Does the Multiple-Participant Innovation Improve Regional Innovation Efficiency? A Study of China’s Regional Innovation Systems

Abstract

The triple helix (university-industry-government) is a dynamic cooperation model in the area of regional innovation. With the application of this model, the authors aim to address the mechanism of how multiple participants affect innovation efficiency against the research background of China Fuzzy-Set Qualitative Comparative Analysis (fs-QCA) is applied to conduct a multiple-case analysis. With the evaluation of 31 provinces in China, which have various innovation performances and degrees of regional cooperation, the authors find strong evidence to support the positive effects of multiple participants on regional radical innovation. Moreover, this article also finds two typical configurations of factors that contribute to high innovation efficiency. Based on the results, the authors propose policy implications to improve China’s regional innovation in different areas. Finally, this paper concludes with the discussion of future research orientations, which could focus on the differences in the triple helix among different industries in China.

1. Introduction

With the increasing importance of knowledge in the development of the global economy, innovation has expanded its meaning from Schumpeter’s definition, the creation and application of a new product or technique, to the creation of approaches through which people treat innovation [1]. Both policy and strategy makers seek to break the inherent mechanism of classical regional development and improve the efficiency of innovation. However, a region is a complex system, which includes tangible and intangible influence factors in the process of innovation. In this way, the traditional static scope of innovation cannot explain the dynamic connections between different factors [2]. The triple helix model is a knowledge-based innovation tool with the interdependent principal parts of government, university, and industry. The mechanism of the model is to optimize the regional innovation environment through the promotion of regional cooperation and fostering a feedback loop to act on the innovation efforts of different participants. Meanwhile, the triple helix focuses on bringing dynamic innovation ability to regional innovation systems and accelerating the transition of knowledge, labor, and technology capital, as well as management experience. This regional innovation model has attracted the attention of many scholars; therefore, the effect of regional cooperation on innovation efficiency has become a new orientation in regional development research [3,4,5].

In the economics research field, it is imperfect to evaluate regional innovation by input-output and resource endowment approaches without the consideration of spatial issues [6]. Thus, the triple helix provides a way of solving problems related to economic geography, such as what factors induce the imbalance among different areas in their evolutionary development routines and which cooperative model would be suitable for regions in different phases of the development life cycle. China is a typical research target for the analysis of regional innovation because it has gone through a gradual economic reform since the 1980s and each province has different degrees of openness, technical devotion, and regional industrial policy. In recent years, the gap between different regions’ economic growth, which is reflected by their innovation ability, has become larger; this unique phenomenon raises the research value of exploring the situation and the causes of regional innovation systems’ heterogeneity.

To explore the mechanism of the triple helix in China’s specific background, considering the roles of government, universities, and organizations in regional synergistic innovation, the authors apply Fuzzy-Set Qualitative Comparative Analysis (fs-QCA) to evaluate China’s provincial innovation using an outcome-casual approach: this research approach combines qualitative and quantitative methods. Our research adds to a rather small body of literature related to regional multiple-participant innovation in China. However, some empirical studies focus on cooperative innovation in emerging countries [7,8]. We propose variables for measuring casual factors of regional innovation. Additionally, our research focusses on radical innovation rather than exploited innovation.

The structure of this paper is organized as follows: the literature review section summarizes the meaning of the triple helix and puts forward hypotheses based on the theoretical background and China’s specific policy and technical environment; the third part is concerned with the methodology, which contains case selection, data sources, and variable explanation; the results and discussion give an in-depth analysis. The paper ends with the policy implications and conclusions.

2. Theoretical Review

2.1. The Triple Helix Model

The triple helix was used to describe the water support system for the Hanging Garden in Babylon; Etzkowitz and Leydesdorff applied this creation to demonstrate three participants’ coordination in regional innovation because they support each other dynamically just like a water screw. As the founder of the theory of the triple helix innovation model, Etzkowitz combined the multiple participants’ devotion with the spatial issue and attributed regional coordination as the source of the high innovation efficiency. As the pioneer, Etzkowitz applied the case study approach to analyze the regional innovation environments of Silicon Valley, Cambridge, and Linkoping, which are affected by the triple helix model. He proposed that these high-tech clusters, which are agglomerated in a region, would benefit from frequent communication between government, corporations, and universities. Besides, in the field of analytical study in different countries and regions, researchers face different political, technical, and cultural environments, so dynamic situation analysis is indispensable for the analysis of the triple helix model [9].

2.2. Regional Innovation and Triple Helix Studies’

Innovation is an important output of industry, which is treated as a critical index of economic development performance by evolutionary economics researchers. Differing from new classical economics, evolutionary economics concentrates on the process of creation rather than transaction. Moreover, evolutionary economics gives researchers a dynamic scope to evaluate economic activity from an uneven development approach, which requires more attention on the different participants’ involvement and the connections between factors in a system [10]. Along with geographical specific innovation research receiving increasing attention from researchers, Cooke’s regional innovation system theory also emphasizes the importance of inner systematic cooperation between different participants [11]; this theory breaks the “marginal economics” restriction of established resources and calls for critical thinking about how to regenerate new resources. Instead of tangible factors, researchers in this field prefer to explore intangible issues such as entrepreneurship, regional linkage, and local culture [12,13,14]. In this way, the previous research concerning regional innovation systems gradually transformed the scope to recognize the optimization of regional coordination for innovation output, government policy, entrepreneurship, technical, and educational environments.

As mentioned above, the triple helix model was created by Etzkowitz and Leydesdorff to analyze regional innovation systems. Besides the change in scope mentioned above, there also exists a transition in the research tools and targets of this model. At the beginning, researchers examined the promotion of regional cooperation to innovation efficiency through case studies [15,16,17]; the research targeted industrial clusters, systematic environments, and new participants in regional innovation. On the other hand, the quantitative research related to the triple helix aims to evaluate the triple helix’s effect on innovation efficiency from an empirical approach. D’Este and Patel examined different channels of cooperation’s influence on knowledge-based innovation by a large-scale survey of UK universities and found that individual characteristics of scholars have potential effects on innovation processes [18]. Besides, in recent years, the research concerning the triple helix has become various, as Martin summarized; along with the shifting of technology, scholars treat innovation in a different scope which requires open, evolutionary, systematic, and multi-factor explanations in terms of regional innovation [19]. Now, scholars can find multi-case analyses, meta-analyses, and simulations in the triple helix research field.

2.3. Research Review and Hypothesis

2.3.1. Government Participation and Regional Innovation

In Triple Helix dynamic systems, even if the three-part degrees of participation are varied among different regions, there is no doubt that the dynamic equilibration relies on the high synergism power. Firstly, government policy is a vital factor influencing regional innovation. Innovative government, which was proposed by Keynesianism, and welfare government have become the core of China’s political structure reformation. In many countries with low levels of government control, innovation, and planning development, these factors are determined by the market; research centers are therefore in charge of technical exploration and invention commercialization. Besides, for high-tech and high-risk events, the government may offer subsidies and foundations for the relevant organizations to cover the risk and cost in the R-D (research and development) process. America’s SBIR (Small Business Innovation Research) and Israel’s Yomza foundation play a vital role in creating these nations’ competitive advantages, and these foundations are mostly conducted by the central government. Compared to the above government pattern, some emerging countries with high levels of government control rely more on the government’s industrial policy and intervention during the high-tech innovation process [20]. In China, the roles of central government and local government are complementary to each other; a central government-issued policy to encourage specific industry innovation gives the local government a signal to support such industries [21]. Moreover, China’s central government has also established national high-tech science parks to promote high-tech technology development [22]. The establishment of industrial policy and science parks has motivated local governments to make efforts in promoting such industries’ competitive advantages through technical devotion. Due to the government’s role of planning, China’s Triple Helix system starts with a germination stage in the science parks. Through a case analysis in Tongji Science Park in Shanghai, Cai and Liu pointed out that the government’s direct participation in regional innovation systems is later than the industry–university coordination, but the results from a single case can hardly explain the emergence of science parks in the whole of China [23]. Although many case analyses in China have examined the government’s engagement in regional cooperative innovation as having a positive effect on regional innovation, due to the sample selection bias, many underdeveloped regions were not included. The mechanism of the government’s participation in the triple helix remains to be explored in a larger sample. Based on the research objective and previous explorations, the authors make the following hypotheses:

Hypothesis 1 (H1).

Central government and industry coordination would positively affect regional innovation efficiency.

Hypothesis 2 (H2).

Local government and industry coordination would positively affect regional innovation efficiency.

2.3.2. Enterprise Participation and Regional Innovation

Modern enterprise’s critical problem is to establish and support an R-D department and for each member in this department, the bread and butter they earn is determined by the achievement of research [24]. In the mid-20th century, high-tech innovation mostly came from large enterprise and emerging enterprise, which was the reflection of Schumpeter’s innovation theory [25]. After the 1950s, along with the economic globalization, the value chain of industry tended to spread all over the world. SMEs (Small and medium enterprises) became the vital part of innovation because of the specialization division. Therefore, enterprises’ engagement in innovation would under the pressure of budget which would drive them seeking financial support to reduce the risk of failed innovation attempt [26]. Despite the government subsidy, the collaboration with university would also save the expenditure cost, with the diffusion of the value chain and waves of industrial diversity, and some large companies also coordinated with famous universities to improve their R-D performance [27,28]. In such a partnership, even though the enterprise could not provide enough intelligent capital, their product commercialization capacity and sharp eyesight to the market also benefitted the other participants in the triple helix. For the specific background of China, the reform and opening-up policy gives enterprise more opportunities to engage in the global value chain and transform their strategy of innovation. More and more enterprises seek universities for collaboration, with the enterprise offering the capital and research plan, and the research center and science laboratory in the university being guided by the industry’s demands. In emerging countries, especially China, even though the government is on the way to transforming into an entrepreneurial government and the institutions are becoming multi-functional, the activation of enterprises’ motivation toward regional development is still the foundational policy orientation. Based on a case analysis of regional cooperative innovation in China, Kim and Lee proved that both high-tech enterprises and universities promote regional innovation, but the critical source of regional competitive advantages were contributed by the coordination between the two entities [29]. According to the review of the dynamic relationship between enterprise and other participants, the authors propose the following hypothesis:

Hypothesis 3 (H3).

The coordination between industry and universities would positively influence regional innovation capacity.

2.3.3. University Participation and Regional Innovation

The transformation from an academic to an explorational focus makes it easier for universities to engage in regional innovation systems; the engagement process is also treated as the reconstruction of the spatial economic structure [30,31]. Local universities have huge effects on the development of regions, mainly through shaping industry clusters’ competitive advantages. In a case study of Denmark’s high-tech wireless cluster, Ostergard and Park found that the only university (Aalborg University) in the industrial agglomeration region played an important role in creating a new generation of products. Moreover, the young institution (built in 1974) provided a great deal of skilled labor as well as infinite creativity to the local companies to support the cluster [32]. Even though the revolution of institutions improves their exploration ability, sometimes there exists a gap between invention and innovation. In the case of giant magneto-resistance, invented by two famous scientists from two well-known universities, the idea was not transformed into an innovation outcome until 10 years’ work from IBM’s team; the reason for this summarized by the authors was mainly because of a lack of coordination with government and industry [33]. The above-mentioned cases prove that a local university’s participation brings positive spillover to the local cluster and a lock-in of development occurs without correct interaction with government. In China’s specific situation, universities’ dynamic engagement with innovation systems took place after the reform and opening-up policy (from the year 1978) and the initial collaboration partner was focal companies. After that, the government, as the main resource allocators, became the new partner of Universities in terms of cooperative innovation. In China, universities’ connections with central government and local government in the Triple Helix system are totally different. Similar to a subsidy to local government, China’s central government also assigned the foundation to encourage universities’ knowledge production and application, which generates positive spillover in regional innovation systems [34]. On the other hand, the tight connection between the local government and the university is not limited to the foundation: the incubator is a new way to improve regional innovation environments, and its establishment is seen as an important motivation for innovation in China [35]. The major formats of university-based incubators include science and technology parks, start-up labs as well as strategic think-tanks. Each kind of incubator plays an important role in regional innovation systems. With a study in Australia and Israel, Rubin et al. proved that the incubator, as a multiple-participant platform, not only accelerates the spillover of knowledge, management experience, and venture capital but also promotes the linkages between research agencies and enterprises [36]. In an empirical study, Mas-verdu et al. investigated 47 firms to examine the sustainable support to local enterprises by incubators and the result showed that the incubator-combined export activity positively affected business survival and increased the innovation efficiency [37]. Besides, with the framework of input-output, a knowledge production process that was used to evaluate a regional incubator’s effect on regional development, the research results indicated that the incubator promoted the amount and the efficiency of Greek government expenditure on R-D [38]. In the specific environment of China, along with the economic reform and the promotion from local government, some university-based incubators established in the 1990s have turned into important innovation sources [39]. In summary, the university’s explorative transition and engagement in regional cooperation bring many positive spillover effects in China’s regional innovation systems. Based on this section’s literature review, we propose the following hypotheses:

Hypothesis 4 (H4).

Regional universities’ coordination with central government would positively affect regional innovation efficiency.

Hypothesis 5 (H5).

Regional universities’ coordination with local government would positively affect regional innovation performance.

3. Methodology

3.1. Research Method

With the review of the previous research concerning regional cooperation and industrial innovation, the application of a quantitative approach would not be able to summarize and compare dynamic connections between different participants. On the other hand, even though qualitative research can figure out the interactions between multiple participants, the bias of case selection and shortage of universality make this approach quite limited. In this way, our research chooses multiple-case analysis by applying qualitative comparative analysis (QCA). Through testing the previous hypotheses, we can have a better understanding of the situation of China’s triple helix innovation and how it is affected by the three participants.

Qualitative comparative analysis (QCA) applies the concepts of “set” and “aims” to find configurations of variables with higher explanatory power, which can better express outcome variables. Besides, QCA uses the systematic quantification method to handle the cases, which enables complex factors to be considered, with the configurations of variables rather than single variable in the calculation process; this approach improves the traditional statistical method’s limitation of variable groups’ blind spot [40]. The QCA approach has casual variables and outcome variables; if one casual variable always presents with a particular outcome, it can recognize that this casual variable could contribute to the results in a large extent. In this paper, the authors apply professor Regin’s fuzzy-set qualitative comparative analysis (fs-QCA). The amendments of fs-QCA compared to QCA is to put a constant (X = 1) as the outcome variable and then to look at the truth table in terms of typologies or trajectories [41]. Fs-QCA is an influential method in social science studies, which is especially suitable for small- and medium-scale samples. Previous applications of fs-QCA in social science studies include community resident resistance effects [42], high-tech industry exploration [43], and regional characters’ effects on innovation efficiency [44]. The previous research received good reviews because it offers a new scope of social science study. Fs-QCA makes comparisons of different cases’ outcomes through two main indicators: the first indicator is consistency and its expression is shown as follows:

$$\mathrm{Consistency}\left({\mathrm{X}}_{\mathrm{i}}\leqq {\mathrm{Y}}_{\mathrm{i}}\right)=\sum \left[\min \left({\mathrm{X}}_{\mathrm{i}},{\mathrm{Y}}_{\mathrm{i}}\right)\right]/\sum {\mathrm{X}}_{\mathrm{i}}$$

(1)

In Equation (1), X is the casual configuration, Y is the outcome variable, consistency is the degree of how X would be the solution of Y, it is quite like the goodness of fit in quantitative study. A consistency higher than 0.8 means that more than 80% of cases meet the conditions and tend to accept the relationship. Another index is coverage; the expression is listed below:

$$\mathrm{Coverage}\left({\mathrm{X}}_{\mathrm{i}}\leqq {\mathrm{Y}}_{\mathrm{i}}\right)=\sum \left[\min \left({\mathrm{X}}_{\mathrm{i}},{\mathrm{Y}}_{\mathrm{i}}\right)\right]/\sum {\mathrm{Y}}_{\mathrm{i}}$$

(2)

Coverage can demonstrate the percentage of cases that can support the relationship; a high coverage stands for more cases being included in a such configuration. It is also worth noticing that high coverage Stands for the casual configuration is exclusive solution to the outcome variable.

3.2. Case Selection

The study of regional innovation systems in China has received increasing attention from researchers all over the world. Even though the economic reform changed China’s developing path, the institutional legacies are still influential after a long period of reform [45]. The provincial unit is quite frequent in the research because each province has a statistical yearbook, which increases the feasibility of research data. Moreover, China’s recent reform policy induced a high degree of regional development imbalance, which brought significant research value for evolutionary economics in terms of resource regeneration and multiple-participant coordination [46]. This research chose China’s 31 provincial units (except for Taiwan, Hong Kong, and Macau) as the sample and by applying fs-QCA we were able to find out the regional multiple participants’ influence on innovation efficiency.

3.3. Variable Design

Based on the research objective, the authors set the regional high-tech innovation capacity as the outcome variable. Among the relevant studies, regional innovation efficiency (RIE) is usually measured by number of patents, number of new technologies applied by national projects, number of new products listed in catalogues, and profits from the new products [47,48,49]. Moreover, some studies also apply questionnaire survey to measure innovation decision of managers to investigate the motivations of innovation [50]. A patent of invention is the most corresponded with high-tech innovation because it comes directly from radical innovation rather than exploitation innovation. Each region’s patents of invention are available at the China statistical yearbook, published by the bureau of statistics.

In our hypotheses, we assume that the five bonds between different participants of the Triple Helix system (Figure 1) would have positive influences on regional innovation efficiency. In this way, our research tries to find suitable variables to indicate the five casual factors. The first two bonds are concerned with government’s interaction with industry. In China’s specific political and economic environment, industry policy pushed by the government is quite conclusive to motivate regional innovation and is also treated by scholars as the measurement of interaction between government and industry [51]. The authors use the percentage of each region’s high-tech subsidy out of China’s total subsidy as the relationship [52]. Moreover, local government’s connection with industry is measured by the percentage of each province’s technical devotion out of the total government expenditure. The third bond, between industry and universities, is widely explored in regional development areas; the variables applied by researchers include dummy variables of collaborative activities [53] and the amount of collaboration channels [54]. Furthermore, considering that the organization’s devotion to the university would immediately fulfill the mission of cooperation and promote the effects of learning [55], the authors of the present study measured the third bond through the university exploration foundation’s percentage out of the total enterprise’s R-D expenditure. A high rate indicates a tight connection between the industry and the university. In terms of the dynamic coordination between universities and government, with regard to the related research in emerging countries, this research calculated the number of a regional university’s Natural National Science projects followed by some classical studies [56,57] as the index of bond 4 (university and central government). Incubators play a vital role in regional innovation and organizations’ performance [58], In China, most of the incubators are held by the combination of local government and university [59]. In this way, some empirical study use the numbers of incubators to evaluate the connections between university and local government [60,61], and the authors applied each region’s university-based incubators in the year 2016 as a variable of the bond between university and local government. As to the source of the five explanatory variables, bonds 1, 2, 3, and 5 are acquired from China’s Regional Innovation Capacity Report published by the Science and Technology Department of PRC. Besides, data on the university foundations in bond 4 are available on the website of the National Natural Science Foundation of China.

To meet the analytical demand of fs-QCA, our research mostly applies the percentage as the variable. Moreover, the regional innovation efficiency and university’s foundation project are standardized. Details of indexes as well as their sources are listed in

Table 1. Indexes of outcome variables and casual variables.

Variables	Variable Descriptions	Source
Regional innovation efficiency (Rie)	Percentage of region’s patents out of the total patents in China	Griliches (1990), Wang (2015)
Bond 1	Percentage of region’s high-tech subsidy out of China’s total high-tech subsidy	Hong (2016), Yu et al. (2017)
Bond 2	Local government’s devotion to technology development (percentage of all financial expenditure)	Shi et al. (2009), Zhou et al. (2012)
Bond 3	Enterprises’ devotion to regional universities for exploration (percentage of total R-D expenditure)	Lee et al. (2014)
Bond 4	Percentage of subjects from central government	Boardman (2009) Lei et al. (2012)
Bond 5	Number of regional incubators (percentage of all China’s incubators)	Etzkowitz (2002) Wang and Li (2011)

.

To examine the effects of individual factors on the outcome variable, we determine a single variable and the absence of single variable’s influence on the study and the results are indicated in

Table 2. Analysis of single conditions’ effects on innovation efficiency.

Condition Tested	Consistency	Coverage
Bond 1	0.714079	0.838160
~Bond 1	0.679245	0.186232
Bond 2	0.757765	0.565550
~Bond 2	0.651947	0.583767
Bond 3	0.418810	0.664932
~Bond 3	0.700156	0.583767
Bond 4	0.553080	0.632934
~Bond 4	0.688296	0.590531
Bond 5	0.462039	0.664932
~Bond 5	0.700156	0.583767

(“~” means the absence of a variable). This approach tests whether single bond could affect the regional innovation efficiency through the calculation of single casual variable’s consistency. As it can be seen, no single condition can explain the outcome variables (no variable’s consistency is more than 0.8). Obviously, a single factor cannot directly promote regional innovation in China. In other words, China’s high regional innovation efficiency is induced by a combination of different factors, which is discussed in the following section.

Considering the research objective, the authors conducted an analysis of the conditions by possible casual configurations that may explain high regional innovation efficiency and the model was identified as followed:

$$\mathrm{RIE}=\mathrm{f}\left(\mathrm{bond}1,\mathrm{bond}2,\mathrm{bond}3,\mathrm{bond}4,\mathrm{bond}5\right)$$

(3)

This process can evaluate which configurations would bring higher innovation efficiency and in which typical regions these configurations are present.

4. Results and Discussion

With the model raised before, the authors apply the fs-QCA program to process the data and create a Truth Table Algorithm, which roughly estimates different configurations of participants’ effects on the innovation outcomes. As indicated in Table 1, according to the different permutations about the five casual factors, the program calculates 32 different configurations. Given that the research contains 31 cases, some configurations could not be covered by any case, so our research ignored the blank configurations and connected the rest of the configurations with the outcome variable. What is important to be noted is that the regulation of truth-table analysis—the primary result of the system—has a threshold score of 0.5. The truth table algorithm is a reflection of the result in first step which includes all the configurations of casual factors and their consistency to the outcome. The primary outcomes are included in

Table 3. Truth Table Algorithm of innovation efficiency with the standardized outcome variable. RIE: regional innovation efficiency.

Bond 1	Bond 2	Bond 3	Bond 4	Bond 5	RIE	Cases	Consistency
1	1	1	1	1	1	3	0.925843
0	1	0	1	1	1	2	0.818073
0	1	1	1	1	0	1	0.689266
0	0	1	1	1	0	1	0.632841
1	0	1	0	1	0	1	0.620192
0	1	0	0	0	0	1	0.429224
0	0	1	0	0	0	1	0.271513
0	0	1	0	1	0	4	0.257853
0	0	0	0	1	0	7	0.182469
0	0	0	0	0	0	9	0.130664

. Results showed that China’s provinces have quite different degrees of regional triple helix cooperation. In Table 3, 10 different combinations exist and at least 16 provinces stay at a low level of triple helix coordination, also having the lowest consistency to the high-tech innovation efficiency. Basically, the higher consistency is correlated with the multiple-participant triple helix model.

To meet the demand of fs-QCA’s standard analysis, our research set the threshold of consistency as 0.8. Then, we labeled consistency values higher than 0.8 as 1 (positive case), with other cases labeled as 0 (negative case). In the process of standardization, we set the positive case as true and the negative case as false, the program produced the solution configurations and we listed the solutions in

Table 4. Adequacy Analysis: High innovation efficiency regions (Presented with solutions).

Truth Table Solution
	Raw Coverage	Unique Coverage	Consistency	Case
Cgs *lgs *os *ins *uni	0.599419	0.240929	0.926009	Guangdong, Zhejiang, Jiangsu
~cgs *lgs *~os *ins *uni	0.490566	0.132075	0.799055	Beijing, Shanghai
Solution coverage	0.731495
Solution consistency	0.926009

.

In Table 4, it can be recognized that two configurations of regional coordination could be the solution to the outcome variable. What is more, the total solution coverage and solution consistency are quite high because the two configurations are beneficial to the regional innovation. Besides, the results indicate that more than one pattern causes high innovation efficiency in China. In regard to this, we will discuss the two typical modes and their pathways to high innovation efficiency regions.

Firstly, the highest consistency solution is related to the configurations of five variables. The completed five factors have a consistency of 0.926009, which almost reached the maturity level. This refers to a developed regional innovation system, and the cases include Guangdong, Zhejiang and Jiangsu. In an economic-geographic scope, both the three cases are located in Southeast China. Moreover, they also have prior advantages in relation to the central government’s reform and opening-up policy. Guangdong has two special economic areas to attract foreign direct investment and the Pearl River Delta contains the earliest electronic manufacturing clusters. With the aim of improving high-tech innovation, Guangdong Government introduced the famous university international companies to promote the regional innovation network, and they gradually narrowed down the subsidy to the labor-intensive industries and encouraged them to migrate to inland China [62]. Besides, Jiangsu and Zhejiang are located in the Yangzi River Delta economic cycle and are both affected by the mega-city Shanghai; the tight spatial cooperation in high-tech industries gave the regional innovation participants channels to share information and build platforms to strengthen and accelerate triple helix cooperation. Moreover, the internationalization of these areas lowered the cost of knowledge interactive searching and the export demand promoted the regional innovation efficiency.

The second configuration has advanced universities and incubators and receives adequate support from the local government. However, the central government’s subsidy and industry participation are relatively less. Considering that the regions belonging to this configuration are two municipalities (Shanghai and Beijing) and are directly under the control of China’s government, the economic gross is relatively low compared to other provinces, and the percentage of subsidy out of the whole country also ranks at the bottom among all the cases. On the other hand, Beijing is the central city in the Bohai Sea Economic Cycle, while Shanghai is the core city in the Yangzi River Delta Economic Cycle. The industry structures of the two cities are quite biased toward the service sector and high-tech departments. The outcome of this differs a little from what we expected before: the regional cooperation in the two cities counts on local government devotion, the participation of universities and incubators rather than industry’s devotion and central government’s subsidy to industry innovation. Beijing and Shanghai have the highest innovation density in China, which is indicated by the highest patents per million inhabitants. However, the percentage of enterprises’ R-D expenditure in the two cities is rather low relative to other regions. Meanwhile, the subsidies from the central government are lower than the rest of the sample. This phenomenon is mainly because of the flow-back effect of university-industry cooperation in the two cities. The fast development of state-owned incumbents and multiple-national enterprises has enabled the industry to establish their own platforms, which attract the university’s devotion in cooperative innovation [63]. This new tendency makes the university–industry collaborations blurry to measure; in this way, triple helix’s different bonds in China’s municipalities need further empirical studies.

Additionally, our research applied the fuzzy-set variable computation to evaluate the relationship between the innovation outcomes and the combination of variables. We combined the two positive configurations mentioned above into two single variables. Through fs-QCA, we labeled configuration 1 as C5 (the configuration with five variables) and configuration 2 as C3 (the configuration with three variables and two absent variables). Then, we used a scattergram to present the result, as shown in Figure 2. The high devotion endowment and frequent regional cooperation bring better innovation efficiency to such regions. In addition, most spots are distributed above the X = Y line; Figure 3 indicates the same trends as Figure 2, but a little difference that needs to be considered is that most spots in Figure 3 are around the X = Y line, as the two absent variables induced a decrease in the innovation efficiency. In this way, the five hypotheses were examined. Another phenomenon worth noticing is that, except for the regions in the two configurations, the other regions’ synergetic innovation is at a relatively low level. The imbalanced industry level and the shortage of intelligent capital caused the failure of regional collaboration and innovation [64]. Meanwhile, the free-rider tendency in these regions decreased the efficiency of multi-participant innovation [65,66].

Considering the fact that the data processing of fs-QCA may have induced error, this paper created a robust check. The main dispute regarding fs-QCA is that it is not a perfect statistical system due to its assignment of the data as positive and negative. In this way, the authors applied a hierarchical regression to evaluate the original data and combined the data’s effects on the dependent variable. The robust test also enlarged the sample scale by adding the data in the years 2014 and 2013. The results are demonstrated in the

Table 5. Robust test (dependent variable: Innovation Efficiency).

	Model 1	Model 2	Model 3
Cgs	0.422 ***	0.673 ***	0.633 **
Lgs	0.315 ***	0.346 ***	0.323 ***
Os	0.234 *	0.261	0.266
Uni	−0.15	−0.025	−0.017
Ins	0.503 ***	0.623 ***	0.6 ***
Constant	0.01	0.013	0.012
Configurations 3		0.52 *
Configurations 5			0.339 ***
R2	0.869	0.878	0.872
Adjusted R2	0.843	0.848	0.84
Observations	93	93	93

*** p < 0.01; ** p < 0.05; * p < 0.1.

, demonstrating that both the central and local government participants played vital roles in promoting regional innovation efficiency. In the aspect of industry–institution collaboration, local incubators brought positive effects to innovation efficiency, while universities did not bring positive effects; the reason for this being mainly due to the process of variable design, which chose the number of universities to describe universities’ participation in regional collaboration, being inappropriate. Above all, the configurations of variables showed positive effects, which indicated that a high degree of triple helix cooperation would induce high technical innovation output in a region. Compared to the fs-QCA, most of the results remained the same.

5. Implications

Based on the clustering and casual analyses of fs-QCA, we have summarized the following implications:

First, as the truth-value table (Table 3) demonstrated, the configuration that represents five low variables has the lowest consistency in relation to the innovation efficiency; the typical cases include most western provinces, which were restricted by their poor infrastructure as well as under-developed concepts of development. Meanwhile, although some regions located in mid-China have relatively higher densities of educational institutions than western regions, the innovation efficiency is relatively low. In the triple helix’s scope, regions in the midlands are still under-developed in terms of co-operative innovation, the university’s process of industrialization is quite slow, and the local Science Parks are mostly founded by local governments, with limited incubating capability. What is more, the industry structure is excessively dependent on primary sectors; in this way, the regional enterprise average scale is becoming larger, which may reduce the innovation capability and induce low-end lock-in [67]. For the regions in western and midland China, the foremost assignment is to upgrade the industrial structure and prevent path dependence, and to adapt the regional innovation environment requires governmental guidance in raising regional industrial diversity and supporting small and medium enterprises.

Second, some regions, such as Liaoning, Hunan, and Hubei, used to be industrial centers of China in the 1950s to 1980s, so they have better industry infrastructure and mature local industry–university cooperation systems. However, along with the economic reform, China’s central government pushed industry transfer and made new industry-supporting policies to promote economic structure upgrading. Because of the connatural human capital and location endowment in these areas, they could not support the demand for dense high-tech clusters, in terms of high-tech innovation, the local government and central government’s devotion and subsidies are limited. In this way, these regions’ innovation outcomes were mostly from the institutions’ and enterprises’ participation. To raise the innovation efficiency, these provinces should make more flexible industry policies to embed themselves in the global value chain, which would bring opportunity to local industry in terms of absorbing new technology and improve the capability of exploitation innovation; like Germany’s industry zones, local government could hold innovation or start-up competitions to activate the regional creativity and foster the exploration of innovation.

Last but not least, the high innovation performances of the two mega cities and three provinces are mainly based on multiple-participant co-operation and the upgrading of regional industrial clusters. Nowadays, China’s central government aims to take further steps in the processes of industry upgrading and cross-regional synergetic development. To promote high-tech innovation and cultivate the growth points of the industrial economy, the government should focus on encouraging the exploration of innovation by protecting intellectual property rights and properly reducing the subsidies to strategic innovation. Innovation incubators have become a useful tool in knowledge-intense innovation in many emerging countries’ regions and increasing both the regional linkages by local institutions and the international cooperation through multinational enterprise’s foreign direct investment are of vital importance for regional development [68].

6. Conclusions and Recommendations

In the 2010s, China slowed down its economic development speed and stepped into a phase called the “new normal”, with innovation-based industrial transformation becoming the most important assignment. The triple helix model provides a new approach, which is treated as the evolution of traditional innovation, since the essential meaning of “framework broken” echoes Schumpeter’s description of innovation (creative and disruptive). Based on the theory of the triple helix and regional innovation systems, our research applied the fs-QCA approach to explore the relationship between regional radical innovation efficiency and the degree of multi-participant collaboration. The authors reached several findings: firstly, we found a tremendous degree of heterogeneity in the innovation efficiency and regional cooperation, with the imbalance of regional development affecting the transformation and upgrading of industrial structures. Secondly, the research outcomes suggested that a multiple-participant governance of the process of innovation would increase the innovation efficiency. Additionally, the local university’s functional transformation would increase the ties between industry and institutions. Finally, in terms of the Chinese specific systemic framework, central government and local government have complementary roles in motivating regional radical innovation. Above all, facing path dependence and the imbalance of regional innovation endowments, like many emerging countries, China’s government should apply industrial policies to shape regional innovation environments and multi-participant cooperation.

Our research also has several limitations; firstly, at the level of the primary data, even if we applied the standardized data, a barrier to accurate computing exists, mainly because of the specific four provincial level cities in China. When the authors applied the variable for the total amount, the numbers of the two cities were quite small; on the other hand, applying the variable per million inhabitants would enlarge the gap between provincial cities and other provinces. This puzzle is hard to solve in terms of fuzzy-set QCA because the outcomes are sensitive to the threshold. Another limitation is the fact that the authors did not address the spatial effects in computing the multiple-participant cooperation. Besides, much research concerning regional innovation divides the research groups into categories for different industries, mainly because different industries have various knowledge densities and institutional environments. In this way, future research could set boundaries for different high-tech industries and their different industrial life cycles to analyze the triple helix’s promotional effects in regional innovation.