent grant/invention patent	1.092
	Government support	GOV	Government accounts generally pay out/gross regional product	2.030
	Urbanization	UR	Urban population/total population	6.031
	Opening up	OPEN	The actual amount of foreign capital utilized	2.233

presented the definitions and measurements of these influencing factors. To ensure the absence of multicollinearity, the variance inflation factor (VIF) was evaluated and estimated. All VIF values were found to be within the acceptable threshold (i.e., below 10, as shown in Table 2), confirming that multicollinearity was not a concern in the utilized models.

3.3. Research Methods

3.3.1. The Entropy-Weight TOPSIS Model

The entropy-weight method was employed for objective weight determination, leveraging its ability to quantify the informational utility of each indicator. This approach assigned different weights based on the degree of variation in the data—indicators exhibiting greater variability across cities and time periods are assigned higher weights, as they provided more discriminatory information for the evaluation. The method was grounded in information theory, where the entropy value measured the degree of disorder in the data distribution. Lower entropy signified more organized information, thereby leading to higher weight assignments. This data-driven method effectively eliminated the subjective bias in weight determination, ensuring that the evaluation results accurately reflected the actual disparities in development levels across the YREB cities [75,76,77].

The computational procedure followed three sequential steps.

Step 1: The weights of the indicators were calculated with the entropy weighting method:

$${x’}_{ij}=(x_{ij}-min(x_{ij}\left)\right)/\left(max\right(x_{ij})-min(x_{ij}\left)\right) \left(\mathrm{Positive\; indicator}\right)$$

(1)

$${x’}_{ij}=\left(max\right(x_{ij})-x_{ij})/\left(max\right(x_{ij})-min(x_{ij}\left)\right) \left(\mathrm{Negative\; indicator}\right)$$

(2)

$$w_j=(1-e_j)/\sum _{j=1}^n \left(1-e_j\right)$$

(3)

Note: $e_j={-\left(lnm\right)}^{-1}\sum _{i=1}^m p_{ij}ln\left(p_{ij}\right),$ $P_{ij}={x’}_{ij}/\sum _{i=1}^m {x’}_{ij}$.

Step 2: The study constructed the weight evaluation matrix $V$, and identify the positive ideal solution $V^{+}$ and the negative ideal solution $V^{-}$ according to the following method:

$$V=R^{’}\times W_j= (V_{ij}{)}_{n\times p}$$

(4)

$$V^{+}=\left\{max{V}_{ij}\right|j=\mathrm{1,2},3,\dots ,p\}$$

(5)

$$V^{-}=\left\{min{V}_{ij}\right|j=\mathrm{1,2},3,\dots ,p\}$$

(6)

Step 3: The study calculated the evaluation index according to the following method:

$$D_i^{+}=\sqrt{\sum _{j=1}^p (v_{ij}-V^{+}{)}^2}$$

(7)

$$D_i^{-}=\sqrt{\sum _{j=1}^p (v_{ij}-V^{-}{)}^2}$$

(8)

$$C_i=D_i^{-}/ (D_i^{+}+D_i^{-}) (0⩽C_i⩽1)$$

(9)

Here, $x_{ij}$ and ${x’}_{ij}$ denoted the pre-normalization and post-normalization values of the $i$th index in the $j$th city, while max ($x_{ij}$) and min ($x_{ij}$) meant the maximal and minimal values, respectively. Similarly, $w_j$ represented the computed weight for each indicator, $P_{ij}$ was the proportion of the value of the $i$th city under the $j$th indicator within the total value of the set indicator. In addition, $m$ indicated the total number of cities covered in the study, and $e_j$ corresponded to the entropy measure of $j$th indicator.

3.3.2. The Coupling Coordination Degree (CCD) Model

The coupling coordination theory is a well-established method for analyzing the interaction mechanisms between multiple systems. By assessing the degree of information and parameter interdependence across various modules, this theory reflects the level of coordinated development between systems. It quantifies the extent of energy exchange and parameter reliance between systems, offering valuable insights into their dynamic relationship [78].

In this study, the CCD model was utilized to quantify the synergistic relationship between the CEI and HQED systems. The theoretical motivation for coupling these two systems stemmed from the recognition that economic development and environmental pressure were not independent, and they existed in a complex, interactive relationship.

The CCD model advanced beyond simple correlation by measuring both the intensity of interaction (coupling degree, C) and the level of harmonious development (coordination degree, D) between the systems. A high CCD value indicated that reductions in carbon emission intensity were achieved in tandem with advancements in HQED, supporting regional sustainable development. This result aligned with the core objective of balancing economic growth with ecological constraints.

The formula for calculating the coupling coordination was developed as follows:

$$C=\sqrt[2]{{\eta }_1+{\eta }_2}/\left({\eta }_1+{\eta }_2\right)$$

(10)

$$T=\alpha {\eta }_1+\beta {\eta }_2$$

(11)

$$D=\sqrt{C\times T}$$

(12)

Here, ${\eta }_1$ and ${\eta }_2$ were the system indicators, while $C$ meant the coupling degree, and $T$ was the overall coordination. Similarly, $D$ represents the CCD value, whereas $\alpha$ and $\beta$ denoted the system contribution parameters. Under the framework of Chinese ecological civilization construction, the policy emphasis on the carbon emission reduction and high-quality economic development was balanced and equally important, this study adopted symmetric weighting with the parameter setting $\alpha =\beta =0.5$. Based on [79], the CCD was categorized into the following five stages in the study:

• Forced Coordination: 0 ≤ CCD ≤ 0.2
• Low Coordination: 0.2 < CCD ≤ 0.4
• Moderate Coordination: 0.4 < CCD ≤ 0.6
• High Coordination: 0.6 < CCD ≤ 0.8
• Extreme Coordination: 0.8 < CCD ≤ 1.0

The above five stages of CCD could help classify the observed level of urban long-term coordination between the CEI and HQED systems, from the minimal interaction to the complete alignment.

3.3.3. Spatial Autocorrelation

Spatial autocorrelation analysis quantitatively evaluated the similarity of attributes across neighboring spatial units through geostatistical modeling. This technique revealed the underlying spatial dependencies that shaped regional development patterns. It facilitated the identification of spatial clustering tendencies and heterogeneity gradients, offering empirical insights to guide spatially targeted policy interventions.

This study investigated the spatial autocorrelation characteristics of the CCD level between CEI and HQED by a spatial weighting method based on the K-nearest neighbors [80]. The K value was set to 4, signifying that each city was connected to its four nearest neighboring cities. This selection of K value aimed to create a balanced and stable spatial connectivity structure. A smaller K value (e.g., K = 2 or 3) might fail to adequately capture the spatial dependence in regions with lower urban density, whereas a larger K value (e.g., K = 5 or 6) could introduce the excessive spatial linkages. To assess the robustness of the results to the choice of K, sensitivity analyses were conducted using the alternative K values (K = 3 and K = 5). The trend and significance of the global Moran’s I remained consistent across different K values, indicating that the spatial dependence findings were not sensitive to the selection of K (

Table 3. Global spatial autocorrelation of CCD.

Years	K = 3			K = 4			K = 5
	Moran’s I	Z-Statistic	p-Value	Moran’s I	Z-Statistic	p-Value	Moran’s I	Z-Statistic	p-Value
2010	0.495	4.852	0.000	0.490	4.841	0.000	0.485	4.721	0.000
2011	0.409	6.112	0.000	0.404	6.086	0.000	0.401	5.942	0.000
2012	0.398	5.976	0.000	0.396	5.934	0.000	0.389	5.877	0.000
2013	0.355	5.254	0.000	0.351	5.199	0.000	0.344	5.137	0.000
2014	0.324	3.301	0.000	0.320	3.225	0.001	0.312	3.148	0.003
2015	0.164	2.425	0.010	0.157	2.351	0.019	0.149	2.276	0.020
2016	0.114	1.729	0.059	0.108	1.657	0.097	0.100	1.584	0.092
2017	0.144	2.167	0.033	0.138	2.098	0.036	0.130	2.028	0.043
2018	0.267	3.502	0.000	0.266	3.439	0.001	0.259	3.375	0.011
2019	0.247	3.568	0.000	0.240	3.511	0.000	0.233	3.453	0.000
2020	0.219	3.182	0.000	0.212	3.123	0.002	0.205	3.063	0.012
2021	0.208	3.061	0.000	0.201	3.005	0.003	0.194	2.948	0.012
2022	0.314	5.098	0.000	0.307	5.045	0.000	0.300	4.990	0.000

Note: * p < 0.10, ** p < 0.05, *** p < 0.01. Standard errors in brackets.

). This spatial weight matrix was applied and integrated with both the global and local Moran’s indices, with the results also being validated for statistical significance [80].

Global Moran’s I index:

$$I=\sum _{i=1}^n \sum _{j=1}^n W_{ij} (x_i-\overline{x})\left(x_j-\overline{x}\right)/\sum _{i=1}^n \sum _{j=1}^n W_{ij}\sum _{i=1}^n (x_i-\overline{x})$$

(13)

Local Moran’s I index:

$$I_i=[n (x_i-\overline{x})\sum _{j=1,j\ne i}^n W_{ij} (x_j-\overline{x})]/\sum _{i=1}^n (x_i-\overline{x}{)}^2$$

(14)

Here, $n$ represented the number of cities, while $x_i$ and ${ x}_j$ denoted the CCD level between region $i$ and region $j$. At the same time, $\overline{x}$ meant the average CCD level of all cities, whereas ${ W}_{ij}$ was the spatial weight matrix.

4. Results

4.1. Temporal Evolution Characteristics of the Targeted Variables

The temporal evolution analysis of CEI and HQED revealed distinct regional patterns across the YREB from 2010 to 2022 (Figure 3). The overall CEI for the entire YREB region declined from 0.868 in 2010 to 0.435 in 2022, with all sub-regions showing downward trends, though varying in magnitude and volatility. The upper reaches exhibited the most volatile evolution, yet maintained the highest overall CEI, experiencing a notable rebound of 37.31% in 2022. The middle reaches followed a consistent downward trajectory, with a sharp decline of 31.36% in 2022 compared to 2021. The lower reaches demonstrated the most significant carbon reduction effect, with CEI dropping from 0.705 to 0.284, reaching a record low in 2022.

The composite HQED score for the entire YREB region showed significant growth, increasing from 0.190 in 2011 to 0.266 in 2022, with an average annual growth rate of 3.10%. The lower reaches consistently outperformed in regional HQED, with a growth rate of 41.36% from 2010 to 2022 and an annual growth rate exceeding 10.50% after 2020. The middle reaches exhibited gradual growth, with the growth rate rising to 4.23% post-2020. The upper reaches showed steady progress, narrowing the gap with the middle reaches, although their score of 0.248 in 2022 still lagged behind the lower reaches’ score of 0.297.

4.2. Characterization of the Temporal Evolution of CCD

The temporal evolution characteristics of the YREB’s CCD were analyzed using kernel density estimation surface maps (Figure 4). In terms of distribution location and morphology, the kernel density curve exhibited a steady, gradual shift to the right, signaling a continuous upward trend in CCD. The curve’s shape evolved such that the main peaks consistently rose, and the coverage expanded, highlighting an increasing disparity in CCD across the region. Regarding distribution extension, the tail of the kernel density curve shifted from a left-tailed pattern at the start of the study to a right-tailed pattern by its conclusion, accompanied by a slight widening of the distribution. This right-tailed pattern suggested a growing concentration of cities reaching higher CCD levels, implying that CEI reduction and HQED improvement were becoming more synchronized across an increasing number of cities. Concerning polarization trend, the kernel density curve transitioned from an initial structure with a single highest peak in the early stages to a bimodal structure with both primary and secondary peaks. This shift indicated the emergence of divergent coordination patterns, with cities clustering into groups of either high or low CEI-HQED synchronization. The significant elevation disparity between the primary and secondary peaks in this bimodal distribution highlighted a tiered progression in CCD levels across urban agglomerations, characterized by varying gradients.

4.3. Analysis of the Spatial Correlation of CCD in the YREB

4.3.1. Characterization of the Spatial Evolution Patterns

To explore the spatial distribution patterns of CCD in CEI and HQED, the CCD data from the years 2010, 2014, 2018, and 2022 were estimated and compiled (Table 2, Figure 5). Over the study period, the regional coordination levels showed slight improvement, with most cities making significant progress toward higher coordination stages. However, spatial differences remained notable, giving rise to a trend of diversified development patterns. Shanghai emerged as the core, with lower-reach cities such as Nanjing and Hangzhou serving as sub-cores (Figure 5).

In 2010, most cities exhibited low coordination levels (Figure 6), with Lincang, Baoshan, and Lijiang reaching the forced coordination stage. However, the lower reaches showed a significant relative advantage, with a few cities advancing to the moderate coordination stage. By 2014, no cities remained in the forced coordination category (Figure 5, Figure 6). The lower reaches had experienced a notable breakthrough, with all but six cities (including Huainan and Huaibei) achieving moderate coordination. By 2018, the YREB had made substantial progress in overall CCD levels, with most urban areas advancing to moderate coordination. However, five cities, including Tongren and Zhaotong, remained exceptions to this trend. By 2022, the spatial differentiation became even more pronounced. Particularly, Shanghai had fully transitioned into a high-coordination region, further solidifying its position as the core growth pole of the region. Meanwhile, Nanjing, Hangzhou, and other provincial capitals in the lower reaches formed a sub-core cluster. As shown in Figure 5 and Figure 6, high-coordination cities were scattered in the dotted patterns across the developed areas of the lower reaches, while low-coordination cities remained concentrated in regions like Lincang, Lijiang, and Zhaotong. The coordination levels in underdeveloped areas showed a distinct spatial contraction trend.

4.3.2. Characterization of the Spatial Distribution Trends of CCD

To explore the spatial clustering patterns of CCD between CEI and HQED in the YREB, a global spatial autocorrelation analysis was conducted (Table 3). Moran’s I remained positive throughout all the studied years, and the results consistently passed the significance tests. These findings revealed a significant positive spatial autocorrelation for CCD, indicating the presence of the distinct clusters with high and low values. Notably, the global Moran’s I index steadily increased from 0.305 in 2010 to 0.339 in 2022, suggesting a strengthening pattern of spatial dependence over the observed period. In terms of dynamic trends, the global Moran’s I index exhibited a ‘W’-shaped evolution from 2010 to 2022, reflecting fluctuating upward and downward trends in the spatial autocorrelation of CCD.

To examine the variations in spatial patterns within local areas, we constructed LISA cluster maps (Figure 7). Overall, high-high clusters were predominantly located in the lower reaches, gradually expanding over time. In 2010, the lower reaches of the YREB gave rise to high-high clusters, encompassing major cities such as Shanghai, Suzhou, Wuxi, and Nanjing. Simultaneously, resource-based cities like Baoshan and Pu’er on the periphery were classified as low-low clusters, signaling the onset of regional polarization.

By 2014, high-high clusters extended to capital cities such as Hangzhou and Changsha. Meanwhile, cities like Guiyang and Zunyi joined the low-low clusters, and Nanchang transitioned from a high-high to a low-high cluster, highlighting the development gap in the middle reaches. Some cities like Bozhou and Fuyang in Anhui emerged as isolated high-low anomalies. By 2016, high-high clusters were predominantly dominated by economically robust cities like Shanghai, Suzhou, and Wuhan. Nanchang and Hengyang alternated between high-high and low-high clusters. In 2022, a discernible pattern re-emerged, with four economically advanced cities—Shanghai, Nanjing, Suzhou, and Wuxi—forming high-high clusters. Low-low clusters were primarily concentrated in underdeveloped regions, including Pu’er and Panzhihua in the upper reaches. However, Liupanshui no longer belonged to the low-low cluster, due to early signs of poverty alleviation, while Baoshan shifted from low-low to high-low. This indicated that enhancing regional coordination depended not only on addressing economic disparities but also on leveraging spatial spillover effects and implementing targeted policy interventions.

5. Empirical Validated Results

5.1. Results of the Benchmark Regression

To mitigate gauge effects and facilitate the comparison of different influencing factors, all explanatory variables were standardized using the Z-score method before being included in the two-way fixed effects model. The regression results (Model I) are presented in

Table 4. Results of the benchmark regression and robustness test.

Explanatory Variable	Model I	Model II	Model III	Model IV	Model V	Model VI	Model VII
RGDP	0.024 ***	0.025 ***	0.024 ***	0.024 ***	0.018 ***	0.027 ***	0.023 ***
	(0.005)	(0.005)	(0.005)	(0.004)	(0.005)	(0.006)	(0.009)
HM	0.035 ***	0.037 ***	0.036 ***	0.035 ***	0.039 ***	0.122 *	0.371 *
	(0.009)	(0.009)	(0.009)	(0.008)	(0.010)	(0.156)	(0.303)
IS	−0.006 ***	−0.006 ***	−0.006 ***	−0.006 ***	−0.010 ***	−0.014 **	−0.014 **
	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)	(0.007)	(0.007)
GI	0.009 ***	0.003 **	0.007 ***	0.009 ***	0.011 ***	0.000 **	0.000 **
	(0.001)	(0.002)	(0.002)	(0.001)	(0.002)	(0.000)	(0.000)
GOV	−0.007 ***	−0.006 ***	−0.006 ***	−0.007 ***	−0.005 **	−0.009 *	−0.049 *
	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)	(0.014)	(0.029)
UR	0.012 ***	0.012 ***	0.011 ***	0.012 ***	0.012 ***	0.088 ***	−0.024
	(0.003)	(0.003)	(0.003)	(0.003)	(0.003)	(0.020)	(0.037)
OPEN	−0.011 ***	−0.010 ***	−0.010 ***	−0.011 ***	−0.010 ***	−0.003 ***	−0.005 ***
	(0.002)	(0.002)	(0.002)	(0.002)	(0.002)	(0.001)	(0.002)
TEC		−0.016 ***
		(0.004)
_cons	0.430 ***	0.430 ***	0.430 ***	0.273 ***	0.436 ***	—	—
	(0.001)	(0.001)	(0.001)	(0.026)	(0.001)
Urban and time fixed effects	YES	YES	YES	YES	YES	YES	YES
N	1430	1430	1404	1430	1320	1430	1430
adj.R2	0.715	0.720	0.709	—	0.699	—	—

Notes: The values in parentheses are z statistics. * p < 0.1, ** p < 0.05, and *** p < 0.01.

. Analyses of RGDP, HM, IS, GI, GOV, UR, and OPEN revealed statistical significance at the 1% level. In terms of their effects, RGDP, HM, GI, and UR had positive impacts on the CCD, while IS, GOV, and OPEN exhibited significant negative impacts. Regarding the strength of these effects, HM had the most substantial positive impact on CCD, followed by RGDP, UR, GI, OPEN, GOV, and IS, in decreasing order of influence. These results emphasized that HM was a critical factor in promoting coupled and coordinated development between CEI and HQED.

The negative coefficient for the variable OPEN, though seemingly counterintuitive, could be explained by the specific industrial structure and development stage of the YREB. The region’s openness had historically been characterized by export-oriented manufacturing that remained energy-intensive and resource-dependent. Foreign direct investment and international trade in these sectors might reinforce existing carbon-intensive production patterns rather than facilitate the transfer of green technologies. Moreover, the “pollution haven” effect might be at play, where openness led to the concentration of pollution-intensive industries across the entire YREB region. This finding suggested that the quality and composition of openness, rather than its scale alone, might determine its impact on environmental-economic coordination.

5.2. Results of the Robustness Test

To ensure the robustness of the results, the relationship between the explanatory and dependent variables was analyzed by six distinct empirical methodologies (Table 4). First, recognizing that science and education expenditures directly influence clean technology innovation, which significantly contributes to GDP growth, Model II incorporated science and education expenditures as an additional explanatory variable in the regression analysis [81]. Second, due to the special developmental policy authorities held by Shanghai and Chongqing as national municipal administrative units, Model III excluded these cases to mitigate potential biases [82]. Furthermore, since the dependent variables were continuous but restricted, Model IV replaced the original model with the Tobit model to improve regression accuracy [83]. Moreover, acknowledging that the CCD between CEI and HQED could involve multiple influencing factors, Model V introduced a one-period lag in the explanatory variables to address potential endogeneity issues arising from bidirectional causality [84]. The results of Model V showed that lagged effects generally maintained the same direction and significance as the contemporaneous relationships, though with slightly attenuated coefficients. This pattern suggested that, while these factors had persistent impacts on CCD, their immediate effects were somewhat stronger than their lagged influences. The consistency across specifications enhanced confidence in the causal interpretation of the targeted relationships. Model VI and Model VII eliminated the influence of endogeneity through the direct and indirect effects of spatial spillover analysis. The robustness test results indicated that, except for the indirect effect of spatial spillover, all other regression results demonstrated strong robustness.

5.3. Heterogeneity Analysis

5.3.1. The Period Heterogeneity Characterization

In 2016, the YREB adopted a development strategy that prioritized ecological conservation and sustainable growth. Consequently, this study used 2016 as a temporal benchmark to divide the CCD between CEI and HQED into two distinct phases. Model VI revealed that during the first phase, the relative influence of each driver was ranked as follows: IS > HM > RGDP > GI > UR > OPEN > GOV. During this period, IS had a significant inhibitory effect, primarily driven by policy-induced capacity reductions and the downsizing of energy-intensive industries, which quickly lowered regional CEI. Meanwhile, HM remained constrained by population size and had not yet established a systematic support structure. Model VII showed that in the second phase, the relative influence shifted to the following order: HM > RGDP > GI > UR > GOV > IS > OPEN. As the development paradigm transitioned to innovation-driven growth, human capital, as the core enabler of innovation activities, gradually emerged as the central driving force.

5.3.2. The Regional Heterogeneity Characteristics

This study conducted subgroup analyses by categorizing YREB cities into the upper, middle, and lower reaches. The regression results for stratified samples were presented in

Table 5. The heterogeneity analysis of regression results.

Explanatory Variable	Different Periods		Different Regions
	Model VIII	Model IX	Model X	Model XI	Model XII
RGDP	0.021 ***	0.016 **	0.019 *	0.012	0.014 **
	(0.007)	(0.008)	(0.011)	(0.008)	(0.007)
HM	0.029	0.028 *	0.059 ***	0.015	0.006
	(0.025)	(0.015)	(0.017)	(0.011)	(0.017)
IS	−0.030 ***	−0.007 ***	−0.011 **	0.001	−0.010 ***
	(0.008)	(0.002)	(0.005)	(0.002)	(0.004)
GI	−0.010	0.016 ***	−0.000	0.006 **	0.010 ***
	(0.006)	(0.001)	(0.005)	(0.003)	(0.001)
GOV	0.000	0.008 **	−0.010 ***	−0.008	0.010 **
	(0.003)	(0.004)	(0.003)	(0.005)	(0.004)
UR	0.008	0.016 ***	0.012 **	0.023 ***	0.007
	(0.006)	(0.005)	(0.005)	(0.006)	(0.005)
OPEN	−0.001	−0.000	−0.011 ***	−0.040 ***	0.012 ***
	(0.004)	(0.002)	(0.003)	(0.005)	(0.004)
_cons	0.400 ***	0.442 ***	0.451 ***	0.447 ***	0.421 ***
	(0.006)	(0.004)	(0.010)	(0.003)	(0.010)
Urban and time fixed effects	YES	YES	YES	YES	YES
N	660	770	429	468	533
adj.R2	0.785	0.714	0.736	0.754	0.757

Notes: The values in parentheses are z statistics. * p < 0.1, ** p < 0.05, and *** p < 0.01.

(Models X, XI, and XII). The findings indicated that the factors influencing coupled and coordinated development exhibit regional heterogeneity (Table 5).

Notably, the estimated level of the variable RGDP had a more pronounced impact on promoting the development levels in the upper reaches compared to the lower reaches. The positive effect of the variable HM was significant only in the upper reaches, likely due to the relatively abundant human resources in the middle and lower reaches, which diminished the marginal contribution of human capital and weakened its impact. The inhibitory effect of the variable IS was primarily observed in the upper and middle reaches. In these regions, the decline of resource-based industries resulted in factor substitution costs, whereas in the lower reaches, the modernization of traditional manufacturing was hindered by path dependency.

The variable GI was not significant in the upper reach regions, suggesting that low-carbon technologies were insufficiently compatible with the region’s heavy chemical industry-dominated structure. The variable GOV demonstrated a significant inhibitory effect in the upper reaches, in contrast to findings from other periods and regions. This highlights the importance of aligning policy tools with market signals to ensure their relevance and effectiveness. However, the level of the variable UR had no significant impact in the lower reaches, revealing the latent costs associated with spatial development polarization. The variable OPEN exhibited a notable inhibitory effect across all regions, reflecting an over-reliance on resource-intensive, export-oriented industries, which could undermine both environmental and economic performance.

6. Discussion

The spatiotemporal evolution pattern of coupled and coordinated development (CCD) between CEI and HQED in the YREB revealed remarkable complexity. The coexistence of high coordination with significant spatial heterogeneity highlights the inherent contradictions in the transformation of regional development models. Over the research periods, CCD estimations consistently improved, yet regional differentiation became more pronounced. In all the YREB research regions, the growth rate of CCD in the lower reaches outpaced that of the middle and upper reaches, aligning with the theory of regional economic gradient development, which asserts that development is driven by differences in factor endowments. This spatial pattern presents a dual challenge to sustainable urban development in the YREB, marked by the coexistence of polarization effects and regional locking [85].

Spatial heterogeneity and the spatial spillover effects of carbon emissions or emission efficiency have been identified and explored in numerous reports. These spatial analyses have some significant socio-economic implications. In this study, the persistent “high-high” agglomeration of CCD in the lower reaches, centered on Shanghai and extending to provincial capitals like Nanjing and Hangzhou, underlined the synergistic benefits of economic agglomeration and knowledge spillovers in the more advanced regions [86]. The spatial analysis presented a polycentric development pattern, with Shanghai as the central core, supported by secondary cores, such as Nanjing and Hangzhou, alongside expanding the high-high agglomeration areas primarily concentrated in the lower reaches. The spatial evolution highlighted a clear trend of regional concentration in high-coordinated agglomerations. This trend aligns with the theoretical EKC framework, frequently employed to explore the relationship between economic development and environmental governance, assessing whether pollution first increases before being addressed in the regional socio-economic and environmental development with carbon consumption and carbon emissions [51,52,53,54]. This pattern also suggested that governance policies promoting city clusters and enhancing connectivity within core areas could amplify positive spillover effects. Conversely, the concentration of “low-low” clusters in less developed upper-reach cities, such as Pu’er and Zhaotong, emphasized the risk of being trapped in a vicious cycle of low development coordination [87]. Therefore, it was concluded that CEI may have spatial spillover effects correlated with HQED.

High-CEI regions and low-CEI regions might cluster in different spaces, consistent with the spatial distribution analysis within urban agglomerations [35,41,45,46,47,50,64,65,66]. This necessitates targeted regional assistance and policies, such as ecological compensation transfers and tailored green industrial support, to prevent widening regional disparities and ensure equitable development. The observed spatial autocorrelation and the expansion of high-value clusters indicate that the development path of one city is not isolated but influences its neighbors. Stronger inter-jurisdictional cooperation mechanisms, such as cross-administrative carbon markets, joint green technology R&D platforms, and coordinated regional environmental regulations, are necessary to internalize spatial spillovers and promote collective action towards regional sustainability goals.

From a socio-economic perspective, spatial divergence may reflect underlying inequalities in resource allocation, human capital mobility, and access to green financing. Policymakers must address these structural imbalances by directing investments towards human capital development and green infrastructure in lagging regions to break the spatial lock-in of low coordination and foster more inclusive, regionally balanced growth within the YREB.

The empirical analysis of this study largely aligns with theoretical expectations and findings from comparable reports, revealing some nuanced contextual specificities. The dominant positive influence of human capital (HM) on CCD strongly resonated with endogenous growth theory, which posits that knowledge accumulation and human capital are key drivers of long-term economic development and technological progress, capable of reconciling economic and environmental goals. This finding is consistent with prior research emphasizing the role of human capital in facilitating green innovation and structural transformation [71].

The significant positive effect of RGDP supports the notion that higher income levels provide the necessary financial resources for investing in cleaner technologies and environmental infrastructure, a premise central to many ecological modernization models. However, the negative coefficient for IS in the baseline model and certain sub-regions initially seemed counterintuitive. This could be interpreted within the distinctive context of the YREB’s transitional phase, potentially reflecting the short-term costs and efficiency losses associated with the aggressive restructuring and decommissioning of traditional, often energy-intensive industries, before newer, greener industries fully matured and optimized their productivity [88]. This phenomenon, commonly observed in other industrial transition regions [89], might be addressed by carbon trading reforms and improvements in industrial clustering strategies within YREB’s urban agglomerations.

Furthermore, OPEN could trigger the industry shifts that hindered the clean energy transition, potentially limiting the development of CCD, particularly in the economically active areas of the midstream region. The current heterogeneity analysis underscored the dynamic nature of these influencing factors, demonstrating their significance and possible fluctuating interplays.

7. Conclusions and Recommendations

This study employed an integrated analytical framework combining the entropy-weight TOPSIS model, the CCD model, spatial autocorrelation, and two-way fixed effects to explore the spatiotemporal patterns and influencing factors of carbon emissions across the YREB from 2010 to 2022. Based on the coordination and driving mechanism of the CEI and HQED coupling, the study reached the following key conclusions.

In terms of temporal evolution, the CEI level in the YREB exhibited a consistent annual decline, whereas the HQED level showed steady growth. The region-wide kernel density curve gradually shifted to the right, indicating progressive improvements in the CCD between CEI and HQED. Notably, the lower reaches experienced more rapid enhancements in CCD compared to the middle and upper reaches.

In terms of spatial evolution, the YREB displayed significant regional disparities in the CCD level between CEI and HQED. This resulted in diversified development patterns, with Shanghai emerging as the primary core, and cities in the lower reaches, such as Nanjing and Hangzhou, serving as secondary cores. The high-high agglomeration zones of CCD showed annual expansion, with a clear concentration in the lower reaches.

Concerning influencing factors, the impacts of the ranked variables—HM, RGDP, UR, OPEN, GI, GOV, and IS—gradually diminished in sequence. Heterogeneity analyses further revealed temporal and regional variations in these influencing factors. Notably, the effect of the variable HM was most significant in the upper reach regions and during the later study period, while IS had a greater influence in the earlier stages. The variable OPEN had a more pronounced inhibitory effect in the middle reaches, whereas the variable RGDP exhibited a stronger positive impact in the lower reaches.

Based on these research findings, three main policy recommendations were proposed to promote sustainable urban development in the YREB.

First, based on the functional positioning of the YREB, government sectors should formulate differentiated regional action plans. The local government in the downstream regions should focus on fostering original innovations in green technologies, establishing carbon neutrality laboratories in collaboration with high-level universities and research institutions, and cultivating green technology enterprise clusters with global competitiveness. The local government in the midstream regions should prioritize the green transformation of traditional industries, establish special funds for the circular transformation of industrial parks, and build regional industrial internet platforms to optimize energy dispatch. In the upstream regions, local governments should innovate mechanisms for realizing the value of ecological products, develop ecological label products and carbon sink projects, and prioritize the use of ecological engineering technologies and renewable energy in infrastructure upgrades.

Second, government sectors should establish a long-term mechanism for the coordinated development of human capital and green innovation. Local governments should set up a green skills training system that covers the entire YREB basin, with advanced green technology training centers in the downstream area. The government should advocate for industrial transformation skills training bases in the middle reaches and the professional development of ecological protection talent in the upstream regions. Furthermore, the government should assist in improving the joint training mechanisms for green technology talents, such as establishing the Yangtze River Green Innovation Fund, implementing a tiered reward system for green patents, and promoting the cross-regional universal and redeemable usage of science and technology innovation vouchers.

Third, government sectors should improve the carbon emission reduction system and optimize carbon emission reduction policies based on market mechanisms. Local governments should expand the coverage of the carbon market throughout the entire YREB basin to include all high energy-consuming industries, such as steel and cement. At the same time, local governments should explore and establish a cross-regional carbon quota adjustment mechanism and innovate green financial products. Government think tanks should promote an eco-environmental and development-oriented model to facilitate the coordinated trading and control of pollution discharge rights and carbon emission rights. Examples of this include establishing a green project database for the basin, implementing differentiated interest subsidies for green credit, and encouraging the issuance of blue bonds and transition finance bonds.

In summary, local governments should more emphasize the importance and urgence of coordinated development strategies across the entire YREB region, taking into account the regional dynamics and fostering the green economic transformations aligned with the national ecological and environmental sustainability.