Studying carbon emission effects on ecological welfare performance in Chengdu–Chongqing urban agglomeration using super efficiency network SBM

Abstract

Urban agglomerations are crucial for achieving dual-carbon goals, and the interaction between carbon emissions and ecological welfare performance (EWP) remains unclear. EWP refers to the efficiency of converting ecological resources into human welfare. Taking the Chengdu-Chongqing Urban Agglomeration (CCUA) during 2012–2022 as a case, this study decomposes EWP into two sub-stages, constructs an evaluation system including ecological input, economic level, undesirable output and welfare output, and calculates EWP of 176 samples using the Super-Efficiency Network Slack-Based Measure Model (Super-NSBM). This study further analyzes EWP’s evolution mechanism through panel Tobit model and spatial econometric analysis, revealing the temporal and spatial characteristics and driving factors of EWP in CCUA. Results show: (1) The average EWP of CCUA was 0.629, mainly limited by low economic welfare efficiency. (2) Spatially, EWP of eastern cities in CCUA was higher and that of western cities was lower, with polarization in both EWP stages. (3) CCUA achieved high economic growth at the cost of low EWP, while carbon emissions exerted positive impacts and positive spillover effects on EWP.

Introduction

With the rapid development of China’s economy and accelerating urbanization, urban agglomerations, as the core carriers of economic development, have driven severe ecological degradation and complicated environmental governance. In particular, greenhouse gas emissions are particularly prominent, threatening ecosystem health and stability and adversely affecting residents’ quality of life and well-being^1,2. China has proactively introduced two objectives: achieving peak carbon emissions and attaining carbon neutrality, and urban agglomerations, which are characterized by high concentrations of carbon emissions, have been entrusted with the important task of leading the realization of carbon peaking, which contains huge pressure to reduce emissions³. Under the dual carbon targets, balancing economic development with ecological protection, enhancing human well-being, and finding innovative ways to address limited natural resources amid urban expansion have become focal points of global concern⁴. As the only national-level urban agglomeration in western China, the CCUA’s sustainable development is of particular importance. On the one hand, high-carbon industries such as iron and steel, and chemicals serve as the core pillars of the region’s economy and employment, directly influencing the supply of people’s livelihood welfare. On the other hand, ecological pressure caused by high carbon emissions restricts the transformation of ecological resources into welfare. Meanwhile, the dual-core driven model has led to the special flow and distribution of resources, population, and industries within the region, providing a typical scenario for studying the contradictions between cross-administrative collaborative development and carbon emission reduction. How to safeguard the ecological security of the upper Yangtze River and enhance residents’ overall well-being while advancing high-quality development stands as a core challenge confronting the CCUA. Therefore, rigorously examining the coordinated development and spatial interactions among the CCUA’s ecological, economic, and well-being systems can help identify the root causes and evolutionary patterns of regional imbalance and provide a western China case study for addressing green, low-carbon development challenges in megaregions.

EWP refers to the effectiveness with which natural capital is converted into well-being of people’s livelihood, and it can systematically evaluate the dynamic coupling relationship between sustainable development capacity, ecosystem services, and human well-being⁵. The concept was initially introduced by Daly⁶ in 1974 to evaluate the sustainable development levels of countries, but it failed to gain widespread application due to the difficulty in quantification. It was not until Rees⁷ put forward the theory of ecological footprint in 1992 that research on EWP gradually expanded. Research related to EWP mainly falls into three categories. The first category focuses on the quantification of EWP. Initially, scholars utilized the ratio-based method to assess the EWP at the national scale, and the particular formula is defined as the ratio of welfare level to ecological consumption^8,9,10. The welfare level is mostly measured by the Human Development Index is often adopted¹¹, or by comprehensive indicators including subjective welfare and objective welfare. Regarding ecological input, ecological footprint is typically regarded as the most commonly utilized metric^12,13 and Xu et al.¹⁴ adopt comprehensive indicators that cover resource utilization and environmental pollution. However, this method can only reflect the simple relationship between two dimensions and cannot cover the non-linear correlation between resource utilization efficiency and welfare distribution. In order to avoid the inherent limitations, researchers constructed a multivariate input-output EWP system for the assessment of EWP. Xiao and Xiao¹⁵ used the stochastic frontier analysis method to evaluate EWP, but this approach is constrained by the uniqueness of expected outputs. Bian et al.¹⁶ applied the Super-Efficiency Slacks-Based Measure model to measure the EWP of 278 cities in China. Long et al.¹⁷ used a modified Data Envelopment Analysis (DEA) model to study 35 large and medium-sized cities in China. Although these measurement methods can effectively further rank cities with high efficiency, the EWP process is regarded as a black box, which lacks the dynamic decomposition of multi-stage efficiency. Therefore, it is difficult for these methods to identify the critical weak links that hinder improvements in EWP.

The second category is research on spatial distribution and spatial correlation, which mainly explores the distribution characteristics of EWP and spatial correlation across regions from the perspective of different scale scopes. Zhang et al.¹³ constructed an Index of Ecological Well-being Performance using the ratio method and calculated the EWP of 82 countries in 2012, revealing that the EWP values for developed countries and G20 countries were comparatively modest. Long¹⁸ conducted an international comparative analysis of EWP utilizing cross-section data from 42 countries. Liu et al.¹⁹ examined the dynamic evolution of EWP in Guangdong Province, China. Lan et al.²⁰ used spatial correlation analysis to carefully study spatiotemporal dynamics characteristics of EWP in seven urban agglomerations near China’s Yellow River Basin. Additionally, Fang and Xiao²¹ utilized the Moran index, and Feng et al.²² applied Local Indicators of Spatial Association, all of which indicated a significant spatial dependence of provincial EWP in China.

The third category concerns the influencing factors of EWP. Bergougui and Satrovic²³ calculated the ecological efficiency and environmental governance efficiency of nine Organization for Economic Co-operation and Development countries, and revealed the connections between excessive resource consumption, environmental governance, and ecological efficiency. Zang et al.²⁴ assessed the EWP of 288 cities in China, concluding that administrative level and city size exerted a notable positive impact on EWP. Furthermore, Behjat and Tarazkar²⁵ analyzed the determinants affecting Iran’s EWP from 1994 to 2014, finding that energy consumption negatively impacted EWP, while a positive correlation existed between economic growth and EWP. Li et al.²⁶ employed the spatial error model to discuss various factors at the inter-provincial scale in China, including technological advancement, greening efforts, social expenditure, medical grade, urbanization, industrial structure, and environmental regulation. Yasmeen et al.²⁷ estimated the ecological efficiency of 30 regions in China using a super-efficient DEA model and found that technological innovation has a significant positive impact on ecological efficiency. Existing studies tend to conduct simple horizontal comparisons at the urban agglomeration or regional scale, whereas research that analyzes the causes of the spatiotemporal differences and dynamic evolution trends of EWP from a more refined micro-scale remains insufficient.

The “dual carbon” goals have been elevated to a major national strategy, whose core essence is highly aligned with high-quality economic development and serves as a crucial pillar for social sustainable development. In studies investigating the welfare implications of carbon emissions, most empirical evidence suggests that higher carbon emissions are associated with improved human well-being^28,29. Compared with other G20 economies, China exhibits lower carbon emission welfare performance (CEWP) with sluggish growth, and there exists an inverted U-shaped relationship between government size and CEWP³⁰. However, there is a dearth of research on the internal nexus between carbon emissions and EWP. The economically developed urban agglomerations in eastern China mostly focus on paths like industrial upgrading, technological progress, and spatial coordination to improve the EWP^31,32. They have not addressed the conflicts between “carbon emission reduction” and welfare improvement, nor have they discussed the performance optimization paths under the low-carbon goal. Yang et al.³³ regarded the EWP of the Chengdu-Chongqing Economic Circle as a single efficiency value, failing to reveal its internal transformation mechanism and clarify the distribution pattern of EWP, thus providing relatively limited spatial guidance for regionally differentiated policies. Overall, research focusing on the optimization mechanisms and pathways of regional EWP against the backdrop of achieving the “dual carbon” goals remains limited, and this research gap will precisely hinder the scientific exploration of its sustainable development pathways.

To address limitations in current research, this study proposes a theoretical framework, “EWP under dual-carbon goals,” grounded in the strong sustainability paradigm. The framework integrates carbon-emission reduction with welfare enhancement and analyzes EWP through two components: ecological economic efficiency and economic welfare efficiency. This paper employs a Super-NSBM model that accounts for undesirable outputs to elucidate the conversion processes between ecological consumption and livelihood welfare. The dynamic evolution of EWP in the CCUA is revealed, and gaps in understanding its internal structure are clarified. Temporal changes in EWP are characterized using kernel density estimation. Moreover, a panel Tobit model and the Spatial Durbin Model (SDM) are used to quantify the impacts of carbon emissions on EWP and to identify spatial spillover effects. This framework embeds carbon constraints into the welfare-evaluation system—rather than focusing solely on emission totals—to promote simultaneous low-carbon growth and welfare enhancement (Fig. 1).

Fig. 1

Technology roadmap. This figure presents the overall analytical framework of the research, covering three core modules: indicator selection, analytical models, and result analysis. In the indicator selection module, ecological input (water, land, energy consumption), intermediate variable (GDP), undesirable output (pollution), and welfare output (green environment, medical and health care, education level) are included to construct the EWP evaluation system. The analytical model’s module adopts the Super-NSBM model (implemented in MaxDEA) to calculate EWP, and uses kernel density estimation, panel Tobit model, and Spatial Durbin Model for temporal evolution, spatial distribution, and spillover effect analysis. The result analysis module focuses on EWP’s time variation, spatial pattern, and association with carbon emissions, providing a systematic technical path for the study.

Materials and methods

Study areas

The specific geographical scope of the CCUA includes 27 districts (or counties) within Chongqing Municipality, in addition to sections of Kaixian and Yunyang, as well as 15 cities in Sichuan Province, while excluding specific districts and counties in Dazhou, Yaan, and Mianyang. The total area covered by this region is approximately 185,000 square kilometers³⁴. Considering the differences in statistical yearbook statistics and research scales, this study designates the entire Chongqing Municipality and 15 cities in Sichuan Province as the research area (Fig. 2).

Fig. 2

Location of study area. Left map: Standard Map Service System (http://bzdt.ch.mnr.gov.cn), Review No. GS(2019)1686. Elevation data: ASTER GDEM V3. This map shows the geographical scope of the research region, which includes the entire Chongqing Municipality and 15 cities in Sichuan Province (excluding specific districts/counties in Dazhou, Yaan, and Mianyang), with a total area of approximately 185,000 square kilometers. The figure marks the administrative boundaries of CCUA, its neighboring provinces, and key latitude/longitude lines, providing a clear spatial reference for understanding the study area’s location and scope.

Indicator selection and data sources

Indicator selection

EWP is generally an efficiency value of achieving overall social welfare under conditions of limited resource consumption and ecological pressures. The primary aim is to assess how to maximize social welfare while minimizing ecological costs, emphasizing the dynamic equilibrium between resource utilization and human well-being. Given that pollutant emissions resulting from urban economic development adversely affect residents’ well-being³⁵, this research integrates environmental pollution factors into the comprehensive evaluation system for EWP. This paper draws upon existing scholarly research^36,37,38 on the EWP indicator system and adheres to principles of scientific rigor, data comparability, and representativeness. Guided by the theory of strong sustainability and the idea of system coupling, it constructs an evaluation index system including four dimensions: ecological input, economic level, undesirable output, and welfare output (Table 1). The development of cities inevitably consumes natural resources, with such consumption primarily manifested in three aspects: electric energy, land resources, and water resources. Electricity serves as the core energy form for urban production and daily life, and energy consumption data can comprehensively reflect the energy demands across multiple sectors, including industry and residential life. Urban built-up areas act as the core carrier of land resource transformation during urban development, which can directly indicate the actual scale of land resource occupation. Water resources are the fundamental guarantee for both ecosystem functioning and urban operation, and per capita water consumption directly reflects the efficiency of urban water resource utilization. Therefore, per capita electricity consumption, per capita built-up area and per capita water consumption are selected to characterize the input of energy, land and water resources in the process of urban development. Economic performance is employed as an intermediate indicator, which is measured by GDP. Environmental pollution is evaluated through metrics such as industrial wastewater discharge, emissions of exhaust gases, solid waste generation, and the volume of domestic waste removal.

Human well-being can be categorized into subjective well-being and objective well-being^39,40,41, and subjective well-being focuses on individuals’ cognitive appraisals and emotional experiences of their surrounding social environment. which is highly subjective⁴². This paper uses objective well-being to measure the overall social welfare level, and selects the number of school students, average life expectancy, and per capita green space area to represent the level of educational development, healthcare, and environmental quality, as these factors can directly reflect the welfare of individuals and hold significant practical implications. Given that cities do not serve as a statistical unit for measuring life expectancy, the number of hospital beds per 10,000 individuals is utilized as a proxy for average life expectancy^43,44. Finally, this paper analyzes the sensitivity of the indicators representing the average life expectancy, and finds that the difference rate between the number of beds per 10,000 health institutions and the number of health technicians per 10,000 people on the core variables EWP, ecological economic efficiency, and economic welfare efficiency is 5.32%, 3.13%, and − 1.55%, respectively.

Table 1 EWP evaluation index system.

Data sources

Taking into account the integrity and representativeness of the data, the study period is defined as spanning from 2012 to 2022. This timeframe encompasses China’s 12th and 13th Five-Year Plan periods and coincides with the development phase of the Chengdu-Chongqing Economic Circle, thus ensuring high representativeness and comparability for the research. The administrative vector data utilized in this research were sourced from the Chinese Academy of Sciences. The Socioeconomic data were primarily derived from the Statistical Yearbook of Urban Construction in China, the Statistical Yearbooks of the respective years for the 16 cities, the bulletin of water resources, the bulletin of the state of the ecological environment, and the information on the prevention and control of environmental pollution by solid waste. Carbon emission data were acquired from the Emissions Database for Global Atmospheric Research (EDGAR). Among them, the EDGAR dataset revises China’s coal emission factor from 2.35 tons of CO₂ per ton of coal to 1.45 tons of CO₂ per ton of coal to enhance the dataset’s accuracy and align it more closely with the actual combustion efficiency of China’s low-quality coals⁴⁵. Additionally, to mitigate potential biases in the evaluation results stemming from the large total values of each index, per capita consumption metrics for each index are taken. The collection of the above data was completed as of February 5, 2025. The specific data sources are shown in Table 2.

Table 2 Data sources.

Methods

Super-NSBM model

Traditional DEA models measure efficiency from a radial perspective, yet this approach fails to account for slacks, which tends to result in overestimated efficiency values⁴⁶. To address this issue, Tone and Tsutsui⁴⁷ developed a slack-based network DEA model that enables the evaluation of the overall efficiency of decision-making units (DMUs) while simultaneously assessing the efficiency of their sub-stages. This study employs the MaxDEA v.9.1 software, and adopts a non-directed, non-radial Super-NSBM model considering undesirable output, based on a two-stage perspective under the assumption of variable scale remuneration (equation solver Gurobi 11.0.3, orientation: Nonoriented, Returns to Scale: variable, Extended Options: SuperEfficiency-Undesirable, the remaining options are default settings). Consider a scenario involving n DMUs, with each DMU consisting of P stages, m_p represents the number of input indicators at stage p, while v_ep and v_up denote the quantities of expected and undesired outputs at the same stage. The variable $\:{x}_{i}^{p}$ signifies the value of the input variable at stage p, y_ep represents the original inputs of the expected outputs at stage p, and y_up represents the original inputs of the unexpected outputs at stage p. Additionally, s_p, s_ep, and s_up represent the relaxation of the input variables, expected output and undesired output in stage P, respectively. λ_p is defined as the model weight assigned to the p-th stage, while ω_p serves to represent the corresponding weight parameter for the same p-th stage. Consequently, the overall efficiency formula for a DMU is articulated as follows:

$$\begin{gathered} \rho _{0}^{{}} = \min \frac{{\sum\nolimits_{{p = 1}}^{P} {w^{p} } \left[ {1 + \frac{1}{{m_{p} }}\left( {\sum\nolimits_{{i = 1}}^{{m_{i} }} {\frac{{s_{i}^{p} }}{{x_{{i0}}^{p} }}} } \right)} \right]}}{{\sum\nolimits_{{p = 1}}^{P} {w^{p} } \left[ {1 - \frac{1}{{v_{{ep}} + v_{{up}} }}\left( {\sum\nolimits_{{r = 1}}^{{V_{{ep}} }} {\frac{{s_{r}^{{ep}} }}{{y_{{r0}}^{{ep}} }}} + \sum\nolimits_{{r = 1}}^{{V_{{up}} }} {\frac{{s_{r}^{{up}} }}{{y_{{r0}}^{{up}} }}} } \right)} \right]}} \hfill \\ \left\{ {\begin{array}{*{20}l} {x_{{i0}}^{p} \ge \sum\limits_{{j = 1,j \ne 0}}^{n} {x_{{ij}}^{p} } \lambda _{j}^{p} + s_{i}^{p} } \hfill \\ {y_{{r0}}^{{ep}} \ge \sum\limits_{{j = 1,j \ne 0}}^{n} {y_{{rj}}^{{ep}} } \lambda _{j}^{p} + s^{{ep}} } \hfill \\ {y_{{r0}}^{{up}} \ge \sum\limits_{{j = 1,j \ne 0}}^{n} {y_{{rj}}^{{up}} } \lambda _{j}^{p} - s^{{up}} } \hfill \\ {\varepsilon \le 1 - \frac{1}{{v_{{ep}} + v_{{up}} }}\left( {\sum\limits_{{r = 1}}^{{V_{{ep}} }} {\frac{{x_{r}^{{ep}} }}{{y_{{r0}}^{{ep}} }}} + \sum\limits_{{r = 1}}^{{V_{{up}} }} {\frac{{s_{r}^{{up}} }}{{y_{{r0}}^{{up}} }}} } \right)} \hfill \\ {z^{{(ph)}} \lambda ^{h} = z^{{(ph)}} \lambda ^{p} ,\sum\limits_{{j = 1,j \ne 0}}^{N} {\lambda _{j}^{p} } = \sum\limits_{{p = 1}}^{P} {w^{p} } = 1} \hfill \\ {\lambda ^{p} \ge 0,s^{p} \ge 0,s^{{ep}} \ge 0,s^{{up}} \ge 0,w^{p} \ge 0} \hfill \\ \end{array} } \right. \hfill \\ \end{gathered}$$

(1)

In Eq. (1), $\:{\rho\:}_{0}$ represents the ecological welfare performance of DMU₀, s is determined by the number of desired or undesired outputs in each stage. The Super-NSBM model with a two-stage perspective is used, so P = 2. As consecutive critical links in realizing value transformation, ecological economic efficiency and economic welfare efficiency jointly determine the level of ecological welfare, making the two stages equally important⁴⁸. Guided by the core principle of “synergy between ecological conservation and people’s livelihood improvement,” this study adopts a setting with equal stage weights, which not only simplifies the model solution process but also ensures the objectivity and reproducibility of efficiency measurement. Equations (1) and (2) illustrate the computational method for determining the efficiency at each stage.

$$\rho _{{01}}^{{}}=\frac{{1+\frac{1}{{{m_1}}}\left( {\sum\nolimits_{{i=1}}^{{{m_1}}} {\frac{{s_{i}^{{1 - *}}}}{{x_{{io}}^{k}}}} } \right)}}{{1 - \frac{1}{\eta }\left( {\sum\nolimits_{{r=1}}^{\eta } {\frac{{s_{r}^{{1+*}}}}{{{z_{ro}}}}} } \right)}}$$

(2)

$$\rho _{{o2}}^{{}}=\frac{{1+\frac{1}{\eta }\left( {\sum\nolimits_{{r=1}}^{\eta } {\frac{{s_{r}^{{1*}}}}{{{z_{ro}}}}} } \right)}}{{1 - \frac{1}{{{v_{e2}}+{v_{u2}}}}\left( {\sum\nolimits_{{r=1}}^{{{v_{e2}}}} {\frac{{s_{r}^{{e*}}}}{{y_{{ro}}^{{e*}}}}} +\sum\nolimits_{{r=1}}^{{{v_{u2}}}} {\frac{{s_{r}^{{u*}}}}{{y_{{ro}}^{{u*}}}}} } \right)}}$$

(3)

In Eqs. (1) and (2), $\:{\rho\:}_{01}$ is the ecological economic efficiency of DMU₀, and $\:{\rho\:}_{o2}$ is the economic welfare efficiency of DMU₀. $\:{s}_{i}^{1-\text{*}}$ is the slack variable for the input indicator in stage 1, $\:{s}_{r}^{1+\text{*}}$ is the slack variable for the output indicator in stage 1, and z is the value of the intermediate variable. $\:{s}_{r}^{e\text{*}}$ is the slack variable for the desired output in stage 2, $\:{s}_{r}^{u\text{*}}$ is the slack variable for the undesired output in stage 2, $\:{v}_{e2}$ is the number of desired output indicators in stage 2, and $\:{v}_{u2}$ is the number of undesired output indicators in stage 2. $\:\eta\:$ is the number of intermediate indicators.

By reading the literature⁴⁹, it has been determined that the Super-NSBM model should not include an excessive number of input indicators, as this can negatively impact the precision of the measurement outcomes. The multicollinearity test indicates that the Variance Inflation Factor values of industrial wastewater discharge, sulfur dioxide emissions, general industrial solid waste generation, and municipal solid waste removal volume are 1.42, 1.67, 1.30, and 1.46, respectively. This result demonstrates that there is no significant multicollinearity among these four undesirable output indicators. Therefore, this paper uses the entropy weight method to reduce the dimension of the undesirable output indicators that represent environmental pollution. The weights of the four factors are 0.144, 0.445, 0.094 and 0.317, respectively. The entropy weight method effectively avoids interference from subjective judgments, making it widely applicable in scenarios involving multi-index evaluation and dimension integration. Notably, a critical limitation arises when strong correlations exist between indicators: the overlapping information among indicators can lead to repeated calculations in weight assignment, which in turn amplifies the distorting impact of redundant information on the final evaluation results⁵⁰. The weights are entirely data-driven, which makes it difficult for this method to be applied to factors that rely on subjective evaluation.

Kernel density estimation

Kernel density estimation is a statistical technique that does not rely on parametric assumptions and is mainly used to calculate the probability density function of continuous variables⁵¹. The core idea is to smoothly estimate the distribution of data across space by weighting the neighborhoods around data points. Compared with the traditional parametric method, kernel density estimation does not necessitate presuppositions regarding the distributional characteristics of the data, and can capture the true distribution characteristics of the data more flexibly, especially for complex and irregular data distribution. This study employs kernel density analysis with a Gaussian kernel function to explore the dynamic evolution patterns of EWP in the CCUA. Let x₁, x₂, ⋯, x_i represent the samples drawn from the overall population, and the formula is as follows:

$$f(x)=\frac{1}{{nh}}\sum\limits_{{i=1}}^{n} K \left( {\frac{{{x_i} - \bar {x}}}{h}} \right)$$

(4)

$$K(x)=\frac{1}{{\sqrt {2\pi } }}\exp \left( { - \frac{{{x^2}}}{2}} \right)$$

(5)

In Eqs. (4) and (5), f(x) represents the kernel function, n denote the total number of samples involved in the evaluation, $\:\stackrel{̄}{x}$ represents the mean, k(x) is the Gaussian kernel function, and the variable h represents the bandwidth, which is also known as the smoothing parameter.

Panel Tobit model and spatial Durbin model

Tobit panel regression analysis represents a statistical approach that combines the features of the Tobit model with panel data, specifically designed to address panel data with truncated or censored dependent variables, which is especially suitable for studying the variation of restricted dependent variables in time and individual dimensions⁵². Its core advantage lies in being closer to the characteristics of real data and more accurate identification of policy effects. In this study, the Tobit model of the constrained dependent variables is used to estimate parameters to study the connection between the EWP and carbon emission. The panel Tobit random-effects econometric model is established, and the basic equation form is as follows:

$$EW{P_{itp}}={\beta _0}+{\beta _1}{X_{1it}}+{\beta _2}{X_{2it}}+ \cdots +{\beta _j}{X_{jit}}+{\varepsilon _{it}}$$

(6)

In Eq. (6), EWP_itp represents the EWP value of certain city in the CCUA in years t and stage p, X represents explanatory variable, β₀ represents the intercept term, β_j represents the coefficient of X, and ε_it is a stochastic error term.

Starting from the core logic of the First Law of Geography, different regions do not exist in isolation but are always in a state of dynamic connection characterized by mutual association and interdependence⁵³. In terms of the analysis of spatial spillover effects, the commonly used spatial econometric models in academia mainly include the Spatial Autoregressive Model (SAR), Spatial Error Model (SEM), and SDM.

$${\text{SAR}}: \rho \delta y = \lambda Wy + \varepsilon$$

(7)

$${\text{SEM}}: \begin{gathered} y = X\beta + u \hfill \\ u = \rho Mu + \varepsilon \hfill \\ \end{gathered}$$

(8)

$${\text{SDM}}:y = \lambda Wy + X\beta + WX\delta + \varepsilon$$

(9)

In the formula, y denotes the explained variable, W represents the Rook first-order contiguity spatial weight matrix, X stands for explanatory variables and control variables, λ is the spatial autoregressive coefficient which measures the impact of the spatial lag term Wy on y, both ρ and δ are coefficients, and ε indicates the random error term.

Results

Spatiotemporal evolution of the EWP in the CCUA

Temporal variation of characteristics and its sub-stages

In this study, the Super-NSBM model was employed to assess the EWP, ecological economic efficiency, and economic welfare efficiency of the CCUA (Fig. 3). From 2012 to 2022, the ecological welfare level of the CCUA was found to be suboptimal, with the value fluctuating around 0.629, a geometric mean of 0.632, and a 1.22% yearly growth rate on average. Overall, the EWP demonstrated a generally upward trend. Additionally, by comparing the EWP values derived from other DEA models, the Wilcoxon signed-rank test revealed significant disparities between the two methods (Table 3). While the Slack-Based Measure model (SBM) demonstrated superior statistical performance in statistical tests, its efficiency failed to effectively distinguish between high-welfare scenarios achieved through high ecological costs and those attained via low ecological costs, particularly in regions with unbalanced development such as the CCUA. In contrast, the results of the Super-NSBM model more precisely capture the core contradictions inherent in EWP. Therefore, the two-stage model that accounts for undesirable outputs aligns more closely with the actual conditions of the CCUA than the traditional SBM model in terms of measurement outcomes.

Fig. 3

Temporal changes in EWP and decomposition stages. This line chart depicts the annual variation trends of three key efficiency indicators in CCUA over the study period: overall Ecological Welfare Performance (EWP), ecological economic efficiency (Stage 1), and economic welfare efficiency (Stage 2). The vertical axis represents efficiency values (ranging from 0.55 to 0.95), and the horizontal axis represents the year (2012–2022). The chart shows that EWP fluctuated around 0.629 with an overall upward trend; ecological economic efficiency (average 0.843) was consistently higher than economic welfare efficiency (average 0.773).

Table 3 Wilcoxon signed-rank test.

The geometric mean values for ecological economic efficiency and economic welfare efficiency were determined to be 0.843 and 0.773, respectively, with annual average growth rates of 2.86% and − 0.8%, indicating a slow growth trend. Notably, in 2016, the ecological economic efficiency peaked at 0.951, while the economic welfare efficiency reached its lowest point at 0.707. Consequently, the study interval was divided into two intervals from 2012 to 2017 and 2017 to 2022 for Mann-Kendall trend analysis. The findings (Table 4) reveal that there is a significant upward trend in ecological economic efficiency and a downward trend in economic welfare efficiency between 2012 and 2017. From 2017 to 2022, ecological economic efficiency and economic welfare efficiency showed a trend of fluctuation. From 2012 to 2022, the fluctuation directions of ecological economic efficiency and EWP are almost the same, indicating that the ecological economic efficiency stage is the key link to affect EWP. After 2014, the efficiency value of the first stage is significantly greater than that of the second stage, which indicates that the low efficiency of economic welfare is the main reason for the low level of overall EWP of CCUA. Improving the efficiency of economic conversion to welfare is an important aspect for the improvement of EWP in CCUA.

Table 4 Mann–Kendall trend analysis.

Based on the time series data from 2012 to 2022, the changes of EWP in the cities of the CCUA are presented through spatial visualization (Fig. 4). Figures (a) to (e) show the annual change rates of EWP at two-year intervals, while Figure (f) presents the average annual change rate of EWP in the CCUA. From 2012 to 2014, both Suining and Luzhou recorded growth rates exceeding 50%. The change rate of EWP in Meishan and Nanchong is as high as 144% and 135%, and Chongqing, Deyang, Guang’ an and Leshan decreases by less than 25%. Between 2014 and 2016, The four cities of Meishan, Luzhou, Nanchong and Suining did not maintain a high growth trend, and the EWP showed a negative growth. In the eastern region, only Chongqing achieved a positive growth rate. The change rate of EWP in urban agglomerations has declined, entering a relatively stable period. From 2016 to 2018, the growth area was concentrated in Chengdu and its northern cities. Yibin and Chongqing ‘s EWP level weakened and failed to maintain a high level of efficiency value. The EWP of Suining and Nanchong further deteriorated, with a rate of change greater than − 25%. At this stage, the development of urban agglomerations is differentiated, and some cities are facing ecological and welfare coordination problems. From 2018 to 2020, the area of EWP collapse appeared in the western and central parts of CCUA, and the decline rates of Chengdu and Mianyang were 49% and 48% respectively. From 2020 to 2022, the efficiency values of Chengdu and its surrounding cities improved, the collapse area of EWP formed a continuous area in the north and south, and the EWP of Chongqing continues to improve. Overall, the EWP of CCUA is characterized by periodic fluctuation and unbalanced development, reflects that the EWP of the CCUA is gradually shifting in a favorable direction amidst the interplay of economic construction and ecological environment governance. However, the improvement of ecological welfare performance is insufficient in terms of sustainability and coordination.

Fig. 4

Time change of EWP. This set of spatial variation maps (a–f) presents the EWP change rates of CCUA cities at five two-year intervals (2012–2014, 2014–2016, 2016–2018, 2018–2020, 2020–2022) and the average annual change rate (2012–2022). The figure uses a color gradient to represent change rate magnitudes (positive values for growth, negative values for decline).

Spatial distribution characteristics

In this study, 2012, 2014, 2016, 2018, 2020 and 2022 were selected as the time nodes, and the equal interval classification method is adopted to divide EWP into five grades: poor, low, medium, good, and high. (Fig. 5). Through a comparative analysis that the dynamic change process of EWP can be clearly observed. The spatial distribution of EWP in the CCUA revealed a pattern characterized by better levels in eastern cities, poorer performance in western cities. In 2012, the EWP showed an obvious east-west distribution, forming a high-value zone in Guang’an and Chongqing in the east and a low-value aggregation area in Meishan, Zigong and Leshan in the southwest. In 2014, The quantity of cities exhibiting poor EWP diminished, forming a large continuous area with efficiency in the central and western part of CCUA. In 2016, the EWP of CCUA exhibited a degree of stability, while the EWP in the Meishan and Leshan areas deteriorated by one grade. In 2018 and 2020, Chongqing continued to exhibit high EWP, positively influencing neighboring cities. This diffusion effect significantly enhanced the EWP levels in the eastern region, while the EWP of the low-value regions in the southwestern urban agglomerations has not been significantly improved. In 2022, the cities characterized by poor EWP did not expand, and Meishan’s EWP transitioned from poor to medium efficient. With the background of the advancements attributed to policy initiatives and ecological improvements, the EWP of western cities shows a certain improving trend, however, a disparity persists when compared to the eastern region. From a spatial perspective, the EWP in Chongqing and Guang’an has consistently exhibited elevated levels. Conversely, Neijiang, Zigong and Leshan have persistently been identified as poor efficiency cities or medium efficiency cities, with their EWP consistently below 0.4 over an extended period. This indicates significant issues, such as the irrational utilization of ecological resources and severe ecological damage. While the EWP of other regions has experienced fluctuations, it has shown an overall improvement trend in the later stage. In general, from 2012 to 2022, the EWP of CCUA formed relatively high-value area around Chongqing and Chengdu, while radiating outward to influence the formation of low-value zones in surrounding cities. The EWP of each city improved to a certain improvement throughout the study.

Fig. 5

Spatial distribution characteristics of EWP. This series of spatial distribution maps (a–f) classifies CCUA’s EWP into five grades (poor, low, medium, good, high) using the equal-interval method for six key years. The color gradient (from light to dark) corresponds to increasing EWP levels (0 to 1.5). The figure shows a consistent “east-high, west-low” spatial pattern: Chongqing and Guang ’an in the east maintained high EWP, while Meishan, Zigong, and Leshan in the southwest remained low-value areas. By 2022, western cities’ EWP showed a slight improvement but still lagged behind the east, with persistent regional disparities.

Trend analysis of time evolution

This section utilizes the Kernel density estimation method to conduct a systematic analysis of the absolute disparity in the EWP from a time series viewpoint. The dynamic evolution characteristics of these differences across two distinct stages are thoroughly explored. As shown in Fig. 6, during the stage of ecological economic conversion efficiency, each density curve is characterized by an initial increase followed by a slight decrease, with continuous improvement throughout the study period. This reflects the evolutionary process where ecological economic efficiency, after experiencing growth and adjustment, gradually develops into a new steady state. The kernel density values show that the peak fluctuation of the main peak of the curve rose to the maximum in 2017, and then decreased to a stable level. At the same time, the overall width of the curves displays a slight narrowing, indicating that the absolute differences of EWP in the ecological economic efficiency stage between cities has a tendency to decrease. In terms of the curve direction, the distribution curve from 2012 to 2018 all exhibit both a main peak and a secondary peak, indicating that there is a clustering phenomenon of ecological economic efficiency in low and high efficiency. After 2018, the double-peak feature vanishes, the curve becomes single-peaked, and the polarization phenomenon converges. The reason for this is that cities with better economic strength, such as Chengdu and Chongqing, have a leading advantage, while some cities have an unbalanced industrial structure due to the lack of ecological compensation mechanism, which greatly widens the gap with other cities. After the implementation of policies such as the Chengdu-Chongqing City Cluster Development Plan, the Red Line Ecological Protection System, the gap between high- and low-efficiency cities gradually narrowed and polarization disappeared, which strengthened the balanced development of ecological economic efficiency of each city.

Fig. 6

Kernel density curves of ecological economic efficiency. This kernel density plot illustrates the dynamic evolution of absolute differences in ecological economic efficiency (Stage 1 of EWP) among CCUA cities. The horizontal axis represents efficiency values, and the vertical axis represents kernel density (reflecting data concentration). The figure shows that density curves first rose then slightly fell, with the main peak reaching its maximum in 2017 and stabilizing afterward. Before 2018, curves exhibited double peaks (indicating low- and high-efficiency clustering); after 2018, double peaks disappeared and curves became single-peaked, meaning polarization in ecological economic efficiency converged due to regional coordinated policies.

The economic welfare conversion stage has a similar trend to the ecological economic conversion stage (Fig. 7). In term of time, the position of the main peak shows a trend of moving left and then right, and this trend show a significant consistency with the evolution trajectory of economic welfare efficiency. The kernel density values show that the height of the main peak showed a fluctuating downward trend, while the width of the curve expanded year by year from 2016 to 2022, revealing that the absolute disparity in economic welfare efficiency between cities has a slight expansion trend. In terms of the curve direction, there are double peaks in the kernel density curve every year, which indicates that the development level of economic welfare efficiency among cities is inconsistent, and high-efficiency aggregation is formed in the northeast of the CCUA, while a low-efficiency aggregation is formed in the Chengdu metropolitan area, and the polarization is extremely serious. Overall, most of the cities in the CCUA show a low level of aggregation at the economic welfare conversion stage, which reflects the problems of uneven regional development, lagging ecological and economic synergy mechanisms, and insufficient transformation capacity of small and medium-sized cities.

Fig. 7

Kernel density curves of economic welfare efficiency. This kernel density plot depicts the dynamic evolution of absolute differences in economic welfare efficiency (Stage 2 of EWP) among CCUA cities. The horizontal axis represents efficiency values, and the vertical axis represents kernel density. The figure shows that the main peak of curves shifted left then right, consistent with the overall trend of economic welfare efficiency. Curve width expanded annually after 2016, indicating a slight expansion of absolute differences among cities. Notably, all years showed double peaks, reflecting persistent polarization: high-efficiency aggregation in northeastern CCUA and low-efficiency aggregation in the Chengdu metropolitan area.

Analysis of EWP and per capita carbon emissions

Carbon emissions play a significant role as one of the major drivers behind global climate change⁵⁴, so how to achieve the improvement of EWP and the effective control of carbon emission have become the two core issues for urban agglomerations to realize sustainable development.

The traditional single evaluation model is difficult to precisely analyze the development contradictions. This study establishes an analytical framework that incorporates the dual dimensions of EWP and carbon emissions to investigate the principal pathways for promoting Low-carbon economic development and green transformation within urban agglomerations. The EWP of the CCUA generally varies between 0.552 and 0.678, but the disparity in development efficiency at each stage is evident because of the notable variations in resource endowment across the various cities. In this section, the 16 cities in the CCUA are classified into different clusters based on the numerical characteristics of their EWP and carbon emissions. Four types of urban groups are identified, namely low efficiency-low emission (L-L), low efficiency-high emission (L-H), high efficiency-low emission (H-L), and high efficiency-high emission (H-H), as shown in Fig. 8.

Fig. 8

Cluster Analysis of EWP and Carbon Emissions. This scatter plot classifies 16 CCUA cities into four types based on standardized EWP (horizontal axis) and per capita carbon emissions (vertical axis): low efficiency-low emission (L-L), low efficiency-high emission (L-H), high efficiency-low emission (H-L), and high efficiency-high emission (H-H). Each point represents a city, with different colors marking different clusters.

The two cities, Chongqing, Guang’an, belong to the H-H type, with stronger resource utilization efficiency or ecological service capacity compared to other cities, but also high per capita carbon emissions, which should prioritize accelerating the optimal restructuring of the energy mix and the innovative practice of low-carbon technologies to crack the high emissions bottleneck while maintaining the EWP advantage. Chengdu, Nanchong and Ziyang belong to H-L type, but their EWP is only slightly higher than average, indicating that they are transforming into green industries. Suining, Zigong, Leshan, Deyang, Yaan, Luzhou, Meishan and other cities belong to the L-L type, and these cities have a lot of room to improve their EWP and avoid the vicious cycle of “low development-low welfare” while maintaining low carbon. Dazhou, Neijiang, Mianyang and Yibin belong to L-H type, with low EWP and high per capita carbon emissions, which exposes the lack of industrial greening, and it is imperative to advance the green transformation of industries in order to mitigate emissions and enhance the value of EWP. It is noteworthy that the ideal development model of “high efficiency-low emissions” shows no urban clusters identified at the high level, reflecting that the CCUA has not yet realized the optimal balance between ecology and economy. This distribution not only highlights the pressure on urban agglomerations to control carbon emissions but also reveals the current situation of insufficient ecological welfare transformation efficiency in some cities.

Spatial panel regression analysis and spillover effect decomposition

The impact of dual carbon on EWP under the static panel model

To explore the relationship between EWP and carbon emissions, the total carbon emissions (ce) and carbon emission intensity (ci) of the CCUA are selected as explanatory variables. Meanwhile, to mitigate the impact of omitted variables on EWP, the following indicators are chosen as control variables. (1) Economic growth level (pgdp). The improvement of economic development level not only serves as a crucial guarantee for residents’ basic material life but also exerts a positive promoting effect on the expansion of medical resource supply and the enhancement of education service quality. It is represented by the per capita GDP of each city. (2) Industrial structure optimization (his). The process of industrialization has significantly driven the improvement of economic levels. However, traditional industrialization mostly takes heavy industries such as chemical engineering, energy, and iron and steel as its core pillars, and the pollution problems caused by these industries have restricted the sustainable growth of regional economies to a certain extent. The service-oriented economic structure is becoming an important symbol of the transformation and upgrading of China’s industrial structure, so it is represented by the ratio of the output value of the secondary industry to that of the tertiary industry. (3) Science and technology (sci). For enterprises reliant on research and development (R&D) investment and government-funded science and technology expenditures, advancements in science and technology serve to enhance their independent innovation capabilities and elevate production efficiency. However, for enterprises with low-level environmental pollution control, they lack responsiveness to science and technology, and the cost of technology investment will instead exert a crowding-out effect on them. This indicator is represented by government science and technology expenditure.

As the Hausman test fails to reject the null hypothesis that the coefficient differences are not systematic, therefore this section utilized panel regression analysis to methodically investigate the internal association mechanism between the EWP and carbon emissions within the CCUA, which provides valuable insights for regional high-quality development (Table 5). The results indicate that all factors, except for his and sci, have passed the significance test at the 1% confidence level. The rise in carbon emissions has a strongly driving impact on the improvement of EWP. The EWP is negatively correlated with per capita GDP and carbon emission intensity, which is consistent with the findings of existing relevant studies⁵⁵. This indicates that the CCUA has transitioned into the descending segment of the inverted U curve describing the relationship between EWP and economic growth during the timeframe of the study. From 2012 to 2022, the per capita GDP of the CCUA grew by 133%, while the growth rate of EWP was only 23%. The CCUA’s EWP is low, with economic growth as the focus of the CCUA, and the efficiency and economic growth is not coordinated, leading to a situation where EWP is reduced in exchange for economic growth.

Table 5 Tobit model regression results (n = 176).

The panel Tobit model performs its analysis based on the assumption that a unidirectional causal relationship exists between the explained variable and its respective driving factors. To mitigate the endogeneity issue between carbon emissions and ecological welfare performance, this paper employs the exogenous instrumental variable method for robustness testing. When undertaking industrial transfers, the CCUA prioritizes economic benefits far above environmental and ecological benefits, which has accelerated the pace of industrial relocation and exacerbated the local predicament of carbon lock-in⁵⁶. The higher the historical gross industrial production, the more entrenched a city’s high-carbon development path becomes. Meanwhile, this variable is exogenous to the current ecological welfare performance, satisfying the core requirements for an instrumental variable. Therefore, historical gross industrial production (IV) is selected as the instrumental variable for carbon emissions. In Table 5, the coefficient of the instrumental variable on carbon emissions shows a significant positive correlation. The Kleibergen-Paap (K-P) test statistic is substantially higher than the empirical threshold of 10, indicating the absence of a weak instrumental variable problem. The two-stage least squares (2SLS) estimation results demonstrate that carbon emissions exert a significant positive impact on ecological welfare performance, which verifies the robustness of the panel regression results.

The spatial effect of the dual carbon goal on EWP

Traditional static regression models, which neglect spatial factors, tend to cause deviations between research results and actual situations. Furthermore, in the measurement of EWP, they only focus on the influencing factors within the region itself, making it difficult to effectively identify the spatial spillover effects that neighboring regions exert on the local region—effects arising from ecological, economic, and welfare linkages between regions. Therefore, to further explore the spatial correlation of EWP in the CCUA, this section, based on spatial correlation analysis and relying on spatial econometric models, uses Stata 18.0 software to conduct modeling and measurement of the spatial spillover effects of EWP in the CCUA. Table 6 shows that the Moran’s I indices from 2012 to 2021 basically passed the significance test at the 5% confidence level, while there was an insignificant positive correlation in 2022. The Moran’s I indices were positive in all years, with an average value of 0.22, indicating that the EWP of various cities is not spatially random; instead, there is similarity between cities with adjacency relationships. Therefore, based on the static panel model, a spatial adjacency matrix is introduced to study the spatial effects of EWP in the CCUA.

Table 6 Moran’s I indices of EWP in the CCUA.

Regarding the selection of spatial econometric models, this paper sequentially conducted the Lagrange Multiplier Test (LM), Likelihood Ratio Test (LR), and Hausman test to select the optimal spatial econometric model for analyzing the spatial effects of EWP in the CCUA (Table 7). In the LM test, both the SEM and the SAR passed the significance test. The LR test rejected the null hypothesis, and the SDM will not degenerate into the SEM or SAR, indicating that the SDM can be initially applied. Meanwhile, the results of the Hausman test showed that the fixed effect was superior to the random effect. Therefore, this paper selected the dual-fixed-effect Spatial Durbin Model with a geographical adjacency matrix, and the model results are shown in Table 8.

Table 7 Spatial econometric model selection test.

As shown by the regression results of the SDM (Table 8), first, when controlling for other relevant variables to keep them constant, a one-unit increase in carbon emissions led to a 0.029 increase in EWP, indicating that total carbon emissions could improve EWP in the CCUA. Meanwhile, the spatial lag term (Wx) of the model could reflect the spatial relationship between the variable and the EWP of neighboring regions. The coefficient of the lag term for the core explanatory variable (ce) was 0.078, indicating that it had a positive spatial spillover effect. In other words, there was a positive correlation between carbon emissions in geographically adjacent regions and the ecological welfare level of the local region. Second, the coefficients of the direct effect and indirect effect of economic development level (pgdp) were − 0.892 and 0.917 respectively, revealing the mechanism that per capita GDP inhibited the EWP of the local region while improving the EWP of surrounding regions. Essentially, the region bears the ecological costs of economic growth and spills over the distributed benefits (technological, industrial, and governance dividends) to adjacent regions. This phenomenon is particularly common in regions with close spatial connections, which also reflects that the CCUA needs to establish cross-regional coordinated policies to improve EWP. Third, the indirect effect of carbon emission intensity failed to pass the significance test. The reason is that carbon emission intensity affects economic welfare efficiency and needs to go through a sequential process: carbon emission intensity influences technological innovation, which in turn affects production efficiency, and ultimately impacts economic welfare. However, this hierarchical transmission has a lag, resulting in an insignificant neighboring effect. Meanwhile, the reduction of carbon emission intensity is also accompanied by the offset between increased costs and environmental benefits, leading to an insignificant net effect in the short term. The regression coefficients of industrial structure optimization and science and technology are not significant, indicating that the versatility of both is fully reflected in the process of converting ecology into welfare.

Table 8 SDM regression results.

Discussion

To address the root causes of low EWP efficiency and to identify bottlenecks in the CCUA, the study decomposed EWP into two sub-stages. This approach overcomes the limitations of conventional single-stage evaluations and provides a micro-level framework to elucidate the structural contradictions within the Chengdu-Chongqing region, characterized by abundant ecological resources yet uneven welfare distribution. The findings indicate that the EWP value remains relatively low, with an average annual growth rate of 1.2%. Notably, the EWP declined in the years 2016, 2018, and 2020 compared to the preceding years. This conclusion aligns with Yang’s research⁵⁷, which suggested a slight overall growth in EWP for the CCUA with minor annual fluctuations. Similarly, Cui et al.⁵⁸ identified a relatively low EWP level in the CCUA, highlighting a significant gap compared to urban agglomerations in eastern China. These findings corroborate the results of the present study, validating the research methodologies employed. Additionally, a sensitivity analysis of the evaluation indicators was conducted, revealing that the variance in the indicators “number of hospital beds” and “number of health technicians” relative to EWP, ecological economic efficiency, and economic welfare efficiency was 5.32%, 3.13%, and − 1.55% respectively. This outcome reaffirms the scientific validity of the EWP evaluation indicators utilized in this study.

To explore the spatial spillover effects of carbon emissions and other factors on EWP within the CCUA, this study employs the panel Tobit model and spatial econometric analysis methods. This method addresses the limitation of previous studies that often overlooked the spatial interdependence and spillover effects between regions. The results from Tobit panel and SDM regression further reveal that the CCUA has decreased its dependence on traditional industries, advancing the development of green industries. Among existing literature, Zang et al. ²⁴ examined the impacts of administrative hierarchy and population size on EWP from governmental and market perspectives, concluding that “high carbon emissions and high economic levels correlate with high EWP,” a finding that resonates with the analysis presented in this study. From the perspective of new urbanization, Yang et al.³³ noted a shift in urban development focus towards the efficient utilization of ecological resources, which enhances the comprehensive welfare of residents, thereby improving EWP. Zhang et al.⁵⁹ linked low EWP levels to high carbon emissions in China, identifying high carbon emission intensity as a key factor contributing to the low EWP. Furthermore, while Yang et al.⁶⁰ analyzed the effects of carbon emissions trading policies on EWP, previous studies did not profoundly investigate the relationship between carbon emission intensity and EWP. This study comprehensively assesses the impact of carbon emissions on the transition from ecological wellness to social welfare, breaking new ground beyond the traditional environmental Kuznets curve. It establishes a multivariate analytical framework that integrates constraints related to carbon emissions, ecological consumption, economic factors, and welfare distribution, thereby expanding the investigative scope of EWP and providing a theoretical foundation for achieving decoupling between carbon emission reduction and EWP. By integrating the two-stage Super-NSBM model with regression methodologies, this research not only fills the existing research gap concerning the internal mechanisms of EWP and its correlation with carbon emissions in the CCUA but also methodologically advances EWP research from result evaluation to process optimization.

Conclusion and policy suggestions

Conclusions

This study constructed an evaluation system for EWP, measured the integral and multistage efficiency value of the CCUA from 2012 to 2022. Meanwhile, the Tobit model and SDM were employed to explore the impact of carbon emissions on EWP. The study finds that the EWP of the CCUA exhibits an evolutionary trend of rising first, then declining, and finally increasing slightly, and improving the efficiency of economic welfare efficiency a key link in the process of the growth of EWP. In terms of spatial distribution, the EWP is high in the east and low in the west, featuring severe east-west polarization and considerable disparities among cities. The polarization trend of each city in the ecological economic transformation stage has weakened, while the spatial heterogeneity in the economic welfare transformation stage has gradually increased. There is a significant positive correlation between the carbon emission process and EWP in the CCUA. Both GDP and carbon emission intensity exert an inhibitory effect on EWP, indicating that the growth-maximization-oriented economic development model fails to yield high levels of EWP. Decomposition results of the SDM show that per capita GDP generates a positive spillover effect on the EWP of neighboring cities. Carbon emissions not only exert a positive promoting effect on local EWP but also produce a positive spatial spillover effect across regions.

Policy suggestions

Based on the aforementioned research, four key measures are proposed to enhance the EWP of the CCUA and accelerate the achievement of the “dual-carbon” goals. First, focusing on the shortcomings in economic welfare efficiency, the assessment system centered on people’s livelihood indicators such as education and healthcare should be reconstructed to promote the efficient whole chain transformation of ecological welfare performance. Second, a cross-regional ecological compensation and benefit-sharing mechanism should be established. Through the transfer of capital and technology as well as the sharing of public services from Chengdu and Chongqing to surrounding cities, regional gaps can be narrowed. Meanwhile, differentiated carbon emission control should be implemented to advance industrial green upgrading and foster welfare-oriented green industries. Finally, efforts should be made to strengthen cross-regional low-carbon collaborative development, unify standards and share resources, and establish a collaborative governance alliance among geographically adjacent cities to address regional challenges and guide cities to exert positive radiating effects.

Type H-H cities should prioritize carbon emission reduction efforts, implement the R&D of carbon capture, utilization and storage technologies, incorporate iron and steel as well as chemical enterprises into pilot projects to strengthen emission reduction constraints. Chongqing and Guang ’an may integrate the carbon emission rights trading market with the electricity market and green certificate trading to develop green manufacturing clusters. They should give priority to the development of urban digital trade. The growth of digital trade can not only promote the development of the green industry, but also reduce the difficulty of technology spillover, thereby maximizing regional emission reduction benefits. For H-L type cities, digital technology should be applied to optimize public services, extend characteristic low-carbon industrial chains, and promote cross-regional mutual recognition of the carbon-inclusive mechanism. Centered on the transformation of ecological values and cross-regional ecological compensation, L-L type cities should develop eco-tourism, carbon sink trading, and establish a mechanism for realizing the value of ecological products to unlock ecological potential. Tourism-oriented cities such as Leshan and Yaan should build zero-carbon scenic spots. Suining and Meishan should undertake green industries from Chongqing to develop the carbon sink economy. L-H type cities should strengthen the administration of pollution taxes, expand the tax scope to cover more pollutants, establish a reward and punishment mechanism linked to tax payments, and encourage high-emission industries to pursue low-carbon transition. Yibin should incorporate liquor-making enterprises into the Sichuan-Chongqing carbon trading pilot, and Mianyang should promote carbon footprint management for high-energy-consuming equipment manufacturing enterprises. These enterprises can obtain quotas through carbon sink purchasing and technological emission reduction, thereby facilitating the transformation of industries towards green and high-end equipment manufacturing.

Limitations and future study

The principal contribution of the study lies in combining the dual‑carbon goals with a two‑stage efficiency‑decomposition framework. This method is particularly well suited to regions with abundant ecological endowments but uneven welfare distributions, as it isolates underperforming sub‑stages and thereby pinpoints targeted pathways for improvement. However, at the large spatial scale, municipal‑level data on subjective well‑being and life satisfaction remain scarce. Subsequent research that integrates subjective and objective assessments of human well‑being would likely improve the accuracy and robustness of EWP measurements.