Spatial differentiation and influencing factors of urban economic resilience: an empirical study from China Authors: Wang Hongyue, Wang Bida, Li Xuerong Published: 2026 ABSTRACT Given the intensifying geopolitical tensions, public health emergencies and frequent natural disasters, urban economic stability has become increasingly challenging, and determining how to achieve stable and high-quality urban economic development is a critical issue. In this study, the economic resilience of 286 Chinese prefecture-level cities from 2008 to 2022 is evaluated via the CRITIC-TOPSIS method. Natural breakpoint classification, kernel density estimation, and geographical detector and MGWR models are adopted for hierarchical division, spatial evolution and heterogeneous driving factors. The key findings are as follows: (1) Urban economic resilience grows steadily with obvious hierarchical differentiation; the high-resilience areas are mainly large cities, whereas the level of resilience of small cities remains low. (2) Among the resilience subdimensions, resistance capacity is dominant, followed by recovery and innovation capacity, while innovation capacity shows the strongest growth momentum. (3) Spatially, eastern and central cities exhibit gradient agglomeration characteristics, whereas western and northeastern cities show prominent polarization. (4) The factors that influence display evident spatial heterogeneity. Digitalization, medical care and transportation positively affect resilience, and the effects increase from east to west. Urbanization has stronger positive effects in the eastern and central regions, whereas industrialization has inhibitory effects, particularly in resource-dependent cities. Transportation accessibility has dual effects on economic resilience.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1038/s41598-026-58841-z. Introduction In 2020, the sudden onset of the COVID-19 pandemic generated pervasive shocks to global economic systems, public healthcare infrastructure, and corporate operational activities1,2 There were heterogeneous economic responses to this exogenous shock across countries, with significant disparities in the areas of shock resistance and resilience. According to the International Monetary Fund (2024), the economic growth rates of the United States were − 2.28%, 5.9%, and 2.1% in 2020, 2021, and 2022, respectively. Germany’s economic growth rates were − 3.8%, 3.2%, and 1.8%, whereas Japan experienced more significant fluctuations, with rates of − 4.3%, 2.0%, and 1.0%, respectively. In contrast, China’s economic growth rates were 2.2%, 8.4%, and 3%, respectively, over the same period, which fully demonstrates the strong resilience and development vitality of a major economy. As the core carriers of national economic development, cities concentrate key factors such as labour, capital, and technology. As they rely on diversified industrial systems, efficient resource allocation mechanisms, and innovation ecosystems, cities not only possess risk resistance capabilities but also have remarkable potential for self-repair, transformation, and upgrading after crises, becoming the core engine driving the sustainable development of regional economies. On December 5, 2024, the General Office of the State Council issued the Opinions on Promoting the Construction of New-Type Urban Infrastructure and Building Resilient Cities, which explicitly presented the overall requirement of “promoting the digital transformation of urban infrastructure, continuously enhancing urban facility resilience, management resilience, and spatial resilience, and advancing the safe development of cities.” This strategic deployment underscores the significant importance of urban economic resilience for the safeguarding of national economic security and the promotion of high-quality development. Literature review Research on economic resilience The concept of resilience, which is originate from engineering, was first used to describe the recovery time after a system is affected by shocks3. Holling further extended this concept to ecology, emphasizing multiple equilibria rather than original state restoration4. However, this concept still cannot be adapted to systems that change over time. To address this issue, Walker proposed evolutionary resilience, which emphasizes the dynamic adaptability of the environment5. Since then, spatial economics has been brought into evolutionary resilience to account for agglomeration phenomena6. Since the 1990s, the rise of evolutionary economic geography has provided solid theoretical support for resilience theory. Many researchers are interested in economic resilience. From the short-term perspective, resilience is focused on resistance and recovery. However, from the long-term perspective, the concept of resilience highlights innovative development7. Overall, resilience has been summarized as the capacity for resistance, recovery, and innovation from an adaptive perspective8,9. Variable selection for economic resilience To select the variables for economic resilience, single-index and multi-index evaluation metrics were used. The former highlights the dynamic evolution and recovery process of the system. By depicting the time changes before and after the disturbance through a single variable, the system’s adaptability to the shock can be intuitively reflected. The variables selected include sensitivity indices such as the regional employment level, real regional GDP, the change rate of employees, and the volume of foreign trade10–13. However, because of the complex and multidimensional characteristics of the urban economic system, comprehensively covering the complex changes in the system with a single indicator is difficult. Hence, the multi-indicator assessment method can be used to more comprehensively reflect the level of resilience. From a macro perspective, the economic resilience index includes industrial structure, the labour market, financial development, government institutions, and infrastructure14,15. In accordance with Martin’s definition of resilience, the measurement framework includes resistance and recovery capability, adaptation capability, and innovation transformation capability16–19, which are variables that can be used to more systematically and comprehensively assess economic resilience. Measurement methods and objects of economic resilience For measurement methods, most existing studies construct indicator systems are first constructed in most existing studies, followed by the adoption of the entropy method20–22, coefficient of variation method23, and subjective evaluation methods, which are widely applied because of their objectivity in weight calculation, to assess economic resilience24. However, these methods may suffer from issues such as expert bias, cognitive bias, ignoring the interrelationships between indicators and consistency testing problems, which can introduce bias into the evaluation results. Hence, selecting appropriate evaluation methods is important. The effectiveness of measurement methods depends on the definition of the research object. The initial focus of the literature was on the quantitative evaluation of economic resilience at the national and regional macro scales25–28. Considering the inadequate interpretation of the microscopic mechanisms embedded in macrolevel analyses, scholarly attention has gradually shifted towards industrial and firm-level micro investigations29–32. Additionally, macro and micro analytical perspectives have been integrated in contemporary scholarship. Such integrated frameworks combine the behavioural autonomy of microeconomic agents with the regulatory effects of macro policies, thereby facilitating the improvement of economic resilience via multidimensional synergistic governance33,34. Factors that influence urban economic resilience Scholarly works have extensively investigated the determinants of urban economic resilience, including the digital economy35,36, industrial diversification37, financial development38, technological innovation39, and governmental regulation40. Among the multifaceted influencing factors, the core enablers are factor agglomeration and the digital economy. As proposed by Krugman, the fundamental determinants of economic agglomeration are increasing returns to scale, transportation costs, and factor mobility. Driven by self-reinforcing mechanisms, agglomeration continues to persist and evolve, ultimately forming a typical core-periphery spatial structure41,42. Empirical evidence further confirms that industrial agglomeration can increase urban productivity, strengthen the ability of urban areas to withstand external shocks and achieve transformative recovery, thereby underpinning an improvement in economic resilience43. In the context of the digital economy, industrial agglomeration leads to the optimisation of factor allocation and the consolidation of economic foundations to reinforce the level of urban risk resistance risk. Moreover, digital platforms decrease transaction costs and facilitate precise supply–demand coordination, which contributes to post-shock economic recovery44–46. Additionally, the knowledge spillover effect of internet technology promotes corporate technological advancement and stimulates urban innovative momentum47,48. In addition to economic and industrial factors, the literature highlights the notion that infrastructure acts as a crucial safeguard for the mitigation of economic systemic risk to cities. On the basis of the abundant research achievements in economic resilience, a solid theoretical foundation has been laid by the literature, which provided methodological references for this paper. However, there are still areas in the existing research that need further exploration and improvement. In terms of evaluation methods, traditional approaches ignore the correlations among indicators. In terms of research scope, most of the existing studies are focused on countries, provinces and urban agglomerations49–53, with insufficient analyses of the 286 prefecture-level cities54–57. The existing research on economic resilience is focused mostly on the overall level, while systematic investigations into its subdimensions remain relatively scarce. In terms of driving factor identification, when a single factor is taken as the main research focus, traditional OLS and WGR models are constrained by spatial limitations and become unable to precisely quantify the magnitude of influencing factors58. On this basis, the marginal contributions of this paper are as follows: First, the CRITIC-TOPSIS framework is constructed, which comprehensively considers the information content and inherent correlations among indicators, thus yielding more objective and reliable weighting results. Second, with 286 prefecture-level cities in China as the research objects, urban economic resilience is decomposed into three dimensions—resistance, recovery, and innovation capacity—to enable refined regional comparative analysis. Third, the inclusion of factors such as agglomeration, digitalization, and infrastructure in the empirical research enriches the study, and an analytical framework that integrates the geographical detector and MGWR is developed, breaking the spatial homogeneity assumption embedded in conventional models and systematically revealing the spatiotemporal evolution, driving determinants, and heterogeneous characteristics of urban economic resilience. Fig. 1Research framework of urban economic resilience. Research framework of urban economic resilience. Indicator measurement, research methodology and data sources Indicator measurement On the basis of complex adaptive system theory, evolutionary economic geography and resilience theory59,60 and in conjunction with previous studies, urban economic resilience is defined in this paper as resistance, recovery, and innovation capacity in the face of external shocks. On this basis, a comprehensive evaluation system comprising 15 specific indicators is established (see Table 1). Table 1Indicator system for the development level of urban economic resilience.Objective levelPrimary indicatorsSecondary indicatorsTertiary indicatorsVariablesAttributeWeightUnitsRationale for selectionEconomic resilienceResistance abilityEconomic foundationPer capita regional gross domestic productx1+0.064Yuan 17 Unemployment levelThe number of registered unemployed persons in urban areasx2−0.073Person 16 Residents’ income levelTotal wages of employeesX3+0.091104yuan 15 Fiscal revenue levelProportion of local general budgetary revenue in GDPX4+0.115% 26 Degree of dependence on foreign countriesProportion of total import and export trade volume in GDPX5+0.065% 15 Recovery abilityMarket potentialProportion of the total retail sales of social consumer goods in GDPX6+0.105% 25 Investment levelProportion of fixed asset investment in GDPX7+0.069% 26 Savings levelProportion of the balance of deposits in financial institutions at the end of the year in GDPX8+0.03% 25 Industrial diversificationProportion of the added value of the tertiary industry in GDPX9+0.108% 21 Innovation abilityInnovation potentialThe number of students enrolled in regular institutions of higher learningX10+0.127People 17 Innovation inputProportion of scientific expenditures in GDPX11+0.04% 18 Innovation outputNumber of patent authorizationsX12+0.051Pieces 17 The number of artificial intelligence enterprisesX13+0.040Units 23 “+” denotes a positive indicator; “−” denotes a negative indicator. Table 1Indicator system for the development level of urban economic resilience.Objective levelPrimary indicatorsSecondary indicatorsTertiary indicatorsVariablesAttributeWeightUnitsRationale for selectionEconomic resilienceResistance abilityEconomic foundationPer capita regional gross domestic productx1+0.064Yuan 17 Unemployment levelThe number of registered unemployed persons in urban areasx2−0.073Person 16 Residents’ income levelTotal wages of employeesX3+0.091104yuan 15 Fiscal revenue levelProportion of local general budgetary revenue in GDPX4+0.115% 26 Degree of dependence on foreign countriesProportion of total import and export trade volume in GDPX5+0.065% 15 Recovery abilityMarket potentialProportion of the total retail sales of social consumer goods in GDPX6+0.105% 25 Investment levelProportion of fixed asset investment in GDPX7+0.069% 26 Savings levelProportion of the balance of deposits in financial institutions at the end of the year in GDPX8+0.03% 25 Industrial diversificationProportion of the added value of the tertiary industry in GDPX9+0.108% 21 Innovation abilityInnovation potentialThe number of students enrolled in regular institutions of higher learningX10+0.127People 17 Innovation inputProportion of scientific expenditures in GDPX11+0.04% 18 Innovation outputNumber of patent authorizationsX12+0.051Pieces 17 The number of artificial intelligence enterprisesX13+0.040Units 23 “+” denotes a positive indicator; “−” denotes a negative indicator. Indicator system for the development level of urban economic resilience. “+” denotes a positive indicator; “−” denotes a negative indicator. Additionally, the digitalization level, urbanization level, industrialization level, traffic accessibility, transportation level, and medical level are utilized as impact factors for the urban economy (see Table 2 for details). Table 2Factors influencing the economic resilience of Chinese cities.Primary indicatorsSecondary indicatorsUnitsVariablesDigitalization levelNumber of international internet users%z1Urbanization levelUrbanization rate%z2Industrialization levelAdded value of secondary industry divided by GDP%z3Medical levelNumber of beds in hospitals and health centresUnitsz4Traffic accessibilityPassenger volume of road transportation104peoplez5Transportation levelTotal volume of postal service104yuanz6 Table 2Factors influencing the economic resilience of Chinese cities.Primary indicatorsSecondary indicatorsUnitsVariablesDigitalization levelNumber of international internet users%z1Urbanization levelUrbanization rate%z2Industrialization levelAdded value of secondary industry divided by GDP%z3Medical levelNumber of beds in hospitals and health centresUnitsz4Traffic accessibilityPassenger volume of road transportation104peoplez5Transportation levelTotal volume of postal service104yuanz6 Factors influencing the economic resilience of Chinese cities. This table presents the factors influencing the economic resilience of 286 Chinese cities from 2008 to 2022, including specific measurements, units and alphabetical representations of indicators such as agglomeration, digital and infrastructure factors. Research methodology CRITIC-TOPSIS methodology The CRITIC-TOPSISI methodology was adopted to measure the development level of urban economic resilience in this study. First, CRITIC was applied to assign weights to the index. Second, TOPSIS was employed to approach the ideal solution. Therefore, using this mixed method, the correlations and variation within and between indices are comprehensively considered, thereby enhancing the objectivity of the measurement results. The detailed steps are as follows: First, standardized processing of the original data is conducted.1\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}{\mathrm{X}}_{\mathrm{i}\mathrm{j}}=\frac{{\mathrm{x}}_{\mathrm{i}\mathrm{j}}-\mathrm{m}\mathrm{i}\mathrm{n}\left({\mathrm{x}}_{\mathrm{j}}\right)}{\mathrm{m}\mathrm{a}\mathrm{x}\left({\mathrm{x}}_{\mathrm{j}}\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left({\mathrm{x}}_{\mathrm{j}}\right)}\left(\mathrm{p}\mathrm{o}\mathrm{s}\mathrm{i}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\:\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{o}\mathrm{r}\right) \end{array}$$\end{document}2\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}{\:\mathrm{X}}_{\mathrm{i}\mathrm{j}}=\frac{\mathrm{m}\mathrm{a}\mathrm{x}\left({\mathrm{x}}_{\mathrm{j}}\right){-\mathrm{x}}_{\mathrm{i}\mathrm{j}}}{\mathrm{m}\mathrm{a}\mathrm{x}\left({\mathrm{x}}_{\mathrm{j}}\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left({\mathrm{x}}_{\mathrm{j}}\right)}\left(\mathrm{n}\mathrm{e}\mathrm{g}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{v}\mathrm{e}\:\mathrm{i}\mathrm{n}\mathrm{d}\mathrm{i}\mathrm{c}\mathrm{a}\mathrm{t}\mathrm{o}\mathrm{r}\right) \end{array}$$\end{document} The standard deviation of each indicator is calculated.3\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}\begin{array}{c}{{\upsigma\:}}_{\mathrm{j}}=\sqrt{\frac{1}{\mathrm{m}-1}\sum\:_{\mathrm{i}=1}^{\mathrm{m}}{\left({\mathrm{x}}_{\mathrm{i}\mathrm{j}}-{\stackrel{-}{\mathrm{x}}}_{\mathrm{j}}\right)}^{2}}\:\end{array} \end{array}$$\end{document} The correlation coefficient between indicators is calculated.4\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}{r}_{jk}=\frac{{\sum\:}_{i=1}^{m}\left({X}_{ij}-{\stackrel{-}{X}}_{j}\right)\left({X}_{ik}-{\stackrel{-}{X}}_{k}\right)}{\sqrt{\sum\:_{i=1}^{m}{\left({X}_{ij}-{\stackrel{-}{X}}_{j}\right)}^{2}}\sqrt{{\sum\:}_{i=1}^{m}{\left({X}_{ik}-{\stackrel{-}{X}}_{k}\right)}^{2}}} \end{array}$$\end{document} The information volume of the indicators is calculated.5\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}{\mathrm{C}}_{\mathrm{j}}={{\upsigma\:}}_{\mathrm{j}}\sum\:_{\mathrm{k}=1}^{\mathrm{n}}\left(1-\left|{\mathrm{r}}_{\mathrm{j}\mathrm{k}}\right|\right) \end{array}$$\end{document} The weight is determined through normalization.6\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}{\mathrm{w}}_{\mathrm{j}}=\frac{{\mathrm{C}}_{\mathrm{j}}}{\sum\:_{\mathrm{k}=1}^{\mathrm{n}}{\mathrm{C}}_{\mathrm{k}}}\end{array}$$\end{document} A weighted standardized matrix is constructed.7\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}{R}_{ij}={X}_{ij}\ast\:{w}_{j}\end{array}$$\end{document} The optimal and worst Euclidean distances D are calculated.8\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}{D}_{i}^{+}=\sqrt{{\sum\:}_{j=1}^{n}{\left({R}_{ij}-max\left({R}_{ij}\right)\right)}^{2}} \end{array}$$\end{document}9\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}{D}_{i}^{-}=\sqrt{{\sum\:}_{j=1}^{n}{\left({R}_{ij}-min\left({R}_{ij}\right)\right)}^{2}} \end{array}$$\end{document} The composite index \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{C}_{i}$$\end{document} is calculated for each region.10\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}{C}_{i}=\frac{{D}_{i}^{-}}{{D}_{i}^{+}+{D}_{i}^{-}} \end{array}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{C}_{i}$$\end{document} represents the resilience level within the range of [0, 1]. The level of economic resilience is positively correlated with the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{C}_{i}$$\end{document} value. Kernel density estimation The kernel density estimation method, which is a nonparametric estimation technique enables the smooth estimation of the probability density of a random variable to reveal its distribution pattern. Its formula is as follows:11\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}f\left(x\right)=\frac{1}{nh}\sum\:_{i=1}^{n}K\left(\frac{x-{X}_{i}}{h}\right) \end{array}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:K(\cdot\:)$$\end{document} is the kernel function, h is the bandwidth, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:n$$\end{document} is the number of observations, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{X}_{i}$$\end{document} is a sample value, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:x\:$$\end{document} is the average value of the observation samples. Geographical detector As a nonparametric statistical tool, the geographical detector is widely used to investigate the spatial differentiation and driving mechanisms of geographical phenomena. It’s use overcomes the constraints of linear assumptions and the data distribution limitations, thereby effectively filling the methodological gap in spatial heterogeneity analysis. In this study, the geographical detector is used to explore the relationships between multiple influencing factors and urban economic resilience to extract dominant determinants. The specific formula is as follows.12\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}q=1-\frac{\sum\:_{h=1}^{L}{N}_{h}{{\sigma\:}_{h}}^{2}}{N{\sigma\:}^{2}} \end{array}$$\end{document} where the symbol q represents the detection power value of the detection factor, L represents the count of samples in the secondary region, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{N}_{h}$$\end{document} indicates the number of samples in the primary region. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{{\sigma\:}_{h}}^{2}\:$$\end{document} represents the variance in the primary region, N represents the total number of samples in the study area, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\sigma\:}^{2}\:$$\end{document} represents the overall variance in the observation area. Multiscale geographically weighted regression (MGWR) The multiscale geographically weighted regression (MGWR) model is a regression model for analysing spatial data. Developed from the classical geographically weighted regression model, this model allows different spatial processes to have their own scales of operation, thereby enabling a more refined and accurate understanding of how each influencing factor varies across different geographic locations. The calculation formula of the multiscale geographically weighted regression is as follows:13\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\begin{array}{c}{Y}_{i}=\sum\:_{j=1}^{k}{\beta\:}_{b{w}_{j}}\left({u}_{i},{v}_{i}\right){X}_{ij}+{\epsilon\:}_{i} \end{array}$$\end{document} where the symbol Y represents the economic resilience level in city i, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\left({u}_{i},{v}_{i}\right)$$\end{document} refers to the geographic coordinates of city i, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:b{w}_{j}$$\end{document} is the bandwidth used for the jth variable, and the differences in its magnitude represent the sensitivity of different variables to distance. \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\beta\:}_{b{w}_{j}}$$\end{document} is represented by the regression coefficient obtained from the regression of variable j, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{X}_{ij}$$\end{document} is the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:j\:$$\end{document}variable of the city\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:i.$$\end{document}. Geometric average growth rate The dynamic evolution of urban economic resilience reflects the temporal evolutionary regularity of urban economic resilience when confronted with internal and external disturbances. In this paper, the geometric average growth rate is used to characterize evolutionary momentum, which can intuitively reflect the change in direction and rate of resilience foundation and constitutes the core connotation of evolutionary momentum. The specific calculation formula is as follows:14\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${ \mathrm{g}}_{\mathrm{n}}=\left[{\frac{{\mathrm{c}}_{\mathrm{n}}}{{\mathrm{c}}_{0}}}^{\frac{1}{\left(\mathrm{n}-1\right)}}-1\right]\times\:100\mathrm{\%}$$\end{document} where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{c}}_{0}$$\end{document} represents the economic resilience level in the base period, \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:{\mathrm{c}}_{\mathrm{n}}$$\end{document} denotes the level of the given year, and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\:{\mathrm{g}}_{\mathrm{n}}\:$$\end{document} is the compound annual growth rate over n years. Logical connections between methods First, the CRITIC-TOPSIS method was used to construct the urban economic resilience index. Kernel density estimation was conducted to visualise the spatial dynamics of the resilience index, revealing distribution patterns, agglomeration cores, and evolutionary trends. Geographical detectors quantitatively screened the key driving factors behind the spatial differentiation of resilience. The resilience index was used as the dependent variable to measure the explanatory power of various factors and identify the core set of drivers. This not only helped explain the spatial patterns but also served as a preliminary screening step for the MGWR to avoid the inclusion of nonsignificant variables and ensure model simplicity. Multiscale geographically weighted regression (MGWR) was then used to refine the analysis by characterizing the local effects and spatial scales of each key driver. MGWR enables each variable’s relationship to vary by geographic location and estimates its own optimal bandwidth. Thus, how strongly each driving factor influences resilience in each specific city can be precisely quantified, thereby providing empirical support for differentiated regional policies. These four methods constitute a complete and rigorous logical analytical chain from “assessment (what)” to “description (where)”, then to “diagnosis (shy)”, and finally to “analysis (how).” The methods complement each other and progressively deepen the analysis, collectively revealing the comprehensive level, spatial pattern, underlying causes, and local operating mechanisms of urban economic resilience (see Fig. 1). Study area, data sources and descriptive statistics On the basis of the accessibility and feasibility of the data, 286 Chinese cities were selected as the basic spatial units for sampling (see Fig. 2). The main focus of the research is the resilience intensity demonstrated by cities in the face of internal and external shocks. Therefore, the research period ranges from 2008 to 2022. The data are mainly obtained from the China City Statistical Yearbook (2009–2023), ESP data, night light data, and statistical reports on the national economic and social development of each city. The linear interpolation method is used to complete the missing data of individual cities. The study presents descriptive statistics for the original indicators, including sample size, average value, standard deviation, and minimum and maximum values. (see Table 3). Fig. 2Study area. This map was created on the basis of the standard map from the Standard Map Service Website of the Ministry of Natural Resources of the People’s Republic of China (No. GS (2023)2767), without any modifications made to the boundaries of the base map. Study area. This map was created on the basis of the standard map from the Standard Map Service Website of the Ministry of Natural Resources of the People’s Republic of China (No. GS (2023)2767), without any modifications made to the boundaries of the base map. Table 3Descriptive statistics of the original data.Variable N MeanStandard deviationMinMaxx1429050,764.45934,426.60799467,749x2429025,686.63431,818.3081686600,501x3429056,659.4125,909.7584958221,497.6x442900.0740.0270.020.238x542900.2230.40205.717x642900.3720.10701.013x742900.9470.6240.0049.577x842901.4370.6840.37120.1x9429041.51510.13411.883.87x10429095,109.711169,396.54781,186,154x1142900.0030.00300.063x1242906662.81317,206.6610279,177x134290686.9142929.295061,966z1429013.3031.0895.46817.762z242900.5440.1580.1311.007z342900.4580.1110.1070.851z442909.6130.7317.15412.134z542908.1691.1823.43412.566z6429010.711.3686.98816.795This table presents descriptive statistics of the 13 indicators of economic resilience in 286 Chinese cities during the period of 2008–2022 and the 6 factors influencing economic resilience, including the names of the variables, sample numbers, average values, variance values, and minimum and maximum values. Table 3Descriptive statistics of the original data.Variable N MeanStandard deviationMinMaxx1429050,764.45934,426.60799467,749x2429025,686.63431,818.3081686600,501x3429056,659.4125,909.7584958221,497.6x442900.0740.0270.020.238x542900.2230.40205.717x642900.3720.10701.013x742900.9470.6240.0049.577x842901.4370.6840.37120.1x9429041.51510.13411.883.87x10429095,109.711169,396.54781,186,154x1142900.0030.00300.063x1242906662.81317,206.6610279,177x134290686.9142929.295061,966z1429013.3031.0895.46817.762z242900.5440.1580.1311.007z342900.4580.1110.1070.851z442909.6130.7317.15412.134z542908.1691.1823.43412.566z6429010.711.3686.98816.795This table presents descriptive statistics of the 13 indicators of economic resilience in 286 Chinese cities during the period of 2008–2022 and the 6 factors influencing economic resilience, including the names of the variables, sample numbers, average values, variance values, and minimum and maximum values. Descriptive statistics of the original data. This table presents descriptive statistics of the 13 indicators of economic resilience in 286 Chinese cities during the period of 2008–2022 and the 6 factors influencing economic resilience, including the names of the variables, sample numbers, average values, variance values, and minimum and maximum values. Overall level, multidimensional development, and dynamic evolution of urban economic resilience Overall level of urban economic resilience With the help of the ArcGIS natural breakpoint approach, the index was classified into five grades: low level, relatively low level, medium level, relatively high level, and high level. Through spatial visualization technology, the geographical spatial distribution of urban economic resilience was plotted to systematically reveal the spatial–temporal evolution characteristics in 2008, 2012, 2016 and 2020. As shown in Fig. 3, the average value of urban economic resilience increased steadily from 0.344 to 0.400 between 2008 and 2020, indicating the continuous increase in systemic risk resistance and adaptive capacity of China’s urban economies. Spatially, urban economic resilience followed a typical core-periphery agglomeration structure. Megacities, including Beijing, Shanghai, Guangzhou and Wuhan, functioned as high-resilience cores, with their resilience score rising from 0.417 to 0.531 at an annual growth rate of 1.73%. In comparison, peripheral large and medium-sized cities experienced mild growth, rising from 0.342 to 0.398, with an annual growth rate of 1.1%. Hierarchical differentiation was prominently observed across the urban samples. High-tier megacities maintained superior resilience endowments that were driven by solid economic bases and locational advantages. Upper-middle-level cities such as Tianjin and Hangzhou strengthened their crisis responsiveness capacity through technological innovation and industrial restructuring, with their share expanding from 3.85% to 10.14%. The medium-resilience cities, which are predominantly concentrated in eastern and southern China, were strongly constrained by regional economic gradients. Notably, the number of lower-tier cities increased from 20 to 83, highlighting the daunting pressure on such cities to improve their levels of resilience. In contrast, lagging cities, including Hegang and Heihe, suffer from infrastructural shortages, industrial rigidities and talent scarcity, resulting in their inferior shock resistance and post-disruption recovery capacity. Overall, China’s urban economic resilience presents pronounced hierarchical stratification and aggravated polarization. Persistent intercity disparities present considerable challenges to balanced regional development and refined urban governance. Fig. 3Development level and spatial distribution of economic resilience in Chinese cities. Created on the basis of the standard map from the Standard Map Service Website of the Ministry of Natural Resources of the People’s Republic of China (No. GS (2023)2767), with unaltered boundaries of the base map. Development level and spatial distribution of economic resilience in Chinese cities. Created on the basis of the standard map from the Standard Map Service Website of the Ministry of Natural Resources of the People’s Republic of China (No. GS (2023)2767), with unaltered boundaries of the base map. Multidimensional development of urban economic resilience In this paper, urban economic resilience is analysed from three dimensions: resistance, recovery, and innovation capacity (Fig. 4). The time series results indicate that all three dimensions steadily increased from 2009 to 2022. The average values of resistance, recovery, and innovation capacity rose from 0.442, 0.283, and 0.051 to 0.508, 0.381, and 0.093, with annual growth rates of 1.00%, 2.15%, and 4.39%, respectively. Such evolutionary trends reflect China’s sustained achievements in consolidating economic foundations, expanding market scale, and facilitating innovative transformation. Urban economic resilience presents a hierarchical gradient of “resistance > recovery > innovation”, while its growth momentum follows the reverse order of “innovation > recovery > resistance”. Metropolises, including Shanghai, Beijing and Shenzhen, exhibit prominent resistance capacities because of their solid economic foundations and superior average employment and income, which together construct a robust risk defence system. The recovery capacity displays a multicentred and hierarchical spatial pattern with evident north–south differentiation; most northern cities remain at low recovery levels, whereas the southern cities generally have medium-to-high recovery capacity. External shocks such as the financial crisis and COVID-19 accelerated resource reallocation and industrial restructuring during the sample period, which improved the urban recovery capacity through consumption expansion, investment optimization and industrial adjustment. Additionally, the capacity for urban innovation remains generally underdeveloped, following a typical core-periphery spatial pattern. Megacities such as Beijing, Shanghai and Chengdu constitute innovation agglomeration cores, whereas most other cities lag, reflecting distinct scale heterogeneity in innovation development. Therefore, constructing a complete innovation system driven by technological progress and high-end human resources is essential for urban economic stabilization, transformation and upgrading amid external disturbances. Fig. 4Spatial distribution of urban economic resilience development in China. Created on the basis of the standard map from the Standard Map Service Website of the Ministry of Natural Resources of the People’s Republic of China (No. GS (2023)2767), with unaltered boundaries of the base map. Spatial distribution of urban economic resilience development in China. Created on the basis of the standard map from the Standard Map Service Website of the Ministry of Natural Resources of the People’s Republic of China (No. GS (2023)2767), with unaltered boundaries of the base map. Dynamic evolution of urban economic resilience As presented in Fig. 5, the national kernel density curve of economic resilience displays obvious right-tailed characteristics and generally shifted rightward from 2008 to 2022, indicating a steady improvement in urban economic resilience. A declining peak height indicates a divergent developmental trend without stable convergence. After 2019, the curve changes from unimodal to bimodal, suggesting apparent stratification between high-resilience and low-resilience cities across China. Regionally, the eastern, western and northeastern regions show a fluctuating peak trajectory of “decline followed by growth”, which reflects an initial expansion and subsequent contraction of intraregional disparities. Conversely, the central region exhibits continuously increasing peaks with narrowing internal gaps. In terms of distribution extensibility, the right-tailed curves in the eastern and central regions denote prominent hierarchical differentiation, whereas the annually extended right tails in the western and northeastern regions reveal widening urban disparities. In terms of polarization features, the eastern and central regions consistently maintain unimodal distributions with mild internal gaps and high-level agglomeration. By comparison, the western and northeastern regions shift from unimodal to bimodal patterns, which indicates intensifying polarization and increasingly distinct stratification between the high-resilience and low-resilience city groups. Fig. 5Dynamic development and evolution of economic resilience levels in Chinese cities and the four major regions for 2008 to 2022. Dynamic development and evolution of economic resilience levels in Chinese cities and the four major regions for 2008 to 2022. Robustness test of the measurement results To test the robustness of the measurement results, an additional sensitivity test was conducted. The data normalization process was clarified using the min–max scaling method (Eqs. 1–2) to ensure comparability, after which the urban economic resilience index was recalculated using the entropy and coefficient of variation methods. The correlation coefficient between the resilience indices was derived from the CRITIC-TOPSIS, entropy methods, and over the fact that the coefficient of variation is greater than 0.8 (see Table 4), the coefficient is positive and it is statistically significant at the 5% level. Moreover, the spatial patterns remained highly consistent. These findings confirm that our primary findings are robust to the choice of weighting scheme. Table 4Pearson correlation coefficient across different methods.MethodMeanStandard deviationCRITIC-TOPSISEntropyCoefficient of variationCRITIC-TOPSIS0.3780.04210.804**0.916**Entropy0.2770.0540.804**10.861**Coefficient of variation0.2070.0420.916**0.861**1This table presents the Pearson correlation coefficients between the CRITIC-TOPSIS, entropy, and coefficient of variation methods. Table 4Pearson correlation coefficient across different methods.MethodMeanStandard deviationCRITIC-TOPSISEntropyCoefficient of variationCRITIC-TOPSIS0.3780.04210.804**0.916**Entropy0.2770.0540.804**10.861**Coefficient of variation0.2070.0420.916**0.861**1This table presents the Pearson correlation coefficients between the CRITIC-TOPSIS, entropy, and coefficient of variation methods. Pearson correlation coefficient across different methods. This table presents the Pearson correlation coefficients between the CRITIC-TOPSIS, entropy, and coefficient of variation methods. Analysis of the factors influencing urban economic resilience development Influencing determinants of urban economic resilience development The results, from a national perspective, are shown in Table 5. The q values of the medical (\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:q\:$$\end{document}= 0.483) and the digitalization levels ( q = 0.477) are relatively high, indicating that these two factors have the most significant influence on urban economic resilience. The transportation level ( q = 0.426) is next; the influence of the urbanization level ( q = 0.268) and traffic accessibility ( q = 0.218) is relatively weak, and the influence of the industrialization level ( q = 0.116) is the weakest. Judging from the q values of each factor, the degree of effect of every factor fluctuates across different periods. Table 5Detection results of the influencing factors on the development of urban economic resilience from 2008 to 2022.Variable2008–20122013–20172018–20222008–2022QRankQRankQRankqRankZ10.41810.41820.41820.4772Z20.18740.30140.30140.2684Z30.12460.13450.13450.1176Z40.40420.43010.43010.4831Z50.18650.11960.11960.2195Z60.28730.33830.33830.4273This table shows the degree of influence of digitalization, urbanization, industrialization, medical care level, transportation accessibility and logistics level on the economic resilience of 286 Chinese cities from 2008 to 2022, including the q values and rankings of the six influencing factors in the three stages of 2008–2012, 2013–2017 and 2018–2022. Table 5Detection results of the influencing factors on the development of urban economic resilience from 2008 to 2022.Variable2008–20122013–20172018–20222008–2022QRankQRankQRankqRankZ10.41810.41820.41820.4772Z20.18740.30140.30140.2684Z30.12460.13450.13450.1176Z40.40420.43010.43010.4831Z50.18650.11960.11960.2195Z60.28730.33830.33830.4273This table shows the degree of influence of digitalization, urbanization, industrialization, medical care level, transportation accessibility and logistics level on the economic resilience of 286 Chinese cities from 2008 to 2022, including the q values and rankings of the six influencing factors in the three stages of 2008–2012, 2013–2017 and 2018–2022. Detection results of the influencing factors on the development of urban economic resilience from 2008 to 2022. This table shows the degree of influence of digitalization, urbanization, industrialization, medical care level, transportation accessibility and logistics level on the economic resilience of 286 Chinese cities from 2008 to 2022, including the q values and rankings of the six influencing factors in the three stages of 2008–2012, 2013–2017 and 2018–2022. Among them, the medical level consistently ranks first across multiple periods, which highlights the fundamental role of medical and health resources in urban economic resilience. Digitalization is also ranked relatively highly, suggesting that the internet facilitates information transmission, technological innovation and industrial upgrades, thereby strengthening urban antishock capacity. Urbanization is fourth, indicating its moderate and stable influence on economic resilience. In contrast, industrialization and transportation have lower rankings in certain periods, indicating their relatively limited contributions to overall resilience construction. Nevertheless, such influencing effects vary across economic stages. For instance, the q value of industrialization increased from 2013 to 2017 but declined to 0.092 from 2018 to 2022. In the early stage, industrialization promoted industrial agglomeration and economic expansion to enhance resilience. Along with economic structural adjustment, the growing importance of the service industry and innovation-driven development gradually weakened the relative explanatory power of industrialization. Analysis of the multiscale geographically weighted regression (MGWR) results To explore the spatial heterogeneity of the driving factors, a multiscale geographically weighted regression (MGWR) model with six explanatory variables was adopted for this study. Although transportation level and traffic accessibility both reflect infrastructure conditions, the former characterizes logistics scale and transportation capacity, whereas the latter emphasizes personnel mobility and spatial connectivity. To eliminate estimation bias, the variance inflation factor (VIF) was applied for collinearity diagnosis. All the VIF values are lower than 7.5, which satisfies empirical criteria61. Moreover, the maximum correlation coefficient between variables is less than 0.5, which is below the critical threshold of severe multicollinearity (see Table 6). Hence, the selected variables are statistically valid. Table 6Pearson correlation coefficient between influence factors.Variablesz1z2z3z4z5z6z11z20.361z3-0.1660.0151z40.4960.168-0.1021z50.04-0.1080.1420.2431z60.4220.33-0.1850.3690.0281 Table 6Pearson correlation coefficient between influence factors.Variablesz1z2z3z4z5z6z11z20.361z3-0.1660.0151z40.4960.168-0.1021z50.04-0.1080.1420.2431z60.4220.33-0.1850.3690.0281 Pearson correlation coefficient between influence factors. Compared with global OLS regression, GWR and MGWR are better at capturing local spatial heterogeneity. Compared with OLS and conventional GWR, the MGWR has an R2 of 0.82 and a lower AICc value, which indicates better explanatory performance. Unlike the fixed bandwidth setting of the GWR (bandwidth = 112), the MGWR allows the variables to possess distinct optimal bandwidths. The local effects of urbanization, industrialization and traffic accessibility are sensitive, with bandwidths of 51, 56 and 165, respectively, indicating obvious spatial divergence regarding their influencing intensity. In contrast, the medical, digitalization and logistics levels share a bandwidth of 285, indicating their relatively homogeneous impacts over broader spatial scales. Table 7Multiscale geographically weighted regression model result summary.Model R 2 Adjusted R2AICc \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\sigma\:$$\end{document} 2 VariableBandwidthCoefficientVIFMeanMinMaxOLS0.4880.482620.894285GWR0.6100.602558.8050.674112MGWR0.8200.785436.6220.464Z2510.5540.2820.9951.587Z356− 0.588− 1.288− 0.1191.171Z42850.1090.1020.1334.385Z51650.030− 0.1210.1371.433Z62850.2310.2220.2554.772Z12850.1160.1090.1365.475C520.125− 0.3400.591This table represents the coefficients of the factors that influence urban economic resilience, including the regression results of the least squares method, weighted geographic regression, and multiscale weighted geographic model, as well as the goodness-of-fit value, AICc value, σ2 value, VIF value, and bandwidth equivalent values of each model. Table 7Multiscale geographically weighted regression model result summary.Model R 2 Adjusted R2AICc \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:\sigma\:$$\end{document} 2 VariableBandwidthCoefficientVIFMeanMinMaxOLS0.4880.482620.894285GWR0.6100.602558.8050.674112MGWR0.8200.785436.6220.464Z2510.5540.2820.9951.587Z356− 0.588− 1.288− 0.1191.171Z42850.1090.1020.1334.385Z51650.030− 0.1210.1371.433Z62850.2310.2220.2554.772Z12850.1160.1090.1365.475C520.125− 0.3400.591This table represents the coefficients of the factors that influence urban economic resilience, including the regression results of the least squares method, weighted geographic regression, and multiscale weighted geographic model, as well as the goodness-of-fit value, AICc value, σ2 value, VIF value, and bandwidth equivalent values of each model. Multiscale geographically weighted regression model result summary. This table represents the coefficients of the factors that influence urban economic resilience, including the regression results of the least squares method, weighted geographic regression, and multiscale weighted geographic model, as well as the goodness-of-fit value, AICc value, σ2 value, VIF value, and bandwidth equivalent values of each model. Spatial heterogeneity of the factors that influence urban economic resilience development In this study, the ArcGIS method is used to determine the area distribution of the factor coefficients of the six economic factors influencing urban economic resilience in 2021. The area distribution of the six economic factors is shown in the MGWR model. The results for the determinants shown in Fig. 6 reveal that the urbanization level, digitalization level, medical level, and transportation level are positively correlated with economic resilience across the country, indicating that these four determinants enhance economic resilience. Conversely, industrialization has an inverse relationship with resilience, indicating that reducing the degree of industrialization can increase resilience. Traffic accessibility has a two-way effect. These factors are characterized by spatial heterogeneity. Fig. 6Spatial distribution of the main factors that influence urban economic resilience development in 2021. Created on the basis of the standard map from the Standard Map Service Website of the Ministry of Natural Resources of the People’s Republic of China (No. GS (2023)2767), with unaltered boundaries of the base map. Spatial distribution of the main factors that influence urban economic resilience development in 2021. Created on the basis of the standard map from the Standard Map Service Website of the Ministry of Natural Resources of the People’s Republic of China (No. GS (2023)2767), with unaltered boundaries of the base map. Influence of medical level on urban economic resilience The results are shown in Fig. 6 (6 − 1). In 2021, the average influence coefficient of the medical level reached 0.218 across the country, which underscores the notion that the medical level plays a relatively large role in enhancing Chinese economic resilience. In terms of geographical space, the influence coefficient shows a regional distribution characteristic as it moves from east to west: “eastern region (0.213) < central region (0.217) < western region (0.22) < northeastern region (0.227).” Since the outbreak of the pandemic in 2020, the degree of impact of the urban medical level on Chinese economic resilience has significantly increased. The coefficient in the northeast was 6.6% greater than that in the east, which underscores the critical role of robust healthcare systems in stabilizing the economy. Influence of the digitalization level on urban economic resilience In 2021, the national average coefficient of the effect of digitalization on urban economic resilience was 0.072. The data presented in Fig. 6 (6 − 2) reveal a significant east‒west gradient. In the economically mature eastern region (e.g., Shanghai and Nanjing), the average coefficient was 0.064. In contrast, the coefficient increased to 0.073 in the central region, 0.078 in the southwest region and 0.084 in the northwest region. This means that a one-unit increase in the digitalization index had a 31% stronger effect on economic resilience in the northwest (0.084) than it did in the east (0.064), which highlights the higher marginal returns of digital investment in less developed western regions. Influence of the urbanization level on urban economic resilience In 2021, the average value coefficient of the urbanization level across countries reached 0.541, as shown in Fig. 6 (6 − 3), indicating that the urbanization process has an important positive effect on urban economic resilience. In terms of regional patterns, the influence of the urbanization level clearly reflects the spatial differentiation characteristics of the “centre-periphery” type. Among them, the “centre” region covers those cities in the eastern and central regions, while the “periphery” region includes the cities in the northeastern, northwestern, southeastern, and southwestern regions. The average influence coefficient of urbanization in the “center” region is 0.648, while that in the “periphery” areas is only 0.401. The reason for this is that the “center” region, with its superior geographical location, complete infrastructure system, and excellent business environment, has become the core area for the agglomeration of various economic elements. Influence of the transportation level on urban economic resilience In 2021, the national average coefficient for the transportation level was 0.150, as shown in Fig. 6 (6 − 4). The impact intensified from the east to the west, with coefficients ranging from as low as approximately 0.14 in the eastern coastal cities to a high of 0.159 in western cities such as Chengdu and Lanzhou. The 13.6% greater impact in the west confirms that efficient logistics systems are more binding constraints and more potent drivers of economic resilience in China’s interior regions, where traditional manufacturing and resource industries still dominate. Influence of traffic accessibility on urban economic resilience Traffic accessibility clearly demonstrated a dual-effect mechanism. Its coefficient was negative (− 0.06 on average) in highly developed provinces such as Shanghai, Jiangsu, and Guangdong, which shows a saturation effect where further investment can yield diminishing returns or even resource misallocation. Conversely, in less developed areas, the coefficient was positive (0.076 on average). The absolute difference of 0.136 between these two contexts quantitatively captures this pivotal shift from a positive enabler to a potential liability, guiding radically different infrastructure investment strategies. Influence of the industrialization level on urban economic resilience The national average coefficient for industrialization was − 0.600, indicating a substantial negative correlation. The impact was more pronounced in the south, with coefficients reaching as low as − 1.288 in some highly industrialized cities, compared with a milder negative impact in the north. This more than twofold difference in the magnitude of the negative effect for southern cities underscores the severe challenges that they face in terms of path dependence and global supply chain vulnerabilities. Robustness test In this study, an econometric model was constructed using a two-way fixed effects panel data model to identify the specific effects of various factors on urban economic resilience. The regression results presented in Table 8 reveal that the digitalization level, medical service level, urbanization level, and logistics capacity all have significant positive effects on urban economic resilience. In contrast, the industrialization level and transportation accessibility have significant negative effects. These empirical findings are largely consistent with the earlier descriptive analysis, thereby enhancing the reliability and practical interpretability of the findings of this study. Table 8Empirical results of urban economic resilience.VariablesResilienceResilienceResilienceResilienceResilienceResilienceDigitalization0.054** (25.068)Urbanization0.215** (25.203)Industrialization− 0.128** (− 16.828)Traffic accessibility− 0.017** (− 26.17)Transportation level0.021*** (34.978)Medical level0.075*** (34.25)R20.5270.4690.4160.3710.4280.564N426242624262426242624262Year fixed effectYesYesYesYesYesYesProvince fixed effectYesYesYesYesYesYesThis table reports the regression results of Eq. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively, with t values displayed in parentheses. Table 8Empirical results of urban economic resilience.VariablesResilienceResilienceResilienceResilienceResilienceResilienceDigitalization0.054** (25.068)Urbanization0.215** (25.203)Industrialization− 0.128** (− 16.828)Traffic accessibility− 0.017** (− 26.17)Transportation level0.021*** (34.978)Medical level0.075*** (34.25)R20.5270.4690.4160.3710.4280.564N426242624262426242624262Year fixed effectYesYesYesYesYesYesProvince fixed effectYesYesYesYesYesYesThis table reports the regression results of Eq. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively, with t values displayed in parentheses. Empirical results of urban economic resilience. This table reports the regression results of Eq. *, **, and *** indicate significance at the 10%, 5%, and 1% levels, respectively, with t values displayed in parentheses. Summary of the key characteristics of urban economic resilience On the basis of the above analysis, the key findings of this study are summarised from five aspects: overall development level, multitiered structure, dimensional differences, regional dynamics, and main influencing factors. The national average resilience index increased from 0.344 to 0.400. Megacities and large cities have demonstrated high-resilience, with the proportion of such cities rising from 0.35% to 8.04%. Regional dynamics indicate that the eastern and central regions exhibit high-level agglomeration with narrowing internal disparities, whereas the western and northeastern regions show “high–low” polarization with widening internal disparities (bimodal distribution). The medical and digitalization levels are the most important factors, and the impact coefficient of digitalization displays spatial heterogeneity (see Table 9) Table 9Summary of key characteristics of urban economic resilience in China (2008–2022).Feature dimensionKey findingSpecific contentOverall development levelNational average resilience indexIncreased from 0.344 (2008) to 0.400 (2022)Multitiered structureHigh-resilience citiesConcentrated in megacities and large cities (e.g., Beijing, Shanghai, Guangzhou), proportion increased from 0.35% to 8.04%.Low-resilience citiesMostly small and medium-sized cities, constrained by weak infrastructure and lagging industrial transformation (e.g., Hegang, Heihe)Dimensional differencesAbsolute scale rankingResistance capacity (0.482) > Recovery Capacity (0.342) > Innovation Capacity (0.07).Growth momentum rankingInnovation Capacity (4.39%) > Recovery Capacity (2.15%) > Resistance Capacity (1.0%) (Average Annual Growth Rate)Regional dynamicsEastern and central regionsHigh-level agglomeration, internal differences narrowing (Kernel density curves show a unimodal shape)Western and northeastern regions“High-Low” polarization, internal differences expanding (Kernel density curves show a bimodal shape)Main influencing factorsMost influential factorsMedical Level (q = 0.483), Digitalization Level (q = 0.477)Example of spatial heterogeneityDigitalization Impact Coefficient: Eastern 0.064 vs. Western 0.084This table presents the main findings on urban economic resilience during the 2008–2022 study period. “Megacities” refer to cities with a resident population in excess of 10 million, while “small and medium-sized cities” refer to those with a resident population of less than 3 million. Table 9Summary of key characteristics of urban economic resilience in China (2008–2022).Feature dimensionKey findingSpecific contentOverall development levelNational average resilience indexIncreased from 0.344 (2008) to 0.400 (2022)Multitiered structureHigh-resilience citiesConcentrated in megacities and large cities (e.g., Beijing, Shanghai, Guangzhou), proportion increased from 0.35% to 8.04%.Low-resilience citiesMostly small and medium-sized cities, constrained by weak infrastructure and lagging industrial transformation (e.g., Hegang, Heihe)Dimensional differencesAbsolute scale rankingResistance capacity (0.482) > Recovery Capacity (0.342) > Innovation Capacity (0.07).Growth momentum rankingInnovation Capacity (4.39%) > Recovery Capacity (2.15%) > Resistance Capacity (1.0%) (Average Annual Growth Rate)Regional dynamicsEastern and central regionsHigh-level agglomeration, internal differences narrowing (Kernel density curves show a unimodal shape)Western and northeastern regions“High-Low” polarization, internal differences expanding (Kernel density curves show a bimodal shape)Main influencing factorsMost influential factorsMedical Level (q = 0.483), Digitalization Level (q = 0.477)Example of spatial heterogeneityDigitalization Impact Coefficient: Eastern 0.064 vs. Western 0.084This table presents the main findings on urban economic resilience during the 2008–2022 study period. “Megacities” refer to cities with a resident population in excess of 10 million, while “small and medium-sized cities” refer to those with a resident population of less than 3 million. Summary of key characteristics of urban economic resilience in China (2008–2022). This table presents the main findings on urban economic resilience during the 2008–2022 study period. “Megacities” refer to cities with a resident population in excess of 10 million, while “small and medium-sized cities” refer to those with a resident population of less than 3 million. Conclusions and discussions Discussion On the basis of resilience theory and the literature, economic resilience is conceptualised in this study into three dimensions—resistance, recovery, and innovation capacity—with a focus on the meso-level perspective of cities. Constructing a three-dimensional evaluation index system of “resistance–recovery–innovation” enables the classification of cities into different tiers. In this study, the overall development level and evolutionary trends of each dimension of urban economic resilience is systematically analysed, the strengths and weaknesses of different cities in terms of resilience development are revealed, and the spatiotemporal evolution process is fully presented. Furthermore, this study incorporates agglomeration factors, digital economy development, infrastructure, and other elements into the driving factor system of urban economic resilience. The varying impacts of these factors on the resilience of cities of different sizes are empirically examined to strengthen the explanation of the underlying causes of spatial heterogeneity in urban economic resilience levels. This study has several limitations. On the one hand, owing to constraints such as data availability, the influencing factors included may not comprehensively cover all of the relevant variables, which could affect the completeness of the model to some extent. On the other hand, the model constructed on the basis of the current data has limited applicability to international cities, and its generalizability requires further validation. Future research should be focused on expanding the data sources, improving the variable system, and exploring the method’s applicability across different contexts and international cities to enhance the comprehensiveness and broader relevance of the study. Conclusions and policy recommendations In this study, the level, dimensional structure, spatial distribution, and drivers of urban economic resilience in China (2008–2022) are analysed. The main conclusions are as follows. First, urban economic resilience steadily increased from 0.344 to 0.400, with temporary slowdowns caused by the 2009 financial crisis and the COVID-19 pandemic. Second, compared with small cities, large cities have greater resilience. The dimensional hierarchy is “resistance > recovery > innovation”, while that of growth momentum follows a “innovation > recovery > resistance” pathway, indicating that innovation is becoming the key driver of long-term resilience. Third, the spatial distribution shows a dual polarization between high- and low-resilience cities. Eastern and central China exhibit high-level agglomeration with shrinking disparities, whereas the western and northeastern regions face widening gaps. Fourth, six spatially heterogeneous drivers are identified. Digitalization, medical resources, and transportation have positive effects, with increasing influence from east to west; urbanization has positive north–south differentiation; industrialization has negative effects; and traffic accessibility has bidirectional effects. On the basis of the above MGWR empirical results, targeted policy recommendations for the improvement of urban economic resilience are proposed in this section from four perspectives. First, given the higher digitalization coefficients in central and western China, local governments should strengthen digital infrastructure construction and promote industrial digital transformation to leverage digital dividends. Second, prioritized medical investment is needed in the northeastern and western regions because of the dominant positive effect of medical resources, particularly for ageing northeastern cities. Third, differentiated transportation strategies should be implemented: Eastern megacities need intelligent traffic optimization, whereas Central and Western cities require continuous transportation investment for the promotion of regional integration. Fourth, given the negative industrialization effect, northern resource-based cities should pursue industrial diversification, whereas southern industrialized cities should increase innovative investment to reduce industrial lock-in risk. These differentiated optimization paths can provide practical references for the coordinated development of resilience in heterogeneous cities. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 1