Disaster Response Mechanisms for Key Technology Innovation in China’s Emergency Industry Under the New National System

Abstract

The emergency industry refers to a comprehensive industrial system of products, technologies, and services aimed at preventing, responding to, and mitigating emergencies. The emergency industry is primarily oriented toward disaster prevention and mitigation, providing direct support to enhance societal resilience. Given the frequent occurrence of natural disasters and the strategic layout of global emergency technologies, it is of great practical significance to study how the science and technology systems of disaster-prone countries respond. Based on the theories of disaster economics and innovation geography, this paper constructs a mediation effect model to investigate how China improves the key technological capabilities of its emergency industry through three response pathways—demand stimulation, technological advancement, and educational enhancement—following natural disasters. The stepwise testing approach, which integrates the mediation effect model with the spatial Durbin model, consists of three stages. The first stage tests the total effect model to assess how disasters impact local key technologies and their spatial spillover on adjacent regions. The second stage examines the direct influence of disasters on the three pathways and their spatial spillover using the mediator equation. The third stage uses the outcome equation with the mediator to evaluate how the pathways affect local key technologies and neighboring regions after controlling for disaster impacts. We offer both theoretical insights and empirical evidence to support specialized research on technological diffusion induced by disasters. The result shows that although the direct negative impact of disasters is inevitable, the institutional advantages of China’s emergency rescue and innovative collaborative efforts have played a significant role in promoting key technologies. Under the new national system, China is progressively establishing a spatial framework wherein emergency products are allocated across regions, key technologies are synergistically integrated, and the development of emergency-related disciplines is promoted through regional collaboration in response to the frequent occurrence of natural disasters. This demonstrates that the advancement of key technologies in China’s emergency industry is significantly supported by inter-regional cooperation and linkage mechanisms.

1. Introduction

The current global situation of natural disasters is complex, with compound disasters causing more severe losses and showing distinct regional characteristics. Humanity engages in profound reflection on its relationship with nature. In response, people are working to make breakthroughs in science and technology, aiming to bridge the gap between innovation and emergency response practices to reduce the negative impact of disasters. The emergency industry plays a crucial role in these efforts, requiring significant technical investments [1,2]. The emergency industry refers to the industrial system developed to prevent, respond to, and recover from various public emergencies, encompassing the integrated provision of products, services, and technologies to ensure social security and stability. Key emergency technologies are essential for accurately predicting and responding to disasters. They enable swift recovery based on local needs, help save lives, minimize damage to disaster relief infrastructure, and reduce economic losses [2]. Emergency technologies also play a crucial role in maintaining regional stability and national security, positioning them as a key focus for developing advanced productive forces. From a global perspective, countries and regions such as the United States, Europe, Japan, and Australia have successfully implemented diverse versions of the “whole nation system” to effectively address and prevent public emergencies. These countries leverage a comprehensive emergency industry technology support system to focus on key technology research, aiming to take the lead in global emergency science and technology innovation under the backdrop of climate change. The key emergency technologies primarily encompass the advanced application of next-generation information technology in critical emergency management functions, including risk assessment, preparedness and prevention, monitoring and early warning, as well as disaster and accident response and rescue [3]. In this new era, the “whole nation system” has been tasked with the goal of achieving key technological breakthroughs in core areas [4]. The new national system emphasizes the integration of market orientation and government coordination, concentrating efforts on overcoming core technological challenges and promoting independent innovation and breakthroughs in key strategic technological fields. China is currently at a pivotal point in achieving its medium- and long-term strategic goal of comprehensive disaster alleviation. Significant breakthroughs have been made in the fundamental theories and key technologies of emergency industry, allowing China to maintain resilient development despite the impacts of catastrophic events [5]. However, there is still a significant gap between the advancement of technology and the practical requirements for effective disaster response [6].

The emergency industry encompasses a broad range of activities, services, and products designed to manage and respond to unexpected situations that pose immediate risks to life, property, and the environment [7]. The emergency industry is primarily oriented toward disaster prevention and mitigation. Existing studies have largely overlooked the technical impact of natural disasters on the emergency industry, and there are limited empirical methods to assess how various countries and regions implement technical response measures within the industry. In light of the current competitive landscape in international emergency technologies and the practical demands of disaster response, exploring from both theoretical and empirical angles how China addresses these challenges under the new national system holds substantial significance. This paper aims to offer actionable recommendations for disaster-prone countries to focus their resources and efforts on overcoming key technological challenges in the emergency industry, thereby fostering a robust collaborative framework for advancing key technology breakthroughs. Our research focuses on rapidly restoring the output of emergency key technologies after disasters and provides concrete evidence to support humanity’s efforts in addressing frequent disasters and climate change through technological innovation.

2. Literature Review

There has been considerable attention given to how innovation systems respond to exogenous shocks [8]. Natural disasters align with the concept of exogenous shocks because of their unpredictability and sudden occurrence [9]. The focus of this paper closely relates to the literature on how innovative systems respond to exogenous shocks, specifically natural disasters. Existing studies have explored the direct impact of natural disasters on technology and the transmission mechanisms of human responses. In comparison to the transmission mechanisms, the literature on the direct impact has reached relatively consistent conclusions. Some scholars have highlighted the force majeure losses caused by disasters, illustrating the negative effects of exogenous shocks on technology and discussing how natural disasters hinder innovation [10], often considering factors like the level of economic development [11], trade openness [12], financial openness [13] and government stability [14]. However, there are significant differences in the research paradigms and indicators used to study the response transmission mechanisms, with some studies even reaching contradictory conclusions. Existing literature primarily examines the impact of natural disasters on technological innovation through three transmission paths: technological demand, the education pathway, and the innovation effect [15,16].

In terms of technological demand, the destructive impact of disasters, as an exogenous variable, forces businesses to adjust their supply of specialized emergency products in response to changes in the external market environment. This, in turn, can either stimulate or hinder innovation within enterprises [17,18,19]. Therefore, the impact of disasters on emergency innovation is not absolute, and significant variations can occur depending on the market environment. The innovation effect only emerges when certain local factors related to the disaster reach a specific threshold [20].

The study of the innovation effect is rooted in Schumpeter’s theory of creative destruction. When natural disasters and other destructive events occur, enterprises are driven to pursue technological innovation as a means of survival [20]. Disaster economics emphasizes that while the losses from natural disasters cannot be fully avoided, humans can still benefit from these events, with technological advancements playing a key role in mitigating the damage. With the rapid innovation of emerging technologies such as the Internet, big data, cloud computing, artificial intelligence, and blockchain, disaster emergency practices have raised higher demands for digital technologies. This, in turn, has driven the advancement of emergency technologies [21].

The education pathway indicates that the occurrence of disasters can enhance the willingness of affected populations to acquire emergency knowledge and skills, thereby better preparing them for future disaster scenarios [16]. The literature of the education path can be traced back to endogenous growth theory. The destruction caused by disasters to capital or the labor force may accelerate the accumulation of human capital, thereby stimulating technological growth and boosting productivity during the subsequent reconstruction phase [13]. Governments frequently enhance the scope of training, recruitment, and investment in professionals within the disaster relief sector. These professionals play a crucial role in supporting and driving innovative activities, particularly in disaster response and recovery efforts [15].

There is limited direct evidence on the interaction among the three pathways. In a few studies, Crespo [20] mentioned that the impact of disaster-induced disruptive innovation on technology transfer varies among countries with different market environments, which are highly related to institutions and supply-demand relationships. From this information, we may infer that the influence of disruptive innovation on local innovation is affected by local market demand, and the extent of this influence varies depending on the local market demand environment. In the research on the transmission effects of different pathways, the use of interaction terms is rarely involved, as this might lose the clear understanding of each pathway and thus affect the reliability of research conclusions [21].

On the other hand, the literature closely related to the theme of this paper focuses on the national innovation system. The concept of a national innovation system (NIS) is a framework used to understand and promote the interactions between various actors and institutions that contribute to technological innovation and economic development within a country [22]. It emphasizes the relationships between government, universities, and industry as key components driving innovation performance and has a significant influence on the allocation of innovation resources [23]. China’s innovation system is called the new national system. A substantial body of literature has demonstrated that the “new national system,” established under centralized leadership of the Communist Party of China, exhibits remarkable political and institutional strengths in addressing uncertain shocks, including severe disasters [24]. Many scholars have substantiated that the new national system is a critical factor enabling China’s industries to maintain technological resilience and sustain development in the face of external shocks [25]. The new mechanism, by leveraging digital and intelligent technologies alongside a unified large-scale market, not only effectively integrates national resources but also addresses the dual challenges of “market failure” and “government failure” in the governance of scientific and technological innovation [26,27,28].

The contributions of this paper to the literature include the following aspects. First, existing research has shown that technical demands, innovation effects, and educational initiatives collectively function as key strategies to mitigate the impact of disasters via technological means, serving as intermediary mechanisms through which disasters influence technological advancement. However, the literature remains fragmented, lacking a comprehensive examination of how these mechanisms interact. Based on the harm-to-benefit transformation theory in disaster economics, this study integrates three types of mechanisms to examine how China, under the new national system, employs technological policies to address the fundamental national condition of frequent disasters. Second, the existing literature on how innovation systems respond to the technological impacts of disasters predominantly focuses on general technological innovation, while largely neglecting key technologies. Meanwhile, the emergency industry plays a crucial role in directly supporting disaster response efforts; however, there is currently a lack of literature that explicitly connects disasters with technologies specific to the emergency industry. Key technologies that hold a central role and possess strategic importance within the emergency technology system warrant significant attention in today’s era of global scientific and technological revolution. Thus, we are the first to conduct an empirical study that tests direct impact of disasters on key technologies and the intermediate transmission mechanisms, taking the emergency industry as the research sample. Finally, as far as our knowledge, there is a lack of research on the diffusivity of technology specifically induced by disasters, while previous literature only generally acknowledges that technology can be diffusive. When disasters occur, the strong and effective coordination capability of China’s central government in allocating emergency resources across regions is likely to significantly influence the demand for technology. In order to solve the potential issue of technology diffusion, we introduce spatial factors to investigate the impact of disasters on local key technologies and their spatial spillover effects on adjacent areas. The technology diffusion effect triggered by disasters has been empirically validated from the perspective of regional interconnection.

3. Theoretical Basis and Research Hypothesis

The emergency industry embodies the dual attributes of ensuring “emergency” support and generating “industrial” economic benefits [29]. Theoretically, we analyze the destructive impact of disasters on key technologies from the perspective of industrial development and expounded on the development opportunities brought to this industry by disasters from the perspective of emergency support. Furthermore, this paper integrates the theoretical implications of the new national system into the context of key emergency technology responses to natural disasters. By integrating the demands, technologies, and educational pathways discussed in the literature, a specific research framework is developed, as illustrated in Figure 1.

3.1. The Direct Impact of Natural Disasters on Key Technologies in the Emergency Industry

The emergency industry consists of numerous enterprises. Natural disasters represent a significant source of operational risks for Chinese enterprises, and such risks constitute a critical factor that must be considered when enterprises make decisions regarding breakthroughs in key technologies [30]. The Action Plan for the Development of Key Areas in Emergency Industry (2023–2025), released in September 2023, highlights the importance of industry chain security for key technologies [31]. The direct impact of natural disasters on the critical technologies within the safety emergency industry can be systematically analyzed from an industrial chain perspective, focusing on three key dimensions: disaster-bearing entities, synergy between the technology and industrial chains, and financial crowding-out effect. Firstly, enterprises serve as the most fundamental disaster-bearing units and key nodes in the industry chain. When disasters damage the chain-master enterprises in affected areas, this will result in the collapse of the entire innovation system [32]. Secondly, the industry chain acts as the foundation for the technology chain’s survival, representing the materialization of technology across each node of the industry chain. Key technologies serve as the connection hub that links all elements of the technology chain, ensuring smooth integration and functionality throughout the system [33]. When extreme disasters cause physical damage to the affected areas, it can exacerbate the instability of the technology chain, which is built upon the industry chain. This disruption can impact the key technology hub nodes within the technology chain. The instability of the industrial chain will feed back to the technology chain, forming a negative cycle. Thirdly, most key technologies emphasize the front-end basic research of both the innovation and industry chains. The government’s ongoing investment in innovative resources—such as talent, knowledge, technology, and R & D funding—is crucial for ensuring the achievement of key technological breakthroughs [34]. After a disaster occurs, although new investment may generate a certain “update-driven effect” that promotes technological iteration, local governments face constraints in fiscal resources, which must be prioritized for disaster prevention, mitigation, and post-disaster recovery efforts. This reallocation of public expenditure, driven by the fiscal crowding-out effect, diminishes funding for long-term research and development of core technologies, thereby undermining sustained basic research and extended R & D cycles at the front end of the industrial chain. Consequently, breakthroughs in key core technologies are significantly hindered [35]. This kind of path dependence will solidify the structure of fiscal expenditure, squeeze research funds for a long time and weaken the innovation incentives at the institutional level [36].

Hypothesis 1:

Natural disasters exert a negative impact on the level of key technologies in the emergency industry.

3.2. Demand Path

The demand path is characterized by an increase in the demand for emergency products and services as a direct result of disasters. The expansion of the market size encourages enterprises and other entities to invest more resources into the organized research and development of key technologies, often with government support. The demand for emergency response products and services driven by disasters can be seen in two key aspects. On one hand, the formation and growth of the emergency industry is rooted in China’s comprehensive industrial system [29] and aligns with the general principles of industrial growth. As a result, people’s demand for emergency support has increased. As disaster-related losses accumulate to a significant extent, the resulting human-environment conflicts can exacerbate modern social risks. Furthermore, China’s vast territory and the resulting geographical and climatic diversity have led to a broader social demand for specialized emergency products and services. Such demands encompass prediction and early warning systems, personal protective equipment, communication technologies, as well as medical rescue capabilities [37]. On the other hand, in response to public safety events such as natural disasters, China established the Ministry of Emergency Management in 2018, reforming and improving the emergency management system and mechanisms. This has created a favorable political environment for the advancement of the emergency industry. The government plays a central role as the primary purchaser of emergency products and services. China’s government-led disaster relief model, which includes rapid response, emergency command, early warning, additional testing, emergency information dissemination, and evacuation efforts, has significantly driven the widespread societal demand for rescue and relief products [38].

As the fundamental unit of the national economy, enterprises are the most dynamic innovators of key technologies. Since the 19th century, planned and organized innovation within enterprises has become a hallmark of scientific and technological development across various countries [39]. Major breakthroughs in emergency key technologies represent one of the critical application domains for the new national system. For emergency industries with a strong public service nature, the Chinese government has always played a leading and safety-net role. On one hand, within a favorable political environment for the emergency industry, the large-scale demand for emergency products and services triggered by disasters enables enterprises to generate profits from their investments in safety technologies [29]. The increasing profitability allows enterprises to boost their R & D investment, thereby enhancing their ability to achieve breakthroughs in key technologies. On the other hand, building upon their existing technological endowments, enterprises systematically expand their production capacity and market reach through iterative trial-and-error processes and the detailed analysis of technological workflows, thereby validating the efficacy of technological breakthroughs. Enterprises systematically iterate through the processes of technological upgrading and capacity expansion, progressively realizing the iterative advancement of key technological innovation [40].

Hypothesis 2:

The occurrence of disasters boosts the demand for emergency products and services, thereby enhancing the level of key technologies in the emergency industry.

3.3. Technology Path

Schumpeter’s theory of creative destruction suggests that in the face of disruptive events, enterprises are driven to innovate for survival. Within the framework of endogenous growth theory, the destruction of capital or labor caused by disasters can stimulate technological development during the subsequent reconstruction phase. Post-disaster technological advancement can contribute to mitigating short-term losses, with the pace of such progress often contingent upon the level of pre-disaster investment in research and development. Natural disasters are frequent external destructive shocks. In the era of the new round of technological revolution and industrial transformation characterized by intelligence and informatization, disasters expedite the evolution of emergency technologies, driving the integration of digital technologies into emergency practices [41]. Disasters motivate the development and integration of new technologies into disaster management, such as big data, machine learning, and artificial intelligence, which are typically key emergency technologies for enhancing situational awareness and response times [42]. Moreover, digitalization is the foundation for promoting the application and development of artificial intelligence and machine learning technologies in emergency industry [41]. Theoretically, the low search costs and high feedback efficiency enabled by digitalization facilitate the dissemination of technology diffusion information across the upstream and downstream of the industrial chain, thereby significantly mitigating information asymmetry among various entities within the innovation system. Digitalization empowers innovation entities to more effectively ascertain the breakthrough directions of critical technologies [43]. In practical applications, disasters act as catalysts for the advancement of digital technologies in the emergency industry by highlighting the need for rapid, efficient, and effective communication and decision-making tools [44]. A variety of cases have demonstrated the stimulating impact of disasters on the advancement of key technologies within the emergency industry from diverse application domains. Below are some common and specific examples. Disasters have accelerated the integration of key digital technologies into disaster prevention and mitigation, as evidenced by the widespread deployment of smart city emergency systems and artificial intelligence tools in real-world scenarios [45]. Information communication technology (ICT) plays a crucial role in supporting effective communication and decision-making during disasters by enhancing the cognitive capacity of emergency managers [46]. Real-time communication tools facilitate better route optimization for emergency response units, reducing transportation time and improving the overall efficiency of emergency response [47]. The use of drones for aerial data collection and glocalization has improved disaster response and damage assessment [48]. The implementation of integrated information systems that combine data from multiple sources can streamline emergency management processes and improve coordination among different agencies [44]. The implementation of integrated information systems that combine data from multiple sources can streamline emergency management processes and improve coordination among different agencies [49].

Hypothesis 3:

Disasters exert a positive reverse effect on the advancement of key digital technologies within the emergency industry. The impact of disasters on the key technologies of the emergency industry is mediated and manifested through digitalization.

3.4. Education Path

Under the new national system, technological breakthroughs, education supply, and talent development work together to address the nation’s major strategic needs. In this process, higher education plays a pivotal role by integrating human resources and guiding key technology breakthroughs, helping to align the country’s innovation efforts with its long-term development goals [50]. All supporting departments involved in emergency rescue must possess professional training knowledge and access to training opportunities across all stages of disaster management—early warning, rescue, recovery, and reconstruction. Academic certification and education play a key role in enhancing the overall professional competence of emergency personnel, ensuring a well-prepared and capable workforce [51]. In terms of academic certification, China began offering undergraduate programs in emergency management in 2005 [52], and since then, the government has targeted the expansion of emergency talent cultivation, introduction, and investment. In 2012, the Ministry of Education supported qualified universities and colleges in establishing independent emergency management programs. By integrating resources from disaster science, emergency engineering, and related disciplines, while effectively leveraging local endowment advantage, these institutions have not only expanded their emergency management programs but also established new schools dedicated to this field. This has led to a significant increase in the number of professionals with specialized disaster knowledge and enhanced the level of scientific research and technology in the field [52]. Additionally, under the leadership of the government, high-level research universities, innovative enterprises, and other entities have collaborated closely in technological innovation. This synergy has led to a continuous emergence of key technological achievements in disaster prevention and mitigation, driving further advancements in the field [15].

Hypothesis 4:

The frequent occurrence of natural disasters drives China to enhance the key technology level of the emergency industry through emergency discipline construction.

3.5. Spatial Effect

Geographic proximity and technology networks are often cited as factors that explain the spillover effects of key technologies. The regional distribution of emergency industry development is closely tied to the frequency and severity of local disasters. Additionally, the disaster prevention and mitigation mechanisms, along with scientific and technological innovations in this field, demonstrate distinct regional characteristics, which reflect the specific needs and challenges faced by different regions [53]. The White Paper on China’s Safety Emergency Industry Cluster in 2022 highlights that, with guidance from governments at all levels and active participation from enterprises, the emergency industry has developed a distinct regional cluster layout, featuring numerous key technologies [31]. On the one hand, the agglomeration of the emergency industry accelerates technology spillover, supplying key technological elements to the innovation-driven enterprises. Simultaneously, it facilitates the diffusion of innovative knowledge and high-level technology professionals to neighboring underdeveloped regions. This not only enhances the innovation capabilities of various entities but also promotes the coordinated development of technological elements [1], thereby establishing stable inter-regional technology chains and industrial chains across regions. Key technologies stand at the hub of connecting technology chain with industry chain [32]. The inter-provincial flow of technical factors triggered by disaster shocks may impart a spatial effect to the emergency key technologies of neighboring provinces. On the other hand, in response to disaster impacts, the government-led new national system can quickly mobilize and allocate national resources for post-disaster relief. The emergency industry plays a crucial role by providing specialized products and services for safety prevention, emergency preparedness, monitoring, and early warning. When disasters strike, they inevitably affect the supply and demand in the emergency industry’s macro market in neighboring areas. This stimulates demand for products and services, offering innovative entities—such as governments and enterprises—the opportunity to test whether their technology iteration products significantly improve disaster prevention and control efficiency. Numerous rounds of technological trial and error contribute to breakthroughs in key technologies.

Hypothesis 5:

The occurrence of natural disasters generates a spatial spillover effect on the key technologies of the emergency industry in adjacent provinces.

4. Data and Methodology

4.1. Model Setting

To investigate the direct impact of disasters on key technologies in the emergency industry and the responses under China’s new national system, the following econometric model is formulated:

$$\ln ectec{h}_{it}={\beta }_0+{\beta }_1\ln disaste{r}_{it}+{\alpha }_1\ln X_{it}+{\lambda }_i+{\epsilon }_{1it}$$

(1)

$$\ln me{d}_{it}={\mu }_i+b\ln disaste{r}_{it}+{\alpha }_2\ln X_{it}+{\epsilon }_{2it}$$

(2)

$$\ln ectec{h}_{it}={\varphi }_i+{\beta }_2\ln me{d}_{it}+cdisaste{r}_{it}+{\alpha }_3\ln X_{it}+{\epsilon }_{3it}$$

(3)

Here, $ectech$ is the key technology level of the emergency industry; $disaster$ represents the disaster loss variable, $X$ stands for all control variables in the model, $\lambda$ is the fixed effect, $med$ is the mediator variable, $b$ is the influence coefficient of natural disaster on the mediator variable; $c$ represents the impact of natural disasters on the key technologies of the emergency industry, while controlling for the mediator variable. $\epsilon$ is the random disturbance term. We use panel data, where i represents the index of different provinces, and t represents the time dimension. All variables are taken in logarithms to reduce the influence of heteroscedasticity.

4.2. Data and Variables

4.2.1. Dependent Variable: Key Technology Level of the Emergency Industry ($ectech$)

The identification methods of key technologies can fall into the four categories: qualitative expert experience evaluation, quantitative patent indicators, patent networks and text mining. The quantification of key technologies can be achieved by integrating patent indicators with text mining techniques [45]. On the one hand, patents fall into the three categories: invention, utility model and external design. Their applicants are diverse and not limited to enterprises. Among them, invention patents is marked with a higher authorization threshold, and serve as new technical solutions in their fields, which have been deemed by many scholars as proxy variables of key technology-related concepts such as bottleneck technology, key technology and disruptive technology [54]. On the other hand, annual reports and research reports of enterprises have fixed text formats, and structured quantitative text analysis is made to quantify non-patent innovation activities of enterprises. However, both patent index and text mining have their weaknesses. Patents may have the weakness of quantity and quality mismatch, and the annual report texts are only limited to measuring innovation entities acting as an enterprise. Therefore, the combination of the two methods can overcome their weaknesses and enhance the accuracy of quantifying the level of key technologies.

Patent quantification. Numerous key products and services of the emergency industry are listed in the Chinese strategic emerging industries [1]. We refer to the indirect measurement method of the National Bureau of Statistics of China for the strategic emerging industries. The added value is calculated based on key industries and enterprises. Listed companies represent the core competitiveness of China’s economy. Therefore, there is some feasibility to measure the key technology level of the emergency industry by taking listed companies as samples. Drawing the research findings of Liu Chengliang [54], we conducted a cross-referencing analysis between the “Reference Relationship Table of Strategic Emerging Industries Classification and International Patent Classification (2021)” and the “Guidance Catalogue of Classification for Emergency Industry (2021 Edition)”. Consequently, we identified that China’s emergency industry encompasses the following sub-sectors of manufacturing. Specifically, it encompasses the specialized equipment manufacturing industry (C36), the transportation equipment manufacturing industry (C37), the electrical machinery and equipment manufacturing industry (C39), the computer, communication and electronic equipment manufacturing industry (C40), and the instrument and meter manufacturing industry (C41) within the manufacturing sector (C). Based on the above classification, we selected patents with a value higher than 8 in IncoPat as the proxy variable for key technologies in the patent aspect. Then, they are sorted out and collected according to the applicant’s province [54].

Text quantification. Since emergency statistics or the list of enterprises have not been made public in China, it is necessary to independently screen listed enterprises that offer emergency products and services. First and foremost, we conducted a search for the characteristic keywords that comply with the “Guidance Catalogue for the Classification of Emergency Industry (2021 Edition)” of the Ministry of Industry and Information Technology and the “Classification of Strategic Emerging Industries (2018)” of the National Bureau of Statistics of China. Through this process, 692 listed emergency enterprises were determined. Second, we searched for the brokerage research reports of these 692 enterprises on the Individual Stock Research Report section of Eastmoney.com from 2000 to 2022. We found a total of 9689 research reports from 692 enterprises during the period from 2000 to 2022. In the third step, referring to Bellstam G’s research we conducted LDA topic modeling on 9689 brokerage research reports [55]. Based on the theme-based interpretability, 15 themes were obtained. Then, on the China National Knowledge Infrastructure (CNKI), we conducted keyword searches with the themes of “strategic emerging industries”, “new quality productivity”, “bottleneck technology” and “emergency technology”, and referred to highly cited papers in the Chinese Social Sciences Citation Index to calculate the KL divergence. By minimizing the probability of the search topic words to filter out the themes most relevant to “key technologies”, we identified the seventh topic word and extracted the probabilities of the top 30 key technology topic words from it. See

Table 1. Words and distribution probabilities for key technology topics (top 30).

Key Words	Frequency	Key Words	Frequency	Key Words	Frequency
Forewarning	0.01512	Man-machine	0.00461	Communication	0.00339
Celerity	0.00878	Security and protection	0.00457	Space	0.00318
Merge	0.00823	Prediction	0.00435	Synergy	0.00315
Cloud	0.00757	Urgency	0.00431	Detection	0.00308
Evaluating	0.00538	Operation	0.00414	Dynamic	0.00293
Green	0.00532	Satellite	0.00412	Automation	0.00292
Platform	0.00512	Decision-making	0.00402	Intelligence	0.00292
Digitization	0.00494	Network	0.00375	Infrastructure	0.00284
Transducer	0.00484	Modularization	0.00361	Real-time	0.00283
Monitor	0.00467	internet	0.00352	Rescue	0.00276

for these 30 words. Finally, by combining the probability with the enterprise text data, we can calculate the key technology scores by accumulating and averaging the scores from all enterprise reports within the same province and year. These scores will serve as a measurement index for the key technologies of listed emergency companies.

$$I\left(f,g\right)={\displaystyle \int f\left(x\right)}{\log }_e\left({\displaystyle \frac{f\left(x\right)}{g\left(x|\theta \right)}}\right)dx$$

(4)

$I\left(f,g\right)$ represents the KL divergence, which is used to measure the information loss between the word frequency distribution $g$ of the references and the modeling result distribution $f$ of topic words. The most appropriate value is selected from a series of parameters to minimize KL divergence.

Given the discrepancies in nature, dimension, and order of magnitude between the number of patents and the text score, these two indicators are initially subjected to standardization using Equation (5) for Z-scores. In the process of data standardization, to mitigate the occurrence of negative values following Z-score transformation, which could potentially influence subsequent analyses and interpretations, a moderate shift was applied to ensure that all data points remain within the positive range. Existing literature indicates that empirical values on the lower end typically fall within the 0–0.1 range. As such, this study adopts 0.05 as the mean after the shift. This approach not only preserves the stability of the data distribution but also enhances the comparability and consistency of the results’ interpretation. The original values of the patent and text indicators are shown in the spatiotemporal distribution of representative years in Figure 2. Based on provincial text scores and high-quality patent data, China’s emergency industry key technologies exhibit clear temporal evolution and spatial differentiation. In the temporal dimension, both text scores and high-quality patent counts have continued to rise, yet significant regional disparities remain: eastern coastal provinces (such as Shandong, Jiangsu, and Zhejiang) lead in growth, with Zhejiang’s text score increasing from 0.22 to 0.48 and its patent count rising from 2801 to 4750. Western provinces (such as Ningxia and Tibet), although showing relatively rapid growth, still maintain comparatively low absolute levels. In terms of spatial distribution, high-value clusters have formed in the Beijing–Tianjin–Hebei region, the Yangtze River Delta, and the Pearl River Delta; the central and western regions display a “multi-core driving” pattern, with Sichuan and Shaanxi standing out, while provinces such as Qinghai and Gansu remain relatively weak. Overall, the development of China’s emergency industry key technologies demonstrates a gradient pattern of “eastern leadership, central rise, and western catch-up,” highlighting the need to further optimize mechanisms for regional collaborative innovation. In view of the advantages and disadvantages of patents and texts in measuring key technologies, the entropy method is selected to combine the two indicators. There are a total of 31 provinces, with patent and text data spanning 23 years. Therefore, the overall data matrix size is 713 × 2. First step, the proportion of each sample on indicator j should be calculated, as shown in Formula (6). The second step is to calculate the entropy value as shown in Formula (7). Due to the large sample values, the probability distribution becomes smoother after normalization, resulting in entropy values close to 1, specifically 0.9652 and 0.9707. The third step is to calculate the redundancy $d_j$, which represents the effective amount of information, as shown in Formula (8). The last step is to calculate the weights, as shown in Formula (9). The formula of the entropy method is shown in Equation (10). According to the measurement, the patent index has the weight of 0.753, and the text measurement index has the weight of 0.247.

$$Z={\displaystyle \frac{x-\stackrel{-}{x}}{\sigma }}$$

(5)

$x$ stands for the observed value, $\stackrel{-}{x}$ for the mean and $\sigma$ for the standard deviation.

$$p_{ij}={\displaystyle \frac{x_{ij}}{{\displaystyle \sum _{i=1}^n{x}_{ij}}}}$$

(6)

$p_{ij}$ is the proportion of each sample on indicator j.

$$e_j=-k{\displaystyle \sum _{i=1}^n{p}_{ij}}\ln (p_{ij})$$

(7)

$e_j$ represents the entropy value, where $k={\displaystyle \frac{1}{\ln (n)}}$

$$d_j=1-e_j$$

(8)

$$w_j={\displaystyle \frac{d_j}{{\displaystyle \sum _{j=1}^m{d}_j}}}$$

(9)

ectech = 0.753 patent index standardization + 0.247 text index standardization

(10)

4.2.2. Primary Explanatory Variable ($disaster$)

Some indicators such as disaster economic loss, affected or dead population, and frequency of disasters are used to measure the degree of disaster loss. However, the above indicators exhibit significant discrepancies in their attributes, dimensions, and orders of magnitude. Typically, a single indicator is insufficient to accurately reflect the impact level of natural disasters. Therefore, drawing on Joo’s research, we applied the objective entropy weight method to assign weights to the aforementioned three indicators and construct a comprehensive disaster loss index [56].

4.2.3. Mediator Variables

Emergency products and services ($eps$). China has not released the statistical data on the number of emergency products and services available at the provincial level. According to the specific products and services listed in the Ministry of Industry and Information Technology’s Guidance Catalogue for the Classification of Emergency Industry (2021 edition), we matched the listed companies, small-giant enterprises and high-tech enterprises in the CNRDS database with the emergency business scope of enterprises in the website of www.qcc.com. Qichacha is a commercial public database, whose data is sourced from the National Enterprise Credit Information Publicity System of China. It enjoys high credibility and reference value and is widely used in economic-related research [57]. A search was conducted using the four primary categories, 21 medium-sized categories, and 117 subcategories outlined in the “Guidance Catalogue for the Classification of the Safety and Emergency Industry (2021 Edition)” as keywords. An OR logic search was applied across fields such as enterprise name, introduction, business scope, products, and trademarks. The target enterprises were those registered as active and operational as of 31 December 2023. Finally, the list of emergency enterprises is determined, and the revenue of such enterprises is allocated and aggregated at the provincial level, which makes it possible to gain the proxy variables of the products and services of the emergency industry in each province.

Digitalization ($tech$). We selected the number of digital economy invention patents applied for in the Digital Economy Research Database in the same year as an indicator to measure the digitalization level of each province. In the database, the accurate matching and processing of patent data is performed in the three aspects: digital innovation, digital industry and digital platform. Finally, the total number of digital economy invention patents is used as a proxy variable of technology path.

Emergency discipline construction ($ecc$): In essence, an emergency response to natural disasters is also a matter of disaster management and a response to a social security event, which is sufficient to establish first-level and second-level disciplines [58]. Since 2005, Chinese universities, colleges, and research institutes have leveraged resources from their respective advantageous disciplines to systematically establish emergency management disciplines [52]. In 2012, the Chinese Ministry of Education supported qualified universities to independently set up relevant majors in respect of emergency management. In 2020, The Office of Academic Degrees Committee of the State Council released the Notice on Promoting Some degree-granting Units to Strengthen the Construction of Emergency Management Disciplines, supporting the establishment of second-level disciplines of emergency management under the first-level discipline of public management. With reference to the Results of Registration and Approval of undergraduate majors in colleges and Universities from 2013 to 2023, we combined with the emergency management talent training plan of public administration majors in colleges and universities, and finally obtained the number of schools offering emergency discipline in each province over the years.

4.2.4. Control Variable

Institutional adaptability. Institutional adaptability under the new national system refers to the degree of organic integration between China’s meta-institutions and variable economic institutions. The meta-institutional framework comprises China’s ownership system, distribution system, and the socialist market economy system. Institutional adaptability can ignite the creativity and innovation capacity of various entities, thereby elevating the ability to confront uncertain shocks in specific application scenarios [1]. For one thing, in the face of disaster shocks, government departments play a regulating role by means of government investment, tax incentives, financial incentives, transfer payments and targeted aid. For another, Enterprises can strengthen their ability to fulfill economic and social responsibilities by harnessing the complementary strengths of state-owned and social capital. With reference to the research results of Wang Susu, this paper constructs the indicators of government departments’ institutional adaptability ($aeg$) and enterprises’ institutional adaptability ($aee$) [59].

Fiscal decentralization and urban scale. Local government behavior and urban scale can also affect key technologies of the emergency industry. With reference to the research of Liu, the principal component analysis method is applied to determine fiscal decentralization. Fiscal decentralization ($fpd$) represents a proxy variable of local government behavior. The per capita gross regional product ($pgrp$) and the number of high-tech enterprises ($nltc$) are used as proxy variables of city size [54].

Internationalization strategy ($I\mathrm{s}$). An internationalization strategy can systematically integrate domestic and international technological resources, thereby expanding access to strategic resources in frontier science and technology and enhancing the global core R & D capabilities of enterprises, universities, and other entities in scientific innovation. With reference to the measurement index of key technology internationalization strategy proposed by Song Yan et al., the number of leading enterprises engaging in international business through subsidiaries, establishing overseas production bases, and conducting acquisitions and investments in international markets is considered as an indicator of enterprise innovation internationalization [60]. In order to directly indicate the impact on the emergency industry, we gain internationalization information from the selected annual reports of listed emergency enterprises and also acquire the internationalization data of non-listed emergency enterprises from the website of www.qcc.com.

Disaster sensitivity ($dsi$). Sensitivity acts as a core factor in evaluating regional disaster vulnerability, reflecting the extent to which the human-earth system is prone to sustain damage when exposed to disaster shocks. The regional distribution of the emergency industry’s development is highly correlated with the frequency and severity of local disasters. Disaster prevention and control technologies still have regional characteristics [53]. Therefore, in this paper, the disaster sensitivity index is selected as a control variable affecting the key technology of emergency industry. This indicator is sourced from the Climate Risk module (CRRD) of the Chinese Research Data Services Platform (CNRDS). It comprises fundamental dimensions including population density, building density, and engineering value.

Industry-university-research ($per$). Cooperation among industry, academia and research institutions shares the cost and risk arising from the breakthrough process of key technologies. With reference to the research results of Liu, we regard joint patent applications as a proxy indicator for the integration of industry, academia, and research [61]. Patents jointly filed by enterprises, universities, and research institutions are defined as collaboration patents. We excluded the data of utility model and design patents from the PatSnap website. Firstly, we carried out a search with the keywords “disaster, emergency, disaster prevention, disaster mitigation and disaster relief”, and then expanded the range of the above keywords by utilizing the keyword assistant tool. Then, the quantity of patents in the field of natural disaster emergency industry cooperation is obtained as the industry-university-research variable in accordance with the main gist of this paper.

The above data comes from China City Statistical Yearbook, Provincial and municipal Statistical Yearbook, National Tibetan Plateau Data Center, Eastmoney.com, WIND database, CNRDS database, IncoPat IPR database, PatSnap and the website of www.qcc.com from 2000 to 2022.

Table 2. Descriptive statistics of variables.

Types of Variables	Variables	Observations	Max	Min	Average	SD
dependent variable	$ectech$	713	0.82	0.05	0.31	0.18
Core variable	$disaster$	713	7865	0	111.74	316.05
Mediator variable	$eps$	713	399.863	0.548	16.45	20.54
	$tech$	713	32,269	5	2228.2	4453.27
	$ecc$	713	86	1	11.16	13.9
Control variable	$aeg$	713	0.95	0.002	0.12	0.12
	$aee$	713	0.95	0.11	0.55	0.20
	$fpd$	713	0.47	0.03	0.24	0.1
	$pgrp$	713	190,313	1018	30,206.77	30,821.96
	$nltc$	713	12,372	5	831.36	1383.65
	$Is$	713	11,042	2	1038.38	1485.47
	$dsi$	713	1	0.001	0.09	0.21
	$per$	713	461	1	31.68	38.35

provides the descriptive statistics for each variable.

5. Empirical Analysis and Results

5.1. Benchmark Regression Analysis

Table 3. Results of benchmark regression.

Variables	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6	Model 7	Model 8	Model 9
$\ln disaster$	−0.096 *** (0.008)	−0.097 *** (0.01)	−0.032 *** (0.009)	−0.031 *** (0.009)	−0.401 *** (0.016)	−0.063 *** (.015)	−0.076 *** (0.025)	−0.015 *** (0.004)	−0.227 *** (0.007)
$\ln aeg$		0.048 (0.061)	0.078 *** (0.174)	0.105 *** (0.019)	0.113 *** (0.021)	0.075 *** (0.018)	0.093 *** (0.018)	0.015 ** (0.008)	0.013 * (0.008)
$\ln aee$			1.133 *** (0.063)	1.124 *** (0.063)	1.001 *** (0.051)	0.41 *** (0.161)	0.135 (0.155)	0.931 *** (0.042)	0.784 *** (0.971)
$\ln fpd$				0.055 *** (0.031)	0.061 *** (0.03)	0.062 ** (0.028)	0.072 *** (0.026)	0.036 *** (0.012)	0.032 *** (0.012)
$\ln pgrp$					0.023 (0.015)	0.053 *** (0.019)	0.093 *** (0.018)	0.002 (0.005)	0.009 * (0.005)
$\ln Is$						0.177 *** (0.037)	0.181 *** (0.032)	0.046 *** (0.01)	0.036 *** (0.012)
$\ln nltc$							0.078 *** (0.024)	0.059 *** (0.006)	0.068 *** (0.007)
$\ln dsi$								0.304 *** (0.007)	0.296 *** (0.009)
$\ln per$									0.041 *** (0.012)
Intercept	−0.612 * (0.033)	−0.073 * (0.047)	−0.192 *** (0.009)	−0.205 *** (0.01)	−0.13 *** (0.05)	−0.718 *** (0.127)	−0.691 *** (0.129)	−0.317 *** (0.033)	−0.359 *** (0.051)
Fixed effect	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Yes
$R^2$	71.93%	71.93%	92.01%	92.06%	92.16%	92.91%	98.8%	98.8%	98.85%
Observations	713	713	713	713	713	713	713	713	713

Note: The results in parentheses are robust standard errors; *, ** and *** indicate the statistical significance at the time of p < 0.1, p < 0.05 and p < 0.01, respectively.

presents the benchmark regression results. Based on the constructed basic model (1), the panel regression model is used to test the impact of natural disasters on key technologies of the emergency industry. With an aim of ensuring the stability of key variables, this paper has gradually introduced control variables to probe into the relationship between natural disasters and key technologies of the emergency industry under different configuration variables.

The estimated results from Model 1 to Model 9 exhibit strong robustness across various specifications. After the stepwise inclusion of control variables, the regression coefficients of disaster losses and control variables did not change significantly. When accounting for all independent variables, a 1% increase in disaster losses is associated with a 0.227% reduction in the key technologies of the emergency industry. In the absence of considering the mediating effect, it is indicated that the total effect of disaster shocks on the key technologies of the emergency industry is negative, validating the negative assumption of the direct impact of natural disasters in hypothesis 1. The 8 control variables are basically significantly positive in the model, manifesting that institutional adaptability efficiency, fiscal decentralization, urban scale, internationalization strategy, disaster risk and industry-university-research institute cooperation all exert a positive effect on the key technologies of the emergency industry. This shows that national support, urban size and scientific & technological development level play an increasingly prominent role in incubating the innovation activities of the emergency industry [54].

5.2. Analysis of Mediating Effect

Based on the aforementioned theoretical analysis, under the new national system, the frequent occurrence of natural disasters has created a strong demand from both the government and society for a larger quantity of products and services in the emergency industry, as well as for advancements in digital technologies and the development of emergency disciplines. This demand, in turn, has triggered positive effects on the key technologies of the emergency industry. The breakthrough innovations in the emergency industry arise from the organic integration of these three approaches. This paper further adopts empirical data to examine the intermediary transmission mechanism of these three approaches. The mediating effect is typically assessed using two methods: test of individual regression coefficients and coefficient product. Specifically, the test methods for the coefficient product include the Sobel-Goodman and Bootstrap tests. The mediating effect model is shown in

Table 4. Results of mediating effect analysis.

Variable	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
	$\mathbf{ln}\mathit{e}\mathit{p}\mathit{s}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$	$\mathbf{ln}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{c}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$
$\ln eps$		0.111 *** (0.006)
$\ln tech$				0.032 *** (0.007)
$\ln ecc$						0.050 *** (0.008)
$\ln disaster$	0.724 *** (0.055)	−0.002 * (0.002)	1.277 *** (0.082)	−0.019 *** (0.006)	0.655 *** (0.056)	−0.017 *** (0.005)
Control variable	Yes	Yes	Yes	Yes	Yes	Yes
Fixed effect	Yes	Yes	Yes	Yes	Yes	Yes
$adj.R^2$	94.98%	99.37%	90.05%	99.75%	98.88%	98.94%
Observations	713	713	713	713	713	713
Sobel	−0.014 *** (0.001)		−0.003 *** (0.001)		−0.002 *** (0.001)
Mediating effect	Partial mediating		Partial mediating		Partial mediating
Bootstrap	Significance of the indirect effect		Significance of the indirect effect		Significance of the indirect effect

Note: The results in parentheses are robust standard errors; *, ** and *** indicate the statistical significance at the time of p < 0.1, p < 0.05 and p < 0.01, respectively.

.

The significantly positive coefficient in Model 1 confirms that the disaster shock has heightened the demand for emergency products and services directly aimed at disaster prevention, mitigation, and relief. The regression coefficient of emergency products and services on the key technologies in the estimation results reported by Model 2 is also significantly positive. The negative impact of natural disasters on the key technologies is offset by the positive mediating effect of the demand for products and services in the industry, which passes the Sobel test and plays a partial mediating role. Therefore, hypothesis 2 is verified. The estimation results of Model 3 confirm that natural disasters prompt traditional innovation to transition towards digitalization, and the urgent demand of disaster emergency response generates a reverse force effect on the key technologies of the emergency industry. In Model 4, the regression coefficient of digital technologies on the key technology is significantly positive, which also offsets the negative impact of natural disasters on the key technologies of the emergency industry, thereby further verifying hypothesis 3. Sobel test results indicate that digital technology plays a partial mediating role in the relationship between natural disasters and innovation transformation. The coefficient shown in Model 5 is significantly positive, which verifies that China’s universities and research institutes have strengthened the emergency management discipline construction and continuously set up emergency management-related majors so as to better deal with frequent natural disasters. In the estimated results reported by Model 6, the regression coefficient of the emergency management discipline construction on the key technology of the emergency industry is significantly positive, and the coefficient of disaster loss is negative, which passes the Sobel test and plays a partial mediating role. Therefore, hypothesis 4 is verified. The Bootstrap test results in Models 1–6 are consistent, and the mediating effect is robust. In conclusion, under the framework of the new national system, to better address disaster prevention and control, China can effectively facilitate the development of key technologies in the emergency industry by enhancing the demand for emergency products and services, driving digital technological innovation, and advancing the construction of emergency-related disciplines.

5.3. Spatial Econometric Analysis

Natural disasters and innovative activities both exhibit distinct regional characteristics. Therefore, the demand for emergency products and services, digital technologies, and the construction of emergency disciplines not only directly influence the breakthrough innovation of the local emergency industry but also absorb the innovative resources from surrounding areas. This can trigger an indirect impact on the key technology innovation activities in the neighboring areas through siphoning and agglomeration effects. On the basis of the spatial autocorrelation test, this paper combines the spatial factors with the regression coefficient test methods in the mediating effect. Firstly, we use the dynamic panel space Durbin model (SDM) to analyze the spatial spillover effect. Then, LR and Wald are further used to test whether SDM degenerates into spatial autoregressive model (SAR) or spatial error model (SEM). When constructing the dynamic spatial Durbin model, it is crucial to determine the appropriate lag order of the spatial term. To this end, we estimated models with first-, second-, and third-order spatial lags, and compared their Akaike Information Criterion (AIC) and Bayesian Information Criterion (BIC) values. As shown in

Table 5. Comparison of AIC and BIC for different lag models.

Model Specification	AIC	BIC
First-order Lag Model	1250.36	1295.82
Second-order Lag Model	1268.91	1315.44
Third-order Lag Model	1285.77	1334.52

, the first-order lag model yields the lowest AIC and BIC values, indicating the best trade-off between model fit and parsimony. Therefore, we adopt the first-order dynamic spatial Durbin model as the baseline specification for subsequent empirical analysis.

The intermediate stepwise test model with spatial terms is as follows:

$$\begin{array}{l}\ln ectec{h}_{it}={{\beta }_0}^{\prime }+{\rho }_{1}W\ln ectec{h}_{it}+{{\beta }_1}^{\prime }\ln disaste{r}_{it}+{\delta }_{1}W\ln disaste{r}_{it}+{{\alpha }_1}^{\prime }\ln X_{it}\\ +{\varphi }_{1}W\ln X_{it}+{{\lambda }_i}^{\prime }+{{\epsilon }_{1it}}^{\prime }\\ \ln me{d}_{it}={{\mu }_i}^{\prime }+{\rho }_{2}W\ln me{d}_{it}+b^{\prime }\ln disaste{r}_{it}+{\delta }_{2}W\ln disaste{r}_{it}+{{\alpha }_2}^{\prime }\ln X_{it}+\\ {\varphi }_{2}W\ln X_{it}+{{\epsilon }_{2it}}^{\prime }\\ \ln ectec{h}_{it}={{\varphi }_i}^{\prime }+{\rho }_{3}W\ln ectec{h}_{it}+{\beta }_2\ln me{d}_{it}+{\rho }_{4}W\ln me{d}_{it}+c\ln disaste{r}_{it}+\\ {\delta }_{3}W\ln disaste{r}_{it}+{\alpha }_3\ln X_{it}+{\varphi }_{3}W\ln X_{it}+{{\epsilon }_{3it}}^{\prime }\end{array}$$

Here $W$ is the spatial weight matrix. We select reciprocal geographical distance matrix ($W1$) and economic distance matrix ($W2$) of per capita gross regional product to estimate the spatial autocorrelation.

5.3.1. Spatial Autocorrelation Test

The Moran index values are shown in

Table 6. The Moran’s I values of the main variables.

Variables	$\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$		$\mathit{d}\mathit{i}\mathit{s}\mathit{a}\mathit{s}\mathit{t}\mathit{e}\mathit{r}$		$\mathit{e}\mathit{p}\mathit{s}$		$\mathit{t}\mathit{e}\mathit{c}\mathit{h}$		$\mathit{e}\mathit{c}\mathit{c}$
	$W1$	$W2$	$W1$	$W2$	$W1$	$W2$	$W1$	$W2$	$W1$	$W2$
2000	0.038 ***	0.9088 ***	0.058 ***	0.1758 ***	0.18 ***	0.298 ***	0.0968 ***	0.2968 ***	−0.106	−0.281
2001	0.1358 ***	0.9128 ***	0.0418 ***	0.1938 ***	0.0998 ***	0.3128 ***	0.0888 ***	0.2978 ***	0.0958 ***	−0.2938 ***
2002	0.1518 ***	0.924	0.0478 ***	0.1978 ***	0.1058 ***	0.2628 ***	0.0868 ***	0.2598 ***	0.109	−0.2558 ***
2003	0.154 ***	0.945 ***	0.013 ***	0.206 ***	0.096 ***	0.285 ***	0.048 ***	0.294 ***	0.096 ***	−0.276 ***
2004	0.141 ***	0.968 ***	0.037 ***	0.209 ***	0.103 ***	0.293 ***	0.083 ***	0.306 ***	0.099 ***	−0.268 ***
2005	0.136 ***	0.979 ***	0.05 ***	0.224 ***	0.097 ***	0.267 ***	0.084 ***	0.22 ***	−0.098 ***	−0.28 ***
2006	0.147	0.98 ***	0.076 ***	0.247 ***	0.107 ***	0.285 ***	0.106 ***	0.27 ***	0.098 ***	−0.297 ***
2007	0.151 ***	0.981 ***	0.065 ***	0.227 ***	0.097 ***	0.27 ***	0.094 ***	0.263 ***	0.084 ***	−0.263
2008	0.163 ***	0.983 ***	0.076 ***	0.235 ***	0.1 ***	0.293 ***	0.098 ***	0.286 ***	0.084 ***	−0.305
2009	0.139 ***	0.992 ***	0.033 ***	0.23 ***	0.088 ***	0.279 ***	0.061 ***	0.292 ***	−0.077 ***	−0.264
2010	0.147 ***	0.988 ***	0.032 ***	0.243 ***	0.083 ***	0.288 ***	0.066 ***	0.293 ***	0.068 ***	−0.273
2011	0.151 ***	0.989 ***	0.048 ***	0.245 ***	0.089 ***	0.289 ***	0.072 ***	0.277 ***	0.072 ***	−0.26
2012	0.168 ***	0.992 ***	0.062 ***	0.253 ***	0.083 ***	0.279 ***	0.077 ***	0.258 ***	−0.071 ***	−0.265 ***
2013	0.167 ***	0.996 ***	0.055 ***	0.267 ***	0.083 ***	0.289 ***	0.069 ***	0.263 ***	0.074 ***	−0.268 ***
2014	0.178 ***	0.992 ***	0.043 ***	0.271 ***	0.079 ***	0.299 ***	0.063 ***	0.299 ***	0.075 ***	−0.29 ***
2015	0.181 ***	0.995 ***	0.044 ***	0.27 ***	0.074 ***	0.278 ***	0.061 ***	0.268 ***	−0.069 ***	−0.279 ***
2016	0.164 ***	0.986 ***	0.083 ***	0.256 ***	0.093 ***	0.253 ***	0.086 ***	0.253 ***	−0.086 ***	−0.251
2017	0.177 ***	0.999 ***	0.069 ***	0.236 ***	0.077 ***	0.272 ***	0.073 ***	0.258 ***	−0.073 ***	−0.257 ***
2018	0.177 ***	1 ***	0.05 ***	0.272 ***	0.043 *	0.304 ***	0.047 ***	0.304 ***	−0.051	−0.291 ***
2019	0.181 ***	1.002 ***	0.091 ***	0.246 ***	0.069 ***	0.28 ***	0.078 ***	0.27 ***	−0.091 ***	−0.253 *
2020	0.169 ***	1.005 ***	0.063 ***	0.266 ***	0.052 ***	0.297 ***	0.067 ***	0.293 ***	−0.066 ***	−0.284 ***
2021	0.177 ***	1.009 ***	0.094 ***	0.275 ***	0.055 ***	0.323 ***	0.078 ***	0.303 ***	−0.082 ***	−0.313 ***
2022	0.187 ***	1.011 ***	0.09 ***	0.288 ***	0.047 ***	0.281 ***	0.07 ***	0.286 ***	−0.077 ***	−0.285 ***

Note: The results in parentheses are robust standard errors; *, ** and *** indicate the statistical significance at the time of p < 0.1, p < 0.05 and p < 0.01, respectively.

. The Moran’I index of the explained variable, the mediator variable and the core explanatory variable is almost significantly positive at least at the level of 10%, while the Moran’I index of emergency discipline construction is almost negative. The Moran’I indexes of the key technology and natural disaster loss in the emergency industry present an upward trend, which means that the key technology and natural disasters in China’s emergency industry are increasingly concentrated in few regions, and the phenomenon of spatial polarization is becoming more prominent.

5.3.2. The Results of the Spatial Econometrics

Firstly, we evaluate and compare the spatial lag model and the spatial error model through the LM test.

Table 7. Results of spatial test.

Method	$\mathit{W}1$	$\mathit{W}2$
LM-lag	44.25 ***	64.48 ***
R-LM-lag	10.39 ***	20.6 ***
LM-error	35.29 ***	22.06 ***
R-LM-error	10.78 ***	13.41 ***
Wald-lag	11.25 ***	0.38
LR-lag	17.21 ***	17.38 ***
Wald-error	6.54 ***	1.15
LR-error	27.35 ***	13.83 ***

Note: *, ** and *** indicate the statistical significance at the time of p < 0.1, p < 0.05 and p < 0.01, respectively.

shows the estimated results. In the two weight matrices, LM-lag, LM-error, Robust lM-lag and Robust lM-error all pass the significance test at the level of 1%. It can be preliminarily concluded that the spatial Durbin model can be used. In the Wald test and LR test, the Wald test of economic weight matrix is not significant. Therefore, in this paper, the spatial Durbin model is selected for analysis.

Table 8. The results of the spatial econometric model incorporating the mediating effect ($W1$).

Variables		Benchmark Model	Mediator Variable
		$\mathbf{ln}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$	$\mathbf{ln}\mathit{e}\mathit{p}\mathit{s}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$	$\mathbf{ln}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{c}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$
$\ln disaster$		−0.016 ** (0.006)	0.15 *** 0.056	−0.002 ** (0.005)	0.628 *** (0.162)	−0.012 * (0.007)	0.092 * (0.059)	−0.013 ** (0.005)
$W1\ln disaster$		0.004 * (0.002)	0.103 *** (0.043)	0.005 * (0.009)	0.045 * (0.062)	0.004 ** (0.002)	0.124 *** (0.047)	0.004 ** (0.002)
$\ln med$		—	—	0.119 *** (0.018)	—	0.026 ** (0.008)	—	0.037 *** (0.013)
$W1\ln med$		—	—	0.018 *** (0.032)	—	0.016 *** (0.012)	—	0.018 * (0.01)
Control variables		Yes	Yes	Yes	Yes	Yes	Yes	Yes
Fixed effects		Yes	Yes	Yes	Yes	Yes	Yes	Yes
Direct effect	$\ln disaster$	−0.016 *** (0.007)	0.192 *** (0.059)	0.005 *** (0.005)	0.676 *** (0.155)	−0.013 ** (0.006)	0.133 ** (0.063)	−0.013 *** (0.005)
	$\ln med$	—	—	0.118 *** (0.015)	—	0.026 ** (0.008)	—	0.038 *** (0.013)
Indirect effect	$\ln disaster$	0.001 * (0.003)	0.460 *** (0.097)	−0.002 * 0.01	0.6 *** (0.096)	0.002 (0.003)	0.45 *** (0.107)	0.002 (0.002)
	$\ln med$	—	—	0.021 *** (0.025)	—	0.022 * (0.012)	—	0.029 * (0.016)
$R^2$		95.92%	97.12%	99.37%	98.73%	99.76%	91.72%	98.21%
Observation		713	713	713	713	713	713	713

Note: The results in parentheses are robust standard errors; *, ** and *** indicate the statistical significance at the time of p < 0.1, p < 0.05 and p < 0.01, respectively.

and

Table 9. The results of the spatial econometric model incorporating the mediating effect ($W2$).

Variables		Benchmark Model	Mediator Variable
		$\mathbf{ln}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$	$\mathbf{ln}\mathit{e}\mathit{p}\mathit{s}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$	$\mathbf{ln}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{c}$	$\mathbf{ln}\mathit{e}\mathit{c}\mathit{t}\mathit{e}\mathit{c}\mathit{h}$
$\ln disaster$		−0.016 ** (0.007)	0.086 *** (0.034)	−0.002 * (0.002)	0.542 *** (0.148)	−0.012 * (0.007)	0.024 * (0.036)	−0.013 ** (0.006)
$W2\ln disaster$		0.024 ** (0.01)	0.287 *** (0.079)	0.02 ** (0.007)	0.315 ** (0.15)	0.029 ** (0.011)	0.299 *** (0.083)	0.031 *** (0.002)
$\ln med$		—	—	0.108 *** (0.009)	—	0.029 ** (0.007)	—	0.037 *** (0.013)
$W2\ln med$		—	—	0.015 *** (0.024)	—	0.05 * (0.3)	—	0.021 * (0.015)
Control variables		Yes	Yes	Yes	Yes	Yes	Yes	Yes
Fixed effects		Yes	Yes	Yes	Yes	Yes	Yes	Yes
Direct effect	$\ln disaster$	−0.015 ** (0.007)	0.104 *** (0.033)	−0.001 * 0.001	0.561 *** (0.145)	−0.012 * (0.006)	0.042 ** (0.036)	−0.012 *** (0.006)
	$\ln med$	—	—	0.109 ***(0.009)	—	0.03 *** (0.007)	—	0.036 *** (0.011)
Indirect effect	$\ln disaster$	0.027 * (0.018)	0.74 *** (0.123)	0.024 ** 0.011	0.95 *** (0.161)	0.039 * (0.021)	0.741 *** (0.132)	0.042 * (0.023)
	$\ln med$	—	—	0.048 *** (0.044)	—	0.086 * (0.045)	—	0.051 * (0.044)
$R^2$		98.46%	78.69%	97.33%	78.1%	95.95%	74.6%	96.02%
Observations		713	713	713	713	713	713	713

Note: The results in parentheses are robust standard errors; *, ** and *** indicate the statistical significance at the time of p < 0.1, p < 0.05 and p < 0.01, respectively.

show the results of the stepwise test spatial Durbin model introducing the spatial terms of geographical and economic distance, respectively. The estimated coefficient of the direct effect in the basic model is significantly negative in all models of the stepwise regression, which demonstrate that natural disasters generate an inhibiting effect on the development of the key technologies of the local emergency industry. They are in line with the results of baseline regression in hypothesis 1. However, in the two spatial weight matrices, the coefficients of indirect effect and the spatial matrix term are positive, which shows that the increased natural disaster loss in the province has a driving effect on the number of emergency key technologies in neighboring provinces. The influence coefficient of such driving effect is much higher and more significant in the regression results of economic weight matrix. The possible reason is that the shock triggered by natural disasters to local market does harm to the formation of local strategic technology factors and capital. China has the good tradition of united efforts against disasters under regional linkage. According to the dispatching optimization goal of minimizing the total cost of emergency rescue efforts, in case of disasters, local provinces will urgently call for necessary human, material and technical support from neighboring regions, thereby promoting neighboring provinces to conduct more technological investment in the emergency industry. This has expedited technological advancement in the emergency industry and fostered the aggregation of strategic technological factors for disaster relief in adjacent provinces. We conclude that the level of economic development exerts a polarizing driving effect, meaning that neighboring provinces with higher levels of economic development experience faster growth in key technologies of the emergency industry. The spatial econometric results presented above demonstrate that the institutional strengths of China’s emergency rescue system have effectively facilitated the advancement of key technologies.

According to the results of the spatial econometric model incorporating the mediating effect, the demand triggered by natural disasters, the technology path, and the education path all generate positive spatial effects. In addition, the direct, indirect and total effects of the three mediator variables on the key technologies are all significantly positive, and the direct effect is higher than the indirect effect. This means that under China’s new national system the impact of natural disaster on the key technologies of the emergency industry is non-linear, with the coexistence of mediating effect and spatial spillover effect. Specific manifestations are that the occurrence of disasters not only stimulates the development of the local market demand for the emergency industry, digital technology and the construction of emergency disciplines, but also has a positive spatial spillover effect on adjacent areas. The above factors can ultimately generate an impact on the key technologies of the emergency industry by exerting a positive effect on the local and adjacent areas. China is striving to gradually establish the mode of mutual dispatching of emergency products and services, mutual integration of digital technologies, and common development and orderly operation of emergency discipline construction among regions under the background of the new national system. From the perspective of regional linkage, this paper provides an in-depth analysis of the intrinsic factors enabling China to maintain technological resilience despite being impacted by disasters. As a result, the testing methods of regression coefficients of spatial factors combined with mediating effects have verified hypothesis 5 in this paper.

5.3.3. Robustness Test

We applied the Delphi method to assign weights to the three indicators—economic loss, affected or deceased population, and disaster frequency. A panel of 17 experts in disaster management and statistics participated in two rounds of anonymous consultation. In the first round, mean scores were 7.5 ± 1.3, 8.6 ± 1.1, and 5.9 ± 1.5, respectively, with Kendall’s W = 0.44 (p < 0.01). After feedback and re-evaluation in the second round, the mean scores improved to 8.2 ± 0.8, 8.8 ± 0.6, and 5.5 ± 1.0, with W = 0.71 (p < 0.001). The normalized results yielded final weights of 0.36, 0.39, and 0.24, which were used to construct the comprehensive disaster loss index for 31 provinces during 2001–2022.

To ensure the robustness of the empirical findings, we conducted a series of robustness checks by re-estimating the spatial econometric models with alternative specifications. The results are reported in

Table 10. Robustness check results of the spatial econometric model (with mediating effect).

Model Form	Benchmark Model		Mediator Variable		Direct Effect		Indirect Effect
	$\mathit{W}\mathbf{1}$	$\mathit{W}\mathbf{2}$	$\mathit{W}\mathbf{1}$	$\mathit{W}\mathbf{2}$	$\mathit{W}\mathbf{1}$	$\mathit{W}\mathbf{2}$	$\mathit{W}\mathbf{1}$	$\mathit{W}\mathbf{2}$
Disaster coefficient	−0.013 ** (0.046)	−0.013 ** (0.088)	0.318 *** (0.132)	0.692 *** (0.096)	−0.015 ** (0.006)	−0.009 ** (0.007)	0.001 (0.005)	0.013 * (0.029)
Control variables	Yes		Yes		Yes		Yes
Fixed effects	Yes		Yes		Yes		Yes

Note: The results in parentheses are robust standard errors; *, ** and *** indicate the statistical significance at the time of p < 0.1, p < 0.05 and p < 0.01, respectively.

. As shown in Table 10, the benchmark model and the extended model incorporating mediating variables yield consistent coefficients, with the key explanatory variables remaining statistically significant at conventional levels. Both the direct and indirect effects are stable across specifications, indicating that the mediating mechanism is valid. Table 9 further confirms the robustness of these findings. When alternative model settings are employed, the estimated coefficients exhibit similar signs and magnitudes, and the statistical significance is largely preserved. The proportion of the mediating effect also remains high, suggesting that the results are not driven by model-specific assumptions. Overall, the robustness tests confirm that our main conclusions are stable and reliable.

To further verify the robustness of our results, we replaced the disaster loss index constructed by the entropy method with the one derived from the Delphi method. The results in Table 10 show that the coefficients under the two weighting schemes share the same signs and exhibit consistent effects, indicating that the estimation results remain stable. Although the entropy method may assign greater weights to indicators with higher variability and thus not fully capture the social severity of disasters, the consistency between the two approaches suggests that the entropy-based index still provides reliable estimates in this study.

6. Conclusions and Policy Implications

Global scientific and technological innovation has entered a phase of rapid activity, with emerging technologies increasingly integrating into the emergency industry. Considering the frequent occurrence of natural disasters and the forward-looking arrangements in international emergency technology strategies, tackling the technological challenges posed by these disasters represents a critical issue for the global community. The emergency industry is directly oriented toward disaster prevention and control, providing essential support for mitigating impacts and enhancing preparedness. This paper aims to explore how natural disasters affect the key technology of the emergency industry under China’s new national system, and the strategies China employs to respond. By applying the theories of innovative geography and disaster economics, this study analyzes demand, technology, and education pathways using province-level panel data from 2000 to 2022. The research integrates text analysis technology and employs models such as the mediating effect model, dynamic spatial Durbin model, and Stepwise Model with spatial terms to examine how natural disasters directly impact key technologies and why these technologies continue to develop resiliently in the face of external shocks under the new national system. The main findings include: (1) Natural disasters have a direct negative impact on key technologies in the emergency industry. (2) Disasters indirectly promote innovation in key technologies by driving demand for emergency products and services, encouraging digital technology adoption, and supporting the development of emergency management majors. These factors create a mediating effect across demand, technology, and education pathways. (3) Innovation in China’s emergency industry and the impact of natural disasters exhibit regional concentration, with a pronounced spatial polarization. Disasters in one province can stimulate technological innovation in adjacent areas, with the influence being more pronounced in regions characterized by higher levels of economic development. China’s macro-control mechanism, exemplified by the principle of “mutual assistance in times of disaster”, has effectively facilitated the advancement of key technologies. (4) The relationship between natural disasters and key technologies is non-linear, involving both mediating and spatial spillover effects. In the field of key technological breakthroughs for the emergency industry in response to natural disasters, China is gradually forming a pattern of orderly operation among regions in terms of mutual dispatch of emergency products and services, digital technology integration, and emergency discipline construction under the new national system. This collaborative approach explains how China maintains technological resilience despite the challenges posed by natural disasters, driven by regional collaboration and linkage. A recent example of inter-provincial cooperation in post-disaster response in China can be seen in the coordinated relief efforts. Following the severe rainfall event in Zhengzhou on 20 July, Shandong province deployed high-capacity drainage vehicles capable of pumping 3000 cubic meters of water per hour, providing critical support in the domain of high-tech emergency resource dispatch. In parallel, Zhuhai Yunzhou Intelligent Technology Co., Ltd., a company based in Zhuhai, Guangdong, China, allocated 118 units of surface rescue robots to conduct remote rescue operations for trapped individuals, representing a key application of core emergency technologies.

The results of this research are novel, and our results contribute to emergency management and disaster economics literature. The new national system has showcased China’s institutional strengths in achieving breakthroughs in key emergency industry technologies for disaster response. Therefore, this paper enriches the research scenarios of the theory of mutual transformation of interests in disaster economics. Firstly, few empirical studies investigate the influence of natural disasters on the key technologies in the emergency field. Therefore, this paper extends and enriches the relevant theories and empirical research in the aforementioned domains. Moreover, this study integrates technological demands, innovation effects, and educational initiatives to investigate how China utilizes technology policies under the new national system to address the fundamental national characteristic of frequent disasters. It organically couples natural disaster and key technologies to make up for the current lack of research and fills this gap in the above-mentioned fields along the direction of technological innovation. Finally, research on the diffusion effect of key emergency industry technologies triggered by disasters within the domain of emergency management remains scarce. This study achieves notable theoretical advancements and empirically validates the diffusibility of technologies induced by disasters.

This paper also puts forward targeted policy implications for the governments of countries prone to natural disasters. The first suggestion is that governments assistance should actively encourage the establishment of a new research and development pattern dominated by the leading enterprises in the emergency industry which, in the Chinese context, is supported by the new national system that integrates state-led design with enterprise-driven innovation. These chain master enterprises are encouraged to undertake major national scientific and technological research projects, addressing key technical challenges that span multiple industries and disciplines. Moreover, multi-disaster-prone countries should make good use of the reverse promotion effect brought by natural disasters and thereby give priority to the innovation of digital technologies. Strive to establish a nationally unified comprehensive management service platform for the emergency industry which should be understood as a comprehensive industrial system covering disaster prevention and mitigation, emergency equipment, post-disaster recovery, and digital emergency services through digital exploration, thereby forming a strong and sustained supply capacity of common technologies that support the green development. In addition, government ought to construct an ecosystem for emergency industry technologies so as to intensify the coordinated development of the agglomeration and radiation effects among regions. Finally, countries prone to multiple disasters should fully capitalize on the spatial spillover effects of key technologies and the strategic opportunities offered by the global emergency science and technology landscape. They should increase investment in talent development and capital allocation, leveraging the advantages of the new national system to expedite the establishment of unified standards, public service platforms, and systematic talent training within the emergency industry, while actively constructing an emergency discipline framework that aligns with their national context.

7. Limitations

This study measures the development of key technologies in the emergency industry by combining text-based scores with the number of high-quality patents, thereby reflecting knowledge accumulation and innovation output. However, this measurement approach has limitations. For instance, public R & D funding could serve as an important alternative indicator to capture the potential for technological advancement and to provide robustness checks. Due to the lack of systematically available data on public R & D investment in the emergency industry, such analysis could not be undertaken in this study. Future research should incorporate public R & D funding and other multidimensional indicators once data become available, in order to improve the measurement and validation of key technology development.