Gatekeepers, Knowledge Brokers

31
GATEKEEPERS, KNOWLEDGE BROKERS AND INTER-FIRM KNOWLEDGE TRANSFER IN BEIJINGS ZHONGGUANCUN SCIENCE PARK MATIAS RAMIREZ * SPRU- Science and Technology Policy Research University of Sussex, The Freeman Centre, Brighton, BN1 9QE, UK [email protected] PETER DICKENSON Department of Manufacturing, Craneld University Craneld, MK43 0AL, UK r.p.dickenson@craneld.ac.uk An important part of industrial policy in China has been directed towards improving the degree and effectiveness of knowledge transfer between key rms in Chinas innovation system. Amongst these policies, the creation of regional science parks that encourage labour mobility and inter-rm collaboration on innovation projects have been central. Learning through inter-rm knowledge transfer focuses the attention on at least two key factors, improving absorptive capability (Cohen and Levinthal, 1990), which relies on the development of specialised skills in the rm and the establishment of inter-organisational networks through which knowledge is transferred. This paper contributes to this analysis through a detailed study of the relationship between learning and knowledge transfer of knowledge workers working on innovation projects in Chinese ICT companies located in Beijings Zhongguancun (ZGC) high- technology park. A major advantage of analysing knowledge transfer through the activities of R&D employees is that it highlights the process by which specic competencies and network relations are built. A skills prole of R&D employees is developed that, amongst other features, includes three different networks Chinese knowledge workers use to access and share knowledge: formal organisational networks, personal networks and scanning networks. Empirical data based on two unique surveys in China of senior R&D managers and R&D employees was collected and analysed. This suggests that a skills prole combining knowledge within and outside of the company and scanning activity positively / Corresponding author. International Journal of Innovation Management Vol. 14, No. 1 (Feb. 2010) pp. 93122 © Imperial College Press DOI: 10.1142/S1363919610002568 93

Transcript of Gatekeepers, Knowledge Brokers

Page 1: Gatekeepers, Knowledge Brokers

GATEKEEPERS, KNOWLEDGE BROKERS ANDINTER-FIRM KNOWLEDGE TRANSFER IN BEIJING’S

ZHONGGUANCUN SCIENCE PARK

MATIAS RAMIREZ*

SPRU- Science and Technology Policy ResearchUniversity of Sussex, The Freeman Centre, Brighton, BN1 9QE, UK

[email protected]

PETER DICKENSON

Department of Manufacturing, Cranfield UniversityCranfield, MK43 0AL, UK

[email protected]

An important part of industrial policy in China has been directed towards improving thedegree and effectiveness of knowledge transfer between key firms in China’s innovationsystem. Amongst these policies, the creation of regional science parks that encouragelabour mobility and inter-firm collaboration on innovation projects have been central.Learning through inter-firm knowledge transfer focuses the attention on at least two keyfactors, improving absorptive capability (Cohen and Levinthal, 1990), which relies on thedevelopment of specialised skills in the firm and the establishment of inter-organisationalnetworks through which knowledge is transferred.

This paper contributes to this analysis through a detailed study of the relationshipbetween learning and knowledge transfer of knowledge workers working on innovationprojects in Chinese ICT companies located in Beijing’s Zhongguancun (ZGC) high-technology park. A major advantage of analysing knowledge transfer through the activitiesof R&D employees is that it highlights the process by which specific competencies andnetwork relations are built. A skills profile of R&D employees is developed that, amongstother features, includes three different networks Chinese knowledge workers use to accessand share knowledge: formal organisational networks, personal networks and scanningnetworks. Empirical data based on two unique surveys in China of senior R&D managersand R&D employees was collected and analysed. This suggests that a skills profilecombining knowledge within and outside of the company and scanning activity positively

⁄Corresponding author.

International Journal of Innovation ManagementVol. 14, No. 1 (Feb. 2010) pp. 93–122© Imperial College PressDOI: 10.1142/S1363919610002568

93

Page 2: Gatekeepers, Knowledge Brokers

impact both the innovation projects and the labour market position of the knowledgeworkers. Policy recommendations in terms of training and development in R&D follow.

Keywords: Gatekeepers; clusters; China innovation system; knowledge brokers; knowl-edge transfer.

Introduction

A substantial body of literature emanating from innovation studies has emphasisedthat, with the development of the ICT paradigm, the organising principles in high-technology industries have evolved to take into account greater degrees ofcollaboration and knowledge transfer. This process has emerged because of themore diverse and complex nature of knowledge associated with the developmentof new technologies and the greater range of actors involved in the generation ofnew knowledge (Gibbons et al., 1994; Castells, 1996; Chesbrough, 2006).

Studies that focus on the specific role that scientists and engineers play inknowledge transfer have tended to emphasise two types of studies. Firstly, there arethose that stress the importance of mobility and migration as a form of “embodied”knowledge flow (Zimmerman, 1995; Vandamme, 2000; Crescenzi et al., 2007).This has some important implications for patterns of skill and career formation, asexpressed in the emergence of “boundaryless careers” in high-technology clusters(Arthur and Rousseau, 1996; Saxenian, 1996). A second related approach focuseson the networks employees form during their careers in different firms. For example,Casper and Murray (2005) compare the career paths of scientists in biotech clustersof Cambridge in the UK and San Diego in the US, and reveal that US scientists buildup much higher levels of social capital through their careers because of the higherlevel of mobility. The significance of this approach is that knowledge worker net-works become an alternative medium through which knowledge is transferred andcan potentially be an effective mechanism of knowledge transfer.

A further point emphasised is that, over and above the different conduits thatexist for knowledge transfer, the ability of knowledge workers to act as effectiveintermediaries of knowledge for the firm will be dependent on the development ofa specific skills profile associated with the ability to scan, translate and evaluate theusefulness of external information and to apply this to a different organisationalenvironment (Macdonald and Williams, 1992; Leonard-Barton, 1995). Creatingeffective bridges between the worlds inside and outside the firm is, therefore, basedon developing common understandings between different groups of practitionersas a prior condition for learning. The paper builds on the above discussion byarguing that effective processes of knowledge transfer involve the development ofa specific division of labour that combines experiential learning of knowledge

94 M. Ramirez & P. Dickenson

Page 3: Gatekeepers, Knowledge Brokers

workers inside and outside the firm. It goes one step further by identifying a seriesof networks and channels of communication Chinese knowledge workers willengage with and tests whether participation in these positively impacts bothinnovation projects and the labour market position of knowledge workers them-selves.

This study is highly relevant for China’s emerging ICT research system, thelabour markets of which have changed rapidly in the course of the last decade, buthave not been studied in great detail. The methodology of the study is built upon aunique survey undertaken in 2006 of 381 R&D employees working on ICTinnovation projects in 71 Chinese high-technology firms located in Beijing’sZhongguancun (ZGC) high-technology park, China’s largest high-technologyscience park.

The paper is structured as follows. Firstly, the broader literatures tracing themechanisms by which effective networks of knowledge transfer are built at thefirm level are discussed. The next section discusses how R&D employees act asagents of knowledge transfer in innovation. Emphasis is placed on the develop-ment of specific competencies and access to networks. A series of hypotheses aregenerated that cover skills and network activities of R&D workers. Next, theevolution of scientific labour market structures in China is discussed, emphasisingtheir rapid transformation since the years of the planned economy and thehigh rates of turnover currently experienced in the high-technology clusters.The empirical section analyses the data gathered on a recent project on knowledgeworkers in Beijing and discusses the findings in the light of the hypotheses.The final section reflects on the implications of the analysis and findings forknowledge transfer and the management of scientific labour markets in China.

Capacity Building Through Internal andExternal Knowledge Transfer

In their comparative analysis of clusters, Bresnahan et al. (2001) suggest that anew feature of high-technology clusters involves the development of “neweconomy” firm behaviour to complement traditional business activity associatedwith building economies of scale. Activities such as entrepreneurship, economiesof scale at the level of the region and external effects, i.e. learning from interactionwith other firms or organisations are, it is argued, essential to innovation intechnologically dynamic environments. The emphasis on technology transfer hasbeen taken to heart by policy makers in China through the establishment ofscience parks in every region that, it is hoped, will enhance synergies betweenknowledge-intensive companies (Walcott, 2003). The financial inducements for

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 95

Page 4: Gatekeepers, Knowledge Brokers

high-technology companies to locate in science parks in terms of tax breaks areconsiderable and this has attracted significant numbers of enterprises. Threequarters of all high-technology activity in China takes place in its science parks.Moreover, it is estimated that more that 4 million people are employed within thescience parks (China Science and Technology Statistics Databook, 2004) while inthe year 2000, science parks accounted for more than 25% of China’s value-addedoutput (Cao, 2002).

There is nevertheless some debate regarding the effectiveness of these policies(Katila and Ahuja, 2002), particularly in Asia. While Massey et al. (1992) ques-tion the existence of networking and point instead to the opportunity real estatedevelopers have to benefit from booming land revenues, Walcott (2003) suggeststhat in developing countries such as China, underlying institutions, such as gov-ernment support, intellectual property protection and solvent financial institutionsare required to derive benefits from industrial clustering. Similarly, for networkingto be effective, mediating actors and institutions for the rules of engagement needto exist. Therefore, given the dominant position they have achieved in China,understanding how inter-firm knowledge transfer within science parks haveinfluenced the development of the ICT sector is clearly important.

The enthusiasm for new economy type activity invites a brief recount of howthe building of organisational capabilities through inter-firm knowledge transferhas been conceptualised. This begins by grouping a body of literature aroundresource-based explanations of competence building that suggest learning organ-isations build capabilities based on hard-to-copy and largely tacit internal rou-tines (Penrose, 1959; Nelson and Winter, 1982; Barney, 1991; Teece and Pisano,1994). The original emphasis of this work was on explaining differentialperformance between organisations through the development of internal compe-tencies. Cohen and Levinthal’s (1990) highly influential work on “absorptivecapacity” extended the concept of internal competence to argue that learning fromexternal sources will depend on the firm’s prior ability to assimilate new knowl-edge. Absorptive capacity, therefore, is a specific resource that has to be generatedand built through strategic intent, for example, by specifically targeting humanresources that have the experience of working outside the firm. A number of othercontributions underline the importance of coupling of internal and externalcapabilities. Rigby and Zook (2002), for example, argue that combining internaland external information sourcing through “open market” innovation is a criticalsource of comparative advantage, while Arora and Gambardella (1994) noted thatinternal know-how was required to be able to screen external projects.

A related but more recent set of studies highlight “knowledge ambiguity” asone of the most important predictors of the effectiveness of organisationalknowledge transfer (Levin and Cross, 2004; Szulanski et al., 2004). Knowledge

96 M. Ramirez & P. Dickenson

Page 5: Gatekeepers, Knowledge Brokers

ambiguity refers to the uncertainty of what the underlying knowledge componentsand sources are (Reed and DeFilippi, 1990). Although high ambiguity clearly actsas a barrier to knowledge transfer, van Wijk et al. (2008) also suggests that highlevels also contribute to protecting knowledge from being imitated by rivals.Therefore, firms will often search for a balance between high and low levels ofambiguity.

A third set of studies emphasises the types of networks firms engage in, withparticular emphasis on differences between strong and weak ties (Granovetter,1973). Strong ties implies regular interaction and a number of studies haveemphasised their importance for the building of trust and long-term collaborativecommitments. van Wijk et al. (2008), for example, in their meta-analytical reviewof the contextual factors that positively contribute to innovation performance,found that the degree of relational capital firms hold through tie strength and trustwas highly significant. Weak ties on the other hand imply access to disparateheterogeneous knowledge through a range of networks. They have been shown tobe invaluable for introducing novelty (Burt, 2005), and have also been emphasisedin Rodan and Galunic’s (2004) study of managers and Sammarra and Biggiero’s(2008) study of multiple networks.

Finally, it is possible to identify an approach that focuses on the governanceof networks as a mechanism of coordination for the transfer of knowledge.Unlike resource-based approaches that emphasise the organisation as the relevantunit of analysis, network-based theories suggest that social phenomena does notexist independently, but are in fact brought into existence by the relationshipsthey establish (Thompson, 2003; Castells, 1996). It follows that an understandingof the governance of networks will provide important information as to thenature of the collaboration. If attention is focussed specifically on the governanceof social networks, the relationships that are established between the actors thatconform these will be strongly shaped by the rules, conventions and standardsthat are formed within the network. In this sense, the governance of inter-firmnetworks can vary a great deal. On the one hand, it is possible to conceive ofhighly formalised contractual inter-firm alliances, such as strategic or supply-chain type collaborations, the governance of which will be defined largely byhierarchical control (Powell and Grodal, 2005) and where transfer of knowledgeis likely to be regulated to a high degree by formal codified arrangements overthe scope, aims and outputs of collaboration. On the other hand, there existinformal quasi-horizontal networks of practitioners, such as the open sourcenetworks that flowed from development of the Linux software, the existence ofwhich hinges around norms for problem-solving that exist parallel to formalorganisational structures (DeFillippi et al., 2006). Here governance may bedominated by concepts of reciprocity, where favours are provided on the

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 97

Page 6: Gatekeepers, Knowledge Brokers

expectation of future reciprocal “gifts” of equal or higher value. Alternatively, itmay be wrong to conceive gifts as an exchange at all in some networks, butinstead a service that is provided around values of cooperation and goodwillcommon to members of the network (Thompson, 2003). Although studiesrelating informal networks to innovation are less common in innovation research,Powell and Grodal (2005) note that informal relations may actually undergridformal ties. Indeed, a body of work around action-based perspectives on learning,such as the communities of practice literature (Wenger, 1998) that emphasises theconstruction of bottom-up networks, have increasingly found their way intostudies of innovation.

Combining Capabilities and Networksin Knowledge Worker Skills

The previous section outlined how the effectiveness of knowledge transfer ininnovation projects is mediated strongly by a number of factors, amongst whichfirm-specific capabilities and experience of engagement and participation inexternal networks are prominent. Ambiguity of knowledge was also highlighted,as was the degree of formality of the network. The following section discusses aseries of hypotheses related to the manner in which the skills profile and work-related activities of knowledge workers operationally builds the firm-specificcompetencies and network links discussed above.

A starting point would be to point to a literature that has highlighted a keydivision of labour that will underpin the organisation’s ability to interact with andabsorb external knowledge within the organisation. This involves individualsworking at the interface between the firm and the external environment who havealternately been referred to as gatekeepers and boundary spanners and holdspecialised skills in monitoring, translating, assimilating and recognising thevalue of external information (Allen, 1977; Macdonald and Williams, 1992;Leonard-Barton, 1995; Bouty, 2000; Haunschildt and Schewe, 2000). The deve-lopment of gatekeeping skills is particularly important where the cognitive dis-tance between an organisation’s expertise and that sought from the externalenvironment is large (Cohen and Levinthal, 1990). It can be assumed that the keyskills required to integrate knowledge inside and outside the firm are developedthrough a history of inter-firm mobility and broad work experience. This willwiden the individual’s perspectives, broaden the knowledge of best-practices andbuild a wide range of external contacts. At the same time, experience andknowledge of the firm itself provides the tacit knowledge and authority to takethrough and implement new ideas within the organisation. It can, therefore, be

98 M. Ramirez & P. Dickenson

Page 7: Gatekeepers, Knowledge Brokers

argued that the existence of effective gatekeepers will underpin essential inte-grative capabilities for successful knowledge transfer. It is, therefore, proposedthat innovation projects will be more successful to the extent that knowledgeworkers within the project combine high levels of experience and knowledge ofthe organisation with practical experience of participating in wider industrypractitioner networks:

Hypothesis 1Innovation projects will be more successful when the skill profile of key knowledgeworkers on these projects combines experience and skills from inside and outsideof the firm.

A number of studies have indeed referred to the above proposition. Forexample, project SAPPHO (Freeman and Soete, 1997), one of the first majorstudies aimed at identifying successful firm-level practices for innovation, foundthat the business innovator (whether in the form of an entrepreneur, gatekeeper ormanager) played a critical role through their understanding of how to get thingsdone in the firm, and through an understanding of the marketing needs of thetechnology. The “coupling” concept is extended in Hypothesis 1 to encompassexperience of lead managers outside the company. This is backed up in morerecent studies, including Powell (1998) who promoted the concept of “networkmanagers,” and Burt’s “knowledge brokers” (2002), acting as intrapreneurswithin the firm.

A second key area of analysis involves differentiating the external networks inwhich knowledge workers participate and learn from. Three such knowledgeworker networks are identified in this paper. These distinguish both formal andinformal forms of governance, but also, perhaps more importantly, different formsof governance.

1. It is possible to identify a learning activity associated with keeping theorganisation abreast of best-practice, lead thinking and on the technologicalfrontier. Such work, that can be labelled as searching and scanning, has beenanalysed through different lenses. For example, Granovetter (1973) empha-sised the importance of developing “bridging agents”, as means to connect to awide range of ‘distant communities,’ that may be beneficial to gaining access tobroader sources of information. In the context of a regional concentration ofeconomic activity, scanning may involve attendance at conferences and talksand informal meetings, while general visits to Internet chat rooms arebecoming a common way of scanning material. The characteristic of scanningis the relatively low cost of engagement and low reciprocal commitment,however, engagement through a “narrower bandwidth” (Powell and Grodal,

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 99

Page 8: Gatekeepers, Knowledge Brokers

2005) can also be argued to be a crucial first step to building stronger inter-firmrelationships.

2. A second type of networking activity groups together individuals that play aprominent role in bridging the cognitive gaps that can exist between firmscollaborating in formal innovation projects. As discussed earlier, inter-firmcollaboration requires firms to nurture “boundary spanners,” individuals thatare able to understand the world of the source and the world of the receiver(Leonard-Barton, 1995; Tushman and Scanlan, 1981) as well as disseminateknowledge. Employees involved in these activities are described as involved ininter-organisational problem-solving activities. In contrast to scanning andsearching, these employees will develop specialised knowledge of differentorganisations and the governance of this network will be characterised byformal control by organisations.

3. A third major source of activity that may impinge on inter-firm learning takesplace within networks built around personal relations of knowledge workers.Macdonald and Piekkari (2005) suggest that personal networks emerge becauseof the difficulty of getting the right information at the right time, hence marketfailure, and to overcome the limitations of contractual arrangements that restrictcollaboration. These networks also resemble Granovetter (1973) embeddednessof labour markets, where effectiveness of career mobility depends on socialnetworks that transcend the boundaries between economy and local social life. Inother words, professional networks minimise the costs of search and the costs ofswitching jobs. These networks are termed relational networks and provide theinformation signals needed to ensure success outside of internal labour markets.

From this discussion, a second hypothesis is developed concerning theimportance of the involvement of knowledge workers in different networks for thesuccessful outcome of innovation projects.

Hypothesis 2Innovation projects will benefit in performance from the direct involvement ofknowledge workers in external networks, including formal inter-organisationalnetworks, relational networks and scanning activities.

A further, more general point emphasised is that, from the point of view of theorganisation, encouraging individuals to develop skills that specialise in knowl-edge brokering, i.e. working at the frontier of the organisation rather than at itscore, requires the establishment of non-standard career routes and recruitmentstrategies that are open to new skills. Flexible HR practices will, therefore, providea key institutional support for the emergence of knowledge brokering activitiesthat underpin initiatives to transfer knowledge across organisations.

100 M. Ramirez & P. Dickenson

Page 9: Gatekeepers, Knowledge Brokers

Chinese Knowledge Worker Networks

Despite its growing importance for studies of innovation, a dearth of material hasbeen published in English language journals or books outlining the institutionalmake-up of scientific labour markets or human resource practices of Chinese sci-entists and engineers. Moreover, with the exception of a small number of studies,such as Saxenian (2005), that make indirect reference to knowledge transfer, fewempirical studies analysing the relationship between knowledge transfer andknowledge workers exist in English.

Yet, the benefits accruing from the so-called “new economy” are built on quiteimportant assumptions regarding work organisation and strategy in R&D envi-ronments. Emphasis on the development of the so-called “open labour markets,”it is argued, have been critical in encouraging diversity of skills and opportunitiesfor inter-firm mobility (Saxenian, 1996; Best, 2001). Further work in this areaemphasises the need for individuals to experiment with different career optionsthat derive from their participation in networks (Arthur and Rousseau, 1996).Similarly, the existence of intermediary labour market services, such as specialisedcontractors, online job search and staffing agencies to encourage effectivematching of supply and demand of skills will encourage new skill and careerpatterns from both firms and employees (ILO, 2001).

An important question, therefore, is posed regarding the ability of Chinese HRpractices to develop specialised skills amongst key knowledge workers. Thissection briefly traces the evolution of the scientific labour market in China sincethe introduction of the market reforms some twenty years ago and discusses somespecific characteristics that impinge upon the earlier discussion of networkrelations.

The emergence of what can be called a genuine labour market within thescience and technology system in China is closely tied to the growth of newtechnology enterprises, the arrival of high-technology multinational enterprisesand the reforms associated to the Chinese higher education (HE) system. China’sS&T structures were traditionally modelled closely on those of the Soviet Union inthe 1950s. The Soviet innovation model was of a simple linear nature, with somesimilarities to the early generation demand-pull model proposed for Westernsocieties (OECD, 1969; Holloway, 1982).

The S&T system was vertically integrated with very few, if any, formal hori-zontal linkages between production units on the one hand, and the R&D institutesresponsible for innovation and HE bodies with industrial ministry affiliation on theother. The role of such HE bodies was to provide a pool of highly specialisedknowledge workers for the production units and R&D institutes. Other organis-ations existed outside the industrial ministries that had S&T functions, such as the

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 101

Page 10: Gatekeepers, Knowledge Brokers

institutes of the various academies of science and universities under non-industrialministries, but they played little role, if any, in industrial innovation. Labourmobility was very limited and labour markets for knowledge workers did not exist,since jobs were allocated administratively, and were usually for life. In China, thestructures for innovation were extensive, with 800 industry branch R&D institutesexisting before the market-orientated changes began in the mid-1980s (Suttmeier,1997).

The 1985 policy initiative concerning the S&T sector was accompanied bysimilar measures concerning higher education, since the HE sector was seen as themain driver in Chinese high-tech development (Yin and White, 1994). The aims ofthe HE changes were to expand the autonomy, financial and otherwise, of insti-tutions, to strengthen links with production organisations and to develop a labourmarket for HE workers. The greater financial autonomy was to be achieved in twoways. Firstly, by charging students for their tuition fees and secondly, by devel-oping commercial structures to sell expertise or to invest in spin-off ventures ofvarious kinds. The main vehicles for the new approach were the new vocationaluniversities, which first appeared in the 1980s. All students paid fees, in one formor another, to attend these bodies and the curriculum was flexible and market-driven, with a strong vocational content (Fang, 1991).

Dramatic changes in the make-up of R&D labour markets appear in thegrowing mobility of employees. There is evidence from the business and pro-fessional press that Chinese high-tech firms are experiencing double digit labourturnover and difficulty in retaining qualified staff, leading to the need to concedehigh salary rises (Raatikainen, 2003; Leininger, 2004), indicating that there is anactive and mobile labour market in this sector. The new state HE policy discussedabove could have facilitated this, because as well as trying to promote labourmobility in universities it also had an aim of developing a labour market in theS&T sector. A key feature of the new policy was that the system where jobs forgraduates were allocated by the state was replaced in 1990 by the “two-wayselection” process. As the term implies, the prospective employer and employeeboth had a say in the transaction (Lewin and Xu, 1993).

The development of spin-off ventures may, however, have been the main driverin creating a high-tech labour market in China. The growth of this sector has beenexplosive, as revealed by the figures in the previous section that there are now23,000 NTEs in the ‘Torch’ programme. These firms were spun-off both byresearch institutes affiliated to the academy of sciences or former industrial branchministries and by HEIs. By the early 1990s, in some HEIs, most of the staff heldconcurrent jobs in spin-offs from their own institutions. For example, 80% of theteachers in Jiamusi Technology Institute were also working in its 12 affiliatedventures (Yin and White, 1994). The same study also reported that there was

102 M. Ramirez & P. Dickenson

Page 11: Gatekeepers, Knowledge Brokers

pressure on staff to leave the HE sector completely for business. The state activelyencouraged this trend through financial mechanisms, i.e. budget cuts on HEIs, andhas continued to do so.

Another driver of labour market mobility was the transformation of researchinstitutes into private companies, where the ability to hire/fire staff was seen as afactor influencing the change (Suttmeier, 1997). A study of one of the leading ICTfirms, Stone, noted that all the R&D team had acquired their skills in the stateR&D sector. The same paper also concluded that mobility of key researchers fromstate institutes to the new spin-offs was a factor in accounting for their success(Lu, 2001).

The combination of rapid growth of China’s high-technology sector, togetherwith a high level of concentration of employment growth in government-spon-sored science parks, has clearly created a highly dynamic labour market in terms ofmobility. What is less known is the degree to which a specialised division oflabour has emerged in high-technology areas that encourages cross-organisationlearning or, indeed, if the high levels of mobility are beneficial or detrimental tothe firm’s innovation projects. The next section of the paper, therefore, formallytests the hypotheses laid out earlier in the paper.

The Data

The empirical study is based on two parallel and linked surveys undertaken in2006–7, one of senior R&D project managers, the other of R&D employees inthe Zhongguancun (ZGC) science park in Beijing. Examples of questions from theoriginal surveys are included in appendices 1 and 2. It was decided to use theopportunity of a congress of ICT firms located in the ZGC park taking place inBeijing, to invite attending R&D managers to participate in the survey. Aside fromfacilitating access to firms, this non-probability method of identifying firmsintroduces some randomness into the sample, although some bias is possible, sincethose firms attending the congress may have been part of a particular network oforganisations. A quota system was used, whereby a target of firms fulfilling certaincriteria was established and collection of data was stopped once that target wasreached. The population of interest and criteria for participation in the survey wasdetermined specifically to be Chinese-owned ICT firms undertaking specificinnovation projects located in the ZGC area. Given the relative lack of knowledgeof the practices of Chinese-owned ICT firms, vis-a-vis multinationals in this area,it was felt that a study of Chinese firms would be more revealing. Moreover,because of budget limits of the projects and the consequent sample size, it was alsofelt that comparative analysis would be facilitated through a more homogenous

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 103

Page 12: Gatekeepers, Knowledge Brokers

ownership sample. It was also decided to choose firms located in one industry,ICT. The choice (ZGC) is based on the fact that there exists a highly skilledcommunity of both academic and non-academic practitioners, including 68universities and 213 scientific research institutes. Moreover, firms located in thepark share certain common characteristics. The Chinese state demands that allfirms located in the park derive at least half of all revenue from high-technologyproducts, and R&D expenditure must account for no less than 3% of total revenue.This facilitated the task of comparison of firms.

In total, 71 different innovation projects in separate organisations participated inthe survey. The protocol used to undertake the survey consisted of approaching asenior R&D manager (or if he/she was not available, a senior person in R&D) toparticipate in the research. If the answer was affirmative, the senior R&D managerwas then emailed the address of a website that they could log on to where thesurvey was available. The respondents were asked to nominate a major innovationproject in the company over the past three years and to answer questions in relationto this. The survey answers could be submitted on-line to a server that collected thedata. Senior R&D managers were also asked to nominate up to ten R&Demployees that worked in the above project, who in turn were asked to answerquestions in a different survey and to submit these on-line. This method ofchoosing knowledge workers is more likely to suffer from bias, since it is based onthe recommendations of the R&D manager. Some care, therefore, needs to takenwhen generalising these results to a wider population of ICT innovation in theZGC park.

Despite these limitations, it was felt that this research represents a novelapproach to measuring the building of capabilities across organisational frontiers.Finally, analysis of the limited missing data in both surveys showed a randomiseddistribution with respect to both respondents and survey items. Therefore, to makefull use of the results, mean substitution method was adopted for the multipleregression (MR) analysis. This technique is considered appropriate where missingvalues are randomly distributed across the data set (Cohen and Cohen, 1983). Itcan be regarded as a relatively conservative approach (Tabachnick and Fidell,1996), i.e. the probability of identifying statistically significant outcomes on aspurious basis is relatively low.

Innovation Performance and Skills:Senior R&D Manager Responses

Hypothesis 1 was evaluated through multiple regression analysis based on theR&D manager survey that drew responses from 71 leaders of R&D projects. In

104 M. Ramirez & P. Dickenson

Page 13: Gatekeepers, Knowledge Brokers

operational terms, the model is the following:

(Model 1)Y ¼ �0 þ �1OWNþ �2SIZEþ �3SQUALþ �4SEXPþ �5SINT

þ�6SAGEþ �7SRECþ ":

The dependent variable Y, was quantified by senior R&D manager averageratings of project success in terms of meeting deadlines, levels of product sales andtechnical capability of the final product in the innovation projects (see Table 1).

Hypothesis 1 specifies the importance of skills within and outside of theorganisation. The variables in model (1) therefore measure the importance linemanagers attach to different sets of skills and experience of knowledge workerswithin and outside of the organisation for the success of the innovation projectthey are working on.

Table 1. Survey item allocations to project success and networking practices constructs usinginnovation project as unit of analysis.

Construct Items Source Available n �

Project Success Success in meeting deadlines Project Leader 55 0.62Success in market share of product Project LeaderSuccess in technical capability of

productProject Leader

Inter-organisationalproblem-solving

Sharing knowledge with researchinstitutes

Team members 56 0.73

Sharing knowledge with founderbodies

Team members

Sharing knowledge with standard-setting bodies

Team members

Relational problem-solving

Sharing knowledge with formerclassmates

Team members 57 0.91

Sharing knowledge withcolleagues

Team members

Internal problem-solving

Sharing knowledge withmarketing dept

Team members 57 0.60

Sharing knowledge withother depts

Team members

General networking Attending conferences Team members 56 0.66External communication via

chat rooms, etc.Team members

Informal contact with externalacquaintances

Team members

Source: Survey of firms in ZGN (2006).

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 105

Page 14: Gatekeepers, Knowledge Brokers

SQUAL represents the importance senior R&D managers attach to R&Demployee academic qualifications to project success.1

SEXP refers to the importance senior R&D managers attach to prior experienceand familiarity of the firm by the R&D employee to project success.2

SINT is the importance senior R&D managers attach to prior experience ofworking on specific projects outside the firm by R&D employees.3

SAGE refers to the senior R&D manager’s perceptions of the reliance thecompany has on external recruitment agencies.4

SREC is the significance senior R&D managers place on newly appointed staffto work on the project for the success of the projects. Both SAGE and SREC areproxies for the “openness” of labour markets in ZGC.

Two additional variables were introduced. OWN represents ownership. Giventhe importance of state and non-state enterprises in China, OWN is a controlvariable that distinguishes between wholly and partly state-owned firms, such ascooperative enterprises and privately-owned companies. In practice, the vastmajority of the state-owned firms in the survey were cooperative enterprises. In theChinese context, cooperatives are enterprises where a significant but minoritystake is owned by the state. Management is relatively autonomous and answers tothe stakeholders, a majority of which are private individuals (Jing and Tylecote,2005). It is also usual in innovation studies to distinguish the size of the firm, as apotential factor differentiating innovation performance. SIZE therefore is a secondcontrol variable that differentiates firm size. Breakdown of the total sample interms of “size” and “ownership” is reported in Table 2.

1Manager survey question 21, see appendix.2Manager survey question 21, see appendix.3Manager survey question 21, see appendix.4Manager survey question 22, see appendix.

Table 2. Breakdown of sample in terms of size of firm and ownership.

Ownership Size Categories <100 500–499 >500 Total

Cooperative 29 (73%) 11 (65%) 4 (67%) 44 (70%)

Private 11 (27%) 6 (35%) 2 (33%) 19 (30%)

Total 40 (100%) 17 (100%) 6 (100%) 63a

aEight cases omitted from table due to missing values on size and/orownership.Source: Derived from survey of firms in ZGC, 2006.

106 M. Ramirez & P. Dickenson

Page 15: Gatekeepers, Knowledge Brokers

A multiple step modelling protocol was adopted with the control variables, sizeand ownership, forced into the solution as independent control variables at step 1.The five items concerning knowledge worker skills were then considered foraddition to themodel at subsequent steps with entry determined using an empiricallydriven stepwise strategy. The inter-correlations, means and standard deviations of allthe independent variables plus the dependent variable are reported in Table 3.

The results of the regression model are outlined in Table 4. This yields asubstantial overall fit (the F-test is significant at the 1% significance level). The R2

value of 0.42 suggests this model is a reasonably good explanatory model for thedependent variable. Final results identify “experience gained by R&D employeesof the firm”, “experience gained by R&D employees outside the firm” and“ownership” as making a significant contribution to prediction of project success.The first two variables have positive weights. The finalised model also suggeststhat higher project success levels tend to be reported by cooperative enterprises.No evidence is found for an association between project success and size oforganisations. The regression suggests that senior R&D managers look highlyfavourably upon high levels of experience within and outside of the firm, con-firming from the R&D manager’s perspective, hypothesis 1.

Networking and Project Performance

This section investigates hypothesis 2 and the importance of networking activityfor innovation projects. In this instance, knowledge worker responses are analysedto measure the impact of inter-firm collaboration across organisational frontiers.Creating the set of independent variables to represent the responses of eachknowledge worker team was a multi-step process. Firstly, means were calculatedacross responses of team members from each organisation on items relevant tonetworking and knowledge sharing. Values on these means were treated as validonly if they could be calculated on responses from at least two team members.These criteria were employed in order to optimise a balance between reliability ofmean ratings and the sample size available for analysis.

As a second step, exploratory factor analysis was undertaken to explore theconceptual similarity of variables. The choice of variables on the surveywas stronglyinfluenced by the classification of three communication channels discussed insection 3 of the paper. On this basis, itemswere allocated to sub-sets that were judgedto represent specific practices. Finally, allocation of items to sub-sets was checkedobjectively on the basis on their internal consistency. The constructs, items allocatedto themand their levels of internal consistency, as quantified byCronbach’s� values,are reported in Table 5.An observation of each of the variables that go up tomake the

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 107

Page 16: Gatekeepers, Knowledge Brokers

Table

3.Means,standard

deviations

andinter-correlations

offirm

levelsurvey

itemsandprojectsuccess(n

¼71).

Item

Mean

SD

Correlatio

ns

O’ship

Size

AcQual

IntExp

Ext

Exp

EA

Use

ImpRec

Success

Ownership(O

’ship)

a0.29

0.44

1.00

0.00

−0.21

−0.02

0.09

0.20

0.05

−0.33

**

Size

218

450

1.00

−0.15

−0.11

−0.01

0.18

−0.29

*−0.20

Academic

qualificatio

ns(A

cQual)

1.85

0.71

1.00

0.33

**−0.14

−0.19

−0.11

0.27

*

Exp

erience/kn

owledg

eof

firm

(Int

Exp

)2.44

0.55

1.00

0.16

−0.05

−0.20

0.45

**

Externalprojectexperience

(Ext

Exp

)2.47

0.53

1.00

−0.15

−0.07

0.34

**

Employ

mentAgencyUse

(EA

Use)

1.59

0.64

1.00

0.22

−0.31

**

Impo

rtance

ofnew

recruits

(ImpRec)

1.63

0.85

1.00

−0.14

Project

Success

2.25

0.41

1.00

a Ownershipcode

Coo

perativ

eenterprises,1¼

Privately-ownedenterprises.

*Indicatescoefficientsign

ificancep<

0:05,**Indicatescoefficientsign

ificancep<

0:01.

Source:Derived

from

survey

offirm

sin

ZGC,20

06.

108 M. Ramirez & P. Dickenson

Page 17: Gatekeepers, Knowledge Brokers

constructs show the alpha values are all above 0.6, except for searching and scanning,with a value of 0.57. Although it is a low value, there is precedent for using suchvalues and it was decided to go ahead with the construct.

In order to test hypothesis 2, that innovation projects will gain in performancefrom the direct involvement of knowledge workers in external networks, a

Table 4. Result of regression analysis derived from R&D leaders.Y ¼ �0 þ �1X1 þ �2X2 þ �3X3 þ �4X4 þ �5X5 þ �6X6 þ �7X7 þ ".

IV and status Reg. Wt � t value p

Variables enteredOwnership − 0.33 − 0.35 − 3.73 p < 0:01Size 0.00 − 0.16 − 1.65 n.s.Experience in firm 0.29 0.38 3.96 p < 0:05Experience outside firm 0.24 0.31 3.21 p < 0:05CONSTANT 1.07 4.43 p < 0:01

Variables not enteredAcademic qualifications 0.12 1.14 n.s.Use of employment agencies − 0.17 − 1.71 n.s.Recruitment of new staff − 0.08 − 0.83 n.s.

R2: 0.42, F;12:07ð4;66Þ, p < 0:01.Source: Derived from survey of firms in ZGC, 2006.

Table 5. Survey item allocations to networking practices constructs using knowledge worker as unitof analysis.

Construct Item Available Crombach’s alpha

Inter-organisationalproblem-solving

Sharing knowledge with researchinstitutes

343 0.74

Sharing knowledge with founder bodiesSharing knowledge with standard-setting

bodiesRelational network Sharing knowledge with former

classmates350 0.82

Sharing knowledge with colleaguesSearching and

scanningAttending conferences 360 0.57

External communication viachat rooms, etc.

Informal contact with externalacquaintances

Source: Derived from knowledge workers survey in ZGC, 2006.

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 109

Page 18: Gatekeepers, Knowledge Brokers

three-step MR modelling protocol was adopted and the following multipleregression model fitted. The following model was constructed.

Model (2)

Y ¼ �0 þ �1OWNþ �2SIZEþ �3IOPSþ �4RELþ �5SCANþ �6INTþ ":

Y is the dependent variable and is measured in the same way as model 1, i.e.degree of success on the specific project as reported by R&D leaders.

OWN denotes ownership as discussed earlier in relation to model l.SIZE denotes size of firm as discussed earlier in relation to model l.

In the next step, four variables were constructed to represent different forms ofknowledge transfer.

IOPS is made up of three variables that reflect formal mechanisms of knowl-edge sharing: These were the involvement of the knowledge worker in collabor-ation with other firms, institutes and standard-setting institutions. Following thecategorisation of variables in the third section, this variable has been termed “inter-organisational problem solving” (IOPS).5

REL brings together two variables that emphasise problem-solving using net-works built on personal contacts and relations around ex-colleagues and class-mates. Following the third section, this is a form of “relational problem-solving.”6

SCAN brings together three variables: attending conferences, sharing infor-mation outside of the firm via chat rooms and informal contact with externalacquaintances.7 This variable operationalises an important distinction discussed inthe third section of the paper, where knowledgeworkers act as bridging agents with awider collection of practitioners. The variable is labelled searching and scanning.

INT represents the degree to which knowledge workers collaborate withdepartments outside of their own but within the same company. In this case,marketing and “other departments” have been specified. This variable is labelled“internal problem solving.” Correlations of the constructs with each other and withthe project success composite measure, after mean substitution had beenimplemented, are reported in Table 6.

One major advantage of the above model is that unlike model 1, the dependentvariable is drawn from R&D managers, while the independent variables are drawnfrom the knowledge worker survey. Given that these are two separate sources, thiseases concerns regarding potential problems of self-report. The control variables,

5Knowledge worker survey, see appendix question 10.6Knowledge worker survey, see appendix question 10.7Knowledge worker survey, see appendix question 12.

110 M. Ramirez & P. Dickenson

Page 19: Gatekeepers, Knowledge Brokers

size and ownership, were forced into the model at step one as control variables. Inthe next step, the four variables were constructed to represent different forms ofknowledge transfer.

Table 7 shows the results of the regression model.The regressionmodel yields a substantial overall fit (the F-test is significant at the

1% significance level). The R2 value of 0.21 suggests that this model is a reasonably

Table 7. Result of regression analysis with knowledge workers.Y ¼ �0 þ �1X1 þ �2X2 þ �3X3 þ �4X4 þ �5X5 þ �6X6 þ ".

IV and status Reg. Wt � t value p

Variables enteredSize 0.00 − 0.15 − 1.32 n.s.Ownership − 0.35 − 0.36 − 3.31 p < 0:01General Networking 0.29 0.24 2.14 p < 0:05CONSTANT 1.82 6.85 p < 0:01

Variables not enteredInter-organisational problem-solving (IOPS) − 0.15 − 1.40 n.s.Relational problem-solving − 0.12 − 1.11 n.s.Internal 0.01 0.04 n.s.

R2; 0.21, F; 5:76ð3;67Þ, p < 0:01.Source: Derived from survey of firm and knowledge workers in ZGC, 2006.

Table 6. Means, standard deviations and inter-correlations of networking constructs with projectsuccess (n ¼ 71).

Correlations

Mean SD IOPS RPS INT GNW Success Size Ownership

Inter-organisationalproblem-solving(IOPS)

1.1 0.46 1.00 0.38** 0.46** − 0.01 − 0.22* − 0.05 0.20*

Relational problem-solving (RPS)

1.2 0.53 1.00 0.18 − 0.04 − 0.18 − 0.04 0.16

Internal problem-solving (IPS)

1.6 0.45 1.00 0.17 0.09 − 0.21* − 0.04

General Networking(GNW)

1.9 0.35 1.00 0.23 − 0.22* 0.13

Project Success 2.2 0.41 1.00 − 0.20* − 0.33**

Size 218 450 1.00 0.00

Ownership 0.29 0.44 1.00

*Indicates significance p < 0:05, **Indicates significance p < 0:01.Source: Derived from survey of firm and knowledge workers in ZGC, 2006.

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 111

Page 20: Gatekeepers, Knowledge Brokers

good explanatory model for the dependent variable. Significant variables are own-ership at the 1% level (given the coding strategy, its positive weight suggests thatcooperative enterprises tend to report higher levels of project success than privately-owned firms) and searching and scanning with a positive coefficient at the 5% level.

The empirical findings so far can be grouped into two sets of results. Firstly,analysis of senior R&D manager responses suggest that a combination of internaland external experience will have a positive impact on innovation performance onprojects as reported by senior R&Dmanagers. Thisfinding is important as it suggeststhat Chinese firms consider human resources from outside the firm as an importantcontributing factor for innovation success. It also underlines the importance of “neweconomy” efficiencies in China, as discussed earlier by Bresnahan et al. (2001).

Secondly, the results also suggest that inter-firm collaboration, whether throughformal business-to-business channels or informal networks based on personalrelations, appear to have no impact on innovation performance. However,searching and scanning activity, associated with establishing broader, “non-dense”network arrangements, has a positive impact on performance. Finally, dis-tinguishing between ownership types appears consistently to discriminate betweensuccess and failure in both models. Cooperative enterprises appear more suc-cessful than privately-owned firms.

A Human Capital Approach: Individual Rewards to Networking

The preceding analysis relied on R&D manager perceptions of relative skills andperformance on innovation projects. Although this method has been used in large-scale studies of innovation performance in the past, including the different versionsof the European Community Innovation Survey, there are clear and obviouslimitations to reliance on self-report measures of innovation performance related tothe bias of the respondent.

An alternative approach might be to measure the importance that particularskills and networking activity has on individual performance of R&D managers,by comparing earnings of R&D managers. This is possible to do by relying on theknowledge worker survey. Employee earnings is commonly used to measure thedegree to which particular skills are rewarded in the labour market. The humancapital8 approach in particular suggests that if individuals are paid a premium for aparticular qualification, skill set or specific responsibility, this is because this helps

8The view that earnings regressions measure the relative productivity of different skills sets hasrightly drawn criticism to the fact that this approach ignores salaries that are determined by widerinstitutional, rather than market, factors, such as the existence of Trade Unions, professionalassociations and industry differences that have little to do with individual productivity.

112 M. Ramirez & P. Dickenson

Page 21: Gatekeepers, Knowledge Brokers

to raise their productivity, or alternately that this skill set is highly regarded in thelabour market (Becker, 1962).

The results of the earnings regression can be analysed in parallel with theresults of the innovation studies and help to reinforce or alternatively underminethe findings in the two earlier models regarding the impact that particular skillshave on innovation projects. The earnings regression uses the individual knowl-edge worker as the unit of analysis, therefore it is not necessary to average outknowledge worker responses by firms. This allows the full sample of 381knowledge respondents to be used, thereby improving the model as a whole.

The following hypotheses were developed to reflect the research questionsposed in the paper.

Hypothesis 3Knowledge workers that have high levels of tenure and high levels of labourmarket experience will attract an earnings premium in the ZGC park.

Hypothesis 4Involvement of knowledge workers in formal inter-organisational networks,relational networks and scanning activities in the ZGC park will attract anearnings premium for knowledge workers.

One further hypothesis also tests the impact that labour mobility has on earnings:

Hypothesis 5Inter-firm mobility in the ZGC park will attract an earnings premium for knowl-edge workers in the ZGC park.

In the next step, a regression model is built with wages as the dependentvariable. The regression model is shown below:

Model (3)W ¼ constantþ a1EXPiþ a2 EXP2iþ b1TENUREiþ b2TENURE2i

þc1 Number of PREVIOUS JOBSþ d1 SENþ d2 IOPSþ d3 RELþd4 SCANþ f1 IOPS�SENþ f2 REL�SENþ f3 SCAN�SENþ ":9

The dependent variable W is a banded response to gross annual wages includingbonuses.10 In earnings regressions, the dependent variableWages is usually logged.However in this case, earnings are in bands, therefore this is not required.

EXP is years of experience prior to joining present firm.11

9Conventional earnings regression would, in addition to tenure and experience, control for factorssuch as qualifications and age.10Knowledge worker survey question 5.11Calculated as the number of years working since finishing formal education minus number of yearsin current employment.

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 113

Page 22: Gatekeepers, Knowledge Brokers

TENURE is years in current firm.EXP2 and TEN2 are the square terms of experience and tenure and are intended

to show if there are diminishing returns to experience and tenure over time.SEN is level of seniority (a dummy variable that distinguishes between man-

agerial/senior engineers and non-management R&D employees).Number of PREVIOUS JOBS specifies number of previous jobs and is

included to signify mobility.IOPS is inter-organisational networking.REL is relational networking.SCAN is scanning and searching activity.A first order interaction was introduced between networking activities to

capture the differentials in the return to networking depending on the position ofseniority in the company. These are IOP*SEN, REL*SEN and SCAN*SEN. Theresults are shown in Table 8.

The regression provides strong evidence to support hypotheses three and five,with positive and significant returns to tenure, outside experience and mobility.The coefficients also show that tenure is rewarded more highly than experience.However, although returns to tenure appear to diminish at some point, this is notthe case with experience. This may be because the workforce is very young andtherefore have yet to experience negative returns to experience. If this were the

Table 8. Results of multiple regression gross, monthly wage plusbonuses dependent variable based on Knowledge Workers surveyZGN, 2006.

� t value p

Constant 7.692

EXP 0.278 2.119 p < 0:05EXP2 − 0.120 − 0.937 n.s.TENURE 0.698 5.165 p < 0:01TENURE2 − 0.412 − 3.073 p < 0:01SEN − 0.214 − 1.302 n.sNu of PREVIOUS JOBS 0.174 3.174 p < 0:01IOPS 0.028 0.506 n.s.REL − 0.028 − 0.504 n.s.SCAN − 0.049 − 0.762 n.s.SCAN*SEN 0.371 2.207 p < 0:05IOPS*SEN 0.052 0.445 n.s.REL*SEN − 0.013 − 0.129 n.s.

Dependent Variable: Model summary: R2 ¼ 0:253, adjusted R2 ¼0:229, F ¼ 4:872, Sig ¼ 0.028 n ¼ 381.

114 M. Ramirez & P. Dickenson

Page 23: Gatekeepers, Knowledge Brokers

case, however, it also suggests that at some point the returns to experience willoutpace the rewards for tenure.

With regards to networking activities, the regression shows no significant returnsto any type of networkingwithin the entire sample. However, the interactive variableshows that for senior employees, positive and significant returns to scanningactivities exist. The results suggest that skills to establish formal competencies thatfacilitate formal networking do not attract a wage premium. Similarly, employeesthat turn to their personal networks to problem solve also fail to attract a premium.Onthe other hand, knowledge workers that focus on keeping up with managerial andtechnological developments do attract a premium, irrespective of their experienceand tenure, but this occurs only amongst management and/or senior Chinese R&Demployees. Non-management employees failed to receive a wage premium.

Conclusions and Discussion

The empirical findings can be grouped into three sets of results. Models 1 and 3support the proposition that the profile of “successful” knowledge workers com-bines high levels of firm experience within and outside of the organisation,suggesting that important skills gained in R&D environments will not be limited tofirm-specific experience and will extend beyond the immediate place of work.Clearly, many of the institutional features that characterised the labour market ofscientific and technical employees in China’s state-controlled planned economyhave given way to a highly fluid labour market.

A second finding is that, despite attracting increased salaries (see model 3),higher labour mobility does not appear to impact innovation performance (model 1).Hence, although rapid firm growth and the subsequent demand for high-skilledlabour will clearly push up earnings, this does not necessarily indicate improvedperformance for projects. This conclusion points to the importance of dis-tinguishing between experience (as a learning variable) that knowledge workersgain over time through work, from labour mobility. As such, it casts doubt oversome studies that stress the advantages of unfettered mobility in high-technologyindustries (for example, OECD 2001) and supports those that highlight its possibledrawbacks (Ramirez, 2007; Tomlinson, 1999).

Finally, the discussion investigated whether a new division of labour associatedwith gaining knowledge from outside the organisation’s environment has emergedamongst ChineseR&Demployees (models 2& 3). Of the three networking variablesconstructed in model 2, only searching and scanning activity has a positive effect onperformance. Searching and scanning was also the only networking variable thatattracted a wage premium in model 3, although only amongst senior and managerialR&D staff, suggesting a robust finding. Somewhat surprisingly, no evidence

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 115

Page 24: Gatekeepers, Knowledge Brokers

emerged to suggest that knowledge sharing in formal inter-organisational problem-solving or from informal personal networks differentiates success or failure oninnovation projects. Neither did these activities attract a wage premium.

The methodology and the findings of this study provide some useful reflectionsfor our understanding of how “new economy” practices referred to earlier byBresnahan et al. (2001) and others are being adopted in Chinese high-technologyfirms. The skills analysis certainly underlines the importance firms attach togaining experience outside the organisation, but perhaps more significantly, thevalue of combining complementary skills in gatekeeping-type functions. Thisresult backs up the points made by Harryson et al. (2008) and Easterby-Smithet al. (2008) that we should be focusing not just on knowledge transfer, but also onintegrative skills. Given the importance “bridging” functions play in gatekeeperroles, a greater understanding of the skills required for such a bridging role, forexample, the balancing between communication and interpretive skills and “hard”technical knowledge in different national contexts, would be highly valuable.

A second reflection concerns the effectiveness of contrasting networks. Theresults would appear to support Granovetter’s (1985) depiction of the importanceof relatively “narrow bandwidth” but open channels of communication that helpresearchers’ rapidly access different types of knowledge. Further analysis aroundfluid modes of communication, such as Assimakopoulos and Yan (2006) study ofthe popularity of IT forums amongst Chinese engineers are highly relevant here.

By contrast, the ineffectiveness of “inter-organisational problem-solving”suggests support to those studies that are dubious about the generalised benefitsthat can be derived from formal inter-firm collaboration in innovation projects(Zucker et al., 1995; Hakanson, 2005). Following the earlier discussion, highlevels of knowledge ambiguity may create obstacles to the effectiveness of con-crete knowledge transfer, although this in itself may be due to the absence ofeffective gatekeepers who are able to bridge cognitive gaps between projects.

The fear of losing proprietary knowledge from collaboration has also beenalluded to (Zucker et al., 1995) as a reason for limited inter-firm collaborations andmay be a relevant factor. However, some research suggests that cultural factorsaround Guanxi type relationships and the burden associated with reciprocalfavours (Saxenian, 2005) may place major obstacles to longer-term collaborationin the Chinese context. A recent empirical study of Chinese knowledge workercollaboration patterns would appear to support this view (Assimakopoulos andYan, 2006) and may explain in part why scanning type activities, such as tech-nology forums, that involve little reciprocal engagement, are popular and effective.

Future studies in this area might also note that the emphasis on skills and work-based actions highlights the value of an action-based experiential approach toanalysing the relationship between firm capabilities and networks. While much

116 M. Ramirez & P. Dickenson

Page 25: Gatekeepers, Knowledge Brokers

work in this area has rightly emphasised the tacit-based intangible capabilities atthe level of the firm (Nelson and Winter, 1982; von Hippell, 1988), research on thedivision of labour and skills is able to provide tangible evidence and support forinvestment in some specific human resource practices as opposed to others. Fur-ther research on the development of skills-based indicators in innovation wouldthus further our understanding of firm capabilities substantially.

Finally, a note of caution is merited. The results discussed in this analysis arebased on a small sample and, therefore, need to be extended with larger samplesand probability sampling that would allow stronger inferences and generalizationsto be made from this analysis. Moreover, comparative labour market data thatinterplays the activities of knowledge workers with innovation data at a regional/industry level would provide highly useful insights into the intangible factorsdistinguishing different innovation systems.

Acknowledgements

The authors wish to thank the Economic and Social Research Council for itssupport for this project, grant number: RES-160-25-0030. Also many thanks to theresearchers of the National Research Centre for Science and Technology forDevelopment in Beijing for their work in the compilation of the survey.

Appendix

Examples of questions from manager survey

21. Indicate how important do you value the following characteristics ofemployees who play a significant role in innovation, by ticking the relevant boxbelow, where 0 ¼ Not important at all, 1 ¼ Not very important, 2 ¼ Desirable,3 ¼ Essential

0 1 2 3Their academic qualificationThe status of the institution they were educated atTheir experience and knowledge of the firmOutside networks the individual can bring to the firmProfessional qualification (e.g. IEEE membership)Experience of working/studying overseasExperience in specific projects outside of the firm

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 117

Page 26: Gatekeepers, Knowledge Brokers

22. Indicate what relative importance do you attach to the following methods torecruit employees who play a significant role in innovation, by ticking the relevantbox, where 0 ¼ unimportant, 1 ¼ not very important, 2 ¼ quite important, 3 ¼very important.

0 1 2 3 AdvertisingDirect approach to top scientists/engineers from universities/research institutesHead hunt from leading firmsApprenticeships via schools/universitiesInformal peer network Employment agencies If spin off, direct approach to founder body

Examples of Questions from Knowledge Worker Survey

10. To what extent has “sharing knowledge” between you and any of the followinginfluenced the output of the “innovation project”?

Below, ‘Founder body’ refers to a university or research institute that originallyestablished the firm.

0 ¼ no knowledge shared 1 ¼ small influence 2 ¼ some influence 3 ¼ veryinfluential (please tick one box for each question)

0 1 2 3 Your project teamColleagues from the marketing department Colleagues from departments other than your own and marketing Individuals from academic departments and/or research institutes Individuals from Founder body, if applicable Colleagues participating in standard-setting bodies orinstitutions.Individuals with whom the firmhas a formal relationship (forexample customers, suppliers, joint venture, affiliates) Former class-mates from the universityFormer colleagues

118 M. Ramirez & P. Dickenson

Page 27: Gatekeepers, Knowledge Brokers

12. Have any of the following been important sources of learning for you aboutnew technologies or managerial methods?

Please indicate the degree of importance attached to each question by ticking0 ¼ not important 1 ¼ slightly important 2 ¼ moderately important 3 ¼ veryimportant (please tick one box for each question)

0 1 2 3 Attending conferences Communication with people outside the company (e.g. group email,message board, chat room)Email communication inside teamInformal communication with people you know outside the company dealing with similar problemsOverseas visitsTraining outside of company inside the ZGC ParkInternal training

References

Allen, TJ (1977). Managing the Flow of Technology: Technology Transfer and theDissemination of Technological Innovation within the R&D Organization(Cambridge, MA, MIT Press).

Arthur, MB and DM Rousseau (1996). The Boundaryless Career, a New EmploymentPrinciple for a New Organizational Era, Oxford: Oxford University Press.

Arora, A and A Gambardella (1994). Evaluating technological information and utilizing it:Scientific knowledge, technological capability and external linkages in biotechnol-ogy. J Econ Behaviour Organ, 24(1), 91–114.

Assimakopoulos, D and J Yan (2006). Sources of knowledge acquisition for Chinesesoftware engineers. R&D Management, 36(1), 97–106.

Barney, JB (1991). Firm resources and sustained competitive advantage. Journal ofManagement, 17, 99–120.

Becker, G (1962). Investment in human capital: A theoretical analysis. Journal of PoliticalEconomics, 70(5), S9–S49.

Best, M (2001). The New Competitive Advantage: The Renewal of American Industry.Oxford: Oxford University Press.

Bresnahan, T, A Gambardella and A Saxenian (2001). “Old economy” inputs for “neweconomy” outcomes: Cluster formation in the new Silicon Valleys. Industrial andCorporate Change, 10(4), 835–860.

Bouty (2000). Interpersonal and interaction influences on informal resource exchangesbetween R&D researchers across organisational boundaries. The Academy of Man-agement Journal, 43(1), 50–65.

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 119

Page 28: Gatekeepers, Knowledge Brokers

Burt, R (2005). Brokerage and Closure: An Introduction to Social Capital. Oxford:Oxford University Press.

Cai, J and A Tylecote (2005). A healthy hybrid: The technological dynamism of minoritystate-owned firms in China. Technology Analysis and strategic Management, 17(3),257–278.

Cao, C (2002). Strengthening China through science and education: China’s developmentstrategy towards the Twenty First Century. Issues and Studies, 38(3), 122–149.

Casper, S and F Murray (2005). Career and clusters: Analyzing the career networkdynamic of biotechnology clusters. Journal of Engineering and TechnologyManagement, 22 (1–2), 51–74.

Castells, M (1996). The Rise of the Network Society. Oxford: Blackwell.Chesbrough, HW (2006). Open Business Models: How to Thrive in the New Innovation

Landscape. Harvard: Harvard Business School Press.Cohen, J and P Cohen (1983). Applied Regression/Correlation Analysis for the Behavioral

Sciences (2nd Ed.). New Jersey: Lawrence Erlbaum Hillsdale.Cohen, WM and DA Levinthal (1990). Absorptive capacity: A new perspective on

learning and innovation. Administrative Science Quarterly, 35, 128–152.Crescenzi, R, A Rodriguez-Pose and M Storper (2007). The territorial dynamics of

innovation: A Europe-United States comparative analysis. Journal of EconomicGeography, 7(6), 673.

DeFillippi, R, MB Arthur and VJ Lindsay (2006). Knowledge at Work: Creative Col-laboration in the Global Economy. MA: Blackwell Publishing.

Easterby-Smith, M, ML Lyles and EWK Tsang (2008). Inter-organizational knowledgetransfer: Current themes and future prospects. Journal of Management Studies, 45(4),677–690.

Fang, Y (1991). Follow-up survey of the graduates of seventeen short-term universitiesand useful points learned. Chinese Education, Summer.

Freeman, C and L Soete (1997). The Economics of Industrial Innovation. (3rd Edition)London: Pinter.

Gibbons, M, C Limoges, H Nowotney, S Schwartzman, P Scott and M Trow (1994). TheNew Production of Knowledge. London: Sage.

Granovetter, M (1973). The strength of weak ties. American Journal of Sociology, 78(6),1360–1380.

Hakanson, L (2005). Epistemic communities and cluster dynamics: On the role ofknowledge in industrial districts. Industry and Innovation, 12(4), 433–463.

Harryson, SJ, R Dudkowski and A Stern (2008). Transformation networks in innovationalliances — the development of Volvo C70. Journal of Management Studies, 45,730–758.

Haunschildt, J and G Schewe (2000). Gatekeeper and process promotor: Key persons inagile and innovative organizations. International Journal of Agile ManagementSystems, 2(2), 96–103.

Holloway, D (1982). Innovation in the defence sector, In Industrial Innovation in theSoviet Union, R Amann and J Cooper (eds.), Harvard University Press.

120 M. Ramirez & P. Dickenson

Page 29: Gatekeepers, Knowledge Brokers

ILO (2001). World Employment Report: Life at Work in the Information Economy.Geneva: ILO Publications.

Katila, R and G Ahuja (2002). Something old, something new: A longitudinal study ofsearch behavior and new product introduction. Academy of Management Journal, 45,1183–1194.

Leininger, J (2004). The key to retention: Committed employees. China Business Review,31(1), 16–39.

Leonard-Barton, D (1995).Wellsprings of Knowledge: Building and Sustaining the Sourceof Innovation. Cambridge, Mass: Harvard University Press.

Levin, DZ and R Cross (2004). The strength of weak ties you can trust: the mediating roleof trust ineffective knowledge transfer. Management Science, 50, 1477–1490.

Lewin, K and H Xu (1993). Higher education in transition: Some strategic planning issuesin China and Britain. In Proceedings of the Sino-British Conference on HigherEducation, C Wang and H Xu (eds.), March 1992, Hangzhou University Press.

Lu, Q (2001). Learning and innovation in a transitional economy: The rise of science andtechnology enterprises in the Chinese information technology industry. InternationalJournal of Technology Management, 21(1–2), 76–92.

Macdonald, S and R Piekkari (2005). Out of control: Personal networks in Europeancollaboration. R&D Management, 35(4), 441–453.

Macdonald, S and C Williams (1992). Survival of the gatekeeper: Proceedings of theinternational product development conference on new approaches to developmentand engineering. EIASM, Brussels, 349–364. 18–19 May.

Massey, D, P Quintas and D Wield (1992). High Tech Fantasies: Science Parks in Society,Science and Space. London: Routledge.

Nelson, RR and SG Winter (1982). Evolutionary Theory of Economic Change. Harvard:Harvard University Press.

OECD (1969). Science Policy in the USSR. Paris: OECD.OECD (2001). Innovative people: Mobility of skilled personnel in national innovation

systems, Proceedings from the OECD Workshop on Science and Technology LabourMarkets, DSTI/STP/TIP(99)2/FINAL, OECD, Paris.

Penrose, ET (1959). The Theory of the Growth of the Firm. New York: Wiley.Powell, W (1998). Learning from collaboration: Knowledge networks in biotechnology

and pharmaceutical industries. California Management Review, 40, 228–241.Powell, WW and S Grodal (2005). Networks of Innovators. In The Oxford Handbook of

Innovation, J Fegeberg, DC Mowery and RR Nelson (eds.), Oxford: OxfordUniversity Press.

Raatikainen, P (2003). Training China’s skilled workforce for future success. China Staff,9(7), 24.

Ramirez, M (2007). Redefining firm competencies, innovation and labour mobility: A casestudy in telecommunication services. Industry & Innovation, 14(3), 325–347.

Reed, R and RJ DeFilippi (1990). Causal ambiguity, barriers to imitation, and sustainablecompetitive advantage. Academy of Management Review, 15, 88–102.

Rigby, D and C Zook (2002). Open market innovation. Harvard Business Review, 80(10),80–89.

Inter-Firm Knowledge Transfer in Beijing’s Zhongguancun Science Park 121

Page 30: Gatekeepers, Knowledge Brokers

Rodan, S and D Galunic (2004). More than network structure: How knowledgeheterogeneity influences managerial performance and innovativeness. StrategicManagement Journal, 25, 541–62.

Sammarra, A and L Biggiero (2008). Heterogeneity and specificity of inter-firm knowl-edge flows in innovation networks. Journal of Management Studies, 45(4), 800–829.

Saxenian, A (1996). Beyond boundaries: Open labour markets and learning in SiliconValley. In The Boundaryless Career, A New Employment Principle for a NewOrganisational Era, BM Arthur and DM Rousseau (Eds.), Oxford: OxfordUniversity Press.

Saxenian, A (2005). Government and Guanxi: The Chinese software industry in transition.In The Software Industry in Emerging Markets, S Commander and E Elgar.

Suttmeier, R (1997). Emerging innovation networks and changing strategies for industrialtechnology in China: Some observations. Technology in Society, 19(3–4), 305–323.

Szulanski, G, R Cappetta and RJ Jensen (2004). When and how trustworthiness matters:Knowledge transfer and the moderating effect of causal ambiguity. OrganizationScience, 15(5), 600–613.

Tabachnick, BG and LS Fidell (1996). Using Multivariate Statistics. New York: HarperCollins.

Teece, D (1988). Capturing value from knowledge assets: The new economy, markets forknow-how and intangible assets. California Management Review, 140, 1–25.

Teece, D and G Pisano (1994). The dynamic capabilities of firms: An introduction.Industrial and Corporate Change, 3(3), 537–556.

Thompson, GF (2003). Between Hierarchies and Markets: The Logic and Limits of Net-work Forms of Organization. Oxford: Oxford University Press.

Tomlinson, M (1999). The learning economy and embodied knowledge flows in GreatBritain. Journal of Evolutionary Economics, 9, 431–451.

Tushman,M andT Scanlan (1981). Boundary spanning individuals: Their role in informationtransfer and their antecedents. Academy of Management Journal, 24(2), 289–305.

Vandamme, F (2000). Labour mobility within the European Union: Findings, stakes andprospects. International Labour Review, 139(4), 437–455.

van Wijk, R, JP Jansen and MA Lyles (2008). Inter and intra-organizational knowledgetransfer, a meta-analytic review and assessment of its antecedents and consequences.Journal of Management Studies, 45(4), 830–853.

Von Hippell, E (1994). Sticky information and the locus of problem-solving: Implicationsfor innovation. Management Sci, 40(4), 423–439.

Walcott, S (2003). Chinese Science and Industrial Technology Parks. Ashgate, Hampshire.Wenger, S (1998).Communities of Practice. NewYork, Cambridge,Mass: University Press.Yin, Q and G White (1994). The “marketisation” of Chinese higher education: A critical

assessment. Comparative Education, 30, 217–37.Zimmerman, K (1995). Tackling the european migration problem. Journal of Economic

Perspectives, 9, 45–62.Zucker, L, MR Darby and M Torero (1995). Labor Mobility from Academe to Commerce,

NBER Working Paper Series 6050.

122 M. Ramirez & P. Dickenson

Page 31: Gatekeepers, Knowledge Brokers

Copyright of International Journal of Innovation Management is the property of World Scientific Publishing

Company and its content may not be copied or emailed to multiple sites or posted to a listserv without the

copyright holder's express written permission. However, users may print, download, or email articles for

individual use.