Evaluating Urban Economic Resilience in the Face of Major Public Health Emergencies: A Spatiotemporal Analysis

Abstract

The COVID-19 pandemic severely impacted China’s economic stability. This study assesses the economic resilience of 2843 Chinese counties from 2019 to 2021 by constructing a comprehensive evaluation index system. Using the Projection Pursuit Model to generate index weights, we analyze resilience across four key dimensions: resistance, stress, recovery, and innovation. Our analysis reveals that urban economic resilience first declined during the pandemic’s peak before recovering in 2021. Spatially, eastern coastal regions demonstrated stronger resilience, supported by robust infrastructure, advanced industries, and flexible markets. In contrast, central and western regions were less resilient due to their reliance on traditional industries. A deeper sub-dimensional analysis showed that eastern regions consistently outperformed the west across all four metrics. This research establishes a rigorous framework for evaluating urban economic resilience and offers targeted strategies for policymakers to build more resilient cities in the face of future public health emergencies.

1. Introduction

The COVID-19 pandemic triggered one of the most severe global economic contractions in modern history. According to the International Monetary Fund and the World Bank, global GDP declined by approximately 3.0% to 5.2% in 2020, culminating in estimated cumulative output losses of nearly USD 9 trillion between 2020 and 2021 [1,2]. This downturn severely impacted international trade, with merchandise trade volumes contracting by 5–8% and commercial services collapsing by over 20%, primarily due to the suspension of tourism and transport [3,4]. Scholarly consensus underscores the heterogeneous impacts across nations and sectors, emphasizing the critical role of containment measures and countercyclical policies in shaping recovery trajectories [5,6].

Against this global backdrop, China presented a notable case of rapid recovery. Following initial significant disruptions to social and economic activities [7,8], the Chinese economy rebounded steadily. By 2023, its GDP had grown to CNY 129.43 trillion, a 30% increase from 2019, highlighting remarkable resilience [9,10]. However, this aggregate recovery masks substantial disparities at the sub-national level. Evidence suggests that the pace and success of economic recovery varied significantly among Chinese cities, with some achieving transformative growth while others struggled to overcome the initial shock [11,12]. This heterogeneity reveals a critical research gap: the need to systematically quantify these divergent urban economic resilience pathways and to develop targeted optimization strategies.

Despite the acknowledged importance of urban economic resilience in modern development theory [13,14], current research, particularly in the Chinese context, faces two primary limitations. First, domestic studies on urban economic resilience are still in their infancy. Second, and more critically, there is a lack of integration of emerging multidimensional big data—such as internet activity, traffic flows, and night-time light data—into the analytical framework [15]. These data sources offer unprecedented, high-resolution insights into real-time urban dynamics across sectors like healthcare, transportation, and commerce [15,16]. Therefore, the central research problem is the absence of a fine-grained, empirically grounded evaluation system for urban economic resilience in China that leverages these novel data streams. This gap impedes an accurate assessment of resilience and the formulation of effective, context-specific policies.

To address this problem, this study aims to achieve three clear objectives: (1) to conceptualize urban economic resilience through the four dimensions of resistance, stress, resilience, and reorientation; (2) to construct a comprehensive rating system for China’s urban economic resilience by integrating traditional social statistics with multidimensional big data (e.g., traffic, night-time lights) for 2019–2021; and (3) to propose targeted optimization strategies for different types of Chinese cities based on the evaluation results.

The justification for this research is twofold. Theoretically, it contributes to the literature on resilient cities by advancing the methodological integration of big data into economic resilience evaluation, thereby enriching the transition from theory to practice. Practically, the findings are intended to provide valuable insights for policymakers to enhance urban preparedness and response capabilities against future major public health emergencies and other systemic shocks.

2. Literature Review

2.1. Conceptual Evolution and Indicator Framework of Urban (Economic) Resilience

The theoretical foundation of urban resilience is derived from the ecological concept of system resilience [17,18]. This framework, now central to urban planning and regional economics, defines resilience as a system’s capacity to withstand, adapt to, and recover from shocks and stresses [19].

Academics generally agree on the multidimensional nature of resilience, encompassing four core capacities: resistance, recovery, adaptivity, and transformability or creativity [13,18]. Indicator frameworks have rapidly developed across areas like disaster management and economic shocks [20,21], with international bodies such as the OECD proposing structured dimensions including governance, infrastructure, economic structure, and social capital [22].

The primary gap in the literature lies in spatial resolution. The bulk of existing resilience studies are concentrated at the macro level, typically focusing on national, provincial, or metropolitan scales [13,23]. Consequently, there is a notable deficit in systematic, fine-grained resilience assessment at the county level (or sub-regional scale). While county-level studies have begun to emerge [24], a comprehensive, nationwide, and fine-scale measurement framework remains an unmet necessity, forming a key point of departure for the present study.

2.2. Common Weighting Methods for Resilience Assessment and Their Limitations

Constructing a composite resilience index requires robust methods for indicator weighting and aggregation. While common techniques like Principal Component Analysis (PCA), the Entropy Weight Method, and the Analytic Hierarchy Process (AHP) are widely used [25,26], they face inherent limitations. Objective methods (e.g., PCA and Entropy) rely heavily on linear assumptions, often failing to capture the complex non-linear relationships characteristic of urban systems. Conversely, subjective methods (e.g., AHP), while integrating expert knowledge, are susceptible to subjective bias, which compromises the generalizability of the findings. Given the multidimensional and intricate nature of urban systems, these traditional methods often struggle to fully reveal the latent structure within the indicators, underscoring the necessity for more flexible, non-linear analytical methods to achieve higher measurement accuracy [25].

2.3. Application of the Projection Pursuit Model (PPM) in Comprehensive Evaluation

The Projection Pursuit Model (PPM), introduced by Friedman and Tukey (1974), is designed to find the optimal low-dimensional projection in high-dimensional space that best reveals data structure [27]. The PPM’s effectiveness in handling complex data has led to its application in comprehensive evaluations across areas such as ecological quality [28], regional development [29], and disaster resilience [30].

The PPM offers significant advantages over PCA and the Entropy Weight Method: it effectively processes high-dimensional and non-linear indicator systems, minimizes subjective bias, and extracts more informative projection directions [29]. Despite these benefits, the application of PPM to urban economic resilience, particularly at the county level, is still limited. Most related studies default to PCA or the Entropy Weight Method. Therefore, employing the PPM for the weighting and aggregation of county-level resilience indicators constitutes a crucial methodological innovation of this research.

2.4. The Theoretical Status and Research Gap of “Stress Response Capacity”

Stress Response Capacity (SRC), or emergency capability, is rooted in ecology and emergency management [17,31]. It emphasizes a system’s immediate reaction speed and absorptive capacity at the moment a shock occurs. In urban resilience, SRC includes fiscal emergency reserves, rapid public health system response, and the temporary substitution of critical functions [32,33].

However, a critical limitation of the prevailing urban economic resilience paradigm is its disproportionate focus on mid-to-long-term structural capacity (e.g., diversification, innovation) and post-shock recovery [15,21]. This orientation results in an insufficient characterization of the immediate emergency response and buffering capability during the shock phase. Therefore, including Stress Response Capacity as an explicit and distinct dimension enhances the process understanding and mechanistic explanatory power of resilience assessment, thereby allowing for a more holistic reflection of a region’s coping abilities across the shock timeline.

3. Research Methodology and Design

3.1. Study Scope, Period, and Unit of Analysis

This study utilizes a comprehensive dataset of 2843 county-level administrative regions across 31 provinces in China, excluding Hong Kong, Macao, and Taiwan. The selection of the county-level administrative scale is justified because it represents a fundamental and relatively independent unit of China’s administrative division and urban governance, providing a more granular and accurate reflection of economic dynamics [17,18,19,20]. The research period spans from 2019 to 2021 to capture the impact of the initial COVID-19 pandemic outbreak and the subsequent economic recovery following the lifting of control measures in 2021. The geographical scope of this study is illustrated in Figure 1.

3.2. Data Sources and Observation Materials

The data for this study is compiled from a variety of sources to ensure a robust analysis. Point of Interest (POI) data, used as a proxy for urban activity, is sourced from Amap1. Night light data, a widely used indicator of economic activity and urbanization, is derived from the DMSP-OLS and SNPP-VIIRS integrated dataset [20]. Industry data is obtained from the enterprise registration website, while enterprise data—including addresses, registration times, and types—is from the China Industrial and Commercial Enterprise Registration Data2 for 2019–2021 [22,23]. Urban characteristics data, such as regional economic aggregates, patent statistics, and scientific investment, are drawn from the China Urban Statistical Yearbook3. Epidemic infection and patent data come from a government statistics website4. Finally, geospatial data, including administrative boundaries and urban administrative centers, are provided by the National Basic Geographic Information Center’s 1:1,000,000 national basic geographic database5. This multi-sourced dataset allows for a multifaceted examination of urban economic resilience by integrating data on economic activity, demographics, business health, and governance.

4. Research Method and Procedure

4.1. Construction of the Evaluation Index System

Urban economic resilience is a circular transmission process where economic shocks or recoveries can change the resilience of cities, which in turn feeds back to the city’s economic system, affecting its robustness and sustainability. Based on the concept of economic resilience and combined with the connotation of urban resilience [25,26,27], this study takes acceptability, applicability, evaluability, and availability as the basic principles and combines the above four stages. Resistance, stress, resilience, and creativity are selected as core indicators, and the projected weight values of the Projection Pursuit Model are integrated to quantify the weights of each indicator from multiple perspectives, including average, scientific, and simplicity, thereby constructing an evaluation system for urban economic resilience in China [28,29,30,31]. The evaluation system aims to comprehensively reflect all aspects of urban economic resilience in China through multidimensional indicators. The specific index system is shown in

Table 1. Index system and quantitative methods for the evaluation of urban economic resilience in China.

Target Layer	Dimension Level	Index Level	Index Calculation and Meaning	Projected Weight Value	Weight
Urban economic resilience	Resistance index	Industrial diversity	$$\begin{array}{c}\mathrm{H}\prime =-{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}}{\mathrm{p}}_{\mathrm{i}}\ln ({\mathrm{p}}_{\mathrm{i}})\end{array}$$ (1) $\mathrm{H}\prime$ is the industry diversification index, ${\mathrm{p}}_{\mathrm{i}}$ is the proportion of the i-th enterprise type, and n is the total enterprise type.	0.65	0.089
		Core industry contribution ratio	$$\begin{array}{c}\mathrm{CR}={\displaystyle \frac{{\mathrm{P}}_{\mathrm{core}}}{{\mathrm{P}}_{\mathrm{total}}}}\end{array}$$ (2) CR is the contribution rate of the core industry, ${\mathrm{P}}_{\mathrm{core}}$ is the number of enterprise registrations in the end year of the core industry, and ${\mathrm{P}}_{\mathrm{total}}$ is the number of enterprise registrations in the end year of all industries.	0.42	0.059
		Economic vitality	$$\begin{array}{c}{\mathrm{DN}}_{\mathrm{mean}}={\displaystyle \frac{{\mathrm{DN}}_{\mathrm{total}}}{{\mathrm{N}}_{\mathrm{grid}}}}\end{array}$$ (3) ${\mathrm{DN}}_{\mathrm{mean}}$ is the average DN of the city’s night light brightness, which represents economic vitality, ${\mathrm{DN}}_{\mathrm{total}}$ is the sum of the DN values of all the light grids in the city area, and ${\mathrm{N}}_{\mathrm{grid}}$ is the number of grids in the city area.	0.40	0.059
		Regional centrality	$$\begin{array}{c}{\mathrm{BC}}_{\mathrm{i}}={\displaystyle \frac{2}{(\mathrm{N}-1)(\mathrm{N}-2)}}{\displaystyle \sum _{\mathrm{s}=1}^{\mathrm{N}}}{\displaystyle \sum _{\mathrm{t}=1}^{\mathrm{N}}}{\displaystyle \frac{{\mathsf{\delta }}_{\mathrm{s}\mathrm{t}}\left(\mathrm{i}\right)}{{\mathsf{\delta }}_{\mathrm{s}\mathrm{t}}}}\end{array}$$ (4) ${\mathrm{BC}}_{\mathrm{i}}$ is the intermediate centrality of city i in a complex urban network, $\mathrm{N}$ represents the number of city nodes in the urban network, ${\mathsf{\delta }}_{\mathrm{s}\mathrm{t}}\left(\mathrm{i}\right)$ represents the number of shortest paths from node s to node t through city node i, and ${\mathsf{\delta }}_{\mathrm{s}\mathrm{t}}$ represents the number of shortest paths from city node s to city node t.	0.22	0.030
	Stress index	Early warning accuracy	$$\begin{array}{c}\mathrm{f}(\mathrm{X})={\displaystyle \frac{1}{\mathrm{B}}}{\displaystyle \sum _{\mathrm{b}=1}^{\mathrm{B}}}{\mathrm{T}}_{\mathrm{b}}(\mathrm{X})\end{array}$$ (5) f(X) is the number of infections predicted by the random forest model, and X is the feature vector, including the impact level, the number of urban population, the number of urban beds, and the proportion of the tertiary industry. B is the number of decision trees (4). Tb (X) is the prediction result of the decision tree b.	0.66	0.089
		Emergency cure rate	$$\begin{array}{c}{\mathrm{R}}_{\mathrm{k}}={\displaystyle \frac{{\sum }_{\mathrm{t}={\mathrm{t}}_{\mathrm{k}-1}}^{{\mathrm{t}}_{\mathrm{k}}}{\text{number of cured cases}}_{\mathrm{t}}}{{\sum }_{\mathrm{t}={\mathrm{t}}_{\mathrm{k}-1}}^{{\mathrm{t}}_{\mathrm{k}}}\text{number of added cured case}}}\end{array}$$ (6) ${\mathrm{R}}_{\mathrm{k}}$ represents the emergency cure rate.	0.75	0.109
		Stress stability	$$\begin{array}{c}\mathrm{D}={\displaystyle \frac{2{\displaystyle \sum _{(\mathrm{u},\mathrm{v})\in \mathrm{E}}}\mathrm{w}(\mathrm{u},\mathrm{v})}{\mathrm{n}(\mathrm{n}-1)}}\end{array}$$ (7) $$\begin{array}{c}{\mathrm{D}}^{\prime }={\displaystyle \frac{2{\displaystyle \sum _{(\mathrm{u},\mathrm{v})\in {\mathrm{E}}^{\prime }}}\mathrm{w}(\mathrm{u},\mathrm{v})}{(\mathrm{n}-1)(\mathrm{n}-2)}}\end{array}$$ (8) $$\begin{array}{c}\mathrm{\Delta }{\mathrm{D}}_{\mathrm{i}}=\mathrm{D}-{\mathrm{D}}^{\prime }\end{array}$$ (9) D is the original network weighted density, ${\displaystyle \sum _{(\mathrm{u},\mathrm{v})\in \mathrm{E}}}\mathrm{w}(\mathrm{u},\mathrm{v})$ is the sum of all edge weights, E is the edge set, $\mathrm{w}(\mathrm{u},\mathrm{v})$ is the weight of edge $(\mathrm{u},\mathrm{v})$, and n is the number of nodes. ${\mathrm{D}}^{\prime }$ is the weighted density of the new network after the node is removed. ${\displaystyle \sum _{(\mathrm{u},\mathrm{v})\in {\mathrm{E}}^{\prime }}}\mathrm{w}(\mathrm{u},\mathrm{v})$ is the sum of the weights of all sides after the node is removed. ${\mathrm{E}}^{\prime }$ is the edge set after the removal of nodes, n − 1 is the number of nodes after the removal of a node, and $\mathsf{\Delta }{\mathrm{D}}_{\mathrm{i}}$ represents the stress stability of the city.	0.64	0.089
		Amplified infection rate	(1) Calculation of maximum influence scope is used to calculate the maximum influence scope of each node and set the direct influence of the node’s GDP on its influence. The formula is as follows: $$\begin{array}{c}\mathrm{M}(\mathrm{i})={\displaystyle \frac{{\mathrm{GDP}}_{\mathrm{i}}-{\mathrm{GDP}}_{\min }}{{\mathrm{GDP}}_{\max }-{\mathrm{GDP}}_{\min }}}\times \left({\mathrm{N}}_{\max }-{\mathrm{N}}_{\min }\right)+{\mathrm{N}}_{\min }\end{array}$$ (10) ${\mathrm{N}}_{\min }=1$ and ${\mathrm{N}}_{\max }=15.$ This reflects the number of cities that city nodes may affect in the network. (2) For snowball effect simulation propagation, one must calculate the total influence of nodes in the propagation process. The formula is as follows: $$\begin{array}{c}\mathrm{S}\left(\mathrm{i}\right)={\displaystyle {\displaystyle \sum }_{\mathrm{j}\in {\mathrm{T}}_{\mathrm{t}}}}{\mathrm{GDP}}_{\mathrm{j}}\end{array}$$ (11) $\mathrm{S}\left(\mathrm{i}\right)$ represents the cumulative GDP value of infected nodes and quantifies the snowball effect of nodes. ${\mathrm{GDP}}_{\mathrm{j}}$ represents the total GDP of city j in the propagation process. By setting the maximum number of propagation steps t = 30, the propagation range and influence degree of nodes in the network can be observed. (3) Normalization: Formula (12) is used to normalize the snowball effect in order to compare the effects between different nodes: $$\begin{array}{c}{\mathrm{S}}^{\wedge }(\mathrm{i})={\displaystyle \frac{\mathrm{S}\left(\mathrm{i}\right)-{\mathrm{S}}_{\min }}{{\mathrm{S}}_{\max }-{\mathrm{S}}_{\min }}}\end{array}$$ (12)	0.56	0.080
	Resilience index	Economic resilience	$$\begin{array}{c}\mathrm{FYi}={\displaystyle \frac{\text{Total GDP before impact}}{\text{Total GDP after impact}}}\end{array}$$ (13) FYi represents the city’s economic resilience in year i and pre-impact. GDP represents the city’s total economic income in the year prior to the pandemic impact. The GDP of the post-impact city indicates the total economic income of the city in year i of the pandemic impact.	0.50	0.069
		Global approach efficiency	$$\begin{array}{c}{\mathrm{E}}_{\mathrm{glo}}={\displaystyle \frac{1}{\mathrm{N}(\mathrm{N}-1)}}{\displaystyle \sum _{\mathrm{s}\ne \mathrm{t}}}{\mathsf{ϵ}}_{\mathrm{st}}={\displaystyle \frac{1}{\mathrm{N}(\mathrm{N}-1)}}{\displaystyle \sum _{\mathrm{s}\ne \mathrm{t}}}{\displaystyle \frac{1}{{\mathrm{d}}_{\mathrm{st}}}}\end{array}$$ (14) $$\begin{array}{c}{\mathrm{E}}_{\mathrm{loc}}={\displaystyle \frac{1}{\mathrm{N}}}{\displaystyle \sum _{\mathrm{w}\in \mathrm{G}}}\mathrm{E}({\mathrm{G}}_{\mathrm{w}})\end{array}$$ (15) ${\mathrm{G}}_{\mathrm{w}}$ represents the subgraph formed by adjacent nodes of node w. Among them, the calculation method of $\mathrm{E}\left({\mathrm{G}}_{\mathrm{w}}\right)$ and the calculation method of ${\mathrm{G}}_{\mathrm{w}}$ are the same.	0.28	0.040
		Road density	The road density is calculated as the road length within the unit area of each county-level city. The roads used are the superposition summaries of rural roads, national roads, and highways of all levels in OSM data.	0.22	0.030
		Attack scenario stability	$$\begin{array}{c}\mathrm{GCR}={\displaystyle \frac{\left|\mathrm{largest}\text{\_}\mathrm{cc}\right|}{\left|\mathrm{graph}\right|}}\end{array}$$ (16) $$\begin{array}{c}\hspace{1em}\mathrm{ASPL}={\displaystyle \frac{{\displaystyle \sum _{\mathrm{i},\mathrm{j}\in \mathrm{largest}\backslash \text{\_}\mathrm{cc}}}\mathrm{d}(\mathrm{i},\mathrm{j})}{\left|\mathrm{largest}\text{\_}\mathrm{cc}\right|\cdot \left(\right|\mathrm{largest}\text{\_}\mathrm{cc}|-1)}}\end{array}$$ (17) $$\begin{array}{c}\mathrm{Efficiency}={\displaystyle \frac{1}{\left|\mathrm{graph}\right|\cdot \left(\right|\mathrm{graph}|-1)}}{\displaystyle \sum _{\mathrm{i}\ne \mathrm{j}}}{\displaystyle \frac{1}{\mathrm{d}(\mathrm{i},\mathrm{j})}}\end{array}$$ (18) $$\text{Composite score}=\mathrm{GDP}\ast (0.4\times \mathrm{GCR}+0.3\times (1-\mathrm{ASPL})+0.3\times \mathrm{Efficiency})$$ (19) $\mathrm{GCR}$ is the proportion of the giant connected component; $\mathrm{ASPL}$ is the average shortest path length; $\mathrm{Efficiency}$ is the network efficiency; the Composite Score is the stability of the attack scenario; $\mathrm{largest}\text{\_}\mathrm{cc}$ is the number of nodes in the maximum connected component; $\mathrm{graph}$ is the total number of nodes in the county; and GDP is the economic total of the county.	0.09	0.010
	Innovation ability index	Degree of technological innovation	$$\begin{array}{c}{\mathrm{H}}_{\mathrm{i}}=-{\displaystyle \frac{1}{\log \left(\mathrm{n}\right)}}{\displaystyle \sum _{\mathrm{j}=1}^{\mathrm{n}}}{\mathrm{p}}_{\mathrm{ij}}\log ({\mathrm{p}}_{\mathrm{ij}})\end{array}$$ (20) $$\begin{array}{c}{\mathrm{w}}_{\mathrm{i}}={\displaystyle \frac{1-{\mathrm{H}}_{\mathrm{i}}}{{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{m}}}(1-{\mathrm{H}}_{\mathrm{i}})}}\end{array}$$ (21) $$\begin{array}{c}{\mathrm{S}}_{\mathrm{j}}={\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{m}}}{\mathrm{w}}_{\mathrm{i}}\cdot {\mathrm{X}}_{\mathrm{ij}}^{\prime }\end{array}$$ (22) ${\mathrm{H}}_{\mathrm{i}}$ is the information entropy of the i city, ${\mathrm{w}}_{\mathrm{i}}$ is the index weight, and ${\mathrm{S}}_{\mathrm{j}}$ is the degree of scientific and technological innovation of the county-level urban area.	0.83	0.119
		Patent application activity	We collect and analyze the number of patent applications and grants for each year, evaluate them using innovation output indicators, and calculate the number of regional patent applications.	0.39	0.059
		Nuclear density of high-tech enterprises	$\mathrm{f}(\mathrm{x},\mathrm{y})$ is the kernel density value of high-tech enterprises, $\mathrm{h}$ is the search radius, ${\mathrm{x}}_{\mathrm{i}}\mathrm{and}{\mathrm{y}}_{\mathrm{i}}$ and ${\mathrm{x}}_{\mathrm{i}}\mathrm{and}{\mathrm{y}}_{\mathrm{i}}$ are the coordinates of interest points within the search radius, $\mathrm{n}$ is the number of high-tech enterprises within the search radius, $\mathrm{x}$ and y are the coordinates of the center of the grid pixel, and the search radius is 10 km.	0.53	0.069

.

4.1.1. Resistance Index

This paper selects industry diversification index, core industry contribution ratio, economic vitality, and regional centrality as the main indicators through which to measure resistance. The industry diversification index reflects the degree of diversification of different types and quantities of industries in an economy, embodying the robustness of the urban economic structure [32,33,34]. The core industry contribution ratio refers to the proportion of a specific industry in the total economic output of a region or country [35], evaluating the degree of urban economy’s dependence on core industries and revealing the potential risk of economic shock resistance [36]. Economic vitality represents the degree to which a city supports different resources and development factors such as its own function, economy, and society, reflecting the ability and potential of urban development to a certain extent [37,38]. Regional centrality represents the centrality of regional network nodes, with urban nodes within the network having smoother flow of factors, closer industrial division of labor and cooperation, and more frequent policy exchanges [39,40,41]. Therefore, cities associated with core nodes can share production costs and act as organizational parts of the industrial system, resolving the negative impact on the whole and showing strong resistance resilience [42,43].

4.1.2. Stress Index

Early warning accuracy, emergency cure rate, emergency stability, and amplified infection rate are selected as the main indexes to measure stress. Early warning accuracy refers to the ability of an early warning system to accurately predict and forecast the occurrence of an emergency before it happens [44]. A highly accurate early warning system can enhance a city’s economic resilience by improving response efficiency, optimizing resource allocation, and enhancing public confidence [45]. The emergency cure rate measures the processing ability and recovery speed of a system when it suffers from the impact of an emergency, with a high rate indicating greater regional economic resilience [46]. Stress stability refers to the ability of the economic system to maintain stability in the face of significant shocks, involving not only the economic stability of a single city but also the resilience and sustainable development of the entire economic network [47]. The amplified contagion rate describes the ability of a node in the network to spread information or influence, also known as the “snowball” effect [48]. Understanding this effect allows researchers and policymakers to identify key nodes in the economic network and formulate targeted policies to enhance economic resilience [49].

4.1.3. Resilience Index

Economic resilience, global approach efficiency, road density, and attack scene stability are selected as the main indexes through which to measure resilience. Economic resilience refers to the transformation of each city’s development from the unbalanced stage to the balanced stage and the gradual recovery of economic construction in the organizational stage after the city is impacted [50]. Global approach efficiency reflects the structural toughness of the traffic composite network. Road density indicates that the urban traffic system can provide functions of distribution, transportation, and convenience while being affected by infectious diseases, possessing certain disaster-bearing capabilities [51]. The stability of the attack scenario usually refers to the ability of the city’s economy to maintain normal operation after being hit [52].

4.1.4. Indicators of Creativity

The degree of scientific and technological innovation, the degree of patent application activity, and the kernel density of high-tech enterprises are selected as the main indicators through which to measure creativity [53]. The development of scientific and technological innovation significantly promotes the improvement of urban economic resilience, effectively enhancing the city’s ability to resist external shocks and interference [54,55]. This paper evaluates the activity of patent application and the nuclear density of high-tech enterprises, reflecting the strength of cities in technological innovation, which can effectively improve urban economic resilience.

4.2. Research Technique: Projection Pursuit Model

The Projection Pursuit (PP) Model is a widely applied statistical method for handling multifactor, high-dimensional, and non-linear data. It has been especially useful in the analysis of datasets that are high-dimensional, non-linear, and non-normal [24]. The model is recognized for its strong stability robustness against noise, and high precision in evaluating relative advantages among different research objects. Consequently, it has been employed in fields such as hydrological forecasting, measurement of land use intensity, and other forms of quantitative evaluation.

The central idea of the PP model is to employ modern information-processing systems (such as DPS) to project high-dimensional data into a lower-dimensional subspace through specific linear combinations. A projection objective function is then constructed to measure how well the projection reveals the underlying structure of the data. By imposing certain constraints, the optimal projection direction vector is identified as the one that maximizes the objective function. Based on this optimal direction, the structural features of high-dimensional data can be effectively analyzed in the lower-dimensional space.

Computational Steps of the Projection Pursuit Model

1: Standardization of Evaluation Indicators

Let the sample set be X = {${\mathrm{x}}_{\mathrm{ij}}$}, where ${\mathrm{x}}_{\mathrm{ij}}$ denotes the raw value of the j-th indicator for the i-th observation; $\mathrm{n}$ is the number of observations; and $\mathrm{p}$ is the number of indicators.

Since different indicators may vary in dimension and range, standardization is required to ensure comparability. Using the min–max (range) method, the standardized values are computed as follows.

For positive indicators,

$${\mathrm{x}}_{\mathrm{ij}}={\displaystyle \frac{{\mathrm{x}}_{\mathrm{ij}}-\min \left({\mathrm{x}}_{\mathrm{j}}\right)}{\max \left({\mathrm{x}}_{\mathrm{j}}\right)-\min {\mathrm{x}}_{\mathrm{j}})}}$$

(23)

For negative indicators,

$${\mathrm{x}}_{\mathrm{ij}}={\displaystyle \frac{\min \left({\mathrm{x}}_{\mathrm{j}}\right)-{\mathrm{x}}_{\mathrm{ij}}}{\max \left({\mathrm{x}}_{\mathrm{j}}\right)-\min {(\mathrm{x}}_{\mathrm{j}})}}$$

(24)

where $\max \left({\mathrm{x}}_{\mathrm{j}}\right)$ and $\min \left({\mathrm{x}}_{\mathrm{j}}\right)$ represent the maximum and minimum of the j-th indicator across all observations. The standardized values ${\mathrm{x}}_{\mathrm{ij}}$ are dimensionless and lie within the interval [0, 1].

2: Construction of the Projection Objective Function

Let $\mathrm{a}={({\mathrm{a}}_1,{\mathrm{a}}_2,...,{\mathrm{a}}_{\mathrm{p}})}^{\mathrm{T}}$ denote the projection direction vector, where $\Vert \mathrm{a}\Vert$ = 1. The one-dimensional projection value of sample $\mathrm{i}$ in direction $\mathrm{a}$ is

$${\mathrm{y}}_{\mathrm{i}}={\displaystyle \sum _{\mathrm{j}=1}^{\mathrm{p}}}{\mathrm{a}}_{\mathrm{j}}{\mathrm{x}}_{\mathrm{ij}}$$

(25)

To optimize the one-dimensional projection, the projection values ${\mathrm{y}}_{\mathrm{i}}$ should simultaneously exhibit local concentration (forming tight clusters) and global dispersion (maximizing separation among clusters). Accordingly, the projection objective function is defined as

$$\mathrm{Q}\left(\mathrm{a}\right)={\mathrm{S}}_{\mathrm{y}}{\mathrm{D}}_{\mathrm{y}}$$

(26)

where ${\mathrm{S}}_{\mathrm{y}}$ is the standard deviation of the projection values, and ${\mathrm{D}}_{\mathrm{y}}$ denotes local density:

$${\mathrm{S}}_{\mathrm{y}}=\sqrt{{\displaystyle \frac{1}{\mathrm{n}}}{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}}{\left({\mathrm{y}}_{\mathrm{i}}-\stackrel{-}{\mathrm{y}}\right)}^2}, {\mathrm{D}}_{\mathrm{y}}={\displaystyle \frac{1}{\mathrm{n}\left(\mathrm{n}-1\right)}}{\displaystyle \sum _{\mathrm{i}=1}^{\mathrm{n}}}{\displaystyle \sum _{\mathrm{j}=1}^{\mathrm{n}}}\mathsf{\phi }\left(\mathrm{r}-\left|{\mathrm{y}}_{\mathrm{i}}-{\mathrm{y}}_{\mathrm{j}}\right|\right)$$

(27)

In these formulas,

$\stackrel{-}{\mathrm{y}}$ is the mean of the projection values.

$\mathrm{r}$ is the neighborhood window radius for density estimation.

$\left|{\mathrm{y}}_{\mathrm{i}}-{\mathrm{y}}_{\mathrm{j}}\right|$ is the distance between samples $\mathrm{i}$ and $\mathrm{j}$.

$\mathsf{\phi }$ is the unit step function:

$$\mathsf{\phi }\left(\mathrm{u}\right)=\left\{\begin{array}{lll}1& \mathrm{if}& \mathrm{u}\ge 0\\ 0& \mathrm{if}& \mathrm{u}<0\end{array}\right.$$

(28)

The optimization problem is thus to find the projection direction vector ${\mathrm{a}}^{*}$ that maximizes $\mathrm{Q}\left(\mathrm{a}\right)$.

3: Optimization and Analysis

By solving for ${\mathrm{a}}^{*}$, the high-dimensional dataset is projected into a low-dimensional space where structural patterns, such as clustering and separation among groups, are revealed. These projection values are then analyzed to assess the comparative strengths and weaknesses of the objects under study.

5. Results

5.1. Temporal Changes in Urban Economic Resilience

By summarizing the urban economic resilience data using Excel, this study calculates the mean changes in urban economic resilience over the years. Specifically, the average economic resilience in 2019 is 0.107; in 2020, it is 0.093; and in 2021, it is 0.114. It can be seen that the resilience of China’s urban economy shows a trend of first declining and then rising. In 2019–2020, the economic resilience was in a declining stage, with the average total economic resilience decreasing by 0.014, indicating a declining trend in China’s economic resilience under the impact of the pandemic. From 2020 to 2021, as the COVID-19 pandemic was initially controlled, the economic resilience entered a rising phase, with the average total economic resilience increasing by 0.021, indicating the recovery and improvement of urban economic resilience in China.

5.2. Spatial Characteristics of Urban Economic Resilience

In recent years, global resilience has become a key indicator of regional economic health and resilience, particularly in the face of external shocks. As the world’s second-largest economy, the performance of China’s economic resilience in different periods and regions profoundly reflects the country’s adaptability and resilience in responding to various internal and external challenges, especially sudden social and economic crises such as the novel COVID-19 pandemic. Therefore, this study evaluates the economic resilience of Chinese cities, as shown in Figure 2.

As illustrated in Figure 2, China’s eastern region has demonstrated robust economic resilience against external shocks such as the COVID-19 pandemic. This resilience is largely attributable to its advantageous geographical location, a highly developed industrial base, and relatively comprehensive social security systems. In contrast, the western region and parts of the northeast, long constrained by resource-dependent economic structures, exhibited weaker resilience and slower recovery during the pandemic. Notably, disparities in provincial economic resilience had already begun to emerge by 2019. The eastern coastal provinces showcased remarkable resilience through the synergistic development of high-tech industries, advanced manufacturing, and modern services, while the western region lagged due to its reliance on traditional, resource-based industries. These differences can be explained by two primary factors: first, the eastern region’s improved infrastructure and diversified industrial structure allowed for greater adaptability to external shocks; second, the western region’s dependence on a singular economic model and lack of endogenous growth momentum rendered it more vulnerable. In this context, agglomeration effects and industrial innovation capacity play critical roles in enhancing resilience. The nationwide outbreak of COVID-19 in 2020 had profound impacts on both China’s economy and the global economic system. The pandemic intensified uncertainty and delivered severe shocks to regional economies. Figure 2 highlights that regional economic resilience in China fluctuated dramatically in 2020, with sharp declines in some of the hardest-hit provinces such as Hubei, Guangdong, and Zhejiang. More detailed explanations, however, reveal important differences: while these provinces experienced significant short-term downturns, their resilience trajectories diverged due to local structural and institutional capacities. For instance, despite the initial shock, economically advanced eastern provinces such as Shanghai and Jiangsu rapidly restored production thanks to strong industrial bases and efficient public health systems, thereby partially strengthening resilience. Conversely, central and western regions, which are heavily reliant on resource-based industries and characterized by weaker foundations, struggled to mount a swift recovery, facing greater socio-economic pressures. By 2021, as the pandemic came under control and global recovery gained momentum, the economic resilience of Chinese cities once again diverged markedly. The eastern region further strengthened its resilience, driven by digital transformation and the rise of green industries, while government-supported infrastructure and ecological projects improved resilience in the western region. These divergent trajectories underscore that COVID-19 did not exert uniform effects but instead amplified pre-existing structural differences across regions. In the eastern coastal areas, resilience was consolidated through innovation and sustained investment in science and technology. In the west, gradual improvements came from industrial diversification and enhanced infrastructure, though overall resilience remained relatively weaker. A comprehensive assessment of changes in China’s economic resilience from 2019 to 2021 demonstrates that resilience reflects not only economic fundamentals and industrial structures but also governance capacity, public service provision, and social capital. The eastern region’s strong resilience stems from sound infrastructure, sophisticated industrial systems, and flexible markets. In contrast, the structural vulnerabilities of western and some central regions, rooted in long-standing reliance on traditional industries, continue to undermine resilience. Moreover, the effectiveness of government policies proved decisive in shaping resilience outcomes during the pandemic, as timely emergency management, social security measures, and financial support determined recovery trajectories. Ultimately, the differentiated impacts of COVID-19 on regional resilience highlight the uneven nature of China’s economic development; while the eastern region maintains a leading position both in growth and in resistance to external shocks, the western and central regions, despite recent improvements, still face significant resilience gaps. Enhancing resilience thus requires not only short-term crisis management but also long-term structural reforms, particularly through infrastructural development, industrial upgrading, and technological innovation in the western region.

5.3. Spatial Autocorrelation Analysis of Urban Economic Resilience

This study also focuses on the spatial autocorrelation analysis of economic resilience to reveal the spatial heterogeneity and resilience evolution process of regional economic development, providing a theoretical basis and empirical support for future policy design and regional development strategies. Generally, global spatial autocorrelation is used to analyze the spatial distribution pattern of data in the entire region, determining whether all regions show a unified spatial aggregation phenomenon. Moran’s I spatial autocorrelation index is one of the most common global spatial autocorrelation indexes, with values ranging from −1 to 1. A value closer to 1 indicates significant positive spatial autocorrelation in regional economic resilience, meaning that similar economic resilience or other variables tend to cluster geographically. A value closer to −1 indicates negative spatial autocorrelation, meaning that similar values are relatively dispersed. A value of 0 means that the data has no significant correlation in space and presents a random distribution [56]. The global autocorrelation index of this study is shown in Figure 3.

From Figure 3, it can be seen that from 2019 to 2021, the global spatial autocorrelation index of national economic resilience showed a certain degree of spatial aggregation, and both passed the significance test of 0.001. In

Table 2. Global autocorrelation index of economic resilience, 2019–2021.

Year	Moran Index	p Value
2019	0.677	0.001
2020	0.669	0.001
2021	0.723	0.001
Mean value	0.691	0.001

, in 2019, Moran’s I spatial autocorrelation index was 0.677, indicating that the national economic resilience that year showed a significant spatial clustering feature. In 2020, due to the impact of the novel COVID-19 pandemic, the global economy experienced sharp fluctuations, but China’s economy showed a certain degree of resilience [57]. In Figure 3, the global spatial autocorrelation index for 2020 is 0.649, slightly lower than in 2019, reflecting the varying degrees of impact of the pandemic on economic resilience in different regions. In 2021, as the pandemic was effectively controlled and the national economy gradually recovered, Moran’s I spatial autocorrelation index rose to 0.723, reflecting the overall recovery of national economic resilience.

In addition, based on the economic data from 2019–2021, this study also uses the spatial autocorrelation analysis method and combines ARCGIS 10.8 to draw a spatial aggregation diagram of economic resilience of various regions in China in different years, as shown in Figure 4. In Figure 4, red areas represent regions with high economic resilience that are similar to their surrounding areas (“high–high” clusters), and blue areas represent regions with low economic resilience that are in line with their surrounding areas (“low–low” clusters). The light blue region and light red region show a “low–high” and “high–low” clustering pattern, respectively, with great differences in toughness from neighboring regions. The details are as follows:

Looking at Figure 4, the spatial evolution of economic resilience reflects the multiple characteristics of China’s regional economic development. Among them, the high-resilience areas in the southeast coast show strong anti-pressure ability, but at the same time, they also expose the risk of industrial simplification and excessive external dependence. In the central and western regions, economic resilience has gradually increased through policy support, industrial diversification, and infrastructure construction. As an external shock, the pandemic not only tested the economic resilience of various regions but also promoted the adjustment of policies and industrial structure. In terms of concentration, the overall spatial distribution features “high in the east–low in the west” and “high in the coast–low in the inland” patterns. The eastern region has a high concentration of high-tech enterprises and financial services. It has a relatively diversified economic structure, which provides strong support for its economic recovery, while the low economic resilience of the northwest and southwest regions may be related to the single economic structure and dependence on traditional industries. This makes these regions significantly more vulnerable to external shocks.

In 2019, China’s economic development was relatively stable. As can be seen from Figure 4a, the southeast coastal areas, including the Pearl River Delta, Yangtze River Delta, and other economically developed regions, show a strong high–high aggregation pattern. In the process of globalization, regions with high concentration have accumulated a strong industrial base, scientific and technological innovation capacity, medical service capacity, and efficient infrastructure construction; they also have high economic resilience. Of them, Beijing, Tianjin, Shandong, Jiangsu, Shanghai, Anhui, Guangdong–Hong Kong–Macao Greater Bay Area, among others, with their developed high-tech industries and financial services, have demonstrated a strong ability to withstand the impact of major public health security. In contrast, parts of the western and northeastern regions, such as Tibet, Xinjiang, Gansu, Qinghai, Heilongjiang and other provinces, show a low–low aggregation model with low economic resilience. Such regions rely on traditional resource-based industries, have a single industrial structure, and are relatively lacking in innovation capacity, which makes them more vulnerable to external shocks. In 2019, the distribution of economic resilience in geographical space showed significant regional differences, which highlights the current situation of uneven regional economic development in China. The agglomeration effect of the eastern coastal region is in sharp contrast to the relative lag of the western region, which reveals that there are significant differences in the ability of different regions to withstand pressure when dealing with global uncertainties and regional economic fluctuations.

In 2020, the outbreak of the novel COVID-19 pandemic hit the global economy unlike any previous event [58]. As one of the first countries to respond to the pandemic, China’s economic resilience was tested on a large scale. Figure 4b shows the economic resilience of each region during the pandemic. As can be seen from Figure 4b, the southeast coastal areas, especially some cities in the Yangtze River Delta and Pearl River Delta, still maintain a strong high–high clustering mode, and their economic resilience is relatively solid. However, cities that are highly dependent on traditional manufacturing and external demand were hit hard by the outbreak. Among them, in Wuhan and some cities affected by the blockade, although economic resilience before the outbreak of the pandemic was strong, due to excessive reliance on specific industries and markets, the short-term economic recovery showed fragility. In contrast, in some western regions, where the economic resilience was originally low (such as Guizhou, Yunnan, etc.), a low–high clustering pattern appeared. This shift reflects the country’s policy-based support for these regions and the positive effects of industrial transformation. Support policies issued by the government, including tax and fee reductions, financial assistance and infrastructure construction, have provided strong support for areas with relatively weak economic foundations. In particular, the cultivation of emerging industries has promoted the gradual strengthening of economic resilience. At the same time, many small and medium-sized cities and counties have improved their ability to respond to emergencies through the development of the online economy and digital transformation, thus enhancing the resilience of the economy. The spatial aggregation map of 2020 reveals the dynamic changes of economic resilience during the crisis. Some previously disadvantaged regions have enhanced their resilience with the help of policy support and industrial diversification, while some previously strong regions have shown greater vulnerability during the crisis, which highlights the risk of a single economic structure and external dependence.

In 2021, the pandemic was brought under initial control, and China entered a recovery stage of economic development. Figure 4c shows the change in economic resilience by region after the pandemic. It can be seen that many regions with low economic resilience (such as some western regions) gradually entered a state of low–high concentration through policy-based support and industrial upgrading after experiencing the pandemic, and economic resilience significantly improved. The economic resilience of Inner Mongolia, Sichuan, Shaanxi, and other provinces has been improved in the process of recovery from the pandemic, and this is closely related to increasing support for high-tech industries, strengthening infrastructure construction, and improving the structure of the industrial chain. At the same time, although the developed areas of the southeast coast still maintain strong resilience, showing a high–high aggregation, the growth rate of economic resilience in some cities has slowed down. In view of the fact that the industrial upgrading of these regions has reached a saturation state, the potential for economic growth has gradually become limited. High-density cities along the southeast coast are likely to encounter more challenges such as resource constraints, environmental pressures, and population mobility, factors that could have an impact on their long-term economic robustness.

In general, under the impact of the COVID-19 pandemic, China’s economic resilience between 2019 and 2021 shows clear regional differences. The spatial differences in economic resilience among different regions are relatively significant, with the southeast coastal region generally showing strong economic resilience, while the western region and some remote areas are relatively weak. In addition, the impact of the pandemic has prompted the adjustment of policies and structures. As an external shock, the pandemic not only revealed the vulnerability of some regions but also provided opportunities for some regions to enhance economic resilience, especially in terms of policy support and industrial diversification; it also promoted the coordinated development of local regions.

5.4. Dimensional Analysis of Economic Resilience

(1) Resistance analysis

As can be seen from Figure 5, there is significant spatial heterogeneity in resistance across different regions of China, showing a spatial pattern of a decreasing trend from east to west. The eastern coastal areas and some central developed regions have strong resistance, while parts of the western and inland areas are relatively weak. This geographical difference is closely related to factors such as the level of economic development, infrastructure construction, social governance capacity, and the distribution of natural resources. Specifically, the eastern coastal areas (such as Beijing, Shanghai, Guangzhou, and other major cities) generally show higher resistance. The medium-resistance areas are in the middle reaches of the Yangtze River and the surrounding areas of the Pearl River Delta, such as Wuhan, Nanjing, and Hangzhou. Although these areas are relatively resilient, they are slightly less resilient than the core cities on the eastern seaboard. The main reasons are as follows. First, in terms of economic growth, these cities have experienced rapid urbanization and economic growth in the past few decades; although their infrastructure and social governance levels are less than those of some developed cities in the east, they have strong economic resilience in their respective regions. Second, in terms of industrial structure optimization, the industrial structure of these regions is more diversified, with service and high-tech industries developing rapidly in addition to traditional manufacturing. Nevertheless, these regions are slightly less resistant than the eastern coastal regions due to differences in technological innovation and the degree of capital concentration. Finally, in terms of local governance and policy support, compared with the eastern region, there is still room for improvement in the policy support and social governance capabilities of some central cities, but overall, local governments are more responsive and able to deal with challenges more effectively. Low-resistance areas are mainly in the west and some remote inland areas (such as Xinjiang, Tibet, Gansu, etc.), showing low resistance.

(2) Stress force analysis

As can be seen from Figure 5, the stress levels in different regions of China show obvious geographical differences. Compared with the distribution of resistance, the distribution of stress levels presents different characteristics: the stress levels in the eastern and some central regions are stronger, while the stress levels in the western and some remote areas are relatively lower. Among them, the Yangtze River Delta, Pearl River Delta, Beijing–Tianjin–Hebei, and other developed regions are high-stress areas, while the western and remote areas (such as Xinjiang, Tibet, Qinghai, etc.) are low-stress areas in China. The characteristics of high-stress areas usually include advantages in economic development, a sound social and medical service system, rapid response to the pandemic, and an efficient governance structure.

(3) Resilience analysis

As can be seen from Figure 5, during 2019–2021, urban economic resilience in China is characterized by high resilience in the east and low resilience in the west. It is worth noting that the economic resilience of districts and counties in Hebei, Henan, and Anhui is even higher than that of other places with higher levels of economic development. The reason for this may be that these places have a large number of out-of-town employment populations and are close to economically developed areas such as Beijing and Shanghai. After the impact of the pandemic, a large number of migrant workers returned home, making these areas more resilient. At the same time, the private economy in the coastal areas of Fujian is relatively developed, and its economic resilience is also strong.

(4) Analysis of innovation power

As can be seen from Figure 5, from 2019 to 2021, the innovation capacity of China’s urban economic resilience shows a “high coastal–low inland” pattern. Relevant studies show that high innovation is usually closely related to urban economic development, technological intensity, and human capital. As can be seen from Figure 5, the distribution pattern of innovation power is roughly similar to that of China’s total economic aggregate and population density, differentiated according to the Heihe–Tengchong line. The innovation vitality of the eastern region is more significant, while the western region is relatively low in this metric, further confirming that financial development plays a positive role in promoting innovation and entrepreneurship activities and has a significant spillover effect. The interaction between innovation and economic development is mutually reinforcing, and the synergistic effect is consistent with economic resilience. Specifically, cities with high innovation vitality are mainly concentrated in the Beijing–Tianjin–Hebei city cluster, the Shanghai metropolitan area, and the Pearl River Delta region. Except for provincial capital cities (such as Xining and Urumqi), the innovation vitality of the western region is generally not strong.

6. Discussion

This study contributes to the growing body of literature on urban economic resilience by introducing several methodological and analytical innovations. First, by incorporating the dimension of stress response under pandemic shocks, the study extends existing resilience frameworks and provides a more comprehensive measure that captures the dynamic impact of COVID-19. Second, the use of county-level administrative units as the scale of analysis allows for a finer-grained understanding of spatial heterogeneity across China, supplementing previous research that primarily focused on provinces or metropolitan regions. Third, the integration of multiple big data sources, including POI, night light, road network, and industrial datasets, enhances the reliability and representativeness of the resilience indicators, allowing for a multidimensional reflection of local economic resilience.

The empirical results demonstrate both temporal dynamics—a decline in resilience at the onset of the COVID-19 pandemic followed by recovery—and spatial disparities, with the eastern coastal regions exhibiting stronger resilience compared to the central and western regions. These findings resonate with earlier studies emphasizing the advantages of infrastructure, industrial diversification, and market flexibility in coastal China, while also highlighting the structural vulnerabilities of regions reliant on traditional industries. Importantly, the evidence from the sub-indicators (resistance, stress, resilience, and innovation) confirms that resilience is multidimensional: eastern areas benefit not only from stronger industrial systems but also from higher levels of creativity and innovation, whereas western regions remain constrained by weaker adaptive capacity.

Despite these contributions, the study faces several limitations. First, while regional differences in resilience are evident, the underlying policy factors driving these differences merit deeper exploration. During 2019–2021, local COVID-19 governance largely aligned with central policies, limiting the observable impact of policy heterogeneity. However, after 2021, the implementation of the “dynamic zero-COVID” strategy introduced greater regional variation in pandemic governance, suggesting that future research should incorporate this policy dimension to better explain resilience disparities. Second, reliance on available datasets limits the measurement of some indicators, particularly those related to informal economic activities and social governance capacity, which may also significantly influence resilience outcomes.

Taken together, this study advances the measurement and understanding of urban economic resilience in China, while also pointing to the need for further investigation into the role of governance flexibility and policy differentiation, especially in the context of prolonged shocks such as the COVID-19 pandemic.

7. Conclusions

This study addressed the scientific gap identified in the Introduction by developing a comprehensive, county-level evaluation system for urban economic resilience in China. Specifically, it aimed to quantify the spatiotemporal dynamics and multidimensional capacity of local economies under the disruptive shock of the COVID-19 pandemic (2019–2021). By successfully integrating multiple big data sources (POI, night light, road network) and introducing a key methodological innovation—the dimension of stress response—this research achieved its core objectives and provided a novel, fine-grained understanding of resilience heterogeneity across China.

The empirical findings confirm two major patterns central to urban resilience theory. First, temporal dynamics show a characteristic shock-absorption-and-recovery trajectory, where resilience initially declines and then stabilizes, confirming the adaptive nature of the urban system. Second, spatial heterogeneity is significant, with eastern coastal regions exhibiting systematically stronger resilience compared to central and western regions. This resilience is not monolithic; sub-index analysis robustly demonstrates that the eastern advantage stems from superior performance not only in an industrial resistance and resilience capacity but also in innovation and creativity—a critical dimension of long-term adaptive capacity. This finding underscores that high resilience is a composite outcome of structural advantages and high-level knowledge systems, rather than mere initial size.

From a scientific methodological point of view, this study makes three significant contributions to academia and relevant studies. First, the use of the county-level unit breaks the scale limitation of previous provincial or metropolitan analyses, offering a more granular perspective that informs local management. Second, the integration of the stress response indicator provides a more complete measure of system dynamics under external shocks, advancing the theoretical framework of resilience. Third, the integration of multi-source big data enhances the reliability and multidimensionality of resilience indicators, setting a new standard for future empirical research on this topic.

Despite these contributions, the study acknowledges its scientific limitations. Chief among them is the difficulty in isolating the effect of local policy heterogeneity from centralized national governance during the 2019–2021 period, which limits the causal inference regarding policy drivers of resilience disparity. Furthermore, data availability constrains the full measurement of certain indicators, such as informal economic activities. Therefore, future research should focus on two key areas: (1) incorporating the policy dimension following 2021’s regional variation in governance to quantitatively assess the impact of policy differentiation and (2) exploring the micro-mechanisms of resilience at the firm or community level to deepen our understanding of adaptive behavior.

8. Policy Recommendations

• Strengthening Regional Structural Balance: Policymakers must actively drive the economic transformation and industrial upgrading of less developed regions to narrow the economic gap and enhance the nation’s overall risk resistance.
• Fostering Industrial Diversification and Innovation: Policymakers should avoid single industrial structures and external market over-dependence and prioritize the development of high-tech and modern service industries to bolster adaptive capacity.
• Implementing Differentiated Policy Strategies: Local policymakers need to tailor strategies based on regional economic characteristics, particularly by strengthening infrastructure, industrial diversification, and digital transformation in the central and western regions.
• Leveraging Core Cities: Guided by market mechanisms, policymakers should strengthen urban connectivity and leverage the radiating and driving role of high-resilience coastal cities to facilitate regional joint force formation.
• Enhancing Emergency Governance: Risk forecasting and emergency drills should be institutionalized by building multi-level monitoring and early warning systems. This will ensure a rapid response mechanism for economic shocks and public health emergencies, making resilience strategies practical for citizens and urban managers.