Coupled coordination study of inter-provincial carbon emission efficiency and ecological infrastructure in China

Abstract

Understanding the coupling relationship between ecological infrastructure (EI) and carbon emission efficiency (CEE) is critical for China to achieve its 2030 carbon peak target. This study develops a comprehensive evaluation index system to assess inter-provincial EI development and employs the Non-Radial Directional Distance Function (NDDF) method to quantify provincial-level CEE. Subsequently, the coupling coordination degree (CCD) between EI and CEE from 2004 to 2022 is calculated using a coupling coordination model. Based on these findings, the development trend of the CCD for the period 2023–2030 is projected using the Grey Prediction GM(1,1) model. The results indicate that both EI and CEE have exhibited steady improvements between 2004 and 2022, albeit with significant regional disparities. The eastern region demonstrates the highest levels of CEE, while the western region surpasses others in EI development. The CCD between EI and CEE has increased consistently over time, revealing distinct spatial autocorrelation and clustering effects. Notably, regions characterized by low CCD (Cold spots) have decreased, whereas those with high CCD (Hot spots) have become more pronounced, although no continuous geographical pattern has emerged. Projections for 2023–2030 suggest that maintaining the current trajectory will lead to slower progress. However, targeted policies could substantially enhance CCD and mitigate regional disparities. This study provides valuable insights into the dynamic interaction between ecological infrastructure and carbon efficiency, offering strategic implications for promoting regional sustainable development, advancing environmental justice, and enhancing long-term climate resilience.

Introduction

Global climate change and accelerated urbanization have positioned cities as primary contributors to carbon emissions¹, making urban low-carbon transitions essential for advancing global climate governance and achieving China’s “dual carbon” goals. Recent studies suggest that more than 70% of global carbon emissions stem from urban activities², highlighting the pressing need to develop low-carbon cities and implement effective mitigation strategies. As one of the world’s largest emitters, China’s success in urban low-carbon transformation is critical to fulfilling its national climate commitments.

Within the urban low-carbon agenda, enhancing carbon emission efficiency (CEE) has emerged as a key pathway for aligning economic growth with emission reduction objectives^{3, 4}. CEE seeks to optimize industrial structures, foster technological innovation, and improve energy efficiency, thereby supporting a systemic transition toward low-carbon economies. Empirical studies have shown that improving CEE not only reduces carbon emission intensity per unit of output, but also facilitates green upgrading along the supply chain through technological spillovers⁵. Nevertheless, debates persist regarding the accurate measurement and effective optimization of CEE. Traditional single-factor indicators such as carbon intensity^{6, 7} and carbon productivity^{8, 9} reflect the relationship between emissions and output but overlook the joint influence of capital, labor, and energy inputs. In contrast, total-factor CEE frameworks, employing methods such as stochastic frontier analysis (SFA)¹⁰ and data envelopment analysis (DEA)^11,12,13, provide a more holistic assessment of low-carbon efficiency at the system level. More recently, advanced frontier techniques such as the non-radial directional distance function (NDDF)^{14, 15} have been utilized to address the challenges of undesirable outputs (such as carbon emissions) and environmental heterogeneity, offering more robust tools for evaluating CEE.

In parallel, ecological infrastructure (EI) has gained increasing attention as a nature-based solution for promoting low-carbon urban development¹⁶. Since UNESCO introduced the concept of EI in 1984 through the “Man and the Biosphere” program¹⁴, its definition and scope have continuously evolved. Da Silva and Wheeler expanded the concept to integrate ecosystems into infrastructure planning¹⁷; Li et al.¹⁸ proposed an integrated framework emphasizing the sustainability of urban ecosystem services; and Childers et al. further advocated for a broader conceptualization of EI¹⁹, encompassing all urban elements that support ecological functions and enhance human well-being. Empirical studies demonstrate that EI can directly reduce urban carbon emissions through mechanisms such as vegetation-based carbon sequestration, soil carbon storage, and the enhancement of ecosystem service values. For instance, Chen²⁰ found that urban green infrastructure in 35 major Chinese cities offset approximately 3.3% of China’s total carbon emissions in 2010, underscoring its potential contribution to climate mitigation. However, the full carbon reduction potential of EI remains insufficiently explored, particularly in terms of its synergistic interactions with urban economic systems.

Although both CEE and EI contribute to urban low-carbon transitions through economic efficiency and ecological service pathways respectively, their synergistic interactions have not yet been systematically explored. Existing research tends to analyze CEE and EI in isolation, overlooking their coupled dynamics across spatial scales, policy instruments, and implementation frameworks. For example, ecologically advantaged regions with high forest coverage may underutilize their carbon sink resources²¹, while industrially dominated areas may intensify ecological degradation²². This spatial heterogeneity highlights the limitations of single-dimensional strategies and underscores the need for integrated approaches that simultaneously promote environmental and economic goals. Therefore, elucidating the synergistic mechanisms linking CEE and EI, as well as developing tailored policy toolkits, has emerged as a critical challenge for advancing sustainable urban development.

This study aims to investigate the synergistic relationship between CEE and EI through empirical analysis and predictive modeling. Specifically, it seeks to uncover the coupling effects between these two dimensions and to develop a theoretical framework alongside actionable insights for policymakers striving to reconcile environmental sustainability with economic development objectives.

The research framework is presented in Fig. 1. First, a theoretical analysis is conducted to examine the synergy–tradeoff dynamics between EI and CEE. Second, panel data are utilized, with CEE measured using the Non-Radial Directional Distance Function (NDDF) method, and EI levels quantified through remote sensing data and Geographic Information System (GIS) techniques. The spatial coupling coordination between the two dimensions is then evaluated using a coupling coordination degree model. Finally, the grey prediction GM (1,1) model is employed to simulate the co-evolutionary trajectories of CEE and EI under different scenarios.

The key contributions of this study are as follows: (1)It constructs an indicator system based on conceptual foundations and environmental performance, thereby enriching the conceptual framework of EI development; (2) It integrates EI and CEE within a unified analytical framework to uncover their synergistic enhancement mechanisms; (3)It proposes practical strategies for advancing low-carbon urban development under diverse future scenarios. Collectively, these contributions enhance the theoretical foundation concerning the interlinkages between urban carbon emissions and ecological infrastructure, while offering actionable insights to support sustainable urban transformation.

Fig. 1

The article’s research route and framework.

Theoretical analysis

The synergy between CEE and EI

This study develops a bidirectional coordinated model of low-carbon development linking EI and CEE. In the direction from EI to CEE, coordination operates through two primary mechanisms:

• 1. Ecological value conversion path: Based on ecosystem services theory, the expansion of urban green spaces and wetlands not only improves environmental quality but also enhances urban livability. These improvements attract green industry clusters, creating a positive feedback loop in which ecological investment contributes to improved economic efficiency. Over time, this feedback is amplified as environmental benefits, such as improved air quality, enhanced biodiversity, and aesthetic value, continue to support economic activity and foster the emergence of sustainable industries²³.

• 2. Carbon asset monetization path: Drawing on carbon asset theory and ecological compensation mechanisms, carbon sink assets generated by EI, such as urban forests and wetland restoration projects, can be monetized through carbon trading markets. This monetization generates economic returns that can be reinvested into further ecological development, thereby enhancing carbon-related economic efficiency and establishing a self-reinforcing cycle of ecological investment and financial benefit²⁴.

In the reverse direction, from CEE to EI, three feedback mechanisms are identified:

• 1. Technological innovation effect: Based on Porter’s Hypothesis and innovation diffusion theory, the adoption of energy-efficient and environmentally friendly technologies accelerates green urban transformation and strengthens the carrying capacity of EI. These technological advances promote the integration of green technologies into urban infrastructure, supporting the expansion, resilience, and sustainability of EI systems²⁵.

• 2. Spatial resource reallocation effect: As high-carbon industries are gradually phased out, valuable urban land is freed and can be repurposed for EI development, such as urban parks, green roofs, or renewable energy installations. This spatial reallocation creates critical physical space to support the sustainable expansion of ecological infrastructure in urban contexts.

• 3. Regulatory pressure effect: Carbon performance assessment mechanisms—such as legally binding emissions targets and carbon reduction goals—create regulatory pressure that incentivizes local governments to increase investment in ecological governance and EI development. These regulations ensure that ecological infrastructure receives sustained financial and policy support, even amid competing development priorities.

This bidirectional coordination, through the dual dynamics of the “economic-ecological” value cycle and “technology-institution” spatial restructuring, simultaneously enhances carbon efficiency and increases the carbon sequestration capacity of EI. Ultimately, this forms a synergistic loop of “efficiency enhancement and ecological expansion” that supports regional low-carbon development.

The trade-offs between CEE and EI

While the synergy between EI and CEE holds significant potential for advancing sustainable development, their joint promotion inevitably encounters inherent trade-offs, particularly in resource allocation, temporal alignment, and spatial equity.

• 1. Resource Competition: One major challenge stems from competition over limited resources. Both EI initiatives^{26, 27} and CEE measures²⁸ require substantial investments in capital, land, and technology. However, as many EI projects are typically treated as public goods, they often face budgetary constraints that limit funding availability for carbon reduction technologies and industrial transformation. Consequently, governments and organizations must make difficult decisions when prioritizing between short-term carbon efficiency improvements and long-term ecological investments. To reconcile these competing demands, innovative policy instruments, such as green bonds and hybrid financing models, are essential²⁹.

• 2. Temporal Mismatches: Temporal asymmetries further complicate the coordination between EI and CEE. The benefits of EI, including biodiversity enhancement, carbon sequestration, and climate resilience, typically accumulate over an extended period. By contrast, technological carbon mitigation measures, including the deployment of renewable energy systems and energy-efficient infrastructure, often deliver immediate emissions reductions. This temporal mismatch presents a critical policy dilemma: decision-makers must weigh the urgency of short-term emissions targets against the need for sustained investment in ecological infrastructure. Achieving a balanced approach that aligns short-term action with long-term sustainability is key to fostering a resilient, low-carbon future³⁰.

• 3. Spatial Disparities: Spatial heterogeneity in the distribution of EI benefits presents another significant challenge. Some regions directly benefit from ecological interventions, such as carbon sinks and improved urban environments, while others, particularly high-emission urban centers, may remain neglected. This exacerbates regional disparities in the outcomes of low-carbon transitions. To address this imbalance, targeted policy frameworks and spatially adaptive interventions are needed to ensure that carbon-intensive regions are not excluded from the sustainable transition. Moreover, integrating advanced technologies such as GIS and satellite-based monitoring can help identify priority areas for EI investment and improve spatial coordination^{24, 31}.

Methods

Study area

China is situated on the eastern coastline of the Eurasian continent, extending from 73°33′ to 135°05′E longitude and from 3°51′N to 53°33′N latitude. Its diverse geography and climate range from humid coastal regions to arid interior areas. The country is divided into three climate zones: tropical, subtropical, and temperate. With a high-west, low-east topography and several mountain ranges, China experiences significant regional climate variation. This study focuses on mainland China, analyzing 30 provinces, excluding Tibet, Hong Kong, Taiwan, and Macao (Fig. 2).

Fig. 2

Study area.

CEE measurement model: NDDF

The NDDF method enables asymmetric adjustments between inputs and outputs, allowing for a more accurate representation of the different resources’ roles in carbon reduction. Its flexibility effectively addresses the complex relationship between carbon emissions and reduction measures, providing more detailed efficiency evaluations and enhancing the accuracy of CEE calculations^{14, 15, 32}:

$$P(x) = \left\{ {\left( {{x^t},{y^t},{b^t}} \right):\sum\limits_{j = 1}^J {z_j^t} x_{jm}^t \leqslant x_m^t,\sum\limits_{j = 1}^J {z_j^t} y_{jn}^t \geqslant y_n^t,\sum\limits_{j = 1}^J {z_j^t} b_{jk}^t} \right.n\left. { = b_k^t,\forall j,m,n,k} \right\}$$

(1)

$$\vec {D}(x,y,b;{\mathbf{g}})=\sup \left\{ {{{\mathbf{\omega }}^ \top }{\mathbf{\beta }}:(y,b,x)+{\mathbf{g}} \times \operatorname{diag} (\beta ) \in P(x)} \right\}$$

(2)

In the formula: P(x) represents the Production Possibility Set of the DMU, encompassing all feasible input-output combinations. z^t_j is the weight for each cross-section of the technology frontier, where z^t_j ≥ 0, and w is the weight vector related to the number of input-output factors. g is the direction vector, while β is a scaling factor representing the ratio of desired output expansion and undesirable input/output reduction. $\vec {D}(x,y,b;{\mathbf{g}})$ measures the maximum efficiency improvement along direction g, considering inputs x, desired outputs y, and undesirable outputs b. diag(β) constructs a diagonal matrix from vector β, and sup denotes the supremum, the least upper bound of a set.

The specific calculation results can be obtained using linear programming, which is expressed as follows³³:

$$\begin{aligned} \vec D(K,L,E,Y,C;g) & = \max \left( {{w_K}{\beta _K} + {w_L}{\beta _L} + {w_E}{\beta _E} + } \right.\left. {{w_Y}{\beta _Y} + {w_C}{\beta _C}} \right){\text{ }} \\ & {\text{s}}{\text{.t}}{\text{. }}\sum\limits_{j = 1}^J {{\lambda _j}} {K_j} \leqslant K + {\beta _K}{g_K} \\ & \sum\limits_{j = 1}^J {{\lambda _j}} {E_j} \leqslant E + {\beta _E}{g_E} \\ & \sum\limits_{j = 1}^J {{\lambda _j}} {Y_j} \geqslant Y + {\beta _Y}{g_Y} \\ & \sum\limits_{j = 1}^J {{\lambda _j}} {C_j} \leqslant C + {\beta _C}{g_C} \\ & {\lambda _i} \geqslant 0,{\beta _K},{\beta _L},{\beta _E},{\beta _Y},{\beta _C} \geqslant 0 \end{aligned}$$

(3)

In this formula, K, L, E, Y, and C represent capital, labor, energy, desired output, and undesirable output, respectively. The ratios of inputs (capital, labor, energy), desired output, and undesirable output are each set to 1/3. Thus, the weights w_k, w_l, w_e, w_t, and w_c reflect their relative importance in the efficiency evaluation: w_k = w_l = w_e = 1/9, and w_t = w_c = 1/3. The coefficients β_k, β_l, β_e, β_t, and β_c adjust the influence of each factor. λ_j denotes the weight of each DMU, representing its contribution to efficiency. The components of the directional vector are: g_k = g_l = g_e = -1, g_t = 1, and g_c = -1, which guide the optimization by minimizing undesirable outputs and maximizing desired outputs.

The CEE for each province is calculated by comparing actual performance with optimal performance over time. The formula incorporates desired output (Y_it), inputs (capital K_it, labor L_it, energy E_it), and undesirable output (C_it). Scaling factors (β_K, β_L, β_E, β_C) adjust the impact of each input and output on the efficiency score. This approach enables the assessment of how effectively provinces convert inputs into outputs while accounting for environmental costs. The optimal scaling factors ($\beta _{{Y,it}}^{*}$,$\beta _{{K,it}}^{*}$,$\beta _{{L,it}}^{*}$,$\beta _{{E,it}}^{*}$,$\beta _{{C,it}}^{*}$) for each province are derived over time.

$${\operatorname{CEE} _{it}}=\frac{1}{4}\left( \begin{gathered} \frac{{{Y_{it}}/{K_{it}}}}{{\left( {{Y_{it}}+\beta _{{Y,it}}^{*}{Y_{it}}} \right)/\left( {{K_{it}} - \beta _{{K,it}}^{*}{K_{it}}} \right)}}+\frac{{{Y_{it}}/{L_{it}}}}{{\left( {{Y_{it}}+\beta _{{Y,it}}^{*}{Y_{it}}} \right)/\left( {{L_{it}} - \beta _{{L,it}}^{*}{L_{it}}} \right)}} \\ +\frac{{{Y_{it}}/{E_{it}}}}{{\left( {{Y_{it}}+\beta _{{Y,it}}^{*}{Y_{it}}} \right)/\left( {{E_{it}} - \beta _{{E,it}}^{*}{E_{it}}} \right)}}+\frac{{{Y_{it}}/Cit}}{{\left( {{Y_{it}}+\beta _{{Y,it}}^{*}{Y_{Yt}}} \right)/\left( {{C_{it}} - \beta _{{C,it}}^{*}{C_{it}}} \right)}} \\ \end{gathered} \right)$$

(4)

Coupling model

The Coupling Model is widely utilized in ecological and environmental studies, particularly for analyzing the interactions between various systems, such as the relationship between human activities and the environment.

• 1. Coupling Degree Model

  The coupling degree describes the level of interaction between systems; a higher coupling degree indicates a higher level of interaction between the systems. The formula is:

  $$C=2\sqrt {\left( {{U_1} \times {U_2}} \right)/{{\left( {{U_1}+{U_2}} \right)}^2}}$$

  (5)

  Where C represents the coupling degree, which ranges from 0 to 1. U₁ and U₂ are the values of the subsystems under study.

• 2. CCD Model

  The CCD reflects the intensity of interactions between systems; a higher CCD indicates a higher level of coordination between systems.

  $$CCD=\sqrt {C \times T} ,T=\alpha {U_1}+\beta {U_2}$$

  (6)

  In this formula, CCD represents the degree of coupling coordination between two systems, ranging from 0 to 1. T represents the comprehensive coordination index of the subsystem under study. α and β are coefficients to be determined, and in this study, both subsystems are considered equally important; hence, α = β = 0.5.

• 3. Coupling Coordination Types

  According to previous research³⁴, coupling coordination types are divided into two main categories: " Dysfunctional Decline” and “Coordinated Development”, as detailed in Table 1.

Table 1 Types of coupled development.

Spatial autocorrelation model

Spatial autocorrelation analysis effectively characterizes the spatial pattern of coordinated development, addressing the limitations of traditional coupling coordination models in spatial dimension analysis. This method is applied at both global and local scales: global spatial autocorrelation is measured using Moran’s I index to assess the overall spatial correlation of neighboring units, revealing regional collaboration patterns; local spatial autocorrelation, based on the Getis-Ord Gi*(d) index, precisely identifies spatial heterogeneity within the region, providing a spatial perspective for the detailed analysis of coordination degree differences. The formula is as follows:

$$\begin{gathered} {\text{Moran's I =}}\sum\limits_{{{\text{i=1}}}}^{{\text{n}}} {\sum\limits_{{{\text{j=1}}}}^{{\text{n}}} {{{\text{W}}_{{\text{ij}}}}} } \left( {{{\text{x}}_{\text{i}}} -\bar{\text{ {x}}}} \right)\left( {{{\text{x}}_{\text{j}}} -\bar{\text{{x}}}} \right){\text{/}}{{\text{s}}^{\text{2}}}\sum\limits_{{{\text{i=1}}}}^{{\text{n}}} {\sum\limits_{{{\text{j=1}}}}^{{\text{n}}} {{{\text{W}}_{{\text{ij}}}}} } \hfill \\ {\text{G}}_{{\text{i}}}^{{\text{*}}}({\text{d}}){\text{=}}\sum\limits_{{{\text{j=1}}}}^{{\text{n}}} {{{\text{W}}_{{\text{ij}}}}} {{\text{x}}_{\text{j}}}{\text{/}}\sum\limits_{{{\text{j=1}}}}^{{\text{n}}} {{{\text{x}}_{\text{j}}}} \hfill \\ \end{gathered}$$

(7)

Where $\bar{\text{{x}}}$ and s² represent the mean and standard deviation of the variable x, respectively. n is the number of research units. x_i​and x_j are the attribute values of spatial units i and j, respectively. W_ij​is the spatial weights matrix.

Grey prediction model

The grey prediction model is a forecasting method suitable for situations with incomplete information. It establishes mathematical models and makes predictions using limited, incomplete data. The core idea involves processing the original data sequence to reduce randomness and reveal inherent laws in the data³⁵.

First, we assume the original sequence as X₀. By accumulating X₀, we generate a new series X₁:

$${X_0}=\left[ {{x_0}(1),{x_0}(2), \ldots ,{x_0}(n)} \right]$$

(8)

$${X_1}=\left[ {{x_1}(1),{x_1}(2), \ldots ,{x_1}(n)} \right]{\text{ , }}{x_1}(k)=\sum\limits_{{i=1}}^{k} {{x_0}} (i),k=1,2 \ldots ,n$$

(9)

Then, we use a differential equation for calculation:

$$\frac{{d{X_1}}}{{dt}}+a{X_1}=b$$

(10)

Where a represents the developmental grey action, and b represents the intrinsic control grey number. Let’s assume the parameter vector to be estimated and solve the equation using the least squares method, resulting in the following prediction model:

$$x_{1}^{T}(k+1)=\left[ {{x_0}(1) - b/a} \right]{{\text{e}}^{ - {\text{ak}}}}+b/a,k=1,2, \ldots ,n$$

(11)

Finally, before proceeding with the forecasts, the validity of the prediction model is verified. The Mean Absolute Percentage Error (MAPE) is an ideal indicator for measuring the accuracy of model predictions. The smaller the MAPE value, the more accurate the model is. The expression and evaluation standards for MAPE can be found in the formula and Table 2.

$$MAPE=\frac{1}{{n - 1}}\sum\limits_{{k=2}}^{n} {\left| {\frac{{{x_0}(k) - \hat {x}(k)}}{{{x_0}(k)}}} \right|} \times 100\%$$

(12)

Where $\hat {x}(k)$ represents the predicted value.

Table 2 Model evaluation criteria by MAPE.

Variable selection and data sources

Variable selection for CEE

This study focuses on panel data from 30 Chinese provinces (excluding Hong Kong, Macao, Taiwan, and Xizang) from 2004 to 2022. The CEE for each province is calculated using the NDDF model. Input indicators include total fixed asset investment, employment, and total energy consumption, with data sourced from the “China Statistical Yearbook” and provincial yearbooks^{15, 36}. The desired output indicator, GDP, is sourced from the “China Statistical Yearbook.” The undesired output indicator, total carbon emissions, is calculated using the IPCC method. For missing data, interpolation methods like averaging or extrapolation are used to ensure data completeness and continuity.

EI development index

With the advancement of research on urban sustainability, the concept of EI has undergone continuous expansion and refinement^17,18,19, its scope has broadened to include a wide array of urban components that support ecological functions and human well-being, encompassing both natural systems and built structures.

Building on this evolving understanding, this study defines EI as a networked system composed of blue, green, grey, and other infrastructure elements that perform ecological functions³⁷. This definition emphasizes the integration of natural and artificial systems, and underscores the role of EI in supporting ecosystem services, enhancing environmental resilience, and promoting urban livability.

Accordingly, this study establishes a multidimensional indicator system for evaluating EI, comprising the following five key dimensions:

• 1. Blue Infrastructure: Includes natural water bodies (lakes, rivers, canals, oceans) that provide water management, purification, flood regulation, and biodiversity protection. Its health is reflected by changes in the water body area^{38, 39}.

• 2. Green Infrastructure: Consists of forests, parks, and green spaces, contributing to climate regulation, carbon sequestration, oxygen release, and residents’ well-being. Key indicators include per capita green space, green space coverage, forest coverage, park accessibility, and convenience^{38, 39}.

• 3. Grey Infrastructure: Comprises man-made structures (buildings, roads) requiring careful planning to maintain ecological balance. Indicators include construction land area, road area, and nighttime light data for energy consumption and carbon emissions^{38, 39}.

• 4. Other Ecologically Functional Infrastructures: Handle urban pollutants’ collection, treatment, and recycling. Indicators include waste treatment rate, resource recycling rate, and environmental governance investment proportion to GDP.

• 5. Environmental performance: In constructing the EI development index system, environmental performance directly reflects ecological protection effectiveness⁴⁰. This is measured by air quality and pollution emission intensity, using specific indicators such as annual average pollutant concentrations ( PM₁₀, SO₂, NO₂), SO₂ emission intensity³⁵, COD emission intensity, and ammonia nitrogen emission intensity.

The assessment system uses the Analytic Hierarchy Process (AHP) and Delphi method to determine weights. Logarithmic transformation and entropy weighting methods ensure accurate indicator weights. The overall indicator system and the weight values and data sources of secondary indicators are shown in Tables 3 and 4.

Table 3 Evaluation system for EI development level.

Table 4 Data sources and processing methods for EI.

Results

Spatiotemporal variation of CEE and EI development index

Spatiotemporal variation of CEE

Fig. 3

National, regional and provincial trends of CEE in China. (a) National CEE trends and growth rate of CEE. (b) CEE in Eastern, Western, and Central provinces in China and growth rate of CEE (refer to the footnotes, again as follows). (c) CEE of Chinese provinces from 2004 to 2022. Eastern Provinces: Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, Hainan. Central Provinces: Shanxi, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan. Western Provinces: Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, Xinjiang.

From 2004 to 2022, China’s CEE demonstrated a general upward trend. However, regional disparities and polarization among provinces persisted throughout the period. The national CEE increased from 0.1693 in 2004 to 0.3422 in 2022, with significant fluctuations observed early on and steady growth in the later years. (Fig. 3(a)). These results align closely with previous studies conducted at the national level from 2000 to 2016³⁶. The growth was slow and unstable from 2004 to 2010, followed by consistent improvement from 2011 to 2016. The most substantial increase occurred between 2017 and 2022, peaking in 2021. Regionally, the eastern region exhibited the fastest growth, while the central region showed rapid but volatile growth. The western region showed the slowest growth, marked by significant fluctuations. (Fig. 3(b)). At the provincial level, Beijing and Shanghai achieved the highest CEE efficiencies, reaching 1 and 0.76, respectively, while Henan, Hunan, Anhui, and Hainan showed efficiencies between 0.3 and 0.4. Provinces such as Hebei, Ningxia, Qinghai, and Xinjiang had efficiencies around 0.2, indicating the need for stronger green transition policies (Fig. 3(c)).

Spatiotemporal variation of EI development index

Fig. 4

National, regional and provincial trends of EI Development Index in China. (a) National EI trends. (b) EI of Chinese provinces from 2004 to 2022. (c) EI and growth rate of EI in Eastern, Western, and Central provinces in China.

Between 2004 and 2022, China’s National EI Development demonstrated consistent growth, although significant regional divergence persisted.

The national EI development index increased from 0.3965 to 0.4985 between 2004 and 2022 (Fig. 4(a)). This trend aligns with the findings of Sun et al.³⁷, which reported similar patterns at the prefecture-level cities scale between 2000 and 2010. Growth was rapid from 2004 to 2010, followed by a slowdown from 2011 to 2015, and notable fluctuations from 2016 to 2022. Key increases occurred in 2007 and 2016, while declines were observed in 2019 and 2022.

Regionally, the western region exhibited the highest and most stable growth, rising from 0.4046 in 2004 to 0.5158 in 2022 (Fig. 4(c)). In contrast, the central region saw more volatility, with its index increasing from 0.3868 to 0.4948, particularly marked by fluctuations in 2007, 2009, and 2016. The eastern region showed consistent growth, increasing from 0.3955 to 0.4839.

At the provincial level, Hainan, Shaanxi, Xinjiang, and Tibet displayed substantial growth, particularly between 2010 and 2016, and from 2016 to 2022 (Fig. 4(b)). In contrast, provinces such as Henan, Jiangxi, and Anhui exhibited considerable fluctuations in growth. Beijing, Tianjin, Shanghai, Guangdong, and Hunan demonstrated steady improvements, while Hebei, Shandong, Liaoning, Jiangsu, and Zhejiang displayed continuous development.

Coupling degree measurement and feature analysis

Coupling coordination

Fig. 5

National, regional and provincial trends of CCD Development Index in China.

From 2004 to 2022, the national and regional CCD in China generally showed an upward trend. The eastern region exhibited the highest coordination, and the national coordination level progressed from basic coordination to intermediate coordination. However, the gap between regions gradually widened, and the number of provinces at different coordination stages fluctuated over time.

Figure 5(a) shows the trend of CCD changes at the national and regional levels from 2004 to 2022. During this period, the national average CCD, as well as the CCD for the eastern, central, and western regions, all exhibited a gradual increase. However, a significant decline occurred in 2010, followed by a slow recovery, with a slight dip in 2015 before steadily rising again.

In terms of numerical comparison, the eastern region had the highest CCD, followed by the national average, the central region, and the western region. Specifically, the national average CCD increased from 0.5052 in 2004 to 0.6315 in 2022. Developmentally, the national CCD transitioned from the Basic Coordination stage to the Intermediate Coordination stage, reflecting an overall improvement in coordination.

Figure 5(b) indicates significant regional disparities in the CCD of provinces from 2004 to 2022, with a diffusion trend over time, suggesting that disparities between provinces have been widening. Figure 5(c) illustrates the distribution of provinces across different coordination levels: provinces with mild dysfunction steadily decreased to zero, while the number of provinces with basic coordination first increased and then decreased, reaching 11 in 2022. The number of provinces with intermediate coordination continued to rise, reaching 16 by 2022. Provinces with good coordination emerged in 2015, with the number increasing to 3 by 2020. High-quality coordination provinces appeared in 2021, with 2 provinces reaching this level by 2022.

Coupling feature analysis

The coupling features between CEE and EI development index can be summarized as follows: (1) There is a three-segment spatial distribution, progressing from the northwest to the central region, and then to the southeast. This pattern has evolved from a non-centralized distribution to a single-center pattern, and finally to a multi-center distribution. Core cities like Beijing and Shanghai drive this trend, although regional development remains uneven. (2) Nationwide, the CCD has shown an upward trend. However, only a few provinces have reached the “good coordination” stage or higher. Over time, there has been a decrease in the number of provinces in the mild dysfunction and basic coordination stages, while the number of provinces in the intermediate coordination stage has increased. By 2022, only Shanghai and Beijing achieved high-quality coordination. (3) Provinces in similar coupling coordination stages tend to cluster, with some exceptions. CCD values are lower in the northwest, moderate in the central regions, and higher in the southeast.

Classification of provincial CEE-EI coupling patterns

The CCD assesses the level of coordinated development between systems but doesn’t reveal the internal reasons for coordination. Provinces are categorized based on median CEE and EI values and further subdivided by CCD values (Fig. 6; Table 5):

• 1. High-High Type: Provinces that excel in both CEE and EI, achieving high CCD. For instance, Shanghai has significantly reduced emissions from industrial and transportation sectors while advancing urban greening and water management. Similarly, Jiangsu has made notable progress in industrial transformation and ecological protection.

• 2. Low-Low Type: Provinces that fall below the median in both CEE and EI, indicating neglect in environmental protection and EI during economic development, resulting in low CCD. For example, Ningxia has high carbon emissions and low energy efficiency, with insufficient EI investment. Shanxi, a coal-rich region, faces significant challenges in economic restructuring and ecological protection.

• 3. Carbon Efficiency Leading Type: Provinces with high CEE but below-median EI, excelling in energy conservation and emission reduction but needing improvement in EI construction to enhance CCD. Beijing, for example, has reduced carbon emission intensity significantly but needs to expand urban green areas and improve water quality.

• 4. Ecological Infrastructure Advantage Type: Provinces with high EI but low CEE, performing well in ecological protection and infrastructure but needing better carbon emission control and energy efficiency. Yunnan, for example, has high forest coverage and abundant water resources but requires improvement in CEE.

Table 5 Distribution of coupling classification by province.

Fig. 6

CEE and EI development index by province.

Spatial distribution layout

Global autocorrelation

Table 6 presents the Moran’s I index and the corresponding P-values. From a statistical significance perspective, the P-values consistently fall below 0.05, indicating that the distribution of the CCD across space is not random but rather exhibits spatial clustering. However, over time, the P-values show an increasing trend, suggesting that the significance of spatial autocorrelation is gradually diminishing.

Meanwhile, the Moran’s I index initially increased and then decreased, indicating that spatial correlation first strengthened and then weakened. This suggests that regions with similar CCD values initially became more clustered. As regions began to exhibit greater variation in CEE and EI, the clustering effect weakened, although it remained present.

Table 6 Distribution of Moran’s I index and P-values.

Global moran’s index analysis

An analysis of the CCD for CEE and EI development across 30 provinces in China was conducted using clustering and outlier analysis. The results from 2004 to 2022 reveal two prominent clusters: low-low and high-high, while other provinces did not exhibit significant clustering characteristics (Fig. 7).

Fig. 7

The clustering distribution of CCD in different provinces in 2004, 2010, 2016, 2022.

Initially, the low-low cluster included Gansu, Shaanxi, Yinchuan, Ningxia, and Inner Mongolia. However, by 2022, it was reduced to only Inner Mongolia and Gansu, regions characterized by lower coordination in CEE and EI development. This reduction can likely be attributed to the challenges associated with transitioning resource-based economies. On the other hand, the high-high cluster, concentrated in Zhejiang, Jiangsu, Shanghai, Fujian, and Jiangxi, exhibited high levels of coordination in both CEE and EI. Notably, since 2016, Zhejiang has consistently remained in this high-performing cluster. In 2022, Sichuan emerged as a high-low outlier, demonstrating a strong CCD while its neighboring provinces displayed lower levels of coordination.

Local autocorrelation

Fig. 8

Hot spot and cold spot analysis of CCD in 2004, 2010, 2016, 2022.

Global spatial autocorrelation reveals the spatial clustering of CCD. The Getis-Ord (G*i) values for CCD between CEE and EI were calculated for the years 2004, 2010, 2016, and 2022. These values were classified into seven categories: high-significance Hot Spot, medium-significance Hot Spot, low-significance Hot Spot, random distribution areas, low-significance Cold Spot, medium-significance Cold Spot, and high-significance Cold Spot.

Figure 8 illustrates the characteristics of spatial distribution. Cold Spot represent areas with low CEE and EI coordination, reflecting delayed progress in green development, particularly in the northwestern regions (such as Gansu, Shaanxi, Ningxia). In contrast, Hot Spot correspond to regions with high CEE and EI coordination, signifying significant progress in sustainability, mainly found in the southeastern coastal regions (such as Zhejiang, Jiangsu, Shanghai). Over time, the Hot Spot have become more dispersed, while the Cold Spot have decreased but remain clustered. Notably, Cold Spot are never adjacent to Hot Spot, highlighting the regional disparities in coordination levels and the unequal progress of green development across different areas.

Prediction of coupling development level

Model validation

Data related to the CCD of various provinces from 2004 to 2017 were input into the Grey System GM (1,1) model to simulate and forecast the coupling coordination values for 2018 to 2022. The predicted values were compared with the actual values, and the Mean Absolute Percentage Error (MAPE) was calculated. The results, shown in Table 7, indicate that except for Shanxi, the MAPE values for all other provinces were less than 10%, demonstrating high predictive accuracy. The forecast period extends from 2023 to 2030.

Table 7 Forecasted MAPE (%) for each province.

Scenario setting

Scenario 1 (Maintenance): Maintain the current development state without implementing new measures, keeping existing carbon emission reduction policies and EI development unchanged.

Scenario 2 (Tailored Strategies for Different Development Types):

• 1. High-High Type: Sustain high performance in both CEE and EI development, reinforcing their leading position.

• 2. Low-Low Type: Implement measures to improve both CEE and EI development by 1% annually.

• 3. Carbon Efficiency Leading Type: Preserve the advantage in CEE while enhancing EI development through increased investments and optimized management, aiming for 1% annual growth.

• 4. Ecological Infrastructure Advantage Type: Boost CEE through technological and model innovations, targeting a 1% annual improvement.

Scenario forecasting

Using the GM (1,1) forecasting model, the evolution of the coupling between CEE and EI development under two different scenarios was simulated. Here are the projected outcomes:

Scenario 1 (Maintenance): Under this scenario, the CCDs of provinces develop well, with an average value of 0.6668, indicating an overall good coordination stage. By 2030, still, 9 provinces will remain in the basic coordination stage, 13 provinces will have reached a good intermediate stage, 6 provinces will have entered a high-quality coordination stage, and only Beijing and Shanghai have entered a high-quality stage (Table 8). This scenario suggests that without any new measures, while most provinces can maintain or improve their coordination levels, some areas still face issues with lower coordination degrees.

Table 8 Distribution of provincial coupling stages from 2023 to 2030 based on scenario 1 prediction.

Scenario 2 (Tailored Strategies for Different Development Types): Following the implementation of targeted enhancement measures across different provinces, the previously observed uneven development has shown significant improvement, with disparities between provinces further narrowing. This positive trend has accelerated the overall evolutionary pace, projecting that by 2030, the national CCD will reach 0.6734 (Table 9). Although this remains within the intermediate coordination stage, it represents a notable improvement.

By 2030, Beijing and Shanghai are expected to maintain their leading positions, remaining in the high-quality coordination stage, with their CCDs approaching 100%. Additionally, only seven provinces will be at the basic coordination stage, while 15 provinces will reach the intermediate coordination development stage.

To transition to a higher level of good coordination development, further stimulation and promotion measures will be essential. These efforts will help bridge the gaps and promote balanced and sustainable growth across all regions, ultimately enhancing the overall CCD nationwide.

Table 9 Distribution of provincial coupling stages in 2023–2030 based on scenario 2.

Under the scenario 1, although the level of coordinated regional development in China has improved, there are still problems of relatively low coordination in some regions. Under the customized strategy scenario, the imbalance in regional development has been improved, and the overall level of coordinated development has significantly increased.

Discussions

Recommendations

• 1. Based on panel data from 30 Chinese provinces over the period 2004–2022, this study systematically examines the coupling coordination relationship between CEE and EI. The results reveal a steadily strengthening synergy nationwide, indicating a fundamental shift in China’s urban development paradigm under the dual-carbon strategy. This transformation from a resource-intensive model to a green, low-carbon, and high-efficiency trajectory also reflects the practical implementation of the “lucid waters and lush mountains are invaluable assets” concept. Compared with prior studies that often focused on static or single-dimensional relationships²⁰, this work contributes a temporally dynamic, spatially differentiated perspective on the evolving interaction between ecological and efficiency metrics.

• 2. Forecasting via the GM(1,1) grey model suggests that, without targeted policy intervention, the CEE–EI synergy may grow slowly and fail to generate sustained endogenous momentum. In light of considerable regional disparities in development stages and resource endowments, a classification-based, context-specific policy framework is required. For high-CEE but low-EI provinces (such as Beijing, Shanghai), ecological connectivity should be strengthened through enhanced green–blue infrastructure^{38, 39, 43}. For high-EI but low-CEE regions (such as Yunnan, Inner Mongolia, Qinghai), policies such as “clean energy base” initiatives can foster the integration of ecological and low-carbon sectors⁴⁴. Provinces with strong EI but moderate CEE (such as Guangdong, Fujian, Zhejiang) should focus on improving energy efficiency¹¹ across key industrial and transport domains by leveraging policy platforms like low-carbon city pilots. For dual-low regions (such as Gansu, Shanxi, Guizhou), establishing green industrial parks and zero-carbon demonstration zones, supported by fiscal transfers and green investments, may provide the structural impetus needed for ecological and carbon efficiency breakthroughs. This typology provides a co-evolutionary framework distinct from traditional region-based policies, offering more actionable guidance for differentiated ecological transition strategies.

• 3. Achieving long-term coordination between CEE and EI also requires institutional synergy across key policy instruments—particularly green finance⁴⁵, carbon emissions trading⁴⁶, and ecological compensation⁴⁷. These tools together form the primary pathway for ecological asset value realization. Green finance facilitates long-term capital allocation through credit, bonds, and investment funds targeting restoration, renewable energy, and low-carbon technologies. Carbon trading promotes efficient resource distribution by incentivizing emissions reductions through market mechanisms. Ecological compensation redistributes costs and benefits between environmental stewards and beneficiaries. Synergistic implementations of these mechanisms are emerging in practice; for example, in Fujian Province, forestry-based carbon sink projects have simultaneously accessed carbon markets, received fiscal compensation, and obtained green financing, demonstrating a closed-loop system linking ecological asset value, market operation, and financial support. To scale such models, standardizing green asset valuation and carbon sink accounting is essential. Integrating ecological projects into both carbon markets and compensation schemes, alongside financial support, would institutionalize ecological asset monetization under a multi-instrument governance framework. Future research could explore how these co-evolving mechanisms perform across different regulatory regimes, spatial scales, and institutional capacities.

Limitations

• 1. This study focuses solely on exploring the coupling relationship between carbon emission systems and EI development systems in various provinces, without considering the impact and changes of other systems such as the socio-economic system. Consequently, predictions of future CCD are limited to these two systems. However, changes in external factors such as urbanization^{36, 48} and population dynamics^{49, 50} can also profoundly affect the performance of these two systems.

• 2. The CCD method is a valuable tool for analyzing interactions and coordination between systems, capable of revealing coupling changes. However, when dealing with complex system relationships, this method often overlooks the influence of external environmental factors, making it challenging to accurately elucidate causal and dominant relationships between systems⁵¹. Although this study has revealed the coupling effects between the two systems by classifying various provinces, future research should focus on enhancing and refining this method to better understand and address the interaction mechanisms in complex systems.

Future research

The spatial heterogeneity of CEE and EI across different regions in China is likely shaped by a complex interplay of factors, including the stage of economic development, industrial structure, and the effectiveness of policy implementation. However, existing research lacks systematic empirical validation, and this theoretical hypothesis requires further refinement through multi-dimensional data analysis and cross-regional comparative studies.

At the policy implementation level, instruments such as the national unified carbon market and renewable energy subsidies have generated stronger emission reduction incentives in the eastern region. By contrast, the western region—due to limited fiscal transfer payments and weaker technological absorption capacity—has experienced delays and diminishing effects in responding to these policies. This regional asymmetry in policy effectiveness may further exacerbate spatial disparities in achieving national carbon neutrality goals.

Future research should therefore focus on mechanisms for enhancing cross-regional cooperation and policy coordination, with particular emphasis on strategies that support the western region in overcoming barriers to green transformation. Promoting balanced and coordinated low-carbon development at the national level remains a critical research priority for achieving equitable and sustainable climate outcomes. Additionally, future research will delve into exploring the temporal relationship between CEE and EI more comprehensively using city-level data, which will help uncover finer-scale dynamics and deepen our understanding of regional transitions.

Conclusion

This study analyzed the distribution characteristics and evolutionary process of CCD between CEE and EI, leading to the following conclusions:

• 1. From 2004 to 2022, both CEE and EI development indices showed an upward trend across Chinese provinces. CEE was highest in the eastern region, followed by the central and western regions. EI was highest in the western region, followed by the central and eastern regions.

• 2. From 2004 to 2022, The overall CCD in China increased, with an initial acceleration that later slowed down, resulting in a widening gap between provinces. By 2022, China’s coordination level was moderate, with higher CCD in the east, relatively lower in the central region, and lowest in the west, forming a multi-center pattern with Beijing and Shanghai as cores.

• 3. Significant spatial autocorrelation was observed, with spatial agglomeration first strengthening and then weakening. Global autocorrelation revealed low-low and high-high clusters, while local autocorrelation showed evident Hot Spot and Cold Spot. Hot Spot increased and dispersed, while Cold Spot decreased and concentrated.

• 4. Using GM (1,1), the CCD of each province is predicted to show a slow upward trend. By 2030, overall coupling coordination between CEE and EI in China is expected to reach a good stage. Precise strategies are projected to significantly accelerate CCD improvement, especially in provinces with lagging CEE and EI development.