Building Resilient Supply Chains: Evidence from Climate-Adaptive City Construction in China

Abstract

Supply chain resilience (SCR) is the core support for building a modernized industrial system and guaranteeing industrial security. In this paper, we aim to evaluate the impact of the Climate-Adaptive City Construction (CACC) pilot policy on SCR and to investigate its underlying mechanisms using a quasi-natural experiment based on the 2017 pilot initiative. We employ the difference-in-differences (DID) model on the panel data of 249 prefectural-level cities in China from 2006 to 2023. The results show that CACC significantly improves SCR, and this effect is stronger on the southeastern side of the Hu Huanyong line, as well as in non-resource cities and non-“two-control” cities. The mechanism analysis reveals that CACC enhances the resilience of the urban industrial chain by fostering environmental awareness, increasing the level of green technology innovation, and reducing the extent of urban sprawl. In addition, the positive spatial spillover effect of the pilot policy on SCR is further verified using the Spatial Durbin Model (SDM). The results presented in this paper serve as a reference for the continued promotion of CACC and offer policy optimization recommendations.

1. Introduction

In light of the complexities and fluctuations within the international landscape, the transformation of the global production system exacerbates tensions in economic and trade relations. The COVID-19 pandemic exposed critical vulnerabilities in supply chains, leading to widespread shortages in semiconductors and medical supplies. These disruptions were further compounded by extreme weather events, such as the 2021 floods in Henan, which severely impacted manufacturing and logistics hubs in central China. Such incidents highlight the recurring problems of “stuck chains” and “broken chains,” underscoring the urgent need to enhance the resilience of China’s industrial systems [1]. Supply chain resilience (SCR) denotes the capacity to maintain stability, recover from disruptions, and adaptively upgrade, a crucial element for economic growth and industrial transformation [2]. Despite China’s relatively comprehensive industrial foundation, persistent structural issues such as an imprecise system and shortages in key technologies and components continue to hinder its efficient and secure operation. Ensuring a secure and stable industrial chain has therefore become essential for addressing external uncertainties and safeguarding economic security [3]. It also represents a vital pathway for industrial advancement and high-quality economic development, forming a fundamental prerequisite for China’s new development paradigm [4,5]. In recent years, international research teams have begun exploring similar intersections between climate adaptation and economic resilience. For instance, the authors of [6] demonstrated how extreme weather events significantly impact economic systems through financial channels, highlighting the vulnerability of modern economies to climate disruptions. Similarly, the authors of [7] examined urban resilience strategies across multiple countries, finding that integrated approaches yield better economic outcomes. In our study, we build upon this emerging international literature by specifically examining how CACC influences SCR within the Chinese context. Enhancing the resilience of the industrial chain has emerged as a critical issue requiring immediate resolution.

SCR is a comprehensive reflection of internal structure, technological innovation, external resources, emergency response capacity, and market demand that must be enhanced through the optimization of structure, technological autonomy, diversified layout, institutional safeguards, and other multi-dimensional synergistic enhancements. Climate change intensification has led to more frequent and severe extreme weather events, significantly impacting global GDP growth and undermining urban economic stability. As key hubs for industrial chains, cities experiencing such instability directly disrupt normal production operations. The results of previous studies confirm that increased climate disasters weaken the urban capacity to support industrial activities [8]. Extreme weather disrupts production processes and climate-induced economic instability reduces industrial productivity, making adaptation an urgent priority in economic development. In response, the UN’s 2030 Agenda explicitly advocates building resilient and sustainable cities. Accordingly, China has implemented the CACC program to enhance urban adaptive capacity and reduce climate vulnerability [9]. In this context, one of China’s top concerns is how to improve the capacity for climate resilience while simultaneously fortifying the stability of the domestic industrial chain.

A growing body of studies address the complexities of climate governance, yet distinct conversations have evolved without sufficient integration. Current research efforts have evolved along two relatively independent trajectories. On the one hand, studies on urban climate adaptation focus on integrating resilience concepts into urban policy [10,11], evaluating the impact of climate-resilient city construction on sustainable development [12] and proposing targeted strategies for specific risks such as sponge city initiatives [13]. These investigations primarily address the capacity of urban systems as a whole to withstand climate shocks. Conversely, SCR research has progressed from conceptual definitions [14,15] to analysis of determinants, particularly the enhancing effects of government actions and digital transformation. Studies demonstrate that governance tools like public data openness [16] and supply chain digitalization [17] effectively strengthen the risk resistance of enterprises and supply chain networks. This field is also witnessing methodological innovation, with scholars beginning to employ technologies such as machine learning to predict disruption risks in specific industries [18,19], driving the research toward multidimensional and forward-looking development.

While all of the above literature sources are robust, a critical theoretical and empirical gap exists at their intersection. The authors of studies on climate-adaptive cities often assume that urban resilience naturally fosters economic stability yet rarely demonstrate how city-level policies specifically enhance supply chain robustness [20,21]. Conversely, the authors of studies on SCR typically treat climate risk as external, overlooking how proactive urban adaptation might systematically mitigate disruptions [22,23]. While CACC policies theoretically strengthen SCR through green technology promotion and infrastructure improvements [24], empirical evidence remains scarce regarding how these urban interventions translate to resilience within complex industrial networks [25,26]. We bridge this divide by investigating whether building climate-resilient cities directly enhances SCR, thereby integrating urban and supply chain perspectives [27].

In summary, an examination of the current literature reveals that scholars are more inclined to discuss climate governance or climate resilience as a means of indirectly enhancing SCR [28] and tend to examine climate governance and SCR as separate domains, leaving critical gaps in understanding. First, empirical evidence on the causal relationship between CACC and SCR remains scarce. Second, most authors focus on measuring SCR without investigating its transmission mechanisms, particularly from an urban perspective. Third, the coupling mechanism between urban and industrial resilience requires deeper theoretical elaboration. To address these gaps, we investigate institutional, dynamic capability, and complex adaptive systems theories to analyze how CACC pilot policies reshape urban governance and industrial behavior [29], enhance resource reconfiguration capacity, and explain non-linear interactions within urban systems [30]. We systematically evaluate the policy’s impact on SCR, identify mediating roles of environmental awareness, green innovation, and urban spatial restructuring, and examine its spatial spillover effects. By clarifying these mechanisms and heterogeneous effects across regions, the findings of this study advance the theoretical understanding of the city–industry resilience nexus and provide empirical support for policy evaluation [31].

Potential marginal contributions of this study include addressing the inadequate focus on SCR regarding climate resilience in the existing literature. Through this study, we provide an early empirical investigation into the effects of CACC on SCR at the city level in China. We utilize data from prefecture-level cities to capture changes in economic activities, aiming to integrate climate change and SCR into a cohesive analytical framework. This analysis helps to address a gap in the literature concerning the coupling between “city-industry” resilience. Regarding research methodology, the double difference model serves as a quasi-natural experimental instrument to quantitatively evaluate the policy’s net effect. In contrast, the spatial Durbin model is employed to confirm the positive spatial spillover effect of the policy, thereby offering comprehensive methodological support for the analysis of SCR in terms of causal inference and spatial analysis. We systematically examine the specific pathways of CACC aimed at enhancing SCR through cognitive transmission, technological advancement, and spatial governance, focusing on value embodiment across three dimensions: environmental concern, green technology innovation, and urban sprawl patterns. This analysis offers a scientific foundation for “mechanism identification–path design” in policy optimization.

The remainder of this paper is organized as follows: In Section 2, policy context and theoretical hypothesis are introduced, followed by a comprehensive study of the methodology in Section 3. In Section 4, we present the results and discussion, and in Section 5, we present a further analysis in terms of spatial effects. Lastly, in Section 6, we summarize the findings presented in this paper and provide recommendations.

2. Policy Context, Theoretical Analysis, and Hypotheses

2.1. Policy Context

China has developed a systematic approach to climate adaptation through progressive policy initiatives, with the institutional foundation being laid in 1994 when the country’s Agenda 21 first formally incorporated climate adaptation concepts [32]. Significant milestones followed, including the 2009 National Program that decentralized adaptation responsibilities to local governments, the 2013 National Climate Change Resilience Strategy that elevated resilience as a national priority, and the 2016 Urban Climate Change Resilience Action Plan that established a comprehensive implementation framework. The pivotal 2017 launch of the CACC pilot program marked China’s first practical investigation into urban climate resilience, designating 28 cities to develop climate-resilient infrastructure [33]. CACC represents a governance approach that integrates climate risk management into urban planning, infrastructure development, and early-warning systems. By strengthening urban systems’ capacity to withstand climate disruptions, CACC indirectly enhances SCR through multiple pathways, resilient infrastructure minimizes production delays from climate shocks, and green technology incentives help industries adapt to environmental changes [34]. Such features position CACC as both a climate adaptation strategy and a foundation for maintaining industrial stability amid climate uncertainties.

2.2. Theoretical Analysis and Hypotheses

From the standpoint of complex adaptive systems theory, CACC facilitates the prompt modification of industrial chains via systemic reconfiguration. This perspective views cities and their embedded supply chains as complex systems that can learn, adapt, and self-organize in response to external shocks [35]. The theory of system resilience posits that conventional urban systems might result in increased system vulnerability, as urban development is entirely reliant on pre-existing infrastructure and management frameworks [36]. Building on the above, the concept of evolutionary resilience emphasizes the capacity of systems not just to return to a pre-shock state, but to adapt and transform into more desirable configurations [37]. The augmentation of the industry chain’s resistance, resilience, and innovation depends on the policy’s direct capacity to counter climate change disruptions while also achieving multidimensional integration through cognitive, technological, and spatial conduits [38]. Furthermore, drawing on institutional theory, CACC pilot policies can be seen as creating new formal institutions and normative pressures that reshape the behavior of firms within the supply chain, compelling them to align with climate resilience objectives [39]. Internationally, researchers have developed theoretical frameworks that align with our approach. The authors of [40] proposed the REPAIR framework for European cities, emphasizing the importance of systemic interventions similar to our cognitive-innovation-layout pathway. In addition, the authors of [41] established comprehensive assessment indicators for urban climate adaptation that complement our theoretical foundation. Herein, we articulate the subsequent research hypotheses derived from the objectives of the pilot policy development mentioned above. Based on the above theoretical analysis, the impact path of CACC on SCR is mapped below (Figure 1).

Figure 1 illustrates the conceptual framework linking CACC to SCR through three primary mediating pathways: environmental awareness, green technology innovation, and urban sprawl patterns. The model posits both a direct effect of the policy on SCR and significant indirect effects channeled through these mediators. The direct effect stems from the policy’s role in fostering a systemic, risk-informed governance environment that directly bolsters the chain’s inherent capacity to anticipate, withstand, and recover from disruptions. The cognitive pathway reflects how heightened environmental concern prompts more resilient decision-making across supply chains [42]. The innovation pathway shows how green technology enhancement improves adaptive capacity and resource efficiency [43]. The spatial pathway demonstrates how polycentric urban development disperses physical risks while fostering knowledge spillovers [44]. These interconnected mechanisms collectively enhance SCR.

Figure 1 delineates three mediating pathways through which CACC influences SCR, each corresponding to our research hypotheses. The environmental awareness pathway aligns with H4, suggesting that enhanced environmental concern improves SCR. The green technology innovation pathway corresponds to H3, positing that technological advancement strengthens resilience. The urban sprawl pathway relates to H2, which proposes that optimized spatial patterns benefit SCR. Together, these pathways reinforce H1, which posits that CACC has a positive impact. To test these relationships empirically, we operationalize each element into measurable variables: government environmental concern captures environmental awareness, green invention applications represent technological innovation, and the urban sprawl index quantifies spatial patterns. The explained variable SCR is measured through the comprehensive evaluation system presented in Table 1.

In summary, we propose the following research hypotheses for the purpose of the pilot policy.

H1. 

The implementation of the pilot policy of CACC significantly improves the resilience of the industrial chain [45].

Industry is the driving force of urban sprawl. The population density and land use type of cities will be altered by the upgrading of industrial structure and changes in employment structure, and the transformation and upgrading of the overall industrial chain will unavoidably affect urban land use and spatial patterns [46]. The impact of urban sprawl on the industrial chain has been the subject of previous research, with a primary emphasis on the stage characteristics of the impact and the distinct effects that are generated by various sectors within the industry. The influence of industrial development on urban spatial expansion exhibits an “S” curve after the accelerated urbanization stage, which initially increases and subsequently decreases. The policy effect is regulated by the degree of urban sprawl through the degree of industrial spatial agglomeration. We hypothesize that CACC enhances SCR by reshaping urban spatial structure. The policy promotes polycentric development and climate-resilient zoning, creating more compact urban forms that mitigate inefficient sprawl. This spatial reorganization strengthens SCR through two mechanisms: dispersing critical industrial assets to reduce concentrated climate risks [47] while concurrently fostering compact clusters that enable knowledge spillovers and resource sharing [48]. Such planned agglomeration balances resilience benefits against over-concentration vulnerabilities. Additionally, it promotes the compactness of industrial clusters, strengthens technological spillover and resource sharing, and enhances the innovation power of the industrial chain [49]. Based on the preceding analysis, the subsequent research hypotheses are proposed in this paper:

H2. 

The development of SCR is facilitated by CACC, which enhances the extent of urban sprawl [50].

The competitiveness of SCR is contingent upon the extent of green technology innovation. From the perspectives of intellectual property protection, application scenario opening, and R&D subsidies, pilot policies for climate-resilient cities activate the “supply and demand cycle” of green technology innovation. The CACC encourages enterprises to invest in the research and development of low-carbon technologies and smart monitoring technologies in order to overcome the production bottleneck in traditional industries. Standardized regulations for CACC provide financial and technical support for the development and strengthening of SCR, accelerate technology commercialization, and establish green technology application scenarios. The synergistic upgrading of the industrial chain is facilitated by the carryover effect of green production technology, which also improves the overall industrial chain’s risk-resistant capacity [51]. In one instance, the industry chain’s resilience is enhanced by the rapid dissemination of green technology innovation information within the network, which enables enterprises to reduce decision-making costs and risks [52].

Consequently, the industry chain cohort effect is generated by the green technological innovation behaviors of enterprises embedded in the industry chain, which are observed, learned from, and imitated, laying a solid foundation for the improvement of the overall SCR. This dynamic reflects the concept of absorptive capacity and knowledge spillovers within industrial networks, where one firm’s innovation can elevate the capabilities of interconnected partners [53]. In light of the above, we present the following research hypothesis:

H3. 

CACC enhances SCR by promoting green technological innovation [54].

Both government attention and public attention are scarce cognitive resources, and the government and the public focus their attention on a limited number of key points, which in turn affects their decision-making behavior. We posit that CACC strengthens SCR by elevating environmental awareness among key stakeholders. As a scarce cognitive resource, heightened attention to climate issues reshapes decision-making behaviors [55]. Through policy advocacy and information disclosure, CACC signals strong government commitment, prompting stricter environmental regulations and strategic investments in resilient infrastructure [56]. Concurrently, firms respond by optimizing supply chain structures, increasing strategic reserves, and adopting green technologies to align with emerging “green legitimacy” expectations [57]. Enhanced public supervision further reinforces this trend, creating collective pressure for sustainable production practices [58]. These cognitive and behavioral shifts collectively build a more adaptive and resilient industrial system. This public pressure mechanism is a key element of stakeholder theory, which posits that firms must manage relationships with all entities that affect or are affected by their operations, including the public [59]. Based on the above analysis, the fourth research hypothesis proposed in this paper is as follows:

H4. 

CACC promotes the development of SCR by improving environmental concerns [60].

3. Methodology

3.1. Research Design and Data Source

In this study, we employ a quasi-experimental research design to evaluate the causal effect of CACC on SCR. The phased implementation of the CACC pilot policy, starting in 2017, creates a suitable quasi-natural experiment. The core identification strategy is the DID model. Our analysis involves a balanced panel of 249 prefecture-level cities from 2006 to 2023, with 2006 chosen as it marks the consistent availability of key indicators in statistical yearbooks and provides sufficient pre-policy baseline data.

The final sample comprises 21 pilot cities and 228 control cities after the exclusion of seven non-prefecture-level pilot areas due to administrative mismatches or data limitations. Data are primarily drawn from the China Urban Statistical Yearbook (CUSY), China Urban Construction Statistical Yearbook (CUCSY), CNRDS, Wind, and CSMAR databases, with missing values addressed through linear interpolation and municipal statistical bulletins.

3.2. Variable Selection

Explained Variable: Supply Chain Resilience (res). In this study, we adopt the three-dimensional framework of resistance, resilience, and innovation to evaluate SCR, following established approaches in the recent literature [61,62]. This structure effectively captures both the defensive capacities against disruptions and adaptive capabilities through recovery and innovation. To construct the composite index, we employ the entropy method for objective weight determination. This approach is excellent for highlighting discernible differences between cities. The resulting Supply Chain Resilience index is therefore a robust and unbiased composite measure. By using this approach, subjective bias is avoided through the assignment of weights based on the informational content of each indicator, making it particularly suitable for multidimensional resilience assessment [63]. The construction of the SCR index follows a structured process involving indicator standardization, weight assignment, and linear aggregation. First, the positive and negative indicators identified in Table 1 are normalized to eliminate unit differences. For a positive indicator $X_{ij,}$ the standardization is as follows:

$$x_{ij}={\displaystyle \frac{x_{ij}-\mathrm{m}\mathrm{i}\mathrm{n}\left(x_j\right)}{\mathrm{m}\mathrm{a}\mathrm{x}\left(x_j\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left(x_j\right)}}$$

(1)

For a negative indicator, it is as follows:

$$x_{ij}^{\mathrm{\prime }}={\displaystyle \frac{\mathrm{m}\mathrm{a}\mathrm{x}\left(x_j\right)-x_{ij}}{\mathrm{m}\mathrm{a}\mathrm{x}\left(x_j\right)-\mathrm{m}\mathrm{i}\mathrm{n}\left(x_j\right)}}$$

(2)

where $i$ denotes the city and $j$ denotes the indicator. Subsequently, the entropy value method objectively determines the weight $w_j$ for each indicator $j$. The calculation involves the proportion of city $i$ in indicator $j$:

$$p_{ij}=x_{ij}^{\prime }/{\sum }_{i=1}^m{p}_{ij}=x_{ij}^{\prime }/{\sum }_{i=1}^m{x}_{ij}^{\prime }$$

(3)

The entropy value $e_j$ for indicator $j$ is as follows:

$$e_j=-{\displaystyle \frac{1}{\mathrm{l}\mathrm{n}\left(m\right)}}{\sum }_{i=1}^m{p}_{ij}\mathrm{l}\mathrm{n}\left(p_{ij}\right)$$

(4)

with $m$ being the number of cities. The weight is derived as

$$w_j=(1-e_j)/{\sum }_{j=1}^n(1-e_j)$$

(5)

where $n$ is the total number of indicators. Lastly, the comprehensive SCR index $re{s}_i$ for city $i$ is computed by integrating the normalized indicators with their respective weights across all dimensions using the linear weighted sum method:

$$re{s}_i={\sum }_{j=1}^n{w}_j\cdot x_{ij}^{\mathrm{’}}$$

(6)

The value range of the SCR index is 0~1, and a larger value indicates that industrial supply chain resilience is higher, and vice versa. The SCR evaluation index system is presented in Table 1. In this study, we measure SCR through a composite index based on city-level statistical data. While this approach enables systematic cross-city comparison, it offers an indirect assessment that may not fully capture firm-level supply chain dynamics. The three-dimensional framework of resistance, resiliency, and creativity provides a comprehensive static measure of cities’ inherent capacity to withstand and adapt to disruptions [64]. However, this structural measure does not explicitly track dynamic recovery processes following specific shocks, which would require higher-frequency operational data not available at the city level. Our index thus reflects preparatory resilience capacity rather than real-time adaptive performance, suggesting that investigating dynamic recovery pathways represents a promising direction for future studies [65].

Table 1. SCR evaluation index system.

Level 1 Indicators	Level 2 Indicators	Level 3 Indicators
Resistance	infrastructural support	Road freight volume (tons)
		Railroad freight (tons)
		Civil aviation cargo and mail traffic (tons)
	industrial system	Value added of secondary and tertiary industries as a share of GDP (%)
Resiliency	financial synergies	Balance of various bank loans (10,000 yuan)
	industry benefits	Profit margin on total assets of industrial enterprises above designated size (%)
	government regulation	Fiscal expenditure per capita (10,000 yuan)
	industrial stability	Gross industrial output value above scale (10,000 yuan)
	market stability	Total retail sales of consumer goods (10,000 yuan)
Creativity	innovative inputs share of fiscal expenditure on science and technology in total local fiscal expenditure (%) innovation outputs	Share of fiscal expenditure on science and technology in total local fiscal expenditure (%)
	innovation outputs	Technology market turnover as a share of GDP (%)

Note: The construction of this index system draws on the methodologies outlined in [61,62].

Table 1. SCR evaluation index system.

Level 1 Indicators	Level 2 Indicators	Level 3 Indicators
Resistance	infrastructural support	Road freight volume (tons)
		Railroad freight (tons)
		Civil aviation cargo and mail traffic (tons)
	industrial system	Value added of secondary and tertiary industries as a share of GDP (%)
Resiliency	financial synergies	Balance of various bank loans (10,000 yuan)
	industry benefits	Profit margin on total assets of industrial enterprises above designated size (%)
	government regulation	Fiscal expenditure per capita (10,000 yuan)
	industrial stability	Gross industrial output value above scale (10,000 yuan)
	market stability	Total retail sales of consumer goods (10,000 yuan)
Creativity	innovative inputs share of fiscal expenditure on science and technology in total local fiscal expenditure (%) innovation outputs	Share of fiscal expenditure on science and technology in total local fiscal expenditure (%)
	innovation outputs	Technology market turnover as a share of GDP (%)

Note: The construction of this index system draws on the methodologies outlined in [61,62].

The selection of indicators in Table 1 is grounded in established theoretical and empirical research on SCR. Each indicator proxies a specific resilience dimension as follows. For the resistance dimension, infrastructural support captures logistics connectivity through three transport indicators. Road freight volume reflects regional logistics efficiency and enables material flow adjustment during disruptions [66]. Railroad freight indicates bulk material stability, while civil aviation cargo represents high-value goods transport. Together, these variables measure the logistics system’s capacity to maintain supply chain continuity.

The resiliency dimension focuses on post-shock recovery. Industrial stability, measured through gross industrial output value above scale, tracks production capacity restoration [67]. Market stability, captured by total retail sales of consumer goods, monitors demand recovery and distribution channel functioning.

The creativity dimension addresses adaptive capacity. Innovative inputs, represented by science and technology expenditure share, signal commitment to developing technological solutions. Innovation outputs, measured through technology market turnover relative to GDP, reflect knowledge commercialization and industrial upgrading potential [68], both essential for long-term resilience enhancement.

Core Explanatory Variable: CACC Pilot Policy (policy). The explanatory variable in this study is the climate-resilient city construction pilot policy, denoted as policy. This variable is constructed as a dummy variable based on the official list of pilot cities released by the Chinese government in 2017 under the “Pilot Work Program for the Climate-Adaptive Cities Construction”. The policy variable takes the value of 1 for cities included in the pilot program from the year of their designation onward, and 0 otherwise. The selection of pilot cities was formally announced by the Ministry of Ecology and Environment and other relevant national departments, providing a clear and publicly documented basis for assigning the policy variable across the sample period.

Control Variables. In order to control the influence of other factors on SCR, we introduce the following control variables based on [68,69,70]: population density ($ppl$), which is measured as the ratio of the total population at the end of the previous year to the average of the total population at the end of the current year to the land area of the administrative region. Ratio measure. Human capital ($hmc$), measured as the share of the number of people with higher education in the total number of people in the region. Degree of openness to the outside world ($oud$), measured based on expenditure through total exports and imports as a share of GDP. Transportation level ($tsl$), reflected by the logarithm of road freight per capita in prefecture-level cities. Degree of local government intervention ($dlgi$), measured by local fiscal expenditure as a share of GDP. The level of science, science and technology innovation ($sti$), calculated from the share of fiscal expenditure on science and technology in total local fiscal expenditure. The level of economic development ($gdp$), measured by the logarithmic indicator of GDP per capita.

Mediating Variables. Regarding the environmental concern indicator, we use text analysis to analyze the keywords in annual government work reports to form the government’s environmental concern indicator. We constructed an environmental concern thesaurus based on [71] and then calculated the ratio of thesaurus words in government work reports from 2012 to 2022 for 249 cities. This ratio is multiplied by 100 to express it as a percentage, which makes the environmental concern indicator more intuitive and easier to interpret in practical terms. Using a percentage scale helps standardize the indicator for comparison across cities and aligns with common practices in policy text analysis.

Regarding green technology innovation, we adopt a number of green invention applications. Green technological innovation is measured by the number of green invention applications [72], and green patents can ensure the availability and accuracy of green innovation data. Considering that there is a certain degree of delay between the application of patented technologies and their granting and that patents may already have an impact on the enterprise during the application process, the number of green patents can reflect the timeliness and stability of the enterprise’s green innovation.

Regarding urban sprawl, we follow the method presented in [73], which is formulated as $St = 0.5 \times \left({ LA}_i - {LH }_i\right)+ 0.5$, where ${ LA}_i$ represents the proportion of city area with below-average population density and ${LH }_i$ denotes the proportion with above-average density. This metric captures the degree of urban dispersion, with lower values indicating monocentric compactness and higher values reflecting polycentric sprawl. Empirically, more sprawling patterns characterized by polycentric development can decentralize economic activities and critical supply chain nodes. This spatial configuration mitigates concentration risks from localized climate disasters while fostering multiple industrial clusters that enhance knowledge spillovers and resource sharing. Such decentralization strengthens the innovation capacity and inherent flexibility of industrial chains, ultimately contributing to SCR.

3.3. Model Setting

First, the construction pilot policy of CACC adopts a method for estimating group causal effects. Its basic principle is to treat the policy as a “quasi-natural experiment”. By distinguishing between the pilot policy group and the non-pilot control group, it explores the impact of policy intervention on specific groups, which meets the modeling conditions of the Difference-in-Differences (DID) model. Therefore, the DID model was used to identify the policy net effect of the construction pilot program on SCR. The model is specified as follows:

$${res}_{it}={\beta }_0+{\beta }_1{policy}_{it}+{\beta }_2{X}_{it}+{\mu }_i+{\eta }_t+{\epsilon }_{it}$$

(7)

In Formula (7), ${res}_{it}$ denotes the SCR index of city $i$ in year $t$; ${policy}_{it}$ is the treatment variable, that is, the CACC pilot policy; when city i belongs to the policy pilot city in year t then it takes the value of 1, and vice versa it is 0, and its coefficient ${\beta }_1$ reflects the influence effect of the CACC on SCR; $X_{it}$ is the control variable, and ${\beta }_2$ is the coefficient of the control variable; ${\beta }_0$ is a constant term; ${\mu }_i$ is the city fixed effect; ${\eta }_t$ is the time fixed effect; and ${\epsilon }_{it}$ is the random perturbation term.

Second, we investigated the transmission channels outlined in our theoretical framework. To test hypotheses H2, H3, and H4, we employed a mediation analysis framework. Model (2) was established to mediate the mediating role of the degree of urban sprawl, green technology innovation, and environmental concern in industrial chain resilience. Based on the practice of [74], we constructed a mediation effect model as follows:

$${mid}_{it}={\alpha }_0+{\alpha }_1{policy}_{it}+{\alpha }_2{X}_{it}+{\mu }_i+{\eta }_t+{\epsilon }_{it}$$

(8)

Descriptive statistics for all of the main variables are presented in Table 2.

Table 2. Descriptive statistics.

$\mathbf{V}\mathbf{a}\mathbf{r}\mathbf{N}\mathbf{a}\mathbf{m}\mathbf{e}$	Obs	Mean	SD	Min	Max
$\mathrm{r}\mathrm{e}\mathrm{s}$	4482	0.021	0.024	0.002	0.151
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	4482	0.027	0.161	0	1
$\mathrm{o}\mathrm{u}\mathrm{d}$	4482	0.005	0.010	0	0.068
$\mathrm{t}\mathrm{s}\mathrm{l}$	4482	3.005	0.748	1.174	4.884
$\mathrm{p}\mathrm{p}\mathrm{l}$	4482	5.842	0.842	2.890	7.167
$\mathrm{d}\mathrm{l}\mathrm{g}\mathrm{i}$	4482	0.170	0.074	0.064	0.423
$\mathrm{s}\mathrm{t}\mathrm{i}$	4482	10.155	1.559	6.415	14.080
$\mathrm{i}\mathrm{s}\mathrm{l}$	4482	0.919	0.678	0.003	9.565
$\mathrm{u}\mathrm{r}\mathrm{b}$	4482	0.508	0.216	0.104	3.206
$\mathrm{h}\mathrm{m}\mathrm{c}$	4482	1.596	2.066	0.021	10.985

Table 2. Descriptive statistics.

$\mathbf{V}\mathbf{a}\mathbf{r}\mathbf{N}\mathbf{a}\mathbf{m}\mathbf{e}$	Obs	Mean	SD	Min	Max
$\mathrm{r}\mathrm{e}\mathrm{s}$	4482	0.021	0.024	0.002	0.151
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	4482	0.027	0.161	0	1
$\mathrm{o}\mathrm{u}\mathrm{d}$	4482	0.005	0.010	0	0.068
$\mathrm{t}\mathrm{s}\mathrm{l}$	4482	3.005	0.748	1.174	4.884
$\mathrm{p}\mathrm{p}\mathrm{l}$	4482	5.842	0.842	2.890	7.167
$\mathrm{d}\mathrm{l}\mathrm{g}\mathrm{i}$	4482	0.170	0.074	0.064	0.423
$\mathrm{s}\mathrm{t}\mathrm{i}$	4482	10.155	1.559	6.415	14.080
$\mathrm{i}\mathrm{s}\mathrm{l}$	4482	0.919	0.678	0.003	9.565
$\mathrm{u}\mathrm{r}\mathrm{b}$	4482	0.508	0.216	0.104	3.206
$\mathrm{h}\mathrm{m}\mathrm{c}$	4482	1.596	2.066	0.021	10.985

4. Results and Discussion

4.1. Baseline Regression Results

In this study, we used the DID model to analyze the effect of CACC on SCR, with the baseline regression results presented in Table 3. Column (1) presents the regression results excluding control variables, column (2) incorporates year-fixed effects, and columns (3) and (4) display the estimation results with the incremental addition of control variables, demonstrating that the estimated coefficient of the policy remains significantly positive. The estimation findings presented in column (4) of Table 3 dominate, followed by a brief examination of the control variables. The estimated coefficient for $ppl$ is significantly positive, indicating that it positively influences SCR. The results show that the CACC significantly promotes SCR regardless of the inclusion of control variables, in addition to the inclusion of time and city fixed effects, thus validating hypothesis H1.

Table 3. Baseline regression.

	Explained Variable: Supply Chain Resilience (Res)
$\mathbf{V}\mathbf{a}\mathbf{r}\mathbf{i}\mathbf{a}\mathbf{b}\mathbf{l}\mathbf{e}$	(1)	(2)	(3)	(4)
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	0.023 ***	0.010 ***	0.010 ***	0.009 ***
	(12.81)	(6.03)	(6.97)	(6.30)
$\mathrm{p}\mathrm{p}\mathrm{l}$			0.096 ***	0.095 ***
			(22.11)	(22.12)
$\mathrm{h}\mathrm{m}\mathrm{c}$			0.004 ***	0.004 ***
			(25.87)	(22.96)
$\mathrm{o}\mathrm{u}\mathrm{d}$			0.012	0.011
			(1.49)	(1.39)
$\mathrm{t}\mathrm{s}\mathrm{l}$			−0.001	−0.000
			(−1.40)	(−0.92)
$\mathrm{d}\mathrm{l}\mathrm{g}\mathrm{i}$			−0.027 ***	−0.022 ***
			(−6.10)	(−3.52)
$\mathrm{s}\mathrm{t}\mathrm{i}$			0.001 ***	0.002 ***
			(2.64)	(4.67)
$\mathrm{g}\mathrm{d}\mathrm{p}$			0.006 ***	0.007 ***
			(7.58)	(4.69)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}$	0.021 ***	0.010 ***	−0.645 ***	−0.666 ***
	(91.20)	(11.76)	(−25.94)	(−20.72)
$\mathrm{City} \mathrm{fixed}$	YES	YES	YES	YES
$\mathrm{Year} \mathrm{fixed}$	NO	YES	NO	YES
N	4482	4482	4482	4482
R2	0.0373	0.2810	0.4440	0.4506

Note: *** 1%, level; the values in parentheses are t statistics.

Table 3. Baseline regression.

	Explained Variable: Supply Chain Resilience (Res)
$\mathbf{V}\mathbf{a}\mathbf{r}\mathbf{i}\mathbf{a}\mathbf{b}\mathbf{l}\mathbf{e}$	(1)	(2)	(3)	(4)
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	0.023 ***	0.010 ***	0.010 ***	0.009 ***
	(12.81)	(6.03)	(6.97)	(6.30)
$\mathrm{p}\mathrm{p}\mathrm{l}$			0.096 ***	0.095 ***
			(22.11)	(22.12)
$\mathrm{h}\mathrm{m}\mathrm{c}$			0.004 ***	0.004 ***
			(25.87)	(22.96)
$\mathrm{o}\mathrm{u}\mathrm{d}$			0.012	0.011
			(1.49)	(1.39)
$\mathrm{t}\mathrm{s}\mathrm{l}$			−0.001	−0.000
			(−1.40)	(−0.92)
$\mathrm{d}\mathrm{l}\mathrm{g}\mathrm{i}$			−0.027 ***	−0.022 ***
			(−6.10)	(−3.52)
$\mathrm{s}\mathrm{t}\mathrm{i}$			0.001 ***	0.002 ***
			(2.64)	(4.67)
$\mathrm{g}\mathrm{d}\mathrm{p}$			0.006 ***	0.007 ***
			(7.58)	(4.69)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}$	0.021 ***	0.010 ***	−0.645 ***	−0.666 ***
	(91.20)	(11.76)	(−25.94)	(−20.72)
$\mathrm{City} \mathrm{fixed}$	YES	YES	YES	YES
$\mathrm{Year} \mathrm{fixed}$	NO	YES	NO	YES
N	4482	4482	4482	4482
R2	0.0373	0.2810	0.4440	0.4506

Note: *** 1%, level; the values in parentheses are t statistics.

When compared to published literature sources, our results corroborate the work of [75], who highlighted the role of climate governance in fostering industrial resilience. In contrast, however, we extend this work by demonstrating the direct impact of urban-level climate adaptation policies on SCR, a dimension less explored in existing research. Similarly, the strong positive association between scientific innovation and SCR reinforces the notion that technological advancements drive industrial upgrading and buffer against external pressures, consistent with findings presented in [76]. From a managerial and policy perspective, these results underscore the importance of integrating climate resilience into urban planning and industrial strategy. Policymakers should consider scaling up CACC initiatives to bolster national supply chain security, particularly in regions vulnerable to climate-induced disruptions. Moreover, the alignment of empirical results with theoretical hypotheses enhances the contribution of this study, providing a validated framework for future research on climate–industry interactions.

4.2. Robustness Test

4.2.1. Parallel Trend Test

The parallel trend test assesses if a significant difference exists in the trends of the target variables between the treatment group and the control group prior to the implementation of the pilot strategy for climate-resilient cities [77]. We employed the event study methodology to further examine the issue by constructing the subsequent model:

$${eff}_{kt}={\alpha }_0+{\sum }_{t=1}^6{\gamma }_{-t}{pre}_{kt}+{\sum }_{t=1}^6{\gamma }_t{post}_{kt}+{\alpha }_1{x}_{kt}+{\omega }_k+{{\eta }_{it}+\epsilon }_{kt}$$

(9)

In Equation (9), t represents the year preceding the adoption of policy k for climate-resilient cities and ${post}_{kt}$ denotes the year t following the implementation of policy k for low-carbon Figure 2 presents the findings of the parallel trend test, utilizing the seventh period preceding CACC as the reference period $\left(current\right)$. None of the coefficient estimates for the periods preceding CACC are statistically significant, indicating that the research sample satisfies the parallel trend test. Furthermore, following the policy’s resilience, the coefficient values are positive and more substantial in the first, second, and fourth periods, indicating that CACC significantly enhances SCR.

4.2.2. Placebo Testing

Individual placebo tests are undertaken to prevent unpredictable, time-varying, and otherwise unobservable individual factors from affecting the estimation results. Based on the methods presented in [78,79], pilot cities are presumed to be randomly generated, policies are allocated, and DID analyses are performed, with the baseline regression analysis replicated 500 times. Figure 3 illustrates that the coefficients of the pilot policies derived from the random sample estimation are predominantly centered around 0, with a significant majority positioned to the left of 0.112, and are approximately normally distributed, thus satisfying the individual placebo test.

4.2.3. Propensity Score Matching (PSM)—DID Method

Figure 4 illustrates the distribution of kernel densities prior to and subsequent to matching. The curves for the experimental and control groups exhibit greater proximity post-matching than pre-matching. Thus, it can be asserted that this method effectively mitigates sample selection bias by eliminating observations that do not conform to the common support assumption and reapplying the baseline regression model to the matched samples for estimation. To address potential sample selection bias and the endogeneity issue associated with the DID model [80], four control variables exhibiting significant coefficients in the baseline regression, namely, population density, human capital, scientific and technological innovation, and economic development level, were initially utilized as covariates. Subsequently, propensity score kernel matching was conducted by employing the logit model for both the treatment and control group cities. The findings presented in column (2) of Table 4 indicate that, following the removal of any sample selection bias, the coefficient of policy remains significantly positive at the 10% level, hence affirming the model’s robustness. The post-matching p-value is not significant, indicating successful balance test results.

4.2.4. Instrumental Variable Approach

To address potential endogeneity from non-random policy placement, we employed a city’s terrain ruggedness as an instrumental variable. This choice satisfies the relevance condition since topographically flatter cities, with more concentrated infrastructure networks, face greater systemic exposure to climate risks like flooding, creating stronger incentives for policy adoption. The first-stage result confirms this relationship, and the geographical nature of the instrument supports the exclusion restriction [81]. The IV estimate in column (3) of Table 4 shows a significant coefficient of 0.068, indicating that CACC increased SCR by about 6.8%. This effect is larger than the baseline DID estimate, suggesting possible negative selection bias whereby more vulnerable cities were prioritized for the pilot. The IV approach corrects this bias and confirms the causal positive impact of CACC on resilience.

4.2.5. Winsorization

To eliminate the impact of extreme values in the explanatory variable, the SCR index, on the regression outcomes, reduced-tail regression is employed to mitigate the interference of outliers on the coefficients [82]. Column (1) of Table 4 presents the regression outcomes of the CACC concerning the contraction of tails from 1% to 99%. The regression coefficient for policy is significantly positive, thereby satisfying the robustness test.

4.2.6. Exclusion of Other Policy Interferences

Throughout the study period, the government concurrently enacted additional pilot policies that could influence the resilience of the industrial chain. To assess the net effect of the pilot policies for CACC, the pilot initiatives for waste-free cities and sponge cities will be incorporated into the multi-period DID model to conduct an interference test of the other policies. Concerning the waste-free city pilot programs, referencing [83], eleven cities were designated as “waste-free city” construction pilots in 2019. The second cohort of waste-free city pilot notifications was released around mid-2022. Thus, the second batch of waste-free city pilots will be deemed influenced by policy shocks from 2022 onwards. Consequently, the second cohort of experimental waste-free towns will be deemed to be influenced by the policy starting in 2022. The findings presented in columns (4) and (5) of Table 4 indicate that, after accounting for the effects of the pilot waste-free cities and sponge cities on SCR, the influence of CACC on SCR remains positive and significant, thereby reaffirming the robustness of the baseline regression results.

4.2.7. Addition of Control Variables

Incorporating control variables can mitigate the issue of omitted variables and diminish the impact of endogeneity on the results, hence enhancing the models’ accuracy and dependability [84]. We additionally incorporate two control variables: infrastructure level and financial self-sufficiency. The results presented in column (6) of Table 4 indicate that the estimated coefficients of policy remain significantly positive following the inclusion of new control variables, and the significance of the other variables remains largely unchanged, suggesting that the coefficients of the new control variables align with theoretical expectations. The above findings indicate that, upon evaluating other potential factors, the conclusions of this paper remain valid, and the results are highly resilient.

4.2.8. Changing the Time Window Period

The selection of the window period preceding and succeeding the policy may influence the regression outcomes, as there may be delayed effects from the shocks induced by the policy. To enhance the robustness of the regression results and mitigate errors arising from sample selection, the time frames in the baseline regression were modified. Two distinct window periods, (2010–2022) and (2012–2020), were superimposed on the foundational window period to assess the extent of variations in policy impacts across time. The results presented in Columns (7–8) of Table 4 demonstrate that the treatment effects across several time frames align with the results of the baseline regression, suggesting the robustness of the findings.

Table 4. Robustness checks.

Variables	Winsor2	PSM-DID	IV 2SLS	Other Policy Interference		Increase Control Variable	Adjustment Window
	1~99%			Zero-Waste City	Sponge City		2010~ 2022	2012~ 2020
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	0.008 ***	0.0076 ***	0.068 **	0.007 ***	0.008 ***	0.007 ***	0.007 ***	0.004 ***
	(7.89)	(7.66)	(2.14)	(6.89)	(7.50)	(7.21)	(7.14)	(5.14)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}$	−0.156 ***	−0.1726 **	−0.306 ***	0.010 ***	0.010 ***	−0.280 ***	−0.300 ***	−0.307 ***
	(−20.51)	(−23.82)	(−19.01)	(18.16)	(17.09)	(−12.61)	(−10.25)	(−9.45)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}$	YES	YES	YES	YES	YES	YES	YES	YES
$\mathrm{C}\mathrm{i}\mathrm{t}\mathrm{y}$ $\mathrm{f}\mathrm{i}\mathrm{x}\mathrm{e}\mathrm{d}$	YES	YES	YES	YES	YES	YES	YES	YES
$\mathrm{Y}\mathrm{e}\mathrm{a}\mathrm{r}$ $\mathrm{F}\mathrm{i}\mathrm{x}\mathrm{e}\mathrm{d}$	YES	YES	YES	YES	YES	YES	YES	YES
N	4482	4482	2683	4482	4482	4482	3237	2241
R2			0.5179	0.4640	0.3952	0.5372	0.4970	0.4816

Note: ***, **, and * indicate significance at the 1%, 5% and 10% levels; the values in parentheses are t statistics.

Table 4. Robustness checks.

Variables	Winsor2	PSM-DID	IV 2SLS	Other Policy Interference		Increase Control Variable	Adjustment Window
	1~99%			Zero-Waste City	Sponge City		2010~ 2022	2012~ 2020
	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	0.008 ***	0.0076 ***	0.068 **	0.007 ***	0.008 ***	0.007 ***	0.007 ***	0.004 ***
	(7.89)	(7.66)	(2.14)	(6.89)	(7.50)	(7.21)	(7.14)	(5.14)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}$	−0.156 ***	−0.1726 **	−0.306 ***	0.010 ***	0.010 ***	−0.280 ***	−0.300 ***	−0.307 ***
	(−20.51)	(−23.82)	(−19.01)	(18.16)	(17.09)	(−12.61)	(−10.25)	(−9.45)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}$	YES	YES	YES	YES	YES	YES	YES	YES
$\mathrm{C}\mathrm{i}\mathrm{t}\mathrm{y}$ $\mathrm{f}\mathrm{i}\mathrm{x}\mathrm{e}\mathrm{d}$	YES	YES	YES	YES	YES	YES	YES	YES
$\mathrm{Y}\mathrm{e}\mathrm{a}\mathrm{r}$ $\mathrm{F}\mathrm{i}\mathrm{x}\mathrm{e}\mathrm{d}$	YES	YES	YES	YES	YES	YES	YES	YES
N	4482	4482	2683	4482	4482	4482	3237	2241
R2			0.5179	0.4640	0.3952	0.5372	0.4970	0.4816

Note: ***, **, and * indicate significance at the 1%, 5% and 10% levels; the values in parentheses are t statistics.

4.3. Mechanism Test

Following the confirmation of a substantial and sustained positive impact of the pilot policy for CACC on SCR, an analysis of its underlying mechanisms is conducted, focusing on urban sprawl, environmental awareness, and green innovation. These mechanisms reflect the core dimensions of city–industry resilience coupling, whereby urban form, public–corporate cognition, and innovation systems interact to shape the adaptive capacity of industrial chains [85]. Consequently, hypothesis H2 is evaluated based on the development of a mediation effect model. The results of the mechanism analysis are shown in column (1) of Table 5, where CACC significantly increases the degree of sprawl, and increased urban sprawl further strengthens SCR. The results presented in column (1) of Table 5 indicate that, after accounting for the influence of sprawl on industry chain resilience, the estimated coefficient of $policy$ is significantly positive at 0.005, and urban sprawl is an important channel for CACC to enhance SCR. The hypothesis H2 is confirmed, indicating that in CACC, the transition from chaotic growth to structured and resilient expansion, while concurrently enhancing urban climate resilience during spatial development, transforms the “passive response to climate risk” into the active utilization of spatial expansion to bolster resilience. This spatial reorganization transforms industrial structures from centralized monocentric models into distributed polycentric networks. By dispersing economic activities across multiple hubs, it significantly lowers the vulnerability associated with geographic concentration while enhancing the overall capacity of urban-industrial systems to withstand climate disruptions. Consequently, CACC fosters efficient urban sprawl, which in turn facilitates the transition of industrial organization from monocentric agglomeration to polycentric and networked distribution, thereby preventing the excessive concentration of critical links in the industrial chain within the primary urban area and significantly bolstering urban resilience. CACC transforms chaotic urban expansion into structured, resilience-focused development by promoting polycentric urban layouts. This approach strategically relocates climate-vulnerable industrial segments, such as moving water-sensitive manufacturing to well-drained suburbs, while retaining high-value functions in central areas. Such spatial reconfiguration prevents single climate events from paralyzing entire supply chains, embodying the shift from passive risk response to active resilience building proposed by the authors of [86]. For managers in this field, the above findings underscore the importance of dispersing critical supply chain components across climate-adapted locations to mitigate concentration risks.

While our mediation analysis identifies three transmission channels between CACC and SCR, we acknowledge potential endogeneity concerns. Although fixed effects mitigate time-invariant confounders, unobserved time-varying factors could still affect both policy implementation and mechanism variables. Furthermore, while comprehensive, our composite resilience indices may not fully capture supply chains’ dynamic adaptation to recurring disruptions. The authors of future studies would benefit from employing more granular resilience metrics and robust causal identification designs.

Regarding green technological innovation, the policy coefficient for green technology innovation in column (2) of Table 5 is 0.619 and statistically significant, indicating that CACC promotes industrial chain resilience by stimulating green patent applications and low-carbon technology adoption. This finding supports hypothesis H3 and aligns with existing evidence on how environmental policies drive sustainable innovation [87], thus highlighting a key technological pathway in the city–industry resilience coupling, whereby climate-adaptive urban policies stimulate green innovation ecosystems that directly enhance the industrial chain’s capacity for upgrading and risk absorption. CACC can facilitate the transition of the industrial chain towards low-carbonization and sustainability, invigorate the stakeholders within the industrial chain to engage in green innovation, and encourage the integration of additional green resources into the development of the industrial chain, thereby bolstering SCR [88].

Regarding environmental awareness, the estimated coefficient of policy in column (3) of Table 5 is 3.921, which meets the 5% significance threshold. Environmental awareness also plays a pivotal role, with the policy elevating governmental and public attention to climate issues. This heightened awareness translates into stricter regulations, corporate green transformations, and increased investment in resilient infrastructure, collectively enhancing SCR [89]. This cognitive mechanism underscores how city-level climate governance aligns corporate and public priorities with resilience goals, fostering a shared responsibility for industrial chain stability and creating a coordinated city–industry response system. Specifically, through reinforcement of the government’s stringent approach to environmental governance and its regulatory measures [90], thereby validating H4.

Table 5. Mechanism decomposition.

	Urban Sprawl	Green Technology Innovation	Environmental Concerns
	(1)	(2)	(3)
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	0.005 *	0.169 **	3.921 **
	(1.66)	(2.54)	(2.38)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}$	1.393 ***	−4.320 *	52.968
	(11.78)	(−1.66)	(0.82)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}$	YES	YES	YES
$\mathrm{N}$	2241	2241	2196
${\mathrm{R}}^2$	0.4839	0.6605	0.1325

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels; the values in parentheses are t statistics.

Table 5. Mechanism decomposition.

	Urban Sprawl	Green Technology Innovation	Environmental Concerns
	(1)	(2)	(3)
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	0.005 *	0.169 **	3.921 **
	(1.66)	(2.54)	(2.38)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}$	1.393 ***	−4.320 *	52.968
	(11.78)	(−1.66)	(0.82)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}$	YES	YES	YES
$\mathrm{N}$	2241	2241	2196
${\mathrm{R}}^2$	0.4839	0.6605	0.1325

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels; the values in parentheses are t statistics.

4.4. Heterogeneity Test

4.4.1. Heterogeneity of Urban Locations

We further examined regional heterogeneity by grouping cities according to their position relative to the Hu Huanyong Line. Our regression results indicate that CACC significantly enhances SCR in southeastern cities; in comparison, its effect is negligible in northwestern regions. This divergence stems from their distinct economic foundations. Southeastern cities benefit from established industrial agglomeration and innovation ecosystems, where climate adaptation measures can readily integrate with existing production networks and knowledge spillovers [91]. In contrast, northwestern cities often exhibit fragmented industrial structures and weaker inter-firm connections, limiting the diffusion and synergistic impact of resilience policies. These findings underscore that the effectiveness of CACC depends critically on regional economic structure and industrial density.

4.4.2. Heterogeneity of Resource Endowments

To examine how urban resource endowments shape the effect of CACC on industrial chain resilience, we categorize cities into resource and non-resource types following official classifications. The regression results presented in columns (3) to columns (4) of Table 6 show that CACC significantly strengthens SCR in non-resource cities, with a coefficient of 0.01 at the 1% level. In contrast, the policy shows a negative and significant effect in resource-based cities. This divergence likely stems from structural differences. Non-resource cities typically host diverse and interconnected industries, enabling cross-sector synergy and smoother integration of climate adaptation measures into existing upgrading strategies [92]. Resource-based cities, however, often face industrial lock-in and fragmented funding priorities, which weaken the urban–industrial resilience link. These results underscore that industrial structure and policy alignment critically shape how effectively climate adaptation translates into SCR.

Table 6. Heterogeneity analysis.

	North-Western Provinces	South-Eastern Provinces	Resource-Based City	Non-Resource-Based Cities	“Two Control” Cities	Non-“Two-Control” Cities
	(1)	(2)	(3)	(4)	(5)	(6)
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	0.001	0.007 ***	−0.003 **	0.010 ***	0.002	0.011 ***
	(0.002)	(0.001)	(0.001)	(0.001)	(0.002)	(0.001)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}$	−0.283 ***	−0.262 ***	−0.143 ***	−0.233 ***	−0.355 ***	−0.202 ***
	(0.029)	(0.033)	(0.025)	(0.037)	(0.045)	(0.026)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}$	YES	YES	YES	YES	YES	YES
$\mathrm{C}\mathrm{i}\mathrm{t}\mathrm{y}$ $\mathrm{f}\mathrm{i}\mathrm{x}\mathrm{e}\mathrm{d}$	YES	YES	YES	YES	YES	YES
$\mathrm{Y}\mathrm{e}\mathrm{a}\mathrm{r}$ $\mathrm{f}\mathrm{i}\mathrm{x}\mathrm{e}\mathrm{d}$	YES	YES	YES	YES	YES	YES
$\mathrm{N}$	1224	3258	1602	2880	1980	2502
${\mathrm{R}}^2$	0.610	0.521	0.7820	0.4350	0.5700	0.4530

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels; the values in parentheses are t statistics.

Table 6. Heterogeneity analysis.

	North-Western Provinces	South-Eastern Provinces	Resource-Based City	Non-Resource-Based Cities	“Two Control” Cities	Non-“Two-Control” Cities
	(1)	(2)	(3)	(4)	(5)	(6)
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	0.001	0.007 ***	−0.003 **	0.010 ***	0.002	0.011 ***
	(0.002)	(0.001)	(0.001)	(0.001)	(0.002)	(0.001)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{s}\mathrm{t}\mathrm{a}\mathrm{t}$	−0.283 ***	−0.262 ***	−0.143 ***	−0.233 ***	−0.355 ***	−0.202 ***
	(0.029)	(0.033)	(0.025)	(0.037)	(0.045)	(0.026)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}$	YES	YES	YES	YES	YES	YES
$\mathrm{C}\mathrm{i}\mathrm{t}\mathrm{y}$ $\mathrm{f}\mathrm{i}\mathrm{x}\mathrm{e}\mathrm{d}$	YES	YES	YES	YES	YES	YES
$\mathrm{Y}\mathrm{e}\mathrm{a}\mathrm{r}$ $\mathrm{f}\mathrm{i}\mathrm{x}\mathrm{e}\mathrm{d}$	YES	YES	YES	YES	YES	YES
$\mathrm{N}$	1224	3258	1602	2880	1980	2502
${\mathrm{R}}^2$	0.610	0.521	0.7820	0.4350	0.5700	0.4530

Note: ***, **, and * indicate significance at the 1%, 5%, and 10% levels; the values in parentheses are t statistics.

4.4.3. Heterogeneity of Environmental Regulation

Based on China’s environmental policy, we categorize cities as “two-control” if they belong to designated acid rain or sulfur dioxide control zones [93]. The regression results presented in columns (5) to columns (6) of Table 6 show that CACC significantly improves SCR in non-two-control cities; in comparison, its effect in two-control cities is not significant. This divergence may stem from the dual financial burden faced by industries in two-control areas, which must simultaneously fund emission reduction and climate resilience measures, thereby diluting resource concentration and policy effectiveness. In contrast, non-two-control cities can focus resources on resilience-building, allowing the policy to achieve a more targeted impact. The above findings suggest that overlapping regulatory pressures may weaken the city–industry resilience link, underscoring the need for coherent environmental governance.

4.5. Theoretical Implications and Boundary Conditions

The results of our heterogeneity analysis reveal that the effect of CACC on SCR depends strongly on local economic structure and geography. The policy proves more effective in non-resource-based cities and those in southeastern China, where diversified industrial systems enable cross-sector collaboration and smooth integration of green innovations. In contrast, resource-based cities often face industrial lock-in and fragmented funding, limiting CACC’s impact without deeper structural transformation [94]. The regional gap further highlights the role of agglomeration: dense networks in the southeast amplify policy benefits, whereas the northwest’s sparse economic geography offers fewer synergies. These results suggest CACC works best in diversified, well-connected areas and requires tailored support where structural conditions are less favorable.

5. Further Analysis

While the results of our baseline analysis confirm that CACC strengthens local SCR, modern supply chains are deeply interconnected across cities. This finding suggests that policy impacts likely extend beyond a single city’s borders, potentially spilling over to neighboring areas through knowledge sharing, industrial cooperation, and policy imitation. To fully evaluate the policy’s reach, we examine its spatial effects in the following section. We first applied Moran’s I to verify spatial dependence in SCR, a prerequisite for spatial econometric analysis. Our results showed that Moran’s I index rose from 0.212 in 2010 (Figure 5) to 0.233 in 2020 (Figure 6), indicating increasing spatial clustering. Regions with weaker adaptive capacity and simpler industrial structures continued to exhibit “low-low” clustering across both years, though spatial spillovers grew more pronounced. This pattern suggests that CACC, aided by technology diffusion and industrial synergy, not only enhances local chain resilience but also generates positive feedback to neighboring regions. We then use a Spatial Durbin Model (SDM) to quantify both the direct effects in pilot cities and indirect spillover effects on their neighbors, offering a more complete appraisal of the policy’s efficacy [95].

Note: This comparative visualization of spatial autocorrelation shows that SCR became increasingly clustered between 2010 and 2020. The rise in Moran’s I confirms a strengthening of positive spatial spillovers, where cities with high (or low) resilience are more likely to be located near other cities with similarly high (or low) resilience levels.

The CACC enhances SCR, potentially encouraging neighboring regions to emulate it. Consequently, further investigation into the spatial effects of the CACC on SCR is warranted [96]. As shown in column (1) of Table 7, the Spatial Durbin Model (SDM) regression results indicate significantly positive policy coefficients, consistent with the baseline regression, thus confirming that CACC substantially enhances industrial chain resilience. Using a nested matrix of geographic and economic distance weights, the CACC policy coefficients remain significant at the 1% level, suggesting that adoption of CACC in nearby cities also benefits local SCR. Enterprises tend to enhance their competitive capabilities and learn from pilot cities’ climate resilience experiences, thereby strengthening local supply chains. Further decomposition of the spatial effects presented in columns (2) and (3) reveals that CACC exerts not only a significant direct effect on SCR but also a positive indirect effect on neighboring cities at the 5% significance level. This finding reflects inter-city imitation and competition spurred by the pilot policy, which supports industrial chain transformation in surrounding areas and generates beneficial spatial spillovers.

Table 7. Spatial Effects Regression.

Variable	SDM (1)	Direct (2)	Indirect (3)	Total (4)
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	0.006 ***	0.006 ***	0.005 **	0.012 ***
$\mathrm{W}\mathrm{\ast }\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	(6.47) 0.002	(6.71)	(2.11)	(4.32)
	(0.90)
$\mathrm{r}\mathrm{h}\mathrm{o}$	0.342 ***
	(13.54)
$\mathrm{s}\mathrm{i}\mathrm{g}\mathrm{m}\mathrm{a}2\_\mathrm{e}$	0.000 ***
	(47.00)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}$	YES
$\mathrm{N}$ R2	4482 0.4211

Note: ***, **, and * indicate significance at the 1%, 5% and 10% levels; the values in parentheses are t statistics.

Table 7. Spatial Effects Regression.

Variable	SDM (1)	Direct (2)	Indirect (3)	Total (4)
$\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	0.006 ***	0.006 ***	0.005 **	0.012 ***
$\mathrm{W}\mathrm{\ast }\mathrm{p}\mathrm{o}\mathrm{l}\mathrm{i}\mathrm{c}\mathrm{y}$	(6.47) 0.002	(6.71)	(2.11)	(4.32)
	(0.90)
$\mathrm{r}\mathrm{h}\mathrm{o}$	0.342 ***
	(13.54)
$\mathrm{s}\mathrm{i}\mathrm{g}\mathrm{m}\mathrm{a}2\_\mathrm{e}$	0.000 ***
	(47.00)
$\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}$	YES
$\mathrm{N}$ R2	4482 0.4211

Note: ***, **, and * indicate significance at the 1%, 5% and 10% levels; the values in parentheses are t statistics.

6. Conclusions and Recommendations

In this study, we assess the impact of CACC on SCR using a double difference model based on panel data from 249 Chinese cities from 2006 to 2023. The results show that the policy can significantly enhance SCR. This conclusion still holds after a series of robustness tests such as the parallel trend test, PSM-DID, and placebo test. The mechanism analysis results reveal that the policy works mainly through three paths: enhancing environmental concern, promoting green technology innovation, and optimizing urban spatial expansion. Heterogeneity analysis further reveals that the effects of the policy are more pronounced in the cities southeast of the Hu Huanyong Line, non-resource cities, and cities that are not in the “two control zones”. In addition, the results of the spatial effect test confirm the positive spatial spillover effect of the policy, which not only enhances the SCR of the pilot cities but also drives the development of the neighboring areas through inter-city interactions.

At the academic level, the findings of this study provide the first systematic empirical support for the causal relationship between climate-resilient city building and SCR in the Chinese context, incorporating the issue of climate resilience into the framework of industrial–economic analysis and adopting a causal identification strategy combined with SDM to clarify the three types of transmission mechanisms, namely, environmental concerns, green innovations, and the spatial structure of cities. At the practical level, the study results provide policymakers with a clear basis for decision-making, pointing out that differentiated promotion strategies should be formulated in accordance with regions and city types [97,98].

This study has some limitations that must be acknowledged. At the data level, the analysis only focuses on cities at the prefecture level and lacks evidence at the micro-scale. In addition, the assessment of indicators covers multidimensional capabilities, but it is still a static assessment, which makes it challenging to capture the dynamic responses of supply chains to shocks. Although the mechanism path has been examined, its inherent endogeneity still needs to be further explored [99]. In the future, we can introduce enterprise-level or high-dimensional geographic data, construct a multi-layer “city–industry–enterprise” analysis framework, map the climate risk supply chain to simulate extreme weather shocks, and conduct cross-country comparative studies to enhance the universality and policy value of the conclusions. This is also a cross-country comparative study, which enhances the generalizability and policy value of the conclusions [100].

Although this study is based on the Chinese context, its core findings have important global implications, particularly for cities in the Global South, which face the dual challenges of rapid urbanization and climate risk. China’s strong top-level facilitation and infrastructure investments have enabled rapid policy implementation [101]. For other regions, the key is not to replicate policy forms but to draw on their governance logic, for example, by prioritizing investment in economic hubs with diversified industries and significant agglomeration effects and prioritizing public communication and community engagement as important supports for policy effectiveness, in order to build resilient supply chain systems under their own institutional and resource conditions.