Revisiting the Environmental Kuznets Curve: Does Economic Growth Necessarily Lead to More Carbon Emissions?

Abstract

Under the “dual carbon” strategy, clarifying the relationship between economic growth and carbon emissions and revealing the differences in green transition pathways among different urban tiers within the metropolitan area is of great significance for promoting regional low-carbon development. Based on panel data of prefecture-level cities in 27 national metropolitan areas in China from 2000 to 2020, this paper employs a two-way fixed effects model and a mediation effect model to test the Environmental Kuznets Curve (EKC) hypothesis and to evaluate the mediating role of industrial structure advancement. The results show that, at the national level, carbon emissions and economic growth exhibit a significant inverted U-shaped relationship, but the EKC becomes invalid in non-core cities after dividing the sample into core and non-core cities. Industrial structure advancement significantly curbs carbon emissions in core cities, while its effect is insignificant in non-core cities, indicating insufficient structural transformation capacity. The findings suggest that core cities have initially formed a “structure-embedded” emission reduction pathway, whereas non-core cities face a dual challenge of growth and emission reduction. In terms of policy, excessive reliance on the “automatic decoupling of growth” should be avoided, and a differentiated governance system centred on structural transformation capacity should be established, with particular attention to enhancing the green transition capacity of non-core cities so as to promote regionally equitable and coordinated low-carbon development.

1. Introduction

Amidst the new era of global pursuit of sustainable development, a long-standing and profound dilemma once again confronts cities: whether economic growth can be decoupled from its negative dependence on the environment. This issue is particularly evident in developing countries undergoing rapid industrialisation and urbanisation. Taking China as an example, since the initiation of reforms and opening-up in 1978, a wave of rapid urbanisation has swept across the country on an unprecedented scale, driving sustained and high-speed economic growth, while also bringing an equally unprecedented environmental burden. Over the past four decades, Chinese cities have not only created economic miracles that have drawn global attention but have also borne the world’s highest intensity of carbon emission pressure. Under this development model characterised by “high growth accompanied by high pollution”, whether cities can break through environmental constraints and achieve a “win–win” outcome of both economic growth and environmental improvement has already become a key issue urgently awaiting resolution in global research on urban sustainable development.

The Environmental Kuznets Curve (EKC) theory provides a persuasive classical framework for addressing this issue [1]. The EKC theory suggests that, in the early stages of economic growth, pollution emissions rise rapidly as a result of industrialisation and urbanisation; however, when the economy reaches a certain level of affluence, society gradually increases investment in environmental governance and promotes industrial structure advancement, eventually achieving the decoupling of environmental pollution from economic growth [2]. This theory has been empirically tested in numerous developed countries across the world and is widely regarded as both a theoretical basis and a practical blueprint for developing countries to achieve green transformation.

However, the real-world context is far more complex than that depicted by the classic theory. For China, with its vast territory and highly uneven regional development, the evolution of environmental pollution is not only determined by the economic development stage of a single city but is also profoundly influenced by the spatial structure and governance patterns among cities [3]. Against the backdrop of the “metropolitan area” being established as the core platform for promoting coordinated regional development at the national level, this structural characteristic becomes particularly prominent: differences in resource endowments, governance capacity, and industrial positioning among cities of different tiers may lead to significant divergences in the pathways linking economic growth and carbon emissions [4]. Consequently, a critical yet easily overlooked question arises—does the “growth–decoupling” mechanism predicted by the EKC hypothesis hold equally true for different types of cities within a metropolitan area? Do core and non-core cities truly share the same green transition pathway, or, under the banner of regional coordination, are the benefits of green growth concentrated mainly in core cities while the high-carbon burden is shifted to non-core cities? This question not only directly concerns the equity and sustainability of China’s “dual carbon” strategy but also provides an important opportunity to re-examine the applicability of the EKC hypothesis within a multi-level urban system.

This issue merits closer examination, as China’s urbanisation displays a pronounced ‘core–periphery’ structural feature at the spatial level: core cities concentrate high-end industries, policy resources, and green technologies, gradually achieving economic transformation and decoupling from emissions [5]. At the same time, a large number of non-core cities often passively receive industrial transfers from core cities, becoming concentrations of high-pollution, low value-added industries [6,7]. Under such a zonal structure, does the “growth–decoupling” mechanism, as emphasised by traditional EKC theory, still remain valid when applied to individual cities within this zonal structure? Or is the green transformation of core cities in fact built upon a concealed “carbon transfer” mechanism, where pollution is merely shifted to the peripheral areas of the metropolitan region?

Amid the deepening implementation of the “dual carbon” strategy and in the evolving context in which China’s metropolitan areas are tasked with pioneering regional sustainable development, it becomes imperative to move beyond the prevalent approach of treating individual cities as independent units of analysis when applying the EKC hypothesis at the urban scale. While such single-city analytical frameworks are common in nationwide city panel studies, they are inherently constrained within the highly interconnected spatial systems of metropolitan areas, as they risk neglecting critical mechanisms such as structural linkages between cities, patterns of industrial division of labour, and cross-regional carbon emission transfers. In response, this study adopts a more integrated, structurally informed, and dynamic analytical framework, re-examining the relationship between economic growth and carbon emissions through the lens of core–periphery heterogeneity within metropolitan areas, with the aim of revealing the deeper spatial structures and internal mechanisms that shape the green transition of Chinese cities.

2. Literature Review

Over the past decades, the EKC theory has become one of the core propositions in the field of sustainable development research, owing to its profound revelation of the complex and dynamic relationship between economic development and environmental pollution. Early research on the EKC was mostly conducted at the national or regional scale and widely verified the phenomenon of “economic growth–pollution decoupling” in developed economies during the late stages of industrialisation and urbanisation [8,9]. However, as research has deepened, the academic community has gradually come to realise that the traditional EKC framework, which uses national or provincial scales as units of analysis, may overlook the important impacts brought about by spatial structural differences within and between cities [10,11,12].

In recent years, an increasing number of studies have begun to focus on EKC relationships at the urban scale, particularly in the context of rapid urbanisation in emerging economies, where the differences in environmental pollution and economic development pathways across different types of cities have become increasingly evident. A large number of empirical studies have pointed out that large core cities often cross the EKC turning point earlier, showing a pattern in which pollution emissions first rise and then decline, while small and medium-sized peripheral cities generally show continuously rising pollution emissions [13,14,15]. This phenomenon is commonly interpreted as core cities being able to achieve industrial structure advancement earlier, accumulate more sufficient technological and capital resources, and possess stronger environmental governance capacity, whereas peripheral cities are relatively lagging behind and may face the risk of “pollution lock-in”.

Nevertheless, although existing literature has clearly identified significant differences between cities [16,17,18], most studies remain at the stage of static comparative analysis and have not systematically revealed the dynamic formative mechanisms and spatial evolutionary logic behind this concentric structure. In particular, for the spatial governance model based on the “metropolitan area” unit, the green transformation of core cities and the pollution accumulation in non-core cities may not be isolated phenomena, but rather the result of joint effects of implicit mechanisms such as deep regional division of labour, industrial relocation, and carbon emission transfer. However, academic research on this issue remains limited: most EKC studies have neglected the phenomenon of “carbon transfer” within metropolitan areas, namely the migration and flow of polluting industries between core and non-core cities; at the same time, relatively few studies have explored how industrial structure advancement, as a key mechanism, influences carbon emission trajectories within metropolitan areas at the spatial level.

Furthermore, although some studies have started to explore the intermediary mechanisms between economic growth and environmental decoupling—particularly the role of industrial structure advancement in emission reduction—most of them focus on the national or provincial scale and lack a dynamic analytical perspective on the concentric layers within metropolitan areas [19,20,21]. Existing studies tend to emphasise the independent realisation of industrial transformation or governance capacity enhancement by individual cities, while neglecting the evident interdependence and spatial interaction effects between the two types of cities. This deficiency at the level of research paradigms limits a systematic understanding of the uneven relationship between environmental governance and economic development within metropolitan areas and also weakens the practical value of the existing literature in informing policymaking.

In addition, in recent years, emerging spatial econometric and network analysis methods have begun to be applied in the field of carbon emissions, aiming to reveal implicit carbon transfer and pollution spillover effects between cities [22,23,24]. These studies have preliminarily illustrated the spatial association characteristics of carbon emissions but have not yet been effectively integrated into the classical EKC framework or into mediation effect analyses involving industrial structure advancement. This methodological fragmentation has meant that existing research still fails to fully answer the following key questions: to what extent is the green transformation of core cities within metropolitan areas accompanied by the expansion of polluting industries in non-core cities? And does the spatial structure of metropolitan areas aggravate the imbalance of regional emission reduction responsibilities and lead to environmental injustice?

In summary, three significant gaps remain in current research within this field. First, there is a lack of systematic theoretical explanation for the differences in EKC pathways between “core” and “non-core” cities within metropolitan areas. Second, there is an oversight of the spatial heterogeneity in the impact of industrial structure advancement on carbon emission pathways within metropolitan areas. Third, the carbon transfer mechanism has not been effectively integrated into the EKC theoretical framework to dynamically reveal the essence of regional economic–environmental interactions. It is precisely based on these gaps that this study seeks to propose a new research paradigm and empirical perspective, aiming to address the above-mentioned theoretical and methodological shortcomings and thereby provide a new analytical framework and theoretical perspective for understanding carbon emission differences within metropolitan areas.

Drawing on the existing literature, this study offers four principal contributions. In terms of theoretical advancement, it extends EKC analysis beyond a purely temporal perspective to encompass the core–periphery structure of metropolitan areas, thereby systematically investigating the structural divergence in the economic growth–carbon emissions nexus across different tiers and delineating the applicability boundaries of the EKC hypothesis within a multi-level urban system. With respect to mechanism identification, employing precise city-level carbon emissions data, the study demonstrates that industrial structure upgrading exerts a significant suppressive effect on carbon emissions; however, economic growth not only fails to facilitate such upgrading but, in fact, significantly inhibits it. This “reverse mediation chain” marks a critical departure from the conventional EKC logic: the structural mechanism underpinning the green transition is not inherently triggered by economic expansion but is functionally impaired under a growth-dominated paradigm. Methodologically, the study integrates industrial structure advancement into the EKC framework as a mediating variable and, by combining core–periphery comparative analyses, identifies the differentiated effects of structural upgrading across city types. This addresses a notable gap in the spatial heterogeneity analysis of industrial mechanisms, with multiple robustness tests conducted to verify the reliability of the findings. In terms of policy relevance, and in light of the “dual carbon” strategy and the imperative for coordinated metropolitan development, the study proposes differentiated governance approaches to address the “green core–carbon periphery” configuration, offering fresh empirical evidence and policy guidance to promote a regionally equitable and coordinated low-carbon transition.

Building on the above-mentioned research gaps and theoretical shortcomings, this study aims to move beyond the traditional EKC analytical framework by investigating the concentric heterogeneity between economic growth and carbon emissions among cities within China’s metropolitan areas. It seeks to reveal the spatial mechanisms jointly shaped by metropolitan circle structures and industrial transfer, as well as the mediation effect played by industrial structure advancement in this process. This study focuses on two specific core questions: First, does a typical Environmental Kuznets Curve (EKC) relationship exist between economic development and carbon emissions under the metropolitan circle hierarchy? And does this relationship exhibit clear differences across cities in different concentric layers—that is, do core and non-core cities present distinct EKC pathways? Second, does industrial structure advancement significantly affect the coupling relationship between economic growth and carbon emissions within metropolitan areas? And does the mediation effect of industrial structure advancement show significant concentric differences between core and non-core cities?

To effectively respond to the above research questions, the structure of this study is as follows: Chapter Two provides a review of the relevant literature, beginning with the theoretical underpinnings and subsequently considering their practical applications. Chapter Three outlines the research methodology, elaborating on the empirical strategy, data sources, and the logic of model construction. Chapter Four presents the empirical results, illustrating the characteristics of the relationship between economic development and carbon emissions among cities within metropolitan areas, from both an overall perspective and the perspective of concentric heterogeneity. It also examines in depth the mediation effect of industrial structure advancement. Chapter Five comprises the discussion section, offering theoretical interpretations of the above empirical findings and exploring the deeper logic of carbon transfer mechanisms and spatial imbalance. Chapter Six sets out the research conclusions, systematically reviewing the core findings, theoretical contributions, and policy recommendations, while identifying the study’s limitations and directions for future research.

3. Methodology

3.1. Study Area

This study adopts prefecture-level cities within China’s metropolitan areas as the basic analytical units. Specifically, with reference to the classification system proposed in the China Metropolitan Areas Development Report 2018 published by Tsinghua University, 27 national-level metropolitan areas are identified (as shown in Figure 1, with detailed city lists provided in Supporting Information Table S1). These include traditional core economic regions such as the Yangtze River Delta, Pearl River Delta, and Beijing–Tianjin–Hebei region, and also encompass emerging growth poles in central and western China, including the Chengdu–Chongqing, Wuhan, and Zhengzhou metropolitan areas, thereby demonstrating strong spatial representativeness and structural diversity. Based on the functional positioning and developmental hierarchy of cities within each metropolitan area, the report divides them into two categories: core cities and non-core cities.

Core cities—exemplified by Beijing, Shanghai, Guangzhou, and Chengdu—are typically the principal drivers within metropolitan areas. They not only exhibit a high degree of economic agglomeration and resource command but also possess marked advantages in institutional frameworks and governance structures. For instance, core cities frequently benefit from higher administrative status, more comprehensive planning and regulatory regimes, and stronger cross-sectoral coordination capacity. Their access to green finance, technological innovation, and public resource allocation is also markedly greater than that of non-core counterparts. These institutional and governance advantages enable core cities to advance industrial structure upgrading and low-carbon transition both earlier and more effectively.

By contrast, non-core cities are predominantly situated on the periphery of metropolitan areas. While they play a crucial role in accommodating industrial transfers from core cities and supporting metropolitan expansion, they are disadvantaged in governance capacity, access to green finance, and the allocation of policy resources. Their economies often remain reliant on traditional energy-intensive industries and resource-dependent activities, with limited investment in environmental governance. Consequently, their progress in industrial transformation and carbon emissions reduction is relatively slow, and they face heightened environmental pressures alongside an increased risk of pollution spillover.

The time span of this study extends from 2000 to 2020, covering a critical period of China’s urbanisation and deep industrial restructuring. This enables the examination of the long-term evolutionary paths of the EKC mechanism across different types of cities, as well as the institutional factors underlying these patterns. In the subsequent analysis, comparing core and non-core cities will not only help capture the characteristics of concentric heterogeneity but also provide a basis for understanding intra-metropolitan carbon transfer mechanisms.

3.2. Variables

Centred on the relationship between economic growth and carbon emissions, this study constructs a comprehensive variable system comprising the dependent variable, independent variables, mediating variable, and control variables, forming a balanced panel dataset at the prefecture-level city scale spanning 2000 to 2020. The data are primarily sourced from authoritative statistical yearbooks and publicly available datasets, ensuring consistency, comparability, and replicability of all indicators.

To measure the level of urban economic development, per capita gross domestic product (GDP per capita) is employed as the core explanatory variable. The data are obtained from the China City Statistical Yearbook and are smoothed using constant prices. To capture potential nonlinear relationships and empirically test the Environmental Kuznets Curve (EKC) hypothesis, the squared term of GDP per capita is additionally incorporated.

As the dependent variable, carbon emissions are derived from the Emissions Database for Global Atmospheric Research (EDGAR v8.0) published by the Joint Research Centre (JRC) of the European Commission (data source: https://edgar.jrc.ec.europa.eu/ accessed on 26 August 2025) [25]. EDGAR v8.0 provides annual estimates of global CO₂ emissions for the period 1970–2022, encompassing multiple emission sources such as energy combustion and industrial processes. The dataset is compiled using energy statistics from the International Energy Agency (IEA), the United Nations Statistics Division, and multi-source remote sensing data and is internationally recognised for its authority and cross-country comparability [26,27,28]. With a spatial resolution of 0.1° × 0.1°, EDGAR v8.0 enables multi-scale emission accounting and spatio-temporal analysis from the global to the regional level. In this study, the gridded data are converted to the prefecture-level city scale by means of spatial overlay and aggregation, based on the administrative boundaries of Chinese prefecture-level cities, yielding annual estimates of total carbon emissions for the period 2000–2020. It should be noted that, while EDGAR data have been widely adopted in the international literature and validated as reliable, their estimates may nonetheless be subject to uncertainties arising from the accuracy of underlying energy statistics and from spatial matching errors, which should be taken into account when interpreting the results.

In this study, the independent variable is the annual GDP per capita of prefecture-level cities, while the dependent variable is their annual total carbon emissions. Accordingly, the “economic growth–carbon emissions” relationship examined herein corresponds to a scenario of absolute decoupling, wherein total carbon emissions decrease or increase whilst the overall economy continues to expand or contract. Compared with relative decoupling, absolute decoupling provides a more direct indication of substantive progress in the low-carbon transition and carries greater policy relevance for evaluating the effectiveness of the “dual carbon” strategy.

To explore the underlying transmission mechanism between economic development and carbon emissions, industrial structure advancement is introduced as a mediating variable. Following the method proposed by Gan, Zheng, and Yu [29], this variable is calculated as the ratio of tertiary industry output to secondary industry output, effectively capturing the extent of structural transition from an industrial-based to a service-oriented economy. The calculation formula is as follows:

$$Adv={\displaystyle \frac{{Ind}_3}{{Ind}_2}}$$

where ${Ind}_3$ denotes the output value of the tertiary industry, and ${Ind}_2$ denotes the output value of the secondary industry. A higher value of this index indicates that the city’s industrial structure is more service-oriented, which is generally associated with greater potential for green development.

To control for potential confounding factors and omitted variable bias, this study further introduces several representative control variables covering multiple dimensions such as urban population, infrastructure, and ecological construction. All these variables are derived from the China City Statistical Yearbook. Descriptive statistics and units for each variable are provided in

Table 1. Descriptive analysis.

VarName	Definitions	Obs	Mean	p50	SD	Min	Max
Dependent Variables
CE	Annual carbon emission (106 t)	3073	32.30	22.55	33.60	0.668	288.2
ln CE		3073	3.021	3.116	1.018	−0.404	5.664
Independent Variables
GDP_P	GDP per capita (CNY/person)	3073	41,317	31,705	35,248	1892	468,000
ln GDP_P		3073	10.26	10.36	0.925	7.545	13.06
Mediation Variables
Adv	The ratio of tertiary to secondary industry output	3073	0.931	0.828	0.458	0.0953	5.297
Socioeconomic and public services variables
HP	Annual average house prices (CNY/m2)	3073	4457	3539	4260	375.4	56,000
ln HP		3073	8.104	8.172	0.762	5.928	10.93
Pop	Population (10,000 persons)	3073	469.9	408	341.2	69	3416
ln Pop		3073	5.961	6.011	0.630	4.234	8.136
Land	Built-up Area (km2)	3073	160.3	86	211.1	5	1565
ln Land		3073	4.574	4.454	0.943	1.609	7.356
Bus	Number of buses (unit)	3073	2489	906	4817	11	66,000
ln Bus		3073	6.941	6.809	1.298	2.398	11.10
Doc	Number of licenced (assistant) doctors (person)	3073	11,000	6864	11,000	68.70	120,000
ln Doc		3073	8.901	8.834	0.851	4.230	11.68
Book	Number of books (volume)	3073	377.6	130.9	908.6	2	18,000
ln Book		3073	5.059	4.874	1.169	0.693	9.797
Edu	Regulation Institutions of Higher Education (person)	3073	7002	2278	12,000	0	71,000
ln Edu		3073	7.676	7.731	1.907	0	11.17
NL	Natural Land (m2)	3073	5.8 × 109	3.0 × 109	7.1 × 109	2.0 × 107	4.6 × 1010
ln NL		3073	21.72	21.83	1.427	16.80	24.55
Ind Str3	Tertiary industrial as a percentage of GDP	3073	0.398	0.387	0.0976	0.0850	0.839

. All variables are annual frequency data, consistent with the panel analysis approach employed in this study.

3.3. Methods

To systematically identify the nonlinear relationship between urban economic development and carbon emissions, and to further examine the core–periphery differences within metropolitan areas and their mechanism pathways, this study adopts a staged and multi-model empirical strategy, constructing two-way fixed effects regression models and mediation effect models to rigorously identify the core propositions and test the underlying mechanisms. To preliminarily verify whether the EKC holds at the prefecture-level city scale, the following baseline regression models (1) and (2) are first constructed:

$$\ln \left({\mathrm{CE}}_{it}\right)=\mathsf{\alpha }+{\mathsf{\beta }}_1\ln \left({\mathrm{GDP}\_\mathrm{P}}_{it}\right)+{\gamma X}_{it}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(1)

$$\ln \left({\mathrm{CE}}_{it}\right)=\mathsf{\alpha }+{\mathsf{\beta }}_1\ln \left({\mathrm{GDP}\_\mathrm{P}}_{it}\right)+{\mathsf{\beta }}_2{\left[\ln \left({\mathrm{GDP}\_\mathrm{P}}_{it}\right)\right]}^2+{\gamma X}_{it}+{\mu }_i+{\delta }_t+{\epsilon }_{it}$$

(2)

where

${\mathrm{CE}}_{it}$ denotes the carbon emissions of city $i$ in year $t$;

${\mathrm{GDP}\_\mathrm{p}}_{it}$ represents the per capita gross domestic product, and ${\left({GDP\_p}_{it}\right)}^2$ is its squared term, used to capture the potential nonlinear inverted U-shaped relationship between the core variables;

$X_{it}$ denotes a vector of control variables;

${\mu }_i$ represents city fixed effects, controlling for time-invariant structural differences across cities;

${\delta }_t$ denotes year fixed effects;

$\epsilon$ is the error term.

To assess whether structural disparities exist in EKC pathways within metropolitan areas, this study further stratifies the prefecture-level city sample by concentric layers. Model (2) is estimated separately for core and non-core cities to examine whether systematic differences exist in the relationship between economic development and carbon emissions. By comparing the sign and significance of the coefficients on economic development variables across the two groups, the analysis identifies whether EKC patterns are concurrently present and whether their turning points diverge, thereby revealing concentric asymmetries in metropolitan green transformation.

Upon confirming the presence of a nonlinear EKC relationship, this study proceeds from a mechanistic perspective by introducing industrial structure advancement (Adv) as a mediating variable to examine its role in the ‘economic development–carbon emissions’ pathway. A classical mediation effect model is employed in two stages. The first stage tests whether economic development significantly affects industrial structure advancement, as specified in Equation (3):

$${\mathrm{Adv}}_{it}=\mathsf{\alpha }+{\theta }_1\ln \left({\mathrm{GDP}\_\mathrm{p}}_{it}\right)+{\theta }_2{{GDP\_p}^2}_{it}+{\mu }_i+{\lambda }_t+{\nu }_{it}$$

(3)

Subsequently, the mediating variable is incorporated into the carbon emissions regression model, as specified in Equation (4):

$$\ln \left({\mathrm{CE}}_{it}\right)={\alpha +\vartheta }_1\ln \left({\mathrm{GDP}\_\mathrm{P}}_{it}\right)+{\vartheta }_2{\left[\ln \left({\mathrm{GDP}\_\mathrm{P}}_{it}\right)\right]}^2+{\delta Adv}_{it}+{\gamma }_i+{\gamma }_t+{\epsilon }_{it}$$

(4)

If $\delta <0$, and the coefficients ${\vartheta }_1$ and ${\vartheta }_2$ exhibit significant attenuation after controlling for Adv, this would indicate that industrial structure advancement exerts a negative mediating effect in the ‘growth–emissions’ pathway—where economic growth promotes industrial restructuring, thereby indirectly suppressing carbon emissions. Furthermore, this mediation effect model is separately applied to the subsamples of core and non-core cities to explore concentric heterogeneity in the mediating mechanism. This approach helps elucidate why certain cities succeed in achieving green transformation, whereas others remain trapped in ‘carbon lock-in’.

4. Results

4.1. Spatiotemporal Dynamics of Core Variables

To deepen the understanding of the dynamic interrelations among economic development, carbon emissions, and industrial structure advancement within metropolitan areas, this section first analyses the spatial distribution patterns and temporal evolutionary trends of the core variables, aiming to identify the spatial structural features of green transformation at the city level.

Figure 2 presents the spatial evolution of GDP per capita in China. As shown in Figure 2a, urban economic development follows a typical “high in the east, low in the west” pattern, with coastal regions performing markedly better than central and western areas, and with a clear core–periphery gap within metropolitan areas, where core cities consistently outperform non-core cities. Benefiting from stronger factor agglomeration, more advanced infrastructure, and greater innovation capacity, core cities have maintained higher development levels than their non-core counterparts and have taken the lead in driving regional economic growth. Over time (Figure 2b–f), GDP per capita has continued to rise nationwide, yet the “core–periphery imbalance” remains evident: core cities have sustained stronger growth momentum, while some non-core cities have fallen behind, with regional disparities showing periodic widening. This reflects a long-term differentiation mechanism within metropolitan areas in terms of industrial upgrading and the inflow of capital and technology. This pattern is consistent with earlier studies on regional growth disparities [30,31,32] and suggests that intra-regional economic gradients are shaped not only by the distribution of material resources but also by institutional advantages, governance capacity, and regional innovation ecosystems. In the context of the green transition, such a configuration implies that core cities are more likely to reach the EKC turning point earlier, whereas non-core cities face greater pressure. This core–periphery disparity is therefore both an important consideration for policymaking and a structural basis for the later divergence of carbon emission trajectories and the emergence of carbon transfer mechanisms.

Figure 3 shows the spatial evolution of carbon emissions in prefecture-level cities. Nationwide, the distribution of carbon emissions largely mirrors that of economic development, displaying a “high in the east, low in the west” pattern, with high-value clusters concentrated in key economic zones such as the Capital Metropolitan Area, the Yangtze River Delta, and the Pearl River Delta. Over time (Figure 3a–f), although total national carbon emissions have continued to increase, their spatial distribution has shown a tendency to diffuse from core cities to surrounding areas, forming a “high-value ring” indicative of a spatial spillover effect. In some metropolitan areas (e.g., Beijing–Tianjin–Hebei, Chengdu–Chongqing), the carbon emission intensity of non-core cities has risen markedly. This may reflect the relocation of high-carbon activities driven by industrial migration and points to the structural features of environmental pressure redistribution within regions. This phenomenon is consistent with existing evidence on “carbon leakage” and the “pollution haven” effect [33,34,35], suggesting that while core cities implement local emission reduction measures, they may also shift certain high-carbon production activities to peripheral areas through industrial restructuring and land factor reallocation. The resulting “carbon transfer–periphery lock-in” pattern not only heightens environmental governance pressures in non-core cities but also poses challenges to achieving a coordinated regional green transition. These findings indicate that policy design should place greater emphasis on the equitable allocation of emission reduction responsibilities and on establishing inter-city linkages for industrial transformation.

Figure 4 presents the spatial distribution of the industrial structure advancement index. At the national level, the industrial structure is gradually evolving from being “industry-led” to “service-led” (Figure 4b–f). However, the core–periphery divide within metropolitan areas remains pronounced. Benefiting from a larger share of knowledge-intensive industries, well-developed innovation networks, and high-quality human capital, core cities have consistently maintained higher levels of industrial structure advancement and have taken the lead in the transition towards a green economy [36]. By contrast, many non-core cities continue to be dominated by the secondary sector, with relatively slow rates of transformation; in some cases, they even display industrial “lock-in,” making it difficult to overcome the path dependence of low value-added, high energy-consuming industries. This core–periphery pattern not only reflects the uneven flows of capital, technology, and high-end service resources within metropolitan areas but also closely mirrors disparities in the spatial distribution of carbon emissions. Core cities are better positioned to cross the EKC turning point earlier through industrial upgrading, whereas non-core cities face a “dual burden”: on the one hand, they receive inflows of high-carbon industries from other areas; on the other, they remain constrained by insufficient industrial upgrading. This situation highlights the need for green transition policies to promote coordinated upgrading in both core and non-core cities so as to avoid a “dual divergence” in regional economic development and environmental quality.

Overall, the three core variables exhibit pronounced spatial structural persistence and path dependence. Core cities, capitalising on their industrial bases and policy advantages, have spearheaded economic growth and structural transformation. In contrast, non-core cities—while absorbing economic spillovers—face dual pressures of rising carbon emissions and lagging structural upgrading, thereby further exacerbating spatial disparities in green development within metropolitan areas.

4.2. Overall and Core—Periphery Heterogeneity in EKC Pathways

Before presenting the regression results, this study conducted systematic pre-estimation diagnostic analyses to ensure the appropriateness of model specification and the robustness of empirical findings. First, Pearson correlation analysis was applied to panel data from all prefecture-level cities between 2000 and 2020 (see Supporting Information Table S1), revealing that most variables exhibit statistically significant positive correlations with carbon emissions, with the exception of natural land use. Second, to mitigate concerns regarding potential multicollinearity, Variance Inflation Factors (VIFs) were calculated for all explanatory variables (see Supporting Information Table S2). The results show that all VIF values fall below the empirical threshold of 10, with an average of 4.6, indicating that collinearity is within acceptable bounds and does not pose a substantial threat to model validity. Finally, F-tests and Hausman tests were conducted on the full sample to identify the optimal estimation approach (see Supporting Information Table S3). The F-test returned a p-value well below 0.01, leading to the rejection of the null hypothesis and confirming that the fixed-effects model outperforms the pooled OLS specification. Similarly, the Hausman test also yielded a p-value < 0.01, further supporting the use of fixed-effects estimation over the random-effects alternative.

The two-way fixed effects model regression results for the overall dataset of cities within the 27 metropolitan areas are presented in

Table 2. Benchmark regression results.

	(1)	(2)	(3)	(4)	(5)	(6)
	All Cities	Core Cities	Non-Core Cities	All Cities	Core Cities	Non-Core Cities
ln_GDP_Per	0.117 ***	0.170 ***	0.169 ***	0.429 ***	1.123 ***	0.126
	(0.021)	(0.037)	(0.028)	(0.108)	(0.257)	(0.127)
sq_GDP_Per				−0.016 ***	−0.046 ***	−0.002
				(0.005)	(0.012)	(0.006)
ln_HP	0.016	−0.207 ***	0.034	0.013	−0.174 ***	0.089 **
	(0.031)	(0.065)	(0.036)	(0.031)	(0.061)	(0.035)
ln_Pop	0.065	0.285 ***	0.020	0.082 **	0.285 ***	0.146 ***
	(0.043)	(0.080)	(0.060)	(0.042)	(0.079)	(0.044)
ln_Land	0.053 **	0.099 **	0.073 **	0.058 ***	0.076 *	0.059 **
	(0.021)	(0.042)	(0.030)	(0.021)	(0.042)	(0.023)
ln_Bus	0.003	−0.045	0.025 **	0.008	−0.037	0.008
	(0.012)	(0.036)	(0.010)	(0.012)	(0.037)	(0.013)
ln_Doc	0.131 ***	0.085 *	0.062 ***	0.134 ***	0.092 *	0.135 ***
	(0.022)	(0.050)	(0.010)	(0.022)	(0.052)	(0.022)
ln_Book	0.010	−0.017	−0.009	0.013	−0.001	0.005
	(0.008)	(0.013)	(0.010)	(0.008)	(0.012)	(0.009)
ln_Edu	0.020	0.034	0.010	0.017	0.016	0.006
	(0.016)	(0.048)	(0.007)	(0.016)	(0.045)	(0.017)
ln_NL	−0.159 **	−0.590 ***	−0.033	−0.205 ***	−0.686 ***	−0.177 **
	(0.071)	(0.156)	(0.073)	(0.072)	(0.164)	(0.080)
Ind_Str3	−0.401 ***	−0.837 ***	0.109	−0.351 ***	−0.768 ***	−0.259 **
	(0.099)	(0.229)	(0.130)	(0.101)	(0.221)	(0.116)
_cons	3.305 **	13.898 ***	0.355	2.594 *	10.864 ***	2.651
	(1.532)	(3.315)	(1.765)	(1.556)	(3.333)	(1.775)
Observations	2987	669	2404	2987	669	2318
R-squared	0.969	0.974	0.983	0.969	0.974	0.965

* p < 0.1, ** p < 0.05, *** p < 0.01.

. The full-sample results of Model (4) indicate that the level of economic development has a significant positive effect on carbon emissions (coefficient of the linear term = 0.429, p < 0.01), while the coefficient of the squared term is −0.016 (p < 0.01), confirming the existence of an inverted U-shaped relationship between carbon emissions and economic growth, that is, the EKC hypothesis holds at the overall level. This finding is consistent with EKC results in the economy–environment literature—including Zahra and Fatima [37], Li et al. [38], and other related studies—and provides important empirical support for understanding the overall green transition trajectory of Chinese cities.

However, this overall trend does not indicate that the EKC holds universal spatial applicability. Further analysis of core–periphery heterogeneity shows that core cities (Model 5) present a clear and significant inverted U-shaped relationship, suggesting that these cities are entering a “decoupling” phase in which economic growth and emission reduction are structurally linked to some extent. By contrast, the results for non-core cities reveal that, although the coefficient of the linear term is positive, it is not statistically significant, and the squared term is likewise insignificant. Combined with the findings of Model (3), this indicates that non-core cities remain in a coupled phase, where emissions continue to rise alongside economic growth, and no turning point in the green transition has yet emerged.

This core–periphery contrast underscores the asymmetric evolutionary mechanism of the green transition within metropolitan areas. Core cities, benefiting from relatively stronger policy capacity, high-tech industries, established clean technology foundations, and support from the service sector, are already moving into the stage of low-carbon transition. In contrast, non-core cities—long acting as destinations for high-carbon industries—are more likely to become locked into carbon-intensive development under the combined constraints of resource dependence, technological path dependence, and limited governance capacity. The spatial differentiation of the EKC essentially reflects a “structural mismatch”: while core cities, despite some fluctuations, are progressing towards a low-carbon economy, non-core cities remain on high-emission pathways, forming a distinct “green core–carbon periphery” configuration.

In summary, the EKC does not hold universally across China’s metropolitan areas. Its spatial heterogeneity exposes structural dimensions that are often overlooked in conventional EKC research. This finding not only challenges theoretical assumptions based on a single-city perspective but also provides empirical support for differentiated responsibility allocation, tiered management of emission reduction pathways, and regionally coordinated governance under the framework of the Dual Carbon Strategy.

To enhance the credibility of the model results and strengthen their explanatory power in terms of temporal logic, this study conducts robustness checks incorporating both one-period and two-period lag models of economic growth, as well as a cubic-term extended model (see Supporting Information Tables S4 and S5). This two-pronged approach not only helps to mitigate potential simultaneity bias and endogeneity concerns but also tests whether the EKC relationship holds under a more flexible nonlinear specification.

The lagged model results show that, whether using a one-period or two-period lag, GDP per capita continues to exert a highly significant positive effect on carbon emissions, the squared term remains significantly negative, and the signs and significance levels of the core variables are consistent with those in the baseline model. The model R² remains above 0.97, indicating that the inverted U-shaped relationship is robust to intertemporal testing. Building on this, when the cubic term is introduced, the nonlinear relationship remains valid for the national sample and for core cities, with only a very slight “N-shaped” tendency observed in core cities—suggesting a potential risk of a carbon emissions rebound at the high-income stage. The cubic term for non-core cities is insignificant, indicating that their economic–carbon emissions relationship is still predominantly characterised by a linear pattern. Therefore, neither the lag specifications nor the higher-order nonlinear extension alters the central conclusion regarding the inverted U-shaped EKC relationship. This finding not only confirms the long-term structural stability identified in this study but also reinforces the universality of its policy implications: core cities should focus on sustaining emission reductions at the high-income stage, while non-core cities should accelerate the peaking process to avert the risk of future rebound.

4.3. Mediation Mechanism: The “Decoupling” Pathway of Industrial Structure Advancement

To elucidate the underlying mechanism linking economic growth and carbon emissions, this study introduces industrial structure advancement (AIS) as a mediating variable, with the regression results reported in

Table 3. Mediation model regression results.

	(1)	(2)	(3)	(4)	(5)	(6)
	All Cities		Core Cities		Non-Core Cities
	AIS	ln_CE	AIS	ln_CE	AIS	ln_CE
ln_GDP_Per	−0.189 ***	0.103 ***	−0.084 **	0.160 ***	−0.195 ***	0.088 ***
	(0.024)	(0.021)	(0.043)	(0.037)	(0.027)	(0.025)
AIS		−0.073 ***		−0.120 ***		0.014
		(0.021)		(0.025)		(0.029)
ln_HP	0.077 ***	0.022	0.371 ***	−0.163 ***	−0.002	0.148 ***
	(0.025)	(0.031)	(0.077)	(0.062)	(0.024)	(0.045)
ln_Pop	0.057	0.070 *	0.166	0.305 ***	−0.131 **	0.059 **
	(0.041)	(0.042)	(0.104)	(0.079)	(0.052)	(0.023)
ln_Land	−0.005	0.052 **	−0.167 **	0.079 *	0.002	0.007
	(0.015)	(0.021)	(0.065)	(0.041)	(0.015)	(0.013)
ln_Bus	−0.021 **	0.002	−0.169 ***	−0.066 *	0.001	0.134 ***
	(0.009)	(0.012)	(0.046)	(0.037)	(0.008)	(0.022)
ln_Doc	0.020 *	0.132 ***	0.003	0.085 *	0.023 **	0.005
	(0.011)	(0.022)	(0.071)	(0.050)	(0.009)	(0.009)
ln_Book	−0.057 ***	0.006	−0.062 **	−0.024 *	−0.027 ***	0.006
	(0.009)	(0.008)	(0.025)	(0.013)	(0.008)	(0.017)
ln_Edu	−0.065 ***	0.015	−0.175	0.013	−0.037 ***	−0.173 **
	(0.013)	(0.016)	(0.119)	(0.039)	(0.012)	(0.079)
ln_NL	−0.018	−0.161 **	0.419 **	−0.539 ***	0.037	−0.295 **
	(0.051)	(0.071)	(0.182)	(0.151)	(0.053)	(0.134)
Ind_Str3	2.539 ***	−0.215 *	3.103 ***	−0.465 **	2.267 ***	0.208
	(0.148)	(0.111)	(0.325)	(0.233)	(0.150)	(0.182)
_cons	2.069	3.456 **	−8.311 **	12.902 ***	2.120	0.430
	(1.259)	(1.548)	(3.771)	(3.229)	(1.357)	(1.786)
Observations	2987	2987	669	669	2318	2318
R-squared	0.885	0.969	0.912	0.975	0.829	0.965

* p < 0.1, ** p < 0.05, *** p < 0.01.

. For the full sample, the regression coefficient of economic growth on industrial structure advancement is −0.189 (p < 0.01), which is significantly negative. This indicates that, during the study period, the overall growth trajectory of Chinese cities remained reliant on the secondary industry and had not yet undergone a general transition towards a service-led structure. This outcome is closely related to the construction of the indicator: when the growth rate of the secondary industry exceeds that of the tertiary industry, the AIS value declines—even if the increase in secondary industry output derives from low-energy sectors such as green manufacturing and high-tech industries. Accordingly, a negative coefficient does not necessarily imply a reversion to high-carbon dependence but rather reflects that, over the study period, a substantial share of China’s urban economic expansion continued to be underpinned by the secondary industry, particularly manufacturing.

With respect to the carbon emission effect, industrial structure advancement exerts a significant negative influence on carbon emissions, indicating that structural upgrading contributes, to some extent, to emission reduction. After introducing the mediating variable, the direct effect of economic growth on carbon emissions is lower than that observed without mediation, thereby confirming that industrial structure upgrading plays a partial mediating role in the “growth–emissions” relationship. This finding suggests that structural transformation constitutes an important pathway for mitigating urban carbon emissions, rather than relying solely on the “income decoupling” process associated with economic growth.

The heterogeneity analysis reveals pronounced hierarchical disparities within metropolitan areas. In core cities, economic growth exerts a significant inhibiting effect on industrial structure advancement, suggesting that these cities may already have achieved structural diversification, thereby reducing their dependence on traditional industrial configurations during the growth process. Specifically, core cities demonstrate that part of their economic expansion is realised through the greening of the secondary sector—namely, by relying on technological upgrading, high-tech investment, and industrial chain modernisation to transform traditional energy-intensive manufacturing into modern manufacturing characterised by low energy consumption and high value-added. Under this model, the suppressive effect of industrial structure upgrading on carbon emissions is more pronounced (coefficient = −0.120, p < 0.01), thereby establishing a relatively clear “growth–structure–emission reduction” mechanism along the green transition pathway.

By contrast, the sample of non-core cities exhibits a breakdown in this mechanism. Although their economic growth likewise significantly constrains industrial structure advancement, the regression coefficient of the mediating variable on carbon emissions is not statistically significant, indicating that these cities have yet to establish a stable structural channel for emission reduction. Non-core cities continue to lack high-end manufacturing and green technology support, with secondary-sector growth still dominated by energy-intensive and high-emission industries. Consequently, industrial upgrading has not yet been effectively converted into emission mitigation capacity. This absence of a structural mechanism also explains why, in the preceding EKC heterogeneity analysis, non-core cities have persistently remained in a “growth–carbon lock-in” state.

Overall, the mediation analysis indicates that industrial structure advancement serves as an effective channel for mitigating urban carbon emissions, yet its effectiveness varies markedly across city tiers. Core cities are developing a “structurally embedded emission reduction” model through pathways such as green manufacturing, whereas non-core cities, constrained by structural inertia and limited governance capacity, exhibit only weak mediation effects. This structural mismatch underscores the spatial imbalance in green development trajectories and further supports the central argument of this paper: disparities in carbon emissions within metropolitan areas are not only driven by uneven distributions of resources and technology but are more fundamentally rooted in tier-based differences in structural transformation capacity. Therefore, under the “dual-carbon” strategy, policy design should address both the industrial structure ratio and the greening level of the secondary sector so as to strengthen the functional role of non-core cities in the emission reduction chain and achieve balanced responsibility-sharing alongside coordinated mechanisms.

5. Discussion

5.1. Re-Evaluating the Applicability of the EKC Theory Within Metropolitan Structures

The Environmental Kuznets Curve (EKC) has long been regarded as a key analytical framework for examining the relationship between economic growth and environmental quality. Its core hypothesis posits that as income levels rise, pollution initially increases before subsequently declining, ultimately leading to environmental improvement. This model has been widely validated at the national and provincial scales, particularly in explaining the “pollute first, control later” trajectory observed in developing countries [39,40]. However, the hypothesis inherently treats a single city or country as the unit of analysis, overlooking systemic differences in resource endowments, industrial functions, and governance capacity within polycentric metropolitan areas. The empirical analysis in this study demonstrates that when the EKC framework is applied to China’s metropolitan regions—characterised by pronounced hierarchical spatial structures—its explanatory power is not only substantially diminished but also reveals a structural discontinuity.

At the national level, while the aggregate data can still be fitted to an inverted U-shaped curve between economic growth and carbon emissions, this pattern displays pronounced heterogeneity within urban agglomerations: core cities have generally entered the emission-reduction stage, whereas non-core cities continue along a “high-carbon growth” trajectory with no turning point in sight. This spatial misalignment is not a statistical artefact but rather the entrenched outcome of systemic disparities in structural capacity within urban agglomerations. Benefiting from stronger technological innovation and green investment capacity, a more diversified industrial structure, and higher governance efficiency, core cities are able to cross the emission peak threshold earlier; by contrast, non-core cities, burdened by the relocation of energy-intensive industries, the absence of channels for green technology diffusion, and insufficient institutional resource allocation, remain locked into a carbon-intensive development path [41,42,43].

This finding challenges the universality implied by the EKC’s stage-based hypothesis and raises doubts about its applicability within multi-tiered urban systems. The EKC’s unilinear temporal logic is insufficient to explain the phenomena of “turning point misalignment” and “prolonged high-carbon retention” observed among cities with differing functional roles. In other words, the emergence of a turning point is not an automatic outcome of economic growth but the result of the combined influence of structural position, industrial trajectory, and policy resources. Introducing this spatialised perspective underscores the importance of accounting for intra-hierarchical differentiation mechanisms when formulating EKC-based regional emission-reduction policies; otherwise, policymakers risk overestimating overall progress in the green transition at the macro level while neglecting the structural constraints faced by non-core cities.

More broadly, the findings of this study offer a cautionary note for EKC-based regional policy evaluation models. Traditional trajectory assessments often treat “whether the turning point has been reached” as the primary indicator of a city’s green transition progress, overlooking the questions of “which cities reach the turning point first” and “why certain cities may never reach it.” In this regard, the present study offers two important extensions to the EKC framework. First, within metropolitan regions, the “growth–pollution” relationship should be re-examined in the context of urban networks and functional divisions of labour, rather than extrapolated mechanically from the stage-based evolution of a single city. Second, the timing of the EKC turning point and the drivers of the green transition are determined more by differences in institutional capacity and industrial evolution pathways than by economic growth levels alone. This not only expands the intersection between environmental economics and urban–regional studies but also lays the theoretical groundwork for subsequent discussions on the mediating role of industrial structure advancement (see Section 4.2) and the issues of spatial justice and carbon transfer traps within hierarchical urban structures (see Section 4.3).

5.2. Why Does the Mediating Mechanism of Industrial Structure Upgrading Fail?

Within conventional EKC models, carbon emissions are theorised to follow an inverted U-shaped trajectory with respect to economic growth, wherein the “descending segment” of the curve is typically attributed to enhanced governance capacity, widespread diffusion of green technologies, and the advancement of industrial structure induced by rising income levels [44,45]. However, the mediation analysis in this study challenges this linear narrative of an “income-driven green transition”. Empirical evidence reveals that while industrial structure advancement significantly curbs carbon emissions, economic growth does not facilitate such advancement; rather, it significantly suppresses it, thereby indirectly undermining urban decarbonisation capacity. This pathway indicates a critical departure from the EKC logic: the advancement-oriented structural transformation mechanism that should underpin the green transition is not activated by economic expansion but instead experiences functional impairment under a growth-dominated paradigm. Consequently, the observed reductions in carbon emissions within China’s metropolitan areas should not be interpreted as “spillover dividends” arising from income growth but rather as “governance dividends” generated through deliberate advancement of the industrial structure and targeted structural adjustments.

Specifically, in the full sample, economic growth significantly inhibits the advancement of industrial structure, while industrial structure advancement in turn exerts a significant negative effect on carbon emissions. This forms a typical case of a suppressed mediation mechanism, characterised by a negative X→M path and a negative M→Y path. This outcome does not deny the emission-reduction potential of structural transformation; rather, it reveals an institutional paradox: although cities possess viable structural pathways for decarbonisation, these mechanisms are constrained by prevailing growth-led trajectories, depriving them of institutional realisation. In the context of rapid urbanisation, local governments frequently rely on manufacturing investment to sustain short-term economic momentum, thereby increasing the share of secondary industry and delaying the transition toward service-led, low-carbon development. This indicates that carbon mitigation capacity is not an automatic by-product of economic growth but rather a function of whether structural governance can be effectively enacted.

This structural mechanism breakdown reveals clear differentiation across metropolitan tiers. Although core cities do not exhibit a statistically significant pathway whereby economic growth promotes the advancement of industrial structure, both the suppressive effect of economic growth on industrial structure advancement and the inhibitory effect of such advancement on carbon emissions are more pronounced. This indicates that core cities have developed a relatively mature structure-led green transition mechanism. The existence of this mechanism is underpinned by their service-dominated industrial base, strong agglomeration capacity in high value-added industrial sectors, and a superior ability to integrate governance resources, enabling the structural pathway to sustain a certain degree of autonomous transformation momentum even in the absence of direct growth-driven impetus [46]. By contrast, non-core cities face a dual mechanism failure: economic growth significantly suppresses structural advancement, while the latter exerts no statistically significant effect on emissions. In practice, the mediating mechanism thus operates in a state of functional stagnation. This indicates that non-core cities have yet to establish a viable structural pathway for emission mitigation. Worse still, owing to their role as industrial recipients, technological inertia, and fiscal incentive distortions, they remain locked into carbon-intensive trajectories and unable to escape the “growth–emissions coupling” trap. Crucially, this institutional suppression of structural potential is not incidental. It reflects the intrinsic imbalance embedded in China’s growth pole–driven model of metropolitan development.

Extending this mechanistic phenomenon further, international urban research reveals similarly paradoxical structural patterns. Numerous cities across the Global South have encountered the so-called “service sector paradox,” whereby service sector expansion does not inherently lead to emission reductions. Instead, it often intensifies carbon pressures due to low value-added outputs, high energy intensity, or limited green governance capacities [47,48,49]. This indicates that “industrial structure” and “structural capacity” are by no means synonymous; the true driver of green transition lies not in the superficial alteration of industrial composition, but in the underlying configuration of institutional resources and the exercise of structural power. Accordingly, in the absence of a concurrent enhancement of both structural capacity and governance authority, the mechanism of industrial upgrading—though nominally present—may fail to deliver, or even be counteracted, in terms of its contribution to carbon emission reduction.

5.3. Spatial Justice and the “Green Centre–Carbon Periphery” Trap

If the EKC theory offers a developmental logic for interpreting urban carbon emission trajectories, then the heterogeneity of mediating mechanisms and structural ruptures revealed in this study expose a long-neglected spatial reality: the responsibilities and capacities for green transition are not evenly distributed within metropolitan regions. Instead, they are structurally concentrated in core cities, while non-core cities are marginalised and rendered voiceless within dominant decarbonisation discourses. This tiered asymmetry in the distribution of green capabilities marks a transformation paradox that warrants serious critical attention—namely, the emergence of a “green centre–carbon periphery”.

The mechanism analysis in this study indicates that core cities are progressing through the EKC process, suggesting that they have established a relatively effective emission-reduction pathway underpinned by structural mechanisms. In contrast, non-core cities are locked into a “high-growth–high-emission” trajectory, trapped within a dual failure of mechanisms—namely, “growth-induced structural suppression” and “structural failure in emission reduction”. More critically, this structural divergence is not merely the by-product of differing developmental stages but the outcome of the long-term accumulation of metropolitan power structures, resource allocation patterns, and institutional biases. In the distribution of key resources—such as green technologies, environmental protection investment, industrial support, and public finance—core cities hold a dominant position, while non-core cities primarily serve as recipients of the spillover from high-carbon industries and, in some cases, are incorporated into a “green transition outsourcing system”. This resulting “green core–carbon periphery” configuration thus reflects not only asymmetries in developmental capacity but also asymmetries in the mechanisms for sharing governance responsibility.

This structural configuration constitutes a fundamental challenge to equitable urban development from a spatial justice perspective. Under the urban green governance principle of “collaborative governance and shared benefits”, non-core cities are frequently mandated to shoulder decarbonisation responsibilities that mismatch their capacities, whilst lacking the requisite industrial foundation, fiscal resources, and technological conditions to fulfil these obligations. This state of “burden-capacity mismatch” not only constricts their space for autonomous development but also intensifies internal differentiation within regional development. Consequently, non-core cities find themselves within the ecological civilisation discourse as entities possessing “neither access to green benefits nor the capacity to decarbonise”. Conversely, core cities, by virtue of their higher emission reduction efficiency and earlier structural optimisation, accrue additional policy advantages and enhanced moral legitimacy. This further consolidates their positional advantage in green capital accumulation.

Viewed through an international comparative lens, this phenomenon of “ecological displacement” is not an isolated occurrence within polycentric metropolitan regions. In multipolar nations such as those in Europe and North America, analogous instances of “carbon responsibility spillover” are increasingly garnering attention. For example: the spillover of manufacturing industries from the London–South England metropolitan region to the Midlands [50]; the shift in manufacturing segments to smaller southern cities accompanying the service-oriented transformation of the Boston Metropolitan Area, USA [51]. Both cases manifest the latent ecological inequality inherent in the “green centre–carbon periphery” dynamic. This “institutionalised carbon transfer” not only challenges the efficacy of regional governance but also raises profound questions regarding the ethical foundations of low-carbon metrics: When a city’s carbon performance is predicated upon the elevated emissions of another, can its “green achievements” retain genuine legitimacy?

Consequently, this study posits that the “green centre–carbon periphery” constitutes not merely a spatial phenomenon of environmental performance, but fundamentally, an issue of power configuration within environmental governance. Against the backdrop of the nation’s deepening “dual carbon strategy”, establishing more equitable, feasible, and quantifiable burden-sharing mechanisms for emission reduction within metropolitan regions emerges as a pivotal issue for advancing regional collaborative governance and achieving spatial justice. Such mechanisms should encompass instruments including: green fiscal transfer payments; collaborative industrial transition funds; cross-jurisdictional carbon trading schemes; and differentiated target-setting based on tier-specific capacities. Only when the formulation of emission reduction targets genuinely reflects cities’ structural positioning, developmental capabilities, and historical responsibilities can the green transition evolve into a truly shared just cause across the entire region—rather than representing the glory of some cities achieved at the expense of others.

6. Conclusions and Policy Recommendations

This study takes 27 prefecture-level cities within China’s metropolitan regions as the empirical sample and, from the perspective of metropolitan tier structures, systematically examines the applicability and structural mechanisms of the EKC under the context of rapid urbanisation and regional integration. It aims to address two core questions: first, does an inverted U-shaped EKC relationship between urban economic growth and carbon emissions generally exist? Second, does industrial structure advancement play a mediating role in this relationship, and does this mechanism exhibit heterogeneity across different metropolitan tiers?

To explore these questions, the paper develops a theoretical mechanism model of “economic growth—structural upgrading—carbon emissions”, employing prefecture-level panel data from 2000 to 2020. Using a two-way fixed effects model combined with between-group mediation effect identification, it proposes an extended EKC analytical framework embedded within the spatial tiered structure of metropolitan regions. The principal research findings can be summarised in three points.

First, the EKC relationship receives statistical support at the level of the overall sample, with economic growth and carbon emissions exhibiting a significant inverted U-shaped relationship, thereby confirming the potential for a “growth–emission reduction” transition pathway at the aggregate scale. However, heterogeneity analysis reveals that this pathway is far from universally present: core cities display a clear EKC structure, whereas non-core cities remain in a stage where “carbon emissions continue to rise with growth”, indicating a breakdown in the applicability of the traditional EKC framework within the internal structure of metropolitan regions.

Second, the mediation mechanism analysis uncovers a “reverse mediation chain” that departs markedly from the classical EKC logic: although industrial structure advancement significantly reduces carbon emissions—thus affirming the central role of structural transformation in green development—economic growth not only fails to activate this structural mechanism but significantly suppresses it, thereby weakening urban decarbonisation potential under a growth-dominant paradigm. This finding suggests that the key driver of green transition lies not solely in the elevation of economic levels, but in the establishment of a stable and autonomous foundation of structural capacity.

Third, the structural mechanism exhibits pronounced tier-based heterogeneity: while economic growth in core cities significantly suppresses structural upgrading, these cities nonetheless possess strong “high value-added, low energy consumption” structural emission-reduction capacity. In contrast, non-core cities display a “dual mechanism failure”: first, economic growth significantly inhibits structural upgrading, with urban development still dominated by high-carbon industrialisation; second, the effect of structural upgrading on carbon emissions is statistically insignificant, meaning that even where structural change occurs, it fails to translate into substantive emission-reduction capacity. This indicates that non-core cities have yet to establish a structural pathway towards green transition, with their capacity for green transformation severely lagging behind.

From the above empirical results, it is evident that the spatial rupture of this mechanism chain directly produces a tier-asymmetric distribution of green development capacity, giving rise to a spatial configuration in which core cities function as “green centres” and non-core cities constitute the “carbon periphery”. Core cities, by virtue of their dominance over industrial structure, fiscal resource allocation capacity, and technological spillover potential, occupy an institutional advantage in green governance and may, through industrial relocation or factor syphoning, outsource pollution beyond their metropolitan tier. By contrast, non-core cities become “silent zones”, absorbing high-carbon industries, bearing transformation pressures far beyond their capacity boundaries, yet remaining largely excluded from substantive decision-making power or equitable access to resources in the green transition.

The empirically identified “growth-suppressed structure—structurally ineffective emission reduction” chain constitutes a direct mechanistic reflection of this spatial asymmetry, representing the green manifestation of structural inequality within metropolitan regions.

The aforementioned findings extend and deepen existing theories on urban carbon emissions in several key dimensions. On the one hand, this study proposes embedding the EKC theory within regional structural systems, thereby moving beyond the conventional city-centric EKC framework. It highlights how a city’s tiered position within the metropolitan hierarchy imposes critical institutional constraints and reflects structural differentiation in its carbon emission pathway. On the other hand, the study identifies a previously overlooked barrier to green transition—namely, the failure of the structural mediating mechanism. It shows that green development is not automatically triggered by economic growth but instead depends on specific structural foundations and institutional conditions, which underscores its nature as a conditional governance process. This mechanistic perspective offers a theoretical explanation for why some cities remain trapped in the growth–emissions coupling dynamic. It also provides a foundation for future research on green governance capacity, path dependence, and the spatial governance of industrial transitions. Finally, by empirically identifying the role of urban tiered structures, the study constructs a pathway linking “growth-suppressing structural upgrading” to “structural failure in emission reduction”. This offers strong empirical and institutional evidence from China to inform the advancement of EKC theory, the political economy of green transition, and the study of regional carbon equity. Consequently, urban green transition constitutes not merely a process of industrial structure optimisation but, fundamentally, a process of reconstructing regional spatial governance capacity. Only by recognising that the tiered differentiation of green pathways is deeply embedded within institutional arrangements, the prevailing development paradigm, and the underlying resource structure can a sound foundation for spatially targeted policy interventions be established. This mechanistic synthesis also delineates the trajectory for subsequent policy recommendations: Redistributing the spatial accessibility of green capabilities and instituting a cross-tier structural synergy system emerge as the central challenge for realising an equitable and effective “dual carbon transition”. Based on the above research findings, this study argues that the key to advancing an internal green transition within metropolitan areas and alleviating the imbalance of the “green core–carbon periphery” mechanism lies not merely in controlling total carbon emissions or enhancing the mitigation efficiency of individual cities, but more critically in dismantling the “reverse mediation chain” in which economic growth suppresses structural upgrading, thereby establishing sustainable, structure-based mitigation capacity. Accordingly, low-carbon transition policies at the regional scale should embody a composite orientation characterised by differentiation, structural guidance, and collaborative governance. Policy objectives should shift towards an integrated pathway that is described as “capability-compensatory”, “mechanism-reconstructive”, and “spatially synergistic”.

Policy recommendations are proposed across three key dimensions:

(1)

Establish Differentiated Incentives for Structural Transformation: Construct more accessible pathways for green upgrading in non-core cities through fiscal subsidies, green industry funds, and service sector incubation platforms.

(2)

Implement a Tier-Graded “Dual Carbon Burden-Sharing System”: Link urban carbon reduction targets to their structural positioning, carrying capacity, and transition conditions. This linkage is crucial to mitigate the capacity mismatch and performance distortions arising from “one-size-fits-all” target setting.

(3)

Enhance the Metropolitan Collaborative Governance Platform: Introduce a “carbon spillover compensation” mechanism alongside robust industrial relocation monitoring. Establish a cross-regional cooperative mechanism between core and non-core cities for shared responsibility in emission reduction and equitable distribution of transition benefits.

In summary, advancing green transition within China’s metropolitan regions necessitates shifting from an “outcome-oriented” approach focused solely on emission control towards a “mechanism-oriented” strategy centred on activating structural upgrading and building equitable capacities. Only by dismantling the entrenched tiered bias within current carbon responsibility allocation and achieving a structural reconfiguration of green policies, fiscal resources, and governance capabilities can the tiered traps inherent in carbon peaking pathways be fundamentally overcome. This transformation holds the key to unlocking a collaborative future characterised by a fairer and more sustainable green transition among cities.