Drivers of Green Transition Performance Differences in China’s Resource-Based Cities: A Carbon Reduction–Pollution Control–Greening–Growth Framework

Abstract

Understanding the multidimensional sources and key drivers of differences in green transition performance (GTP) among resource-based cities is vital for accomplishing national sustainable development objectives and facilitating regional coordination. This study proposes a “Carbon Reduction–Pollution Control–Greening–Growth” evaluation framework and utilizes the entropy method to assess the GTP of China’s resource-based cities from 2013 to 2022. The Dagum Gini coefficient and variance decomposition methods are employed to investigate the GTP differences, and the Optimal Parameters-Based Geographical Detector and the Geographically and Temporally Weighted Regression model are applied to identify the driving factors. The results indicate the following trends: (1) GTP exhibits a fluctuating upward trend, accompanied by pronounced regional imbalances. A pattern of “club convergence” is observed, with cities showing a tendency to shift positively toward adjacent types. (2) Spatial differences in GTP have widened over time, with transvariation density emerging as the dominant contributor. (3) Greening differences represent the primary structural source, with an average annual contribution exceeding 60%. (4) The impact of digital economy, the level of financial development, the degree of openness, industrial structure, and urbanization level on GTP differences declines sequentially. These factors exhibit notable spatiotemporal heterogeneity, and their interactions display nonlinear enhancement effects.

1. Introduction

The global momentum toward green transition is rapidly reshaping urban development paradigms [1,2]. The United Nations 2030 Agenda for Sustainable Development emphasizes the need for systemic transformation through goals such as “Sustainable Cities and Communities” (SDG 11) and “Climate Action” (SDG 13). According to the International Energy Agency (IEA), global energy-related carbon emissions reached 36.8 billion tons in 2022, with cities contributing over 70%, highlighting their strategic importance as the primary arena for green transition [3]. As archetypal high-carbon economies, resource-based cities face particularly intense transition pressures due to their reliance on mineral extraction, mono-industrial economic structures, and significant ecological deficits [4,5,6]. China, the world’s largest carbon emitter, is spearheading a comprehensive green transition of its economy and society, with pledges to reach peak carbon emission levels by 2030 and achieve carbon neutrality by 2060. Under this dual-carbon strategy, resource-based cities—marked by abundant resource endowments and energy-intensive industries—encounter especially acute challenges in achieving green transition goals. Data from the China Emission Accounts and Datasets (CEADs) indicate that, in 2020, carbon emissions from China’s resource-based prefecture-level cities made up nearly one third of the national total. Their carbon emissions per CNY 10,000 of GDP were 1.6 times higher, and per capita emissions 16.1% higher than those of non-resource-based cities [7]. Notably, due to divergent development stages, these cities exhibit substantial variation in industrial structure, energy consumption modes, and resource endowments, leading to marked differences in green transition performance (GTP). Excessive differences in GTP risk undermining environmental justice and impeding the equitable distribution of environmental benefits. In this context, scientifically evaluating the GTP of resource-based cities and analyzing the multidimensional sources and key drivers of performance disparities are crucial for achieving national sustainability targets and fostering coordinated regional development.

Based on the research background outlined above, this study addresses the following key questions: (1) How can the GTP of resource-based cities be measured scientifically? (2) What temporal and spatial patterns characterize GTP in these cities? (3) What are the primary sources of differences in GTP across resource-based cities? (4) Which factors drive these differences, and to what extent do their effects exhibit spatiotemporal heterogeneity and interactions? To answer these questions, this study adopts a “Carbon Reduction–Pollution Control–Greening–Growth” evaluation framework, emphasizing policy relevance in the design of the indicator system. The entropy method is applied to measure the GTP of China’s resource-based cities, and, building on this, the Dagum Gini coefficient and variance decomposition methods are used to reveal the multidimensional sources of GTP disparities from both spatial and structural perspectives. Lastly, the Optimal Parameters-Based Geographical Detector (OPGD) and the Geographically and Temporally Weighted Regression (GTWR) model are employed to identify the spatiotemporal heterogeneity and interactive effects of the driving factors.

Theoretically, this study constructs an analytical framework of “spatial decomposition–structural decomposition–causal identification,” systematically uncovering the evolution of GTP differences among resource-based cities and enriching the literature on green transition. Practically, this study provides a detailed characterization of GTP, enabling cities to understand their current relative positioning and potential pathways for improvement, thereby offering evidence to inform policies aimed at promoting coordinated enhancements in GTP.

2. Literature Review

The escalating global ecological crisis has prompted a growing wave of academic inquiry into green transition. Current studies closely related to the topic of this paper mainly address three primary research areas:

The first concerns the quantitative measurement of GTP, which is typically approached in two ways: The first approach relies on single indicators to represent performance. For instance, green GDP incorporates environmental costs into economic activity, serving as a modification of traditional GDP [8,9,10], while green total factor productivity (GTFP) comprehensively considers the relationship between multiple input factors—such as resources and the environment—and economic output [11,12,13]. However, this method has inherent limitations, as it cannot fully capture the multidimensional nature of green transition. The second approach involves the construction of multidimensional evaluation systems, whereby scholars can assess GTP across multiple dimensions—economic, environmental, and social. The economic dimension generally includes indicators related to economic output and market size; the environmental dimension covers aspects such as pollution reduction and resource utilization efficiency; and the social dimension addresses green consumption, eco-friendly travel, and related factors. Constructing indicator systems has become the mainstream approach for quantifying GTP, as they more effectively capture the multidimensional nature of green transition. For example, Wang et al. developed an evaluation framework based on economic development, environmental sustainability, and improved livelihoods [14], while Liu et al. adopted the DPSIR model, incorporating five dimensions: driving forces, pressures, state, influences, and responses [15].

The second primary research area concerns the spatiotemporal characteristics of GTP, which have emerged as a central focus in empirical research. Building upon performance measurement efforts, many scholars have employed kernel density estimation, convergence analysis, and exploratory spatial data techniques to investigate the spatiotemporal evolution of green transition [16,17]. Given the inherent imbalances in regional development, differences in GTP have garnered growing scholarly interest. Early research predominantly used descriptive statistics to compare performance across regions [18,19], and, while this approach revealed the existence of disparities, it offered limited insights into the sources of their differences. With the evolution of analytical methodologies, tools such as the Theil index and the Dagum Gini coefficient have been widely employed to decompose regional differences in GTP. For example, Tian et al. employed the Theil index to examine the spatial imbalance in GTP in the manufacturing sector across China’s three major river basins, finding that internal disparities within the Yellow River Basin were higher than those in the Yangtze and Pearl River Basins [20]. Long et al., using the Dagum Gini coefficient decomposition method, revealed the sources of differences in China’s GTP, showing that inter-regional disparities contributed far more than intra-regional differences and transvariation density [21].

Lastly, the factors influencing GTP have attracted substantial scholarly attention. These factors generally fall into two categories—natural factors (e.g., temperature, humidity, precipitation) [22] and socio-economic elements (e.g., industrial structure, technological innovation, policy frameworks, and population size and quality) [23,24,25]—and have been predominantly examined via linear regression models. Due to their greater measurability, socio-economic factors have received a greater focus, and numerous studies have shown that GTP is shaped by a combination of socio-economic driving forces, with significant regional variation in the impact of the same factor. For instance, Liu and Kong found that digital economy significantly advances green transition in cities located in the lower reaches of the Yangtze River, while its impact in the middle and upper reaches remains limited [26].

Existing studies provide valuable insights for this research but still exhibit several limitations: First, due to differences in scholars’ understanding of GTP, existing evaluation systems vary significantly in terms of dimension selection and specific indicators. This has resulted in considerable discrepancies in measured GTP, limiting its utility for scientifically assessing the effectiveness of green transition policies. With China’s explicit coordinated advancement of the “Carbon Reduction–Pollution Control–Greening–Growth” framework, establishing an evaluation system consistent with this strategic orientation has become a key, pressing task in GTP research. Second, mainstream methods, such as the Theil index and the Dagum Gini coefficient, primarily focus on decomposing the spatial differences and sources of GTP from a geographical perspective. However, they fail to analyze structural disparities among the core dimensions of green transition—such as economic, environmental, and social dimensions—from an internal systems perspective. Amid China’s nationwide shift toward a comprehensive green transition, simply narrowing the spatial disparities in GTP is insufficient to achieve the dual objectives of regional balance and systemic coordination; attention must also be paid to structural alignment across dimensions. Finally, few studies simultaneously consider the spatiotemporal heterogeneity of factors influencing GTP and the potential synergistic or antagonistic interactions among these factors. The strength and direction of these factors may vary across regions and periods, and interactions among factors may cause some factors to reinforce or constrain others. Ignoring these aspects hinders scientific understanding of the key drivers behind regional differences in GTP, thereby reducing the precision and effectiveness of policy design.

In response to the above-mentioned research deficiencies, this study makes three primary contributions: First, it establishes a policy-oriented, multidimensional evaluation framework, and, based on China’s recently proposed green transition paradigm—“Carbon Reduction–Pollution Control–Greening–Growth”—a comprehensive performance evaluation system for resource-based cities is developed. This system utilizes a policy–indicator mapping matrix to ensure direct alignment between evaluation results and government policy agendas, thereby mitigating the subjectivity and narrow scope inherent in existing indicator systems. Second, this study integrates the Dagum Gini coefficient with variance decomposition to analyze differences in GTP from two perspectives: spatial differentiation and structural imbalance. This dual-perspective approach addresses the limitations of conventional research, which often prioritizes regional balance while overlooking systemic coordination. It enables a more holistic understanding of the sources of GTP differences and supports the formulation of region-specific development policies. Third, this study applies the OPGD and the GTWR model to uncover the spatiotemporal heterogeneity and interactive mechanisms of driving factors. This methodological approach enhances our understanding of the underlying causes of GTP differences in resource-based cities, thereby providing more precise and actionable insights for policy design.

3. Materials and Methods

3.1. Evaluation Indicator System for GTP

Unlike developed countries that have completed industrialization, China, in the mid-to-late stages of its industrialization, faces multiple developmental constraints, necessitating a systemic transformation to reshape its development paradigm [27]. From the perspective of climate governance, the “dual-carbon” goal proposed in 2020 represents a stringent constraint, compelling a departure from the traditional high-carbon growth model. In terms of environmental governance, although the battle against pollution has yielded phased achievements, enduring challenges remain—including excessive PM_2.5 concentrations, degraded aquatic ecosystems, and soil pollution risks—intensifying the pressure to transition environmental improvements from incremental to qualitative changes. On the economic front, since 2010, China’s GDP growth rate has declined from around 10% to approximately 5%, and the marginal returns of traditional factor-driven growth have diminished, highlighting the urgent need to foster new growth drivers through green technological innovation and industrial upgrading. To tackle these multifaceted challenges, in 2022, the 20th CPC National Congress set the guiding principle of “coordinated efforts to reduce carbon emissions, control pollution, expand greening, and boost growth.” In 2025, the Government Work Report designated this coordinated approach as one of the top ten national development priorities, signaling a shift from conceptual advocacy to comprehensive policy implementation. As a state-led governance model, China possesses institutional advantages that provide strong political support for green transition. Within this context, the coordinated advancement of carbon reduction, pollution control, greening, and growth not only serves as a practical framework for promoting a comprehensive green transition in economic and social development but also provides a conceptual foundation for reassessing GTP under evolving policy regimes.

From the perspective of synergetics theory, carbon reduction, pollution control, greening, and growth represent four major subsystems within the broader green transition system. The synergy among these subsystems manifests in several ways: carbon reduction and pollution control produce a “co-control effect” due to their shared sources (e.g., coal combustion simultaneously emits CO₂ and SO₂); greening generates a “foundational support effect” through the provision of ecosystem services; and growth creates a “driving force effect” by optimizing resource allocation. These three effects collectively constitute the “three-dimensional pillars” supporting green transition. However, green transition is not a simple linear sum of its parts. Rather, it is driven by nonlinear interactions among subsystems, catalyzed by institutional innovation and technological advancement. This dynamic interaction leads to a systemic emergence effect—“1 + 1 + 1 + 1 > 4”—indicating a holistic transformation of the socio-economic system from a high-carbon, high-pollution, low-efficiency state to one that is low-carbon, clean, and efficient.

Guided by this theoretical framework, this study develops an evaluation system to assess GTP in China’s resource-based cities, focusing on carbon reduction, pollution control, greening, and growth. The establishment of this system draws upon official policy documents and the outcomes of existing academic research [28,29], and the specific indicators employed are elaborated in

Table 1. Evaluation system for measuring GTP in China’s resource-based cities.

Dimension	Sub-Dimension	Calculation	Type 1
Carbon Reduction	Total carbon emissions	CO2 emissions	-
	Carbon emission intensity	CO2 emissions/Year-end total population	-
		CO2 emissions/GDP	-
	Carbon emission growth rate	CO2 emissions growth rate	-
Pollution Control	Air pollution control	Industrial SO2 emissions/GDP	-
		Industrial smoke and dust emissions/GDP	-
		PM2.5 concentration	-
	Water pollution control	Wastewater discharge volume/GDP	-
	Soil pollution control	Pure chemical fertilizer use/Total sown area of crops	-
		Municipal solid waste collection volume /Year-end total population	-
Greening	Natural greening	Urban park green space area/Year-end total population	+
		Vegetation greening area/Built-up area	+
	Social greening	Number of green patent applications /Year-end total population	+
		Total ridership of public buses and trolleybuses/Year-end total population	+
Growth	Qualitative improvement	Total factor productivity index	+
		Technological progress index	+
		Technical efficiency index	+
	Quantitative growth	GDP per capita	+
		GDP growth rate	+
		Total retail sales of consumer goods /GDP	+
		Per capita disposable income	+

¹ “+” = positive type, “-“ = negative type.

.

In the carbon reduction dimension, the evaluation focuses on three key aspects: total carbon emissions, carbon emission intensity, and the carbon emission growth rate [30]. Total carbon emissions represent the overall scale of regional mitigation efforts and aligns directly with the aggregate control targets set by the “dual-carbon” strategy. Carbon emission intensity is measured using two indicators—per capita carbon emissions and carbon emissions per unit of GDP—which characterize the carbon linkage of population size and economic growth, respectively. The former indicates individual carbon responsibility, while the latter captures the efficiency of carbon use in economic output. Including the carbon emission growth rate provides a dynamic view of emission trends and helps assess the long-term effectiveness of mitigation policies, thereby supporting sustained progress toward emission reduction goals.

In the pollution control dimension, the evaluation focuses on three critical areas: air, water, and soil pollution control. The report of the 20th CPC National Congress emphasizes the importance of “intensifying efforts in environmental pollution prevention and persevering in securing victories in the campaigns for blue skies, clear waters, and clean soil,” identifying these areas as key priorities for environmental governance [31,32]. This study uses three indicators to evaluate air pollution control: industrial SO₂ emission intensity, industrial smoke and dust emission intensity, and PM_2.5 concentration. Water pollution control is measured by wastewater discharge intensity, while soil pollution control is assessed using fertilizer application intensity and per capita municipal solid waste collection.

In the greening dimension, the evaluation concerns two aspects: natural greening and social greening. Natural greening stresses boosting urban environmental carrying capacity and bettering ecosystem service functions by expanding green spaces. Social greening, on the other hand, focuses on fostering green production and lifestyle practices at the societal level. This includes embracing environmentally friendly technologies, advancing sustainable transportation systems, and fostering green consumption. Together, these two facets capture the environmental and behavioral dimensions of greening, offering a comprehensive representation of its multifaceted implications [33]. This study employs per capita park green space and the green coverage ratio in built-up areas to evaluate natural greening and per capita green patent applications and per capita passenger volume of public electric and gas-powered buses to measure social greening.

In the growth dimension, the evaluation concerns two aspects: qualitative enhancement and quantitative expansion. The report of the 20th CPC National Congress emphasizes the significance of “effectively upgrade and appropriately expand China’s economic output.” In other words, growth provides essential material support for the other three subsystems. The core objective of green transition is to transform the socio-economic development model from an industrial civilization paradigm to an ecological civilization paradigm, thereby addressing the so-called “green paradox” and illustrating that ecological protection and economic growth are not mutually exclusive [34,35]. This study uses total factor productivity, the technological progress index, and the technical efficiency index to measure the quality of economic growth, while per capita GDP, GDP growth rate, household consumption level, and per capita disposable income are employed to assess its quantitative expansion.

3.2. Methods

3.2.1. Entropy Method

The entropy method, rooted in information entropy theory, is an objective weighting technique that ascertains indicator weights from the extent of data dispersion. The greater the dispersion of an indicator, the more information it carries, the greater its contribution to the comprehensive evaluation, and the greater its assigned weight. As described above, our evaluation of GTP in resource-based cities encompasses 21 underlying indicators across four dimensions—carbon reduction, pollution control, greening, and growth—and the entropy method and its probability density functions are used to eliminate dimensional differences caused by the heterogeneity in the indicator types and units. This avoids the subjectivity associated with expert scoring and analytic hierarchy processes, offering a more objective assessment of each indicator’s importance within the comprehensive evaluation framework.

$$x_{ij}^{\prime }={\displaystyle \frac{x_{ij}-\min \left(x_{ij}\right)}{\max \left(x_{ij}\right)-\min \left(x_{ij}\right)}}$$

(1)

$$x_{ij}^{\prime }={\displaystyle \frac{\mathrm{m}\mathrm{a}\mathrm{x}\left(x_{ij}\right)-x_{ij}}{\max \left(x_{ij}\right)-\min \left(x_{ij}\right)}}$$

(2)

$$P_{ij}={\displaystyle \frac{x_{ij}^{\prime }}{{\sum }_{i=1}^m{x}_{ij}^{\prime }}}$$

(3)

$$e_j=-k{\sum }_{i=1}^m{P}_{ij}\mathrm{l}\mathrm{n}\left(P_{ij}\right)$$

(4)

$$g_j=1-e_j$$

(5)

$$w_j={\displaystyle \frac{g_j}{{\sum }_{j=1}^n{g}_j}}$$

(6)

$$I_{ij}={\sum }_{j=1}^n{w}_j{x}_{ij}^{\prime }$$

(7)

In Equations (1) and (2), $x_{ij}$ stands for the original value of indicator j in year i; $\mathrm{m}\mathrm{a}\mathrm{x}\left(x_{ij}\right)$ and $\mathrm{m}\mathrm{i}\mathrm{n}\left(x_{ij}\right)$ denote the maximum and minimum values of indicator j, respectively; $x_{ij}^{\prime }$ denotes the standardized value; m is the number of years; and n is the total number of evaluation indicators. Equations (3)–(5) are applied to compute the information entropy and utility values of the indicators. Specifically, in Equation (4), if k > 0, then $k={\displaystyle \frac{1}{\ln \left(m\right)}}$, and, when $P_{ij}=0$, we define $P_{ij}\ln \left(P_{ij}\right)=0$ ($0\le e_j\le 1$) to ensure numerical stability. Equation (6) is used to determine the weight of each indicator, while Equation (7) is used to compute the comprehensive index of the samples.

3.2.2. Markov Chain

The Markov chain, a fundamental tool in stochastic forecasting theory, is a discrete-time stochastic process characterized by a finite set of states. It estimates the future state and development trends of a system by constructing a state transition probability matrix [36]. A defining feature of the Markov model is its memoryless property, meaning the future state relies only on the present state and not on the sequence of past states [37]. For this study, the Markov chain approach is particularly suitable, as it captures not only the static distribution of GTP across cities but also the dynamic transition probabilities between performance levels. This enables a deeper understanding of whether cities are likely to maintain their current state, experience gradual improvement, or face decline over time. Moreover, the method facilitates the identification of “club convergence” in the green transition of resource-based cities by quantifying both the persistence within categories and the limited likelihood of leapfrog transitions.

$$\mathrm{P}\left\{X\left(t\right)=j|X\left(t-1\right)=i,\dots ,X\left(0\right)=i_0\right\}=\mathrm{P}\left\{X\left(t\right)=j|X\left(t-1\right)=i\right\},$$

(8)

$$p_{ij}={\displaystyle \frac{n_{ij}}{n_j}},$$

(9)

In Equation (8), the GTP X in the present period t relies on its performance level j in the previous period t − 1. Equation (9) defines the probability of a city transitioning from performance level i to level j in the subsequent period, where n_ij denotes the total number of cities that will shift from performance level i in the present period to performance level j in the subsequent period over the entire study period, and n_j denotes the total number of cities belonging to performance level i throughout the study period. By organizing all transition probabilities into matrix form, the state transition probability matrix for the GTP of resource-based cities is obtained.

3.2.3. Dagum Gini Coefficient Decomposition Method

The Dagum Gini coefficient decomposition method offers an innovative approach to measuring imbalance. By introducing the concept of the directional economic distance ratio, this method breaks down the overall Gini coefficient into three parts: the within-group Gini coefficient (G_w), the between-group Gini coefficient (G_nb), and the intensity of transvariation (G_t). This method fully considers the distribution characteristics of subsamples, effectively addressing issues of data overlap and unclear sources of disparity, and thus overcomes the limitations of the traditional Gini coefficient and the Theil index [38]. Accordingly, this study employs this method to examine the spatial sources and dynamic evolution of GTP differences among China’s resource-based cities.

$$G={\displaystyle \frac{{\sum }_{i=1}^k{\sum }_{m=1}^k{\sum }_{j=1}^{n_i}{\sum }_{r=1}^{n_m}\left|y_{ij}-y_{mr}\right|}{2n^2\mu }},{\mu }_m\le {\mu }_i\le \cdots \le {\mu }_k,$$

(10)

$$G_{ii}={\displaystyle \frac{{\sum }_{j=1}^{n_i}{\sum }_{r=1}^{n_i}\left|y_{ij}-y_{ir}\right|}{2n_i^2{\mu }_i}},$$

(11)

$$G_w={\sum }_{i=1}^k{G}_{ii}{p}_i{s}_i,$$

(12)

$$G_{im}={\displaystyle \frac{{\sum }_{j=1}^{n_i}{\sum }_{r=1}^{n_m}\left|y_{ij}-y_{mr}\right|}{n_i{n}_m\left({\mu }_i+{\mu }_m\right)}},$$

(13)

$$G_{nb}={\sum }_{i=2}^k{\sum }_{m=1}^{i-1}{G}_{im}\left(p_i{s}_m+p_m{s}_i\right)D_{im},$$

(14)

$$G_t={\sum }_{i=2}^k{\sum }_{m=1}^{i-1}{G}_{im}\left(p_i{s}_m+p_m{s}_i\right)\left(1-D_{im}\right),$$

(15)

$$D_{im}={\displaystyle \frac{d_{im}-p_{im}}{d_{im}+p_{im}}},$$

(16)

$$d_{im}={\int }_0^{\infty }d{F}_i\left(y\right){\int }_0^{y}y-xd{F}_m\left(x\right),$$

(17)

$$p_{im}={\int }_0^{\infty }d{F}_m\left(y\right){\int }_0^{y}y-xd{F}_i\left(x\right),$$

(18)

In Equation (10), k denotes the number of resource-based city categories; n denotes the total number of resource-based cities; y_ij (y_mr) represents the GTP of city j (r) in category i (m); n_i (n_m) stands for the number of cities in category i (m); µ_i (µ_m) denotes the average GTP of category i (m). Equations (11) and (13) calculate the Gini coefficient for category i (G_ii) and the Gini coefficient between categories i and m (G_im), respectively. Equations (12), (14) and (15) calculate the G_w, G_nb, and G_t for the GTP of resource-based cities. Equations (16) to (18) calculate the D_im, d_im, and p_im, respectively.

3.2.4. Variance Decomposition Method

The Dagum Gini coefficient is used to identify the spatial sources of disparities in the GTP of resource-based cities; however, the internal structural differences within the performance system remain unclear. Therefore, this study applies variance decomposition to explore the “black box” of GTP differences and further investigate the sources of internal structural disparities. Variance decomposition is a method for analyzing the sources of disparities based on the principles of variance and covariance in statistics. It decomposes the total variance of the subject into distinct components and evaluates the contribution of each component to the overall disparity [39]. Importantly, it assumes an additive relationship between the subject and its components. In this study, the GTP of resource-based cities is calculated using four indices—carbon reduction (CRI), pollution control (PCI), greening (GI₁), and growth (GI₂)—making it a suitable case for variance decomposition.

$$var\left(GTP\right)=cov\left(GTP,CRI\right)+cov\left(GTP,PCI\right)+cov\left(GTP,{GI}_1\right)+cov(GTP,{GI}_2),$$

(19)

$$1={\displaystyle \frac{cov(GTP,CRI)}{var\left(GTP\right)}}+{\displaystyle \frac{cov(GTP,PCI)}{var\left(GTP\right)}}+{\displaystyle \frac{cov(GTP,{GI}_1)}{var\left(GTP\right)}}+{\displaystyle \frac{cov(GTP,{GI}_2)}{var\left(GTP\right)}},$$

(20)

In Equations (19) and (20), var and cov represent variance and covariance, respectively. These equations measure the contribution of differences in the four dimensions to the overall GTP difference. A larger contribution from a specific dimension indicates a greater impact on the disparity in GTP.

3.2.5. Optimal Parameters-Based Geographical Detector

The Geographical Detector is a statistical method for analyzing spatial differentiation and identifying underlying driving factors. Its core principle is to quantify the influence of explanatory variables on spatial heterogeneity by assessing the similarity in the spatial distribution patterns of both independent and dependent variables. Traditional geographic detectors rely on subjective methods for discretizing variables, leading to issues such as bias and poor discretization. In contrast, the OPGD uses algorithms to determine the optimal combination of data discretization methods and partition numbers, offering a set of continuous discretization approaches and breakpoints for each variable, thereby yielding more accurate and objective results [40]. Therefore, this study applies the factor detection and interaction detection models from this method to investigate the driving factors and their interactions in the evolution of GTP disparities in China’s resource-based cities.

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

(21)

In Equation (21), q represents the explanatory power of a factor in the GTP differences in resource-based cities, with values ranging from 0 to 1, where a larger q-value means a stronger influence of the variable on the disparities. L represents the number of independent variables. N_h and N denote the sample sizes of independent variable h and all independent variables, respectively. ${\sigma }_h^2$ and ${\sigma }^2$ denote the sample variance of indicator h and the sample variance of all independent variables, respectively.

3.2.6. Geographically and Temporally Weighted Regression Model

Although the OPGD identifies the key drivers of differences in the GTP of China’s resource-based cities from a global perspective, it fails to reveal the specific direction and strength of these effects and does not fully account for the influence of spatial continuity and proximity on variable relationships. The GTWR model, an extension of the Geographically Weighted Regression (GWR) model, incorporates a temporal dimension to capture dual-weighted spatial heterogeneity within a multivariate linear regression framework. It not only overcomes sample size limitations but also effectively addresses spatiotemporal non-stationarity, thereby remedying the Geographical Detector’s shortcomings in analyzing local effects and enhancing the accuracy of estimation results [41]. Thus, this study applies the GTWR model to comprehensively examine the spatiotemporal differentiation of the driving factors behind the GTP differences among China’s resource-based cities.

$$Y_i={\beta }_0\left(l_i,g_i,t_i\right)+{\sum }_{k=1}^p{\beta }_k\left(l_i,g_i,t_i\right)X_{ik}+{\epsilon }_i,$$

(22)

In Equation (22), l_i and g_i denote the longitude and latitude of sample city i, respectively; t denotes time; ${\beta }_0\left(l_i,g_i,t_i\right)$ is the intercept; ${\beta }_k\left(l_i,g_i,t_i\right)$ are the estimated coefficients; k is the number of explanatory variables; X_ik represents the explanatory variables; and ${\epsilon }_i$ is the random disturbance term.

3.3. Study Area and Data

Based on the classification of resource-based cities outlined in the Plan for the Sustainable Development of Resource-based Cities in China (2013–2020) issued by the State Council, and taking data availability into account, this study selects 111 out of 126 prefecture-level administrative units to comprise the study area. These include 13 ‘growing’-type cities, 60 ‘grown-up’-type cities, 23 ‘recessionary’-type cities, and 15 ‘regenerative’-type cities. The study period spans from 2013 to 2022.

Carbon dioxide emission data come from the Emissions Database for Global Atmospheric Research (EDGAR) of the European Union, while PM_2.5 data are obtained from the China High Air Pollutants Dataset (CHAP). Data on green patent applications are retrieved from the China Research Data Services Platform (CNRDS). To calculate total factor productivity (TFP), capital stock, urban construction land area, and labor force are used as input indicators, with real GDP as the output indicator. The global benchmark super-efficiency SBM model is used for measurement. Following the method of Yuan et al. [42], the TFP index, technological progress index, and technical efficiency index are converted into cumulative growth indices using 2013 as the base year. Nighttime light data are obtained from the study by Chen et al. [43], while public environmental concern is measured using the Baidu search frequency of the keywords “haze” and “environmental pollution” (Beijing, China; https://index.baidu.com/) [44]. Following the method of Zhao et al. [45], the level of digital economy development is assessed using two dimensions—Internet development and digital inclusive finance—measured via the entropy weight method. Internet development is reflected by four indicators, namely, Internet penetration, employment in related industries, output level, and mobile phone penetration. Specifically, these indicators include the number of broadband Internet access users per 100 people, the proportion of employees in computer services and software industries relative to total urban employment, per capita telecommunication services, and the number of mobile phone users per 100 people. The corresponding data are obtained from the China Urban Statistical Yearbook. Digital inclusive finance is measured using the China Digital Financial Inclusion Index, which is sourced from the Digital Finance Research Center of Peking University and the Digital Economy Open Research Platform of Ant Financial. Additional data sources include the China Urban Statistical Yearbook, China Urban and Rural Construction Statistical Yearbook, and the statistical yearbooks and bulletins of various provinces and cities. Missing or anomalous values are supplemented using linear interpolation and moving average methods.

4. Results

4.1. Estimation Results of GTP

4.1.1. Temporal Characteristics of GTP

Overall, GTP exhibits a fluctuating upward trajectory, increasing from 0.078 in 2013 to 0.104 in 2022, with an average annual growth rate of 3.17% (Figure 1). This reflects notable progress in the green transition of resource-based cities, aligning with the findings of Xu et al. and Wang [46,47]. Since the 18th CPC National Congress, ecological civilization has been incorporated into China’s overall plan for building socialism with Chinese characteristics. In response, the government has issued a series of policy documents, which provide scientific guidance for promoting green transition. In terms of sub-dimensional indices, the average values for pollution control (0.03), growth (0.026), carbon reduction (0.021), and greening (0.015) reveal a performance hierarchy of “pollution control > growth > carbon reduction > greening.” Over the study period, all indices except carbon reduction show upward trends. In sum, while significant progress has been achieved in pollution control and growth, improvements in carbon reduction and greening have lagged. Thus, more balanced advancement across all four dimensions is essential to further enhance overall GTP.

The average GTP values for growing, grown-up, recessionary, and regenerative cities are 0.085, 0.09, 0.097, and 0.097, respectively (Figure 2), displaying a hierarchy of “regenerative ≈ recessionary > grown-up > growing.” In terms of temporal evolution, GTP in growing and grown-up cities demonstrates a steadily fluctuating upward trend, with average annual growth rates of 2.78% and 4.46%, respectively. In contrast, GTP in recessionary cities first rises and then falls, while regenerative cities exhibit the opposite trend—an initial decline followed by subsequent growth. These differences can be attributed to several underlying factors: Growing and grown-up cities capitalize on the advantages of the resource development stage by advancing resource-intensive processing, improving energy efficiency through clean energy substitution, and fostering emerging industries, collectively contributing to a gradual increase in GTP [47]. Recessionary cities experience short-term improvements in GTP due to central government transfers and industrial support policies. However, as resource exhaustion deepens—compounded by industrial homogeneity and talent loss—the slow development of alternative industries and the waning effect of policy incentives weaken the momentum for transformation, leading to a downturn in GTP [48]. This “decline–then–rise” pattern in regenerative cities reflects a “creative destruction” transition logic. The initial retreat of traditional industries, combined with a temporal lag in the emergence of new sectors, causes a short-term drop in GTP, while, in the later phase, GTP rebounds significantly, driven by industrial restructuring supported by the deployment of new energy technologies, innovation, and ecological compensation mechanisms.

4.1.2. Spatial Evolution Characteristics of GTP

This study applies the natural breaks method to classify the GTP in 2013 and 2022 into four tiers (Figure 3). The results reveal significant spatial differences in GTP across these cities, with spatial patterns dynamically evolving over time. In 2013, 40 cities fell into the Low-GTP category, comprising 12.5% growing, 57.5% grown-up, 22.5% recessionary, and 7.5% regenerative cities. Only two regenerative cities—Xuzhou and Linyi—were classified as High-GTP. By 2022, the number of cities in the Low-GTP category had increased to 69, consisting of approximately 10.14% growing, 53.62% grown-up, 26.09% recessionary, and 10.14% regenerative cities. It should be emphasized that, because the overall GTP of resource-based cities increased during 2013–2022 (Figure 2), the classification intervals derived from the natural breaks method are dynamic, and the corresponding GTP values for each class in 2022 are higher than those in 2013. Therefore, the observed increase in the number of low-GTP cities reflects a relative reclassification against a background of overall improvement rather than an absolute decline in performance. Only two grown-up cities—Jilin and Yichun—were categorized as High-GTP. These findings indicate that grown-up cities generally encounter greater challenges in achieving green transition, primarily due to their long histories of resource extraction, heavy reliance on traditional industries, and fragile ecological environments, which often means they are relegated to the Low-GTP category. Nevertheless, the emergence of Jilin and Yichun in the High-GTP group in 2022 demonstrates that some grown-up cities achieve significant breakthroughs in green development, highlighting their strong potential for transition. Furthermore, throughout the study period, high-GTP cities are typically scattered, whereas low-GTP cities are concentrated in contiguous clusters in Southwest, North, and Northwest China.

4.1.3. Dynamic Transition Characteristics of GTP

This study employs a Markov chain model to examine the long-term dynamics of GTP in China’s resource-based cities, where the states “Low,” “Medium-Low,” “Medium-High,” and “High” represent relative positions within the annual GTP distribution. (Note: As Figure 1 shows, overall GTP increased from 2013 to 2022, so the absolute GTP values corresponding to each state in 2022 are higher than those in 2013.)

Table 2. Markov chain transition probability matrix of GTP in China’s resource-based cities.

t + 1	Type	Low	Medium-Low	Medium-High	High
Overall	Low	0.532	0.317	0.028	0.123
	Medium-Low	0.048	0.472	0.357	0.123
	Medium-High	0.028	0.048	0.714	0.210
	High	0.053	0.103	0.140	0.704
Growing Type	Low	0.259	0.444	0.185	0.111
	Medium-Low	0.037	0.185	0.556	0.222
	Medium-High	0.000	0.074	0.481	0.444
	High	0.028	0.056	0.111	0.806
Grown-Up Type	Low	0.563	0.304	0.022	0.111
	Medium-Low	0.015	0.481	0.407	0.096
	Medium-High	0.030	0.044	0.667	0.259
	High	0.044	0.104	0.126	0.726
Recessionary Type	Low	0.778	0.204	0.019	0.000
	Medium-Low	0.093	0.593	0.278	0.037
	Medium-High	0.000	0.130	0.722	0.148
	High	0.000	0.022	0.089	0.889
Regenerative Type	Low	0.417	0.250	0.028	0.306
	Medium-Low	0.028	0.472	0.444	0.056
	Medium-High	0.056	0.111	0.611	0.222
	High	0.148	0.074	0.148	0.630

presents the one-year transition probabilities, revealing several notable patterns: Low-GTP cities have a 53.2% probability of remaining in their original category after one year and a 31.7% chance of transitioning upward to the Medium-Low category. For Medium-Low-GTP cities, the probability of remaining stable is 47.2%, while the chances of shifting downward or upward are 4.8% and 35.7%, respectively. Medium-High-GTP cities are most likely to remain unchanged (71.4%), with downward and upward transition probabilities of 4.8% and 21%, respectively. High-GTP cities have a 70.4% likelihood of remaining at the top level and a 14% chance of being downgraded to the Medium-High category. Overall, these figures indicate that upward transitions between adjacent categories are generally more likely than downward shifts.

Transition dynamics vary across city types: Grown-up, recessionary, and regenerative cities largely follow the overall pattern, with a high likelihood of maintaining their current category. In contrast, growing cities exhibit more mobility—for instance, Low → Medium-Low occurs with a 44.4% probability, and Medium-Low → Medium-High with 55.6%—reflecting stronger upward potential. Notably, recessionary cities show a particularly high probability of remaining in the Low category (77.8%), which largely contributes to the overall stability of low-GTP cities.

Two main insights emerge from the analysis: First, a “club convergence” phenomenon is evident, wherein cities are generally more likely to retain their relative GTP category than to transition to another. This stability coexists with the overall growth of absolute GTP (Figure 1), meaning cities in lower categories may still experience absolute GTP increases even if their relative ranking does not change significantly. Second, upward transitions predominantly occur between adjacent categories, whereas cross-tier leaps (e.g., Low → Medium-High or High) are rare. This indicates that, while GTP improves overall, rapid or leapfrog advancement remains limited in the short term, a pattern consistent with the incremental growth of absolute GTP.

4.2. Decomposition of GTP Differences

4.2.1. Decomposition of Spatial Differences in GTP

The Gini coefficient shows a fluctuating upward trend, rising from 0.111 in 2013 to 0.157 in 2022, with an average annual growth rate of 3.94%, reflecting an increasing gap in GTP among these cities (Figure 4). By city type, the average intra-group Gini coefficients for growing, grown-up, recessionary, and regenerative cities are 0.074, 0.127, 0.164, and 0.144, respectively. Recessionary cities exhibit the greatest internal difference, mainly due to a pronounced segmentation in GTP distribution within the group. Notably, cities such as Puyang and Baishan have much lower GTP levels than other recessionary cities, creating a clear high–low gap, particularly when compared with Baiyin, Wuhai, and Luzhou. Growing cities display the smallest intra-group difference, as they are in the upward phase of resource development, characterized by high industrial homogeneity and national policies promoting early transition planning before resource depletion. Consequently, most growing cities implement similar strategies, such as advancing green mining and fostering alternative industries. Compared to other city types, growing cities share more similarities in development stages, resource endowments, and policy environments, leading to a more even distribution of GTP levels within the group. Regarding trends, except for regenerative cities, where intra-group differences narrow, the intra-group disparities for growing, grown-up, and recessionary cities generally increase.

Based on annual averages, the Gini coefficients for the six city–type pairings, ranked from highest to lowest, are as follows: recessionary–regenerative: 0.166, grown-up–recessionary: 0.156, grown-up–regenerative: 0.147, growing–recessionary: 0.131, growing–regenerative: 0.124, and growing–grown-up: 0.104 (Figure 5). Regarding trends, the inter-group differences between growing, grown-up, and recessionary cities and regenerative cities exhibit a fluctuating downward trend, with average annual decline rates of 8.28%, 2.97%, and 7.84%, respectively. This suggests a gradual convergence in GTP between regenerative cities and the other types. Conversely, differences between growing–grown-up, growing–recessionary, and grown-up–recessionary cities demonstrate a clear upward trajectory, with average annual growth rates of 9.16%, 3.02%, and 9.24%, respectively. These findings highlight the uneven evolution of green transition trajectories across different city types.

On average, intra-group differences, inter-group differences, and transvariation intensity account for 35%, 23.27%, and 41.73% of the total differences, respectively, indicating that transvariation intensity is the primary spatial contributor (Figure 6). These results align with the findings of Wang and Li [49]. Transvariation intensity reflects the degree of cross-regional overlap in GTP levels—namely, some cities with relatively low GTP are located in high-performing regions, while some high-performing cities are situated in low-performing regions. For instance, although regenerative cities generally have higher average GTP levels, cities such as Linyi and Anshan exhibit lower GTP than cities like Ordos and Nanchong in the growing type. In terms of trends, the contributions of intra-group differences and transvariation intensity show a generally increasing trajectory. By contrast, the contribution of inter-group differences declines markedly in the study period. These patterns suggest that the growing differences in GTP are mainly driven by the expansion of intra-group variation and transvariation intensity.

4.2.2. Decomposition of Structural Differences in GTP

On average, differences in greening were the dominant structural contributor, accounting for 62.52% of total disparities, followed by growth differences at 38.38% (

Table 3. Structural decomposition of GTP differences in China’s resource-based cities (2013–2022).

Year	Carbon Reduction	Pollution Control	Greening	Growth
2013	−1.28%	−0.53%	59.52%	42.29%
2014	0.66%	0.53%	48.15%	50.65%
2015	0.80%	−0.79%	58.70%	41.29%
2016	−0.34%	−0.33%	67.43%	33.25%
2017	−0.69%	−0.24%	57.16%	43.77%
2018	−0.15%	−0.49%	70.59%	30.05%
2019	−0.67%	−0.44%	71.84%	29.26%
2020	−1.52%	−1.72%	67.63%	35.61%
2021	−0.57%	−0.31%	59.50%	41.39%
2022	−0.57%	−0.36%	64.71%	36.21%

). In contrast, differences arising from carbon reduction and pollution control contributed minimally, with average rates of −0.43% and −0.47%, respectively. In terms of trends, the contribution rates of carbon reduction, pollution control, and greening differences generally increased over time, with average annual growth rates of 9.47%, 4.53%, and 0.93%, respectively. By contrast, the contribution of growth differences declined from 42.29% in 2013 to 36.21% in 2022. These results suggest that greening and growth differences remain the two most significant structural sources of GTP differences, together accounting for more than 98% of the total variation during the study period. Thus, reducing differences in greening and growth is the key to promoting a more balanced and equitable green transition among resource-based cities.

Figure 7a–d illustrates the structural decomposition of GTP differences for growing, grown-up, recessionary, and regenerative cities, respectively. The findings indicate pronounced heterogeneity in the structural sources of GTP disparities across the four city types.

For growing cities, the main sources, ranked by average contribution rate, are growth (77.8%), greening (27.43%), pollution control (3.41%), and carbon reduction (−8.64%). Notably, the contribution rates of greening and carbon reduction exhibit a fluctuating upward trend, while the contributions of growth and pollution control show a declining trend.

For both grown-up and recessionary cities, the structural sources of GTP differences are, from highest to lowest average contribution, greening, growth, carbon reduction, and pollution control. However, the evolution of contribution rates across these four dimensions differs notably. In grown-up cities, the contributions of growth, carbon reduction, and pollution control generally decline, while that of greening increases at an average annual rate of 9.6%, surpassing growth in 2015 to become the dominant contributor. In recessionary cities, the contributions of greening, carbon reduction, and pollution control exhibit a fluctuating downward trend, whereas the contribution of growth rises at an average annual rate of 9.47%, overtaking greening in 2022 as the leading structural source of GTP differences.

For regenerative cities, the structural sources of differences in GTP, ranked by average contribution rate from highest to lowest, are greening (59.2%), growth (41.79%), pollution control (−0.15%), and carbon reduction (−0.84%). Among these, the contribution rates of growth, pollution control, and carbon reduction generally show upward trends, with average annual growth rates of 3.39%, 21.72%, and 1.24%, respectively. Although the contribution of greening declines slightly over time, it consistently remains the dominant structural source of GTP differences.

Moreover, the structural decomposition of GTP differences across the four city types reveals notable commonalities. Specifically, greening and growth consistently emerge as the top two structural sources of GTP disparities, while carbon reduction and pollution control are ranked lower, with their contribution rates often being negative in most years. This pattern can be attributed to three aspects: policy uniformity, local strategic autonomy, and variations in urban development stages. China has established unified standards through policy documents such as the Implementation Plan for Synergistic Pollution and Carbon Reduction, introducing rigid regulatory mechanisms—including pollutant emission standards, carbon emission quota systems, and a national carbon market. These measures have compelled local governments to adopt similar technological pathways in pollution control and carbon emission reduction, such as retrofitting coal-fired power units and substituting clean energy. In particular, the incorporation of environmental indicators into officials’ performance evaluation systems has further driven the “bottom-line convergence” in the dimensions of pollution and carbon reduction [50,51]. In contrast, greening and growth are not governed by a national unified standard and are more influenced by local resource endowments, industrial foundations, and transformation strategies. For instance, some recessionary cities (e.g., Fuxin) have developed cultural and tourism industries through ecological mine restoration, while regenerative cities (e.g., Daqing) focus on new energy projects rooted in emerging industries. These locally differentiated strategies heighten inter-city differences. In summary, policy uniformity promotes standardized governance in pollution control and carbon reduction, while local resource conditions, industrial inertia, and development stage differences lead to significant divergence in greening and growth, driven by varied strategic choices. This highlights both the efficacy of national strategies in environmental governance and the dialectical relationship between “common requirements” and “individual development” in the green transition of resource-based cities.

4.3. Cause Identification

4.3.1. Variables Selection and Description

The GTP of resource-based cities is influenced by a complex interplay of socio-economic factors, with spatial disparities among these factors serving as key drivers of GTP variation. Drawing on theories such as the new economic geography theory and techno-economic paradigms, along with prior research, this study examines the determinants of GTP differences from the perspectives of factor supply, structural evolution, and institutional environment, as shown in Figure 8. Based on this framework, this study employs the OPGD to examine the individual and interactive effects of these factors, and, in combination with the GWTR model, it further uncovers the spatiotemporal heterogeneity of these influences, thereby offering a comprehensive explanation of the mechanisms underlying disparities in GTP across resource-based cities.

The influencing factors in factor supply include digital economy, the level of financial development, and human capital. According to production factor theory, these correspond to technological, capital, and labor inputs, respectively, and together constitute the material foundation for green transition.

(1) Digital economy. Digital economy advances the green transition of energy-intensive industries through technologies such as digital sensing, big data analytics, and artificial intelligence, and also supports the development of emerging green industrial clusters, including smart energy and environmental protection technologies [52]. Furthermore, digital tools enhance the accuracy of environmental regulation and policy implementation—for instance, blockchain technology ensures the traceability of carbon emissions, while integrated land–air–space monitoring systems facilitate pollution source tracking and early warning. This study measures the level of digital economy development using a digital economy index calculated via the entropy method.

(2) Level of financial development. A well-functioning financial system reallocates capital from traditional high-carbon sectors to green industries. Instruments such as green credit and carbon finance provide targeted support for ecological restoration, the deployment of clean technologies, and the growth of emerging industries, thereby alleviating the financing constraints of green transition [53]. Financial development is measured by the ratio of year-end financial institution deposits and loan balances to GDP.

(3) Human capital. Highly educated laborers tend to exhibit stronger awareness of ecological issues. These individuals contribute to clean technology innovation and the low-carbon transformation of industries through their expertise while also serving as behavioral exemplars to promote green consumption and sustainable lifestyles. Human capital is represented by the number of university students per 10,000 people.

The structural evolution dimension includes two key influencing factors: industrial structure and urbanization level. According to structuralism development theory, these factors reflect the evolution of economic and spatial structures, respectively, and collectively shape the green metabolic capacity of socio-economic systems.

(1) Industrial structure. The upgrading of the industrial structure facilitates green transition through factor substitution and technological diffusion. In this study, it is quantified by the ratio of value added in the tertiary sector to that in the secondary sector. On the one hand, the service sector can substitute high-energy-consuming segments of manufacturing by employing knowledge-intensive production, thereby reducing resource consumption per unit of output. On the other hand, when producer services are integrated into the manufacturing value chain, digital coordination can accelerate the diffusion of clean technologies, rebuild closed-loop systems of “R&D–production–recycling,” and enable systemic reductions in pollutant emissions [54].

(2) Urbanization level. Urbanization may exert both positive and negative effects on GTP. This study uses the nighttime light index as a proxy for urbanization. Positively, agglomeration and scale effects associated with urbanization can enhance resource allocation efficiency; for instance, shared infrastructure reduces energy intensity, and centralized pollution management lowers ecological costs. Conversely, the concentration of population and industry can increase the rigid demand for high-carbon infrastructure such as transportation and buildings. Additionally, the expansion of construction land may encroach upon ecological space, damage natural carbon sinks, and diminish ecosystem service functions [55].

The institutional environmental dimension encompasses three key influencing factors: government support, public environmental concern, and the degree of openness. According to the viewpoint of new institutional economics, government support provides formal institutional guarantees, public environmental concern acts as an informal institutional constraint, and openness promotes cross-scale institutional interaction. Collectively, these factors form a multi-level institutional incentive framework for green transition.

(1) Government support. Governments facilitate green transition through policy guidance and resource coordination. In this study, government support is measured by the ratio of fiscal expenditure to GDP. On the one hand, instruments such as ecological compensation and strategic planning address market failures, while targeted transfer payments and investments in green infrastructure alleviate the long-term financial constraints of green transition [56]. On the other hand, as coordinators among diverse stakeholders, governments can utilize cross-regional ecological agreements and collaborative mechanisms involving the public, private, and civic sectors to transform fragmented environmental governance into integrated, systemic action [57].

(2) Public environmental concern. Public sensitivity to ecological issues can exert pressure through public opinion, prompting governments to strengthen environmental regulation and enterprises to fulfill their environmental responsibilities, thereby accelerating pollution control and ecological restoration [58]. Furthermore, increasing preferences for green consumption and environmental participation foster the growth of low-carbon product markets, motivating enterprises to innovate green technologies and improve production practices. This study adopts the Baidu Index as a proxy for public environmental concern.

(3) Degree of openness. In line with the “pollution halo” and “pollution haven” hypotheses, the degree of openness may exert both beneficial and adverse impacts on green transition. In this study, the degree of openness is quantified by the ratio of total imports and exports to GDP. On the one hand, openness facilitates the adoption of advanced technologies and management practices, and participation in global value chains encourages firms to align with international environmental standards, thereby promoting green development [59]. On the other hand, the lock-in effect of international division of labor may increase carbon leakage risks, as some regions attract the relocation of energy-intensive industries, potentially becoming “pollution havens.”

4.3.2. Results of the OPGD for Single-Factor Detection

The results of the OPGD are shown in

Table 4. Geographical detection results of influencing factors.

Variable Name	p Value	Significance Level	q-Value	q-Value Ranking
Digital Economy	0.000	1%	0.046	1
Level of Financial Development	0.058	10%	0.031	2
Human Capital	0.713	—	0.006
Industrial Structure	0.007	1%	0.022	4
Urbanization Level	0.005	1%	0.021	5
Government Support	0.821	—	0.011
Public Environmental Concern	0.333	—	0.010
Degree of Openness	0.001	1%	0.029	3

. Among the eight independent variables, four are significant at the 1% level and one at the 10% level. Digital economy has the highest q-value, indicating that it is the most influential factor driving the GTP of resource-based cities. Such cities are often confronted with challenges including rigid industrial frameworks and ecological shortfalls, and, through technological penetration, digital economy directly addresses the inefficiencies and pollution associated with traditional industries, stimulates the development of green substitute sectors, and reduces reliance on resource-based revenues. Its features—light asset intensity and high coordination—enable it to bypass the capital and talent constraints common in resource-based cities. By leveraging digital infrastructure, it can quickly activate local green innovation networks, offering a breakthrough path for transformation under multiple constraints and fostering a process of “creative destruction” to advance green transition.

The level of financial development ranks second in terms of its q-value, highlighting its importance in promoting the green transition of resource-based cities. These cities often encounter constraints such as funding shortages, high risks in phasing out traditional industries, and significant ecological restoration costs. Financial development helps alleviate local fiscal pressures by attracting external capital through green credit, mitigates transition risks via diversified insurance mechanisms, and guides resource-dependent investments toward green industries through multi-tiered capital markets.

The other three variables that passed significance tests are degree of openness (q = 0.029), industrial structure (q = 0.022), and urbanization level (q = 0.021). Although their explanatory power is somewhat weaker than that of digital economy and the level of financial development, they nonetheless play important supporting roles in advancing the green transition of resource-based cities.

4.3.3. Results of the Spatiotemporal Heterogeneity of Influencing Factors

To further examine the spatiotemporal heterogeneity of factors driving the GTP of resource-based cities in China, this study employs the GTWR model to analyze the five variables that passed significance testing. Multicollinearity diagnostics indicate that all variables have variance inflation factors (VIFs) below 10, confirming the absence of multicollinearity issues. The GTWR model yields an AICc of −1266.73 and an adjusted R² of 0.62, exceeding the corresponding values from the OLS and GWR models (

Table 5. Parameter estimation and diagnostic results of the GTWR model.

Regression Model	AICc	R2	Adjusted R2
OLS	−596.78	0.218	0.214
GWR	−1267.05	0.602	0.601
GTWR	−1266.73	0.621	0.620

), thereby demonstrating superior model fit due to the integration of spatiotemporal features. ArcGIS is used to visualize the spatial distribution of the five influencing factors in 2013 and 2022, revealing notable spatial variation and temporal shifts in their respective effects.

Impact of digital economy on GTP (Figure 9). Temporally, digital economy exerted a suppressive influence on the green transition of most resource-based cities in 2013. This was mainly due to the early-stage development of China’s digital economy, during which the construction of digital infrastructure consumed substantial amounts of fossil energy, intensifying carbon emissions and environmental pollution. By 2022, digital economy’s positive influence on green transition had significantly increased. This shift reflects the adoption of a national strategy that designates digitalization as a key driver of green development, alongside the nationwide deployment of new digital infrastructure, which facilitated the deep integration of digital technologies into green industrial chains. Spatially, the positive effect is concentrated in cities across South China, Central China, and parts of Northwest China. For instance, Nanping in Fujian has capitalized on its coastal digital industry advantages to support an integrated green industrial chain, while Tongling in Anhui has promoted the upgrading of traditional industries through digital regulation. In contrast, several resource-based cities in Northeast and North China exhibit a suppressive effect, where digitalization efforts in heavy industries remain limited to the production phase. In these cases, digital investment has crowded out resources for green R&D, resulting in a “pseudo-transition” scenario.

Impact of the level of financial development on GTP (Figure 10). Temporally, the suppressive effect of financial development on the green transition of resource-based cities has intensified over time. This trend stems from the cities’ long-standing dependence on a “resource mortgage–loan” model. As resource revenues decline, banks tighten credit, and financial institutions tend to favor traditional industries to maintain economic stability, thereby constraining investment in green initiatives. Despite national efforts to promote green credit policies, financial sectors in many resource-based cities remain underdeveloped, hindering the implementation of innovative instruments such as carbon finance and resulting in “nominal greening” [60]. Spatially, the positive effects of financial development have contracted from broader areas of the southwest and northeast to only a few cities in the northeast. For example, cities such as Anshan and Daqing have consistently benefited from policy-driven financial instruments, sustaining a positive influence on green transition. In contrast, some cities in North China and the southwest are caught in a “financial suppression trap,” where financial systems are deeply embedded in traditional industries, and the high cost of financing green projects continues to impede their development.

Impact of industrial structure on GTP (Figure 11). Temporally, the positive influence of industrial structure on the green transition of resource-based cities has broadened in geographic scope but weakened in intensity. In the early stages, the rapid upgrading of extensive industries generated considerable green dividends. However, as the barriers to transformation rise, structural optimization increasingly depends on the depth of industrial integration, resulting in diminishing marginal returns from industrial upgrading. Spatially, regions such as Northwest and South China have achieved positive outcomes by coordinating the development of high-value-added tertiary industries with the decarbonization of secondary industries. For instance, Ordos has advanced the integration of wind, solar, and hydrogen storage and ecotourism, producing notable achievements in green development. In contrast, some cities in Northeast and Southwest China have experienced suppressive effects, as their tertiary sectors remain locked in low-end development and disconnected from secondary industries. This disconnect impedes the formation of a mutually reinforcing model of industrial integration, thereby constraining green transition efforts.

Impact of urbanization level on GTP (Figure 12). Temporally, the driving effect of urbanization on GTP has generally declined. At the initial stage, most resource-based cities experienced rapid expansion. Urbanization stimulated economic growth through population concentration and infrastructure investment, which concealed the environmental costs of traditional industries and fostered an extensive but temporarily effective transition pathway. Over time, however, the adverse effects of conventional urbanization—such as industrial path dependence and increased environmental stress—have become more pronounced, weakening its driving force. Spatially, the suppressive impact of urbanization on GTP persists in North and East China, while South and Southwest China demonstrate a continued positive effect. In contrast, Northeast and Northwest China show internal divergence. This spatial variation is primarily driven by differences in the compatibility between urbanization, local industrial structures, and ecological systems. In North and East China, heavy industries are tightly bound to urban land use, resulting in a dilemma where expansion leads to pollution. In Northwest China, ecological thresholds constrain urban development, and overexpansion undermines the capacity for green transition. Conversely, South China and Southwest China have adopted a “city-driven greening” model supported by low-energy infrastructure and favorable policy measures.

Impact of the degree of openness on GTP (Figure 13). Temporally, both the positive effect and spatial degree of openness on the GTP of resource-based cities have shown a consistent upward trend. This improvement is primarily driven by openness policies—such as the Belt and Road Initiative—that enhance trade and investment facilitation, foster international cooperation and technology transfer, and promote the adoption of green standards, thereby reinforcing the momentum for green transformation. Spatially, openness exerts increasingly positive effects in North, East, and South China, where robust economic foundations, strong innovation capabilities, and stringent environmental regulations create a virtuous cycle of foreign investment spillovers and industrial upgrading. In contrast, cities in Northeast, Northwest, and Southwest China face limitations due to resource dependency and low levels of marketization. In these regions, weak enforcement of environmental regulations during industrial relocation has led to “pollution haven” effects. Additionally, unfavorable geographic conditions and high logistics costs further restrict the absorption of advanced technologies, thereby impeding green transition efforts.

4.3.4. Results of the OPGD for Factor Interaction Detection

Given the potential synergistic and antagonistic relationships among influencing factors, this study further explores the interactions among five variables that passed the significance test: digital economy, level of financial development, industrial structure, urbanization level, and degree of openness, labeled S₁ to S₅, respectively. The results (

Table 6. Interaction detection results of influencing factors.

A∩B	A + B	Comparison Results	q-Value Ranking
S1∩S2 = 0.132	S1(0.046) + S2(0.031) = 0.077	A∩B > A + B	6
S1∩S3 = 0.124	S1(0.046) + S3(0.022) = 0.068	A∩B > A + B	9
S1∩S4 = 0.153	S1(0.046) + S4(0.021) = 0.067	A∩B > A + B	1
S1∩S5 = 0.134	S1(0.046) + S5(0.029) = 0.075	A∩B > A + B	5
S2∩S3 = 0.140	S2(0.031) + S3(0.022) = 0.053	A∩B > A + B	4
S2∩S4 = 0.145	S2(0.031) + S4(0.021) = 0.052	A∩B > A + B	2
S2∩S5 = 0.145	S2(0.031) + S5(0.029) = 0.060	A∩B > A + B	2
S3∩S4 = 0.118	S3(0.022) + S4(0.021) = 0.043	A∩B > A + B	10
S3∩S5 = 0.129	S3(0.022) + S5(0.029) = 0.051	A∩B > A + B	8
S4∩S5 = 0.132	S4(0.021) + S5(0.029) = 0.050	A∩B > A + B	6

) show that the explanatory power of each pairwise interaction exceeds the combined explanatory power of the individual variables. All interaction pairs demonstrate nonlinear enhancement effects, indicating that differences in GTP across China’s resource-based cities stem not from the independent, direct effects of individual factors, but from amplified outcomes arising from their interactions (Figure 14).

Based on Table 6 and Figure 14, it can be seen that four dominant interaction combinations significantly influence the GTP of China’s resource-based cities (q-values ≥ 0.14). Among these, the interaction between urbanization level and the level of financial development is the strongest, with a q-value of 0.153. Although urbanization level exhibits the weakest individual explanatory power in the factor detection analysis, its combined effect with financial development substantially enhances its impact on GTP. This suggests that urbanization level alone may contribute little to green transition—due to challenges such as industrial monocultures and inadequate public services—but, when integrated with the level of financial development, mechanisms like green credit and risk sharing can convert accumulated human and infrastructural capital into momentum for green investment and innovation. This underscores the importance of coordinating urbanization and financial policies in resource-based cities, for example, by channeling capital into low-carbon technology R&D and ecological restoration via industry–finance integration. Furthermore, digital economy forms three additional dominant interaction combinations with the level of financial development, degree of openness, and industrial structure, yielding q-values of 0.145, 0.145, and 0.14, respectively. Although digital economy serves as a core driver of green transition by promoting technological diffusion and efficiency, it also faces constraints such as high R&D costs, limited factor mobility, and inadequate industrial alignment. Interaction with other factors mitigates these challenges; financial development eases funding constraints, openness accelerates the adoption of green technologies and market expansion, and industrial upgrading offers fertile ground for digital application. These synergies significantly strengthen digital economy’s contribution to GTP, highlighting the co-evolution of core and supporting elements in the green development process.

5. Discussion

This study is anchored in two key national strategies: the all-round green transformation of economic and social development and the advancement of coordinated regional development. It systematically investigates the multidimensional sources and driving mechanisms underlying the disparities in GTP among China’s resource-based cities, with the aim of offering strategic insights to boost the synergy of green transition efforts and mitigate environmental welfare inequality.

To advance the evaluation of GTP, this study moves beyond traditional single-indicator approaches, such as green GDP or GTFP [8,9,10], to develop a comprehensive assessment framework encompassing four key dimensions: carbon reduction, pollution control, greening, and growth. The framework is closely aligned with national strategic objectives, thereby boosting both the policy relevance and practical usability of the indicator system, and offers a standardized metric for assessing the progress of green transition in resource-based cities. Importantly, this framework can help guide governments at all levels in fostering healthy competition and coordinated development, thereby facilitating the effective implementation of integrated efforts across all four dimensions.

To analyze the sources of disparities in GTP, this study applies the Dagum Gini coefficient and variance decomposition methods to explore both the spatial and structural sources and evolution of these disparities, revealing the dual nature of such performance gaps; they stem not only from spatial imbalances among cities but also from structural inequalities across the four dimensions. Notably, greening accounts for over 60% of the total disparity, underscoring the vital role of green space development in advancing GTP. In contrast to prior research that primarily emphasized geographic regional differences [20,21], this study highlights the systemic imbalance among performance dimensions, offering fresh insights for designing more targeted and differentiated governance policies.

To identify the cause of disparities in GTP, this study selects influencing factors from three dimensions: factor supply, structural evolution, and the institutional environment. A logically structured analytical framework is developed to examine the driving mechanisms, effectively avoiding the arbitrariness often associated with variable selection. The findings suggest that factors such as digital economy, the level of financial development, the degree of openness, industrial structure, and urbanization level significantly affect GTP disparities. However, the strength and direction of these effects vary markedly across time and regions, emphasizing the need for place-based and differentiated policy interventions. Interaction detection further reveals that combinations of factors exhibit nonlinear enhancement effects. For example, the synergy between digital economy and the level of financial development significantly enhances GTP. This finding complements existing studies that primarily emphasize linear causal pathways [61,62], underscoring the importance of considering policy synergies from a complex systems perspective.

From the perspective of aligning the analytical scale with the research objectives, this study focuses on the macro-level driving mechanisms underlying the differences in GTP among resource-based cities, with particular emphasis on the role of digital economy, level of financial development, degree of openness, industrial structure, and urbanization level. Nevertheless, it should be noted that micro-level factors are equally valuable, both theoretically and practically, constituting another important dimension for understanding transition disparities. Specifically, at the enterprise level, green business negotiations (e.g., low-carbon actions in inter-firm or government–enterprise collaborations) and sustainable business transformation (e.g., ESG implementation and the upgrading of green business models) are key vehicles for promoting micro-level green behavior. Existing studies typically anchor these topics at the firm level, emphasizing the green decision-making logic and implementation pathways of individual organizations or firm clusters [63,64,65,66]. At the individual level, financial literacy (which influences residents’ capacity for green investment and consumption decisions) and household carbon footprint (which reflects the environmental impact of daily behaviors) play a more targeted role. Their core value lies in guiding green consumption preferences and shaping the social structure of green demand, thereby indirectly influencing the process of urban green transition [67,68,69]. Although these micro-level dimensions are not included in our analysis, acknowledging their potential influence provides a more nuanced insight into the mechanisms shaping the disparities in the GTP of resource-based cities.

6. Conclusions and Policy Implications

6.1. Conclusions

Anchored in China’s newly proposed green transition action framework, this study develops an all-encompassing system for evaluating the GTP of resource-based cities across four dimensions: carbon reduction, pollution control, greening, and growth. Based on GTP calculated via the entropy method, this study applies the Dagum Gini coefficient and variance decomposition to identify the spatial and structural sources of GTP differences. Furthermore, it integrates the OPGD and the GTWR model to uncover the principal drivers of GTP variation. The main findings are as follows: (1) The GTP of resource-based cities demonstrates a generally upward but fluctuating trend, alongside pronounced spatial heterogeneity across regions. A clear “club convergence” pattern is observed, with cities tending to transition positively toward adjacent categories. (2) Differences in GTP have increased over time, with transvariation density emerging as the dominant spatial contributor. The intra-group differences among recessionary, regenerative, grown-up, and growing cities decrease in that order, while the inter-group differences between recessionary and regenerative cities are greater than those among other city group pairings. (3) Among the structural sources of disparity, greening contributes the most, followed by growth, carbon reduction, and pollution control. Greening alone accounts for over 60% of the annual structural variation, underscoring its dominant role. (4) The driving factors of GTP differences, ranked from most influential to least influential, are digital economy, the level of financial development, degree of openness, industrial structure, and urbanization level, exhibiting marked spatiotemporal heterogeneity. All factor interactions show nonlinear enhancement effects, indicating that GTP differences arise from the synergistic interplay of multiple factors rather than isolated influences.

6.2. Policy Implications

Based on the above findings, this study puts forward the following recommendations:

First, given the relatively weak performance in carbon reduction and greening, the government should fully implement the action framework of the coordinated promotion of “Carbon Reduction–Pollution Control–Greening–and Growth.” For carbon reduction, highly energy-intensive industries should be required to take the lead in meeting emission reduction targets, while key enterprises should disclose carbon emission data and implement mandatory reporting systems, thereby accelerating industrial low-carbon upgrading. For greening, progress can be achieved through measures such as strengthening the ecological compensation system, promoting integrated urban–rural greening, and accelerating the construction of green infrastructure. By addressing the shortcomings in carbon reduction and greening, the overall performance of the green transition can be substantially improved.

Second, in response to spatial disparities and the “club convergence” phenomenon in green transition, governments should encourage regenerative cities to play a leading role by sharing their advanced experiences with lagging regions and establishing cross-regional partnership and assistance mechanisms. Through measures such as green technology transfer and joint development of industrial chains, growing and grown-up cities can be supported in their transition. Meanwhile, in lagging regions, pilot projects for green transition should be launched, focusing on industrial restructuring, renewable energy development, and green technology transformation, with pilot cities granted priority policy support. By combining external “blood transfusion” assistance with internal “blood making-style” empowerment, a complementary and phased pattern of regional collaborative development can be fostered.

Third, given that structural disparities in GTP among the four types of resource-based cities exhibit both shared and distinct features, greening should be promoted as a foundational initiative at the common level, with increased investment in public green space, ecological restoration, and green spatial governance through central fiscal transfers and green bonds. Simultaneously, the quality and quantity of economic growth should be enhanced to provide a robust material and technological basis for green transition. At the differentiated level, each type of city should develop tailored strategies aligned with its structural decomposition results, adhering to the principle of context-specific and category-based policies and exploring targeted and fine-grained approaches to narrowing structural gaps.

Finally, given the spatiotemporal heterogeneity and nonlinear amplification effects of influencing factors, local governments should tailor policies to regional realities and identify dominant drivers with precision. For example, leveraging digital economy can improve resource allocation efficiency and intelligent environmental governance; financial development can guide green investment and credit; openness can facilitate the introduction and diffusion of advanced low-carbon technologies and green standards; and industrial restructuring and urbanization upgrading can serve as opportunities to promote industrial greening and low-carbon lifestyles. Moreover, when formulating green transition policies, potential synergies and trade-offs among influencing factors should be pre-assessed. By aligning factor interactions with regional weaknesses in green transition, governments can maximize integrated policy effects and ensure sustainable progress in green transition.

6.3. Limitations and Future Research

This study recognizes three primary limitations that merit further investigation: (1) Our assessment of GTP relied predominantly on official statistical yearbook data. Although authoritative, these data are limited in their ability to capture micro-level behavioral shifts and the dynamic environmental responses occurring throughout the transition process. Future research could improve measurement accuracy by incorporating multi-source heterogeneous data, such as remote sensing imagery, corporate carbon emissions, ESG disclosures, and household behavioral information. (2) This study focused on resource-based cities, with influencing factors primarily selected from macro-level variables. Consequently, micro-level factors—such as green business negotiations, sustainable business transformation, financial literacy, and carbon footprint—were not included in the analysis. Current research indicates that green business negotiations and sustainable business transformation are typically examined at the firm or micro level, whereas financial literacy and carbon footprint mainly affect residents’ green behaviors and consumption decisions. Future studies could adopt a micro-level perspective, integrating firm- and resident-level data to examine how these factors influence variations in green transition performance in resource-based cities, thereby providing more precise and actionable evidence for policymaking. (3) This study adopted the OPGD and the GTWR model to identify the drivers of GTP differences and their spatiotemporal heterogeneity. Nevertheless, these models have limitations; the OPGD assesses factor significance based on spatial distribution alignment but cannot establish causal relationships between variables, while the GTWR model captures spatiotemporal variations in driving factors but cannot elucidate their pathways or transmission mechanisms. Future research could combine approaches such as instrumental variables, difference-in-differences, and mediation analysis to uncover the causal effects and mechanisms of these drivers, providing a more systematic understanding of variations in GTP.