Knowledge Spillovers and Integrated Circuit Innovation Ecosystem Resilience: Evidence from China

Abstract

A resilient innovation ecosystem is an important guarantee for enhancing industrial competitiveness. Knowledge spillover is the key driving force to enhance system resilience. Firstly, we use the MEREC-CoCoSo method to calculate the resilience level of the integrated circuit (IC) innovation ecosystem and analyze the evolution trajectory of the resilience before and after the emergence of the “stuck-neck” problem. Secondly, based on the panel data of 30 provinces (autonomous regions and municipalities directly under the central government) in China from 2011 to 2021, this paper analyzes the mechanism of the impact of intra-regional knowledge spillovers on the resilience of IC innovation ecosystems using the fixed-effect model and analyzes the spatial effect of inter-regional knowledge spillovers on the resilience of innovation ecosystems using the spatial Durbin model under the human capital matrix. Finally, we analyze the regulating role of contractual and relational governance mechanisms and try to open the “black box” of governance. The result shows the following: (1) The polarization of innovative ecosystem resilience of integrated circuits is gradually increasing, with strong spatial agglomeration, high–high agglomeration, low–low agglomeration, and low–high dispersion, and there is an obvious “matthew effect” and “siphon effect”. (2) Both intra- and inter-regional knowledge spillovers contribute significantly to the resilience of IC innovation ecosystems. The contractual governance mechanism can effectively enhance the impact of knowledge spillovers on the resilience of innovation ecosystems in the region, and the relational governance mechanism has a positive impact on the resilience of innovation ecosystems in neighboring regions. (3) Heterogeneity results show that knowledge spillovers within the Pan-PRD region have a significant positive impact on innovation ecosystem resilience. Knowledge spillovers between regions with low innovation capacity have a double effect on innovation ecosystem resilience, and knowledge spillovers between regions with “talent highlands” have a facilitating effect on innovation ecosystem resilience. Accordingly, policy recommendations are put forward to open up channels for innovation knowledge spillover, realize effective allocation of innovation resources, and optimize the system of innovation talents.

1. Introduction

Progress in key core technologies has become a national priority and a strategic asset for enhancing China’s economic competitiveness and ensuring national security. There is an urgent need to strengthen the in-depth integration of scientific and technological innovation with industrial innovation, integrate scientific and technological innovation forces and advantageous resources across regions, and create a more globally competitive innovation ecology [1]. The key core technology “stuck-neck” problem stems from the long-term existence of a large technological gap between China’s key core technology field and other countries, and the gap is difficult to narrow through technological innovation in the short term [2]. Compared with other key core technologies, “stuck-neck” technology has a greater strategic threat, stronger monopoly, higher external dependence, longer research and development cycle, and more difficulty in breaking through or finding alternatives in the short term. A resilient innovation ecosystem provides more room for effective interaction of innovation entities and efficient flow of innovation factors [3]. Only by improving the resilience of the innovation ecosystem of key core technologies can we better respond to the challenges of the internal and external environment and realize the innovation-driven development strategy. As the “main battlefield” of the science and technology struggle, integrated circuits directly reflect the modernization level and development potential of China’s industrial base, and studying the resilience of the integrated circuit innovation ecosystem under the pressure of “stuck-neck” is a necessary move to cope with global industrial competition.

In an innovation ecosystem with strong linkages among innovation agents, knowledge spillovers occur when external resources are absorbed while releasing their own value [4]. Knowledge spillover refers to the informal diffusion and flow of knowledge and information among different actors (e.g., enterprises, universities, research institutions, etc.), usually without direct economic exchange. This process implies that knowledge created or possessed by one subject can be acquired, imitated, or applied by other subjects, thus generating positive externalities. Early studies on knowledge spillovers focused on defining its connotation [5], driving factors [6], and other aspects. According to different criteria, knowledge spillovers can be divided into different types. First, in the development of knowledge spillover theory, Ali [7] considered knowledge spillover as a phenomenon caused by FDI and pointed out that knowledge spillover triggered by FDI can improve the productive capacity of local enterprises in host countries. According to the relationship between the source and the receiver of knowledge, knowledge spillover can be divided into internal spillover within the same organization and external spillover between different organizations [8]. Secondly, according to the scope of spillover, knowledge spillover can be categorized into local spillover in the same city or region and spillover across different countries and regions [9]. Subsequently, scholars are more inclined to regard knowledge spillovers as the unconscious knowledge dissemination generated by the flow of technology and talents, which is the embodiment of positive knowledge externalities [10]. Finally, according to the characteristics of the content of knowledge spillover, it is divided into explicit knowledge spillover through patents and literature disclosure and implicit knowledge spillover through personnel mobility and face-to-face communication [11]. Technology, information, talent, and other knowledge-based resources are becoming more and more important to the development of organizations, and enterprises, regions, and governments are becoming more and more aware of the necessity of laying out knowledge resources and the criticality of analyzing the impact mechanism of knowledge spillover.

Most researchers agree that knowledge spillovers have a significant impact on innovation. Tang [12] argues that with the increase in risky shocks, scientific and technological cooperation is an important way of knowledge spillover, which has an important impact on the development of regional innovation system, and that knowledge spillover exists in the city tier effect and geographic proximity effect. Hu [13] explores the impact of OFDI reverse knowledge spillover on the technological progress of invested enterprises. OFDI reverse knowledge spillover refers to the process of domestic enterprises investing production factors in foreign markets and acquiring local knowledge factors through overseas mergers and acquisitions, joint ventures, the establishment of branches, etc. Audretsch [14] puts forward a knowledge-based view of knowledge collaboration and spillovers, explaining how firms’ collaborative decision-making affects their innovation outputs and propensity to innovate. Some scholars, based on the traditional knowledge base view (KBV), encourage firms to view knowledge spillovers as something to be discouraged in order to maintain their technological competitiveness [15]. Excessive knowledge spillovers can lead to “free-rider” behavior, which weakens firms’ innovation dynamics and makes knowledge spillovers less effective as a driver of innovation.

The impact of knowledge spillover on resilience is mainly realized through three mechanisms: First, knowledge spillover can accelerate the accumulation and diffusion of new knowledge. Yuan [16] argues that knowledge spillover can serve as a channel through which digital transformation enhances resilience by facilitating knowledge spillover. Second, knowledge spillover can enhance the synergistic effect between different participants. Liu [17] believes that when enterprises and universities cooperate in joint research and development, universities can transfer cutting-edge research results to enterprises through knowledge spillovers, while enterprises feed market demand back to universities, and the two synergistically promote technological innovation and application. When scientific research institutes and enterprises within the innovation system continue to master the most advanced core technologies and maintain their stable output, they can reduce the uncertainty of the transaction and increase the resilience of the system. Third, knowledge spillover can promote value creation within the innovation ecosystem. Li [18] believes that the knowledge spillover effect of industrial intelligence will enhance the success rate of R&D in key areas, especially in “stuck-neck” technologies, so as to increase resilience and stability. Therefore, knowledge spillover is the key internal motivation for innovation subjects to utilize high-quality innovation resources and enhance resilience.

However, most of the literature focuses on the impact of knowledge spillover factors on regional innovation and industrial resilience, but there are relatively few studies on the resilience of IC innovation ecosystems leading the new round of technological revolution. Most empirical studies focus on enterprise-level analysis, and fewer studies are conducted at the regional level, in multiple dimensions. Third, knowledge spillover does not exist in isolation, and there is a relative lack of research on its interaction with other innovation drivers, which cannot fully reveal the actual effects and spatial effects of knowledge spillover in the innovation ecosystem. Based on this, fundamental questions that need to be addressed now are as follows: How do intra- and inter-regional knowledge spillovers affect IC innovation ecosystem resilience? What are the knowledge governance mechanisms behind this process? What kind of differences exist in different regions?

Accordingly, this paper firstly uses the MEREC-CoCoSo method to calculate the resilience level of the IC innovation ecosystem and analyzes the evolution trajectory of the resilience before and after the emergence of the “stuck-neck” problem. Secondly, based on the panel data of 30 provinces (autonomous regions and municipalities directly under the central government) in China from 2011 to 2021, this paper analyzes the mechanism by which intra-regional knowledge spillover affects the toughness of IC innovation ecosystems using the fixed-effect model and analyzes the spatial effect of inter-regional knowledge spillover on the toughness of innovation ecosystems using the spatial Durbin model under the human capital matrix. Finally, we analyze the regulating role of contractual and relational governance mechanisms and try to open the “black box” of governance. This paper not only has important practical significance for improving the resilience of regional IC innovation ecosystems, but also provides new spatial perspectives for resilience governance research.

2. Theoretical Analysis and Research Hypothesis

2.1. Innovation Ecosystem Resilience

The IC industry plays a pivotal role in economic development and national security [19]. It is a driving force behind the technological revolution and industrial change. China’s IC industry is currently undergoing a period of rapid development. However, due to the highly specialized structure of the IC industry, the innovation system is not yet fully optimized. The highly specialized structure of the IC industry, coupled with imperfect innovation systems and high technical barriers, has resulted in the emergence of “short boards” in numerous pivotal sectors [20]. Furthermore, the lack of effective integration between the primary entities engaged in innovation, coupled with the potential for “chain breakage” in the supply chain, underscores the necessity for a more robust innovation ecosystem. The IC innovation ecosystem utilizes the core role of scientific research institutions to achieve technological innovation based on the coexistence of multiple elements of the innovation system, including the main body of innovation, i.e., innovation producers, innovation consumers, innovation decomposers, and the corresponding environmental factors, namely the policy environment, innovation environment, and social environment [21]. In this context, scientific research institutions, financial institutions, and government departments can be considered as producers, generating the power and energy of innovation [22]. Core enterprises and other innovative enterprises, in addition to colleges and universities, are collectively responsible for the transformation process, which can be considered a form of consumption. In the final analysis, the market and consumers will assimilate the results of the transformation process, thereby fulfilling the role of decomposer and utilizing the resources of the products [23]. Simultaneously, they will convey new innovation needs to the innovation producers. Nevertheless, the successful implementation of this process hinges on the establishment of a robust IC innovation ecosystem.

The term toughness was first used in physics or mathematics to describe the ability of an object to return to its position after displacement. In the 1990s, the concept of resilience began to describe the ability of social systems to self-regulate after disturbances. The development and change of the concept of resilience have experienced stages from “engineering resilience” to “ecological resilience” and then to “evolutionary resilience”. There are also different theoretical frameworks for toughness indicators, as shown in

Table 1. Classification of theoretical frameworks for toughness indicators.

Theoretical Framework	Standardized Layer
evolutionary toughness theory [24]	Diversity–Buffering–Evolution–Mobility–Compatibility
ecosystem theory [25]	Ecological resilience–Economic Resilience–Social Resilience–Infrastructure Resilience
adaptive circulation theory [26]	Develop–Maintain–Release–Reorganize
actor network theory [27]	Reactivity–Adaptability–Resilience

. Carpenter [24] proposed the theory of evolutionary resilience in 2005. Bogers [25] discusses some of the main trends (e.g., digital transformation), challenges (e.g., uncertainty), and potential solutions (e.g., EU funding schemes) in the context of open innovation and innovation policy based on ecosystem theory. Martin [26] presents the concept of resilience and examines its importance in helping regional economies respond to major recessionary shocks. Scholars such as Li [27] have studied community resilience governance in conjunction with actor network theory, which suggests that community resilience is collective action to enhance responsiveness, adaptability, and resilience when collective spaces face shocks.

Compared with engineering toughness and ecological toughness, adaptive cycle theory can show the complexity and non-equilibrium toughness characteristics of integrated circuit systems [28]. Adaptive cycle theory is one of the few theories that explicitly explain the dynamics of complex systems, with a particular focus on the concept of a system’s “resilience” in response to external changes and shocks, and how resilience changes as the system develops. It not only pays attention to the recovery of the initial state of the integrated circuit system, but also pays more attention to the dynamic adaptive adjustment path of the long-term continuous growth of the system, especially the updating ability of the system to maintain key performance. Based on the adaptive cycle theory, the evolution cycle of resilience is divided into four stages: development, maintenance, release, and reorganization. Under the adaptive cycle theory, the integrated circuit industrial cluster resilient system needs to go through four stages of “shock absorption, impact adaption, recovery, and re-organization” to resolve external shocks, which need to focus on system support, resistance, resilience, and renewal, respectively. Support refers to the adaptive ability of the system to reduce the negative impact of risk. Resistance refers to the ability of the system to dissolve the impact force of risk when it is hit by risk. Resilience refers to the ability of the system to restore balance after a risk shock. Renewal power refers to the ability of the system to self-repair and update after receiving risk shocks. During the evolution of the IC innovation ecosystem, the innovation community gradually evolved from a single community to a community network, thus improving system resistance. At the same time, resources are shared and interact among the communities, and the network structure continues to expand, effectively improving the resilience and adaptability of the system. When the system is subjected to external shocks, in order to avoid the system collapse caused by vulnerability, the renewal force in the system is opposed to it, and the four capabilities are interlinked and progressive, jointly promoting the spiral development of the integrated circuit innovation ecosystem. Finally, a toughness index system consisting of four dimensions, namely elemental support, structural resistance, environmental restoration, and functional renewal, was constructed (

Table 2. Resilience index system of IC innovation ecosystem.

Dimension	Indicator	Variable Description	Indicator Source
elemental support	Talent factor allocation	R&D developers [29]	China High-Tech Industry Statistical Yearbook
	Science and technology factorization	Number of new product development projects [30]	China High-Tech Industry Statistical Yearbook
	Allocation of financial elements	Provision for new product development [31]	China High-Tech Industry Statistical Yearbook
		R&D investment intensity [32]	China High-Tech Industry Statistical Yearbook
	Energy factor allocation	Energy consumption per unit of GDP [33]	China Statistical Yearbook
structural resistance	Business diversity	Number of enterprises with R&D activities as a percentage of enterprises [34]	China High-Tech Industry Statistical Yearbook
		Number of enterprises with R&D organizations as a share of enterprises [35]	China High-Tech Industry Statistical Yearbook
	Diversity in higher education	Number of students enrolled in general higher education [36]	China Regional Economic Statistical Yearbook
		Number of higher education institutions [37]	China Regional Economic Statistical Yearbook
environmental restoration	Economic environment	GDP growth rate [38]	China Statistical Yearbook
		Investment efficiency [39]	China Statistical Yearbook
	Social environment	Industrial agglomeration [40]	China Labor Statistics Yearbook
	Ecological environment	Investment in pollution control [41]	China Environmental Statistical Yearbook
functional renewal	Innovative features	Intensity of knowledge protection [42]	China IC Industry Intellectual Property Annual Report
		Local finance expenditure on science and technology [43]	China Statistical Yearbook
	Production function	Technical output [44]	China Statistical Yearbook
		Technical profit [45]	China Industrial Statistics Yearbook

).

In this paper, in the construction and evaluation of the resilience of integrated circuit innovation ecosystem, talent factor allocation, science and technology factor allocation, capital factor allocation, and energy factor allocation are selected as important elemental support indicators. The importance of these four factors is explained in detail as follows:

Talent allocation is the core of the IC innovation ecosystem. Talent is not only the source of technological research and development and innovation, but also the key force promoting industrial upgrading and technological progress. The lack or misallocation of talent leads to a decline in innovation capacity, and even leads to the fragility of the entire ecosystem [46]. Therefore, the resilience of talent factor allocation reflects the ability of the system to cope with challenges and emergencies. Secondly, scientific and technological elements include technical infrastructure, research and development capabilities, technology platforms, etc., which determine the technological level and innovation capability of an ecosystem [47]. The rational allocation of scientific and technological elements directly affects the transformation efficiency of innovation achievements and the speed of technological upgrading. The resilience of the allocation of science and technology elements can measure the adaptability of the system in the face of technological changes and market demand changes [48]. Capital factor allocation is also one of the key indicators of the IC innovation ecosystem. The allocation of capital factors affects the scale and depth of innovation activities. The lack or misallocation of funds can limit the advancement of innovation and affect the vitality and stability of ecosystems. Finally, energy factor allocation is also important for the resilience of the IC innovation ecosystem. Energy is the basis for supporting innovative activities, especially in high-tech industries and manufacturing [49]. The stability and efficiency of energy supply directly affect the continuity of technology development and product production.

Firm diversity and efficient diversity were selected as important structural resistance indicators, the importance of which is detailed as follows: First, firm diversity refers to the presence of many types, sizes, and sectors of firms in the innovation ecosystem, including startups, large enterprises, SMEs, etc., as well as companies involved in different industries and technologies. This diversity is critical to enhancing the resilience of the innovation ecosystem. Corporate diversity can lead to multiple perspectives and solutions, reducing reliance on a single domain or model, and thereby reducing systemic risk [50]. The diverse corporate structure enables the entire ecosystem to better respond to economic fluctuations, technological changes, and changes in market demand, thereby enhancing the overall innovation capacity and resilience. Secondly, different universities have different research directions and disciplinary advantages, and their research results and technological innovation can provide rich knowledge support for the innovation ecosystem. The diversity of universities promotes interdisciplinary cooperation and exchange, which helps to combine knowledge and technology in different fields to produce new research results and applications [51]. Such cooperation not only improves the efficiency of innovation, but also drives more comprehensive and in-depth technological progress. As important indicators of the resilience of the innovation ecosystem, enterprise diversity and university diversity can significantly improve the risk diversification ability, innovation stimulation ability, flexible adaptation ability, and knowledge and technology diversification of the system.

The economic environment, social environment, and ecological environment are selected as important indicators of environmental restoration. The importance of these three factors is explained in detail as follows: The economic environment includes macroeconomic policies, market dynamics, economic growth level, and other factors that directly affect the operation status of enterprises and resource allocation of innovative activities [52]. A stable and healthy economic environment can provide sufficient financial support and market demand to promote innovative development. The social environment includes social policies, public services, education system, etc., which affect the atmosphere of innovation and talent cultivation. The stability and support of the social environment directly affect the smooth progress of innovation activities. A good education system can cultivate a large number of high-quality innovative talents, and effective social policies can promote the improvement of the environment for innovation and entrepreneurship [53]. The instability or insufficiency of the social environment will lead to a lack of innovation resources and brain drain, affecting the overall resilience of the system. The natural environment is also an important indicator of the resilience of innovation ecosystems. The natural environment includes the availability of resources, environmental protection policies, and the health of ecosystems. The quality of the natural environment directly affects the production and innovation activities of enterprises. Changes in the natural environment, such as climate change and environmental disasters, can have an impact on the innovation ecosystem [54]. Assessing the resilience of the natural environment can help to understand the coping capacity of ecosystems in the face of natural disasters and environmental changes, so as to formulate sustainable development strategies.

Innovation function and production function are selected as important indicators of functional renewal, and the importance of these two elements is explained in detail as follows: First, an ecosystem with strong innovation function can respond quickly to changes in market demand and technical challenges. In the face of sudden changes in market demand or technological changes, enterprises and organizations with strong innovation functions can quickly adjust their research and development direction and product strategies to adapt to new market conditions [55]. This flexibility and rapid adaptation ability are an important manifestation of system resilience. Secondly, production function refers to the capacity of each participant in the innovation ecosystem in terms of production and manufacturing, including production efficiency, manufacturing quality, resource utilization, etc. Productive ecosystems ensure a continuous supply of goods and services [56]. This stability is particularly important in responding to external shocks such as supply chain disruptions and market demand fluctuations. Powerful production capabilities can quickly restore the normal operation of the supply chain by optimizing production processes or adjusting the allocation of resources. Efficient production functions can reduce production costs and improve product quality, thereby enhancing the market competitiveness of enterprises and the overall toughness of the system.

2.2. Knowledge Spillover

IC technology presents obvious knowledge characteristics in the innovation process. Knowledge spillover is an important factor affecting the resilience of the innovation ecosystem and a key driving force for improving the connection strength of innovation entities and enhancing the cohesiveness of cooperation. It will promote the aggregation of innovation entities and the concentration of innovation activities and contribute to the integration and interaction of heterogeneous knowledge among innovation entities. The knowledge spillover from the symbiotic network of innovation entities within the region can help to reduce the knowledge gap among universities, enterprises, and research institutes, enhance the information transfer efficiency among innovation entities, and reduce the transaction costs between partners. The free flow of knowledge resources creates a good competitive environment, and enterprises and innovation entities in different regions can carry out fierce competition in an open market environment, which promotes the increase in innovation input [57]. Enterprises and scientific research institutions in the region will increase investment in research and development in order to maintain competitive advantages and improve technology level and innovation quality, so as to attract more high-quality knowledge resources [58]. This positive cycle will further promote the development of regional innovation so that each region can make better use of its unique resources and advantages and form the agglomeration effect of innovation [59]. At the same time, after receiving government support, enterprises can attract more external knowledge inflow and enhance the performance of knowledge transfer by combining with external technology and knowledge. Enterprises can more effectively transform external advanced technology and management experience into their own innovation results, improving the overall technical level and market competitiveness [60].

According to the theory of regional innovation system, innovation entities can not only generate knowledge spillover with regional entities, but also exchange knowledge with non-regional entities to increase knowledge diversity [61]. The receiver of knowledge can combine the acquired knowledge with other knowledge to create new knowledge, but the provider may not enjoy the full benefits [62]. Knowledge spillovers can also occur when the recipient fails to bear the full cost and the provider is not adequately compensated. When a region has significantly more knowledge than other surrounding regions, knowledge spillover is inevitable [63]. As Figure 1 shows, the potential difference between different regions will effectively promote the inward and outward inflow of innovative knowledge in each region. Based on the innovation network, the innovation body constantly absorbs and digesters and then generates new knowledge resources so as to continuously improve the knowledge stock of the entire integrated circuit innovation ecosystem and realize the stable operation of the innovation ecosystem.

2.3. Regional Knowledge Spillovers and IC Innovation Ecosystem Resilience

Based on the logical assumption of “collaborative governance to cope with problems” of existing research, this paper also considers that collaborative governance among multiple subjects is an important tool for innovation ecosystems to cope with risky shocks and that knowledge spillover, as a kind of knowledge externality exerted by unconscious knowledge dissemination, provides a rich knowledge base and innovation impetus for collaborative governance [64]. The theory of collaborative governance emphasizes the behavioral process of collective action, mutual cooperation, mutual coordination, and collaborative progress among multiple subjects to achieve certain public goals. Intra-regional knowledge spillover is the key driving force for innovation subjects to realize the value of innovation, collaborate and cooperate with each other, improve the connection strength of innovation subjects, and enhance the viscosity of cooperation [65]. Collaborative governance among innovation subjects can promote the aggregation of innovation subjects and the concentration of innovation activities, help the integration and interaction of heterogeneous knowledge among innovation subjects, and overcome the “lock-in effect”. David [66] argues that knowledge spillovers are triggered by knowledge aggregation and that the sources of information for knowledge spillovers are usually located in the public domain and depend on the ability of firms to create information flows from patents, publications, and technologies. Ferreira [67] believes that knowledge is non-competitive and non-exclusive and that knowledge spillover is an important source of economic growth and provides opportunities for business development. Wang [68] believes that the innovation main body within the region can collaborate to promote knowledge spillover to achieve R&D output growth through industry–university–research cooperation and promote knowledge marketization. Symbiotic network knowledge spillover of innovation subjects within a region helps to reduce the knowledge potential difference between universities, enterprises, and research institutes, enhances the efficiency of information transfer between innovation subjects, and reduces the transaction costs between the cooperating parties. The knowledge spillover effect will reduce the intrinsic uncertainty of the innovation ecosystem and form a cluster-type innovation network development mode, thus enhancing IC innovation ecosystem resilience [69]. Based on this, this paper proposes the following:

Hypothesis H1.

Intra-regional knowledge spillovers will positively affect IC innovation ecosystem resilience.

Knowledge spillover can also play a role in different spatial scales. Sergio Ivan [70] believes that a region does not carry out activities in isolation, that spatial proximity allows knowledge to flow from one region to another through a variety of ways, such as technology transfer, research cooperation, and labor mobility, and that inter-regional knowledge spillover is important for the development of new technologies in neighboring regions. The theory of endogenous economic growth suggests that because knowledge is characterized by externality and dynamism, there is a higher probability of knowledge spillover effects between regions. Knowledge spillover is usually composed of talent flow, R&D cooperation, entrepreneurial innovation, and trade and investment [71]. There is a strong need to deeply analyze the inter-regional knowledge spillover effect. As spatial factors have been gradually introduced into the mainstream category of technological innovation, more and more studies have begun to study the impact of geographic factors on knowledge spillovers. Mao [72] believes that geographic distance can have a significant impact on knowledge spillovers and further studies the impact of heterogeneous regional knowledge resources on enterprise innovation. Inter-regional knowledge spillovers have a positive impact on innovation in all technological fields, promote regional economic development, and present the two main characteristics of diminishing effect and complementary effect [73]. Knowledge spillover from developed regions to backward regions can avoid repeated investment in knowledge technology and continuously attract a new batch of innovative subjects to carry out knowledge cooperation, thus forming a virtuous cycle and impacting IC innovation ecosystem resilience. In this paper, in order to make up for the shortcomings of the neoclassical economic growth theory that will be exogenous to technological progress, economic growth will be endogenized, highlighting the role of human capital and knowledge spillovers in the promotion of resilient growth. Polidoro [74] believes that the knowledge spillover effect exists between the associated industries, which is a benign result of the mutual exchanges of human capital in the associated industries. When human capital and industrial structure are highly matched, higher levels of human capital will promote the upgrading of industrial structure by increasing the supply of skilled professionals and knowledge spillovers, reducing the time cost of technological learning, and increasing the ability to cope with risks. In solving the “stuck-neck” problem, innovation subjects between different regions will be connected through the upstream and downstream of the industrial chain and realize the flow of human capital and technical information through the symbiotic network [75]. When frequent cross-regional interactions between innovation subjects occur, knowledge spillovers will be triggered between regions [76]. Therefore, inter-regional differences in human capital levels will largely affect knowledge spillovers and technology diffusion, and knowledge spillovers will lead to the resilience governance of innovation ecosystems in neighboring regions.

Hypothesis H2.

There are spatial spillovers in the impact of inter-regional knowledge spillovers on IC innovation ecosystems, and inter-regional knowledge spillovers not only promote the resilience of IC innovation ecosystems in the region, but also act on the resilience of IC innovation ecosystems in neighboring provinces.

2.4. The Moderating Role of Knowledge Governance Mechanisms

Innovative subjects cooperate in different ways, and knowledge spillover methods are also diversified, while whether innovative subjects can absorb this external knowledge and to what extent they can utilize it depend on knowledge governance. The concept of knowledge governance emphasizes the effective management and control of the process of creating, disseminating, storing, using, and protecting knowledge by adopting certain organizational institutions and coordinating mechanisms to achieve organizational goals and enhance competitiveness. Grandori [77] was the first to put forward the theory of knowledge governance on the basis of reflecting on the mainstream enterprise management theories. Min et al. [78] believe that knowledge governance will have a great impact on knowledge sharing, transferring, and absorbing. Moradi [79] believes that knowledge governance will have a great impact on knowledge sharing, transferring, and absorbing. Aslam [80] conducted in-depth research and believes that it is necessary to form appropriate knowledge activities and mechanisms in order to realize the optimization of knowledge value. Zhang [81] believes that knowledge governance is used to regulate the rules and modes in the process of knowledge creation, use, and sharing and expands the object of research on knowledge governance to the entire economic system. Existing research basically agrees that the ultimate goal of knowledge governance is to optimize the value of its own knowledge. Most of the existing research on knowledge governance focuses on the coordination of activities within enterprises and less on the impact of knowledge governance mechanisms on the resilience of the entire innovation ecosystem from the perspective of the knowledge spillover space.

Innovative subjects have a certain tendency to profit. Therefore, in the process of joint cooperation of innovative subjects, certain mechanisms and systems are needed to ensure that both sides maintain a stable knowledge spillover. The contractual governance mechanism is a form of governance that manages and coordinates the relationship between individuals or organizations through a clear contract. It emphasizes the use of contracts as a basis to regulate the behavior of participants and achieve optimal allocation of resources and cooperation. The contract governance mechanism clarifies the rights and obligations of each subject in the innovation ecosystem by entering into a formal agreement. As the depth of open innovation becomes deeper, the more innovative subjects seek external cooperation and share core knowledge, the more they need more complete and stricter contract governance to ensure the rights and interests of their own knowledge resources. Wu [82] believes that effective contract governance solves the problem of goal inconsistency and information asymmetry. Li [83] et al. suggest that core firm contractual governance facilitates knowledge sharing among firms in innovation ecosystems. Scholars such as Li [84] find that contractual governance also plays a complementary role in facilitating knowledge acquisition and furthermore facilitates the role of relational governance mechanisms in influencing tacit knowledge. A perfect contract provides a guarantee for knowledge spillover through a legalized agreement, so as to reduce the opportunism among subjects and avoid the risks caused by the transfer of knowledge and technology from one party to another. Innovative subjects can clearly stipulate the number of resources invested by participating subjects, the type of information sharing, the mode and rules of cooperation, etc., and clarify the direction of knowledge sharing and transfer and the scope of knowledge use and development; the contract governance mechanism is conducive to improving the efficiency of knowledge utilization among innovative subjects. In addition, existing research points out that contract governance provides a favorable foundation for value co-creation and subject linkage within the ecosystem, and formal legal agreements bring good external security for interacting parties to ensure that all parties within the ecosystem are actively achieving win-win goals. Based on this, this paper proposes the following:

Hypothesis H3a.

The contractual governance mechanism can enhance the impact of intra-regional knowledge spillovers on the resilience of IC innovation ecosystems.

Hypothesis H3b.

The contractual governance mechanism can enhance the impact of inter-regional knowledge spillover on the resilience of IC innovation ecosystems.

When partners in a region espouse analogous values and cultural tenets, it is more conducive to the establishment of long-term and stable relationships of trust. Consequently, it is imperative for enterprises and partners to establish a direct and effective information communication mechanism to facilitate the sharing of innovation resources, risks, and costs [85]. This approach will enhance the resilience of the innovation ecosystem. The transfer of knowledge among innovation subjects is primarily evidenced by the exchange between the source of knowledge and the recipient of knowledge [86]. Those who stand to gain from the exchange of knowledge tend to select content aligned with their own development objectives and disseminate it to other innovation subjects. Conversely, those who stand to lose from such an exchange actively seek out partners with whom they can collaborate and share knowledge. However, as the distance between the source and recipient of the information increases, the accuracy and integrity of the information may be compromised. Furthermore, there may be instances where the superior knowledge provider is reluctant to share their knowledge, and this is subsequently misappropriated by the collaborator [87]. This can have a detrimental impact on the competitive advantage of the innovation ecosystem, and consequently, the resilience of the ecosystem as a whole. It is therefore evident that there is a necessity for an effective practice based on reciprocity and mutual benefit in socio-interpersonal relationships within the collaborative process [88]. This process is moderated by the use of informal governance mechanisms, namely relational governance mechanisms. Relational governance mechanisms are commonly employed in repetitive and anticipatory developmental relationships, thereby facilitating the exchange of knowledge to activate unforeseen innovation values and generate greater flexibility in the collaborative process [89]. The coexistence of competitive and cooperative relationships among the subjects of the innovation ecosystem, coupled with the greater emphasis on the coupling of heterogeneous knowledge in this open innovation context, renders relational governance a beneficial approach for core firms and member firms of the innovation ecosystem to realize value co-creation in the process of long-term cooperation. In light of the aforementioned, this paper puts forth the following proposition:

Hypothesis H4a.

The relationship governance mechanism has the potential to augment the influence of intra-regional knowledge spillover on the resilience of IC innovation ecosystems.

Hypothesis H4b.

The relationship governance mechanism can enhance the impact of inter-regional knowledge spillover on the resilience of IC innovation ecosystems.

In light of the fact that knowledge spillover represents a significant aspect of innovation, and given that it is subject to the formation of symbiotic relationships and mutual benefit, it is also a crucial factor influencing the resilience of innovation ecosystems; this paper aims to investigate the impact of both intra- and inter-regional knowledge spillover, as well as the impact of knowledge governance mechanisms, on the resilience of innovation ecosystems in the field of integrated circuits. To this end, the theoretical framework illustrated in Figure 2 is presented.

3. Research Design

3.1. Sample Selection and Data Sources

(1)

Explained Variables

Integrated circuit innovation ecosystem resilience (Icr): The MEREC-CoCoSo method was used to measure the toughness index system established in Section 2.1; MEREC was used to objectively determine the standard weights, and CoCoSo was used to reveal the toughness level of IC innovation ecosystems in each province. This method can effectively reduce subjectivity, improve the objectivity and scientificity of evaluation, and contribute to more objective and reasonable measurement results of the resilience level of the innovation ecosystem in the IC industry.

Scholars have employed a variety of techniques to assess the resilience of innovation ecosystems. These include hierarchical analysis, the entropy weight method, principal component analysis, and TOPSIS, among others. Nevertheless, in the majority of extant studies, the weights assigned to the indicators are typically predetermined by experts, which introduces a significant degree of subjectivity based on their personal preferences. The objective determination of the relative importance of each indicator represents a significant challenge. Furthermore, the evaluation methods employed in existing theoretical studies are relatively straightforward and conventional, with some inherent limitations. Accordingly, the objective of this study is to propose a novel multi-criteria decision-making (MCDM) model for assessing innovation ecosystem resilience. This model, designated as the MEREC-CoCoSo framework, is based on two recently developed MCDM techniques: the method of criteria-removing effects (MEREC) and the gray-based combined compromise solution (CoCoSo) [90]. First, the MEREC method is employed to ascertain the objective weights of the criteria. Subsequently, the CoCoSo approach is utilized to ascertain the resilience level of each regional IC innovation ecosystem, thereby obtaining its resilience index. This enables the ranking of the resilience levels of IC innovation ecosystems in different provinces.

The specific calculation steps of the MEREC method are as follows:

Step 1: Construction of the decision matrix $X={\left(x_{ij}\right)}_{m\times n}$.

$$X=\left(\begin{array}{cccc}{x}_{11}& x_{12}& \cdots & x_{1n}\\ x_{21}& x_{22}& \cdots & x_{2n}\\ ⋮& ⋮& \ddots & ⋮\\ x_{m1}& x_{m2}& \cdots & x_{mn}\end{array}\right)$$

(1)

Step 2: Normalization of the decision matrix $N={\left(n_{ij}\right)}_{m\times n}$.

$$n_{ij}=\left\{\begin{array}{l}{\displaystyle \frac{\underset{1\le i\le m}{\min }{x}_{ij}}{x_{ij}}},\mathrm{for} \mathrm{benefit} \mathrm{criteria}\\ {\displaystyle \frac{x_{ij}}{\underset{1\le i\le m}{\max }{x}_{ij}}},\mathrm{for} \mathrm{cost} \mathrm{criteria}\end{array}\right.$$

(2)

Step 3: Calculation of the overall performance of the alternative ($H_i$).

$$H_i=\ln \left(1+\left({\displaystyle \frac{1}{m}}{\displaystyle \sum _j\left|\ln \left(n_{ij}\right)\right|}\right)\right)$$

(3)

Step 4: Calculation of the performance of the alternatives after the removal of each criterion ($H_{ij}^{\prime }$).

$$H_{ij}^{\prime }=\ln \left(1+\left({\displaystyle \frac{1}{m}}{\displaystyle \sum _{k,k\ne j}\left|\ln \left(n_{ik}\right)\right|}\right)\right)$$

(4)

Step 5: Computation of the summation of absolute deviations ($E_j$).

$$E_j={\displaystyle \sum _i\left|H_{ij}^{\prime }-H_i\right|}$$

(5)

Step 6: Determination of the final weights of the criteria.

$$w_j={\displaystyle \frac{E_j}{{\displaystyle \sum _{j=1}^n{E}_j}}}$$

(6)

The specific algorithm for the COCOSO method is as follows:

Step 1: Construction of the decision matrix $X={\left(x_{ij}\right)}_{m\times n}$.

$$X=\left(\begin{array}{cccc}{x}_{11}& x_{12}& \cdots & x_{1n}\\ x_{21}& x_{22}& \cdots & x_{2n}\\ ⋮& ⋮& \ddots & ⋮\\ x_{m1}& x_{m2}& \cdots & x_{mn}\end{array}\right)$$

(7)

Step 2: Normalization of the decision matrix $R={\left(r_{ij}\right)}_{m\times n}$.

$$r_{ij}=\left\{\begin{array}{l}{\displaystyle \frac{x_{ij}-\underset{1\le i\le m}{\min }{x}_{ij}}{\underset{1\le i\le m}{\max }{x}_{ij}-\underset{1\le i\le m}{\min }{x}_{ij}}},\mathrm{for} \mathrm{benefit} \mathrm{criteria}\\ {\displaystyle \frac{\underset{1\le i\le m}{\max }{x}_{ij}-x_{ij}}{\underset{1\le i\le m}{\max }{x}_{ij}-\underset{1\le i\le m}{\min }{x}_{ij}}},\mathrm{for} \mathrm{cost} \mathrm{criteria}\end{array}\right.$$

(8)

Step 3: Computation of the weighted comparability sequence $S$ and its power weight $P$. The sum of the weighted comparability sequences and the whole of their power weights for each alternative, denoted by $S_i$ and $P_i$, are calculated as follows:

$$S_i={\displaystyle \sum _{j=1}^m{w}_j{r}_{ij}}$$

(9)

$$P_i={\displaystyle \sum _{j=1}^m{\left(r_{ij}\right)}^{w_j}}$$

(10)

Step 4: Calculation of the relative weights of the alternatives. Three appraisal score strategies that have been used to determine the relative weights of all other alternatives are computed using the following aggregation formulae:

$$k_{ia}={\displaystyle \frac{P_i+S_i}{{\displaystyle {\sum }_{i=1}^n\left(P_i+S_i\right)}}}$$

(11)

$$k_{ib}={\displaystyle \frac{S_i}{\underset{1\le i\le m}{\min }{S}_i}}+{\displaystyle \frac{P_i}{\underset{1\le i\le m}{\min }{P}_i}}$$

(12)

$$k_{ic}={\displaystyle \frac{\lambda S_i+\left(1-\lambda \right)P_i}{\lambda \left(\underset{1\le i\le m}{\max }{S}_i\right)+\left(1-\lambda \right)\left(\underset{1\le i\le m}{\max }{P}_i\right)}},0\le \lambda \le 1.$$

(13)

Step 5: Final ranking of alternatives in order of preference. The final preferential ranking of the alternatives is based on the previously calculated $k_i$ scores:

$$k_i={\left(k_{ia}{k}_{ib}{k}_{ic}\right)}^{{\displaystyle \frac{1}{3}}}+{\displaystyle \frac{1}{3}}\left(k_{ia}+k_{ib}+k_{ic}\right)$$

(14)

(2)

Explanatory variables

Intra-regional knowledge spillover (Nksp): Intra-regional knowledge spillover is primarily observed in the realization of value associated with innovation elements within the symbiotic network of innovation subjects. The joint patent data of multiple subjects can, to a certain extent, reflect the pivotal aspect of the innovation ecosystem’s value outcome. In accordance with the studies conducted by Xie [91] and Sargento [92], the number of joint patents held by the government, industry, academia, and research institutions is employed as a measure of knowledge spillover. Specifically, this paper employs the incoPat database to obtain the number of effective inventions of IC patents for each province with a cooperation object greater than one. This figure is then subjected to logarithmic processing.

Inter-regional knowledge spillover (Iksp): Inter-regional knowledge spillover is mainly manifested in the interaction and exchange of innovation subjects between different provinces and the knowledge spillover effect brought about by the integration of knowledge, information, and technology. Therefore, this paper obtains the number of joint patent applications involving agents from different regions in each province from the incoPat database [93]. This figure is then subjected to logarithmic processing.

(3)

Moderating variables

Contractual knowledge governance mechanism (Cgo): The contractual governance mechanism gives the participating subjects responsibilities and rights and provides a reliable guarantee for knowledge activities through contracts and other means, so the total amount of technology contract turnover is selected as a proxy variable for the contractual governance mechanism [94].

Relationship governance mechanism (Rgo): Rgo focuses more on the organizational atmosphere, and the diameter of the cooperation network among innovation subjects in each province is measured year by year in Ucinet software (Ucinet6.0) as a proxy variable for Rgo [95].

(4)

Control variables

The enhancement of innovation ecosystem resilience is a complex process, and factors that may have an impact on it are analyzed as control variables with reference to relevant studies. These include the level of technological development (tec, which is expressed using the number of high-tech enterprises in each province, and the level of transportation infrastructure (tin), which is expressed using the logarithm of the number of railroad and highway mileage owned per square area. Financial development level (fin) is operationalized as the ratio of year-end financial institution loan balances to regional gross domestic product (GDP). Social consumption level (con) is expressed using the logarithm of total retail sales of consumer goods divided by the total population at the end of the year. The control variables have been selected for the following reasons: Firstly, in regions with high technological development, production efficiency is likely to be higher and more innovative, which will affect the results of the research object [96]. Secondly, the transportation infrastructure has a direct impact on the logistics costs, mobility of people, and the degree of market connectivity between regions. A well-developed transportation network may facilitate regional economic growth. Therefore, controlling for this variable may reduce bias in economic or innovation capacity due to differences in transportation conditions. The level of financial development affects the capacity of firms to finance and manage risk. This has an impact on firms’ innovation and R&D investment. The level of financial development is controlled for in order to isolate the effect of this factor and focus on the role of other variables on innovation and business activity. In conclusion, the level of social consumption reflects the purchasing power and consumption habits of the population, which have a significant impact on economic growth and firms’ innovative performance. Consequently, it is used as a control variable in this study.

The sample of this paper covers the period of 2011–2021, and the data are mainly derived from the following sources: (1) IC innovation ecosystem resilience measures are derived from the IC-related panel data of 30 provinces (autonomous regions and municipalities directly under the central government) in China from 2011–2021, and due to the limitation of data availability, this study’s sample does not include Tibet, Hong Kong, Macau, and Taiwan. Data are from the China Statistical Yearbook, China High-Tech Industry Statistical Yearbook, China Science and Technology Statistical Yearbook, and other yearbooks. (2) Knowledge overflow data are from the Incopat Patent Information Database; after literature review and word frequency analysis, IC chips were divided into four parts, namely material, design, manufacturing, and sub-assembly, for searching, and the search formula is shown in the table below. The search time is from 1 January 2012 to 31 December 2021, selecting “China”, “validity”, and “invention patent”, and excluding the single subject of the application. Relevant patents of a single applicant are excluded. (3) The contractual knowledge governance mechanism data come from the China Science and Technology Statistical Yearbook, and the relational knowledge governance mechanism data come from Incopat Patent Information Database. (4) The data of control variables are mainly from the National Statistical Yearbook, China Industrial Statistical Yearbook, and China Science and Technology Statistical Yearbook, and some missing values are made up by the interpolation method.

3.2. Model Construction

(1)

Fixed-Effect Model

In order to test the mechanism of the impact of intra-regional knowledge spillover on the resilience of IC innovation ecosystem, this paper adopts the two-way fixed-effect model (two-way FE) to test the above theoretical hypotheses. The two-way fixed-effect model can combine time series data and cross-sectional data to analyze inter-individual differences and individual dynamics and at the same time effectively expand the sample size. The two-way fixed-effect model can effectively capture the unobservable heterogeneity among individuals as well as the temporal heterogeneity of individuals to improve the estimation accuracy of the model. The model is defined as follows:

$${Icr}_{\mathrm{i},\mathrm{t}}={\alpha }_1+{\beta }_1{Nksp}_{\mathrm{i},\mathrm{t}}+{\delta }_1{Z}_{\mathrm{i},\mathrm{t}}+U_t+V_{\mathrm{i}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(15)

where $\mathrm{i}$ denotes province, $\mathrm{t}$ denotes year, ${\alpha }_1$ is a constant term; $Ic{r}_{\mathrm{i},\mathrm{t}}$ is the IC innovation ecosystem resilience; ${Nksp}_{\mathrm{i},\mathrm{t}}$ is the intra-regional knowledge spillover effect; $Z_{\mathrm{i},\mathrm{t}}$ denotes control variables such as the level of technological development, the level of transportation infrastructure, and the level of social consumption; ${\beta }_1$ denotes the coefficient of the impact of knowledge spillovers on the resilience of IC innovation ecosystems; ${\delta }_1$ denotes the coefficient of the impact of control variables on the resilience of IC innovation ecosystems; $U_t$ denotes controlled time fixed effects; $V_{\mathrm{i}}$ denotes controlled provincial fixed effects; ${\epsilon }_{\mathrm{i},\mathrm{t}}$ is a randomized disturbance term.

(2)

Spatial Durbin Model

In order to explore the spatial effect of inter-regional knowledge spillovers on the resilience of innovation ecosystems, this paper incorporates spatial factors into the study and improves the accuracy of the empirical results by eliminating the regression bias brought about by the spatial correlation of the dependent variable. To this end, the spatial Durbin model is constructed to test its accuracy. Models (16)–(18) represent the spatial autocorrelation model (SAR), the spatial error model (SEM), and the spatial Durbin model (SDM), respectively. The LM test, LR test, and Wald test were used to identify the use of spatial econometric models.

$${Icr}_{\mathrm{i},\mathrm{t}}={\alpha }_1+\gamma {WIcr}_{\mathrm{i},\mathrm{t}}+{\beta }_1{Iksp}_{\mathrm{i},\mathrm{t}}+\mathrm{W}{Iksp}_{\mathrm{i},\mathrm{t}}\theta +{\delta }_1{Z}_{\mathrm{i},\mathrm{t}}+U_t+V_{\mathrm{i}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(16)

When $\theta =0$, the SDM model is transformed into an SEM model, at which point the model expression is

$${Icr}_{\mathrm{i},\mathrm{t}}={\alpha }_1+\gamma {WIcr}_{\mathrm{i},\mathrm{t}}+{\beta }_1{Iksp}_{\mathrm{i},\mathrm{t}}+{\delta }_1{Z}_{\mathrm{i},\mathrm{t}}+U_t+V_{\mathrm{i}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(17)

When $\theta +\rho \beta =0$, the SDM model is transformed into a SAR model, at which point the model expression is

$${Icr}_{\mathrm{i},\mathrm{t}}={\alpha }_1+{\beta }_1{Iksp}_{\mathrm{i},\mathrm{t}}+{\delta }_1{Z}_{\mathrm{i},\mathrm{t}}+U_t+V_{\mathrm{i}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(18)

where $W$ is the spatial weight matrix and $\gamma$ is the coefficient.

(3)

Moderating effect model

In order to explore the moderating effect of the knowledge governance mechanism on intra-regional and inter-regional knowledge spillovers and the resilience of the IC innovation ecosystem, this paper adds the interaction term between the baseline regression model and the spatial Durbin to construct a moderating effect model to test the above theoretical hypotheses.

$${Icr}_{\mathrm{i},\mathrm{t}}={\alpha }_1+{\beta }_1{Nksp}_{\mathrm{i},\mathrm{t}}+\varphi {Nksp}_{\mathrm{i},\mathrm{t}}\times Cgo+{\delta }_1{Z}_{\mathrm{i},\mathrm{t}}+U_t+V_{\mathrm{i}}+{\epsilon }_{\mathrm{i},\mathrm{t}}$$

(19)

$$\begin{array}{l}{Icr}_{\mathrm{i},\mathrm{t}}={\alpha }_1+{\beta }_1{Nksp}_{\mathrm{i},\mathrm{t}}+\varphi {Nksp}_{\mathrm{i},\mathrm{t}}\times Rgo+{\delta }_1{Z}_{\mathrm{i},\mathrm{t}}\\ +U_t+V_{\mathrm{i}}+{\epsilon }_{\mathrm{i},\mathrm{t}}\end{array}$$

(20)

$$\begin{array}{l}{Icr}_{\mathrm{i},\mathrm{t}}={\alpha }_2+{\beta }_2{Iksp}_{\mathrm{i},\mathrm{t}}+{\rho }_{2}Cgo+\varphi {Iksp}_{\mathrm{i},\mathrm{t}}\times Cgo+{\theta }_{2}W{Iksp}_{\mathrm{i},\mathrm{t}}\\ +{\theta }_{2}WCgo+{\theta }_{2}W{Iksp}_{\mathrm{i},\mathrm{t}}{U}_t\times Cgo+{\delta }_1{Z}_{\mathrm{i},\mathrm{t}}+{\epsilon }_{\mathrm{i},\mathrm{t}}\end{array}$$

(21)

$$\begin{array}{l}{Icr}_{\mathrm{i},\mathrm{t}}={\alpha }_2+{\beta }_2{Iksp}_{\mathrm{i},\mathrm{t}}+{\rho }_{2}Rgo+\varphi {Iksp}_{\mathrm{i},\mathrm{t}}\times Rgo\\ +{\theta }_{2}W{Iksp}_{\mathrm{i},\mathrm{t}}+{\theta }_{2}WRgo+{\theta }_{2}W{Iksp}_{\mathrm{i},\mathrm{t}}{U}_t\times Rgo+{\delta }_1{Z}_{\mathrm{i},\mathrm{t}}+{\epsilon }_{\mathrm{i},\mathrm{t}}\end{array}$$

(22)

where $Cgo$ is the contractual governance mechanism and $Rgo$ is the relational governance mechanism.

4. Analysis of Empirical Results

4.1. Analysis of Knowledge Spillover and Innovation Ecosystem Resilience Evolution Trajectory

Intra- and inter-regional knowledge spillovers and IC innovation ecosystem resilience measures are selected from the IC-related panel data of 30 provinces (autonomous regions and municipalities directly under the central government) in China from 2011 to 2021, and the samples of this study do not include Tibet, Hong Kong, Macao, and Taiwan due to the limitation of data availability. In order to more intuitively show the spatial and temporal evolution of knowledge spillovers and IC innovation ecosystem resilience, this paper intends to use the kernel density estimation method to explore the distribution of the main peaks, the polarization trend, and the ductility of the knowledge spillovers and innovation ecosystem resilience during the evolution process. Figure 3a–c show the evolution trajectories of intra-regional knowledge spillover, inter-regional knowledge spillover, and national IC innovation ecosystem resilience from 2011 to 2021.

Figure 3a examines the three-dimensional kernel density distribution of knowledge overflow in the region. Based on the movement of the wave peak, the position of the main peak of the distribution curve demonstrates a tendency to shift to the right and then to the left. This indicates that the level of knowledge overflow in the region initially increased but subsequently exhibited a downward trend in recent years. From the perspective of the polarization phenomenon, the “multi-peak” state of the kernel density curve is gradually diminishing, which suggests that the polarization phenomenon between regions is gradually becoming less pronounced. Figure 3b examines the distribution dynamics of three-dimensional kernel density of knowledge overflow between regions. The peak fluctuation of the main peak initially rises and then falls, exhibiting an inverted U-shaped trend. Figure 3c examines the evolution of the distribution of the three-dimensional kernel density of the national IC innovation ecosystem. The position of the main peak of the distribution curve demonstrates a tendency to shift to the right, followed by a shift to the left. Additionally, the peak of the main peak exhibits a notable downward trajectory in 2017 and 2018. This shift marks a pivotal point in the development of China’s IC industry, with the United States beginning to view China as a formidable competitor. In 2019, China’s technological advancements, which had previously been regarded as “breakthroughs”, were elevated to the status of national strategy. This resulted in a notable increase in the level of competitiveness. The right trailing feature of the kernel density curve is evident, indicating an expansion of the interprovincial disparity. With regard to polarization, the presence of multiple peaks in the kernel density curve indicates a gradual increase in the polarization of the resilience of innovation ecosystems across provinces.

4.2. The Impact of Intra-Regional Knowledge Spillovers on the Resilience of IC Innovation Ecosystems

4.2.1. Descriptive Statistics and Correlation Analysis

After screening, this paper obtains a total of 330 observed variables in 30 provinces (autonomous regions and municipalities directly under the central government) from 2011 to 2021 as the research sample. The descriptive statistics of the main variables are shown in

Table 3. Descriptive statistics of relevant variables.

	Sample Size	Average Value	Upper Quartile	Standard Deviation	Minimum Value	Maximum Values
Icr	330	2.098	1.860	0.921	1.140	6.640
Nksp	330	3.009	3.091	2.172	0	8.928
Iksp	330	0.693	0.555	0.641	0.00600	3.960
Rgo	330	0.650	0.625	0.134	0.410	0.938
Cgo	330	1.320	1.282	0.401	0.100	3.043

.

Table 4. Pearson correlation coefficient matrix.

Variable	Icr	Nksp	Iksp	Rgo	Cgo	tec	tin	fin	con
Icr	1
Nksp	0.686 ***	1
Iksp	0.571 ***	0.394 ***	1
Rgo	−0.100 *	−0.0390	−0.114 **	1
Cgo	0.320 ***	0.497 ***	0.127 **	−0.117 **	1
tec	0.843 ***	0.541 ***	0.461 ***	−0.156 ***	0.305 ***	1
tin	0.0190	−0.0450	0.0380	−0.219 ***	0.0550	0.186 ***	1
fin	0.787 ***	0.716 ***	0.347 ***	−0.0540	0.503 ***	0.785 ***	0.319 ***	1
con	0.251 ***	0.415 ***	0.158 ***	−0.217 ***	0.302 ***	0.162 ***	0.203 ***	0.415 ***	1
VIF		1.37	2.87	1.23	1.50	3.62	1.50	6.22	1.50

Values in brackets are standard errors; ***, **, and * are significant at the levels of 1%, 5%, and 10%, respectively.

illustrates the Pearson correlation coefficient matrix for each variable. The results indicate that there is a statistically significant positive correlation between intra-regional and inter-regional knowledge spillover and IC innovation ecosystem resilience. Furthermore, the direction of the correlation coefficients aligns with the research hypotheses. As a result, the correlation between the variables is preliminarily verified. Furthermore, to circumvent the issue of multicollinearity, this study conducted a multicollinearity test. The results presented in Table 4 demonstrate that the variance inflation factor (VIF) of each explanatory variable is less than the critical value of 10, suggesting that the multicollinearity problem is unlikely to be a significant concern.

4.2.2. Benchmark Regression Analysis

In this paper, the Hausman test was used to determine whether a fixed-effect model could be constructed. The test results show a statistic of 124.65, which is significant at the 1% test level, and it can be preliminarily determined that a fixed-effect model should be chosen to be constructed. On this basis, the variables’ intra-group autocorrelation and inter-group heteroskedasticity were tested, and the statistics were 163.930 and 192.19, respectively, which were both significant at the 1% test level rejecting the original hypothesis, further ruling out autocorrelation between the variables, and the fixed-effect model should be constructed. To ensure the reliability of the test results, a unit root test was conducted for each variable using IPS and LLC; both passed the smoothness test at the 1% level. Further, using the Pedroni cointegration test, it was found that the original hypothesis is rejected at a 5% significance level and there is a long-run equilibrium cointegration relationship (

Table 5. Pedroni cointegration test results.

	Statistic	p
Modified Phillips–Perron t	2.0232	0.0215
Phillips–Perron t	−1.8965	0.0289
Augmented Dickey–Fuller t	−3.3236	0.0004

).

Therefore, a panel fixed-effect model is employed to examine the influence of intra-regional knowledge spillovers on the resilience of IC innovation ecosystems (

Table 6. Influence of knowledge spillover on the resilience of integrated circuit innovation ecosystem in the region.

	(1)	(2)	(3)	(4)
	Icr	Icr	Icr	Icr
Nksp	0.060 ***	0.053 ***	0.026 ***	0.032 ***
	(0.012)	(0.009)	(0.004)	(0.004)
Cgo			0.268 *
			(0.147)
Rgo				0.011
				(0.022)
Nksp × Cgo			0.088 ***
			(0.013)
Nksp × Rgo				−0.015 ***
				(0.004)
tec		0.039	−0.012	0.004
		(0.043)	(0.016)	(0.017)
tin		0.679 ***	0.758 ***	0.779 ***
		(0.157)	(0.104)	(0.108)
fin		0.011	0.034 **	0.019
		(0.020)	(0.013)	(0.014)
con		−0.000	−0.022	0.041
		(0.064)	(0.026)	(0.028)
_cons	1.905 ***	2.028 ***	2.189 ***	2.209 ***
	(0.035)	(0.042)	(0.028)	(0.031)
N	330	330	330	330
R2	0.397	0.488	0.565	0.515
id	yes	yes	yes	yes
year	yes	yes	yes	yes

Values in brackets are standard errors; ***, **, and * are significant at the levels of 1%, 5%, and 10%, respectively.

). The results of Models (1) and (2) indicate that intra-regional knowledge spillovers consistently exert a substantial positive influence on the resilience of IC innovation ecosystems, irrespective of the inclusion or exclusion of control variables. This is due to the fact that intra-regional knowledge spillovers assist in the enhancement of the regional knowledge stock and diversity, as well as the expansion of the cognitive boundaries of innovation subjects through the processes of knowledge flow, absorption, and reorganization. This, in turn, enables the region to deviate from its original path dependence and achieve breakthroughs in IC technology, thereby promoting innovation ecosystem resilience. At this juncture, hypothesis H1 is provisionally validated.

4.2.3. Analysis of Moderating Effects

Model (3) and Model (4) discuss the impact of the relational governance mechanism and contractual governance mechanism on the resilience of IC innovation ecosystems; the coefficient of the interaction term between the moderating variable contractual governance and the explanatory variables is positive (0.088), and the estimated coefficient is significant at the 1% level. This indicates that the contractual governance mechanism can significantly enhance the impact of intra-regional knowledge spillovers on innovation ecosystem resilience, and to a certain extent safeguard the flow of knowledge within the regional innovation ecosystem, overcome the information dilemma faced by the single-subject innovation, and thus positively affect the regional innovation ecosystem resilience, and hypothesis H3a is valid.

The coefficient of the interaction term between relationship governance and explanatory variables is negative (−0.015), and the estimated coefficient is significant at the 1% level, indicating that the relationship governance mechanism has an inhibitory effect on the governance of IC innovation ecosystem resilience in the process of intra-regional knowledge spillovers; hypothesis H4a is not valid. It can be seen that the knowledge governance mechanism also cannot guarantee a sustained and effective role in knowledge governance, and the relationship governance mechanism puts more emphasis on the organizational climate and reduces inter-subjective conflicts. However, the mismatch between the characteristics of knowledge and its transmission method will increase the risk of knowledge spillover to a certain extent.

Intra-regional knowledge spillover needs to rely on advanced technology or institutional culture to visualize tacit knowledge; in the middle technology proximity, technology exchanges cannot generate economies of scale through similar technology interaction, and it is also difficult to form a variety of technology exchanges generated by the scope of the economy, so the knowledge spillover benefits are low. Innovative subjects in a long-term mismatch have inertia in absorbing heterogeneous knowledge, which will “lock in” knowledge in a certain region, thus negatively affecting the resilience of the innovation ecosystem.

4.2.4. Robustness Test and Endogeneity Treatment

Further robustness tests of the regression results were performed to ensure the reliability and rationality of the regression. This paper mainly adopts four kinds of tests: shrinking treatment of the explanatory variables, lagging the explanatory variables by one period, replacing the explanatory variables, and instrumental variables (

Table 7. Robustness and endogeneity test results.

	Tailing Treatment (1)	One-Phase Lag (2)	Alternate Explanatory Variable (3)	Instrumental Variable (4)
Tailing treatment	0.092 ***
	(0.020)
L.Nksp		0.028 ***
		(0.009)
Alternate explanatory variable			0.074 ***
			(0.017)
Instrumental variable				0.0876 ***
				(3.25)
Control variable	Yes	Yes	Yes	Yes
Time variable	Yes	Yes	Yes	Yes
Province variable	Yes	Yes	Yes	Yes
LM statistics				8.38 **
Wald F statistic				173.201 ***
N	272	300	330	300
R2	0.557	0.497	0.492	0.489

Values in brackets are standard errors; *** and ** are significant at the levels of 1% and 5%, respectively.

).

(1) Tailoring treatment. In order to eliminate the impact of extreme outliers on the results of the study, this paper adopts the reduced-tail treatment for robustness testing and bilateral reduced-tail treatment at the 1% quantile for the explanatory variables and then uses the new samples to re-run the regression analysis. As the results in column (1) of Table 7 show, the coefficient of the impact of knowledge spillover on the resilience of the innovation ecosystem is 0.092, which passes the test of significance at the 1% level, with the sign of the coefficient and the level of significance being consistent with those of the benchmark regression, and with only a slight change in the coefficient magnitude, and the basic conclusions of the previous paper still hold.

(2) Explanatory variables lagged by one period. Since the effect of knowledge spillovers requires a longer period of time to show up, this paper thus lags it by one period to re-examine the effect of knowledge spillovers on the resilience of the innovation ecosystem to weaken the potential reverse causation problem. As the results in column (2) of Table 7 show, the regression coefficient (0.028) is positively significant at the 1% statistical level, indicating that the results of this paper are not affected by the reverse causation problem.

(3) Replacing explanatory variables. In this paper, the number of patent applications is used to recalculate the knowledge spillover in the region, and the regression is re-run, as shown in the results of column (3) of Table 7, which passes the test of significance, indicating that the study’s conclusion that the knowledge spillover can enhance the resilience of the innovation ecosystem is relatively robust.

(4) Endogeneity test. There may be an endogeneity issue between knowledge spillovers and innovation ecosystem resilience. This is because the more stable the system is, the more efficient the knowledge flow among innovation agents is, which promotes knowledge spillover. Therefore, this paper addresses this issue with the help of the instrumental variable method, taking the explanatory variable lagged one period as an instrumental variable and using the 2SLS estimation method to re-test the regression model. As the results in column (4) of Table 7 show, in the first period, the instrumental variables are significantly and positively related to innovation ecosystem resilience at the 1% statistical level. Moreover, the F-statistic is 173.201, which is greater than 10 and passes the weak instrumental variable test. The second-stage regression results are still positive and significant at the 1% level, indicating that the key results do not change after controlling for endogeneity issues.

4.2.5. Heterogeneity Test

This paper divides the sample into six regions, namely the Yangtze River Delta, Beijing–Tianjin–Bohai Rim, Pan-Pearl River Delta, Central, Northeast, and Northwest, based on the aggregation of IC industries, and examines the regional heterogeneity of the impact of knowledge spillovers among IC innovation agents on the resilience of innovation ecosystems (

Table 8. Heterogeneity of knowledge spillover on innovation ecosystem resilience in a region.

	Yangtze River Delta	Beijing and Tianjin Ring the Bohai Sea	Pan-Pearl River Delta	Middle Part	Northeast	Northwest
Nksp	0.070	0.032	0.061 ***	0.026	0.039	0.030
	(0.054)	(0.017)	(0.014)	(0.015)	(0.016)	(0.018)
tec	0.429 *	0.191	−0.005	0.102	−0.008	−0.184
	(0.153)	(0.111)	(0.239)	(0.148)	(0.080)	(0.329)
tin	0.268 **	0.105 ***	0.006	−0.073	0.136	0.031
	(0.077)	(0.026)	(0.116)	(0.063)	(0.095)	(0.096)
fin	−0.267 **	0.063	0.106	0.012	0.528 **	−0.085
	(0.060)	(0.072)	(0.098)	(0.076)	(0.112)	(0.342)
con	−0.035	−0.165	−0.081	−0.082	0.126	−0.000
	(0.240)	(0.168)	(0.164)	(0.088)	(0.156)	(0.159)
_cons	2.295 ***	1.963 ***	2.086 ***	1.872 ***	1.971 ***	1.113 *
	(0.131)	(0.038)	(0.210)	(0.127)	(0.125)	(0.438)
N	44.000	55.000	99.000	44.000	33.000	55.000
R2	0.851	0.818	0.389	0.814	0.910	0.301
id	Fixed time	Yes	Yes	Yes	Yes	Yes
year	Provincial fixation	Yes	Yes	Yes	Yes	Yes

Values in brackets are standard errors; ***, **, and * are significant at the levels of 1%, 5%, and 10%, respectively.

). The results show that the Pan-PRD region has the best effect of knowledge spillover among IC innovation subjects on innovative ecosystem resilience enhancement. As the most important production region for the development of the IC industry, the Pan-PRD region has always maintained a strong development trend and has complexity and uniqueness in the institutional environment and industrial development. On the one hand, the spatial pattern of the network of cluster innovators has changed from a “core–edge” structure with Guangzhou and Shenzhen at its core to a multi-center and multi-polar development, so that innovation resources can flow efficiently within the PRD, and the level of innovation cooperation has been significantly improved, thus generating knowledge spillovers to promote resilient development. On the other hand, as the number and strength of enterprises, universities, and research institutions in the Pan-PRD region continue to increase, the intra-regional knowledge spillover effect is enhanced, the exchange of technology and feedback of information are smoother, and the innovation ecosystem can more efficiently adapt to external risks and industrial development requirements.

4.3. The Impact of Inter-Regional Knowledge Spillover on the Resilience of IC Innovation Ecosystems

4.3.1. Spatial Correlation Analysis

It is necessary to test whether the innovation ecosystem resilience of regional ICs in China is spatially correlated before conducting spatial econometric regression. Considering the different impacts of spatial factors on knowledge spillover, and in order to better explore the role of spatial role, the role of knowledge spillover, and innovation ecosystem resilience between the mechanism, this paper constructs the human capital matrix. Human capital is the key carrier of the spatial flow of innovation factors, which determines the absorptive capacity of the country or region to a certain extent, so this paper constructs the human capital spatial weight matrix as follows: The diagonal element is 0 and the non-primary diagonal element is $W_{\mathrm{ij}}=K{\displaystyle \frac{\left|\overline{{\mathrm{y}}_{\mathrm{i}}}-\overline{{\mathrm{y}}_{\mathrm{j}}}\right|}{{\mathrm{d}}_{\mathrm{ij}}}}$, where $\overline{{\mathrm{y}}_{\mathrm{i}}}$ and $\overline{{\mathrm{y}}_{\mathrm{j}}}$ denote the mean value of the labor force for 2011–2021 for provinces $i$ and $j$, respectively; ${\mathrm{d}}_{\mathrm{ij}}$ is the Euclidean distance between the centers of the two provinces; and K is the adjustment parameter 1000.

Subsequently, this paper measures the Moran’s I index of innovation ecosystem resilience of national ICs from 2011 to 2021 (

Table 9. Moran’s I index of IC innovation ecosystem resilience, 2011–2021.

Variables	I	Z	p-Value
a2011	0.063	2.126	0.017
a2012	0.079	2.450	0.007
a2013	0.079	2.473	0.007
a2014	0.076	2.410	0.008
a2015	0.075	2.425	0.008
a2016	0.078	2.517	0.006
a2017	0.074	2.479	0.007
a2018	0.082	2.666	0.004
a2019	0.078	2.648	0.004
a2020	0.086	2.823	0.002
a2021	0.069	2.603	0.005

). The results show that under the human capital matrix, the Moran’s I index values are all greater than 0, and all of them pass the 10% significance test, which indicates that the innovation ecosystem resilience of each region in China has strong spatial agglomeration. This shows that the spatial agglomeration effect of each region in China has a strong spatial agglomeration, and the spatial agglomeration effect shows a dynamic trend of “rising–declining–rising–declining” from 2011 to 2021. The spatial agglomeration effect rose rapidly in 2012 and began to fluctuate in 2014, indicating that during this period, the gap between the agglomeration level of neighboring regions and that of more distant regions with higher levels of toughness of the IC innovation ecosystem was not large. A gradual rise was seen in 2018, indicating that after the emergence of the “stuck-neck” situation, the regions with higher levels of toughness during this period were able to bring about the enhancement of toughness in the neighboring regions through the radiation effect, and the gap between the distant regions and them became more and more obvious. Starting from 2021, the imbalance in the spatial agglomeration of the toughness of China’s IC innovation ecosystem shows a downward trend.

Subsequently, a Moran scatterplot was used to further analyze the spatial correlation between the IC innovation ecosystem resilience level of each province and the IC innovation ecosystem resilience level of the neighboring provinces, and Figure 4 below shows the local Moran scatterplot for 2011, 2014, 2018, and 2021. From the figure, it can be seen that the IC innovation ecosystem resilience level is more often distributed in the first quadrant, the second quadrant, and the third quadrant, with high–high agglomeration characteristics, low–high discrete characteristics, and low–low aggregation characteristics. Specifically, cities such as Shanghai, Jiangsu, Zhejiang, and Fujian, which continue to be located in the first quadrant, have strong technological innovation capabilities and can better drive the development of surrounding provinces. In these regions, there exists the “Matthew effect”, which has the feature of “one prosperity and one prosperity”; that is, the region has a high value, and its adjacent regions also have a high value, which can give full play to its positive spatial spillover effect. On the other hand, Liaoning, Qinghai, and Inner Mongolia are located in the third quadrant, and these cities themselves have low levels of technological innovation and knowledge spillover, and they do not have a driving effect on the surrounding areas, nor can they play a good role in radiation driving; i.e., the region has a low value, and its neighboring regions also have a low value, with the characteristic of “one loses, all lose”, which is also a “Matthew effect”. Hunan, Chongqing, Jiangxi, Guizhou, and other regions are located in the second quadrant of the low–high dispersion zone; the region’s toughness level is low, and the surrounding toughness level of the stronger provinces and regions is not significant for promoting their role. The provinces of Beijing and Guangdong, located in the fourth quadrant, are high–low aggregation areas, showing a significant negative spatial correlation with the surrounding provinces. The innovation competitiveness of these regions is obviously strong, and the absorption and introduction of resources in the surrounding regions attract the inflow of talents, capital, services, trade, etc., so that the resilient development speed of the weak competitiveness of the surrounding regions is reduced. This paper defines this as the “siphon effect”.

4.3.2. Measurement Model Selection

In this paper, the autocorrelation test is used for spatial econometric model selection. First, model selection is carried out using the LM test, and the results are shown in

Table 10. LM test results.

Test	Statistic	df	p-Value
Spatial error
Lagrange multiplier	46.981	1	0.000
Robust Lagrange multiplier	74.624	1	0.000
Spatial lag
Lagrange multiplier	8.737	1	0.003
Robust Lagrange multiplier	36.381	1	0.000

. The results show that the LM−Lag and LM-Error values obtained under the Lagrangian test are significant, so a further robustness Lagrangian test is needed to obtain Robust LM−Lag and Robust LM-Error values, respectively, and it can be seen that both are under the 5% significance level. After that, further LR and Wald tests are performed to determine whether the spatial Durbin model can degenerate into a spatial lag model or a spatial error model. According to the test results demonstrated in Table 10, both the LR statistic and the Wald statistic reject the original hypothesis at the 1% significance level, indicating that the spatial Durbin model cannot be degraded to a spatial lag model or a spatial error model, and therefore, the spatial Durbin model is finally selected.

4.3.3. Analysis of Spatial Effects

The spatial Durbin model is used to explore the spatial effect of inter-regional knowledge spillover on the resilience of the IC innovation ecosystem. As shown in

Table 11. Results of spatial effect analysis.

	Human Capital Matrix	Geographic Distance Matrix
Iksp	0.033 *** [0.0169]	0.019 ** [0.0086]
W × Iksp	0.110 *** [0.0361]	0.090 *** [0.0287]
W × Icr	−0.278 *** [0.1014]	−0.316 *** [0.1036]
Direct effect	0.029 * [0.0169]	0.015 * [0.0086]
Indirect effect	0.086 *** [0.0296]	0.071 *** [0.0223]
Total effect	0.115 *** [0.0324]	0.086 *** [0.0238]
Control variable	Yes	Yes
Time fixed effect	Yes	Yes
Provincial fixed effect	Yes	Yes
Observed value	330	330
R2	0.416	0.177
AIC	−415.5	−415.6
BIC	−312.9	−313.0

Values in brackets are standard errors; ***, **, and * are significant at the levels of 1%, 5%, and 10%, respectively.

, the spatial autocorrelation coefficient of innovation ecosystem resilience is significantly less than zero, and there is a certain “siphon effect”. Under the human capital matrix, the general regression coefficient and spatial regression coefficient of knowledge spillover are significantly positive, indicating that the knowledge spillover among innovative subjects has both a direct contribution to innovation ecosystem resilience and a positive spatial spillover effect. At the same time, considering the influence of different spatial weight matrices, a geographic distance matrix is added to confirm the robustness of the regression results.

The results of the effect decomposition indicate that the direct effect (0.029), indirect effect (0.086), and total effect (0.115) of inter−regional knowledge spillover are all significantly positive, with all values under the 10% significance level. Additionally, the indirect effect is stronger than the direct effect, suggesting that the knowledge spillover among the IC innovation main bodies enhances the resilience of the innovation ecosystem in the province and neighboring provinces. This phenomenon can be described as a “radiation effect”. Hypothesis H2 of this paper is thus verified.

The development of innovation ecosystem resilience in neighboring provinces is promoted through the spatial spillover effect. The influx and consolidation of a considerable quantity of human capital, coupled with the sharing of knowledge and a proclivity for collaboration in innovation, serve to reinforce the inter−regional knowledge spillover effect, enhance the efficiency of the transformation of technological innovation outcomes, facilitate the dissemination of innovation resources within the region, and facilitate rapid dissemination to neighboring provinces. This has led to the emergence of a phenomenon whereby the benefits of innovation are not confined to the immediate region but extend to neighboring provinces. It is therefore necessary to continue to reinforce cross-regional platforms for sharing resources and to facilitate the flow of talent and technology exchanges in order to achieve closer regional cooperation and integrated development, thereby enhancing the resilience of the innovation ecosystem.

4.3.4. Spatial Moderating Effects of Governance Mechanisms

The moderating role of knowledge governance mechanisms in knowledge spillovers and innovation ecosystem resilience is further considered under spatial effects. The effect decomposition results in

Table 12. The moderating effects of knowledge governance mechanisms on knowledge spillover and innovation ecosystem resilience.

Variable	Main	Wx	Direct Effect	Indirect Effect	Total Effect
Icr	0.124 *** [0.0181]	0.111 * [0.0666]	0.122 *** [0.0178]	0.069 [0.0478]	0.191 *** [0.0545]
Cgo	−0.037 ** [0.0146]	0.113 ** [0.0568]	−0.042 *** [0.0122]	0.103 ** [0.0455]	0.061 [0.0474]
Icr × Cgo	0.049 *** [0.0104]	−0.078 [0.0551]	0.053 *** [0.0101]	−0.074 * [0.0436]	−0.020 [0.0454]
Icr	0.110 *** [0.0197]	0.175 *** [0.0649]	0.104 *** [0.0198]	0.110 ** [0.0443]	0.214 *** [0.0486]
Rgo	−0.012 [0.0114]	−0.073 ** [0.0313]	−0.010 [0.0101]	−0.057 ** [0.0253]	−0.068 *** [0.0258]
Icr × Rgo	0.020 * [0.0114]	0.074 ** [0.0326]	0.018 [0.0125]	0.056 ** [0.0255]	0.074 *** [0.0250]

Values in brackets are standard errors; ***, **, and * are significant at the levels of 1%, 5%, and 10%, respectively.

show the direct effect (0.053), indirect effect (−0.074), and total effect (−0.020) of the contractual governance mechanism in the impact of inter-regional knowledge spillovers on innovation ecosystem resilience, with the direct effect being below the 1% significance level, and hypothesis H3b holds. Also shown are the direct effect (0.018), indirect effect (0.056), and total effect (0.074) of relational governance mechanisms in the impact of inter-regional knowledge spillovers on the resilience of innovation ecosystems, with the indirect and total effects being under the 5% significance level, and the direct effect is not significant; hypothesis H4b is partially established.

The analysis finds that inter-regional knowledge spillovers can better enhance the innovation ecosystem resilience of the province under the contractual governance mechanism and enhance the innovation ecosystem resilience of neighboring provinces under the relational governance mechanism. This is due to the fact that the contractual governance mechanism confers responsibilities and rights on the participating entities and provides reliable guarantees for knowledge activities through contracts and contracts. In the process of inter-regional knowledge spillover, the geographical distance between innovation subjects is shorter, the similarity between technologies is higher, and the core enterprises of knowledge spillover are prone to knowledge hiding and concealment of real information to avoid the problem of technological loss, so it is necessary to rely on the contractual governance mechanism to effectively coordinate the knowledge activities of the innovation ecosystem and constrain the management of innovation behavior of innovation subjects, and the contractual governance mechanism can effectively enhance the impact of knowledge spillover on the innovation ecosystem resilience of the province. The contract governance mechanism can effectively enhance the impact of knowledge spillover on the resilience of the innovation ecosystem in the province. The relational governance mechanism places more emphasis on the organizational atmosphere and reduces inter-subject conflicts. The exchange of innovation subjects between different regions is relatively small and cannot form stable cooperative relationships, so through the moderating effect of the relationship governance mechanism, it can enhance the positive impact of knowledge spillover on the innovation ecosystems of neighboring provinces, but at present, it does not have a significant effect on the resilience of the local area.

4.3.5. Heterogeneity Test

Based on the perspective of regional innovation capacity and according to the “China Regional Innovation Capacity Evaluation Report 2021”, the 30 provinces (autonomous regions and municipalities directly under the central government) are divided into high-innovation-capacity regions and low-innovation-capacity regions for heterogeneity analysis, and the regression results are shown in columns (1) and (2) in

Table 13. Heterogeneity effects of inter-regional knowledge spillover on innovation ecosystem resilience.

	(1)	(2)	(3)	(4)
	High Innovation Region	Low Innovation Region	Talent Base	Talent Depression
Direct effect	0.077 * [0.0177]	0.116 * [0.0675]	0.037 ** [0.0164]	0.038 *** [0.0132]
Indirect effect	0.103 * [0.0555]	−0.243 ** [0.1143]	0.150 * [0.0895]	0.040 [0.0388]
Total effect	0.181 *** [0.0612]	−0.127 [0.1361]	0.187 ** [0.0945]	0.078 * [0.0419]
Control variable	yes	yes	yes	yes
Time fixed effect	yes	yes	yes	yes
Provincial fixed effect	yes	yes	yes	yes
Observed value	165	165	110	220
R2	0.223	0.458	0.301	0.228
AIC	−191.8	−110.6	−232.5	−233.9
BIC	−114.0	−155.7	−151.2	−152.6

Values in brackets are standard errors; ***, **, and * are significant at the levels of 1%, 5%, and 10%, respectively.

. Knowledge spillovers among low-innovation-capacity regions have a positive effect on the resilience of local innovation ecosystems but a negative effect on the resilience of innovation ecosystems in neighboring provinces. This “dual nature” stems mainly from the asymmetry of agglomeration effects: low-innovation-capacity regions have smaller innovation clusters, low marketization, and concentration of innovation resources in a few cities. Within the region, the triple agglomeration of knowledge, talent, and technology manifests itself as a clear facilitator. In contrast, in neighboring cities, these agglomeration effects translate into a “siphoning effect” of technology, leading to a loss of capital and talent, resulting in a greater siphoning effect than radiation in spatially connected areas and thus inhibiting the growth of innovation ecosystem resilience.

Based on the perspective of talent attractiveness, according to the “China City Talent Attractiveness Report 2021”, the 30 provinces (autonomous regions and municipalities directly under the central government) are categorized into “talent highlands” and “talent depressions” for heterogeneity analysis, and the regression results are shown in columns (3) and (4) of Table 13. The knowledge spillover effect of the “talent highlands” on the local innovation ecosystem resilience is significantly positive, and the innovation ecosystem resilience of neighboring provinces is also significantly positive. Talent mobility in the IC innovation ecosystem can promote the strong synergistic coupling effect of knowledge elements, provide inter-regional innovation subjects with cooperation paths across time and space, reduce the dependence on geographic location, cut resource consumption, improve innovation efficiency, and avoid the decline in resilience brought about by the homogenization of system knowledge.

4.4. Results and Discussion

Based on the above findings, the hypotheses tested in this study are summarized as shown in

Table 14. Hypothesis validation summary.

Hypothesis	Hypothesis Validation
Hypothesis H1: Intra-regional knowledge spillovers will positively affect IC innovation ecosystem resilience.	support
Hypothesis H2: There are spatial spillovers in the impact of inter-regional knowledge spillovers on IC innovation ecosystems, and inter-regional knowledge spillovers not only promote the resilience of IC innovation ecosystems in the region, but also act on the resilience of IC innovation ecosystems in neighboring provinces.	support
Hypothesis H3a: The contractual governance mechanism can enhance the impact of intra-regional knowledge spillovers on the resilience of IC innovation ecosystems.	support
Hypothesis H3b: The contractual governance mechanism can enhance the impact of inter-regional knowledge spillover on the resilience of IC innovation ecosystems.	support
Hypothesis H4a: The relationship governance mechanism has the potential to augment the influence of intra-regional knowledge spillover on the resilience of IC innovation ecosystems.	no support
Hypothesis H4b: The relationship governance mechanism can enhance the impact of inter-regional knowledge spillover on the resilience of IC innovation ecosystems.	partial support

.

Determining how knowledge spillover effects affect the resilience of the IC innovation ecosystem is one of the keys to achieving technological breakthroughs. This study first confirms that the polarization of resilience in the IC innovation ecosystem is gradually increasing and has strong spatial clustering. Based on this, it is found that both intra-regional and inter-regional knowledge spillovers can significantly promote the resilience of the IC innovation ecosystem. The contract governance mechanism can effectively enhance the impact of knowledge spillover on the resilience of the innovation ecosystem in the region, and the relational governance mechanism has a positive effect on the resilience of the innovation ecosystem in neighboring regions. Based on these findings, this paper has the following theoretical contributions:

Firstly, the adaptive cycle theory is applied to the construction of the resilience index of the IC innovation ecosystem, and the research level of this theory is expanded. Existing studies mainly focus on the resilience of cities and regions, and there are few studies on the resilience of a specific industrial innovation ecosystem.

Secondly, the impact of knowledge spillover on the resilience of the IC innovation ecosystem is no longer limited to the role of a single innovation agent. Most of the previous studies focused on the role of enterprise knowledge spillover, and there was a lack of relevant research on knowledge spillover between innovation entities. After reviewing the literature on knowledge spillover, this paper finds that there are abundant studies on knowledge spillover’s impact on innovation performance, but few studies on how it affects the resilience of an innovation ecosystem. In addition, previous studies have also found that human capital distance has an impact on knowledge spillover, but the relevant empirical studies are relatively few. Therefore, this paper discusses the relationship between inter-regional knowledge spillover and the resilience of the IC innovation ecosystem from the perspective of human capital distance, and it further enriches the research on the relevant spatial effects of the relationship between knowledge spillover and resilience.

Finally, this study also expands the research on knowledge governance mechanisms. Previous studies suggested that formal contract governance mechanisms or informal trust governance mechanisms could be used to coordinate knowledge activities in the innovation ecosystem, but there was a lack of discussion on the suitability of knowledge governance mechanisms. This study finds that these two governance mechanisms have different governance effects under different knowledge spillover models, and their internal influence mechanisms are also different. For example, the relational governance mechanism has an inhibitory effect on the resilience governance of the IC innovation ecosystem in the knowledge spillover process within the region but has a positive effect on the resilience governance of the IC innovation ecosystem in neighboring regions. To a certain extent, this enriches the research on knowledge governance mechanisms.

5. Conclusions and Limitations

5.1. Conclusions

Based on the panel data of 30 provinces (autonomous regions and municipalities directly under the central government) from 2011 to 2021, this paper, on the basis of constructing an IC innovation ecosystem resilience indicator system, utilizes the panel bi-directional fixed-effect model, spatial Durbin model, moderating effects model, and multi-dimensional empirical research on the impact of regional knowledge spillovers on the resilience of the innovation ecosystem and explores the regional heterogeneity and knowledge governance mechanisms of the intrinsic influence. The results show the following:

(1) The polarization of innovation ecosystem resilience is gradually increasing, with strong spatial aggregation. Shanghai, Jiangsu, Zhejiang, and other regions have “high–high” aggregation characteristics; Liaoning, Qinghai, Inner Mongolia, and other regions have “low–low” aggregation characteristics, with neighboring provinces showing significant positive spatial correlation, and there is a “Matthew effect”. The “Matthew effect” is a phenomenon in which the good becomes better and the bad becomes worse. Hunan, Chongqing, Jiangxi, and other regions have “low–low” aggregation characteristics. Hunan, Chongqing, Jiangxi, and other regions have “low–high” discrete characteristics, and Beijing, Guangdong, and other provinces have “high–low” discrete characteristics. Hunan, Chongqing, Jiangxi, and other regions have “low–high” discrete characteristics, and Beijing, Guangdong, and other provinces have “high–low” discrete characteristics, with significant negative spatial correlation with neighboring provinces, and there is a “siphon effect”.

(2) This paper innovatively starts from the spatial perspective and clarifies that intra- and inter-regional knowledge spillovers have a significant contribution to the resilience of innovation ecosystems in integrated circuits. It is found that the contractual governance mechanism can effectively enhance the impact of knowledge spillovers on the resilience of local innovation ecosystems, and the relational governance mechanism has a positive impact on the resilience of innovation ecosystems in neighboring regions, which opens the “black box” of governance.

(3) Heterogeneity results show that knowledge spillovers within the Pan-PRD region have a positive effect on innovation ecosystem resilience; knowledge spillovers between regions with low innovation capacity have a positive effect on the resilience of local innovation ecosystems but a negative effect on the resilience of innovation ecosystems in neighboring provinces. Knowledge spillovers between “talent highlands” have an uplifting effect on local innovation ecosystem resilience and an uplifting effect on innovation ecosystem resilience in neighboring provinces.

5.2. Policy Implications

Based on the above findings, the following targeted policy recommendations are made:

First, open up the spillover channels of innovation knowledge and improve digital infrastructure. The results of this paper show that intra- and inter-regional knowledge spillover promotes the resilience of the innovation ecosystem. However, there are still “blockages” in the knowledge spillover channels of China’s integrated circuits. The government should take effective measures to promote the transformation efficiency of digitalization and Informa ionization infrastructure. In turn, it should realize the stable operation of the innovation ecosystem.

Second, realize the effective allocation of innovation resources and improve the knowledge governance mechanism. In the process of knowledge governance, the knowledge governance mechanism plays an important role in promoting the dual effect of the contract governance mechanism and relationship governance mechanism. To avoid the emergence of the “closed door” innovation model, to strengthen the innovation cooperation between multiple subjects, accelerate the flow of knowledge between innovative subjects to realize the collaborative governance of the innovation ecosystem resilience. In practical application, the governance mechanism should be flexibly adjusted according to the characteristics of industrial knowledge, subject relations, and policy orientation so as to realize the effective allocation of regional innovation resources according to the local conditions and resist the external risks of the system.

Thirdly, we should optimize the system of innovative talents and consolidate the core strength of technology. In order to better solve the “stuck-neck” problem, we should continue to cultivate “high-precision, top-notch and short-listed” talents, and encourage the flow of scientific research talents and knowledge to “locked” regions. Talent introduction policies should be enacted according to local conditions to accurately replenish technologically scarce talents, so as to form an attractive and internationally competitive talent system, accelerate the construction of talent centers and innovation highlands, and lead the technological “unlocking” of lagging regions, so as to promote the resilience of the innovation ecosystem.

5.3. Limitations

There are two limitations in this study that need to be further improved in subsequent studies: First, IC innovation ecosystem resilience is also affected by external factors such as government and market, so the variable selection in this study is not comprehensive enough to cover all the influencing factors. Second, in terms of variable measurement, due to the limitation of data availability, the data measurements of some variables are not precise enough, and subsequent studies should further enhance the data completeness in order to improve the analysis of the enhancement mechanism of IC innovative ecosystem resilience.