Introduction
In the context of global climate change and resource and environmental constraints, the pursuit of economic growth while ensuring effective environment protection has emerged as a pivotal concern for numerous nations worldwide. To address these challenges, the Chinese government has put forward a development concept of ‘lucid waters and lush mountains are invaluable assets’, alongside strategic objectives aimed at achieving carbon peaking (the turning point of maximum carbon emissions) and carbon neutrality (net-zero carbon emissions through emission reduction and offsetting efforts; Ministry of Ecology and Environment, PRC 2019). Meanwhile, the emergence of the digital economy (DE) has engendered novel prospects for China’s economic transformation and green development. This assertion is substantiated by the 2023 Global Digital Economy White Paper, a publication by the China Academy of Information and Communications Technology (2023). The document asserts that in 2022, the scale of China’s DE attained USD 7.5 trillion, representing a year-on-year growth rate of over 10%. The DE accounted for 30–45% of gross domestic product (GDP), further solidifying its position and enhancing its supporting role in the national economy. The 14th Five-Year Plan for National Informatization of the Chinese government explicitly states: ‘Lead green development with digitalization and promote digitalization through green initiatives’; and: ‘Vigorously develop new technologies and industrial systems for the convergence of digital and green to create new momentum for high-quality development’ (Cyberspace Administration of China 2021). The coordinated development of the DE and environmental quality (EQ) is crucial for advancing China’s economy and for it to remain current. Exploring the relationship between the DE and EQ has important theoretical and practical value for achieving an ecological civilization and maintaining high-quality development.
The protection of the environment and sustainable development have long been focal points for governments and academia worldwide. The ‘ecological environment’ refers to a dynamic system comprising living organisms, their physical environment and the complex interactions among them, which collectively influence human survival and development. Some strands of EQ research place particular emphasis on assessing EQ particularly in the context of human impacts and sustainable development. The pressure–state–response (PSR) model is frequently employed when measuring regional environment conditions (Jiao et al. Reference Jiao, Wang, Lu, Fan, Zhang and Wu2023, Zhang et al. Reference Zhang, Zhou and Yin2024a). As environmental challenges continue to evolve, identifying new development models has become crucial for advancing the concept of an ‘ecological civilization’. As an emerging economic paradigm, the DE is increasingly driving China’s sustained growth and economic development, leading to a surge in related research. Some scholars have found that the DE has significantly contributed to environment protection, as digital technologies enhance resource efficiency, reduce pollutant emissions and promote green development. Li et al. (Reference Li, Zhao and Wang2024f) empirically found that the DE reduces carbon emissions by lowering energy intensity. Furthermore, Li et al. (Reference Li, Yang and Wang2024e) further revealed that the DE lowers carbon intensity by optimizing industrial structure, driving energy consumption transitions, enhancing green technological innovation and improving resource allocation. However, some scholars contend that the expansion of the DE leads to increased energy consumption, thereby exerting negative environmental impacts (Ozturk & Ullah Reference Ozturk and Ullah2022).
The relationship between the DE and EQ is characterized by bidirectional interactions. On the one hand, the DE can promote ecological improvement through industrial digitalization and digital industrialization. Industrial digitalization introduces data elements and digital tools to traditional industries, boosting productivity, reducing emissions and optimizing resource allocation (Li et al. Reference Li, Wang and Wang2024c, Zhang et al. Reference Zhang, Zhang, Wu, Song, Pan and Feng2024c). On the other hand, digital industrialization reshapes consumption patterns and lifestyles, encouraging low-carbon behaviours through remote work, online services and smart mobility systems (Richardson Reference Richardson2017, Li et al. Reference Li, Wang, Zhang and Wang2024d).
EQ also influences DE development; it provides natural resources and environmental capacity while imposing ecological constraints. For instance, environmental conditions supply the essential inputs – such as land, water and energy – required for the deployment of digital infrastructure. The degradation of EQ increases operational costs and hinders DE expansion. The example of Gui’an New District, where big data development is facilitated by abundant hydropower, illustrates how EQ determines DE sustainability. Moreover, environmental regulation – especially under China’s dual-carbon goals – acts as a policy driver encouraging enterprises to adopt digital technologies for green transformation, referring to the process of enhancing environmental sustainability through digital means. Consistent with the Porter Hypothesis (Zhang et al. Reference Zhang, Zhu, Li and Yan2024b), environmental regulation can enhance productivity and environmental performance (Yang et al. Reference Yang, Lin, Zhu and Yang2023, Zhu et al. Reference Zhu, Chen, Sun and Lyu2023, Lv & Chen Reference Lv and Chen2024, Wang et al. Reference Wang, Chen, Gao and Li2024a). Environmental degradation also weakens the DE by worsening health outcomes, reducing worker productivity, increasing inequality and undermining the region’s attractiveness to skilled labour (Gehrsitz Reference Gehrsitz2017, Qu et al., Reference Qu, She, Li, Zheng and Albitar2024).
In summary, the DE and EQ are interdependent, mutually reinforcing yet potentially constraining one another; their interactions involve dynamic feedback loops whereby digitalization supports green transformation, while EQ sets the boundaries for sustainable digital expansion (Fig. 1). Failure to protect the environment can impede DE growth, while inadequate digital progress may hinder green innovation and prolong reliance on resource-intensive models.
Figure 1. Analytical framework for the coupling coordination mechanism. The framework shows how the digital economy and environmental quality interact. Squares indicate system components; central arrows show mutual promotion; top and bottom arrows reflect reciprocal constraints.
To uncover the mechanisms driving DE–EQ coordination, this study examines four main influencing factors, outlined in the following subsections.
Economic development level
Economic growth theory suggests that regions with higher economic levels typically possess stronger investment capacity, particularly in digital infrastructure (e.g., high-speed internet, data centres), which accelerates technological innovation and DE development. These regions also tend to allocate more resources to environmental protection, achieving better ecological outcomes. In contrast, less developed regions often prioritize economic growth over environmental concerns, resulting in slower DE growth and weaker environment protection (Wang et al. Reference Wang, Wu and Du2023a). Our first hypothesis is that higher economic development significantly enhances the coupling coordination degree (CCD) between the DE and EQ.
Human capital
According to human capital theory, skilled labour is a key driver of economic and digital innovation. High-quality talent fosters DE-related industries and contributes to environmental sustainability through greater eco-awareness and use of green technologies (Li et al. Reference Li, Zhao, Ma, Shao and Zhang2019, Wang et al. Reference Wang, Chen and Li2024b). A shift towards knowledge-based growth also reduces reliance on traditional energy and supports the green transformation (Wang et al. Reference Wang, Shang and Li2023b). Ecological economics highlights that growth based on human capital promotes synergy between economic and environmental goals. Our second hypothesis is that human capital improvements significantly enhance the CCD between the DE and EQ.
Government intervention
Public choice theory emphasizes the government’s role in infrastructure planning, including the digital infrastructure essential for DE growth. Projects such as smart cities and smart grids enhance both operational efficiency and ecological outcomes (Li et al. Reference Li, Lei and Wang2024a). However, excessive intervention may distort markets, hinder innovation and reduce resource efficiency (Ran et al. Reference Ran, Zhang and Zheng2023). Despite regional progress, China’s DE remains in its early stages and is heavily reliant upon public investment. Given that the growth of the DE relies on early infrastructure investments, which are often initiated by the government’s financial contributions, our third hypothesis is that government intervention has a notable positive effect on the CCD between the DE and EQ.
Industrial upgrading
Industrial upgrading shifts economic structures towards high-end manufacturing and services, reducing environmental pressure as per the Environmental Kuznets Curve (Guo & Shahbaz Reference Guo and Shahbaz2024). It stimulates innovation and improves energy efficiency, facilitating green industries and enhancing both DE development and EQ. Transition and technological change theories also underscore that adopting low-carbon, intelligent production models improves productivity and reduces pollution (Feng et al. Reference Feng, Du, Lin and Zuo2020, Wang et al. Reference Wang, Zhang, Wan and Chen2023c, Li et al. Reference Li, Li and Wang2024b). Our fourth hypothesis is that industrial upgrading has a notable positive effect on the CCD of the DE and EQ.
Although the relationship between the DE and EQ has attracted increasing academic attention, empirical studies examining their coupling coordination remain limited. Most existing research focuses on the provincial level, which overlooks the significant spatial heterogeneity and interaction mechanisms that exist at the city level (Han et al. Reference Han, Fu, Lv and Peng2023, Liu et al. Reference Liu, Lu and Li2024).
This study contributes to the literature by constructing a coupling coordination evaluation framework using panel data from 285 prefecture-level cities in China between 2016 and 2021. In doing so, it captures finer spatial dynamics and provides new insights into the factors influencing DE–EQ coordination. At the urban level, the coordination between the DE and EQ is crucial for sustainable development. This coordination not only enhances resource utilization efficiency, reduces pollution and optimizes economic structures, but also strengthens urban resilience and improves residents’ quality of life. This paper builds on existing research and systematically explores the bidirectional relationship between the DE and EQ, along with their spatial differences in coordinated development and driving factors.
The objectives of this study are threefold. First, it aims to construct a comprehensive evaluation index system to measure the CCD between the DE and EQ in Chinese cities. Second, it seeks to analyse the spatial and temporal evolution of the CCD across 285 prefecture-level cities in China. Third, it aims to identify and quantify the factors driving the CCD through a spatial econometric model, thereby revealing regional disparities and providing theoretical support for coordinated urban development. By shifting the analytical scale from the national or provincial level to the city level, this research offers more refined and practical policy implications than those derived from national- or provincial-level analyses.
Methods
Evaluation of index system construction
This study constructed a DE evaluation system based on the measurement approach of Zhao et al. (Reference Zhao, Zhang and Liang2020) and the framework in China’s 14th Five-Year Plan for Digital Economy Development, and it developed an EQ evaluation index system for Chinese cities using the PSR model, following the methodology of Jiang et al. (Reference Jiang, Shi, Su, Lu, Li and Meng2021) and Zhang et al. (Reference Zhang, Wang and Xiong2023; Table 1). Both were measured using the entropy method.
Table 1. Evaluation index system and weights for the digital economy and environmental quality in Chinese cities. Weights are calculated using the entropy method. A positive (+) sign indicates a positive indicator, whereas a negative (–) sign represents a negative indicator.
GDP = gross domestic product.
Coupling coordination degree model
Reflecting both the interaction (coupling degree) and mutual promotion (coordination degree) between systems and offering a comprehensive measure of system synergy, this study used the CCD model to evaluate the coordinated development of the DE and EQ in the assessed cities. The coupling degree was calculated as in Equation 1:
$$C = 2 \cdot {{{\sqrt {Dig \cdot Env} }}\over{{Dig + Env}}}$$
where
$Dig$
and
$Env$
represent the DE index and EQ index, respectively.
$C$
ranges from 0 to 1, with higher values indicating stronger interaction. However, as C only reflects the degree of connection, and not the quality of connection, this study introduces a coordination component for a more objective assessment, as per Equations 2 and 3:
$$T = \alpha Dig + \beta Env$$
$${CCD = {\sqrt {C \cdot T}}}$$
Here,
$T$
is the comprehensive development index and
$\alpha $
and
$\beta $
represent the contribution coefficients of development of the DE and EQ; assuming their equal importance, both were set to 0.5. The resulting CCD ranges from 0 to 1, with higher values indicating better synergy and coordination between the two systems.
To accurately analyse the coupling coordination status between EQ and DE development, this study followed Liu et al. (Reference Liu, Lu and Li2024) in classifying the CCD of EQ and DE development into four specific types (Appendix S1, Table S1).
Driving factors and spatial econometric models
Based on prior analysis and research, the following driving factors were selected: economic development level measured by taking the logarithm of per-capita GDP (PGDP); human capital measured using average years of education (AVE; calculated by weighting student numbers by years of schooling (6 for primary, 10.5 for secondary, 16 for higher) and dividing by the total student population) and education investment (EDU; measured as the ratio of educational expenditure to GDP); government intervention (G) measured by the proportion of general budget expenditure to GDP; and industrial upgrading (IU) measured by the proportion of the value added of the tertiary industry to GDP.
The dependent variable was the CCD between the DE and EQ. The data used were derived from the annual statistical yearbooks of various provinces and cities (Appendix S1, Table S2).
To account for spatial dependence in the analysis of driving factors, spatial econometric models were adopted that capture spatial interactions often overlooked by traditional models. Specifically, three types were used: the Spatial Error Model (SEM), which addresses omitted variable bias through spatial error correlation; the Spatial Autoregressive Model (SAR), which considers spatial dependence in the dependent variable; and the Spatial Durbin Model (SDM), which integrates both to capture spatial spillovers (Wang et al. Reference Wang, Chen, Long and Li2024c). Hausman, Lagrange multiplier (LM), likelihood ratio (LR) and Wald tests were employed to determine the appropriate model. To enhance robustness, the SAR and SEM with two-way fixed effects were also estimated. The general form of spatial econometric models was as per Equation 4:
$$CC{D_{it}} = \alpha + \beta {X_{it}} + \rho \mathop \sum \limits_{i \ne j}^n {w_{ij}}CC{D_{jt}} + \gamma {\mathop \sum \limits_{i \ne j}^n {_i}{X_{jt}}} + {\mu _{it}}$$
$${\mu_{it}} = \lambda \mathop \sum \limits_{i \ne j}^n {w_{ij}}{\mu_{jt}} + {\epsilon _{it}}$$
where
$CC{D_{it}}$
is the dependent variable,
${X_{it}}$
represents the driving factors,
$\alpha $
stands for the constant term,
$\beta $
stands for the spatial regression coefficient of the independent variable,
$\rho $
stands for the spatial regression coefficient of the dependent variable,
$\gamma $
stands for the spatial regression coefficient of driving factors,
$\lambda $
stands for the spatial error regression coefficient and
${\epsilon_{it}} $
stands for the random error term. When
$\rho $
= 0,
$\gamma $
= 0 and
$\lambda \;$
≠ 0, the model was SEM; when
$\rho $
≠ 0,
$\gamma $
= 0 and
$\lambda $
= 0, the model was SAR; and when
$\rho $
≠ 0,
$\gamma $
≠ 0 and
$\lambda \;$
≠ 0, the model was SDM.
Threshold model
Considering that some driving factors may exhibit threshold effects, Hansen’s (Reference Hansen1999) threshold regression model was introduced, as per Equation 5:
$${CC{D_{it}} = {\mu_i} + {\beta_1}{x_{it}} \cdot {\mathbb{I}}({q_{it}} \le \gamma) + {\beta_2}{x_{it}} \cdot {\mathbb{I}}({q_{it}} \gt \gamma) + {\epsilon_{it}}}$$
where
$CC{D_{it}}\;$
denotes the dependent variable,
${x_{it}}$
represents the independent variable and
${q_{it}}$
is the threshold variable used to segment the sample. The parameter
$\gamma $
is the estimated threshold value that divides the sample into distinct regimes, and
${\epsilon_{it}} $
is the stochastic error term. The indicator function
$\mathbb{I}$
takes the value 1 if the specified condition is satisfied and 0 otherwise. The coefficients
${\beta _1}\;and\;{\beta _2}$
capture the marginal effects of the independent variable in the regimes below and above the threshold, respectively.
Before testing for the threshold effect, it was essential to identify the number of thresholds. A traditional grid search method was employed to identify candidate threshold values, dividing the sample interval into 100 grids and applying 300 bootstrap replications to test the significance of the threshold model.
Data sources
The study was based on balanced panel data from 285 Chinese cities between 2016 and 2021. The digital financial inclusion index for China was obtained from the index jointly compiled by the Digital Finance Research Center of Peking University and Ant Group (https://www.antgroup.com). Data on DE-related patents in each city were sourced from the Chinese Research Data Services Platform (CNRDS), while the number of artificial intelligence enterprises in each city was collected from Tianyancha (https://www.tianyancha.com). Additional economic panel data were taken from the China City Statistical Yearbook, other statistical yearbooks, government reports from various provinces and cities and the China Stock Market and Accounting Research (CSMAR) database. For missing values, linear interpolation was used to fill in the gaps. The detailed data sources are shown in Appendix S1, Table S3.
Results
Measurement of the CCD between China’s DE and EQ
From 2016 to 2021, the average coupling degree and CCD between China’s DE and EQ displayed an upward trend (Fig. 2). The overall coupling degree increased from 0.381 in 2016 to 0.506 in 2021, and the CCD rose from 0.327 in 2016 to 0.393 in 2021. This trend suggests a growing interaction between the DE and EQ. The overall state reached a moderate coupling coordination level, with an average growth of c. 4% over these 5 years.
Figure 2. Trends of coupling degree and coupling coordination degree (CCD) between the digital economy and environmental quality. Error bars indicate standard deviations, reflecting inter-city variability. The yellow and black lines represent parabolic trend fits for coupling degree and CCD, respectively.
The kernel density curve (Appendix S1, Fig. S1) shifts noticeably rightward, signifying an improvement over time in the overall CCD. A slight secondary peak appears on the curve, indicating a ‘one main, one secondary’ bimodal structure of the CCD, suggesting the existence of a ‘club convergence’ phenomenon. This implies that some cities have formed closely aligned groups in terms of coupling coordination, indicating a convergence of the CCD within certain regions or city clusters. Additionally, the trailing characteristic on the right side of the curve shows that some cities were significantly leading others in terms of coupling coordination levels.
From the perspective of urban spatial layout (Fig. 3), the CCD between DE development and EQ in China became increasingly balanced. In 2016, a third of cities were in a low coupling coordination state (Fig. 3, green areas), but by 2021, all cities had reached a moderate coupling coordination level or higher (Fig. 3, yellow areas), with some cities even reaching a high coupling coordination level (Fig. 3, orange areas), and with Beijing achieving an extreme coupling coordination level (Fig. 3, red area). Specifically, in 2016, the CCD between the DE and EQ in China mainly exhibited a ‘high in the east, low in the west’ pattern. By 2021, the pattern had shifted to a higher CCD in the central cities of urban clusters compared to the peripheral cities. For example, Guangzhou and Shenzhen in the Pearl River Delta Urban Cluster, Beijing and Tianjin in the Beijing–Tianjin–Hebei Urban Cluster, Shanghai and Hangzhou in the Yangtze River Delta Urban Cluster and Chongqing and Chengdu in the Chengdu–Chongqing Urban Cluster were all in a high coupling coordination state, whereas the CCD of the peripheral cities was relatively lower.
Figure 3. Spatial evolution of the coupling coordination degree in China in 2016 and 2021. Different colours indicate different levels of coupling coordination. The image is based on the standard map released by the Ministry of Natural Resources of the People’s Republic of China (No. GS (2024) 0650).
From the viewpoint of China’s four regions (Appendix S1, Fig. S2), coupling levels across all areas rose, although regional disparities persisted. Specifically, the CCD in the eastern region significantly exceeded the national average, reaching 0.4 in 2019, whereas other regions remained below this value. The CCD in the north-eastern region stagnated between 2017 and 2018, demonstrating slower growth than in the central and western regions.
The regions showed significant differences in their development, with the CCD varying at different points in time (Fig. 4). The years 2016–2017 marked the initial phase of DE development, during which government and societal awareness and support for the DE gradually increased, laying the foundations for a policy framework. During this phase, 62% and 35% of cities, respectively, were in moderate and low coupling coordination states. From 2018 to 2019, 87% of cities achieved a moderate level of coupling coordination, marking a rapid development period for the DE. Significant progress was made in technological innovation and internet infrastructure, and sectors such as e-commerce, artificial intelligence and big data began to scale up. Additionally, the execution of the ‘Three-Year Action Plan to Win the Battle for a Blue Sky’ strengthened environmental regulations, ingraining the concept of green development and fostering initial coordination between the ecological and economic sectors (The State Council of the People’s Republic of China 2018). During the period 2020–2021, the COVID-19 pandemic triggered explosive growth in the DE, and cities with low coupling coordination disappeared entirely, with 25 cities reaching high coupling coordination levels and one city achieving an extreme coupling coordination level; a growing number of cities reached a new level of coupling coordination. This trend is underpinned by the prioritization of high-quality DE development as a national strategy, with the adoption of digital technologies into traditional industries becoming mainstream. At the same time, the ‘Carbon Peak and Carbon Neutrality Plan’ was integrated into the nation’s sustainable development strategy as a critical component (The State Council of the People’s Republic of China 2021).
Figure 4. Chord diagram presenting the proportions of Chinese cities falling into different coupling coordination categories across the years assessed. Each outer arc denotes either a year or a coordination level (e.g., low, moderate, high), and the chords between them represent the distribution of coordination levels in a given year. The relative thickness of each chord reflects the share of cities at that level during that year.
Driving factors of the CCD between China’s DE and EQ
The LM test was significant, suggesting that the SDM was most suitable for this analysis (Appendix S1, Table S4). Both the Wald and LR tests rejected the null hypothesis, which suggests that the SDM did not reduce to either the SAR or SEM. Moreover, the Hausman test was passed, confirming the appropriateness of the SDM with two-way fixed effects for this analysis.
The spatial regressive coefficient
$\rho $
was positive and statistically significant at the 1% level, signifying a positive spatial spillover effect (Table 2). This implies that an increase in the CCD of one city led to a corresponding improvement in the CCD of neighbouring cities. Furthermore, each driving factor had a significant impact on the CCD between the regional DE and EQ.
Table 2. Results of spatial econometric model (values in parentheses are z-statistics).
***, ** and * indicate significance at the 1%, 5% and 10% levels, respectively.
a Log-likelihood measures model fit, with higher values indicating better fit.
$\lambda $
= spatial error regression coefficient;
$\rho $
= spatial regression coefficient of the dependent variable; σ2_e = variance of the error term; AVE = average years of education; EDU = education investment; G = government intervention; IU = industrial upgrading; N = total number of observations; PGDP = per-capita gross domestic product; R2 = coefficient of determination; SAR = Spatial Autoregressive Model; SDM = Spatial Durbin Model; SEM = Spatial Error Model; w. = spatially lagged explanatory variable, reflecting spillover effects from neighbouring regions in the spatial econometric model.
Although some driving factors had significant coefficients, and the coefficients for the spatial lag terms were also significant, they did not reflect the direct and indirect effects of the driving factors on the CCD between the DE and EQ. Following the method of LeSage and Pace (Reference LeSage and Pace2009), the spatial total effect was broken down into direct and indirect effects, and doing so indicated the following four things (Appendix S1, Table S5): first, both direct and indirect effects of economic development (PGDP) were significantly positive, supporting our first hypothesis. This suggests that local economic strength and regional integration jointly enhance CCD, with developed cities driving surrounding areas through market demand and industrial linkages.
Second, in terms of human capital, education investment (EDU) had a significant positive direct effect, whereas average years of education (AVE) was significantly negative, which contradicts our second hypothesis. This suggests that although an increase in years of schooling indicated an overall improvement in the knowledge level of the labour force, it did not necessarily translate into an enhancement of skills and innovation capabilities that matched the demands of the DE. The current education system still primarily focuses on traditional teaching methods, failing to adequately cultivate professionals who are equipped to meet the needs of digital transformation and environmental protection. Neither EDU nor AVE showed significant indirect effects, possibly due to limited knowledge spillovers constrained by market conditions and local protectionism (Song & Liu Reference Song and Liu2023).
Third, government intervention (G) showed a significantly negative direct effect, contradicting our third hypothesis. Several factors might explain this result: although local governments emphasize DE development and propose investments in digital infrastructure and industrial upgrading, these plans require time to implement. Most cities remain in the early stages of digitalization, facing underdeveloped models and weak policy execution, delaying the impacts of government efforts. The present study also used the fiscal expenditure-to-GDP ratio to measure intervention. As digital infrastructure demands long-term investment, high initial spending may not translate into immediate outcomes. The indirect effect is insignificant, suggesting limited fiscal spillovers from neighbouring cities, probably due to administrative barriers and protectionism that hinder regional cooperation.
Fourth, industrial upgrading (IU) exhibits a significantly positive direct effect, supporting our fourth hypothesis, and also demonstrates a positive indirect effect. This indicates that IU in a given region not only enhances its own coupling coordination between the DE and EQ, but also fosters coordination in surrounding areas through technological advancements and the diffusion of innovation. At the same time, regional environmental pollution is often influenced by neighbouring areas, and a region’s pollution has significant spatial effects (Hosseini & Kaneko Reference Hosseini and Kaneko2013). Therefore, upgrading the industrial structure not only enhances local coupling coordination through technological advances, but also reduces the risk of pollution being transferred to surrounding areas by decreasing the proportion of highly polluting industries. This effect of environmental industry and technology diffusion may create a positive feedback mechanism across regions, further promoting green development in surrounding areas.
Threshold effect
The spatial econometric results indicate that government intervention exerted a significant negative effect, contradicting our third hypothesis. There may be a threshold effect in government intervention; when government investment remains below a critical threshold, its negative effects tend to dominate. However, once investment surpasses this threshold, it is likely to significantly enhance the coverage and quality of digital infrastructure, further stimulating technological innovation and industrial upgrading, and ultimately fostering a virtuous cycle between the DE and EQ in terms of production, lifestyle and ecological benefits. Only the single-threshold effect was significant, with a threshold of 13.31% of GDP (Appendix S1, Table S6), and the regression results (Appendix S1, Table S7) indicate that when the general budget expenditure ratio exceeded 13.31%, government investment had a significant positive influence on the CCD between the DE and EQ. This finding further corroborates a ‘threshold effect’ in government intervention; only when government intervention reached a sufficiently high intensity could it effectively promote the coordinated development of the DE and EQ.
Discussion
The analysis reveals that the CCD between the DE and EQ showed an upward trend during 2016–2021. However, the overall CCD remained at a moderate level, with significant regional disparities: eastern regions and core cities within urban clusters demonstrated notably higher CCD compared to western and north-eastern regions. The CCD in the north-eastern region stagnated between 2017 and 2018 and remained lower than in the central and western regions throughout the study period. This lag may be associated with the region’s reliance on heavy industry and its relatively slow response to economic restructuring and technological upgrading. In addition, sustained population outflow and talent loss to more developed regions have contributed to labour shortages, potentially constraining local digital and green development. Looking ahead, assuming economic development and policy support remain relatively stable, the CCD is expected to reach 0.5 by 2028, indicating a shift towards high coupling coordination. Of the influencing factors, economic development, education investment and industrial upgrading significantly enhance local CCD. Interestingly, human capital – measured by average years of education – had a negative impact, implying a possible mismatch between educational output and the demands of the green digital transformation. Additionally, government intervention exhibited a threshold effect: the construction of digital economic infrastructure requires substantial investment, and policy initiatives were still in their early exploratory stages. When government investment remained below a critical threshold, its negative effects tended to dominate, but when fiscal expenditure exceeded 13.31% of GDP, its influence on CCD became positive. This finding underscores the importance of ensuring that fiscal investment reaches the necessary scale, as insufficient investment may hinder progress. The results also reveal spatial spillover effects, with economic development and industrial upgrading in one region positively affecting the CCD of neighbouring areas. These findings suggest that policy responses should be tailored to regional and city-level characteristics to promote more balanced and effective coordination. To address regional disparities in China, differentiated policies are needed. Eastern regions should continue to strengthen technological innovation and ecological protection to enhance development quality in terms of both economic growth and environmental sustainability. Western and north-eastern regions require greater policy support, improved infrastructure and talent development to advance digital and ecological coordination. Core cities should play a leading role by facilitating cross-regional resource flows and promoting cooperation through technology transfer and platform sharing.
The regional imbalance observed in China – where the east outpaces the west and north-east – is consistent with patterns identified in other emerging economies, where uneven digital infrastructure and policy implementation hinder balanced development (United Nations Conference on Trade and Development 2021). The identification of a government intervention threshold complements earlier research emphasizing the importance of effective governance design, supporting the view that well-calibrated public expenditure can catalyse sustainable digital transitions (Borrás & Edquist Reference Borrás and Edquist2013). Notably, the negative effect of human capital, as measured by education years in our study, contrasts with mainstream findings that regard education as a positive driver of green and digital capabilities. It may be that in rapidly transforming economies, the content and orientation of education systems lag behind technological and environmental needs, echoing concerns raised by Deev et al. (Reference Deev, Gamidullaeva, Finogeev, Finogeev and Vasin2021) regarding skill mismatches. Furthermore, the spatial spillover effects observed in this study are in line with the theory of innovation diffusion (Autant-Bernard Reference Autant-Bernard2001), underscoring the importance of regional cooperation in fostering coordinated progress. Collectively, this study not only validates several globally recognized mechanisms, but also highlights context-specific challenges and policy implications for enhancing the synergy between digital and ecological development in transitional economies.
Our results are constrained by indicator selection and timespan. First, due to the lack of complete time-series data at the city level for key variables such as water availability, biodiversity and carbon emissions of the DE, we adopted more accessible pollution-related proxies. Doing so is likely to underestimate positive ecosystem services (e.g., habitat provision, water purification and carbon sequestration) and to place more emphasis on environmental pressures, leading the model to reflect environmental degradation rather than fully capturing ecological restoration and sustainable development. Second, the 6-year timeframe captures the development of China’s DE and ecological progress during 2016–2021 but could not reflect longer-term fluctuations, policy cycles or the full effects of drivers such as infrastructure and education. As more city-level ecological data become available, future research should extend the study period and improve the indicator system to better assess the evolution of the coupling coordination between the DE and EQ in different policy and economic contexts.
Supplementary material
To view supplementary material for this article, please visit https://doi.org/10.1017/S0376892925100192.
Data availability
The datasets used and/or analysed during the current study are available from the corresponding author upon reasonable request.
Acknowledgements
None.
Author contributions
Liping Wang: conceptualization, formal analysis, supervision, writing – review and editing, funding acquisition; Zhonghao Ye: data curation, resources, writing – original draft, writing – review and editing; Chuang Li: methodology, validation, writing – review and editing.
Financial support
The research is supported by the National Social Science Fund of China (24FJYB037).
Competing interests
The authors declare none.
Ethical standards
Not applicable.
