Geography of knowledge sourcing, heterogeneity of knowledge carriers and innovation of clustering firms: Evidence from China's software enterprises

Habitat International 71 (2018) 60–69

Contents lists available at ScienceDirect

Habitat International journal homepage: www.elsevier.com/locate/habitatint

Geography of knowledge sourcing, heterogeneity of knowledge carriers and innovation of clustering firms: Evidence from China's software enterprises

T

Cassandra C. Wanga,∗, George C.S. Linb a b

School of Earth Sciences, Zhejiang University, 38 Zheda Road, Hangzhou 310027, China Department of Geography, The University of Hong Kong, Hong Kong

A B S T R A C T The increasing complexity of innovation has triggered a plenty of studies on external knowledge sourcing and innovation performance of clustering firms. While existing economic geography literature places much emphasis on the geographical dimension of knowledge flows, the relative importance of different types of knowledge sources and their geographical identity in innovation have been largely undervalued. Drawing upon a firm-level questionnaire survey and face-to-face interviews on the software firms in a Chinese regional economy, we reveal that, contrary to the conventional wisdom of distance decay and geographical proximity, no significant difference is found between local and non-local suppliers, customers, and rivals as the external sources of knowledge for firm innovation. Local universities and research institutions are identified by firms as the better and more effective sources of knowledge than others outside of the region. Whereas suppliers and rivals make no significant difference to firm innovation, knowledge obtained from customers is reported to be highly significant to the innovation performance of software firms. Findings of this research cast doubts over the prevailing uncritical and undifferentiated perception of the functioning of geography and inter-firm linkages in the processes of firm innovation as well as the under-socialized understanding of knowledge flows between different agents.

1. Introduction In the perennial debate about the nature and dynamics of technological innovation, the issue of externality has never ceased to capture scholarly imagination and arouse competing interpretations. It is now widely accepted that firm innovation cannot be completed solely by internal R&D activities and instead must be facilitated by the knowledge obtained from relevant firms and organizations externally (Sun & Zhou, 2011; Trippl, Tödtling, & Lengauer, 2009). In the studies of externality, attention has been overwhelmingly paid to the geographical dimension of knowledge production and spillover based on the cases of advanced economies. By contrast, relatively less is understood about the functioning of various agents in the process of innovation, such as partners, rivals and universities and research institutes (URIs) in developing countries (Zeng, Xie, & Tam, 2010). Identification of the roles played by the different kinds of agents involved in knowledge production and transfer is important not only because the constraints on time, energy and resources do not allow the firms to heavily and deeply seek external knowledge without selection but also because the knowledge affiliated with different agents tends to be in different scientific or applied nature and associated with technology, market or

∗

organizational aspects (Grillitsch, Tödtling, & Höglinger, 2015). It is the heterogeneous nature of the knowledge associated with different types of agents that has intrigued debates over their spillover effects on innovation (Caloghirou, Kastelli, & Tsakanikas, 2004; Tomlinson, 2010). Against the backdrop of current theoretical advancement, this study attempts to investigate the roles played by the knowledge obtained from several main sources (suppliers, customers, rivals and URIs) and their geographical identity in the process of firm innovation, taking China's software industry as a case. External knowledge sourcing is of particular importance for the indigenous firms in many developing countries where firms are normally characterized by inferior technological capability and inadequate internal R&D investment (Lin et al., 2011; Zhou & Tong, 2003). As the “heart of the information society”, the software industry has experienced rapid growth in China over the recent two decades to warrant itself an interesting and significant case for serious investigations. In particular, we attempt to address the following questions. What exactly is the role played by geographical proximity between a firm and external knowledge sources in the process of innovation? What specific type of knowledge sources is central to the innovation of the firms, and why? The rest of the paper is structured as follows. It starts with a critical

Corresponding author. E-mail addresses: [email protected] (C.C. Wang), [email protected] (G.C.S. Lin).

https://doi.org/10.1016/j.habitatint.2017.10.012 Received 16 September 2016; Received in revised form 28 September 2017; Accepted 28 October 2017 Available online 21 November 2017 0197-3975/ © 2017 Elsevier Ltd. All rights reserved.

Habitat International 71 (2018) 60–69

C.C. Wang, G.C.S. Lin

based primarily upon the practices of developed countries, leaving us to wonder how geography and the heterogeneous nature of knowledge carriers in developing economies would affect technological innovation of a firm. For instance, unlike the situation of industrial clusters in many advanced economies where firms interact among themselves intensively and extensively to engage in ground-breaking innovations, many software firms in China have been preoccupied by the simple task of imitations, modification, and application of the technology originating from advanced economies to improve efficiency in production and expand market penetration. These firms are clustered simply to share the costs of regional infrastructure or the benefits of externality (Lin et al., 2011). They are located close-by for convenience and not for cooperation in innovation. Under this circumstance, the notion of geographical proximity or local scale that is central to the theoretical literature of industrial clustering may not work in the same way in China as it has been in many Western advanced economies. Because these firms are mainly involved in a “fine-tuning" of the existing technologies, cognitively distant knowledge from non-local scales is not as significant as what it has been observed from many advanced economies. In a word, geographical scales of knowledge sources make much less sense to the firms in a developing economy such as China. This noticeable gap in the existing literature leads us to formulate the first research hypothesis.

evaluation of the extant literature on external knowledge sourcing and firm innovation based upon which two research hypotheses are made. This is followed by a clarification of our own research design and methodology. Attention is then turned to China where the software industry has experienced such a dramatic growth that it becomes an interesting case for the study of how geographical proximity and different types of knowledge sources affect the innovation performance of the firms. Major findings of this research and their implications are summarized and discussed in the end. 2. External knowledge sourcing, heterogeneity of knowledge carriers and firm innovation: the devil in the details? 2.1. Geography in knowledge sourcing and innovation Over the last decade, technological innovation is increasingly understood as a result of some complex, collective and cumulative processes of knowledge production involving many different kinds of agents from within and outside of a firm (Asheim, Boschma, & Cooke, 2011; Fu, Revilla Diez, & Schiller, 2013). In recognition of the importance of external knowledge for innovation, economic geographers have devoted much of their attention to the geographical dimension of external knowledge sourcing and generated significant insights (Grillitsch & Tripple, 2014; Howells, 2012; Hu & Lin, 2013). On one hand, it has been observed that, since the generation, utilization and distribution of knowledge depend on the frequency and density of interaction among firms, geographical proximity makes these processes more easily and smoothly (Caloghirou et al., 2004; Lin, Yang, & Hu, 2012). On the other hand, while geographical proximity may foster face-to-face contacts and smooth the process of knowledge exchange, it brings along a risk of creating a locally closed mentality that may inhibit innovation (Munari, Sobrero, & Malipiero, 2012). Knowledge from non-local scales, although culturally, cognitively and technologically distant, is believed to be a significant complement of the existing knowledge stock (Sidhu, Commandeur, & Volberda, 2007). The multiscalar nature of knowledge interactions and the different role they play in the process of innovation have been well-recognized and extensively documented (Asheim et al., 2011; Bathelt, Malmberg, & Maskell, 2004; Cooke, Boekholt, & Tödtling, 2000; Tödtling, Grillitsch, & Höglinger, 2012). It is believed that a combination of the knowledge of different scales (local and nonlocal), different forms (codified and tacit) and different kinds (analytical, synthetic, symbolic) would enhance the innovative performance of firms (Halkier et al., 2012; Manniche, 2012; Strambach & Klement, 2012). What has not been well elucidated is, however, from whom (clients/ suppliers/customers etc.) a firm tends to obtain valuable knowledge for innovation. In reality, firms tend to seek specific knowledge through contacting with many knowledge carriers, formally or informally, with little concern over the geographical scale of these particular knowledge providers. We argue, therefore, that identification of the specific types of knowledge sources/providers is no less important, if not greater, than the location of knowledge sources to our understanding of the processes of innovation. The impact of geographical proximity depends upon specific characteristics of different knowledge sources. It is claimed that knowledge sourcing from URIs tend to occur more often at the local scales whereas knowledge sourcing from customers and suppliers are frequently located outside of the region (Grillitsch et al., 2015). As Boschma (2005) has pointed out, geographical proximity per se is neither a necessary nor a sufficient condition for learning and knowledge exchange to take place. Its importance is related to other dimensions of proximity such as cognitive and relational proximity. As it stands, the larger cognitive distance between firms and URIs may entail geographical proximity for a better knowledge exchange. By contrast, the relatively higher cognitive and relational proximity between firms and their customers/suppliers makes geography much less important. In addition, existing theoretical and empirical studies have been

Hypothesis 1. Location or geography of knowledge sources is less important than the heterogeneous nature of the knowledge carriers to the innovative performance of firms in the case of China. 2.2. Agents in knowledge sourcing and innovation Ever since the concept of “open innovation” was introduced, there have been strong scholarly interests in how cooperative ties and networks or cooperation partnerships could facilitate firm innovation (Segelod & Jordan, 2004; Tomlinson, 2010). Research in this vein has generated mixed and contradictory results, however. Some researchers have identified suppliers as the key sources of knowledge because they have better expertise and a more comprehensive understanding of the supplied parts and components that are necessary to fix particular technical problems (Nieto & Santamaría, 2007; Tsai, 2009). It is empirically confirmed that co-operation with suppliers enables firms to reduce lead times of product development and enhance flexibility, product quality and market adaptability (Chung & Kim, 2003). Nevertheless, others maintained an insignificant relationship between cooperations with suppliers and innovation (Ledwith & Coughlan, 2005). For instance, by investigating 597 firms in the UK, Freel (2003) has found that there is no significantly important relationship between cooperation with suppliers and product innovation. It has been argued that cooperation with customers are theoretically significant for product innovation simply because only customers can provide user friendly technical know-how and market information, and the debate continues (Tether, 2002). It is observed that close linkages between software firms and their customers are crucial in the innovation process (Bettencourt, Ostrom, Brown, & Roundtree, 2002). In particular, customers, by virtually integrating into a company's innovation process could provide valuable inputs for new product development (Füller and Matzler, 2007). Segelod and Jordan (2004) have demonstrated that linkages with customers are the most important ones during the whole process of product innovation from the idea proposal, decision to investment in innovation, development and commercialization of the new product. However, Romijn and Albaladejo (2002) have shown that interactions with customers do not enhance the innovative performance of the firms in southeast England. Evaluation of the role played by interactions with rivals is even more controversial because, on one hand, rivals may share common problems and can be teamed up to reduce cost and increase technological efficiency, but on the other hand, collaborations with rivals are potentially 61

Habitat International 71 (2018) 60–69

C.C. Wang, G.C.S. Lin

813). Hence, URIs play a negligible role in the innovation of software firms in China. Furthermore, the less advanced technology adopted by Chinese software firms tends to confine these firms into a similar market niche. A lack of protection of IPR in transitional China also means that software firms see themselves as competitors with suspicions of information leakage. Research has already suggested that a lack of trust among firms had undermined knowledge circulation and exchange in China (Wang & Lin, 2008). Under these circumstances, interactions with rivals do not help with the practices of innovation. Software platform suppliers and equipment manufacturers, through establishing subsidiaries and after-sales office in China, only provide standardized products to software firms as a basic platform for software development. They normally offer no customized information and valuable knowledge to China's software firms. Suppliers therefore play a less important role in innovation. Finally, an emphasis on applied knowledge for instant market expansion has led Chinese software firms to rely upon their customers as the main external source of knowledge to facilitate innovation. The alliance between software firms and their customers out-weighs any other linkages with suppliers, peers, or URIs within or without the region. The special nature of external knowledge sourcing in the process of firm innovation in the context of a developing economy has not been adequately addressed in the existing literature. This gap leads us to make our second hypothesis.

dangerous bearing in mind the similar markets they target and their unwanted outward knowledge spillovers (Miotti & Sachwald, 2003). It has therefore come as no surprise that interactions with rivals are believed to have positive impacts upon firm innovation by some but negative effects by others (Nieto & Santamaría, 2007; Wang & Lin, 2008; Zeng et al., 2010). A case study of the toothbrush cluster of firms in China suggested that the friendly ties with competitors had contributed to the successful and innovative performance of the firms (Li, 2014). Another study of the electronic firms in China's Shenzhen Special Economic Zone reported different findings that interactions with rivals run the risks of knowledge leakage because of the immature institutional and market environment in the developing countries where the problem of imitation remains pervasive (Wang & Lin, 2008). It is acknowledged that knowledge transfers between the firms and competitors may not necessarily to be collaborations, but involves more complex interactions. It is observed that knowledge sourcing from rivals is usually done by monitoring rather than collaborations (Tödtling, Lehner, & Trippl, 2006). Beyond inter-firms interactions, the importance of collaboration with universities and research institutes to firm innovation is another source of confusion and debates (Audretsch, Lehmann, & Warning, 2005). The crux of the matter here basically concerns whether or not URIs have the motivation and capability to spill over valuable knowledge to firms (Tsai, 2009). Proponents claim that government's encouragements on university-industry collaboration, coupled with the financial pressure upon URIs, have pushed URIs to engage with industrial firms to improve productivity and competitiveness (Nieto & Santamaría, 2007). This assertion is contested by others who argue that URIs have the difficulties to create the kind of knowledge applicable to the market (Caloghirou et al., 2004). Empirical analyses have shown a great variation in terms of the relationship between collaborations with URIs and the innovation of firms. Liefner, Hennemann, and Xin (2006) investigated the cooperation patterns of firms in Beijing and found that collaborations with URIs had contributed to the design of new products. Another case study of technology transfer from China's Tsinghua University to industrial firms confirmed that URIs have been an important source of industrial innovation (Liu & Jiang, 2001). Yet other studies have presented evidence to show that there was no significant impact of URIs on the innovation of firms (Ledwith & Coughlan, 2005). The inconsistent and contradictory findings identified above can be attributed to a negligence of the differentiated knowledge bases across a variety of sectors. Those sectors and firms with an analytical knowledge base tend to rely more on codified knowledge and hence benefit from the interactions with URIs to generate radical innovations whereas others with a synthetic knowledge base are often involved in incremental innovations through interactions with customers and suppliers (Asheim et al., 2011; Manniche, 2012). Firms with a symbolic knowledge base require predominantly tacit knowledge and a good understanding of trends and cultural artfacts to improve their competitiveness (Tödtling et al., 2012). We therefore advocate a situational and disaggregate approach toward understanding the importance of external knowledge carriers to firm innovation being mindful of the differentiated knowledge bases across sectors as well as diverse trajectories in different regional contexts and their contingency upon various social, cultural, and economic circumstances. For the firms in the software industry, the innovation process involves more tacit knowledge from customers or suppliers than codified knowledge from URIs. This differentiation is noticeable in many developing countries where firms tend to consider URIs to be slowly responsive or unresponsive at all to industrial needs because URIs usually provide the kind of basic knowledge that cannot be easily and directly transformed into new products (Tether & Tajar, 2008). For a developing country such as China, “the fundamental problem is that even the best universities in developing countries are often behind the curve of commercialized technology in a globalized economy, even if some of their scientific research may be cutting edge” (Wu & Zhou, 2012, p.

Hypothesis 2. Customers play a more important role than other types of knowledge sources (URIs, rivals and suppliers) in the innovation process of China's software firms. 3. Data and methodology To test the research hypotheses, we conducted a detailed investigation of external knowledge sourcing and firm innovation in the software industry in Hangzhou, China. We have selected the software industry in China as our case for two reasons. First, the software industry has been one of the most rapidly expanding industrial sectors within and outside of China. Notwithstanding the increasing significance of the software industry, the existing literature on external knowledge sourcing and firm innovation continues to be preoccupied by research on manufacturing firms, leaving us with an awkward knowledge gap to be filled (Segelod & Jordan, 2004). Second, the software industry is well-known for its knowledge-intensive nature in which knowledge sourcing and inter-firm networks play an important role in the process of innovation. By its virtue, the software industry provides an interesting case highly relevant to the specific issues to be addressed in this research (Grimaldi & Torrisi, 2001). Software firms can be generally divided into platform suppliers and firms focusing on software applications (Segelod & Jordan, 2004). The former delivers generic technology and tools that are the basis for developing software solutions by the latter. The platform suppliers are normally large multinational corporations such as Microsoft, IBM and Oracle etc. who are in the highest position in the value chain focusing on radical innovation. Comparatively, firms focusing on software production and application stand at the lower end of value chain. They mainly provide standard solutions for industrial firms and public organizations (Isaksen, 2004). Software firms in Hangzhou can be largely categorized into the second group. Their suppliers include not only platform suppliers but also equipment providers. They serve their products and services to a widely range of companies in different industrial sectors as well as public institutions such as hospitals, universities and governments etc. (Fig. 1). The selection of Hangzhou as our study site requires explanations. Located in eastern China with a distance of 160 km away from Shanghai (Fig. 2), Hangzhou is the capital city of Zhejiang province where individual and private economies have been growing most rapidly in China (Wei, Li, & Wang, 2007). Among many others, the city is characterized by its central location in coastal China, rich historical and 62

Habitat International 71 (2018) 60–69

C.C. Wang, G.C.S. Lin

cultural endowments as well as a well-established local tradition of entrepreneurship and private enterprises. At the end of 2013, Hangzhou, with a population of 7 million people (15% of the province) generated a GDP of 834 billion yuan—nearly a quarter of what was produced by the province as a whole (ZJSB, 2014). The GDP per capita in Hangzhou is well above the average of the province and almost three times higher than that of the nation (ZJSB, 2014; CSSB, 2014). This is a significant site where the influences of marketization and globalization have been strongly felt and where individual and private entrepreneurship has been fully developed to push the limits within a socialist economy undergoing market transition. Hangzhou has also been identified as “China's Software City” by the Ministry of Industry and Information Technology in 2014 and the “City of E-business" by China Electronic Commerce Association in 2008. Hangzhou is perhaps best known nationally and globally as the home of Ali Baba—China's most important e-commerce enterprise with one of the largest sales volumes in the world. The software industry has been so well developed that it becomes the pillar industry of the regional

Fig. 1. Value chain of software firms in Hangzhou.

Fig. 2. Location of surveyed software firms in Hangzhou.

63

Habitat International 71 (2018) 60–69

C.C. Wang, G.C.S. Lin

economy generating an added value that is equivalent to 8.13% of Hangzhou's GDP at the end of 2013 and a total sales revenue equivalent to 5% of the national total (MIIT, 2014). It should be acknowledged that a study of one single city-region, however significant it may be, cannot generate research findings broad enough to make overall generalizations. Nonetheless, a detailed investigation of the experiences of the software firms in a leading economy such as Hangzhou can contribute significant insights into the dynamics of external knowledge sourcing and firm innovation in the context of a regional economy where sufficient space is allowed for market and global forces to play their roles. The data used in this research are derived from our own questionnaire survey and in-depth face-to-face interviews with software firms, relevant government officials and directors of the association of the software industry in Hangzhou conducted in 2013–2014. According to the online data released by the Ministry of Industry and Information Technology, there were 881 software firms in Hangzhou in 2013. We obtained a name list of Hangzhou's software firms from the Association of Software Industry and then conducted a pilot study of 20 sampled firms to test the response to our questionnaire. Based on the feedbacks of these firms, we further revised our semi-structured questionnaire before it was randomly distributed to 400 software firms with the assistance from the Center of Hangzhou High-tech Human Resources and the Bureau of Development and Reform of Binjiang District, Hangzhou. In the end we obtained valid responses from 110 firms, representing an effective participation rate of 28% and a sample rate of 12.5%. Most of these firms are clustering in the downtown (Fig. 2). Data and information gathered from the questionnaire survey were then further verified and cross-checked through our twenty five indepth interviews. Since firms are reluctant to accept external interview, we completed the interviews by a snowballing approach. Our interviewees not only include several larger innovative software firms, such as Sunyard, Singlee and Insigma but also other less innovative smaller software firms. We have successfully interviewed nineteen technical directors, managers and CEOs from 19 software firms and 3 managers from their customers and suppliers, 1 university faculty member, 1 government officer from Hangzhou Economic and Information Technology Commission and 1 office director from Hangzhou Software Association. Information from firms includes: 1) Basic information and development trajectory; 2) Innovation input and output; 3) Relative importance of knowledge sources at local and non-local scales in firm innovation; 4) Frequency of interactions with customers, suppliers, rivals and URIs in the process of innovation; 5) Importance of customers, suppliers, rivals and URIs in the process of innovation; 6) Trust relationship with customers, suppliers, rivals and URIs. Questions for government and software industrial association mainly include the growth and innovation of Hangzhou's software industry, the competitive and cooperative relationship among software firms, knowledge sourcing of the clustering firms and trust relationship among firms etc. Each interview lasted for about 1–3 h depending on the interviewees and we tape-recorded the interviews with the permission and consent of the interviewees. As shown in Table 1, our sampled firms cover a variety of firms with different sizes. Of the 110 firms, 30% are in small size with a workforce of less than 50 employees, 20% have an employment of 51–100 people and 16% are in large size with a workforce of more than 500 people. The remaining one-third are medium-sized firms. Our sampled firms are also representative of the software industry in Hangzhou in terms of its ownership structure. More than 80% of our sampled firms are domestic firms among which private-owned enterprises are predominant whereas foreign-invested firms only accounted for 18%. Measuring software innovation is a debatable issue, but software copyrights, patents and new products appear to be the three indicators that have been commonly used although they are not flawless (Edison, Ali, & Torkar, 2013). Our questionnaire was originally designed to ask firms about their innovation output by copyrights, patents and new

Table 1 Profile of the sampled software firms. Source: Our own survey. Characteristics Number of employees < =50 51–100 101–300 301–500 > 500 Ownership FIEs SOEs POEs Total

Number of firms

Percentage (%)

33 22 19 18 18

30.00 20.00 17.27 16.36 16.36

19 13 78 110

17.30 11.80 70.90 100

products. Unfortunately, most of the firms we contacted refused to disclose any information about their innovation output. In the end, we had to resort to the online database established by the State Intellectual Property Office (http://www.pss-system.gov.cn/sipopublicsearch/ portal/index.shtml) through which the licensed patent information of firms can be obtained by simply typing the full name of the firms.1 Innovation—the dependent variable of this study—is a dummy variable measured by the patents achieved by the software firms. Innovators refer to the firms holding at least one authorized patent and non-innovators refer to those without any authorized patent. We analyzed four types of knowledge sources in the process of innovation, namely, suppliers, customers, rivals and URIs since most of our sampled firms reported that the role played by other external knowledge sources such as state agencies and professional associations was negligible. To scrutinize the influence of geographical scales, we divided these four types of knowledge sources into local and non-local using the geographical boundary of the Hangzhou Municipality as a watershed.2 Relative importance of each type of knowledge sources to firm innovation is measured by the frequency of interactions with knowledge sources in the process of innovation. The frequency of interactions with external knowledge sources is measured by a five-point likert scale with 1 being the least frequent and 5 the highest. It is noted that the importance of different types of knowledge sources can be affected by the ability of surveyed firms to establish a broad of connections with external sources. For instance, it is possible that a firm regards its interactions with suppliers as not important simply because it fails to establish effective supplier-networks which are able to provide valuable knowledge for innovation. However, even the surveyed firms themselves could not figure out whether or not this possibility stands and to what extent it affects the importance of external sources. Therefore, this study merely focuses on the existing interactions between software firms and external knowledge carriers to investigate the relative importance of different types of knowledge sources. Logistic regression analyses are deployed to evaluate the relative importance of each type of knowledge sources. Five controlled variables are included in the regression equations. 1) Firm size is calculated as the total number of full-time employees in 2013; 2) Firm age is measured by the span of years from the year of establishment to the year of 2013; 3) Ownership is controlled as a dummy variable with 1 denoting that the firm is foreign-invested and 0 otherwise; 4) Affiliated R&D facility is controlled as a dummy variable as well: if a firm reports that it has an affiliated R&D facility, the variable would be 1 and otherwise 0;

1 We acknowledge that using the number of patent as a single indicator of innovation has its limitation. Future studies on innovation of software firms should explore other measures of innovation. 2 We acknowledge that a dichotomy of the geographical scales into local and non-local has its limitations since recent research has highlighted the role of multi-scales for knowledge sourcing (Halkier et al., 2012; Asheim et al., 2011). We have opted for the local/non-local dichotomy for the consideration of simplicity, clarity, and comparability.

64

Habitat International 71 (2018) 60–69

C.C. Wang, G.C.S. Lin

5) R&D intensity is measured by the ratio of R&D expenditure to total turnover.

firms. I don't care whether they are in Hangzhou or elsewhere. Where there is an opportunity for business or useful knowledge, I will chase for it, no matter where they are” (interview notes, 10th of July 2013). “The advanced communication technology allows us to contact our business partners via e-mail, video conference or telephone. We can even fly over to their place if it is necessary. It makes no difference whether my customers are in Hangzhou or in Beijing …... For the case of interactions with universities, we tend to choose local ones, since we know them very well, not to mention that Zhejiang University is quite good” (interview notes, 13th of July 2013). These interesting findings encourage us to further probe into the relative importance of different types of knowledge sources in innovation as well as the role played by geographical proximity in the relationship between external knowledge sources and innovation.

4. Findings and interpretations

4.2. Relative importance of different types of knowledge sources

4.1. Importance of geography and knowledge providers

The importance of customers and URIs in the process of innovation can be further demonstrated by a simple t-test between innovators and non-innovators. As shown in Table 3, innovators indicate a significantly higher frequency of interactions with customers and URIs and give a significant higher appraisal on the importance of such interactions than non-innovators. Information from our interviews provides insights into differentiation. Most of the software firms in Hangzhou engaged in the application software that is relatively less complicated in technology and have a low entry barrier. They worry about knowledge leakage and therefore choose not to have frequent contacts with rivals. A firm manager shared his worries with us about knowledge leakage in their interactions with their rivals: “We certainly do not encourage our employees to discuss company's projects with people from our rivals. First, there is nothing to talk about, because our targeted market is different although we are working in the same applied software industry. Interactions with them do not bring us anything. Second, even if our rivals share with us a common interest, our conversation and discussion with them never involve any core technology or idea for innovation. After all, the most important innovation lies in our understanding of the demands of the targeted industry. Too many contacts with our rivals increase the risk of knowledge and information leakage. Nothing is worthy of the risk” (interview notes, 2nd of September 2013). Since the trust issue impedes knowledge exchange between firms and rivals, does geographical proximity play any role in the relationship between firms and their rivals? Among the surveyed 110 firms, 89 firms (81% of all) indicate that local clustering of rivals does not exert any positive or negative influence on their operation and innovation. More than 76% of surveyed firms report that they do not have frequent linkages with local rivals. Even among the rest of 24% of firms who show a frequent or very frequent linkage with local rivals, a half of them reveal that knowledge from interactions with local rivals does not help to promote their innovation. Thus, geographical proximity does not

Table 2 Relative importance of knowledge sources at local and non-local scales in firm innovation. Source: Our own survey. Knowledge source

Local

Non-local

Equally important

Neither

Total

Rivals Suppliers Customers URIs

10 (9%) 19 (18%) 9 (8%) 36 (33%)

12 (11%) 17 (16%) 14 (13%) 6 (5%)

11 14 87 55

77 (70%) 54 (52%) 0 (0%) 13 (12%)

110 104 110 110

(10%) (13%) (79%) (50%)

(100%) (100%) (100%) (100%)

In our questionnaire survey, respondents are requested to answer the question about which geographical scale (local or non-local) is more important for firm innovation. Four options are provided: local scale, non-local scale, equally important and equally unimportant. The results show an interesting variation depending upon the specific sources of knowledge (Table 2). A lion's share (70%) of the sampled firms did not identify rivals as an important source of knowledge obtained for innovation regardless of their location (whether these rivals are local or non-local). More than half of the firms reported that local and non-local suppliers are equally unimportant. Customers were identified as an important external source of knowledge by all of the surveyed firms and nearly 80% believed that local and non-local customers are equally important. As such, location does not seem to be a factor to affect external knowledge sourcing and firm innovation. The only exception is URIs: although half of the surveyed firms believed that local and non-local URIs are equally important, over onethird of the survey firms identified local URIs as the important external source of knowledge which outnumbers those reliant upon non-local URIs by a large margin. Three important findings can be derived here. First, most of the surveyed firms claimed that the knowledge obtained from rivals and suppliers is less important than that from customers and URIs. Second, location or distance does not make any difference for customers to serve as external sources of knowledge for firm innovation. Finally, local URIs are identified as the external sources of knowledge more important than those located far away. Our interview notes further confirm and explain these findings. A manager explicitly cast his doubts over the importance of geography in the process of firm innovation. “Actually, I do not understand why you are so obsessive with the geographical scales of our partners. It never occurred to me that geography should be considered before we make connections with other Table 3 T-test results between innovators and non-innovators. Source: Our own survey. Mean

Frequency of interactions with rivals Frequency of interactions with suppliers Frequency of interactions with customers Frequency of interactions with URIs Importance of interaction with rivals in process of innovation Importance of interaction with suppliers in process of innovation Importance of interaction with customers in process of innovation Importance of interaction with URIs in process of innovation

Innovators

Non-innovators

2.88 2.38 4.43 3.49 2.90 3.16 4.24 3.51

2.88 2.64 3.84 2.82 2.79 2.89 3.58 2.66

Note: ** Statistically significant at the 0.01 level. *** Statistically significant at the 0.001 level.

65

T-value

P-value

0.021 1.437 3.182*** 3.268** 0.833 1.470 3.382*** 3.883***

0.983 0.155 0.001 0.002 0.407 0.145 0.001 0.000

Habitat International 71 (2018) 60–69

C.C. Wang, G.C.S. Lin

Table 4 Descriptive statistics and correlation matrix of dependent, independent and control variables.

1. Innovation (patent) 2. Firm size 3. Firm age 4. Ownership 5. Affiliated R&D facility 6. R&D intensity 7. Frequency of interactions with rivals 8. Frequency of interactions with suppliers 9. Frequency of interactions with customers 10. Frequency of interactions with URIs Note:

∗∗∗

p < 0.001,

∗∗

Mean

S.D.

1

2

3

4

5

6

7

8

9

0.65 573 10 0.16 1.88 2.79 2.88 2.47 4.23 3.25

0.478 2056 7 0.372 0.324 1.453 0.566 0.87 0.885 1.079

0.153 0.225* 0.166 0.208* 0.609*** −0.002 −0.141 0.318*** 0.297**

0.106 0.008 0.083 0.258** 0.047 −0.197* 0.141 0.193*

−0.016 0.169 0.234* −0.136 −0.140 0.084 0.166

0.086 0.251** −0.077 0.280** −0.003 0.078

0.181 0.019 −0.047 −0.097 0.244*

−0.073 −0.061 0.301*** 0.280**

−0.167 −0.071 0.188*

−0.090 0.014

0.458***

∗

p < 0.01, p < 0.05 (two-tailed).

entry barriers for application software firms, they always keep an eye on their rivals. They are afraid of their innovative ideas being stolen” (interview notes, 6th of August 2013). A manager interpreted how customers help them with product development and innovation. “The interests of our customers are the main sources of our innovation. Since customers know the market best and they can share with us their after-use experiences, they inspire us to be innovative. Sometimes they can even help with solving our technical problems” (interview notes, 15th of August 2013). The customers from different sectors confirm their importance for innovation of software firms. “The online booking system employed by our hospital was designed by a locally newly-established small software firm. We alert this firm about when and how the software bugs appear and offer our advices on how to improve this software product” (interview notes, 8th of May 2017). “The founder of Singlee was our chief of software section of our bank. He knows exactly what we want. He founded Singlee to fulfill our demand on financial software. I believe that we to some level force our suppliers to be innovative and of course, give them useful information and feedbacks to help their development” (interview notes, 8th of May 2017). Although customers play a significant role in firm innovation, the importance of geographical proximity in knowledge flows between firms and their customers is not as outstanding as claimed by the literature of industrial cluster. Almost 70% of surveyed firms show that most of their customers are not located in Hangzhou city and therefore geographical proximity with customers is not their concern. The rest of 30% firms mainly deal with local customers, however, three quarters of them indicate that knowledge from local and non-local customers is equally important for their innovation. We conducted a series of logistic regression analyses to further identify the relative importance of different types of knowledge sources to firm innovation. The VIF of each variable is below 2 and the correlations between variables are low (Table 4). Therefore there is no multicollinearity problem in this study. The results of logistic regression modeling are presented in Table 5. Judging by the correctly predicted percentages (all higher than 80%), the independent and controlled variables selected for modeling can well explain the probability of innovation of software firms. Model 1 presents the impact of all controlled variables, among which only R&D intensity appears to be positively significant. This suggests that innovation requires internal R&D efforts to help enhance firms' absorptive capability to better search, evaluate and utilize external knowledge. The positive impact of R&D intensity is robust even after the independent variables are entered into the models (Models 2, 3, 4). Model 2 examines the impact of inter-firm interactions on the innovation of software firms. The exponential values of the beta coefficients and their significant level suggest that interactions with rivals

enhance valuable knowledge flows between firms and local rivals. In terms of suppliers, software firms do not consider their interactions as an important source for innovation not only because most of the suppliers are platform providers who spill very little valuable knowledge but also because some of the current suppliers may turn into their rivals later on. “In China, many software firms are not satisfied with being just a supplier for other firms because, if they have capability of serving a core modular, they tend to be capable of providing the whole products. It is very easy for them to extend their business downwards to make more profits” (interview notes, 10th of October 2013). “Our suppliers are software platform providers, like Microsoft. They provide us with standardized products without any extra information or knowledge that is useful for our innovation” (interview notes, 7th of August 2013). This information has been verified by an equipment supplier of software firms. “Our company functions as a sales office with a focus on sales and after-sale services. I do not think we can provide useful information or knowledge to our customer for their innovation. Instead, the feedbacks from our customers have efficiently improved our products” (interview notes, 6th of May 2017). The turbulent environment facing China's software firms leads to a lack of trust in not only rivals but also suppliers, even geographical proximity fails to boost a mutual trust relationship between them. By digging deeper into our database, we find, firstly, only 50 surveyed firms (46% of all) inform that a majority of their suppliers comes from Hangzhou city. And among these 50 firms, only 5 firms show that they trust their suppliers for knowledge exchange. In other words, more than a half of surveyed firms locate in Hangzhou not for being close to their suppliers, and most of firms do not forge a trust relationship with local suppliers. Second, 60% of surveyed firms do not have frequent linkages with their local suppliers and only 25% of them indicate that knowledge from local suppliers benefits to their innovation. It implies that geographical proximity does not play an important role in the relationship between firms and their suppliers. Only the customers turned out to be the partners in whom the software firms have the confidence and trust. Furthermore, an emphasis on the application software that characterized the firms in Hangzhou also helps explain the finding that their innovation depends on customers' demand and requirements. The business culture of Zhejiang that values the market economy and demand more than others also highlights the significance of customers to firm innovation. A government official commented: “Zhejiang's economy is dominated by private-owned enterprises with quick and flexible ways to respond to market demands. The software industry makes no exception. Especially for application software companies, they always give the highest priority to customers' requirements. Any innovation would be based on the knowledge and information obtained from customers. By contrast, due to the lower

66

Habitat International 71 (2018) 60–69

C.C. Wang, G.C.S. Lin

spin-offs of URIs, we received very positive comments on the role of URIs in the process of innovation. “Our founders graduated from Zhejiang University ……We have intensive and frequent linkages with Zhejiang University in a variety of forms. I think interactions with universities help our innovation. First, we have collaborations with university faculties through which we not only get technical solutions but also know better about the quality and capabilities of their students who could be our employees later. Second, although technologies from URIs usually cannot be easily transformed and applied directly into the market, we can collaborate and discuss with them to turn their inventions into innovations” (interview notes, 7th of August 2013). This opinion has been further supported by a faculty member of Zhejiang University: “we have established a long-term relationship with a number of local software firms founded by our alumni. We know each other very well and hence our cooperation goes smoothly. We are dedicated to technological development while they mainly focus on sales and marketing. Our students also work for them when they are short of hands” (interview notes, 10th of May 2017). By contrast, a firm without any pre-existing social relationship with URIs disclosed the following information. If the firm does not have any long-term relationship with the URIs or does not know the URIs well enough, they may end up with getting the wrong persons to do collaborations with a disappointing result. “We once collaborated with University of Electronic Science and Technology of China to develop a technology. They claimed they had developed this technology for years and are expertise on it. To our disappointment, their so-called advanced technology turned out to be even less advanced than what we already have! Furthermore, collaborating with URIs is expensive. They do not care about costs but we do” (interview notes, 12th of August 2013). Furthermore, a closer looking at out database reveals that 81% of firms have selected local URIs to make collaborations, which suggests that geographical proximity plays a significant role in the relationship between firms and URIs. It is revealed that the founders of surveyed software firms are inclined to select a location near to their alma mater so as to maintain a close relationship with the URIs. Some of them are even the spin-off of local URIs and tend to locate nearby to take advantages of the excellent technology resources and high-qualified labors.

Table 5 Logistic regression results. Source: Our own survey. DV: Patent

Model 1

Model 2

Model 3

Number of employees Age Ownership Affiliate R&D facility R&D intensity

1.001 0.992 1.378 2.327 3.722***

1.001 0.998 4.735 3.321 4.198***

1.001 1.005 4.385 3.871 4.414***

1.348 0.471 2.184*

1.641 0.497 2.440* 0.806 63.308*** 85.6

Frequency of interactions with rivals Frequency of interactions with suppliers Frequency of interactions with customers Frequency of interactions with URIs Chi-square Correctly predicted percentage

53.512*** 80.9

63.040*** 85.6

Note: Values in this table refer to the exponential values of beta coefficients that describe the factor indicating the change of the estimated probability of innovation. ∗∗∗p < 0.001, ∗ p < 0.05. Table 6 Robustness check results. Source: Our own survey. DV: Patent

Model 1

Model 2

Model 3

Number of employees Age Ownership Affiliate R&D facility R&D intensity

1.001 0.992 1.378 2.327 3.722***

1.001 1.017 3.116 4.264 5.005***

1.001 1.013 3.566 3.830 5.004***

1.530 2.248 3.038**

1.240 2.411 2.762* 1.222 70.344*** 85.6

Importance of interactions with Importance of interactions with Importance of interactions with Importance of interactions with Chi-square Correctly predicted percentage

rivals suppliers customers URIs 53.512*** 80.9

70.061*** 86.5

Note: Values in this table refer to the exponential values of beta coefficients that describe the factor indicating the change of the estimated probability of innovation. ∗∗∗p < 0.001, ∗∗ p < 0.01, ∗p < 0.05.

and suppliers have no significant influence but interactions with customers positively and significantly affect innovation of the firm. Model 3 brought not only firm-level interactions but also interactions with URIs into the equation. It turns out that internal R&D intensity and external interactions with customers are the only significantly positive factors affecting the probability of firm innovation. This probability was 2.44 times higher for firms with frequent contacts with their customers than those without. And more strikingly, the probability of innovation would be increased by 4.4 times if the R&D intensity increased one percent. The results of modeling verify Hypothesis 2 that interaction with customers out-weighs other linkages as the most significant source of knowledge affecting innovation. Given that our sample is not very large, we replace our independent variables “frequency of interactions” with “importance of interactions” to conduct a robustness check. As shown in Table 6, results of the robustness check are highly consistent with our original ones. It shows that our findings are quite robust. There exists an interesting inconsistency between the results of regression modeling and that of the t-test presented earlier in Table 3. While the t-test shows that interactions with URIs set the innovators apart from non-innovators, the results of the regression modeling demonstrate an insignificant impact of the frequency of interactions with URIs upon firm innovation. Information obtained from our interviews provides interesting insights into the dynamics of firms’ interactions with URIs in the process of innovation. Whether or not interactions with URIs can help obtain useful knowledge for firms to engage in innovation pretty much depends upon the pre-existing social relationship between the software firms and the URIs. If the founders, CEOs or top managers of the software firms are alumni of the URIs or the firms are

5. Conclusion and discussion The prevailing intellectual trend of technological innovation takes geography or multi-scalar knowledge interactions as the fundamental conditions that effectively shape knowledge production and transfer for technological innovation. Yet it remains controversial and vague on how location or distance has factored in the process of firm innovation, and what specific external sources of knowledge are important in many developing economies. This study investigates the practices of external knowledge sourcing and innovation in China focusing on the software industry which has not been fully explored due to data deficiency. A systematic analysis of the data obtained from our firm-level questionnaire survey and interviews has generated several interesting findings. First, customers play a most significant role in the process of firm innovation. Because innovation activities of the software industry involved more applied knowledge to be derived directly from the market in China, and customers are able to share their after-use experiences and sometimes even help with solving technical problems, software firms value their customers the most in the process of innovation. Geographical proximity, however, does not facilitate innovation or knowledge exchange between firms and their customers, since ICT adoption and the advancement of transport allow the firms to reach their customers easily, quickly and cheaply without necessarily being close to them. Second, interactions with and knowledge sourcing from suppliers 67

Habitat International 71 (2018) 60–69

C.C. Wang, G.C.S. Lin

make no significant difference to firm innovation. Our interview notes reveal that most of the suppliers are platform providers or equipment suppliers who only provide standardized products and hence spill very little valuable knowledge to software firms in Hangzhou. Also, some of the current suppliers may turn into their rivals later on and therefore software firms are reluctant to interact with suppliers for knowledge sourcing. In our case, geographical proximity does not accelerate knowledge exchange between firms and their suppliers either, not only because knowledge from suppliers is not as valuable as that from customers, but also because a half of firms select their suppliers outside of Hangzhou without taking proximity into account and most of firms are reluctant to seek knowledge exchange with local suppliers due to the mutual distrust in a turbulent market environment. Third, knowledge from rivals is not significant to firm innovation. Most of the software firms in Hangzhou are engaged in the development of software application. This kind of activities is characterized by a relatively less complicated technology and a low entry barrier for firms. Due to the imperfect institutional environment, software firms do not trust rivals and they concern more about knowledge leakage than knowledge input from rivals. This risk increases when the rivals located nearby. Geographical proximity hence fails to forge a mutual relationship between firms and local rivals and a majority of firms do not interact with local rivals. Fourth, URIs are identified as highly significant to innovation by those software firms who have pre-existing social ties with the URIs. Although technologies from URIs are usually hard to be transformed and applied directly into the market, relational proximity between the firms and URIs can remove this barrier through frankly, deeply and constantly interacting and discussing with each other and eventually turn the invention into innovation. In addition, the pre-existing social ties allow the firms to easily recruit some highly qualified students from the URIs which is crucial for software development. Geographical proximity, together with relational proximity facilitates knowledge exchange and innovation. Although our study is based on one single city-region, the findings can be leveraged to explain knowledge sourcing and innovation in other emerging economies where the institutional environment is imperfect and marketization and globalization intensively interwove to welcome a rapid response to market demand. Under this circumstance, firms do not trust their rivals and suppliers in the process of innovation but value the interaction with customers. In consideration of the specific nature of software industry, we believe that our research findings are also relevant to other regions where application software firms are clustering. Empirical analyses made by Weterings and Boschma (2009) on Dutch software industry and Segelod and Jordan (2004) on Swedish software companies have illustrated that customers outweigh other actors in the process of innovation. However, this study has its limitation on the possibly varied relationship between software firms and URIs across different contexts. For instance, Beijing harbors more URIs than Hangzhou and therefore URIs in Beijing may play a larger role in the process of innovation of software firm than those in Hangzhou. Since our study focuses on software industry, although its findings can provide reference to other market-oriented and knowledge-intensive industries, it still requires more empirical studies to further warrant our results, due to the great variety of industrial characteristics. For instance, geographical proximity to suppliers and customers may still be important for those industries which are sensitive to time and require higher transportation costs. Our study has the following theoretical contribution and implications. First, it unfolds an intriguing trajectory that is a significant supplement to the conventional wisdom of distance decay and geographical proximity, by demonstrating that functioning of geography is varied by the attribute of different agents in the context of developing countries. Geographical proximity is found to play a negligible role in knowledge exchange with rivals, suppliers and customers. However, firms enjoyed a more effective and fruitful collaboration in the process

of innovation with local URIs than those located far away. It echoes with the research finding made by Tödtling, Lehner, and Kaufmann (2009) that geographical proximity matters less for collaborations with customers/suppliers than URIs. It also implies that the effect of geographical proximity is dependent upon cognitive and relational proximity (Boschma, 2005). On one hand, the relatively larger cognitive distance between firms and URIs requests geographical proximity to facilitate knowledge exchange (Kaufmann & Tödtling, 2001). On the other hand, relational proximity between firms and URIs directs firms to find the most appropriate and best team of local URIs to help innovation. By contrast, the cognitive proximity between firms and suppliers, customers and rivals does not require a geographical proximity to foster exchange of information and knowledge (Grillitsch et al., 2015). In a broader and conceptual perspective, it suggests that the role of location and geography in the process of external knowledge sourcing and firm innovation cannot be over-simplified and taken for granted. For example, a previous study has pointed out that geographical proximity fails to forge a mutual trust relationship among firms in China (Wang & Lin, 2008). This study confirms this finding and goes further by revealing under what conditions geographical proximity might be able to contribute to knowledge sourcing and innovation. The economic globalization brings a hybrid of firm structure to a cluster in which firms show significant differences in social backgrounds, modes of thinking, power relations as well as cognitive and management styles (Wang, Zhang, & Yeh, 2016; Yuan, Wei, & Chen, 2014). Our study has suggested that the devil is in the details: it is the attributes of the firms and external knowledge providers, the social network that these firms are embedded, rather than simply geography per se that determine whether or not external knowledge sourcing can enhance the innovative performance of the firms. Finally, this study has highlighted the social dimension of knowledge flows between the firms and the URIs. The existing literature tends to treat the firms as a homogenous entity without looking carefully into the social background of firm founders and their relationship with the URIs. A special social, cultural and relational embeddedness of the firms established by alumni or university spin-offs in the URIs allows them to tap into the labor and technology resources of the latter that is unknown to others. Several of our interviewees complained that the collaboration with URIs is futile simply because they did not know to whom they could resort, which research team should be the best and how to remove the barriers to collaboration and communications. It appears that “knowledge clearly involves social and instituted processes in its formation …. Thus, the mutual exchange of knowledge and shared learning, noted above, means that knowledge is intrinsically a socially constructed process” (Howells, 2012, p. 1010). Future studies should pay greater attention to the socially constructed processes of knowledge flows so as to better understand the sophisticated mechanism of external knowledge sourcing and firm innovation. This study also has managerial implications for software firms. First, knowledge from URIs is valuable and can be successfully transformed to support firm innovation. However, the collaboration between software firms and URIs should be bridged and initiated by people who have preexisting ties with and still maintains a good connection with the URIs to avoid the risk of failure. Second, picky customers make the innovative firms. Knowledge from customers is most significant for firm innovation and therefore firms should keep a frequent and close relationship with their customers no matter where the customers are located.

Acknowledgement The work described in this article has been sponsored by the grants from National Natural Science Foundation of China (No. 41101112, 41471101). We would like to express our gratitude to Miss Zihang Zhu, for her assistance in data collection and processing. 68

Habitat International 71 (2018) 60–69

C.C. Wang, G.C.S. Lin

References

Manniche, J. (2012). Combinatorial knowledge dynamics: On the usefulness of the differentiated knowledge bases model. European Planning Studies, 20(11), 1823–1841. MIIT (Ministry of Industry and Information Technology of China) (2014). China's electronics and information industry statistical yearbook 2013 (software). Beijing: China Statistics Press Electronic version. Miotti, L., & Sachwald, F. (2003). Co-operative R&D: Why and with whom?: an integrated framework of analysis. Research Policy, 32(8), 1481–1499. Munari, F., Sobrero, M., & Malipiero, A. (2012). Absorptive capacity and localized spillovers: Focal firms as technological gatekeepers in industrial districts. Industrial and Corporate Change, 21(2), 429–462. Nieto, M. J., & Santamaría, L. (2007). The importance of diverse collaborative networks for the novelty of product innovation. Technovation, 27(6–7), 367–377. Romijn, H., & Albaladejo, M. (2002). Determinants of innovation capability in small electronics and software firms in southeast England. Research Policy, 31(7), 1053–1067. Segelod, E., & Jordan, G. (2004). The use and importance of external sources of knowledge in the software development process. R&D Management, 34(3), 239–252. Sidhu, J. S., Commandeur, H. R., & Volberda, H. W. (2007). The multifaceted nature of exploration and exploitation: Value of supply, demand, and spatial search for innovation. Organization Science, 18(1), 20–38. Strambach, S., & Klement, B. (2012). Cumulative and combinatorial micro-dynamics of knowledge: The role of space and place in knowledge integration. European Planning Studies, 20(11), 1843–1866. Sun, Y. F., & Zhou, Y. (2011). Innovation and inter-firm technological Networking: Evidence from China's information communication technology industry. Erdkunde, 65(1), 55–70. Tether, B. S. (2002). Who co-operates for innovation, and why: An empirical analysis. Research Policy, 31(6), 947–967. Tether, B. S., & Tajar, A. (2008). Beyond industry-university links: Sourcing knowledge for innovation from consultants, private research organisations and the public science-base. Research Policy, 37(6–7), 1079–1095. Tödtling, F., Grillitsch, M., & Höglinger (2012). Knowledge sourcing and innovation in Austrian ICT companies– How does geography matter? Industry and Innovation, 19(4), 327–348. Tödtling, F., Lehner, P., & Kaufmann, A. (2009). Do different types of innovation rely on specific kinds of knowledge interactions? Technovation, 29, 59–71. Tödtling, F., Lehner, P., & Trippl, M. (2006). Innovation in knowledge intensive industries: The nature and geography of knowledge links. European Planning Studies, 14(8), 1035–1058. Tomlinson, P. R. (2010). Co-operative ties and innovation: Some new evidence for UK manufacturing. Research Policy, 39(6), 762–775. Trippl, M., Tödtling, F., & Lengauer, L. (2009). Knowledge sourcing beyond buzz and Pipelines: Evidence from the vienna software sector. Economic Geography, 85(4), 443–462. Tsai, K.-H. (2009). Collaborative networks and product innovation performance: Toward a contingency perspective. Research Policy, 38(5), 765–778. Wang, C. C., & Lin, C. S. G. (2008). The growth and spatial distribution of China's ICT industry: New geography of clustering and innovation. Issues & Studies, 44(2), 145–192. Wang, J., Zhang, X., & Yeh, A. G. O. (2016). Spatial proximity and location dynamics of knowledge-intensive business service in the Pearl River Delta, China. Habitat International, 53, 390–402. Wei, Y. H. D., Li, W. M., & Wang, C. B. (2007). Restructuring industrial districts, scaling up regional development: A study of the Wenzhou model, China. Economic Geography, 83(4), 421–444. Weterings, A., & Boschma, R. (2009). Does spatial proximity to customers matter for innovative performance? Evidence from the Dutch software sector. Research Policy, 38, 746–755. Wu, W., & Zhou, Y. (2012). The third mission stalled? Universities in China's technological progress. The Journal of Technology Transfer, 37(6), 812–827. Yuan, F., Wei, Y. D., & Chen, W. (2014). Economic transition, industrial location and corporate networks: Remaking the Sunan model in Wuxi city, China. Habitat International, 42, 58–68. Zeng, S. X., Xie, X. M., & Tam, C. M. (2010). Relationship between cooperation networks and innovation performance of SMEs. Technovation, 30(3), 181–194. Zhou, Y., & Tong, X. (2003). An innovative region in China: Interaction between multinational corporations and local firms in a high-tech cluster in Beijing. Economic Geography, 79(2), 129–152. ZJSB (Zhejiang Statistical Bureau) (2014). Zhejiang statistical yearbook 2014. Beijing: China Statistics Press Electronic version.

Asheim, B. T., Boschma, R., & Cooke, P. (2011). Constructing regional Advantage: Platform policies based on related variety and differentiated knowledge bases. Regional Studies, 45(7), 893–904. Audretsch, D. B., Lehmann, E. E., & Warning, S. (2005). University spillovers and new firm location. Research Policy, 34(7), 1113–1122. Bathelt, H., Malmberg, A., & Maskell, P. (2004). Clusters and knowledge: Local buzz, global pipelines and the process of knowledge creation. Progress in Human Geography, 28(1), 31–56. Bettencourt, L. A., Ostrom, A. L., Brown, S. W., & Roundtree, R. I. (2002). Client coproduction in knowledge-intensive business services. California Management Review, 44(4), 100–128. Boschma, R. A. (2005). Proximity of innovation: A critical assessment. Regional Studies, 39(1), 61–74. Caloghirou, Y., Kastelli, I., & Tsakanikas, A. (2004). Internal capabilities and external knowledge sources: Complements or substitutes for innovative performance? Technovation, 24(1), 29–39. Chung, S., & Kim, G. M. (2003). Performance effects of partnership between manufacturers and suppliers for new product development: The supplier's standpoint. Research Policy, 32(4), 587–603. Cooke, P., Boekholt, F., & Tödtling (Eds.). (2000). The governance of innovation in Europe. London: Pinter. CSSB (China State Statistical Bureau) (2014). China statistical yearbook 2014. Beijing: China Statistics Press Electronic version. Edison, H., Ali, N., & Torkar, R. (2013). Towards innovation measurement in the software industry. The Journal of Systems and Software, 86, 1390–1407. Freel, M. (2003). Sectoral patterns of small firm innovation, networking and proximity. Research Policy, 32(5), 751–770. Füller, J., & Matzler, K. (2007). Virtual product experience and customer participation– A chance for customer-centred, really new products. Technovation, 27(6), 378–387. Fu, W., Revilla Diez, J., & Schiller, D. (2013). Interactive learning, informal networks and innovaiton: Evidence from electronics firm survey in the Pearl River Delta, China. Research Policy, 42(3), 635–642. Grillitsch, M., Tödtling, F., & Höglinger, C. (2015). Variety in knowledge sourcing, geography and innovation: Evidence from the ICT sector in Austria. Papers in Regional Science, 94(1), 25–43. Grillitsch, M., & Tripple, M. (2014). Combining knowledge from different sources, channels and geographical scales. European Planning Studies, 22(11), 2305–2325. Grimaldi, R., & Torrisi, S. (2001). Codified-tacit and general-specific knowledge in the division of labour among firms: A study of the software industry. Research Policy, 30(9), 1425–1442. Halkier, H., James, L., Dahlström, M., & Manniche, J. (2012). Knowledge dynamics, regions and public policy. European Planning Studies, 20(11), 1759–1766. Howells, J. (2012). The geography of knowledge: Never so close but never so far apart. Journal of Economic Geography, 12(5), 1003–1020. Hu, F. Z. Y., & Lin, G. C. S. (2013). Placing the transformation of state-owned enterprises in North-east China: The state, region and firm in a transitional economy. Regional Studies, 47(4), 563–579. Isaksen, A. (2004). Knowledge-based clusters and urban location: The clustering of software consultancy in Oslo. Urban Studies, 41(5/6), 1157–1174. Kaufmann, A., & Tödtling, F. (2001). Science–industry interaction in the process of innovation: The importance of boundary-crossing between systems. Research Policy, 30, 791–804. Ledwith, A., & Coughlan, P. (2005). Splendid Isolation: Does networking really increase new product success? Creativity and Innovation Management, 14(4), 366–373. Li, P.-F. (2014). Horizontal versus vertical Learning: Divergence and diversification of lead firms in the Hangji toothbrush cluster, China. Regional Studies, 48(7), 1227–1241. Liefner, I., Hennemann, S., & Xin, L. (2006). Cooperation in the innovation process in developing countries: Empirical evidence from Zhongguancun, Beijing. Environment and Planning A, 38(1), 111–130. Lin, G. C. S., Wang, C. C., Zhou, Y., Sun, Y., & Wei, Y. H. D. (2011). Placing technological innovation in globalizing China: Production linkage, knowledge exchange, and innovative performance of the ICT industry in a developing economy. Urban Studies, 48(14), 2999–3018. Lin, G. C. S., Yang, F. F., & Hu, F. Z. H. (2012). The new geography of information and consulting services in China: Comparing Beijing and Guangzhou. Habitat International, 36, 481–492. Liu, H., & Jiang, Y. (2001). Technology transfer from higher education institutions to industry in China: Nature and implications. Technovation, 21(3), 175–188.

69