A Study of Methods to Identify Industry-University-Research Institution Cooperation Partners based on Innovation Chain Theory

Yoon and Song (2014) summarized the methods for partner selection and classified them into three categories: mathematical programming approaches (Soles-vik & Encheva 2010), rating/linear weighting approaches (Wang & Chen 2007), and artificial intelligence techniques (Fischer, Jähn, & Teich, 2004). They utilized patent information to investigate innovation activities by applying morphological analyses (MA) and generative topology maps (GTM) to the process used in identifying technology configurations and visualizing the collected patent information. Park et al. (2015) explored potential R&D collaboration partners through patent analysis based on bibliographic coupling and latent semantic analyses. The potential R&D collaboration partners were visualized in the form of a patent-assignee-level map based on the technological similarity between patents using network analyses. Wang et al. (2017) identified R&D partners through subject-action-object (SAO) semantic analyses in a problem and solution pattern and combination term clumping. Xu et al. (2015, 2016) used multiple indicators to synthetically assess the new IURC situation and identify cooperation partners for collaboration. By extending a multimodal data analysis and by considering the technical similarity of IURC institutions, the competitive position, types, and core-periphery structures of these institutions in the cooperation network, potential IURC partners were identified. Most existing studies on IURC employ a single variable to identify cooperation partners. Such studies still lack comprehensiveness in cooperative partner identification and consider the technology similarity only from a technical point of view. Even if one begins with technology similarity to identify some potential IURC possibilities, it is inappropriate to identify IURC partners using only technical similarity. For example, the IURC cluster—represented by the industrial cluster in China—has not achieved collaboration innovation. The reasons for this are that the degree of industry specialization within the cluster chain is still rather low and fierce competition caused by homogenization make it difficult to achieve collaboration effectively (Huang, 2014). Thus, the similarity of the technology, but not its convergence, is an important factor in cluster innovation and proper IURC. Therefore, it is necessary for institutions to identify proper IURC partners with more complementary capabilities.

There are significant differences between IURC and pure scientific research cooperation. When a sustainable IURC partnership is formed, it needs to consider several market factors beyond research ability from industrial and economic perspectives. When the rules and characteristics of cooperation are known, the IURC can be established. Therefore, only when the factors of collaboration have been recognized, can potential IURC partners be identified effectively. Usually, both cooperation and competition exist at the same time where similar technologies exist. However, for IURC, it is only when cooperation is heavily emphasized that knowledge spillover effects occur.

In an IURC that works well, the expectation from universities and research institutions is to bring their scientific innovations and inventions to market, thereby promoting economic development, and gaining further financial support for R&D (Xu et al., 2017). On the other hand, industry enterprises expect to gain more advanced technology to upgrade their products. Furthermore, some disruptive innovation can possibly generate more competitive advantages and higher profits (Zhang, 2012). The promotion of and demand for scientific innovation in universities, research institutes, and industries are referred to as “knowledge potential.” The essence of knowledge potential is the “push” and “pull” effect in knowledge dissemination. This knowledge potential is then the innovation impetus in the formation of the IURC (Yue et al., 2015).

Cai (2002) defined the innovation chain theory as the commercialized process of science and technology knowledge through the transformation of technical innovation (Cai et al., 2001). The thinking around the innovation chain optimizes the innovation process through systematic analysis. The driving force behind the innovation chain largely comes from demand; it is a functionally linked network structure model that is demand-driven. The functional links are node links that can meet some specific functions (Bamfield, 2006; Larson & Brahmakulam, 2002; Liang, 2007; Timmers, 1999). In this study, the innovation chain is defined as a chain process with five orderly innovation goals: the transformation of scientific problems to scientific theory, the transformation of scientific theory to practical application, the transformation of actual application to product manufacturing, the transformation of product manufacturing to commodities, and the transformation of goods to industrialized production.

Given that the essence of the innovation chain is the dissemination of innovation factors consistent with the IURC process that leads to the knowledge spillover effect, innovation chain theory can effectively describe the characteristics of knowledge spillover and exchange phenomena in IURC. Accordingly, this paper presents a new method to identify IURC partners based on innovation chain theory, attempting to survey the distribution of innovation technology in the innovation chain of different institutions in search of additional complementary capabilities among them.

According to the different goals of innovative activities in the innovation chain, we divide the chain into five phases: basic research, application research, transfer and transformation, commercialization, and industrialization (Figure 1). The different innovation phases have specific innovation activities but are connected to each other and work synergistically. Each innovation phase is divided into several functional nodes including enterprises, scientific research institutions (research institutes and universities), government-involved innovative activities, and other innovation elements. The innovation is driven by the interaction between these functional nodes.

Figure 1

The reason technology transfer does not go smoothly is that there are obstacles to the diffusion of innovation elements in the innovation chain. IURC is one way to effectively promote the diffusion of innovation elements in the innovation chain.

In this study, IURC identification research is established based on the entire innovation process, including a series of orderly innovation targets. The detailed analysis is as follows:

In the basic research phase, scientific problems are formulated into scientific theories. The evaluation index in this phase is a comparative analysis of quantity and quality based on scientific research papers.

The process of IURC partner identification based on innovation chain theory is divided into two modules: (1) core network analysis of the existing cooperative institutions, and (2) institutional competitiveness analysis based on the innovation chain.

The core network analysis of the existing institutions

The cooperative network is a complex one. In the cooperative network, although the number of nodes is very large, its core nodes are very small. From the structural viewpoint of the complex network, these core nodes are connected closely, while other noncore nodes are connected to these core nodes only by a few edges.

The core nodes of a complex network can be found through the k-core network. For a network, if any node has at least k neighbors that are also in the same network, then the network is a k-core network (Baxter et al., 2012; Newman, 2003). K-core nodes are the core nodes in a complex network and k-core analysis is often used to mine core members. In this study, the k-core network is used to analyze the core cooperative institutions in the basic and applied research phases.

Table 1

Evaluation indexes of institutional competitiveness based on innovation chain.

Basic research analysis

To access the relationship between the amount of basic research, the influences on the basic research results, and the basic research areas of the core institutions, the research papers in the database of academic resources are identified, and the quantity, influences on these papers, and the relationships between institutions analyzed.

Both the citation analysis and the citation impact analysis index (CNCI) (Incites, 2015) are used to analyze research influence. CNCI is a relative evaluation indicator in the Thomson Reuters InCites™ database. CNCI is an unbiased indicator that excludes the subject area, publication year, and document type. The influence of collected papers of different sizes and different subjects can be compared using the CNCI. When the value of the CNCI is equal to one, the cited performance of this group of papers is equivalent to the global average. Thus, if the index value is greater than one, the cited performance of this group of papers is higher than the global average. Correspondingly, if the index value is less than one, the cited performance is lower than the global average. If CNCI is equal to N (N≥2), the cited papers of this group are N times higher than the global average.

Listed below are the four hypotheses that identify the IURC partners based on the innovation chain theory.

Hypothesis 1

There is a greater potential for collaboration among agencies with different attributes of spatial distribution in the innovation chain. Thus, the competition within universities-enterprises or research institutes-enterprises is less than enterprises-enterprises.

This hypothesis stems from the belief that the heterogeneity capacity of different institutions in the innovation chain can contribute to cooperation potential. The primary choice in the selection of the cooperation partner is whether the partner to be selected can provide some advantage or expertise, and whether it can bring in complementary resources to achieve a goal that cannot be achieved by the primary establishment relying on its own resources. Kogut and Chang (1991) pointed out that complementary resources are a key driver in cross-organizational cooperation based on resources. Laursen and Salter (2006) argued that the rapid development of technology makes the business model of the enterprise more open, and technology is more dependent on external sources, especially in the high-tech industry.

The gene engineered vaccine (GEV), also known as the “genetically engineered vaccine,” is the vaccine produced by recombinant DNA cloning; it is the expression of the protective antigen gene, or produced by the expressed antigen products, or the recombinant itself (Tang et al., 2006). As the vaccine industry is related to strategic security, it has received extensive attention in many countries. Further, the GEV is a typical high technology with an intact knowledge innovation chain. Therefore, this study chose the major research institutions in the field of GEVs to identify potential cooperation partners.

This empirical analysis is based on the SCI papers indexed by the Web of Knowledge and the Derwent innovations index database and retrieved June 2015. We built a detailed search strategy (not shown here owing to its complexity), and obtained 3962 papers and 999 Chinese patent applications. The Thomson Data Analyzer (TDA) (Thomson Reuters, 2015) was used for text data analyses. We then identified major patent holders, technical topics, and types and names of institutions as the basis for the data analysis, ultimately, obtaining a result with 180 institutions.

Figure 2 is the k-core network based on co-authored papers where the number of co-occurrences is no less than three. Figure 2 shows three cooperative networks involving three domestic research institutions including. The first one is Shanghai Jiaotong University, Fudan University, the Chinese Academy of Sciences. The second one is Chinese Academy of Military Medical Sciences and Chinese Academy of Agricultural Sciences. The third one is Academy of Medical Sciences (Peking Union Medical College) and the Russian Academy of Medical Sciences. The elicited results show that these seven institutions are major cooperative institutions in basic research on GEVs. There are close cooperative relationships within the three cooperative networks. At the same time, cooperative relations between research institutions are often frequently compared with these universities.

Figure 2

A lot of research analyzes the factors blocking the path between R&D results and commercial production and the impact of those factors on patent-to-product transformation. In the area of GEVs, there are many impact factors affecting the decision-making at commercial enterprises in terms of how to acquire and produce the vaccine. Nine patented technology-effect factors considered for the genetic vaccine in its patent-to-product transformation are included in this study: good immunogenicity, production readiness, high safety, good profitability, low cost of production, high-efficiency expression, high purity, active immunization, and good repeatability.

We construct a two mode network of the institutions that have more than 10 GEV patent applications and nine technology-effect, as shown in Figure 3. The larger the institution nodes in the graph, the more technology-effect the institution’s patents have; the larger the technology-effect nodes, the more institutions there are involved in this technology-effect; the thicker connection line between the nodes indicates that the organization has advantages in this technology-effect.

Figure 3

Here, we use community discovery to analyze different technology-effect communities. The community in the network reflects the local characteristics of the network and the interrelation among its internal nodes. Different communities have different structural characteristics. The nodes in the same community are closely connected with each other and the connections among the communities are relatively sparse. The structure or attributes of the same community have some similarities (Newman, 2004). The patents applied by the patent agencies in the community have the same technology-effect and these institutions show a closer connected status. The fast unfolding community discovery algorithm proposed by Blondel et al. (2008), whch is a non-overlapping community discovery algorithm based on a block modulus. The two mode network is divided into five communities: #1, #2, #3, #4, and #5. The patents applied by patent agencies in community #1 have two technologyeffect factors: production-readiness and high-efficiency expression, in #2, active immunization, in #3, high purity, good repeatability, and good profitability, in #4, low cost of production, and in #5, both good immunogenicity and high safety.

To achieve the commercialization of the GEV, all the nine technology-effect factors need to be well developed. There are interdependencies among the agencies in five communities, that is, similarities in technology-effect. Therefore, these patent application agencies have the basis for cooperation with each other.

In addition, the figure shows that the patents applied for by the patent agencies in communities #2 and #4 have fewer patents outside their communities, indicating that the patents of these communities are different from those of other community organizations. In addition, the UNIV PLA THIRD MILITARY MEDICAL organization, except for the unconnected patented effect in #2, has links to the technology-effect factor of patents in other communities, indicating that the patents applied by UNIV PLA THIRD MILITARY MEDICAL are functionally equivalent to those in other communities. The overlap is so strong that this agency may have a greater potential for cooperation with patent agencies in other communities.

Competitiveness analysis in basic research

The top 19 institutions with a high number of SCI papers in GEVs in China (Table 2) include five scientific research institutes (accounting for 26.3%). The total number of papers was 72 (44.4%), and the average citation number was 10.36. The results show that the number of scientific research institutes in the field of domestic GEVs is small but the number of fundamental research studies is large and has a high impact, which constitutes the backbone of basic research in the field of GEVs. In addition to the five listed institutions, the other four institutions, Huazhong Agricultural University, China Agricultural University, Jilin Agricultural University, and China Agricultural University, account for a total of 16 papers (9.9%) with an average citation number of 12.65. Although the number of papers from these agricultural universities is very small, despite the high research influence, these four agricultural universities play an important role in the field of basic research on gene vaccines in China. Moreover, most of the remaining colleges and universities are national key institutions with strong capacities in scientific research. Therefore, these colleges and universities constitute the main research power in the research area of the GEV.

Table 2

Publication of SCI papers of domestic GEV by the amount of top19 institutions.

Among the top 19 institutions, there are four institutions whose CNCI value exceeded 1, including two research institutes and two agricultural universities. The citations of papers from these four institutions exceeded the global level. Among them, the CNCI value for the The National Center for Nanoscience and Technology of China was 6.23. The citation of the paper was 6.23 times the global average. There were also 15 institutions (78.9%) below 1, indicating that most of these papers were not influential.

Analyzing the features of the domestic GEV industry in China, we can see that this industry presents the pattern of a comprehensive innovation chain. Thus, we can infer that there will be a higher possibility of cooperation among enterprises, universities, and research institutes in the IURC partnership.

In the basic research phase in the innovation chain in the domestic GEV industry of China (Table 7), the selection of the category types for the IURC for institutions can be divided into research institutions, agricultural universities, and comprehensive research universities. In the applied research phase, enterprises begin to join the innovation chain, and most of them are the main participants in the transfer and transformation, commercialization, and industrialization phases. The comprehensive innovation chain needs to be populated with all types of industries, universities, and research innovations. At present, there are no enterprises in basic research. This indicates that enterprises have a higher possibility to join basic research networks. There are five elite enterprises in the applied field of GEVs: Aventis Pasteur Company, Beijing Kaiyin Biological Technology Company, Merial Company, Pulaike Biological Engineering Company, and Yuanlun Biotechnology Company. However, the companies that do not have CFDA qualification, and the other 58 companies that are qualified, own fewer patents. Therefore, considering knowledge transformation and the abilities that are complementary to the process of transformation, the five nonqualified companies and 58 qualified companies are more likely to be candidates for cooperation.

Table 7

Category types of IURC in GEV innovation chain in domestic China.

From the spatial distribution of the innovation chain process, the institutions that have greater potential to cooperate are those that have high relevance and complementary technology capacities, are in close geographical position, and are different types of institutions. GEV R&D and production institutions in Guangdong, Beijing, and Shanghai comprise the complete innovation chain. Each institution in the innovation chain in the three locations has high complementary capacities, which shows that IURC institutions in Beijing, Shanghai, and Guangdong have greater cooperative potential. Compared to the other two areas, there are more industry-university-research institutions in Beijing; these types of institutions vary, the possibility of choosing collaborative partners is, thus, higher, and the spatial distances among these institutions are smaller. Therefore, the GEV research institutions in Beijing have increased cooperative potential.

In the Guangdong province, there are more cooperation possibilities among the South China Agricultural University, the Aventis Pasteur Company, the Zhongshan University, and the Shenzhen Kangtai Biological Products Company. Specifically, the quality of both the basic and applied research of the South China Agricultural University is high, while the corresponding research ability of Shenzhen Kangtai Biological Products Company as a commercial and industrial institution is weaker. Thus, there are many more possibilities for cooperation between the South China Agricultural University and the Shenzhen Kangtai Biological Products Company in view of the convenience of the geographical location and the complementary capacities.