Interpreting the relationships between network closure and firms competitive advantages a knowledge based perspective

50 3.3.2 Predicting direct effects – the impacts of social network structure on interorganizational learning and protection .... Firstly, literature in organizational learning is limite

NETWORK CLOSURE AND FIRMS’ COMPETITIVE ADVANTAGES: A KNOWLEDGE-BASED PERSPECTIVE

WANG XIAOYANG

NATIONAL UNIVERSITY OF SINGAPORE

2007

NETWORK CLOSURE AND FIRMS’ COMPETITIVE ADVANTAGES: A KNOWLEDGE-BASED PERSPECTIVE

WANG XIAOYANG

(B Eng Tsinghua Univ., M Sc Tsinghua Univ.)

A THESIS SUBMITTED FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

DEPARTMENT OF INDUSTRIAL AND SYSTEMS ENGINEERING

NATIONAL UNIVERSITY OF SINGAPORE

2007

I am deeply grateful to my supervisor, Dr Chai Kah-hin, not only for his suggestions and advice which were critical to the preparation of this manuscript, but also his emotional support and encouragement throughout the project

I also wish to express gratitude to my co-supervisor, Dr Yap Chee-Meng, for his sharp opinions and input in both research topics and methodologies

I want to further thank members of my research group, in particular Foong Hing Wih (Awie), Xin Yan, Lin Jun, Wang Qi, Chang Hongling, Xing Yufeng (Esther), Yu Dan, Ulf Andreas Hamster, and colleagues from business school, in particular Dr Lim Kwanghui, Dr Soh Pek-Hooi, Manathattai S Annapoornima (Poornima), Ruan Yi (Annie), Zhuang Wenyue, for their valuable suggestions and feedback

I extend thanks to project collaborators in China, in particular Dr Wu Jinxi, Professor Gao Xudong (Tsinghua University), Dr Yu Jiang (China Academy of Science), and collaborators in Taiwan, in particular Professor Liu Shang-Jyh, Chu Mei-Tai (Debbie) (National Chiao Tung University), for their valuable suggestions and kind help in administering large scale survey

I also show my gratitude to the many anonymous industrial association staff, practitioners, interviewees, and friends, for their support and valuable opinions

Finally, I would like to thank my wife, Guo Yan (Helen), and my parents, for their understanding during this most memorable period of time in my life

Acknowledgements i

Table of Content ii

Summary v List of Tables viii

List of figures x

Chapter 1 Introduction 1

1.1 Research background 1

1.2 Research objective 3

1.3 Structure of the dissertation 4

Chapter 2 Literature Review 6

2.1 Introduction 6

2.2 Organizational learning literature 6

2.2.1 Introduction to organizational learning 6

2.2.2 Firm-level interorganizational learning 8

2.2.3 Interorganizational learning through strategic alliance 10

2.2.4 Nexus between interorganizational learning and internal organizational learning 14

2.2.5 Knowledge attributes and its relevance to interorganizational learning 16

2.2.6 Summary of organizational learning review 20

2.3 Social network literature 22

2.3.1 Contradictory results from studies combining organizational learning with social network theories 22

2.3.2 Recent efforts devoted to reconcile the contradictions and further gaps 26

2.3.3 Summary of social network review 27

2.4 Absorptive capacity 27

2.4.1 Classical absorptive capacity model 28

2.4.2 Advanced absorptive capacity models 30

2.4.3 Summary of absorptive capacity review 32

2.5 Conclusion 32

2.6 Research questions 33

Chapter 3 Hypotheses Development 36

3.1 Introduction 36

3.2 Preliminary case study and interviews 36

3.2.1 Background of the case study 36

3.2.2 Data collection 37

3.2.3 Discussion of preliminary findings 38

3.2.4 Additional interviews in other companies 47

3.2.5 Summary of findings from preliminary case study and interviews 49

3.3 Hypotheses developed from existing literature 50

3.3.1 Working definitions of network structure, interorganizational learning and competitive advantages 50

3.3.2 Predicting direct effects – the impacts of social network structure on interorganizational learning and protection 52 3.3.3 Predicting direct effects – from learning and protection to competitive

3.3.5 The moderating effects of social integration and environmental dynamics 62

3.4 Summary 66

Chapter 4 Survey Instrument Development and Implementation 69

4.1 Introduction 69

4.2 Sampling Strategy 69

4.2.1 Network Type Selection 69

4.2.2 Defining Ties 70

4.2.3 Choosing industries 72

4.3 Questionnaire Design 77

4.3.1 Measures: Outcome variables 77

4.3.2 Measures: Predictor variables 80

4.3.3 Measurement variables: Network Variables 82

4.3.4 Measures: Moderating variables: 85

4.3.5 Measures: Control variables: 86

4.3.6 Summary of definitions of the constructs 86

4.4 Survey Implementation 88

4.4.1 Reverse translation issues and pre-test of the questionnaire 88

4.4.2 Survey administration 89

4.5 Summary 94

Chapter 5 Data Analysis 96

5.1 Introduction 96

5.2 Data Bias Analysis 96

5.2.1 Non-response bias analysis 96

5.2.2 Excluding potential biases caused by different collection methods 100

5.3 Descriptive Analysis 102

5.3.1 Firm size 102

5.3.2 Ownership 103

5.3.3 Position and years of experience 104

5.4 Measurement Models 106

5.4.1 Choosing PLS as the analytical method 106

5.4.2 Factor analysis and item reliability 110

5.4.3 Convergent validity and discriminant validity 117

5.5 Structural Models 120

5.5.1 Testing direct effects by bootstrapping 120

5.5.2 The mediating effects between NETDENS and COMPADV 125

5.5.3 The moderating effects of SOCIINTE and ENVIDYNA 130

5.5.4 Analysis of different dimensions of formative structures 139

5.6 Summary 141

Chapter 6 Discussion and Conclusions 143

6.1 Introduction 143

6.2 Discussion of research findings 143

6.2.1 Findings about the impacts of network density on learning process and knowledge protection 143

6.2.2 Findings about the antecedents of firm’s competitive advantages 144

6.2.3 Findings about the mediating effects of knowledge identification and protection 145

6.2.4 Findings about the moderating effects 146

6.2.5 Findings about the differences between the two industries 150

6.2.6 Summary of findings 151

6.3 Implications of the study 151

6.4 Limitations of the study and further directions 156 6.5 Conclusions 159

Appendix A: References 161

Appendix B: Questionnaire (Cover letter)

Appendix C: Questionnaire (English version)

Appendix D: Questionnaire (Chinese version)

With the expansion of the knowledge bases of products/services, the sources of such expertise are becoming more widely dispersed outside a firm’s boundary The current study aimed to examine a set of practical questions raised under such circumstances with regard to how firms operating in network environments could learn from their partner firms while protecting their core knowledge Specifically, literature in organizational learning and social networks were reviewed Several gaps were identified Firstly, literature in organizational learning is limited to addressing the learning and protection dilemma only at dyadic level while the nexus between internal learning and external learning remains under-explored Secondly, social network theories are inadequately incorporated with organizational learning and knowledge management studies More often than not, existing studies applied social network theories only to a fraction of the overall learning process and a fine-grained examination has yet to be conducted Realizing the importance of network structures for interorganizational learning under such network environment, a set of research questions was raised as below:

1) What are the impacts of network structures on a firm’s interorganizational learning process and knowledge protection?

2) What are the effects of interorganizational learning and knowledge protection on a firm’s competitive advantage?

3) What are the contingency factors that differentiate firms’ interorganizational learning outcomes?

Based on literature inputs as well as empirical knowledge obtained from a preliminary

previous two literature bodies by establishing the mediating role of interorganizational learning and knowledge protection process between network closure and firms’ competitive advantages It is hypothesized that network closure has both positive and negative impacts on different stages of interorganizational learning and knowledge protection process, and hence further affect the focal firm’s competitive advantages In addition, internal social integration mechanism was hypothesized as the moderating factor that affects individual firms’ learning outcomes, and environmental dynamics was hypothesized as the moderating factor that affects the relative importance of knowledge identification and protection for enhancing the focal firm’s competitive advantages

Large scale surveys in petrochemical industry (in mainland China) and semiconductor industry (in Taiwan) were conducted, and PLS analysis as implemented in PLS Graph 3.0 yielded statistical evidences in favor of the above hypotheses Specifically, the current study revealed that network closure is positively associated with a firm’s capability in knowledge transfer and protection, but the close structure restrains a firm’s capability in knowledge identification For firms operating in a dynamic environment and with low internal integration mechanisms a sparse network is recommended to exploit the flexibility in knowledge identification without suffering badly from knowledge leakage While for firms operating in a stable environment with strong integration mechanism, a dense network configuration is more appropriate to leverage the interorganizational learning benefit and protect the firm’s core competencies As such managers could purposely design and monitor their social network structure so as to maximize the interorganizational learning benefits while

Table 3-1 Data sources of the case study 38

Table 3-2 A description of Alpha’s interorganizational learning 38

Table 3-3 Summary of preliminary interviews 48

Table 3-4 Summary of Hypotheses sets 67

Table 4-1 Summary of interviews for choosing industries 73

Table 4-2 A summary of definitions of the constructs and corresponding measurement

items 87

Table 5-1 Response status for the survey in three industries 97

Table 5-2 Non-response bias test for petrochemical and semiconductor industries 98

Table 5-3 Descriptive statistics and Chi-square for mail-based and telephone interview-based samples 100

Table 5-4 Firm size stats 102

Table 5-5 Distribution of ownership in mainland China and Taiwan samples 103

Table 5-6 Descriptive data on informants’ position and years of experience 105

Table 5-7 EFA for the three dimensions of COMPADV 110

Table 5-8 EFA for the three dimensions of COMPADV (after trimming) 111

Table 5-9 EFA for constructs with reflective indicators 112

Table 5-10 Environmental dynamics revisited 113

Table 5-11 EFA for constructs with reflective indicators (excluding ENVIDYNA) 115

Table 5-12 Loadings for all measurement items (petrochemical dataset) 116

Table 5-13 Loadings for all measurement items (semiconductor dataset) 117

Table 5-14 Measurement convergent validity 118

Table 5-15 Correlation between constructs (Petrochemical) 119

Table 5-16 Correlation between constructs (Semiconductor) 119

Table 5-17 Multi-collinearity tests for semiconductor industry 120

Table 5-18 Path coefficients and significance 121

Table 5-19 Results of hypotheses testing (direct effects) 123

Table 5-20 Analysis of the dimensions of competitive advantage (petrochemical

Table 5-21 Analysis of the dimensions of competitive advantage (Semiconductor

industry) 125

Table 5-22 Nested model comparison 127

Table 5-23 Significance of mediated paths from NETDENS to COMPADV 128

Table 5-24 Path coefficients and significances for SOCIINTE*KNOWTRAN 132

Table 5-25 Simple effect of knowledge transfer on knowledge institutionalization among different groups 132

Table 5-26 Path coefficients and significances for ENVIDYNA*KNOWIDEN 134

Table 5-27 Simple effect of knowledge identification on competitive advantage among different groups 134

Table 5-28 Path coefficients and significances for ENVIDYNA*KNOWIDEN (with only strategic flexibility indicators for COMPADV) 135

Table 5-29 Simple effect of knowledge identification on competitive advantage among different groups (with only strategic flexibility indicators for COMPADV) 135

Table 5-30 Path coefficients and significances for ENVIDYNA*KNOWPROT 136

Table 5-31 Simple effect of knowledge protection on competitive advantage among different groups 137

Table 5-32 Path coefficients and significances for ENVIDYNA*KNOWPROT (with only innovativeness indicators for COMPADV) 137

Table 5-33 Simple effect of knowledge protection on competitive advantage among different groups (with only innovativeness indicators for COMPADV) 138

Table 5-34 Analyzing different dimensions of competitive advantage 139

Table 5-35 Analyzing different indicators of knowledge transfer 140

Table 5-36 Summary of hypotheses testing 141

Table 6-1 Analyzing environmental dynamics in the two industries 148

Table 6-2 Further tests on the direct impact of network closure on competitive advantage 149

Figure 2-1 Hypotheses represented as a homological network (Schroeder 2002) 14

Figure 2-2 A dynamic model of intra- and inter- organizational learning (Holmqvist 2003) 15

Figure 2-3 A pragmatic framework for KM research (Grover and Davenport 2001) 17

Figure 2-4 Knowledge tacitness, governance difficulty, and learning alliances (adopted from Lubatkin et al 2001) 18

Figure 2-5 A model of Absorptive Capacity (adopted from Zahra and George 2002) 31

Figure 3-1 Alpha’s network information 39

Figure 3-2 Different learning mechanisms at different design stage 41

Figure 3-3 Illustration of Alpha’s internal learning efficiency 46

Figure 3-4 K nowledge protection in open / close networks 55

Figure 3-5 Full model of effects of network closure on competitive advantage, mediated by knowledge identification and protection, and moderated by social integration mechanisms and environmental dynamics 65

Figure 4-1 Working procedures for making follow-up phone calls 92

Figure 5-1 Research model result - petrochemical industry (direct effects only) 122

Figure 5-2 Research model result - semiconductor industry (direct effects only) 122

Figure 5-3 Graphical representation of the full model 131

Figure 5-4 Graphical representation of the Hypotheses testing results (petrochemical) 142

Figure 5-5 Graphical representation of the Hypotheses testing results (semiconductor) 142

Figure 6-1 Matching network structures with industrial and company characteristics 155

More specifically, as traditional multi-national companies (MNCs) are increasingly outsourcing not only production operations, but also R&D processes, and small companies are “networked” to serve end customers and compete with vertically integrated MNCs, learning from networks has become a new issue that needs to be addressed by organizational learning studies

While resource-based view (RBV) of the firm considers knowledge as one of the most important rent-generating resources (e.g Barney, 1991), the credibility of this

argument however “depends critically on the assumption that a firm can protect its

knowledge from appropriation or imitation by its competitors” (Liebeskind 1996, pp

95), and protection of unique knowledge, together with its deployment, underlies sustained competitive advantage (Barney, 1991) The issue of knowledge appropriation

is of greater concern for companies operating under a network environment, where the knowledge base of products becomes both complex and expanding and the sources of

expertise are widely dispersed outside the company (Powell et al, 1996) Companies facing such a situation have to collaborate and learn from each other to avoid severe problems such as the lack of a shared language and routines, which are normally developed within integrated firms (Schilling & Steensma, 2001) However, by

participating in interorganizational learning activities, companies may be “at greater

risk of having its proprietary technologies or competitive tactics disseminated to other firms” (Schilling & Steensma 2001, pp 1153)

Recent literature has paid considerable attention to the dilemma between learning and protection Larsson et al (1998), for instance, applied game theory to address the issue

of interorganizational learning dilemma under the setting of dyadic strategic alliance Kale et al (2000) suggested that relational capital and conflict management could achieve learning and protection simultaneously Oxley and Sampson (2004) reported that some companies protected their core knowledge by restricting the scope of their alliance activities However, these dyadic level studies neglect the possible impact of social networks, and specifically network structures, on learning and protection According to Powell (1990), social networks might contribute critically to organizational learning because it could be the most efficient organizational arrangement for sourcing information, and traditional mechanisms such as market and hierarchy might be less efficient due to the difficulty in pricing information (for market mechanism) and bureaucracies (for hierarchy mechanism) However, though strategy

researchers increasingly view firms as “embedded in networks of social and

professional relationships with other organizations, rather than autonomous entities”

(Gulati et al 2000, pp 203), a research incorporating a social network perspective into organizations’ learning and knowledge protection activities has yet to be conducted

In recognizing the importance of firms’ social networks to their performance, researchers therefore viewed social relationships as social capitals Network structures

as the aggregate level of social relationships were studied intensively, and the two most often discussed mechanisms are “protection” within network closure and “brokerage” across structural holes (Burt, 2000) However, inconsistent results were found for the two mechanisms (Brass et al, 2004), thus raising the need for further theoretical explanation

1.2 Research objective

As more and more companies begin to collaborate in order to survive the fluctuating environment (Hamel, 1991), a set of practical questions became crucial: how would companies, especially those operating under network context, learn? Who can they learn from? What types of learning exist and what are the results of these learning? Can they get a better competition position by learning from others or do they actually lose such kind of a position by giving away core knowledge to their collaborators? The current study was set to tackle such questions from the angle of knowledge-based view and network structure was anchored as the focus of the study In particular, the study aimed to answer the following research question:

How could network structures impact on firms’ competitive advantages from a knowledge-based view?

Specifically, interorganizational learning and knowledge protection were incorporated

in the scope of current research, and I argue that a fine-grained examination on not only interorganizational learning process but also knowledge protection could contribute to the extant literature by offering explanations for inconsistent findings

reported in different studies

1.3 Structure of the dissertation

The dissertation consists of six chapters A brief description of each chapter is listed below:

Chapter 2 - Literature review: In order to address the practical questions listed in Chapter 1, literature in organizational learning and social networks was examined Absorptive capacity was found to be one of the key mechanisms to create synergy between internal and external learning and served as the nexus between the previous two literature bodies Research questions and sub-questions were identified as the result of a thorough literature review

Chapter 3 - Hypotheses development: In order to get an overview of the interorganizational learning in a practical situation, a preliminary case study in a design firm, as well as a few interviews with home appliance and semiconductor firms, were conducted in advance to the literature search for hypotheses development Social network literature was then anchored as the main frame of reference for the current study and relevant research arguments and results were examined for inputs of the hypotheses development In accordance with the research questions raised in Chapter 2 the hypotheses were further developed in three blocks, i.e hypotheses on direct effects, mediating effects and moderating effects

Chapter 4 - Survey instrument development and implementation: Large scale survey was chosen as the research methodology to validate the hypotheses developed in Chapter 3 Firm’s ego network was then chosen as the appropriate subject of study

Furthermore, technological knowledge flows among focal firm and their alter partners were argued to be the most suitable subject in representing the ties of the ego network Based on such a network definition, petrochemical and semiconductor industries were selected The main body of the questionnaire was then developed based largely on literature and further modified by incorporating suggestions and comments from practitioners in the two chosen industries Large scale survey in the two industries was conducted between Nov 2005 and Feb 2006 (for petrochemical), and Mar 2006 and May 2006 (for semiconductor) Follow-up phone calls and email notifications were sent out in an attempt to encourage responses and further understand the industrial characteristics

Chapter 5 - Data analysis: Following the procedures elaborated in Chapter 4, two datasets were collected with 78 and 76 sample size After careful examination of potential data bias and validation of the measurement quality, PLS as implemented in PLS Graph 3.0 was used to test the three sets of hypotheses, i.e hypotheses on direct effects, mediating effects and moderating effects Most hypotheses were supported with convincing statistical results, and hypotheses with partial supports / no support were further discussed and potential reasons were identified

Chapter 6 - Discussion and conclusions: This last chapter summarizes the dissertation

by demonstrating that research questions raised in chapter 2 were satisfactorily answered in the current study Implications for researchers and practitioners were identified, limitations of this study were discussed and areas for future research were further proposed

Chapter 2 Literature Review

2.1 Introduction

As described in the previous chapter, the main concern of the current study was on the dilemma between learning and protection under network contexts, therefore extant literature was reviewed in two main blocks, i.e organizational learning /interorganizational learning issues and social network studies Moreover, absorptive capacity was examined and highlighted separately since it depicts the firm’s capability

in converting its social capitals and transforming them into tangible learning results Therefore absorptive capacity could be viewed as the nexus between the previous two main literature blocks

2.2 Organizational learning literature

2.2.1 Introduction to organizational learning

In a broad sense, organizations should learn because the rate at which individuals and organizations learn may become the only sustainable competitive advantage, especially

in knowledge-intensive industries (Stata, 1989) According to the resource-based view

of the firm (Barney, 1991; Teece, 1987), learning is crucial for companies because it is the way to create idiosyncratic modes of technology at any point in time (Conner and Prahalad, 1996) and could result in proprietary process and equipment that are difficult for others to imitate in the short term Furthermore, such process and equipment could not be acquired through factor markets (Schroeder et al., 2002) Moreover, learning from other companies could serve as a way to acquire complementary knowledge and skills (Scott, 2000)

Despite the importance of organizational learning, there lacks a consensus on its definition, and definitions are as many as there are writers on the subject (Tsang, 1997) Crossan et al (1995) developed a more popular definition regarding organizational learning In their review of influential papers on organizational learning, they classified the literature body along three dimensions, i.e 1), unit of analysis; 2), cognitive and /

or behavioural change as the outcome of learning, and 3), the relationship between learning and performance measures For the first dimension, levels of learners are identified as individual, group and organizational, and to the relevance of this study, interorganizational level For the second dimension, they advocated an integrative learning which reflects both cognitive and behaviour changes As for the third dimension, they concluded that learning might not be positively related to performance due to impacts of environmental factors and implementation time required for learners

to fully exploit the benefits of learning (Huber, 1991)

Although researchers from different disciplines hold different perspectives, Holmqvist (2003) argued that there were four basic assumptions underpinning organizational learning literature, i.e 1) organizational learning was experiential, and thus learning

“is seen as a relatively permanent change in organizational knowledge that is produced by experience”; 2) learning “is a process that relatively permanently alters the character of behaviour”; 3) learning “is individual learning taking place in a social context” and 4) learning “is organized by existing standard operating procedures, practices and other organizational rules” (pp.97) By comparing Crossan

et al (1995) and Holmqvist (2003)’s work, it is clear that organizational learning literature begins to converge in agreeing that organizational learning is an individual-based, socially embedded, multi-level organizational cognitional and

behavioral change (i.e., what), that helps companies to probe the external changing world and compete through internal proprietary capability building (i.e., why), and it is possible for organizations learn better by articulating and facilitating the learning process (Scott, 2000), as well as by cultural building and motivation (Pemberton et al., 2001) (i.e., the how)

However, the above body of literature did not address learning issues raised in

interorganizational context As noted by Crossan et al (1995), “there have been few

attempts to link a concept of organizational learning to interorganizational learning”,

and “fewer the interactions between the levels” This view is strengthened by other

researchers in arguing that few studies had attempted to extend organizational learning theory to an interorganizational level (e.g Larsson et al., 1998; Lane and Lubatkin, 1998) In an attempt to address the practical questions proposed in the previous chapter,

I extended the literature review to interorganizational learning level

2.2.2 Firm-level interorganizational learning

As described above, traditional approaches on learning in organizations had emphasized learning within integrated and coherently bounded entities, which may not

be suitable in an independent, hostile environment (Holmqvist, 1999) Ironically on one hand the goal of organizational learning is to survive the environment, yet on the other hand it has mainly focused within the organization boundary and the links to outside overlooked Scholars acknowledged that an organization's learning and value-creation are inescapably embedded in various forms of partnerships Learning

thus took place in the “interstices between firms, universities, research laboratories,

suppliers, and customers” (Powell et al., 1996, pp.118) In this sense organizations

could be seen as enduring alliances between independently knowledge-creating entities,

be it individuals, teams, or other organizations (Spender, 1996)

While theories have began to converge for organizational learning, studies on interorganizational learning are arguably deviated from their origin, and the two theory bodies are living in partly separate worlds (Holmqvist, 2003) Traditionally, interorganizational learning had been regarded as a way of developing the organizational learning literature by conceptualising another unit of analysis (Holmqvist, 1999, 2003; Lane and Lubatkin, 1998; Larsson et al., 1998) However, Knight (2002) argued that the major focus of existing literature on interorganizational

learning was “on the appropriation of learning by the individual organization – what

each firm can learn (acquire) from the other, or from their interaction.”(pp.435)

Knight also shifted his emphasis to treat a network as a learning body Knight (2002)’s work provided a typology to distinguish interorganizational learning (such as discussed

in Lane and Lubatkin 1998; Larsson et al 1998) from learning concerned with a higher level of unit of analysis (such as collective learning that has been addressed by regional competitive studies and co-evolution researchers, e.g Lawson and Lorenz, 1999; Konstadakopulos, 2000; Knight, 2002) According to such typology the locus of research in interorganizational learning, as defined by learning in the context of groups

or pairs of organizations that were proactively cooperating (Crossan et al., 1995; Larsson et al., 1998), did not shift to another level of unit of analysis, but is still concerned with company-level learning issues The key differentiations of interorganizational learning, as a consequence, focused on knowledge acquisition from other companies, and issues such as knowledge appropriation and trust became more central in the literature body Indeed, by comparing Levinson and Asahi (1995)’s work

on interorganizational learning process (where they proposed 4 steps, i.e 1) being

aware and identifying new knowledge, 2) transferring/interpreting new knowledge, 3) using knowledge by adjusting behavior to achieve intended outcomes, and 4) institutionalizing knowledge by reflecting on what is happening and adjusting alliance behavior) with Crossan et al (1995)’s organizational learning process, one may find the interorganizational learning framework was notably similar to Crossan et al (1995) and Holmqvist (2003)’s argument that organizational learning was behavioral, process focusing, and could be institutionalized by changes in routines and rules

2.2.3 Interorganizational learning through strategic alliance

Although interorganizational learning could take place in “interstices between firms,

universities, research laboratories, suppliers, and customers” (Powell et al., 1996,

pp.118), learning, as an enduring cognitive and behavioral change, necessitates a relatively stable and trustable party Partnership, as by definition implies a relatively long-term, reciprocal relationship, providing a good place for interorganizational

learning to happen Indeed, as summarized by Holmqvist (1999), “there have been

many theoretical approaches to the rise of organizational partnerships, including imaginary organizations (Hedberg et al., 1997), strategic alliances (Harrigan, 1988), virtual organizations (Hale and Whitlam, 1997), networks (Powell, 1990), dynamic networks (Miles and Snow, 1992), strategic networks (Jarillo, 1988), regional networks (Hanssen-Bauer and Snow, 1996), and others”(pp.419-420) Since the

current study was set to focus on individual companies as the subjects of learning, strategic alliance studies turned out to be the most appropriate literature to start with, and it can then be further extended to networks of firms, as elaborated in Section 2.3

Prisoner’s dilemma of firms’ interorganizational learning at dyadic level

Many studies emphasized on how firms could achieve successful alliances and

satisfied performances by behaving as “good” partners, and in consequence practical suggestions for managers were derived, examples including enhancing sharing power and lateral communication (Lei and Slocum, 1992), resolving confliction and building trust (Kale et al., 2000; Parise and Sasson, 2002) However, Hamel (1991) warned that companies behaving as “good” partners with high transparency and collaborative intent would endanger themselves by allowing more selfish partners with lower transparency and more competitive intent to exploit more on collective learning Yet if both of the parties would not cooperate, they would also suffer a loss of learning opportunities and hence potentially better performance Larsson et al (1998) applied game theories to solve such a prisoner’s dilemma under the interorganizational learning scenario Though promising results were derived, their focus was on two parties with non-cooperative gaming assumption, where the two partners could not achieve a powerful, obligatory agreement (Pekec, 2001) Such results can hardly be applied to a network of partners In addition, more often than not the non-cooperative assumption is not held true as strategic alliance studies showed that co-operative gaming arrangements, such as equity joint venture and joint development agreement, were more prevailing than short-term based, non-cooperative gaming forms, say R&D contracts As a matter of fact co-operative games could result in higher technological overlap and hence higher absorptive capacity (Mowery et al., 1996)

Levels of interorganizational learning between strategic alliances

Following classical organizational learning theories, learning within strategic alliance could also happen at multi-levels Jones et al (2003), for example, discussed how salespersons would learn from each other Since salespersons serve as the most frequent contact channel between two partners, such a study was illuminating Jones et

al (2003) also addressed how the individual learning could be leveraged to organization level by cultural construction

Focusing on team-level strategic learning between partners and adopting a longitude

case study, Wagner (2003) concluded that “interorganization project teams are the

mechanisms by which knowledge is converted into new joint capabilities”(pp.97).

Cross-functional teams were proposed as an efficient way of linking customers and suppliers to adapt to the external environment and exploring market opportunities

Noticeably studies from both levels of analysis emphasized the importance of internal knowledge sharing so that knowledge learned from partners could be institutionalized and create value for the firm

Broadening the view of strategic alliance to customers and suppliers

Besides strategic learning between bilateral partners that possess complementary skills and knowledge (Hamel, 1991), other studies added customers and suppliers to the scope of their studies For instance, McIvor and Humphreys (2004) extended traditional supply chain management literature from procurement and value-added activities to new product development and further addressed the prominent phenomenon of design outsourcing They highlighted advantages of early supplier involvement (ESI), as well as the barriers that impeded the integration process Notably, McIvor and Humphreys (2004)’s conclusions were in line with Mowery et al

(1996) in suggesting that “environments which are conductive to highly co-operative

relationships between buyers and suppliers are more likely to lead to supplier involvement”, and “co-operative relationships emphasizes a need to distribute

responsibility and power” (pp.181) In short, the literature in ESI not only broadened

the scope of strategic alliances to suppliers, but also supported the argument made earlier in this section that a long-term, reciprocal relationship could be conducive to interorganizational learning

Flint (2002) focused learning on customers’ side By treating the customer base as an entity where companies could learn from, he highlighted the necessity to identify customer-value-determination process and to keep track of the customer-value-change process By defining interorganizational learning as learning through problem solving with customers and suppliers, Schroeder et al (2002) also set their scopes in interorganizational learning to customers and suppliers It should be emphasized that they also stressed on a long-term supplier relationship and close customer interaction

as the indicators for successful interorganizational learning

Issue of trust under strategic alliance context

Realizing that knowledge appropriation and self-protection issues were becoming more central to the literature (e.g Hamel, 1991; Larsson et al., 1998), many researchers studied trust and commitment as a possible solution for the prisoner’s dilemma For instance, Kale et al (2000) tested two constructs, relational capital and conflict management, on their relationships with interorganizational learning performance and protection of proprietary assets They found that by relational capital building and conscious conflict management, alliance companies could achieve higher levels of learning and at the same time protect themselves Parise and Sasson (2002) explained that trust between partners reduces the need for strict monitoring of the alliance and time-consuming contract renegotiations And as discussed earlier in

customer and supplier relationship management, trust was a critical component

(McIvor and Humphreys, 2004; Schroeder et al., 2002; Wagner, 2003), and “Success

of the relationship between the companies was dependent on shared future transactions within a mutual cooperative culture The level of trust was directly related

to the degree of communication and integration” (Schroeder et al., 2002, pp.110)

2.2.4 Nexus between interorganizational learning and internal

organizational learning

While previous literature review showed a lot of research efforts have been devoted to interorganizational learning through strategic alliance, other researchers focused on the nexus of internal learning and external learning For example, based on a resource-based view, Schroeder et al (2002) suggested that there existed positive interactions between interorganizational learning and intraorganizational learning, and

in this sense they were two complementary tasks that when handled hand in hand could result in higher manufacturing performance (as shown in Figure 2-1) Yet they did not explicitly propose what could be the mechanisms that linked the two types of learning together and created synergy

Figure 2-1 Hypotheses represented as a homological network (Schroeder 2002)

A similar effort was observed in Caloghirou et al (2004)’s study By quantitative

analysis they showed that on one hand “the higher the levels of R&D efforts and

training within a firm are, the more the firm will be able to create and exploit

Internal learning

External learning

Proprietary processes and equipment

Manufacturing performance

novelty”(pp.31), and on the other hand “the more the firm uses the above (technical reports, use of patent databases, attendance at conferences, scientific publications, reverse engineering and use of the Internet) mechanism, the more is its openness to external sources of knowledge and the more it develops information and knowledge sharing with positive effects on its innovative performance”(pp.32) And they drew the

same conclusion with Schroeder et al (2002) that the two tasks of interorganizational learning and intraorganizational learning were not substitutes, but complements In other words, there existed positive interactions between organizational learning and interorganizational learning However, none of them explained where the positive interactions came from, and how to manage it

Holmqvist (2003, 2004) went a step further than the previous studies in terms of theoretical thinking and emphasized on the mechanisms for integrating the two types

of learning By examining Scandinavian PC Systems, he used exploitation and

exploration to represent the essence of intraorganizational learning and

interorganizational learning, as depicted in Figure 2-2:

Figure 2-2 A dynamic model of intra- and inter- organizational learning (Holmqvist 2003)

Exploitative extension

Exploitative internalization

Explorative extension

Explorative internalization

Opening-up Focusing

Joint opening-up

Joint focusing

Opening-up extension Opening-up internalization internalization Focusing

Focusing extension

By defining exploitation as “creating reliability in experience, and thrives on

productivity and refinement”; and exploration as “concerned with creating variety in experience, and thrives on experimentation and free association”, Holmqvist (2004,

pp.70) argued that “opening-up and focusing” were the two types of intermediary learning that tied up exploitation and exploration, and “extension and internalization” were the other two types of intermediary learning that tied up intra- and inter- organizational learning

Though a promising theoretical model, Holmqvist’s framework did not explicitly explain the differences between internal exploitation and external exploration mechanisms, and his works (1999, 2003, 2004) were based on a single case study, which necessitates further construct validation and exploration of the framework

2.2.5 Knowledge attributes and its relevance to interorganizational learning

The organizational learning literature has been further augmented by “the advent of the

now well-established knowledge management research stream” (Daghfous, 2004,

pp.70), and the increasingly used knowledge-based perspective (e.g Steensma and Corley, 2000) As a matter of fact, from a resource-based view the purpose of organizational learning is mainly concerned with knowledge accumulation tasks Gupta and Govingranjan (2000) disaggregated this accumulation process into knowledge creation, acquisition, and retention, which resembles Levinson and Asahi (1995)’s interorganizational learning procedures

Basic concepts in Knowledge Management

Though researchers might hold different views with regard to knowledge transfer process and the way of its incorporation with organizational learning, they do agree on

several concepts that underpin KM literature Grover and Davenport (2001) provide a framework for exploring KM theories in the favoring of interorganizational learning study In their influential work they addressed issues like tacit and explicit knowledge, knowledge processes, knowledge codification and personalization, and the level of knowledge By combining these commonly accepted and discussed concepts, Grover and Davenport (2001) proposed a pragmatic framework for KM research (Figure 2-3):

Figure 2-3 A pragmatic framework for KM research (Grover and Davenport 2001)

The purpose of presenting Grover and Davenport (2001)’s work here was to compare the KM research concepts and boundaries with those of organizational learning It is obvious from the above framework that these two research areas are closely related: 1) while organizational learning researchers advocate 4 levels of learning, there exist 4 levels of knowledge correspondingly; 2) while organizational learning consists of certain manageable processes (Mark, 1993; Levinson and Asahi, 1995), knowledge management further breaks down the process into several stages; 3) both of the literature streams stress on the strategic importance and both favor a resource-based view Therefore the different attributes of knowledge and its codification /

Knowledge process Generation Codification Transfer Realization

Individual

Group

Organization

personalization could serve as complementary literature to organizational learning and thus necessitates detailed review

Knowledge attributes and learning through alliance

As discussed above, knowledge attributes were arguably neglected in traditional organizational learning theory and were more developed through the knowledge management stream Thus along with the development in KM, more researchers began

to shift their focus to incorporate KM into organizational learning, and more specifically, due to the notable neglect of knowledge attributes in organizational learning, recent studies began to examine knowledge attributes and its relationship with organizational learning and strategic alliance (e.g Lubatkin et al., 2001; Kale et al., 2000; Contractor and Ra, 2002; Chen, 2004; Tsai, 2002)

By examining knowledge tacitness and governance difficulty, Lubatkin et al (2001) identified a type of learning that tended to be neglected by interorganizational learning theory, namely reciprocal learning (Figure 2-4):

Figure 2-4 Knowledge tacitness, governance difficulty, and learning alliances (adopted from

Knowledge Grafting Alliances (M&A)

Reciprocal Learning Alliances

Knowledge Absorption Alliance

the interorganizational learning activities was low, organizations tended to choose the vicarious learning form, referred as a “learning-by-watching” process However at the other extreme when both tacitness and difficulty were high, companies tended to choose knowledge grafting alliances, where acquiring knowledge by merger and acquisition (M&A) with partners was the favored option In the middle of the spectrum were the knowledge absorption alliances, which represent a more common situation where knowledge was not ready to be learnt and a certain amount of effort should be paid to ensure a stable and cooperative alliance Lubatkin et al (2001) highlighted that the reciprocal learning alliance was ideal for dealing with the mutual knowledge creation process, rather than the traditional knowledge transfer process And for partners whose core knowledge resided in distinctively different domains, this kind of learning alliance should be adopted to achieve better innovativeness

Contractor and Ra (2002) took a similar perspective to examine the relationship between knowledge attributes and learning alliance forms They found that four types

of knowledge attributes would affect the alliance governance modes, i.e codification, newness, complexity, and teachability By lining up different forms of alliance from

“one time contract” at the left extreme to “wholly owned subsidy” at the right extreme, they argued that the alliance form would more likely lean towards right extreme when knowledge is 1) more tacit, 2) more novel, 3) more complex; and 4) less teachable However, one can argue that actually codification, newness, complexity, and teachability are closely related and may not function as independent constructs for determining alliance governance modes

Knowledge attributes and issue of trust

As tacit knowledge is more difficult to transfer and necessitate extensive learning for its acquisition, researchers believed that an alliance involving tacit knowledge transferring and / or knowledge creation would favor close contact with partners and thus mutual trust became the most important issue For instance, Kale et al (2000)

stated that “know-how… is generally more sticky, tacit, and difficult to codify than

information and thus more resistant to easy transfer, both within and across firms”,

and “strong relational capital usually engenders close interaction between alliance

partners and thus facilitate exchange and transfer of information and know-how across the alliance interface” (pp.221) Chen (2004) showed statistical evidence in

that “trust and adjustment between partners is positively related to knowledge transfer

performance” (pp.314) Tsai (2002) also argued that while formal hierarchical

structure (in the form of centrality) might be negatively related to knowledge transfer performance, informal communication structures such as social interaction could be positively related to knowledge transfer performance because it could generate trust and reduce appropriation concerns

As the practical questions proposed in the introduction chapter were concerned with how organizations could learn from each other, traditional organizational learning literature was firstly examined However, this literature body was mainly concerned with learning within organization boundaries, and as Crossan et al (1995), Larsson et

al (1998), and Lane and Lubatkin (1998) have criticized, little attention was paid to learning dynamics that resided between organizations Interorganizational learning, a stream of literature derived from organizational learning but relatively separated from its origin (Holmqvist 2003), was examined in expectation of covering research gaps in organizational learning Moreover, network learning was differentiated from

interorganizational learning in order to restrict the unit of analysis of the current study

to the company level, and thus avoid potential confusion and incoherence Since a long-term, reciprocal relationship provides an ideal place for knowledge transfer and interorganizational learning (McIvor and Humphreys, 2004), learning under strategic alliances turned out to be a major body of interorganizational learning and was reviewed extensively Although this literature body covered many practical issues such

as competitive cooperation (Hamel, 1991), appropriation and a learning dilemma (Larsson et al., 1998), different levels of interorganization context (Jones et al., 2003; Wagner, 2003), stable supplier and customer base that resembled a strategic alliance (McIvor and Humphreys, 2004; Flint, 2002), and trust (Kale et al., 2000), four key factors were neglected Firstly, strategic alliance studies were mainly concerned on learning between two parties (e.g Larsson et al., 1998; Jones et al., 2003), despite the

fact that in the real world dyadic partnership was “being replaced by one consisting of

complex networks of collaborating organizations, and chains of buyers and suppliers”

(McIvor and Humphreys, 2004, pp 179) Secondly, organizational learning theories were poorly incorporated into the interorganizational learning level, and even though Holmqvist (1999, 2003, 2004) persistently tried to bridge the “separated worlds”, little was understood on what were the positive interactions that existed between organizational learning and interorganizational learning, where they came from, and how to manage them Thirdly, knowledge attribute and its relationship with strategic alliance and organizational learning were neglected at the beginning (Daghfous, 2004), and efforts are needed to continuously track the advances in both streams of literature and further bridge them Fourthly, while much emphasis was devoted to knowledge appropriation, the above literature did not explicitly explain why certain companies could perform better than others even if they treated each other in an equally open and

collaborative manner, i.e with the same degree of protection and learning Studies in social network literature and absorptive capacity were then examined in an attempt to address the above gaps

2.3 Social network literature

As discussed in the previous section, though some studies (e.g Larsson et al., 1998; Kale et al., 2000; Oxley and Sampson, 2004) addressed the dilemma of interorganizational learning and protection, they examined the phenomenon under the setting of dyadic strategic alliance However, these dyadic level studies neglect the possible impact of social networks on learning and protection According to Powell (1990), social networks may contribute critically to organizational learning because it

is the most efficient organizational arrangement for sourcing information, and traditional mechanisms such as market and hierarchy are less efficient due to the difficulty in pricing information (for market mechanism) and bureaucracies (for hierarchy mechanism) Therefore social network literature was examined for the purpose of both extending interorganizational learning from dyadic level to network level, and complementing the deficiencies in market and hierarchy mechanisms Specifically, studies relating network structures with organizational learning were studied in detail In this literature body contradictory results were reported, and though recently some efforts were devoted to reconcile such contradiction, further research gaps still remain unexplored

2.3.1 Contradictory results from studies combining organizational learning

with social network theories

Several studies have implicitly referred to organizational learning and knowledge transfer by associating network structures with organization performance Tsai (2001), for instance, argued that network centrality is positively related to a business unit’s

innovativeness and business performance because a central position “reveals its ability

to access external information and knowledge” (pp 997) Similarly, Baum et al (2000)

found that for biotechnology startups, the larger the alliance network size, and the more ‘efficient’ the network, the higher the performance They reasoned that the larger

size and more efficient network could provide “access to strategic and operational

knowhow … with minimum costs of redundancy, conflict and complexity” (pp 270) In

both cases, interorganizational (as in Baum et al., 2000) and inter-unit learning (as in Tsai, 2001) were assumed to be the causal mechanism linking network structure to firm/ business unit performance However, neither study addressed the detailed learning processes, let alone knowledge protection activities

Other researchers have also referred to organizational learning by studying the links between network structures and innovation Shan et al (1994) argued that the more cooperative relationships a startup possessed, the greater the innovation output; and the wider the range a startup participated in, the greater the innovation output Such an observation was explained through the firm’s access to resources and information, which might not be evenly distributed within the network Similarly, Soh and Roberts (2004) examined the U.S computer networking market and drew the conclusion that both the number of cooperative relationships and centrality of the firm would positively relate to innovation performance This statement was put forward by arguing that cooperative relationships could bring in experience and knowledge to the focal firm However, though the link between network structure and innovation was examined in both studies, they did not explicitly model the association between network structure and interorganizational learning process, and therefore could hardly provide a fine-grained theoretical explanation for effects of network structures on

innovation

A third category of studies that exemplified the relationship between network structures and learning process were concerned with the network effects on product development Hansen (1999), for instance, examined the impact of tie strength on project completion time Following the same thread, Hansen (2002) further tested the effects of direct / indirect ties on project completion time The fundamental idea was that tie strength/ network settings should match with knowledge search and transfer activities and knowledge tacitness Similarly Batjargal (2005) examined the effects of two other network structure elements - network density and homogeneity, on project completion time He suggested that dense networks could facilitate knowledge transfer and hence reduce the product development duration However, the path that bridges network structure and interorganizational / inter-unit learning process is still missing in this literature

In general, while organizational learning and knowledge transfer were implicitly assumed to be the causal relationship between network structure and performance and innovation, it is surprising to note that few studies have examined in detailed attention the underlying process of interorganizational learning, let alone knowledge protection One of the possible reasons could be that such analysis necessitates a firm-level

investigation As Zaheer and Bell (2005, pp 811)) have argued, “one important but

understudied aspect of network research is consideration of focal firm capabilities in explaining network outcomes at the firm level” Methodologically speaking,

researchers could get access to network-level data (e.g inter-firm relationships) by examining industrial journals and periodicals, and in many cases firm innovativeness

may also be assessed through the proxy of aggregated patent data However, how firms get to know sources of new knowledge, how they manage to transfer it, and what firms

do with their acquired knowledge are not observable from analysis only with public data input Therefore I argue that incorporating these within-firm level constructs into research framework could significantly increase the credibility of the model and factors such as individual firms’ internal learning capabilities could then be modeled and avoid confounded results

I further argue that studies overlooking such detailed process may yield contradictory results due to the distinctions between different stages of learning and protection A recent review on networks and organization by Brass et al (2004) has actually highlighted some theoretical contradictions on the effects of network structure on innovation and organization performance The most well-known one would be Coleman (1988)’s proposition on network closure, versus Burt (1992)’s structure holes theory According to Coleman, a close network would generate trust among the companies, which in turn indicate positive effect of information access (Ahuja, 2000), ease the transfer of knowledge (Reagans and McEvily, 2003) and hence increase innovation rates (Ahuja, 2000) In contrast, Uzzi (1997) argued that embedded ties would decrease the width of search range and an over-embedded network should not

be recommended Burt (1992) suggested that firms embedded in sparsely connected networks would enjoy efficiency and brokerage advantages based on their ability to arbitrage non-redundant information exchanges (the power of betweenness) Baum et

al (2000) also found support for the structure holes view by showing positive effects

of networks giving access to diverse information

2.3.2 Recent efforts devoted to reconcile the contradictions and further gaps

Recently some research efforts have been devoted to reconcile these contradictions Burt (2000), for instance, further extended the structural holes theory by suggesting that the two forms of social capital, namely network closure and structural holes, would not necessarily be contradictory, but rather play different roles He argued that network closure could serve as one of the significant contingency factors for extracting the value of brokerage, and the highest performance might be achieved when network closure within the group is high and structure holes outside the group are rich Baum and Ingram (2000) elaborated on the different roles and purposes of network structures, and proposed that close networks served the purpose of knowledge exploitation, and structural holes would be more suitable for knowledge exploration Brass et al (2004)

suggested that by “embedding networks into structures that generate trust” (pp 806)

could solve the tension between Coleman’s close network stance and Burt’s structure holes theory This implies that by studying structural embeddedness together with relational embeddedness could reveal significant theoretical explanation for the debates

These recent efforts in reviewing social network studies (e.g Brass et al, 2004) and exploring contingency factors (e.g Burt, 2000, Baum and Ingram, 2000, Rowley et al, 2000) have made the social network literature internally more consistent and externally more powerful in explaining a broad domain of environmental circumstances However, there still existed a noticeable gap in empirical research on knowledge protection, despite its theoretical and practical importance (McEvily et al 2004), and studies have yet to be done on directly examining the effects of network structures on both learning and protection at inter-organization level Furthermore, how knowledge

protection would moderate a firm’s competitive advantages is not modeled in social network studies Under such circumstances, it is difficult to determine whether the ease

of knowledge transfer (Reagans and McEvily, 2003) or protection of knowledge (Kale

et al, 2000) should be the main benefit a firm could gain from a closure network Only

by examining learning process together with knowledge protection would make it possible to disentangle the complexity of these two activities and their relationship with a firm’s competitive advantages These gaps highlight the importance of a fine-grained examination not only on knowledge sharing, but also on knowledge protection

I mentioned that examining organizational learning from a social network perspective would contribute to the literature Furthermore, there are also distinctive advantages in examining social networks from the perspective of knowledge-based theory (KBV),

for “it has the potential to explain and predict different forms of interfirm

collaboration across a broad domain of environmental circumstances” (Grant and

Baden-fuller, 1995; pp 17) Though several social network studies incorporated elements of KBV (e.g Hansen, 1999; 2002), few studies have looked at the full process of interorganizational learning and protection It is crucial to examine such a fine-grained process because there exist distinctive differences between each stage of the process (ref Spender, 1992), and overlooking such distinctions in research may lead to contradictory propositions

2.4 Absorptive capacity

When two companies learn together but result in different levels of technical innovativeness, it is straightforward to assume that the two learning parties have

absorbed the mutually created knowledge at a different degree Absorptive capacity, an organization-level construct first developed by Cohen and Levinthal (1990), was also addressed from time to time in organizational learning literature For instance, in measuring a company’s capacity in learning new knowledge, Larsson (1998) used the term “receptivity” that functioned similarly with absorptive capacity Several strategic alliances studies also highlighted the importance of absorptive capacity in improving performance (Levinson and Asahi, 1995; Mowery et al., 1996) Given the crucial role played by absorptive capacity in bridging interorganizational learning with internal organizational learning, it was highlighted and discussed separately

In the next section Cohen and Levinthal’ s classic study was first examined, and related topics such as relative absorptive capacity (Lane and Lubatkin 1998), potential and realized absorptive capacity (Zahra and George 2002) were discussed subsequently

2.4.1 Classical absorptive capacity model

Cohen and Levinthal (1990) created absorptive capacity as a new construct, which

referred to “ability to recognize the value of new information, assimilate it, and apply

it to commercial ends” (pp.128) Originated from individual cognitive study, it was

justified that the individual’s previous related knowledge was required to assimilate and use new knowledge Cohen and Levinthal argued that this was also true on an organization level In bridging these two levels of cognitive capacities, they emphasized that 1) on an organization level, absorptive capacity not only referred to acquisition or assimilation of information, but also to the organization’s ability to exploit it; and 2) because an organization is a collective body of individuals with different roles, absorptive capacity was not only the function of the gatekeeper’s

searching and disseminating skill, but also the function of “expertise of those