How Does the Innovation Openness of China’s Sci-Tech Innovation Enterprises Support Innovation Quality: The Mediation Role of Structural Embeddedness

Abstract

Sci-Tech innovation enterprises (STIEs) in China are responsible for improving the quality of national innovation (IQ). Because of their inherent innovation openness (IO), STIEs are facing constantly changing external cooperation channels and gradually optimizing their openness. However, existing research considers external cooperation relationships as established network environments, which may not apply to STIEs’ network relationships that are still under construction. Hence, this study investigates the impact of STIEs’ IO on IQ by exploring the role of structure embeddedness (SE). Empirical findings from 362 sample enterprises suggest that openness breadth and depth have an inverted U-shaped relationship with IQ, while openness balance impacts IQ positively. Moreover, network centrality plays a partial mediation role between openness depth and IQ, and network reach fully mediates the relationship between openness balance and IQ. The results indicate the influence of three openness factors on IQ and further expand the research on the SE of STIEs in the dynamic development stage. These can support STIEs to improve IQ through the adjustment of network centrality and reach by changing their openness depth and balance.

1. Introduction

According to the Global Innovation Index 2022, China’s innovation performance rose from 17th in 2018 to 11th in 2022 in the international ranking. Meanwhile, China ranks first in the world in total R&D personnel and the number of invention patent applications and authorizations. However, over the past few decades, the Chinese government and enterprises have been focusing more on innovation quantity instead of innovation quality [1]. The phenomenon of “quantity over quality” is still prominent in Chinese enterprises’ pursuit of innovation. Compared with its world-leading position in terms of innovative achievements, China has a large gap with major developed countries in innovation quality. Facing fierce global competition in science and technology, a high-quality development strategy was put forward by the Chinese government, and raising innovation quality (IQ) is an urgent task [2]. In this case, the important position of Sci-Tech innovation enterprises (STIEs) in the national innovation-driven development model is increasingly prominent. To promote STIEs to maintain sustainable and healthy growth, the Science and Technology Innovation Board (STAR) was officially established on the Shanghai Stock Exchange in January 2019, to provide a financing platform for innovation. As of 2021, there are more than 300 STIEs listed as published by STAR (http://www.sse.com.cn/, accessed on 12 August 2024). These STIEs come from six major industries including new-generation information technology, biomedicine, high-end equipment, new materials, energy conservation, and environmental protection.

STIEs refer to enterprises that have independent brands and intellectual property rights and are capable of continuous research and development (R&D), technological achievement transformation, and innovation (A definition in “Guidelines for Evaluation of Science and Innovation Attributes” issued by China Securities Regulatory Commission in 2020). Compared with other types of enterprises, they face a faster technology iteration in their industry and rely more on the continuous implementation of innovation activities to get rid of the original technological trajectory, achieve technology leadership, and make new product breakthroughs [3]. Thus, innovation is STIEs’ foundation and the only way to obtain sustainable competitive advantages [3,4]. However, because of the constraints from internal resources and capabilities, it would be more difficult to complete innovation solely by themselves. The innovation in STIEs therefore strongly depends on the support and coordination of external resources and intellectual assets [5,6,7]. Innovation openness (IO) is the basis and essence of innovation in STIEs. By keeping an openness to innovation, STIEs can access and exploit a wider range of external ideas, knowledge, and resources [8,9] and further cope with limited resources, more uncertainties, and barriers to innovation [10].

Although research on the relationship between IO and IQ is limited, previous studies have expressed two opposing opinions about the effect of open innovation on innovation performance. With openness to partners, an enterprise can access and integrate external knowledge to support collaborative innovation, and reduce the costs of R&D and the risks associated with innovation development activities, which further promotes its absorption capacity [11]. All of these are positively related to innovation performance [10,12]. However, some studies also pointed out that openness to innovation does not necessarily lead to positive performance [13,14]. Open innovation may increase the costs of identifying external knowledge, negotiating supply terms, and developing absorptive capacity, which reduces the enterprise’s competitive advantage [15]. Therefore, how to manage the balance between the potential benefits and the emerging risks of innovation openness needs further exploration. Specifically, in China, it is imperative to examine the various effects of IO on IQ in the context of STIEs.

Furthermore, cooperation networks have become an important way to obtain scarce and valuable resources [16,17]. Previous studies have proved that the location and density of an enterprise’s network, namely structure embeddedness (SE), impact external activities and performance significantly [13,18]. In particular, the patent R&D cooperation network between enterprises has become an important means to break through the lack of innovation resources [19]. STIEs can obtain heterogeneous resources for patent R&D innovation through this embedded cooperation network. However, STIEs are mostly composed of small enterprises with short development histories and are still in the early stages of cooperative network construction. The active innovation activities, rapid iteration, and flexible decision-making mechanism in STIEs enable their SEs to be an emerging network structure and result in dynamic change, which is different from the ones of larger organizations that have built relatively stable and extensive network relations. In other words, SE is affected by enterprises’ attitudes towards external resources and the tendency of openness and is the result of STIEs’ innovation cooperation behavior instead of an established objective environment. This way, the findings from research into larger organizations do not apply to STIEs [12].

In this case, this study focuses on STIEs. It considers structural embeddedness as a mediation variable in the relationship between innovation openness and innovation quality, which integrates the particularity of STIE development. In theory, this study reveals and confirms the inverted U-shaped relationship between innovation openness and innovation quality in STIEs. More importantly, the various mediation roles of the structural embeddedness of STIEs are further explored. This shows that the open strategy of STIEs is the foundation of the development of their cooperative relationship network and indirectly affects the quality of innovation through the development and construction of the network. In practice, our findings show that the improvement in the openness breadth and depth of STIEs within a certain range can effectively improve the quality of innovation, while excessive openness will produce the opposite effect. Thus, to ensure and improve innovation quality, STIEs should focus on the openness balance. While expanding cooperation partners, it needs to strengthen cooperation relations to improve the innovation quality. In this process, on the one hand, STIEs should pay attention to shaping their network centrality to better promote innovation cooperation and quality. On the other hand, STIEs can accumulate high accessibility partnerships and then give full play to the positive role of openness balance on the quality of innovation.

The rest of this paper is organized as follows. Section 2 reviews the related literature on innovation openness, innovation quality, and structural embeddedness; The hypotheses on the relationships among variables are analyzed and proposed in Section 3; Section 4 introduces the research methodology and data; Section 5 implements statistical and regression analyses to test the hypotheses; and Section 6 discusses the results of the empirical examination; Finally, the theoretical and managerial implications, some limitations, and directions for future research are presented in Section 7.

2. Literature Review

2.1. Innovation Openness

Innovation openness reflects the enterprise’s tendency to cooperate with external partners [20]. An enterprise is regarded as open if it can search for and use external resources by collaborating with consumers, suppliers, universities, or other institutions [11]. Openness helps to identify and leverage external complementary assets and then enriches the pool of knowledge [21]. Furthermore, Laursen and Salter [20] described innovation openness from breadth and depth, which represent two different usage patterns of outside resources according to external search strategies. Specifically, openness breadth is conceptualized as the number of external knowledge sources used in the innovation and indicates the diversity of external partners [11]. Openness depth means the extent to which an enterprise cooperates with specific partners [22] and emphasizes the intensity of external collaboration [23]. In other words, openness breadth reflects the amount of different external sources that can be reached and used by the enterprise, while openness depth refers to the stability of external sources [24] and embodies the frequency of contact with a particular entity or group [25].

Furthermore, innovation activities involve high uncertainty and risk and require a large and sustained investment of resources [26]. Both kinds of openness represent two different source strategies. However, the diversity and frequency of external sources involved in the process of innovation are not the opposite. Inversely, a single strategy cannot react to the rapid changes and fierce competition; thus, multiple strategies should be applied to acquire adequate resources [27]. An enterprise must decide how to use and allocate its limited internal resources and balance the breadth and depth of openness to a particular entity or group. In this case, concerning previous research [20,27,28], we introduce the third dimension, openness balance, to describe the tension between openness breadth and depth in the process of innovation, and we extend innovation openness into three dimensions, namely breadth, depth, and balance.

2.2. Innovation Quality

Innovation quality (IQ) describes the degree of novelty of an innovation [29] and is the integration of product, service, and process [2,30]. This integration reflects the embodiment of the innovation process quality and the degree to which innovation results meet the market demand [31]. IQ presents an enterprise’s actual and potential innovation ability [32,33,34], which is different from innovation performance, whichrepresents the quantity in the process of innovation activities. A change in IQ guarantees a change in performance with a certain range [34,35], while a change in performance may not lead to a change in quality. Specifically, with high IQ, the influence of innovation output is more significant [36], and the enterprise can constantly provide customers with new and outstanding value [37].

In this study, STIEs engage in scientific and technological innovation activities by acquiring and integrating external knowledge and resources. Their IQ is a comprehensive concept that illustrates all the innovation outputs in this innovation process. Therefore, based on previous research [27,31,38,39], IQ is defined as a comprehensive manifestation of STIEs’ innovation ability and can be measured by innovation outputs. Furthermore, invention patents are the core output of innovation activities in STIEs and judge the level and quality of innovation [27,36]. Thus, this study takes invention patents as a basis to measure IQ.

2.3. Structural Embeddedness

Structural embeddedness is another concept that is closely related to open innovation [25]. It is a sub-dimension of network embeddedness and describes the structure of an entity’s relationships with other partners [40,41,42]. Structural embeddedness reflects the number and degree of cooperation or transaction with the subject. If the level of one subject’s structural embeddedness is high, it indicates that the subject has a closer cooperative relationship with external partners, and cooperation becomes an important support for its development. On the contrary, the operation of the subject is relatively independent, especially in the innovation environment; it is more inclined to independently carry out innovative research and development of new technologies, indirectly reflecting that it may have a higher innovation ability and resource level.

According to Lyu et al. [18], structural embeddedness can be manifested as two parts: network centrality and network reach. The former implies the degree to which an enterprise is close to the center of the cooperation network, while the latter reflects the sum of the distances from an enterprise to other external organizations [43]. Both depict an entity’s position in a pattern of relationships with other members [44] and represent the extent to which an entity is entrenched in a network of relationships [41].

Furthermore, structural embeddedness reflects an enterprise’s opportunity to the source of social capital and enables it to search for and acquire various innovation resources through the relationship network [13]. This way, structural embeddedness defines the number and categories of embedded resources and may further affect the enterprise’s innovation and competitive advantages [18]. Some studies have demonstrated the positive effect of suitable structural embeddedness on innovation outcomes [45,46]. A higher degree of structural embeddedness implies more connections with the external environment and the potential to dominate the flow of important resources in the networks [47].

3. Hypotheses

3.1. Innovation Openness and Innovation Quality

Innovation openness (IO) is the core of open innovation and reflects an enterprise’s external knowledge search strategy for innovation [25]. Although there is currently insufficient empirical evidence for the relationship between innovation openness (IO) and innovation quality (IQ), the positive and negative effects of open innovation have been revealed [11,15,48,49,50,51]. Thus, based on previous studies, this study explores the impact of innovation openness breadth, depth, and balance on IQ in STIEs.

3.1.1. Innovation Openness Breadth and Innovation Quality

The breadth of innovation openness refers to the scope and scale of innovation activities. A broader openness breadth means more extensive channels and promotes the flow and complementarity of knowledge resources inside and outside the organization. This further benefits innovation outcomes [52] and yields more opportunities to effectuate radical innovation [53]. Specifically, the increasing openness breadth can improve the quality of innovation in three ways.

First, from the standpoint of knowledge resources, a higher degree of openness breadth can offer STIEs more diversified channels to search and acquire various types of external knowledge and technologies [11,54] including market sources, general sources, research sources, and regional sources [6]. Then, STIEs can further absorb and integrate different external sources from customers, suppliers, competitors, universities, government institutions, or research institutions into their innovation processes [55,56]. These complementarities and synergies among knowledge sources enable STIEs to generate better new ideas for innovation [57] and improve innovation quality.

Second, a broader openness can reduce potential path-dependence risk [15]. STIEs with more external partners can obtain knowledge from multiple alternative sources and do not need to worry about the inevitable risk due to the uniqueness of the external source [55]. Meanwhile, it also improves the possibility of finding useful knowledge and solving problems more quickly [58], and then developing high-quality innovation [59].

Third, a broader openness can enable STIEs to perceive external changes more accurately and in a timely manner [15,60]. Based on this, STIEs can have a better understanding of the development trends in relevant markets, products, and technologies [61], and even develop reliable forecasts and predictions of future technology trajectories [53]. Then, they can adjust their innovation plans and acquire more advanced technologies and innovation resources in certain promising technology areas [15], which help shorten R&D cycles, reduce R&D risks and costs [50,51,62], and benefit high-quality innovation.

However, openness breadth does not always have a positive effect. First, cooperation with a large number of different partners usually comes with high costs of management and coordination [59]. Specifically, excessive breadth may bring excessive external knowledge and cause organizational knowledge redundancy that cannot be learned and utilized [63] and increases the demand for resource investment [64]. In this case, the breadth of innovation openness cannot be expanded indefinitely because of STIEs’ limited finances and resources. Moreover, excessive breadth may result in the mismatching and wasting of resources, which affects innovation quality negatively [38,55]. Second, STIEs are often in high-tech industries with rapid technological upgrading, and their innovation activities are always at high risk. An excessive openness may require the enterprise to disperse and squeeze the limited innovation resources. This in turn burdens them with higher risks and brings a negative performance to innovation [1].

Overall, with a broader breadth of innovation openness, STIEs can produce more new and valuable ideas, reduce path-dependence risks, and perceive and predict future trends in changing environments. But at the same time, excessive openness breadth may lead to an adverse effect. Therefore, innovation openness breadth needs to be controlled at a moderate and balanced level, and we propose the following hypothesis:

H1a: 

There is an inverted U-shaped relationship between innovation openness breadth and innovation quality in STIEs.

3.1.2. Innovation Openness Depth and Innovation Quality

A high depth of innovation openness usually means an STIE carries out frequent horizontal collaboration with partners and reuses existing knowledge elements to carry out innovation. In this process, frequent cooperation helps enterprises gain a deeper understanding of partners and their knowledge. First, such collaboration not only improves the comprehension of external knowledge and avoids misunderstandings [65] but also improves the efficiency of cross-organizational knowledge integration [66]. This can further create a good condition for transforming knowledge into innovative outcomes, achieving new ideas, and even improving innovation ability [8,67].

Second, close interactions can mutually integrate STIEs’ long-term partners into the innovation processes and create a shared picture of partners’ knowledge and technologies within a field [68]. This is more conducive to the transmission of tacit knowledge, which is an extremely important resource for innovation [69].

Third, openness depth can benefit the trust and stability of external relationships [70], which can help establish long-term cooperative relationships, and enable partners to share their information and knowledge more effectively [71], thus resulting in high-quality innovation [72].

However, excessive openness depth reduces the diversity of resources available outside the enterprise and leads to over-dependence on certain partners and their knowledge resources, namely path dependence. This further restricts STIEs’ innovation. Moreover, a single and fixed source of knowledge will reduce the sensitivity of enterprises to changes in the external environment. In this case, STIEs may be unable to track the frontier of technology in time and fall into the capability trap [73], which hinders the improvement of innovation quality [1]. Only moderate and effective innovation openness depth can guarantee excellent innovation quality. Therefore, we propose the following hypothesis:

H1b: 

There is an inverted U-shaped relationship between innovation openness depth and innovation quality in STIEs.

3.1.3. Innovation Openness Balance and Innovation Quality

Due to the limitation of organizational resources and culture, innovation openness breadth and depth are two opposing innovation cooperation strategies and show competitive demands on innovation resources [1]. STIEs need to decide how to allocate resources to achieve a balanced openness, namely a relatively balanced state between openness breadth and depth and produce maximum benefit.

A more balanced openness can bring richer external knowledge resources to enterprises while ensuring effective knowledge flow and integration [1]. It also means a dynamic selection and adjustment of partners and relations to keep such a balance. Through this continuous optimization of cooperation strategy, STIEs can better allocate and use their innovation resources and enrich and improve knowledge structure, which enables them to acquire a higher level of innovation ability and achieve high-quality innovation.

On the contrary, under the constraints of resources, over-investment in the breadth of openness will lead to a lack of depth, which may result in an inefficient partnership. Too much attention to depth drives the enterprise to concentrate on a single external resource, which makes it difficult to support innovation activities. Furthermore, excessive openness breadth or depth will bring negative influences such as the “path dependence” or “capability trap” mentioned above, instead of their positive effects. This way, an unbalanced emphasis on the breadth or depth of openness is only a partial optimization of innovation cooperation.

Overall, enterprises need to focus on the duality of innovation openness to avoid extreme situations. And with a better balance of innovation openness, STIEs can make more effective use of limited resources to achieve high-quality innovation. Therefore, we propose the following hypothesis:

H1c: 

There is a positive relationship between innovation openness balance and innovation quality in STIEs.

3.2. Innovation Openness and Structural Embeddedness

Some previous studies have confirmed the impact of structural embeddedness on innovation openness [25,74,75,76]. In these studies, a cooperation network is regarded as an established and unchanged external environment, which forms the premise of openness. However, for STIEs, their cooperation network is still in the process of formation. Openness is the core of STIEs’ innovation activities. It is the norm to adjust STIEs’ innovation openness in the process of development to acquire and integrate diverse external innovation resources. This way, innovation openness is manifested as a kind of active choosing and deciding behavior and reflects the strategies of external cooperation and resource search [25]. This further constructs the basis for the formation of external cooperation relations and impacts STIEs’ network positions. Therefore, we believe the structural embeddedness of STIEs is the result of their open behavior strategy and being influenced by innovation openness.

3.2.1. Innovation Openness Breadth and Structural Embeddedness

On the one hand, by expanding the openness breadth, the enterprise can build a self-centered cooperative network to better acquire and use different knowledge resources to complete innovation activities [77,78], which further enables it to be the intersection center of various knowledge resources and improve its network centrality. On the other hand, if STIEs can carry out cooperation with various types of partners, they must have a strong learning and absorption capacity in the cooperation process [77]. These enable STIEs to assimilate and exploit the wide range of knowledge acquired externally [59], while effectively reducing innovation costs and the risks of unsuccessful technology acquisitions, which are caused by cooperating with different types of partners [53]. In this way, STIEs can reach external partners easily and show a better network reach. Comparatively, STIEs that only cooperate with a single type of partner can hardly reach a more diversified network and their network reach will be low. Therefore, we propose the following hypotheses:

H2a: 

There is a positive relationship between innovation openness breadth and network centrality in STIEs.

H2b: 

There is a positive relationship between innovation openness breadth and network reach in STIEs.

3.2.2. Innovation Openness Depth and Structural Embeddedness

STIEs are often faced with greater risks and a more competitive environment [79]; thus, they need to constantly pursue in-depth innovation cooperation with all kinds of partners to better integrate knowledge and improve innovation efficiency [80,81]. With an increase in openness depth, they can create deeper relational involvement, increase trust and commitment with external partners [70], and then form a stable and lasting cooperative relationship. This deep tie with partners in turn enhances the enterprise’s central position in this cooperation network. Moreover, it also enhances critical knowledge transfer [82] and expands innovation resources through existing networks [83,84], which further brings the enterprise closer to its partners and benefits its network reach. Therefore, we propose the following hypotheses:

H2c: 

There is a positive relationship between innovation openness depth and network centrality in STIEs.

H2d: 

There is a positive relationship between innovation openness depth and network reach in STIEs.

3.2.3. Innovation Openness Balance and Structural Embeddedness

In a balanced openness environment, STIEs can carry out extensive and in-depth innovation cooperation with partners to ensure the diversity of external knowledge resources [47] and realize in-depth knowledge absorption and integration [77]. On the one hand, STIEs will try their best to occupy the center of the cooperation network to further utilize the limited resources efficiently and then construct a self-centered network gradually. On the contrary, if the balance is broken, STIEs may be more inclined to cooperate with certain partners or contact several different partners. The former situation may weaken or even terminate the relationship with various partners, and the latter will reduce the enterprises’ control or influence on the network. Both are detrimental to the formation and maintenance of network centrality. On the other hand, with a high level of openness balance, a relatively stable cooperative relationship will be formed gradually, which can improve cooperation efficiency steadily. Then, the distance between the enterprise and all partners can be narrowed, and even an optimal cooperative network can be constructed. Conversely, unbalanced openness may break this benign development, resulting in low network reach. Therefore, we propose the following hypotheses:

H2e: 

There is a positive relationship between innovation openness balance and network centrality in STIEs.

H2f: 

There is a positive relationship between innovation openness balance and network reach in STIEs.

3.3. Structural Embeddedness and Innovation Quality

Previous studies found that structural embeddedness can bring the spillover benefits of knowledge resources for enterprises and facilitate information acquisition and knowledge transfer [38,76,85,86,87]. With high embeddedness, the strong ties between enterprises encourage trust and cooperation, reduce opportunistic behavior, increase the efficiency of cooperation, decrease transaction costs [88], and enhance innovation diffusion in the network [18,89]. All of these can benefit the exchange of valuable information and tacit knowledge that is hard to obtain in other types of networks and help detect and explore market opportunities [90].

Specifically, network centrality offers the advantage of searching for novel combinations and exploring novel technologies [18,91,92]. Getting closer to the center of the cooperation network can facilitate better access to various innovation resources, which provides the basis for innovation and helps improve innovation quality. Moreover, the enterprise occupying the center position of the network will have more structural holes [76] and acquire external knowledge in a faster and more efficient way through the patterns of interaction [93]. This also promotes the transfer and assimilation of knowledge and leads to a high level of innovation quality [18].

Similarly, STIEs with substantial network reach are more active in technological cooperation and have shorter technological distances. This can help STIEs avoid the blindness of innovation search [94], increase the speed of knowledge flow, and reduce maintenance costs [95]. Compared with a dispersed network, a dense one can provide much more insight and new technologies, improve knowledge efficiently, and result in high-quality innovation [38,96]. Thus, with an increase in network reach, STIEs can achieve higher-quality innovation.

Overall, STIEs can benefit from network centrality and reach [97,98]. We propose the following hypotheses:

H3a: 

There is a positive relationship between network centrality and innovation quality in STIEs.

H3b: 

There is a positive relationship between network reach and innovation quality in STIEs.

3.4. Mediation of Structural Embeddedness

Although most current studies emphasize the moderating effect of network structure on the relationship between open innovation and innovation performance [25,76], this study argues that network structure is not an established objective environment but the result of innovation cooperation behavior. STIEs can improve their innovation quality and competitive advantages by implementing appropriate innovation strategies and optimizing the cooperation network. Innovation openness emerges as an enterprise’s innovation strategy and affects the formation of a cooperation network structure. Further, the network position held by the enterprise will affect the quantity, quality, and efficiency of available knowledge resources for the enterprise [76], which forms the basis of innovation and has an important influence on innovation quality.

Specifically, an appropriate breadth of openness is conducive to improving network centrality and reach. STIEs with high network centrality and reach can expand the source channels of external knowledge, new ideas, and new opportunities and facilitate the efficient and high-quality acquisition of knowledge resources. Second, the improvement of openness depth also can construct benign cooperative relations and promote strong structural embeddedness. This position helps the enterprise capture the technologies and knowledge behind market development trends in a timely manner [99] and optimizes the transfer efficiency and quality of knowledge resources. Third, with the improvement of openness balance, STIEs can allocate limited resources more rationally and effectively to form an efficient network relationship. And all of these can facilitate high-quality innovation. Therefore, this study proposes the following hypotheses.

H4a–c: 

Network centrality plays a mediation role in the relationship between innovation openness breadth/depth/balance and innovation quality.

H5a–c: 

Network reach plays a mediation role in the relationship between innovation openness breadth/depth/balance and innovation quality (Figure 1).

4. Methodology

4.1. Variable Measurement

4.1.1. Dependent Variable

Innovation quality is the dependent variable in this study and can be considered as the influence of innovation output [33,36]. For STIEs, the patent is an important output of innovation activity and indicates the level of innovation. Thus, it can be used to assess innovation quality directly [36,38,100,101,102]. Furthermore, some studies have used various patent indicators to measure innovation quality, such as the number of patents, the proportion of invention patents, the number of patents in force, and the number of patent citations [1,103,104]. Among them, patent forward citation is the most direct and objective one to reflect the influence of technology innovation. More citations means a higher level of innovation quality. Relatively, other indicators are more focused on the indirect reflection of innovation quality such as the number of innovation outputs or economic benefits. Thus, we choose patent forward citation and use the average citation number of all patents held by an STIE to measure its innovation quality.

4.1.2. Independent Variables

Openness breadth, depth, and balance are the independent variables and can be measured by their propensity to cooperate with other entities in the environment [105]. According to the study of Laursen and Salter [20], we use the number of categories of partners that cooperate with STIEs to reflect openness breadth. The patentees indicate the cooperators in the R&D process and usually include five categories, namely company, university, scientific research institution, government agency, and individual. If an STIE cooperates with a certain category of partners, the value is 1; otherwise, it is 0. The openness breadth is the cumulative value and is defined as a nonnegative integer ranging from 1 to 5.

Second, openness depth is the intensity of an enterprise’s cooperation with external innovation sources. This study uses the multiplication of cooperation frequency and cooperation distance between different partners’ categories to measure. The cooperation distance reflects the degree of difficulty of cooperation between five partners’ categories and is measured by the percentage of cooperation frequency between all sample enterprises and a certain category of partners in the total frequencies. Last, openness balance shows the state of an STIE cooperating with multiple categories of partners and is measured by the multiplication of the percentages of cooperation frequency between an enterprise and its different categories of partners.

4.1.3. Mediation Variables

According to the study of Lyu et al. [15], we use betweenness centrality and distance-weighted reach to measure the network centrality and network reach, respectively. The formulas are as follows:

$${\mathrm{N}\mathrm{B}\mathrm{C}}_{\mathrm{i}}={\sum }_{\mathrm{j}}^{\mathrm{n}}{\sum }_{\mathrm{k}}^{\mathrm{n}}{\displaystyle \frac{{\mathrm{g}}_{\mathrm{j}\mathrm{k}}\left(\mathrm{i}\right)}{{\mathrm{g}}_{\mathrm{j}\mathrm{k}}}}$$

(1)

where ${\mathrm{N}\mathrm{B}\mathrm{C}}_{\mathrm{i}}$ refers to the network betweenness centrality of node i. j and k refer to the node in the cooperation network of patentees. ${\mathrm{g}}_{\mathrm{j}\mathrm{k}}\left(\mathrm{i}\right)$ refers to the total number of shortest paths that link node j and k, which contains node i, and ${\mathrm{g}}_{\mathrm{j}\mathrm{k}}$ refers to the total number of shortest paths from node j to node k.

$${\mathrm{N}\mathrm{R}}_{\mathrm{i}}={\sum }_{\mathrm{j}=1}^{\mathrm{n}}{\displaystyle \frac{1}{{\mathrm{d}}_{\mathrm{i}\mathrm{j}}}}$$

(2)

where ${\mathrm{N}\mathrm{R}}_{\mathrm{i}}$ refers to the distance-weighted reach of patentee i, ${\mathrm{d}}_{\mathrm{i}\mathrm{j}}$ is the minimum distance from node i to j, and n is the number of reachable nodes in the enterprise’s patent cooperation network. The shorter the path between two nodes, the larger the network reach. We define the path length as infinite when two enterprises do not connect.

4.1.4. Control Variables

Enterprise ownership, named enterprise type, is the first control variable. In China, state-owned enterprises are important participants in economic and innovative activities and have more resource advantages and market shares compared with private enterprises. Thus, enterprise type may have a certain impact on its business performance. And we adopt dummy variables on the indicators of ownership, where 1 represents state-owned enterprises and 0 means others.

Enterprise size is the second control variable. Usually, large-scale enterprises have more advantages in human resources, capital, and risk resistance [20], which indicates an impact on innovation. For STIEs, researchers always make up a large proportion of human resources, whose intellectual assets have a vital influence on the quality of innovation. Therefore, we use the number of employees to measure the enterprise size.

Third, enterprises established earlier can accumulate more internal and external resources. Thus, the age of an STIE (enterprise age) is considered another control variable and is measured by the period from its year of foundation to the current year. Similarly, the listing can also provide more external resources for enterprises, expand the scope of cooperation, and be related to innovation activities. The earlier the listing time, the more conducive the enterprises are to the accumulation of innovation resources and the better the quality of innovation output. Thus, we also take the listed age of the enterprise (enterprise age listed) as a control variable and use the period from the listing year to the current year for measurement.

4.2. Sample and Data Collection

First, we chose STIEs listed in the Science and Technology Innovation Board (STAR) of the Shanghai Stock Exchange (http://www.sse.com.cn/, accessed on 12 August 2024) from 2000 to 2021 as our samples. The STAR was officially launched in China in 2019. Compared with ordinary enterprises, STIEs’ ability for R&D is more prominent. As the forefront enterprises in China’s economic transformation, these listed companies’ innovation quality is worth more attention.

Then, according to the list, we obtained 362 STIEs, including 101 state-owned enterprises and 261 non-state-owned enterprises. Most of them (294) are from the manufacturing industry, with the rest from information transmission, scientific research and technical service, and ecological protection and environmental governance. Additionally, most enterprises (275) have fewer than 2000 employees, and only a few (42) have more than 3000 employees.

Third, we collect all enterprises’ invention patent data from the Incopat database (https://www.incopat.com/advancedSearch/init, accessed on 12 August 2024), which officially joined Clarivate Analytics in 2020 and contains more than 100 million patent data from 112 countries/organizations/regions in the world. This can meet our research needs while ensuring data quality. Actually, we took 362 STIEs’ English and Chinese names as the search keywords and conducted the patent applicant searches on the 1 July 2022. And 110,105 invention patents were obtained and used in our empirical study. These invention patents can reflect the innovation level of an enterprise accurately and objectively [1,27], and the patent documents provide details about patent applications, which also indicate the cooperation relationships and can be used to construct the open innovation network [106].

5. Result

5.1. Descriptive Statistics Analysis

Table 1. Descriptive statistics and correlation.

Variables	Mean	SD	Enterprise Type	Enterprise Size	Enterprise Age	Enterprise Listed Age	OB	OD	OBL	NBC	NR
Enterprise type	0.721	0.44913	—
Enterprise size	1464.7	2577.107	−0.05	—
Enterprise age	193.7459	62.5916	−0.085	0.009	—
Enterprise listed age	13.3425	8.17596	0.191 **	0.058	−0.065	—
OB	1.2514	1.20033	−0.157 **	0.247 **	0.212 **	−0.027	—
OD	0.3764	0.38279	−0.035	0.137 **	0.039	−0.023	0.374 **	—
OBL	0.4437	0.46008	−0.011	−0.018	0.009	0.000	0.061	0.551 **	—
NBC	41.4834	244.54268	−0.099	0.422 **	0.047	0.033	0.221 **	0.173 **	−0.051	—
NR	0.3894	0.36918	−0.013	−0.002	0.069	0.062	0.318 **	0.486**	0.802 **	−0.093	—

Notes: * p < 0.1, ** p < 0.05, *** p < 0.01, “—” on the diagonal means the variance inflation factor is below the threshold of 10.

provides the descriptive statistics and bivariate correlation matrix. The correlation coefficient of any two variables is less than 0.7, and the variance inflation factor (VIF) is below the threshold of 10, indicating that multicollinearity is not a substantive concern; the variables were well measured and subsequent regression analysis was possible.

5.2. Regression Analysis

To test the direct and mediation effect analysis, we construct the following regression models.

$$\mathrm{I}\mathrm{Q}/\mathrm{N}\mathrm{B}\mathrm{C}/\mathrm{N}\mathrm{R}=\mathsf{\alpha }+{\mathsf{\beta }}_1\mathrm{O}\mathrm{B}+{\mathsf{\beta }}_2{\mathrm{O}\mathrm{B}}^2+\sum \mathsf{\delta }·\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}+\mathsf{\epsilon }$$

(3)

$$\mathrm{I}\mathrm{Q}/\mathrm{N}\mathrm{B}\mathrm{C}/\mathrm{N}\mathrm{R}=\mathsf{\alpha }+{\mathsf{\beta }}_1\mathrm{O}\mathrm{D}+{\mathsf{\beta }}_2{\mathrm{O}\mathrm{D}}^2+\sum \mathsf{\delta }·\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}+\mathsf{\epsilon }$$

(4)

$$\mathrm{I}\mathrm{Q}/\mathrm{N}\mathrm{B}\mathrm{C}/\mathrm{N}\mathrm{R}=\mathsf{\alpha }+{\mathsf{\beta }}_1\mathrm{O}\mathrm{B}\mathrm{L}+\sum \mathsf{\delta }·\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}+\mathsf{\epsilon }$$

(5)

$$\mathrm{I}\mathrm{Q}=\mathsf{\alpha }+{\mathsf{\beta }}_1\mathrm{O}\mathrm{B}/\mathrm{O}\mathrm{D}/\mathrm{O}\mathrm{B}\mathrm{L}+{\mathsf{\beta }}_2{\mathrm{O}\mathrm{B}}^2/{\mathrm{O}\mathrm{D}}^2+{\mathsf{\beta }}_3\mathrm{N}\mathrm{B}\mathrm{C}/\mathrm{N}\mathrm{R}+\sum \mathsf{\delta }·\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}+\mathsf{\epsilon }$$

(6)

${\beta }_n$ is the coefficient and Controls represents the control variable. $\sum \mathsf{\delta }·\mathrm{C}\mathrm{o}\mathrm{n}\mathrm{t}\mathrm{r}\mathrm{o}\mathrm{l}\mathrm{s}$ refers to the set of the control variables, and $\mathsf{\epsilon }$ is the random disturbance term.

5.2.1. Direct Effect Analysis

Table 2. Regression analysis results of the direct effect of IO and SE on IQ.

Variables	IQ
	Model 1	Model 2	Model 3	Model 4	Model 5	Model 6
Control variables
Enterprise type	−0.108 **	−0.014	−0.091 *	−0.106 **	−0.099 *	−0.102 **
Enterprise size	0.083	−0.047	0.014	0.087 *	0.034	0.085 *
Enterprise age	0.139 ***	0.005	0.047	0.137 ***	0.134 **	0.118 **
Enterprise listed age	−0.032	−0.03	−0.024	−0.033	−0.035	−0.053
Independent variable/independent variable squared
OB		1.197 ***
OD			2.452 ***
OBL				0.195 ***
OB × OB		−0.597 ***
OD × OD			−2.145 ***
Mediation variables
NBC					0.118 **
NR						0.289 ***
R2	0.044	0.462	0.314	0.082	0.055	0.127
Adjusted R2	0.033	0.453	0.303	0.069	0.042	0.114
F	4.120 ***	137.703 ***	69.926 ***	14.671 ***	4.276 **	33.688 ***

Notes: * p < 0.1, ** p < 0.05, *** p < 0.01.

summarizes the regression analysis of the direct effect of IO and SE on IQ. Model 1 only includes control variables and shows that enterprise type (β = −0.108, p < 0.05) and enterprise age (β = 0.139, p < 0.001) have significant impacts on innovation quality. Based on this, we further comment on our Hypotheses (H1, H2, H3) by constructing Models 2–6. Model 2 includes OB and OB², while Model 3 includes OD and OD². Based on these two models, the relationship between OB/OD and IQ is explored. As the results show, both OB (β = 1.197, p < 0.001) and OD (β = 2.452, p < 0.001) have positive impacts on innovation quality, while their squared terms show a negative influence (β = −0.597, p < 0.001; β = −2.145, p < 0.001). These indicate that there is an inverted U-shaped relationship between openness breadth/depth and innovation quality. As the openness of innovation increases, so does innovation quality. However, when innovation openness breadth or depth reaches a certain level, innovation quality will shift from improvement to decline. Hypotheses 1a and 1b are supported. Meanwhile, Model 4 includes OBL, and the results (β = 0.195, p < 0.001) show it has a positive impact on IQ, which further indirectly proves the existence of the above inverted U-shaped relationship. Hypothesis 1c is supported. Figure 2a,b show these two inverted U-shaped relationships and reveal the turning point of IQ with the increase in OB/OD.

Moreover, compared with Model 1, Model 5 and Model 6 include NBC and NR, respectively. The results show that these two variables (β = 0.195, p < 0.001; β = 0.118, p < 0.05; β = 0.289, p < 0.001) also have significant impacts on IQ. Thus, as network centrality and network reach increase, innovation quality will be improved. Hypotheses 3a and 3b are supported.

Further,

Table 3. Regression analysis results of the direct effect of IO on S.

Variables	NBC				NR
	Model 7	Model 8	Model 9	Model 10	Model 11	Model 12	Model 13	Model 14
Control variables
Enterprise type	−0.08	−0.066	−0.078	−0.08	−0.021	0.023	−0.011	−0.011
Enterprise size	0.417 ***	0.390 ***	0.401 ***	0.416 ***	−0.008	−0.091 *	−0.076	0.007
Enterprise age	0.038	0.016	0.034	0.039	0.072	0.003	0.055	0.065 **
Enterprise listed age	0.026	0.027	0.029	0.026	0.071	0.072	0.083	0.068 **
Independent variable/independent variable squared
OB		0.112 **				0.346 ***
OD			0.115 **				0.496 ***
OBL				−0.045				0.801 ***
OB × OB
OD × OD
R2	0.187	0.198	0.2	0.189	0.01	0.115	0.25	0.651
Adjusted R2	0.177	0.186	0.188	0.177	−0.001	0.102	0.239	0.646
F	20.476 ***	4.866 **	5.744 **	0.878	0.867	42.225 ***	114.038 ***	653.924 ***

Notes: * p < 0.1, ** p < 0.05, *** p < 0.01.

shows the analysis of the direct effect of IO on SE. Model 7 and Model 11 only test the effect of the control variables on NBC and NR, and the results prove that only enterprise size has a positive impact on NBC and none of the control variables can affect NR. We also test the impact of OB, OD, and OBL on NBC and NR using Models 8–10 and Models 12–14. Compared with Model 7, Models 8 and 11 indicate that OB (β = 0.112, p < 0.05; β = 0.346, p < 0.001) has a positive effect on NBC and NR. As OB increases, NBC/NR will be improved. Model 9 and Model 12 indicate that OD (β = 0.115, p < 0.05; β = 0.496, p < 0.001) has a positive effect on NBC and NR. As OD increases, NBC/NR will be improved. However, Models 10 and 14 reveal that OBL only affects NR (β = 0.801, p < 0.001) positively and cannot affect NBC (β = −0.045, p > 0.1). Therefore, Hypotheses 2a, 2b, 2c, 2d, and 2f are supported, while Hypothesis 2e is not.

5.2.2. Mediation Effect Analysis

To test the mediation effect, we carry out regression analysis by constructing Models 15–19. Models 15 and 16 include NBC and NR, respectively. Compared with Model 2, the results indicate neither NBC nor NR can mediate between OB and IQ, and Hypotheses 4a and 4b are not supported. Thus, network centrality and network reach cannot mediate the relationship between OB and IQ.

Models 17 and 18 include NBC and NR, respectively, and compared with Model 3, the results indicate that only NBC (β = 0.105, p < 0.05) can mediate the relationship between OD and IQ. Hypothesis 5a is supported, while hypothesis 5b is not. Thus, network centrality can partly mediate the relationship between openness depth and IQ. Moreover, compared with Model 14, Model 19 tests and proves that NBC cannot have as a mediating role in the relationship between OBL and IQ, while NR acts (β = 0.373, p < 0.001) as a perfect mediator (

Table 4. Regression analysis results of mediation effect of SE.

Variables	IQ
	Model 15	Model 16	Model 17	Model 18	Model 19
Control variables
Enterprise type	−0.01	−0.011	−0.083 *	−0.091 **	−0.102 **
Enterprise size	−0.07	−0.049	−0.028	0.019	0.084 *
Enterprise age	0.004	0.005	0.042	0.047	0.113 **
Enterprise listed age	−0.031	−0.025	−0.027	−0.028	−0.058
Independent variable/independent variable squared
OB	1.199 ***	1.306 ***
OD			2.476 ***	2.369 ***
OBL					−0.104
OB × OB	−0.606 ***	−0.692 ***
OD × OD			−2.181 ***	−2.086 ***
Mediation variables/mediation variables squared
NBC	0.058		0.105 **
NR		−0.063		0.050	0.373 ***
R2	0.464	0.464	0.323	0.316	0.131
Adjusted R2	0.454	0.453	0.31	0.303	0.116
F	92.585 ***	92.423 ***	48.614 ***	46.909 ***	17.656 ***

Notes: * p < 0.1, ** p < 0.05, *** p < 0.01.

). Thus, Hypothesis 5c is supported but Hypothesis 4c is not. OBL can only affect IQ indirectly through NR.

Above all, of all the direct effects, only Hypothesis 2e is not supported and OBL cannot affect NR. Of all mediation effects, Hypotheses 4a, 4b, 4c, and 5b are not supported. In other words, openness breadth and depth have a positive impact on IQ to a certain extent, and openness balance always brings a better IQ. Furthermore, the impact of OB on IQ is direct, while OD can partly affect IQ through the mediation of NBC, and all the influence of OBL on IQ is indirect with the help of NR.

6. Discussions

6.1. The Impact of Innovation Openness on Innovation Quality

Our study provides significant evidence that both openness breadth and depth show an inverted U-shaped relationship with innovation quality, which is in line with the research of Kao et al. [13] and Radicic and Alkaraan [14]. When STIEs’ capacity and innovation resources reach a certain level, the initial positive effect of OB and OD turns into a negative one.

On the one hand, although openness breadth brings external cooperation partners and knowledge resources, the enterprises cannot exploit the overwhelming partners and resources in a timely manner and economically. The excessive dispersion of capabilities and resources is not conducive to innovative cooperation between enterprises and external partners [107] and reduces the in-depth mining and exploration of external knowledge. In return, the knowledge and ideas turn into obstacles to deep innovation.

On the other hand, the strengthening of openness depth is accompanied by continuous investment. Although it is helpful to stabilize the cooperation relationship, it also increases the switching cost [108]. In particular, when external partners can no longer provide new resources or knowledge, it is difficult for STIEs to change the existing relationship structure and thus they will be faced with the problem of insufficient new external resources.

In terms of openness balance, our findings show that openness balance has a positive effect on innovation quality, which extends previous research [1,11,63,73]. A proper proportion of openness breadth and depth can lead to the balance of open innovation. In this case, STIEs can make full use of existing internal capabilities and resources to acquire, absorb, and use diversified external resources as much as possible. Meanwhile, they can also realize timely and dynamic adjustments to the opening relationship as the internal and external environment changes. All of these will benefit STIEs’ innovation quality.

6.2. The Impact of Innovation Openness on Structural Embeddedness

As our results prove, openness breadth and depth have a significant impact on network centrality and network reach, while openness balance only affects network reach positively. In terms of openness breadth and depth, our results are in line with the research of Chen et al. [109] and Lu et al. [11]. The increase in the categories of cooperation partners and frequency enriches the types of external knowledge resources and the diversity of the R&D cooperation network. This further reflects the strong knowledge absorption ability and networkability of the enterprise, which helps improve the enterprise’s reputation and influence [17,109]. As for openness balance, our results consolidate and extend the research of Li et al. [49]. A balanced openness tendency may allow an enterprise’s resources and capabilities to be equally distributed among multiple partners. For STIEs, this may limit their influence on cooperative relationships and become obstacles to the promotion of network position and centrality. Thus, openness balance cannot affect NR significantly. However, openness balance allows enterprises to effectively collaborate with external partners as much as possible, thereby improving network reach.

The above findings expand the research on the relationship between open innovation and network structure. Previous studies mainly explored the significant impact of structural embeddedness on innovation openness or revealed the moderating role of open innovation or innovation openness [11,25,110]. In these studies, the enterprises’ cooperative network is regarded as a static environmental basis before the degree of their openness, and few took innovation openness as the antecedent factor of structural embeddedness. As large enterprises have invested a lot of resources to realize innovation and have formed a relatively stable network structure, the cooperative network has become their basis for further openness to innovation. Their findings are more reasonable and applicable to large enterprises. However, STIEs are still in the initial stage of development on a relatively small scale. As pointed out by Chaochotechuang et al. [12], lessons drawn from studies on large organizations may not apply to small ones. STIEs usually lack internal resources and are endowed with openness to carry out innovation activities. They have to search for cooperation with partners relying on their personal and professional ties. Their cooperation network is gradually constructed under the opening actions and then becomes the premise of structure embeddedness. Therefore, for STIEs, innovation openness is an important antecedent variable of structural embeddedness, and the latter is the phased outcome pursued in the process of its innovation and development and the foundation of the relationship accumulated under the guidance of the open innovation strategy.

6.3. The Mediation Role of Structural Embeddedness

Our study once again proves the significant impact of structural embeddedness on innovation quality, which is in line with previous research [5,76,90]. Network centrality improves STIEs’ position in the cooperation network, while network reach expands the range of cooperation, and they both enable STIEs to ensure the stability and diversity of external resource supply and thus support innovation activities effectively.

More importantly, our study further reveals the mediating role of network centrality and reach in the relationship between IO and IQ. Network centrality plays a partial mediating role between OD and IQ, while network reach plays a complete mediating role between OB and IQ. In terms of network centrality, STIEs can deepen the cooperative relationship with partners to improve network position and coordination ability. This network structure enables them to obtain diversified external knowledge resources and capture the technologies and knowledge [99], which further effectively improves innovation quality. In terms of network reach, with the improvement in openness balance, limited resources can be allocated more rationally and effectively for contact with as many partners as possible. This in turn forms a proper innovation cooperation relationship and benefits the network reach. Meanwhile, it also enables STIEs to expand and ensure the diversity of external knowledge resources. Based on this, they can acquire and apply external knowledge resources at a lower cost with less risk and thereby improve the innovation quality.

However, both network centrality and reach do not mediate the relationship between OB and IQ, and network reach does not mediate the relationship between openness depth or balance and innovation quality, either. The probable reason for this may be the limitation of absorption ability and network capacity. Although it may improve the network centrality and reach of STIEs through cooperating with multiple categories of partners or deepening cooperation relationships, the costs of cooperation, coordination, and communication will also rise. The enterprises are required to obtain higher levels of ability to absorb and apply diversified knowledge resources, which requires further accumulation rather than immediate formation. In this circumstance, STIEs cannot improve innovation quality simply by changing network position or reach. Instead, internal resources and capabilities become the key to making better use of structural embeddedness. Therefore, we propose that only the openness tendency that emphasizes relationship quality and equilibrium can affect IQ by changing structural embeddedness. This also supports the research of Tobiassen and Pettersen [111] that small enterprises need to invest more resources to enhance their legitimate and formal relationships with partners.

7. Conclusions

With the global innovation landscape and industrial pattern being reshaped, high-quality innovation has become the foundation for China to further achieve in-depth national development. As the most dynamic and contributive innovation subjects, Sci-Tech innovation enterprises (STIEs) are faced with the problem of how to improve the quality of innovation through open innovation activities. Considering the characteristics of STIEs, our study explores the impact of innovation openness (IO) on innovation quality (IQ) by revealing the mediation role of structural embeddedness (NE). The empirical results show that openness breadth and depth have an inverted U-shaped relationship with IQ, while openness balance impacts IQ positively. Network centrality plays a partial mediation role between openness depth (OD) and IQ, while network reach fully mediates the relationship between openness balance (OBL) and IQ. These findings reconfirm the existing relationship between innovation openness and innovation quality and further reveal the importance of structural embeddedness and the different mediation roles of network centrality and reach.

In theory, this study confirms that open innovation cooperation can enrich the external knowledge resources of STIEs, reduce the uncertainty of innovation, transfer innovation risk, and improve IQ effectively. However, excessive openness increases acquisition costs and the coordination of external resources, ultimately harming IQ. Furthermore, structural embeddedness does not mediate the effects of all openness on IQ. Instead, only openness depth and balance can affect innovation quality through network centrality and reach, respectively, which reflects the particularity of the innovation network of STIEs and expands the research on the role of NE.

In practice, our findings indicate that, for STIEs, open innovation is not only the essence but also the basis of high-quality innovation. However, excessive opening may bring adverse effects and harm innovation quality. Thus, STIEs need to strengthen the balance of innovation openness and give full play to the positive effect of balanced opening on IQ through appropriate openness breadth and depth. In other words, STIEs should not blindly pursue the number of external partners but need to pay attention to the depth of cooperation between each other to ensure high-quality cooperation within a certain range of cooperation, achieving a balanced innovation openness. Second, embedding the cooperative relationship network is an important way to enrich innovation resources, and STIEs can constantly adjust and improve their network position through open innovation activities to guarantee innovation quality. In particular, it should be noted that openness depth can effectively improve network position, while openness balance can promote the improvement of network reach. Therefore, STIEs should pay more attention to both, rather than just blindly expanding the scope of cooperation partners. When expanding the number of partners, STIEs should pay attention to the shaping of their network centrality and obtain relatively high network centrality by occupying advantages such as emerging technologies and management flexibility to better promote innovation cooperation and quality. On the other hand, in balancing the breadth and depth of cooperation, the accessibility of partners should be an important consideration to give full play to the positive role of openness balance on the quality of innovation.

Last, for STIEs in different countries, the model found may also have some plausibility. For example, in developed countries, STIEs may face more intense competition, but the choice of partners is also increased, and it is even easier to attract the favor of partners from developing countries. STIEs can establish their cooperation networks more quickly and have more open innovation strategies. To improve the quality of innovation more effectively, STIEs need to balance the breadth and depth of cooperation and optimize network centrality and accessibility by their market position.

Despite the above-mentioned contributions, some challenges still deserve attention in future research. First, the object of this study is China’s STIEs, which are different from other types of enterprises. STIEs exist in most developing and developed countries and are growing rapidly. Therefore, companies with similar characteristics are likely to exhibit similar attributes. Conversely, the findings of this study should also apply to a certain extent. But, whether the relationships among variables revealed in this study apply to enterprises in other countries or other types of enterprises remains to be further discussed. Second, as the research results show, STIEs in the process of development have certain particularities, and it is necessary to pay attention to the innovation strategy and cooperation process in the development process. The cross-sectional analysis conducted in this study cannot reveal this dynamic process. Thus, to further reveal the dynamic interaction between innovation openness and network embeddedness and its impact on innovation quality, future research should start from the perspective of processes and conduct dynamic research. Last, this study focuses on the mediating effect of network centrality and reach, while there are still a variety of indicators to measure structural embeddedness. For example, Chen et al. [109] took structural holes as an indicator to discuss their impact on innovation. Future studies can explore the role of other indicators of structural embeddedness in the relationship between innovation openness and quality.