Spatial and Temporal Characteristics and Regional Difference in China’s Provincial Green Low-Carbon Development

Abstract

Since the 18th National Congress of the Communist Party of China in 2012, green and low-carbon development has become a national strategic priority. This study constructs a 39-indicator evaluation system grounded in the DPSIRM framework, which includes six interlinked subsystems. A key innovation lies in incorporating the Digital Inclusive Finance Index as a driver of green transitions and using Baidu search indices for “environmental protection” and “carbon dioxide” as proxies for public awareness. Using a projection pursuit model optimized by simulated annealing, we assess green low-carbon development across 30 Chinese provinces from 2011 to 2021. Temporal and spatial patterns are analyzed via kernel density estimation and Moran’s I, while Theil Index decomposition quantifies regional disparities. Results: First, substantial variations exist among Chinese provinces in both subsystem performance and integrated green low-carbon development levels, and response subsystems have the greatest influence on the overall development level. Second, over time, the gaps in green, low-carbon development between provinces have become more pronounced. Third, geographically, a distinct east-to-west declining gradient characterizes the regional clustering patterns of green low-carbon development. Fourth, the Theil Index for green, low-carbon development exhibits an overall trend of fluctuating increase, indicating that the overall gap in green, low-carbon development is gradually widening, with within-group disparities as the primary cause. This research enhances understanding of China’s green and low-carbon development, actively promoting global sustainable development and environmental improvement.

1. Introduction

Amid accelerating global warming, curbing emissions and steering toward low-carbon trajectories have become urgent worldwide tasks. As the world’s largest developing economy, China has advanced an assertive climate agenda anchored in the “dual-carbon” commitments: peaking carbon output by 2030 and attaining carbon neutrality by 2060. Recent statistics reveal that national emissions stabilized at roughly 12.6 billion tons in 2024, mirroring 2023 levels; however, a nuanced structural reconfiguration is underway—energy-related releases edged up by 0.4%, whereas industrial-process discharges fell by 5%. Responsible for about one-third of global anthropogenic carbon, China recorded a simultaneous fall in carbon intensity and continued economic expansion, highlighting both achievements and enduring hurdles. Green and low-carbon development constitutes a pivotal lever for delivering these dual-carbon ambitions. Echoing this priority, the 20th CPC National Congress in October 2022 emphasized that “advancing green and low-carbon socioeconomic transformation is central to high-quality development.” Likewise, Minister of Ecology and Environment Huang Runqiu’s 2024 report outlined post-14th Five-Year Plan milestones: accelerating green and low-carbon progress by 2027, cutting major pollutant loads, elevating ecological quality, reinforcing ecosystem services, markedly enhancing urban-rural living conditions, safeguarding national ecological security, sharpening environmental governance frameworks, piloting replicable models, and securing visible gains in Beautiful China construction. Against this backdrop, rigorously evaluating provincial green low-carbon performance is indispensable for crafting precision policies and expediting the national transition [1,2].

This study aims to construct a comprehensive evaluation system for green low-carbon development based on the DPSIRM framework, quantify development levels across 30 Chinese provinces using the simulated annealing-projection pursuit (SA-PP) model, and reflect spatiotemporal distribution characteristics and regional disparities through kernel density estimation, Moran’s Index, and Theil Index. The significance of this research lies in furnishing empirical evidence to aid policymakers in addressing regional imbalances, optimizing resource allocation, and advancing the ‘dual-carbon’ agenda. Methodologically, it extends the application of the DPSIRM framework within sustainability science.

The rest of this paper is arranged in the following structure. Section 2 presents a literature review on relevant studies. Section 3 details the indicator system, data sources, and research methods, followed by the measurement results in Section 4. Section 5 discusses the temporal and spatial characteristics, and Section 6 analyzes the regional disparities. The last section summarizes the whole paper.

2. Literature Review

2.1. Measurement Method and Index Selection

Regarding research evaluating green and low-carbon economic transitions, authoritative international research has primarily used the Environmental Performance Index (EPI) produced by Yale University [3] and Columbia University, the green development evaluation indicator system developed by the World Bank [4], and the green growth measurement indicator system published by the Global Green Growth Institute (GGGI) [5]. Existing evaluations predominantly employ multi-criteria frameworks (e.g., AHP, the entropy weighting method [6] and PCA) to construct indicator systems from diverse analytical angles. For instance, Xiang and Zheng (2013) proposed the China Green Economy Development Index from green production, green consumption, and ecological health perspectives, and validated this index using data from the 11th Five-Year Plan period [7]. Zhang et al. (2016) constructed a measurement indicator system for quantifying green development from three dimensions of green and beautiful homes, green production and consumption, and green high-quality development [8]. Xu et al. (2021) assessed the overall level of green economic development in 30 Chinese provinces from 2013 to 2018, considering social development, economic efficiency, innovation drive, ecological construction, benefits for the people, and fairness [9]. This study employs the simulated annealing-optimized projection pursuit (SA-PP) model. Unlike AHP, which relies on expert judgments, or PCA, which assumes linearity, SA-PP derives weights through a global optimization process, minimizing subjective bias and accommodating non-linear interactions. By combining PP for dimensionality reduction and SA for global optimization, it effectively handles high-dimensional, non-linear sustainability data characteristic of green low-carbon development systems.

Scholars have constructed evaluation systems based on theoretical frameworks such as the pressure–state–response (PSR), driving force–state–response (DSR), and driving force–pressure–state–impact–response (DPSIR) models. The European Environment Agency (EEA) proposed the DPSIR model [10], which expanded upon earlier frameworks by categorizing indicators into driving forces, pressures, state, impacts, and responses. The DPSIR framework advances prior models by explicitly capturing bidirectional human–environment feedback loops [11]. Despite applications in ecological security and urban sustainability assessments, debates persist over indicator selection and systemic coherence [12]. Yang et al. (2012) introduced the DPSIRM model by incorporating “management,” forming a six-element framework: driving forces, pressures, state, impacts, responses, and management [13]. This enhanced framework provides a more holistic analytical structure for addressing complex environmental issues. The DPSIRM model has been applied in various fields, including ecosystem health assessments [14,15], water resource carrying capacity evaluations [16,17], industrial sustainability analysis [18], and human settlement security assessments [17]. However, its application in evaluating green low-carbon development levels remains undocumented. Distinguishing this study from prior work, we construct a DPSIRM-based indicator system. Key innovations include (1) introducing the Digital Inclusive Finance Index as a novel ‘driving force’ indicator, capturing the role of digital finance in enabling green transitions; and (2) incorporating Baidu search indices for ‘environmental protection’ and ‘carbon dioxide’ as ‘impact’ indicators. Thus, we provide real-time proxies for public awareness and engagement—aspects often underrepresented in traditional metrics.

2.2. Spatiotemporal Evolution and Regional Disparities

Kernel density estimation, as a nonparametric method, was used to characterize the dynamic distributions of economic and environmental indicators [19]. Spatial autocorrelation analysis holds a significant position in research on the spatial distribution of green low-carbon development. Moran’s Index is commonly employed to measure the degree of clustering or dispersion of spatial elements [20,21,22,23,24,25]. The Theil Index possesses additivity, allowing clear decomposition into within-group and between-group disparities, which facilitates the identification of dominant sources of differences. Numerous studies utilize the Theil Index to delineate development gaps within and between regions across various dimensions [26,27,28,29]. This decomposition enables the further quantification of factors’ contributions to overall disparity—providing more interpretable policy insights than Dagum Gini coefficient decomposition, which includes a transvariation term for distributional overlaps.

Existing studies have frequently employed these techniques in isolation, yet few have attempted to synthesize them into a unified analytical framework. To address this gap, the present study integrates these three methodologies synergistically, establishing a comprehensive evaluation system for green and low-carbon development, synergistically applying (1) kernel density estimation to visualize temporal distribution dynamics and polarization trends; (2) Global/local Moran’s Index to diagnose spatial clustering patterns; and (3) Theil Index decomposition to quantify regional disparity.

2.3. Comparison with Previous Studies

In summary, previous research has employed various methods to quantify and explore green and low-carbon development, providing a solid foundation for this study; however, several considerations require further in-depth exploration. First, in determining the weights of various indicators, some studies have used subjective weighting methods that have notable subjective arbitrariness. Second, when constructing indicator systems, the majority of the existing literature has focused on traditional characteristics and factors, neglecting the role of digital technology as an emerging driver of green and low-carbon development, in addition to the significant value of public sentiment in measuring green and low-carbon development. This makes it difficult to accurately reflect the current configuration of green and low-carbon development in China. Moreover, most studies have primarily considered the completeness of the content covered by indicator systems while neglecting the internal correlations within the indicator system. As a classic causal network model, the driving force–pressure–state–impact–response–management (DPSIRM) model can scientifically reflect internal causal relationships but is rarely used in green and low-carbon development indicator systems. Finally, previous research has not sufficiently used and analyzed measurement results, meaning we are lacking combined analysis of regions and areas, in addition to the absence of analysis and research on spatial and temporal distribution characteristics and regional disparities.

In contrast to previous research, this study offers three possible marginal contributions. First, we employ the DPSIRM model, which can reflect internal causal relationships, to construct an evaluation indicator system for green and low-carbon development from driving force, pressure, state, impact, response, and management perspectives, presenting more comprehensive, scientific, and reasonable evaluation. This approach complements previous methods for constructing green and low-carbon economic evaluation systems and broadens the application scope of the DPSIRM evaluation indicator system. Second, this study introduces the Digital Inclusive Finance Index, as a driving force indicator, and environmental protection and carbon dioxide (CO₂) Baidu indices, as impact indicators, into the indicator system, further refining the DPSIRM model as a more comprehensive indicator system for quantifying green and low-carbon development with contemporary characteristics. Third, we employ the simulated annealing algorithm to optimize a projection pursuit model for the comprehensive evaluation of green and low-carbon development in each province and use multiple methods to conduct in-depth analysis of the evaluation results from provincial and regional perspectives, including spatial and temporal distribution characteristics and gap analysis across regions, providing a valuable reference for future green and low-carbon development in various regions of China.

3. Indicator System, Data Sources, and Research Methods

3.1. Theoretical Framework

Building upon the DPSIRM model, this study establishes a systematic analytical framework for green and low-carbon development levels by analyzing the interrelationships of six constituent elements.

Driving force (D) refers to the causes that result in system changes, including population, economic growth, and technological shifts. Digital Inclusive Finance is a new financial service model that is enabled by digital technology [30]. On the one hand, digital technologies facilitate the adjustment of energy structure, promote the adoption of renewable energy, and reduce dependence on fossil fuels [31]. On the other hand, digital financial inclusion promotes the economic transition to green and low-carbon by supporting enterprises’ green technological innovation [32,33]. It advances the virtuous cycle of green production–increased revenue–low-carbon development. These mechanisms align with the ‘driving force’ dimension in the DPSIRM framework, which uses indicators reflecting socioeconomic catalysts for systemic change. Consequently, the Digital Inclusive Finance Index is introduced to quantify digital finance’s transformative impact on green and low-carbon development. In contrast, the natural population growth rate, as a negative indicator, underscores the strain of demographic expansion on resource carrying capacity.

Pressure (P) represents the direct effects of driving forces acting on the system, manifesting as resource depletion or pollutant emissions resulting from human activities. This paper employs negative indicators such as carbon dioxide emissions and ammonia–nitrogen discharge levels to quantify the systemic environmental burden imposed by the industrialization process.

State (S) reflects the actual physical, biological, and chemical status of the system under current pressure. Positive indicators such as average daily precipitation and forest coverage rate demonstrate ecological resource endowments, whereas negative indicators like average PM2.5 concentration reveal air quality deterioration [34].

Impact (I) refers to the multidimensional consequences of environmental changes on health, economy, and society. Positive indicators such as grain yield and value added in agriculture, forestry, animal husbandry, and fishery reflect the value of ecological services, while negative indicators such as forest disease-affected areas and population mortality rates indicate risks of environmental degradation. Additionally, web search data based on internet queries—which record user behaviors—can capture real-time market attention to specific events, revealing preferences and behavioral intentions [35,36]. As China’s leading search engine analytics tool similar to Google Trends, Baidu Index measures public attention by quantifying keyword search volumes [37]. As the largest Chinese search engine, Baidu has extensive coverage, and its search frequency can partially reflect public attention. Therefore, we incorporate the Baidu Seach Index for environmental protection and also incorporate carbon dioxide into the impact subsystem.

Response (R) refers to the policies and measures adopted by societies or governments to relieve pressure and improve the state. For example, indicators such as the area of forest pest control and the number of industrial wastewater treatment facilities are used to characterize forest pest control measures such as pesticide spraying and technical means such as the operation of industrial wastewater treatment facilities.

Management (M) involves the regulation and optimization of the entire process. Through indicators such as investment in industrial pollution control and full-time equivalent R&D personnel, the closed loop of governance is optimized to promote the transformation of policy response into long-term effect.

This study extends the DPSIRM framework in two aspects. First, the green characteristics and positive environmental spillover effects of digital finance position it as a key driver in promoting green and low-carbon development. Therefore, we introduce the Digital Inclusive Finance Index into the DPSIRM model to quantify the driving effect of digital finance on green and low-carbon development. Second, the Baidu Index for Environmental Protection and Baidu Index for Carbon Dioxide are introduced as impact indicators within the evaluation system, reflecting shifts in public cognition, attitudes, and the prioritization of green development triggered by systemic state changes.

3.2. Evaluation Indicator System

Spanning 2011–2021, our assessment of provincial-level green and low-carbon performance is anchored in established theoretical foundations, while also accounting for policy priorities and data constraints. The selection of indicators followed three principles. These were (1) theoretical relevance, referencing the similar literature to ensure variables directly match the core DPSIRM framework; (2) data availability, prioritizing officially published statistics with continuous annual coverage from 2011 to 2021; and (3) policy alignment, conforming to China’s national strategic priorities. Based on these considerations, we construct an evaluation indicator system to quantify green and low-carbon development using 39 indicators across the six dimensions of the DPSIRM theoretical model, as detailed in

Table 1. Comprehensive evaluation index system for green and low-carbon development level.

Target Layer	Factor Layer	Index Layer	Unit of Measurement	Attribute
Green and low-carbon development	Driving force ($D$)	Gross regional product per capita $D_1$	yuan (RMB)	Positive
		Per capita disposable income $D_2$	yuan (RMB)	Positive
		Social total retail sales $D_3$	hundred million yuan	Positive
		Urbanization rate $D_4$	%	Positive
		Digital financial inclusion index $D_5$	/	Positive
		Natural population growth rate $D_6$	‰	Negative
	Pressure ($P$)	Sulfur dioxide emissions $P_1$	10 kt	Negative
		Carbon dioxide emissions $P_2$	Megaton	Negative
		Ammonia nitrogen emissions $P_3$	10 kt	Negative
		Chemical oxygen demand emissions $P_4$	10 kt	Negative
		Municipal sewage discharge $P_5$	10 km3	Negative
		Per capita water consumption $P_6$	m3/person	Negative
		Consumption of electric power $P_7$	100 million kWh	Negative
		Number of private cars $P_8$	10,000 vehicles	Negative
	State ($S$)	Average daily precipitation $S_1$	m	Positive
		Forest coverage rate $S_2$	%	Positive
		Growing stock $S_3$	10 km3	Positive
		Urban green space $S_4$	10,000 hectares	Positive
		Per capita water resources $S_5$	m3/person	Positive
		Average PM2.5 concentration $S_6$	μg/m3	Negative
	Impact ($I$)	Grain output $I_1$	10 kt	Positive
		Added value of agriculture, forestry, animal industry, and fishery $I_2$	hundred million yuan	Positive
		Environmental protection Baidu Index $I_3$	/	Positive
		Carbon dioxide Baidu Index $I_4$	/	Positive
		Forest disease occurrence area $I_5$	10,000 hectares	Negative
		human mortality $I_6$	‰	Negative
	Response ($R$)	Forest pest control area $R_1$	10,000 hectares	Positive
		Comprehensive use of general industrial solid waste $R_2$	10 kt	Positive
		Number of industrial waste gas treatment facilities $R_3$	Set	Positive
		Number of industrial wastewater treatment facilities $R_4$	Set	Positive
		Total amount of urban sewage treatment $R_5$	10 km3	Positive
		Drainage pipe length $R_6$	km	Positive
		Harmless treatment rate of household garbage $R_7$	%	Positive
	Management ($M$)	Investment in industrial pollution control $M_1$	hundred million yuan	Positive
		Proportion of education expenditure in general budget expenditure $M_2$	%	Positive
		Proportion of expenditure on science and technology in the general budget $M_3$	%	Positive
		Number of domestic patent applications authorized $M_4$	items	Positive
		Full-time equivalent R&D personnel in industrial enterprises above designated size $M_5$	person	Positive
		Number of R&D projects of industrial enterprises above designated size $M_6$	items	Positive

.

3.3. Data Sources

Considering data availability, this study takes 30 provinces, autonomous regions, and municipalities (excluding Hong Kong, Macao, Taiwan, and Tibet) in China from 2011 to 2021 as the research objects. This study also completes the missing data points in specific samples through linear interpolation. The data sources for specific indicators are detailed below:

(1) Digital financial inclusion index $D_5$: Digital Finance Research Center of Peking University [38] (https://idf.pku.edu.cn/).

(2) Carbon dioxide emissions $P_2$: China Carbon Accounting Databases (CEADs) (https://www.ceads.net.cn/).

(3) Average daily precipitation $S_1$: European Center for Medium-Range Weather Forecasts (ECMWF) (https://www.ecmwf.int/).

(4) Average PM2.5 concentration $S_6$: Socioeconomic Data and Applications Center (SEDAC) at Columbia University (https://sedac.ciesin.columbia.edu/).

(5) Environmental protection Baidu Index $I_3$ and carbon dioxide Baidu Index $I_4$: Baidu Index platform (https://index.baidu.com).

(6) All other indicators: The following official yearbooks, accessible via the National Bureau of Statistics (http://www.stats.gov.cn): the China Statistical Yearbook, the China Environmental Statistical Yearbook, the China Urban and Rural Construction Statistical Yearbook, and the China Science and Technology Statistical Yearbook.

* Baidu Index raw data subject to platform terms; standardized results available upon request from correspondence author.

3.4. Research Methods

As described above, we construct a green and low-carbon development index system based on the six DPSIRM dimensions, calculating the green and low-carbon development of 30 provinces, autonomous regions, and municipalities across China by applying a projection pursuit model that is optimized by a simulated annealing algorithm. We determine the evolutionary features of green and low-carbon development via approaches like kernel density estimation and spatial autocorrelation analysis, and also use the Theil Index to probe into regional differences in green and low-carbon development across the nation. Figure 1 outlines our general study workflow. The analysis was conducted by using MATLAB R2021a for the SA-PP model and kernel density estimation, utilizing ArcGIS 10.8 for geographic mapping, and employing Stata 17 to perform spatial autocorrelation analysis and Theil Index decomposition. The core mathematical formulas and computational procedures for each method are detailed below.

3.4.1. SA-PP Model

To address potential scale effects and ensure comparability across provinces of varying sizes, all raw indicators are first standardized to eliminate the influence of measurement units and provincial economic scales, transforming them into dimensionless values. This study adopts the modified min–max normalization method, with the following specific formulas:

For positive indicators:

$$z_{ij}={\displaystyle \frac{0.98\times max(x_j)-x_{ij}}{0.98\times max(x_j)-{0.02\times min(x}_j)}}$$

For negative indicators:

$$z_{ij}={\displaystyle \frac{x_{ij}-{0.02\times min(x}_j)}{0.98\times max(x_j)-{0.02\times min(x}_j)}}$$

where $max(x_j)$ and $min(x_j)$ are the maximum and minimum values of the $j$-th indicator across all provinces, respectively, and $z_{ij}$ is the normalized value. To mitigate the influence of extreme values and ensure computational stability, we employed a modified min–max scaling technique, adjusting the range bounds to [0.02, 0.98] to prevent zero-valued outcomes during projection pursuit modeling.

Standardized data (i.e., $z_{ij}$) is then processed using the SA-PP model, combining the simulated annealing algorithm (SA) [39] and projection pursuit (PP) [40] to construct the provincial green low-carbon development index.

This model projects high-dimensional data onto a low-dimensional subspace. Let the normalized evaluation matrix for $p$ provinces and $m$ indicators be $Z=\{z_{ij}{\}}_{p\times m}$. The projection direction vector is $w=(w_1,w_2,\dots ,w_m)$, where $\Vert w\Vert =1$. The projection value $z_i$ for province $i$ is

$$z_i={\sum }_{j=1}^m w_j·z_{ij}$$

The optimization objective $Q(w)$ simultaneously enhances between-province separation (${\sigma }_z$) and within-cluster compactness (${\rho }_z$):

$${\sigma }_z=\sqrt{{\displaystyle \frac{{\sum }_{i=1}^p (z_i-\bar{z}{)}^2}{p-1}}}, {\rho }_z={\sum }_{i=1}^p {\sum }_{k=1}^p (R-r_{ik})·f(R-r_{ik})$$

where $\bar{z}={\displaystyle \frac{1}{p}}{\sum }_{i=1}^p z_i$ is the mean of all provincial projection values, and $d_{ij}=|z_i-z_j|$ denotes the absolute distance between projection values of provinces i and j.

$$Q(w)={\sigma }_z·{\rho }_z$$

where $R=0.1{\sigma }_z$ (window radius) and $f(u)$ is the unit step function ($f(u)=1$ if $u\ge 0$, else $0$).

The SA optimizes $\mathrm{w}$ to maximize $Q(w)$: initialize $w^{(0)}$, temperature $T_0=100$, and cooling rate $\alpha =0.95$. Perturb $\mathrm{w}$ to generate $w^{\prime }=w+\eta ·\delta$ (random vector $\delta$, step size $\eta$). Accept $w^{\prime }$ if $Q(w^{\prime })>Q(w)$ or with probability $exp\left({\displaystyle \frac{Q(w^{\prime })-Q(w)}{T}}\right)$. Cool $T=\alpha ·T$ until $T<{10}^{-5}$.

The optimal $w^{*}$ yields the comprehensive provincial score $z_i^{*}$. This objective, rational methodology avoids subjective weighting interference when measuring green low-carbon development.

3.4.2. Kernel Density Estimation

For analyzing the dynamic distribution regarding the provincial green low-carbon development index, a nonparametric method, namely kernel density estimation, is employed [41,42]. Provincial score distributions are modeled nonparametrically via kernel density estimation to capture temporal dynamics. Its core process involves constructing a continuous probability density function and using an appropriate bandwidth [43] to perform nonparametric estimation of the data distribution. For scores $z_1,z_2,\dots ,z_p$ in year $t$, the density $\hat{f}(z)$ at point $z$ is

$$\hat{f}(z)={\displaystyle \frac{1}{p·h}}{\sum }_{i=1}^p K\left({\displaystyle \frac{z-z_i}{h}}\right)$$

where $K(u)={\displaystyle \frac{1}{\sqrt{2\pi }}}exp\left(-{\displaystyle \frac{u^2}{2}}\right)$ is the Gaussian kernel, and bandwidth $h$ is selected via Silverman’s rule: $h=0.9·min\left(\sigma ,{\displaystyle \frac{IQR}{1.34}}\right)·p^{-1/5}$. $IQR$ denotes the interquartile range (i.e., the difference between the 75th and 25th percentiles).

This study generates three-dimensional plots of $\hat{f}(z,t)$ based on two-dimensional joint kernel density estimation, illustrating the temporal evolution characteristics of green and low-carbon development.

3.4.3. Spatial Autocorrelation

To examine spatial clustering patterns of the provincial green low-carbon development index, spatial autocorrelation analysils is employed. This statistical method measures spatial connections of spatial data, which can be divided into global and local spatial autocorrelation. As Moran’s Index (Moran’s I) [44] is a commonly used indicator for measuring the spatial autocorrelation of geospatial data, this study analyzes the global and local Moran’s I for the provinces in each region separately.

Global Moran’s I [45,46]:

$$I={\displaystyle \frac{p{\sum }_{i=1}^p {\sum }_{j=1}^p w_{ij}(z_i-\bar{z})(z_j-\bar{z})}{\left({\sum }_{i=1}^p {\sum }_{j=1}^p w_{ij}\right){\sum }_{i=1}^p (z_i-\bar{z}{)}^2}}$$

where $w_{ij}$ is a binary spatial weight matrix ($w_{ij}=1$ if provinces $i$ and $j$ share a border, else $0$). Significance is tested via permutation.

Local Moran’s I [47]:

$$I_i={\displaystyle \frac{(z_i-\bar{z})}{{\sigma }^2}}\sum _{j=1}^p w_{ij}(z_j-\bar{z}), {\sigma }^2={\displaystyle \frac{{\sum }_{k=1}^p (z_k-\bar{z}{)}^2}{p}}$$

Scatter plots classify provinces into HH (high–high), LH (low–high), LL (low–low), and HL (high–low) clusters.

3.4.4. Theil Index and Decomposition

To quantify regional disparities in the provincial green low-carbon development index, the Theil Index [48] is employed. The Theil Index measures overall disparity and decomposes it into within-group ($T_{\mathrm{WG}}$) and between-group ($T_{\mathrm{BG}}$) components [49]. For $k$ regions, the following applies:

$$T={\sum }_{i=1}^p s_{i}ln\left({\displaystyle \frac{s_i}{1/p}}\right), s_i={\displaystyle \frac{z_i}{{\sum }_{i=1}^p z_i}}$$

where a higher value of T indicates greater disparities in green low-carbon development between regions and a lower value signifies smaller disparities. Decomposition:

$$T=\underset{Within-group}{\underbrace{{\sum }_{g=1}^k s_g{T}_g}}+\underset{Between-group}{\underbrace{{\sum }_{g=1}^k s_{g}ln\left({\displaystyle \frac{s_g}{n_g/p}}\right)}}$$

where $s_g$ is region $g$’s score share, $n_g$ is its province count, and $T_g$ is its internal Theil Index. The contribution rates of within-group and between-group disparities to the overall disparity are computed as $T_{\mathrm{WG}}/T$ and $T_{\mathrm{BG}}/T$, respectively. This study uses the Theil Index to analyze changes and sources of disparity in this index across 30 provinces, autonomous regions, and municipalities [50].

4. Results and Analysis of Green and Low-Carbon Development Across Chinese Provinces

4.1. Optimal Projection Direction Vector

To eliminate dimensional differences in the data, we conducted separate normalization processing on forward and reverse indicators. We optimized the SA-PP model to measure green and low-carbon development across Chinese provinces [51,52,53,54]. After optimization calculations, the optimal projection direction vector obtained was determined to be $w$ = (0.035, 0.183, 0.142, 0.253, 0.138, 0.172, 0.063, 0.044, 0.131, 0.072, 0.014, 0.144, 0.229, 0.251, 0.080, 0.092, 0.222, 0.018, 0.195, 0.179, 0.022, 0.208, 0.048, 0.247, 0.209, 0.008, 0.165, 0.236, 0.180, 0.089, 0.208, 0.212, 0.171, 0.048, 0.099, 0.205, 0.034, 0.235, 0.174). This vector$, \mathrm{w},$ represents the optimal weights for combining the 39 indicators into a comprehensive score, where each component $w_k$ quantifies the relative contribution of the corresponding indicator (a higher $w_k$ indicates stronger influence, e.g., $D_4$ has the maximum weight of 0.253). The optimization criterion is to maximize the projection index function $Q\left(w\right)$=$\sigma ·\rho$, which ensures both the dispersion ($\sigma$) and cluster separation ($\rho$) of provincial scores. Figure 2 illustrates the proportions of various indicators’ and subsystems’ influence on overall green and low-carbon development. The magnitude of each component in the optimal projection direction reflects the decision-making information of each evaluation indicator, with a higher numerical value indicating the corresponding evaluation indicator’s greater influence on overall green and low-carbon development. Based on the magnitudes of the components in the projection vector, we identify the top 10 indicators that most significantly impact the evaluation of green and low-carbon development, which are $D_4,$ urbanization rate; $P_8,$ the number of private cars; $I_4,$ CO₂ Baidu Index; $R_2,$ the comprehensive use of general industrial solid waste; $M_5,$ industrial enterprises above designated full-time equivalent research and development (R&D) personnel; $P_7,$ electricity consumption; $S_3,$ growing stock; $R_6,$ drainage pipe length; $I_5,$ forest disease occurrence area; and $I_2,$ the added value of agriculture, forestry, animal husbandry, and fishery. Response mechanisms constitute the most influential subsystem (23.12% aggregate weight). This is obtained by summing up the weights $w_k$ of all indicators in this subsystem, and the result indicates that it exerts the highest influence on green and low-carbon development. In addition, the impact subsystem has the least substantial influence on green and low-carbon development, with an impact proportion of 13.60%.

4.2. Comprehensive Green and Low-Carbon Development Score

Provincial green low-carbon scores for 2011–2021 are derived via the SA-PP model. As the core output of this evaluation framework,

Table 2. Partial values, full-period annual averages, and growth rates (%) of green and low-carbon comprehensive development.

Province	2011	2013	2015	2017	2019	2021	Annual Averages	Growth Rates
Beijing	2.269	2.536	2.942	2.673	2.769	2.626	2.641	1.472
Guangdong	2.246	2.426	2.696	2.752	2.736	2.560	2.575	1.317
Shanghai	2.387	2.430	2.750	2.501	2.970	2.279	2.569	−0.462
Jiangsu	2.333	2.306	2.762	2.503	3.045	2.316	2.559	−0.073
Zhejiang	2.374	2.374	2.781	2.485	3.024	2.304	2.552	−0.299
Shandong	2.093	2.420	2.594	2.596	2.432	2.317	2.495	1.022
Tianjin	2.188	2.452	2.352	2.617	2.623	2.365	2.484	0.781
Fujian	2.078	2.292	2.395	2.435	3.018	2.192	2.436	0.536
Liaoning	2.137	2.542	2.485	2.665	2.643	2.464	2.430	1.434
Hebei	2.089	2.362	2.288	2.562	2.529	2.394	2.416	1.372
Shaanxi	2.045	2.485	2.271	2.955	2.560	2.163	2.416	0.563
Anhui	1.969	2.246	2.681	2.384	2.893	2.131	2.385	0.794
Inner Mongolia	2.074	2.407	2.320	2.554	2.544	2.397	2.385	1.458
Sichuan	2.014	2.395	2.653	2.750	2.558	2.101	2.382	0.424
Shanxi	2.079	2.355	2.291	2.493	2.562	2.378	2.371	1.353
Henan	2.060	2.368	2.430	2.544	2.358	2.226	2.357	0.778
Hubei	2.063	2.428	2.514	2.614	2.423	2.230	2.353	0.781
Jilin	2.116	2.115	2.494	2.647	2.610	2.382	2.341	1.191
Heilongjiang	2.173	2.168	2.482	2.194	2.639	2.422	2.337	1.091
Chongqing	2.165	2.393	2.539	2.555	2.488	2.060	2.332	−0.496
Ningxia	2.139	2.129	2.302	3.008	2.512	2.157	2.330	0.084
Qinghai	2.094	2.460	2.273	2.908	2.417	2.138	2.319	0.208
Hunan	2.025	2.180	2.503	2.614	2.419	2.207	2.315	0.864
Gansu	2.094	2.425	2.183	2.980	2.449	2.111	2.311	0.081
Yunnan	1.994	2.379	2.213	2.744	2.449	2.076	2.308	0.404
Jiangxi	1.978	2.252	2.387	2.360	2.205	2.094	2.304	0.572
Guangxi	2.046	2.263	2.468	2.499	2.352	2.005	2.290	−0.202
Xinjiang	2.192	2.146	2.324	2.184	2.546	2.226	2.279	0.154
Guizhou	1.976	2.300	2.107	2.708	2.427	1.999	2.247	0.116
Hainan	2.088	2.207	2.485	2.501	2.364	1.943	2.240	−0.717

To ensure clarity in regional analysis, China’s 30 provinces are categorized into three major regions. Eastern region (11 provinces): Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, and Hainan. Central region (8 provinces): Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, and Hunan. Western region (11 provinces): Guangxi, Inner Mongolia, Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang.

presents the green and low-carbon development levels of various regions for selected years, full-period annual averages (for cross-sectional comparison of overall levels across regions), and annual geometric growth rates (%, to reflect temporal trends of dynamic changes). It should be noted that the arithmetic mean mainly reflects the overall average state during the observation period, which is suitable for data with small fluctuations or no significant trends. However, the green and low-carbon development level in this study showed a fluctuating upward trend from 2011 to 2021. Between 2011 and 2021, the average comprehensive green and low-carbon development across China’s provinces increased from 2.119 to 2.242, exhibiting a fluctuating upward trend that reflects China’s progress on the path to green and low-carbon development. Over more than a decade of development, particularly since the 18th National Congress of the CPC in 2012, China has adhered to a policy system with energy conservation and emissions reduction at its core, continuously supporting the development of an ecological civilization and protecting the ecological environment, with certain results achieved in enhancing green and low-carbon development.

Despite overall improvement in green and low-carbon development, certain disparities in comprehensive development levels remain among China’s provinces. Coastal provinces (Beijing, Shanghai, Jiangsu, Zhejiang, Guangdong) lead in composite scores, all of which have annual averages exceeding 2.500, demonstrating leading positions. The bottom five provinces are Hainan, Guizhou, Xinjiang, Guangxi, and Jiangxi, which lag behind in green and low-carbon development, with annual averages below 2.305. These regions generally face challenges such as resource and environmental constraints, irrational industrial structures, and inadequate technological innovation capabilities, leading to relatively low green and low-carbon development. Additionally, these regions vary in geographical location and economic development, posing more challenges for green and low-carbon development.

From a regional perspective, the top five provinces are all located in the eastern region, reflecting the region’s overall advantages in green and low-carbon development. As the pioneer of China’s economic development, the eastern region boasts considerable economic strength, technological innovation capabilities, and openness, which provides strong support for green and low-carbon development. Among the bottom five provinces, three are in the western region, indicating certain shortcomings in the region’s green and low-carbon development. Furthermore, the average comprehensive green and low-carbon development in the eastern, central, and western regions are 2.490, 2.354, and 2.327, respectively, exhibiting an overall imbalance. The eastern region has achieved higher green and low-carbon development due to its superior economic conditions and innovation advantages. Although the central region lags behind the eastern region, regional governments are actively promoting green transformation and low-carbon development. The western region faces more challenges and difficulties and must increase investment and efforts to drive green and low-carbon development.

4.3. Subsystem Scores for Green and Low-Carbon Development

Subsystem trajectories (Figure 3) reveal divergent trends among DPSIRM dimensions over 2011–2021. The scores of the driving force, state, impact, and response subsystems increased, while those of the pressure and management subsystems declined. The driving force subsystem exhibited a trend of steady development and was the subsystem with the second-highest average score by 2021. The pressure subsystem score exhibited a fluctuating downward trend, decreasing from 0.707 in 2011 to 0.596 in 2021, but remained the subsystem with the highest average score in China. The state subsystem score exhibited complex fluctuations, with a significant decrease in the later period. The impact subsystem showed a favorable development trend, with slight decreases in individual years but an overall upward trend. The response subsystem score increased overall in the early period but experienced a significant decline from 2016 to 2018, followed by another increase in the later period. The management subsystem exhibited a continuous decrease in scores from 2017 to 2021, dropping from 0.276 in 2011 to 0.192 in 2021, making it the subsystem with the lowest average score.

Table 3. Annual average values and annual growth rates (%) of development levels of various green and low-carbon subsystems (2011–2021).

Province	Driving Force (D)		Pressure (P)		State (S)		Impact (I)		Response (R)		Management (M)
	μ	r (%)	μ	r (%)	μ	r (%)	μ	r (%)	μ	r (%)	μ	r (%)
Beijing	0.620	3.414	0.663	−0.985	0.502	0.062	0.237	16.241	0.458	4.057	0.230	−3.343
Tianjin	0.535	2.618	0.668	−0.768	0.518	0.066	0.224	14.334	0.418	3.218	0.278	−3.570
Hebei	0.414	3.971	0.656	−0.748	0.523	0.039	0.241	14.593	0.465	5.114	0.285	−2.748
Shanxi	0.400	4.069	0.657	−0.587	0.537	0.115	0.231	18.716	0.425	4.233	0.278	−3.361
Inner Mongolia	0.437	2.712	0.640	−1.052	0.511	0.009	0.230	17.311	0.434	3.831	0.289	−1.768
Liaoning	0.477	2.861	0.643	−0.703	0.496	−0.024	0.231	15.105	0.450	4.494	0.293	−3.130
Jilin	0.411	3.523	0.650	−0.409	0.505	0.100	0.230	14.135	0.414	3.349	0.282	−3.231
Heilongjiang	0.413	3.351	0.639	−0.275	0.508	−0.157	0.223	12.151	0.432	4.160	0.278	−3.246
Shanghai	0.630	2.896	0.641	−2.261	0.502	0.057	0.216	0.043	0.455	2.852	0.281	−6.683
Jiangsu	0.555	3.813	0.615	−2.363	0.482	0.034	0.231	−0.409	0.512	4.344	0.302	−5.970
Zhejiang	0.538	3.267	0.628	−2.433	0.483	−0.063	0.232	−0.663	0.498	3.280	0.310	−3.752
Anhui	0.420	4.314	0.644	−2.122	0.472	0.134	0.225	14.603	0.452	3.158	0.314	−2.363
Fujian	0.490	3.857	0.641	−1.841	0.482	−0.068	0.226	10.308	0.441	2.537	0.306	−2.957
Jiangxi	0.401	3.498	0.645	−1.911	0.473	0.106	0.209	17.312	0.422	2.523	0.297	−2.077
Shandong	0.501	3.350	0.643	−1.896	0.484	0.132	0.230	16.344	0.491	3.918	0.295	−2.401
Henan	0.423	3.189	0.648	−2.176	0.511	0.033	0.224	10.577	0.442	3.368	0.276	−2.230
Hubei	0.454	3.436	0.648	−2.121	0.508	−0.073	0.223	9.146	0.416	3.048	0.266	−1.728
Hunan	0.417	4.360	0.633	−1.710	0.510	−0.027	0.229	7.516	0.411	3.031	0.261	−2.034
Guangdong	0.571	3.733	0.617	−1.509	0.497	−0.247	0.237	5.007	0.514	4.861	0.275	−2.130
Guangxi	0.371	1.592	0.607	−5.161	0.495	0.693	0.231	7.460	0.407	3.174	0.286	−4.047
Hainan	0.379	0.645	0.612	−5.955	0.518	0.519	0.215	6.818	0.359	2.843	0.273	−3.989
Chongqing	0.438	1.433	0.619	−6.435	0.507	0.148	0.222	5.566	0.395	3.248	0.287	−3.591
Sichuan	0.406	5.276	0.599	−1.945	0.493	1.368	0.242	1.769	0.473	−0.917	0.302	−3.359
Guizhou	0.338	5.704	0.596	−1.154	0.489	1.406	0.232	1.421	0.422	−1.600	0.298	−4.152
Yunnan	0.352	4.903	0.599	−0.793	0.502	1.616	0.243	2.072	0.440	−0.685	0.305	−4.593
Shaanxi	0.404	4.796	0.602	−0.595	0.489	1.837	0.246	2.583	0.479	−0.482	0.331	−4.275
Gansu	0.345	6.110	0.614	−0.746	0.504	2.087	0.232	0.868	0.427	−1.380	0.322	−4.735
Qinghai	0.364	6.654	0.618	−0.438	0.498	2.346	0.244	0.873	0.383	−1.529	0.323	−3.709
Ningxia	0.383	7.260	0.616	0.177	0.469	2.806	0.230	0.174	0.372	−2.874	0.340	−5.010
Xinjiang	0.364	7.058	0.639	0.155	0.500	2.908	0.228	0.468	0.398	−2.319	0.293	−5.733

presents the annual average values (μ) and annual geometric growth rates (r) of the various green and low-carbon subsystems’ development across regions from 2011 to 2021. Table 3 reveals that the annual national average value for the driving force subsystem’s development is 0.442, and Shanghai ranks at the top with an annual average of 0.630 and Guizhou ranks at the bottom with an annual average of 0.338. The national annual average value of the pressure subsystem’s development is 0.631, with Tianjin ranking the highest with an annual average of 0.668 and Guizhou ranking the lowest at 0.596. The state subsystem’s national annual average is 0.499, with Shanxi ranking the highest with an annual average of 0.537 and Ningxia the lowest at 0.469. The impact subsystem’s national annual development average is 0.230, with Shaanxi ranking the highest with an annual average of 0.246 and Jiangxi the lowest at only 0.209. The response subsystem’s national annual development average is 0.437, with Guangdong ranking the highest with an annual average of 0.514 and Hainan the lowest at 0.359. The management subsystem’s national annual development average is 0.294, with Ningxia exhibiting the highest annual average among the provinces listed (despite being lower than the overall national average at 0.261) and Hunan the lowest at 0.340. Significant differences are evident among provinces in terms of scores in the various subsystems supporting green and low-carbon development.

5. Temporal and Spatial Characteristics of Green and Low-Carbon Development Across China’s Provinces

5.1. Temporal Changes in Green and Low-Carbon Development

Referencing the comprehensive scores for green and low-carbon development across various provinces, Figure 4 presents the kernel density estimation plots of the overall distribution of green and low-carbon development nationwide and in the eastern, central, and western regions from 2011 to 2021, respectively, to illustrate the temporal characteristics of green and low-carbon development in different regions. (a) is the kernel density estimation plot for national green and low-carbon development. The significant left-side tailing of the kernel density curve indicates a notable gap in green and low-carbon development among provinces nationwide, with some provinces displaying a long-term lag in green and low-carbon development. The polarization trend of the curve reveals a main peak + small peak distribution in 2011–2012 and 2017–2019, indicating increased polarization in national green and low-carbon development during these two periods. By 2021, this phenomenon had significantly improved, with the kernel density curve transitioning to a single peak, indicating the notable weakening of polarization. However, the peak value of the curve decreases and the width expands, suggesting that the absolute gap between provinces is still widening, and the overall development balance needs to be improved. (b) presents the kernel density estimation plot for green and low-carbon development in the eastern region. The curve initially exhibits a single-peak shape. The most obvious multi-peak characteristics of the kernel density curve in the eastern region reflect the spatial imbalance of its internal development. In 2016, a main peak and multiple side peaks emerged, showing that there were multiple development centers in the eastern region. With the number of side peaks later decreasing, the width of the curve continued to expand after 2018, indicating that the internal gap in the eastern region was still widening. (c) is the kernel density estimation plot for green and low-carbon development in the central region. The kernel density curve in the central region evolved from “multi-peak” to “single-peak” and then to “wide-peak”, reflecting the phased changes in its development gap. The multi-peak shape in 2015 clearly indicates multipolarization in green and low-carbon development in the central region. From 2015 to 2018, the kernel density curve gradually shifts to a single peak, with the peak value rising. This reflects an overall elevation in the development level and a narrowing of internal gaps. After 2018, the single peak of the kernel density curve gradually vanishes; the width expands, peaks decline, and the distribution shape changes from sharp–narrow to flat–wide. This indicates a gradual growth in the disparity of green and low-carbon development in the central region and low-carbon development in the central region. In addition, compared with other regions, the multi-peak stage in the central region is shorter and the number of peaks is smaller, which is closely related to its higher industrial structure homogeneity, stronger regional policy coordination, and smaller differences in resource endowments. These characteristics result in relatively smaller internal differences in green and low-carbon development in this region. (d) presents the kernel density estimation plot for green and low-carbon development in the western region. The kernel density curve shifts to the right amid fluctuations, indicating an upward trend in green and low-carbon development in the western region with notable fluctuations. Later in the observation period, the curve mainly exhibits high and low peaks, with a polarized shape. The low peak on the left has always existed and accounts for a significant proportion, indicating that a large number of provinces in the western region are still lag relatively in green and low-carbon development. To promote overall nationwide green and low-carbon development, the western region requires more attention and support.

5.2. Spatial Characteristics of Green and Low-Carbon Development

5.2.1. Spatial Distribution Characteristics

To visually illustrate the dynamic spatial evolution of green and low-carbon development, we employ ArcGIS software and adopt the Natural Breaks classification method, classifying the 30 provinces into five tiers for 2012, 2015, 2018, and 2021: low-value, medium-low-value, medium-value, medium-high-value, and high-value regions.

As shown in Figure 5, by 2021, the number of provinces in high- and medium-high-value regions for green and low-carbon development increased. This development was primarily concentrated in the eastern regions of Beijing, Guangdong, Liaoning, Hebei, Tianjin, Shandong, and Jiangsu. Meanwhile, Heilongjiang, Jilin, and Shanxi in the central region and Inner Mongolia in the western region entered medium–high-value status. Zhejiang, Shanghai, and Fujian in the eastern region; Hubei, Henan, and Hunan in the central region; and Xinjiang in the western region display medium-value status. Most provinces in the western region are of low- and medium–low-value status, encompassing Shaanxi, Ningxia, Qinghai, Gansu, Sichuan, Yunnan, Chongqing, Guangxi, and Guizhou. Additionally, Hainan in the eastern region and Anhui and Jiangxi in the central region also fall into medium–low-value status. Green and low-carbon development exhibits a decreasing spatial pattern from east to west. The eastern coastal region, leveraging unique geographical advantages and pioneering of reform and opening-up, has made significant progress in green and low-carbon technologies. Conversely, central and western provinces exhibit lower levels, potentially constrained by factors such as less advanced technology, talent shortages, and slower industrial restructuring. Overall, certain disparities in green and low-carbon development are evident among different regions, and China must prioritize efforts to address the issue of imbalanced and uncoordinated regional development.

5.2.2. Spatial Agglomeration Characteristics

To comprehensively investigate the spatial correlation of green and low-carbon development across China’s regions from 2011 to 2021,

Table 4. Global Moran’s I of green and low-carbon development nationwide (2011–2021).

	$\mathit{G}\mathit{l}\mathit{o}\mathit{b}\mathit{a}\mathit{l} \mathit{M}\mathit{o}\mathit{r}\mathit{a}\mathit{n}\mathit{\prime }\mathit{s} \mathit{I}$	$\mathit{Z}$	$\mathit{P}$
2011	0.3213	2.9267	0.0034
2012	0.4211	3.7985	0.0001
2013	−0.0871	−0.4241	0.6715
2014	0.2522	2.3536	0.0186
2015	0.2175	2.0466	0.0407
2016	0.4355	3.7945	0.0001
2017	0.0845	0.9758	0.3292
2018	0.3812	3.4266	0.0006
2019	0.3175	2.8729	0.0041
2020	0.6193	5.2360	0.0000
2021	0.2549	2.3479	0.0189

employs the global Moran’s I index based on a first-order contiguity spatial matrix for sequential testing. The results reveal that from 2011 to 2021, the overall level of spatial correlation in green and low-carbon development across China’s regions exhibited a fluctuating downward trend, indicating that some provinces achieved rapid green and low-carbon development, characterized by diversified development, balanced distribution, and other benefits. With the exception of insignificant global Moran’s I values in 2013 and 2017, the Moran’s I values for other years passed the significance test at the 5% level, with all values greater than 0, indicating a significant positive spatial correlation in green and low-carbon development across regions nationwide.

The global Moran’s I values indicate that while spatial correlation in the green and low-carbon development exists between provinces nationwide, it does not reflect the characteristics of local spatial agglomeration. This study further calculates the local Moran’s I for each year based on the first-order contiguity spatial matrix and generates scatter plots to reveal the types of local spatial correlation in green and low-carbon development in Figure 6 (due to space limitations, the figure only shows the results for 2012, 2015, 2018, and 2021) [55,56].

The first quadrant represents areas with high green and low-carbon development that are surrounded by other high-value regions (high–high agglomeration areas). Taking the Yangtze River Delta region as an example, Shanghai, Jiangsu, and Zhejiang have long been located in the first quadrant. These areas are not only at the forefront in green and low-carbon technology R&D, policy formulation, and implementation, but they also have positive spillover effects on surrounding regions through regional cooperation, industrial transfer, and other means. The second quadrant represents spatial relationships where low-value regions are surrounded by high-value areas (low–high agglomeration areas). These regions often face considerable development pressure but also have enormous potential for improvement and can learn from neighboring high-value regions’ development experiences and strengthen regional cooperation to enhance development. Most provinces in China are in the third quadrant, and the majority of provinces in the western and central regions have long been in low–low agglomeration areas, with low green and low-carbon development levels and surrounded by neighboring provinces with lagging green and low-carbon development. These areas lack sufficient awareness of green and low-carbon development or clear green and low-carbon development goals, policy support, and regulatory mechanisms. It is difficult for such regions to achieve green and low-carbon development solely through internal efforts, and the diffusion effects from adjacent areas have not had an influence. The fourth quadrant represents spatial relationships where high-value regions are surrounded by low-value areas (high–low agglomeration areas). Although these areas have good internal development, the lagging development of surrounding areas can make it difficult to establish effective regional linkage effects.

Most provinces in China fall into the first and third quadrants, indicating the pronounced positive aggregation characteristics of green and low-carbon development across various regions. Meanwhile, a small number of provinces have long been in the second or fourth quadrant, exhibiting high–low or low–high aggregation characteristics. Overall, significant local spatial agglomeration characteristics in the green and low-carbon development are evident among various regions in China, accompanied by severe polarization. In addition, although some provinces have undergone notable transitions in certain years, the local spatial agglomeration pattern of green and low-carbon development among various regions in China remains relatively stable overall.

6. Regional Disparities and Decomposition in China’s Green and Low-Carbon Development

6.1. Theil Index Trends

This study employs an integrated analytical framework combining kernel density estimation, Moran’s Index, and the Theil Index to systematically investigate the spatiotemporal distribution patterns and regional disparities in China’s provincial green low-carbon development. Kernel density estimation captures the temporal distribution dynamics of green low-carbon development levels through probability density curves, visually presenting evolutionary trends. Moran’s Index diagnoses spatial autocorrelation characteristics using global and local statistics to identify significant geographical agglomeration effects. The Theil Index quantifies structural sources by decomposing regional inequality into within-group and between-group components based on the entropy decomposition methodology. Together, these three methods form a complete logical system that extends from temporal distribution characteristics to spatial agglomeration features, and ultimately to the decomposition of disparity structures.

The Theil Index, which is characterized by additivity and decomposability, has become an important method for measuring inequality and disparity. To explore the rationale behind the differentiation among the three major regions in China, this study employs the Theil Index to measure the relative disparities in green and low-carbon development across regions [58]. The results are presented in Figure 7, revealing that the overall Theil Index and the Theil Indices of the three regions exhibit a cross-evolutionary trend. The overall Theil Index roughly follows an M-shaped trend, with disparities widening before 2016, converging between 2016 and 2018, gradually widening from 2018 to 2020, and then narrowing again after 2020. Ultimately, the index increased from 0.0013 in 2011 to 0.0027 in 2021, representing a 0.0014 increase, demonstrating that the overall disparities in China’s green and low-carbon development are on an upward trend, and showing that the issue of addressing imbalanced green and low-carbon development in China remains a long and arduous task.

The trends in the Theil Indices of the three major regions exhibited a certain degree of divergence from 2011 to 2018 but remained relatively synchronized from 2019 to 2021. The Theil Index was the highest for the eastern region in 2021, indicating that compared with other regions, the distribution of green and low-carbon development within the eastern region exhibited the most significant disparities. Provinces with high and medium–high values coexisted with those in other ranges. Beijing and Guangdong, which fell into the high-value range, along with five provinces including Jiangsu and Shandong in the medium–high-value range, elevated the green and low-carbon development in the eastern region but also widened the gap with Hainan, Fujian, Shanghai, and Zhejiang. The Theil Index for the central region showed a fluctuating upward trend, suggesting an increase in disparities in green and low-carbon development among its provinces [59]. By 2021, the Theil Index for the central region ranked second among the three major regions. Heilongjiang, Jilin, and Shanxi were relatively close to meeting the requirements for green and low-carbon development, but most provinces still scored low in comprehensive green and low-carbon development, revealing significant internal disparities within the central region. Although the western region exhibited an upward trend from 2012 to 2017, it experienced a sharp decline in the later period. By 2021, seven provinces in the western region fell into the medium–low-value range of green and low-carbon development, resulting in insignificant disparities between the provinces in the region. Consequently, the western region’s Theil Index was the lowest among the three major regions.

6.2. Theil Index Decomposition Results

To examine the impact of spatial factors on regional disparities in green and low-carbon development, this study analyzes the contribution of within-group and between-group disparities in the Theil Index.

Table 5. Theil Index of green low-carbon development and structural cause decomposition (2011–2021).

Year	Numeric Value			Contribution Rate (%)
	Overall	Between-Group	Within-Group	Between-Group	Within-Group	Within-Group
						East	Central	West
2011	0.0013	0.0008	0.0005	37.88	62.12	38.77	9.42	13.88
2012	0.0011	0.0006	0.0005	46.79	53.21	38.80	7.76	6.99
2013	0.0013	0.0010	0.0002	19.05	80.95	24.41	21.08	35.25
2014	0.0023	0.0023	0.0000	0.43	99.57	50.24	28.72	20.79
2015	0.0032	0.0021	0.0010	32.70	67.30	36.00	7.73	23.56
2016	0.0037	0.0021	0.0016	43.43	56.57	24.35	14.64	17.60
2017	0.0029	0.0022	0.0007	22.65	77.35	8.22	15.88	53.34
2018	0.0015	0.0007	0.0009	57.79	42.21	31.37	4.09	6.38
2019	0.0034	0.0023	0.0011	31.18	68.82	42.31	23.03	3.62
2020	0.0058	0.0038	0.0019	33.39	66.44	29.83	17.44	19.20
2021	0.0027	0.0019	0.0008	30.97	69.03	40.30	12.73	15.86

reveals that the Theil Index for green and low-carbon development from 2011 to 2021 exhibits an overall fluctuating upward trend, indicating that the overall disparity in green and low-carbon development across the country is gradually increasing.

The contribution of within-group disparities is far greater than that of between-group disparities, suggesting that within-group disparities are the primary cause of the differences in green and low-carbon development nationwide. Specifically, the within-group contribution increased from 62.12% to 69.03%, while the between-group contribution decreased from 37.88% to 30.97%. The average contributions of within-group and between-group disparities are 67.59% and 32.39%, respectively, indicating that the disparity in green and low-carbon development between the three regions is relatively small, while that within each region is significant. In the eastern coastal regions, although most provinces have high green and low-carbon development, some provinces still lag behind due to historical, geographical, and other factors. Similarly, in the central and western regions, although overall development is relatively lower than that of the eastern region, the development disparities between the provinces within these regions are also significant. This internal disparity is reflected in economic aggregate as well as various aspects such as green technology, environmental policies, public awareness, and other considerations.

Although the contribution of between-group disparities to regional differences displays a declining trend, it remains a significant factor affecting coordinated regional development. Collaboration and exchange among regions are crucial for promoting national green and low-carbon development. In addition, regarding horizontal regional comparisons, the within-group disparities in China’s green and low-carbon development are primarily driven by those within the eastern region, while the internal disparities in the central and western regions contribute relatively less to the overall disparity. In terms of vertical temporal comparisons, from 2011 to 2021, the contribution of internal disparities within the eastern region increased from 38.77% to 40.30%, that within the central region rose from 9.42% to 12.73%, and that within the western region grew from 13.88% to 15.86%. This trend indicates that the impact of within-group disparities on the overall disparity in China’s green and low-carbon development level gradually intensified.

7. Conclusions and Policy Recommendations

7.1. Research Conclusions

In the context of globalization, environmental issues have become increasingly prominent alongside rapid economic development. As the largest developing country globally, China has significant responsibilities in promoting green and low-carbon development. This study employs the DPSIRM model to select evaluation indicators for green and low-carbon development, using the SA-PP model for measurement and applying kernel density estimation and Moran’s I to reflect spatiotemporal distribution characteristics. Finally, the Theil Index is used to measure regional disparities.

The study examines green and low-carbon development across 30 provinces in China from 2011 to 2021, with the following conclusions. First, notable disparities are evident among Chinese provinces in terms of various subsystems and comprehensive green and low-carbon development, a characteristic echoed by several studies [60], with the response subsystem having the greatest impact on the level of green and low-carbon development. Second, nationwide disparities are evident in green and low-carbon development among provinces, with the overall expanding trend of inter-provincial gaps being consistent with observations from studies focusing on unbalanced regional development in China [61,62]. The eastern region exhibits pronounced multipolar differentiation and spatial imbalance, the central region shows an increase in disparities in green and low-carbon development, and the western region demonstrates a trend of improvement in green and low-carbon development with fluctuations, albeit with some provinces lagging behind. Third, a clear pattern of positive aggregation in green and low-carbon development is evident among various regions in China, with green and low-carbon development levels decreasing spatially from east to west. The east-to-west decline aligns with the broader literature on regional environmental inequality within China [63]. Fourth, the overall disparity in green and low-carbon development is gradually rising, with within-group disparities being the primary reason for nationwide disparities. This conclusion consistent with findings from studies using Theil Index decomposition to explore regional environmental inequality in China [64,65].

7.2. Policy Recommendations

Based on the aforementioned conclusions, we propose three policy recommendations. First, strengthening and improving the response subsystem. The response subsystem refers to the actions or measures taken by human society in response to environmental changes or potential impacts. Our results indicate that among all subsystems, the response subsystem has the highest influence weight, accounting for 23.125%, implying the greatest impact on green and low-carbon development. Therefore, to promote green and low-carbon development in China, efforts should be focused on strengthening and improving the response subsystem. This can be achieved through three policies: (1) strengthen environmental governance frameworks with enforceable compliance mechanisms, strictly penalizing environmental violations, increasing the penalties for violations, and reducing environmental risks; (2) leverage technological innovation and policy instruments such as green finance, environmental taxes, and ecological compensation to promote the implementation of response strategies and encourage cross-sectoral and cross-disciplinary cooperation in formulating and implementing such strategies; (3) enhance environmental education and publicity through social media, online platforms, and other channels to raise public awareness about and draw attention to environmental issues.

Second, it is necessary to facilitate coordinated green and low-carbon development between regions. The research reveals a spatial pattern in which green and low-carbon development decreases from east to west. The eastern region generally exhibits higher green and low-carbon development due to advanced industrial technologies, well-developed infrastructure, and abundant resource endowments. In contrast, central and western regions lag behind in green and low-carbon development due to historical circumstances, natural conditions, and other constraints. To achieve coordinated green and low-carbon development among different regions, a series of measures must be adopted such as the following two recommendations. (1) Targeted policy bundles—combining fiscal incentives, technology transfer, and human-capital programs—should address green tech gaps in less-developed provinces. (2) It is necessary to strengthen resource reciprocity and experience sharing between regions in the various subsystems of green and low-carbon development, and to establish cooperative platforms and enhance talent exchanges in order to facilitate resource sharing, complementary advantages, and coordinated development among the three major regions in China.

Third, it is necessary to narrow the gap in green and low-carbon development between different provinces. The findings of this study indicate that within-group disparities are the primary reason for the national disparity in green and low-carbon development. To effectively address this challenge, policies should focus on narrowing the development gap. We propose three related recommendations. (1) It is necessary to provide targeted support for provinces with low full-period averages and slow growth rates. Specifically, this can be achieved by developing a series of targeted support policies supplemented by financial subsidies, tax incentives, low-interest loans, and other economic instruments for financial support. (2) It is necessary to encourage provinces with lower green and low-carbon development to leverage existing unique resource endowments and industrial bases to develop green and low-carbon industries with regional characteristics to build differentiated competitive advantages. (3) It is necessary to establish a comprehensive cooperation mechanism to promote in-depth exchanges and collaboration between provinces to advance green and low-carbon technologies, talent, policies, and related factors.

The above in-depth discussion on China’s provincial green low-carbon development is only a preliminary verification, and there are still insufficient studies on related aspects. One available study concerns the indicators and the data. When the specific indicators are selected according to the DPSIRM model framework, because some data have not been disclosed or lack of statistics, the corresponding indicators cannot be used for measurement. At the same time, due to the lack of index data in some years, the relevant models are used to fill in and supplement the data, which objectively has a certain deviation from the actual data. Future studies could further evaluate the level of green low-carbon development using more comprehensive indicators and more scientific data. Another concern is related to the sample. Overall, 30 provinces in China were selected as research units and divided into eastern and western regions for analysis. The research results can provide some reference for the carbon reduction work in various provinces and regions of China. Future studies can take cities across the country as research objects in order to obtain more comprehensive results.

8. Discussion

8.1. Future Study Recommendations

Drawing on the results and constraints of this study, future research can be advanced in these directions. First, we must broaden the research scope and data dimensions, extend the sample to all provincial-level administrative regions in China (covering Tibet, Hong Kong, Macao, and Taiwan), and utilize the most recent data to more fully capture the spatiotemporal evolution of China’s green and low-carbon development. Second, we must further supplement relevant emerging indicators to refine the indicator system of the DPSIRM model. Third, we must expand the research framework to include Belt and Road countries or major global economies, identifying distinctive paths and universal patterns of China’s green and low-carbon development via international comparisons. Fourth, we must incorporate alternative spatial weight matrices (e.g., economic, geographic, or economic–geographic mixes) to better grasp complex spatial dependencies that go beyond simple geographical adjacency.

8.2. Study Limitations

This study has certain limitations. First, restricted data coverage for Tibet, Hong Kong, Macao, and Taiwan narrowed the sample to 30 provinces, potentially limiting national-level generalizations. Second, although the indicator system covers six subsystems, it may not fully capture all dimensions of green low-carbon development. Third, the temporal analysis is restricted to 2011–2021, and longer-term trends require further exploration. Fourth, this study only adopted a first-order contiguity spatial matrix in the selection of spatial weight matrices. It failed to cover spatial correlations between regions based on other dimensions such as economic ties. Combining multiple spatial weight matrices such as economic–geographic matrices may be more helpful in revealing complex spatial dependencies in the field of low-carbon development.