Inter-Organizational Knowledge Transfer Difficulty: The Influence of
Organizational Network Type, Absorptive Capacity, Causal Ambiguity
and Outcome Ambiguity
A Dissertation Submitted
In Fulfillment of the Requirements for the Degree of
Doctor of Philosophy in Business Administration
J. Mack Robinson College of Business,
Georgia State University
Jennifer Lewis Priestley
Defended December 13, 2003
Table of Contents
Page
List of Tables………………………………………………………………………….. iv
List of Figures…………………………………………………………………………. v
List of Appendices…………………………………………………………………….. vi
Abstract……………………………………………………………………………….. vii
1 Introduction………………………………………………………………………
1
1.1 Guiding Research Question……………………………………………….
3
2 Theoretical Basis, Literature Review and Hypotheses……………………….
2.1 Theoretical Basis…………………………………………………………
2.1.1 Transaction Cost Economics…………………………………..
2.1.2 Knowledge Based View of the Firm…………………………..
2.2 Knowledge Transfer Literature Review………………………………….
2.2.1 Factors of Knowledge Transfer Difficulty – Part I of Specified
Model………………………………………………………….
2.3 Review of Inter-organizational Network Types…………………………
2.4 Inter-Organizational Network Types and Factors of Knowledge Transfer
Difficulty – Part II of Specified Model…...………………………………
2.5 Chapter Summary………………………………………………………...
5
5
5
8
11
13
31
41
55
3 Methodology…………………………………………………………………….
58
3.1 Construct Operationalization Review and Approach….....………………
3.1.1 Knowledge Transfer Difficulty……………………………..….
3.1.2 Absorptive Capacity……………………………..……………..
3.1.3 Causal Ambiguity………………………….…………………..
3.1.4 Outcome Ambiguity……………………………………………
3.2 Study Approach...………………………………………………………...
3.3 Data Collection…………………………………………………………...
3.3.1 Sample Description…………………………………………….
3.4 Data Analysis and Methodology…..……………………………………..
58
58
60
61
63
63
65
65
66
4 Quantitative Data Analysis and Results.…………………………………...…
70
4.1 Pilot Study Results..……………………………………………………..
4.2 Field Study Results..…………………………………………………….
4.2.1 Description of Populations and Responses…………………..
4.2.2 Assessment of Field Study Scale Validities………………….
4.2.2.1 Validation of Reflective Measures…………..
70
74
74
76
77
ii
4.2.2.2 Validation of Formative Measures…………..
Hypothesis Testing…………………………………………..
4.2.3.1 Hypothesis Testing – Part I………………….
4.2.3.2 Hypothesis Testing – Part II…………………
Non Response Bias and Specification Bias Testing.………...
78
79
80
82
89
5 Discussion and Interpretation of Results……………………………………..
5.1 Absorptive Capacity.…………………………………………………….
5.2 Causal Ambiguity………………………………………………………..
5.3 Outcome Ambiguity……………………………………………………..
91
91
94
96
6 Conclusions……………………………………………………………………..
6.1 Study Limitations………………………………………………………..
6.2 Study Implications……………………………………………………….
6.2.1 Implications for Practitioners………….………………….…
6.3 Suggestions for Further Research………………………………………..
98
98
99
99
101
4.2.3
4.2.4
iii
List of Tables
Page
TABLE 1. Summarization of Knowledge Transfer Difficulty Studies………………..108
TABLE 2. Summarization of Causal Ambiguity Studies…………………………….. 109
TABLE 3. Summarization of Measurement Items from Previous Studies…………… 110
TABLE 4. Interviewee Contact Information…………………………………………..112
TABLE 5. Changes Made to Survey Instrument………………………………………113
TABLE 6. Pilot Study Scale Reliabilities……………………………………………...116
TABLE 7a. Descriptive Statistics for Field Study Respondents(SunTrust)…………..118
TABLE 7b. Descriptive Statistics for Field Study Respondents(CUNA)...…………..119
TABLE 7c. Descriptive Statistics for Field Study Respondents(CU “Independents”)..120
TABLE 8. Field Study Scale Reliabilities...........……………………………………..121
TABLE 9. Intercorrelations for Field Study…………………………………………..122
TABLE 10a. Item-to-Construct Correlation Matrix.………………………………….. 123
TABLE 10b. Sub Construct-to-Construct Correlation Matrix………………………....124
TABLE 11. Hypothesis Testing Results (Part I)…...………………………………….125
TABLE 12. Hypothesis Testing Results (Part II)…...…………………………………126
TABLE 13. Regression Model Results (Part I)…………...…………………………...127
iv
List of Figures
Page
FIGURE 1. Causal Ambiguity………………………………………………………….128
FIGURE 2. Outcome Ambiguity Framework…….…………………………………….129
FIGURE 3. Factors Which Influence Knowledge Transfer Difficulty…………………130
FIGURE 4. Differentiation Among Network Types……………………….……….….131
FIGURE 5. Specified Model Parts I and II: ……………………………...…………….132
FIGURE 6. Construct Development Methodology………...………………………..…133
v
List of Appendices
APPENDIX I Interview Logs……………………………………………………….141
APPENDIX II Explanation of Measurement Items………………………………....147
APPENDIX III Pilot Study Survey Instrument…………………………………….148
APPENDIX IV Field Study Cover Letter Examples and Survey Instrument……….161
vi
Abstract
While most articles dealing with the issue of inter-organizational knowledge transfer take
the position that some type of network alliance is superior to independent entities for the
purposes of knowledge transfer, limited theory, or empirical research, addresses how
different network types experience the factors of knowledge transfer differently. This
dissertation addresses this issue. Specifically, this study develops two distinct parts of a
specified model. The first part of the model identifies the factors expected to affect
knowledge transfer difficulty, including the introduction of a new factor – outcome
ambiguity – and puts forth hypotheses regarding their respective relationships. The
second part of the model hypothesizes how each of these factors is expected to vary with
four different network types. A total of nineteen hypotheses are developed, and eleven
are tested empirically using primary data.
Research overseen by Subhashish Samaddar – Committee Chair
vii
1. INTRODUCTION
Increasingly, knowledge is recognized as not just one of many assets to be managed, but
rather as the single most important asset of the firm. The concept that a firm’s
competitive advantage flows from its unique knowledge base, and then the degree to
which the firm recognizes and values this knowledge as a resource, has been the theme of
many papers on competitive advantage (Prahalad and Hamel, 1990; Barney, 1991;
D’Aveni, 1994; Teece and Pisano, 1994; Spender, 1996; Nonaka and Teece, 2001). For
example, Teece (1998) has identified knowledge as the “essence of modern economic
growth” and “the underpinning of competencies”. Drucker (1995) has argued that
knowledge is quickly displacing natural resources, capital and labor as a firm’s basic
economic resource. Unlike other resources, knowledge will grow rather than diminish
with use. Knowledge tends, therefore, to play an increasingly central role in economic
development over time (Adler, 2001).
The ability to value and leverage knowledge external to the organization, in addition to
an organization’s own knowledge, has become a critical component to a firm’s
innovative assets, in part because this “second-hand” experience can be obtained faster
and more cheaply than “first-hand” experience (Hamel, 1991; Huber, 1991). Networked
alliances represent an important channel through which to access this “second hand”
knowledge. As Stuart observes,
“As a reason to enter alliances, the potential to learn from partners highlights the
fact that alliances are access relationships. Just as scholars of social networks
have observed that social ties purvey access to information possessed by one’s
contacts, so have alliance researchers recognized that strategic coalitions can
convey access to the resources or know how possessed by one’s partners”.
(2000:792)
1
These networks and partnerships represent a critical source of knowledge external to a
participating firm. For instance, Hamel et al. (1989) studied alliances among entities
within the same industry for 15 years. Over this period, the authors identified dozens of
examples in Japan, the US and Europe, where multi-firm alliances (including alliances
with competitors) represented the lowest cost alternative to gaining access to knowledge.
Because knowledge transfer among organizations represents an important source of
knowledge, an entire literature exists examining and isolating the factors of knowledge
transfer difficulty. These factors include absorptive capacity, as defined by Cohen and
Levinthal (1990) and ambiguity, as defined by Mosakowski (1997), Wilcox-King and
Zeithaml (2001) and Barney (1997), among others.
Although researchers in organizational learning have effectively concluded that
networked organizations will realize superior economic gains from their increased access
to knowledge within a network relative to independent or non-aligned firms (Argote,
1999; Carlsson, 2002; Uzzi, 1996; Darr et al., 1995; Powell et al., 1996; Baum and
Ingram, 1998), there has been no attempt to link the different forms that networks can
assume with these factors of knowledge transfer difficulty. This dissertation attempts to
make this link.
Given the recognized importance of knowledge and its transferability, firm managers
would be well served to understand how the factors of inter-organizational knowledge
transfer are affected by network type. Where knowledge is sticky and transfer is difficult,
the implications may threaten a firm’s long-term competitiveness, including new
enterprise formation, the exploitation of technological know-how and the successful
2
development and commercialization of new products and services (Teece, 1998).
Therefore, a better understanding of the factors that impede or enhance interorganizational knowledge transfer, and how these factors are affected by the type of
network in question, can be critical to a firm’s competitiveness.
1.1 Guiding Research Questions
Some network researchers have concluded that “well-structured” networks are the basis
of economic gains (Burt, 1992). Although the positive implications of having a structure
versus not having a structure are generally accepted, the implications of different network
types are less understood. For example, in their study of the biotechnology industry,
Powell et al. (1996) found that organizations embedded within R&D networks generated
more scientific papers, with more citations per paper, and generally experienced greater
sales than did similar firms that were less integrated within a network. Darr et al. (1995)
found that members of a pizza franchise experienced shorter operational learning times,
as measured by decreasing unit costs, than did similar pizza outlets not part of the same
franchise. Ingram and Simons (2002) found that experiential knowledge transfer was
greater among kibbutzim (the plural of “kibbutz”) within the same federation (organized
network), while the kibbutzim outside of a federation did not demonstrate the same
degree of experiential knowledge transfer, with negative economic consequences. In
1999, the multi-billion dollar Mars Climate Orbiter was lost in space because some
engineers from one of several entities involved in the project incorrectly entered data in
English units instead of metric units (Postrel, 2002). The mistake was attributed in large
part to a lack of knowledge transfer among engineers.
3
These four studies examined a network of biotechnology firms, a network of pizza
franchises, an agricultural network of kibbutzim, and a group of engineers respectively –
clearly a highly diversified set of networks. Although all four networks would be
considered “well-structured”, their network forms are actually quite different. The
networks accommodated, for example, different levels of competition, different types of
governance policies, and had different objectives.
The success of knowledge transfer within each of these networks varied. This variation
raises an important question for researchers and practitioners – Are the effects of the
inter-organizational network type on the factors of knowledge transfer difficulty
invariant?
Specifically, the two guiding research questions of this dissertation are:
1. What are the relative effects of absorptive capacity and ambiguity on
knowledge transfer difficulty?
2. How do the factors of absorptive capacity and ambiguity vary with interorganizational network type?
4
2.0 THEORETICAL BASIS AND LITERATURE REVIEW
2.1 Theoretical Basis
Firms organize, in large part, to mitigate risk and uncertainty, and increase the
efficiencies of resource access and utilization. Two theoretical perspectives that scholars
have used to explain and predict how firms organize include Transaction Cost Economics
and the Knowledge-Based View of the Firm. These two theories provide the foundation
for this dissertation.
2.1.1 Transaction Cost Economics
In his address to the Royal Economic Society in 1986, R.C. Matthews noted that, “…the
economics of institutions has become one of the liveliest areas in our discipline”
(Matthews, 1996). He went on to highlight that the study of economic institutions was
based on two concepts – (i) institutional form does matter and (ii) the determinants of
institutional form are susceptible to analysis by economic theory. According to Boerner
and Macher (2003), this assessment was based upon the contributions of transaction cost
economics (TCE).
The basic insight of TCE is recognition that in a world of minimizing transaction costs,
exchange agreements must be governed and, contingent on the transactions to be
organized, some forms of governance are better than others. According to Williamson
(1973) and work extended by Heiman and Nickerson (2002), three of the primary factors
underlying transaction costs, include: (i) the bounded rationality of individuals (and
5
firms); (ii) the risk of opportunistic behavior; (iii) the impactedness of information. All
three factors are affected by inherent uncertainty in exchange.
First, individuals (and firms) are limited by bounded rationality – there is a limit to the
ability to plan for the future, and in spite of best efforts, to deal with the complexity and
uncertainty of the environment. Individuals and firms lack the knowledge, foresight
and/or skill to accurately predict and plan for all the various contingencies that may arise.
In the context of knowledge, it can be said that individuals and firms cannot know all that
is knowable.
Second, transactions are subject to opportunistic behavior by one or more parties. The
most commonly recognized form of opportunism is the use of asymmetrically distributed
information by at least one party to the disadvantage of the first party. As a result,
transactions must be governed (either implicitly or explicitly), which has associated costs.
Where TCE considers the costs of opportunistic behavior, the Knowledge Based View
(see below) does not.
Third, transactions between parties may be subject to the “impactedness” of information,
which is a condition of knowledge asymmetry. When describing this concept,
Williamson places particular emphasis on the term “impactedness”,
It is costly for the party with less information to achieve information parity. To
the extent that it is difficult to distinguish between agents who disclose the
impacted information to which they have access in an opportunistic manner from
those who make good faith representations, agents of the latter type may be
induced to withdraw from the market (1973:318).
6
The concept of information impactedness is particularly important for this dissertation,
because it provides a theoretical foundation for the two types of ambiguity factors of
knowledge transfer difficulty - Causal Ambiguity (the inability to isolate the influence of
factors or inputs on output) and Outcome Ambiguity (the inability to predict an outcome,
in part, because the motivations of the knowledge recipient are unknown). These
concepts will be explored in more detail in Section 2.2.1.
TCE dictates different organizational structures to minimize transaction costs based upon
bounded rationality, risks associated with opportunistic behavior, information
impactedness and general uncertainty. In part, these factors can be used to explain why
transactions shift from a market-orientation to an intra-organizational orientation.
Williamson (1973) simplifies this discussion through his evaluation of two general
governance forms: peer group associations and simple internal hierarchies.
Peer group associations are defined by Williamson as nonhierarchical associations of
individuals or groups, without subordination, involving collective and usually
cooperative activity. The advantages to peer group arrangements include mitigation of
risk, pooled resources and selective and limited membership. Peer group associations
address the issue of costs associated with bounded rationality by extending a firm’s
access to resources to encompass the resources of the entire group. However, costs
related to opportunism and information impactedness are not addressed through this
form. In addition, peer group associations are vulnerable to free-rider abuses from some
members.
7
The second form, simple internal hierarchies, effectively represents the decision to bring
resources within the boundaries of the firm. Although a hierarchical form will provide
the authority to address issues related to opportunism and information impactedness, the
firm continues to be limited by bounded rationality. This concept of hierarchical
authority and its effects on opportunism and impactedness plays a contributing role in the
development of the inter-organizational network forms, where a parallel is intended to be
drawn between the boundaries of the firm and the boundaries of the network.
2.1.2 Knowledge Based View of the Firm
The second theoretical foundation for this paper comes from the Knowledge-Based View
(KBV) of the Firm. Where TCE is grounded in economic theory with influences from
research in strategy, the KBV is a theory grounded in strategy with influences from
economics. The more general Resource-Based View perceives the firm as a bundle of
idiosyncratic resources and capabilities where the primary task of management is to
maximize value through the optimal deployment of existing resources and capabilities,
while developing the firm’s resource base for the future (Barney, 1986, 1991). To the
extent that the KBV focuses upon knowledge as the most strategically important of the
firm’ resources, it is an extension of the resource-based view (Grant, 1997).
Kogut and Zander (1996) argue that a firm can be understood as a social community
specializing in increasing the speed and efficiency of both the creation and transfer of
knowledge. The central competitive dimension of what firms know how to do is to create
8
and transfer knowledge efficiently within an organizational context. Hierarchy, in Kogut
and Zander’s view, offers advantages over markets for the transfer of knowledge. This is
true in part because internal organizations can create identity that leads to social
arrangements that support coordination and communication. These hierarchy-enhanced
coordination and communication capabilities allow firms to create and internalize
complex, tacit knowledge, again in a way that markets cannot.
Grant (1997), an important contributor to the KBV, identifies three mechanisms for
transferring and integrating knowledge. The first mechanism is through rules and
directions, which works best in a highly organized hierarchical structure. Thompson
(1967) referred to a similar concept as “pooled interdependence”. The efficiency of rules
and directions in achieving coordination extends a firm’s ability to improve and lower
costs of communication, and therefore improve the efficiency of knowledge transfer. As
recognized by Demsetz (1991),
“…direction is a low cost method of communicating between specialists and the
large number of persons who are non specialists or specialists in other fields”.
Rules enable the conversion of tacit into explicit knowledge. Nonaka (1991) refers to
this conversion process as “Externalization”.
The second mechanism that facilitates the transfer and integration of knowledge is
sequencing. Grant describes sequencing as “…the simplest means by which individuals can integrate their specialist
knowledge while minimizing their communication and continuous
coordination…” (1997:115)
9
Sequencing occurs when production activities are organized in a time patterned sequence
such that each specialist’s input occurs independently – each specialist makes an
independent contribution to a process or product.
The third knowledge transfer mechanism occurs through routines. An organizational
routine is,
“…a relatively complex pattern of behavior…triggered by a relatively small
number of initiating signals or choices and functioning as a recognizable unit in a
relatively automatic fashion” (Winter, 1987:165).
While routines may be simple sequences, an interesting feature is their ability to support
complex patterns of interactions between individuals in the absence of rules or directives,
or even verbal communications. As a result, routines are highly communication-efficient.
Examples of organizations operating under routines include the navigation crew of a ship,
a surgical team and an auto racing pit crew (Grant, 1996). In these instances, knowledge
transfer is occurring with minimal verbal communication. Thompson (1967) describes a
similar concept as one of “mutual adjustment”.
While these mechanisms for knowledge transfer are necessitated by knowledge
specialization, all depend upon some basis of common knowledge. Grant describes
common knowledge,
“At its most simple, common knowledge comprises those elements of knowledge
common to all organizational members: the intersection of their individual
knowledge sets. The importance of common knowledge is that it permits
individuals to share and integrate aspects of knowledge which are not common
between them”. (1997:115)
10
As with TCE, the KBV plays an important role in supporting the foundational concepts in
this dissertation. Specifically, the idea that firms will organize to minimize the costs
associated with knowledge transfer is a basis for the four inter-organizational types
described in detail in Section 2.3. In addition, the mechanisms described by Grant (1997)
provide a basis for explaining how knowledge transfers (or does not transfer) between
organizations. And, his statement regarding common knowledge will be further
developed in my explanation of absorptive capacity and its significance to interorganizational knowledge transfer difficulty in Section 2.2.1.
2.2 Knowledge Transfer Literature Review
An important point from the KBV described above, is that knowledge is viewed as
moving unencumbered by and transferring without cost within and among organizations
(Grant, 1997; Kogut and Zander, 1992, 1996); although knowledge is recognized as an
asset, unlike other assets its transferability has no associated costs. As von Hippel
described (1994), this may not be the case.
Knowledge has been described as a “sticky” asset that is costly to acquire and difficult to
transfer between locations, even within the boundaries of a single firm. This stickiness is
caused by, among other factors, the form of knowledge being transferred (Is the
knowledge in question explicit? Or is it tacit?), as well as different attributes of the
source(s) and the recipient(s), such as their situational absorptive capacity, their
respective levels of causal ambiguity and the degree of trust or motivation to share
between the source and the recipient (von Hippel, 1994; Szulanski, 1996).
11
In part because of the broad operational definitions of the concept, the study of
knowledge transfer represents a unique challenge for researchers and has been studied
from several vantage points. For example, in his widely cited work on the intraorganizational transfer of best practices, Szulanski (1996) surveyed eight firms across
different industries and determined that the degree to which knowledge transferred was
primarily a function of absorptive capacity, causal ambiguity and the working
relationship between source and recipient. Von Hippel (1994) looked at the cost
implications of knowledge transfer difficulty, and found that tacitness, absorptive
capacity and specialization were important factors influencing knowledge transfer. Using
the work of von Hippel (1994) and Szulanski (1996), Tsai (2001) studied knowledge
transfer across business units in a single firm. He found that a combination of absorptive
capacity and “centrality” of organizational position were the primary contributors to
knowledge transfer. Similarly, Hansen (2002) used survey data collected from business
units of a multi-unit electronics company and found centrality to be a key determinant of
knowledge transfer difficulty, particularly when the knowledge in question was tacit.
Lapre and Wassenhove (2001) and Darr et al. (1995) studied learning curves within
manufacturing environments as indicators of knowledge transfer. Lapre and Wassenhove
(2001) found management “buy in” and knowledge diversity as the factors of knowledge
transfer, while Darr et al. (1995) found network membership to be the primary factor of
knowledge transfer. A summary of studies from the topic of knowledge transfer can be
found in Table 1.
Based upon these studies, when knowledge cannot be transferred from one location to
another, the organization, or network of organizations, may experience negative
12
implications. These implications have been described in terms of the incremental
expenditures required to transfer knowledge from source to recipient (von Hippel, 1994)
as well as the extent to which ad hoc solutions are required or current practices need to be
adapted to facilitate transfer (Szulanski, 1996).
2.2.1 Factors of Knowledge Transfer Difficulty – Specified Model Part I
In this section, drawing from the works of multiple researchers studying the concept of
knowledge transfer difficulty, factors are identified and developed that have an influence
on the extent to which knowledge would be expected to transfer (or not) interorganizationally. These factors and their hypothesized relationships with knowledge
transfer difficulty form Part I of the specified model for this dissertation (Figure 3) and
are explored in turn.
Absorptive Capacity Organizations must possess some degree of absorptive capacity to
first recognize and then realize any value from the external knowledge to which it is
exposed as a member of a network. The concept of absorptive capacity has received a
significant amount of research attention since Cohen and Levinthal’s (1990) seminal
work on the topic. Their definition of the concept is the most widely cited,
“…the ability of a firm to recognize the value of new, external information,
assimilate it and apply it to commercial ends is critical to its innovative
capabilities. We label this capability as a firm’s absorptive capacity…”
(1990:128)
Lane, Koka and Pathak (2002) provide a thematic analysis of 189 different papers that
have studied absorptive capacity from a variety of different perspectives. All of the
papers they studied utilized the Cohen and Levinthal definition of the concept. Other
authors have specifically researched the influence of absorptive capacity on knowledge
13
transfer (Szulanski, 1996; Van den Bosch, 1999; Lane and Lubatkin, 1998; Boynton,
1994; George et al., 2001). All of these authors agree that the recipient’s absorptive
capacity is critical to an effective transfer of knowledge in an intra-organizational
context. If absorptive capacity is considered to be low, knowledge “stickiness” and its
associated costs are expected to be high. Teece (1998) emphasizes that only in those
circumstances where the relevant knowledge is fully understood and appreciated by both
parties can knowledge transfer be collapsed into a simple problem of an inexpensive
transaction. Otherwise, knowledge transfer is often difficult and expensive.
In a networked context, the absorptive capacity of the recipient organization is integral to
the success of the knowledge transfer process. In his work examining the effectiveness
of inter-organizational alliances, Walker (1995) argues that firms that emphasize their
relationships with other firms will be more successful, in large part because of their
ability to recognize and apply new knowledge,
“Firms operated to acquire extensive (external) technological information more
rapidly…and designed to process (external) technological information
better…will be more innovative and profitable.” (195:116)
This ability to sense new external knowledge and have the processes in place to then
bring it internal to the organization quickly becomes a competitive advantage when
translated into economic rents. This “sensemaking” is a critical function (Teece, 1998)
that enables the organization to more effectively connect with its environment and
allocate resources efficiently. However, developing this capacity represents a unique
challenge to most organizations. The paradox of absorptive capacity is that an
organization that does not have it may not understand that they need it (i.e., their
knowledge is incomplete and can be augmented). Organizations with low absorptive
14
capacity, arguably those with the least amount of knowledge, will be less likely to value
external knowledge,
“…the decision maker [or organization] may not know enough to estimate the
costs of his [their] ignorance…it will be difficult to evaluate knowledge [for
acquisition] in the future without possessing this knowledge during the
evaluation.” (Mosakowski, 1997:437)
Cohen and Levinthal (1990) highlight the importance of possessing certain
commonalities to facilitate absorptive capacity and the effective sharing of information
among organizations. Four types of commonalities have been suggested by Cohen and
Levinthal (1990) and others (e.g., Lane and Lubatkin, 1998) to represent the primary
contributors to a recipient’s overall level of absorptive capacity; language, knowledge
base, process, and problem solving.
The first of these contributors involves the commonality of language. Investments in
communication codes – or common language – have been shown to increase the speed
and lower the cost of knowledge transfer (Cohen and Levinthal, 1990). Following from
the KBV discussion above regarding mechanisms for knowledge transfer, where the
“language” (both spoken and unspoken) across entities is well-understood, it is
considered to be communication-efficient – increasing the probability of transfer while
decreasing the associated costs (Grant, 1997).
Consequently, as the economies of scale
and scope of operations rise, the benefits associated with investments in commonality of
language also rise (Heiman and Nickerson, 2002). The example of the failed Mars
Climate Orbiter highlighted in Section 1.1 represents a particularly expensive
15
repercussion of NASA not having invested in creating a common language for data entry
among its engineers and subcontractors (Postrel, 2002).
The second contributor to absorptive capacity is common or base knowledge. Common
knowledge was described by Grant (1997) in the KBV as one of the mechanisms needed
to facilitate knowledge transfer. However, common knowledge translates to an
intersection, not an overlap of knowledge. A complete overlap of knowledge is
inefficient and represents limited opportunity for transfer. This inefficiency is
highlighted by the relationship that exists between a manufacturer and a supplier. For
example, Microsoft requires a base knowledge of the circuitry of the silicon chips made
by Intel for the production of its Windows operating system. However, Microsoft does
not need to have a complete knowledge of the development and production of the silicon
chip – there is an intersection of knowledge, but not an overlap. Based upon TCE, if
Microsoft had a need for a complete overlap of the knowledge required to develop and to
produce a silicon chip then they would bring that knowledge within the boundaries of
their firm. On the other hand, if Microsoft had no knowledge of how a silicon chip
works, they would, most likely, be unable to effectively interface with Intel.
The third contributing commonality for absorptive capacity is a common understanding
(or utilization) of processes. As highlighted in the discussion of the KBV in Section
2.1.2, a common process coordinated through a hierarchical structure improves the
efficiency of knowledge transfer while decreasing the associated costs (Grant, 1997).
16
The fourth contributing commonality to absorptive capacity is one of common problem
solving,
“The more [common] experience the…firms have in solving similar types of
problems, the easier it will be for the [recipient] firm to be able to
find…applications for the newly assimilated knowledge.” (Lane and Lubatkin,
1998:466)
A network of multiple organizations can provide a “platform” for these commonalities to
exist (Argote, 1999) and grow. For example, a network of organizations that engage in
standardized operations, imposed by some hierarchical authority, would be expected to
exhibit some commonalities of process and language, resulting from that standardization.
One example would be compliance with an operations manual provided to all
owners/managers of a fast food franchise to ensure a common customer experience.
Because all franchises are utilizing the processes outlined in a common manual, their
processes should be the same. These common processes enable a richer language to
describe operational details and to facilitate the transfer of knowledge than would be
possible if their processes were not similar.
In summary, absorptive capacity has been demonstrated to be an important factor of
knowledge transfer difficulty, and the four contributors of common language, common
process, common problem solving, common knowledge represent four sub-factors or subconstructs of the larger formative construct absorptive capacity.
Hypothesis 1 - Absorptive Capacity will have a negative relationship with interorganizational network knowledge transfer difficulty.
17
Causal Ambiguity An organization can use a number of inputs (X1 through Xn) in
combination with a number of causal factors (Factor1 through Factor n) to generate some
outcome(s) (O1 through On) (Figure 1). The concept of causal ambiguity centers around
“knowability” (the extent to which something can be known) and “knowness” (the extent
to which something is known) of two sets of elements – (i) the organizational inputs and
(ii) the causal factors that are used in combination to generate outcomes.
Causal ambiguity represents an interesting paradox for organizations. On the one hand,
causal ambiguity impedes a firm’s (or network’s) ability to imitate valuable resources
(knowledge) within its boundaries, limiting the ability to leverage resources to create a
competitive advantage (Reed and DeFillipi, 1990). For instance, organizational inputs
can be understood as the raw materials used to manufacture a product, and the causal
factors can be viewed as the processes used. When an organization does not know what
combination of inputs and process factors cause the final outcome, their knowledge is, at
best, causally ambiguous1. When knowledge is causally ambiguous, transfer is difficult
if not impossible. In his work exploring intra-organizational “stickiness”, Szulanski
(1996) found “fundamentally irreducible” causal ambiguity (which is a characteristic of
the knowledge recipient) to be a key contributing factor to knowledge stickiness.
Alternatively, Wilcox-King and Zeithaml (2001) examined, in part, the tacitness of the
1
For example, in the 1890s, Procter and Gamble had been manufacturing Ivory Soap (outcome) for several years
utilizing the same ingredients (inputs) and the same processes (causal factors). When an employee had inadvertently
left one of the soap making machines on during his lunch break, he returned to a frothy mixture unlike any soap
mixture ever seen at Procter and Gamble. Because none of the inputs had changed, Procter and Gamble elected to
package and distribute the soap as normal. Several months later, Procter and Gamble was inundated with orders for the
“floating soap”. At this point, Procter and Gamble was operating under causal ambiguity – having forgotten about the
frothy accident several months before, they were unclear as to what ingredient (input) or process (causal factor) could
have generated the outcome of floating soap. Eventually the connection to the extra air in the soap making process was
discovered and “It Floats” became an advertising slogan for Ivory Soap. (Ivory.com, 2003)
18
knowledge in question as an indicator of causal ambiguity – the more tacit the knowledge
in question, the greater the associated causal ambiguity. Mosakowski (1997) provides a
detailed framework to describe the concept in the context of organizational decision
making.
On the other hand, causal ambiguity within a firm or a network may represent a positive
characteristic of the organization, because it inhibits inimitability of competencies by
other firms and therefore protects competitive advantage (Wilcox-King and Zeithaml,
2001). Mosakowski (1997) echoes this observation and similarly determined that
although increased causal ambiguity has the impact of decreasing knowledge
transferability within the firm, and by association its application, it also has the potential
to increase competitive advantage by increasing the difficulties associated with imitation
by competitors.
This paradox of causal ambiguity, provides the researcher with two sub-constructs of the
larger formative construct of causal ambiguity for analysis – causal ambiguity related to
the internal inputs and processes that are attributable to specific outcomes and causal
ambiguity related to the internal inputs and processes which are known to create a
competitive advantage.
Hypothesis 2 – Causal Ambiguity will have a positive relationship with interorganizational network knowledge transfer difficulty.
Outcome Ambiguity The deliberations on causal ambiguity focus on uncertainty about
the causal factors or inputs that generate outcomes and assume that the outcome(s) is
19
(are) known. However, “outcome ambiguity”2 exists when the outcomes are unknown or
unknowable to the knowledge source. At the conclusion of her work on causal
ambiguity, Mosakowski (1997) poses the question – “Do they (managers) differentiate
outcome predictability from causal ambiguity?” She leaves the question unanswered.
Outcome ambiguity has not been addressed through the existing research on the topic of
ambiguity and uncertainty. In a development and exploration of environmental
uncertainty, Milliken (1987) describes a typology of uncertainty exogenous to the firm –
state, effect and response uncertainty. She defines the general construct of uncertainty as
“…the perceived inability to predict something accurately. An individual
experiences uncertainty because he/she perceives himself/herself to be lacking sufficient
information to predict accurately or because he/she feels unable to discriminate between
relevant data and irrelevant data” (Milliken, 1987: 136)
Although her work provides insight into different types of environmental uncertainty,
ultimately, Milliken’s work does not differentiate among the different sources of
environmental uncertainty that exist outside of the boundaries of the firm. The
descriptions of state, effect and response uncertainty characterize a generalized
environment that encompasses everything which is external to the firm, without
distinction among the different components of that environment. Because of this
‘general’ orientation of the conceptualization of all the uncertainty concepts, Milliken’s
development is not intended to differentiate among specific sources of uncertainties that
are implicitly included in the constituent members of the general concept of ‘the
2
Outcome ambiguity, although not technically a new term, is put forth as a new concept here within the
context of knowledge transfer. Ho et al. (2001) studied the impact of “outcome ambiguity” and its relation
to managerial decision making. The authors did not develop a definition of the concept, relying on the
reader’s understanding of the generalized concept of “ambiguity”.
20
environment’, which may include alliance partners, competitors, or entities which
encompass both roles simultaneously. Such a general treatment, assumes a completely,
and only, random behavior by the environment.
Although informed by Milliken’s work, understanding of the uncertainties regarding how
behavior of one organization will affect the perspectives of another organization, which
are both members of the same network, and specific constituents within an environment,
is not further developed.
Extending the work of Mosakowski and others in causal ambiguity and the work of
Milliken in perceived environmental uncertainty, the concept of “outcome ambiguity” is
developed as a factor that will influence the transfer of knowledge within an interorganizational network. I distinguish between two sources of outcome ambiguity, which
may exist separately or in combination.
The first source of outcome ambiguity is the “knownness” of the knowledge in question.
Szulanski (1996) develops the concept of “unproveness” or unknownness in his work
examining the factors related to intra-organizational knowledge transfer difficulty.
Unproveness is explained to be present when the knowledge in question has no previous
record of past usefulness – neither the source nor the recipient can know the outcomes
associated with its application (i.e., the outcomes are unknowable). For example,
knowledge of a well-established operational best practice, would be considered to be
proven, with a finite or bounded set of possible applications. I will refer to this set of
21
possible applications as the “Knowledge Usage Set”; [KU1, KU2, KU3]. Alternatively,
the discovery of a new chemical compound would be considered to be unproven
knowledge, with an infinite or unbounded Knowledge Usage Set; [KU1, KU2,
KU3…KU∞]. When knowledge is unproven and the Knowledge Usage Set is unbounded,
a higher degree of outcome ambiguity and knowledge transfer difficulty would be
expected.
The second source of outcome ambiguity is the uncertainty embedded within the
relationship between the knowledge source and the knowledge recipient(s). The basic
premise here is that the knowledge recipient can put the received knowledge to more than
one use. That is, it can choose from multiple possible actions to follow once the
knowledge has been received. I will refer to this set of actions as the “Recipient Action
Set”; [RA1, RA2, RA3]. There are two primary concepts that contribute to the
manifestation of uncertainty in this relationship – partner protectiveness, and trust.
Partner protectiveness, as described by Simonin (1999), is the degree of protectiveness a
knowledge source will assign to its knowledge base. Hamel (1991) explains that some
partners in alliances (and networks) make their knowledge less transparent than others,
creating situations dominated by asymmetry. Similarly, Szulanski (1996) found that lack
of motivation due to fear of losing ownership, inadequate incentives or a lack of
willingness to allocate appropriate resources, all contributed to knowledge transfer
difficulty. Where competition, or potential for competition may exist, a similar lack of
enthusiasm for knowledge transfer may exist for a fear of opportunistic behavior. Where
the possibility of opportunistic behavior exists, the number of elements in the Recipient
22
Action Set increases and the Set potentially becomes unbounded. This is true because,
unlike a situation defined by no competition or a limited probability of opportunism, if
competition is present and/or if opportunism is likely due to limited or weak authority,
the knowledge source cannot limit the possible outcomes associated with knowledge
transfer– contributing to increased outcome ambiguity.
Trust represents the second component of the relationship between the knowledge source
and the recipient. Across the many definitions of trust, the common themes of risk,
expectations and a concept of voluntary vulnerability are consistently present,
“…the willingness of a party to be vulnerable to the actions of another party
based upon the expectation that the other will perform a particular action
important to the trustor, irrespective of the ability to monitor or control that other
party.” (Mayer et al., 1995: 712)
Trust deals with the source’s present beliefs about the recipient(s) upon which it will then
base its future actions with the recipient – specifically in our context to share knowledge
or not (Hosmer, 1995; Zucker, 1986). Researchers have suggested that trust is a
functional prerequisite for knowledge exchange (Lewis and Weigert, 1985; Allee, 2002).
And trust, relative to price and authority, is the most effective mechanism to facilitate the
transfer of knowledge resources within and between organizations, in part because the
presence of trust decreases situational uncertainty (Adler, 2001). However, Ford (2002)
points out that cooperation can (and does) occur without trust – provided that the risk of
an undesirable outcome is low. Korczynki (1996) found in a study of the UK
construction industry, that “low trust network forms” enabled cost improvements, but not
knowledge transfer. Alternatively, “high trust network forms” have been found to excel
at transferring knowledge (Adler, 2001). Facing opportunistic threats, which contribute
23
to an unbounded Recipient Action Set, as might be expected in the former, firms will
prefer to retain their knowledge at the expense of the network, rather than risk
engagement in unknown scenarios where their shared knowledge could be used to their
detriment (Walker, 1995).
Figure 2 provides a typology of outcome ambiguity based on the two sources discussed
above – the proveness of the knowledge in question and the certainty with which the
knowledge source understands the actions of the knowledge recipient.
In a scenario where knowledge is proven (i.e., the Knowledge Usage Set is bounded) and
the actions of the knowledge recipient can be understood with some degree of certainty
(i.e., the Recipient Action Set is bounded), the scenario is considered to be one of Type 1,
or low, outcome ambiguity. Where the two element sets are bounded, the eventual
Outcome Set (designated as [O1, O2, O3]), is also considered to be bounded. And, where
the possible outcomes are finite and known with some degree of certainty, outcome
ambiguity can be considered to be low.
An example of Type 1 outcome ambiguity
could be represented by a network of hotel franchises who share well-documented checkin procedures. In this scenario, the knowledge in question is proven, creating a bounded
Knowledge Usage Set. In addition, through limited partner protectiveness practices, the
duration of the relationship(s) and/or through trust (or trust-like behaviors facilitated by a
strong central office), the Recipient Action Set(s) would be relatively well understood
and considered to be free of opportunistic behaviors. Consequently, the Outcome Set is
considered to be bounded – creating the least amount of outcome ambiguity (Type 1).
24
On the other extreme, if the knowledge in question is unproven (i.e., the Knowledge
Usage Set is unbounded) and the actions of the knowledge recipient are unknown (i.e.,
the Recipient Action Set is unbounded), the scenario is considered to be one of Type 4, or
high, outcome ambiguity. Where the two element sets are both unbounded, the eventual
Outcome Set is also considered to be unbounded. And, where the possible outcomes are
infinite and unknown, outcome ambiguity is considered to be high. An example of Type
4 outcome ambiguity could be represented by a newly organized network of
pharmaceutical firms who co-discover a compound that inhibits the growth of certain
types of cancer cells. In this scenario, the Knowledge Usage Set is unbounded because
the finding is new and unproven. In addition, the Recipient Action Set(s) could be
unbounded, in part because the network is new (duration of relationships is limited),
partner protectiveness may not be fully understood (especially if the firms are in
competition) and trust may exist only as far as contracts or a network governing authority
will punish for opportunistic actions. These unbounded element sets contribute to an
unbounded Outcome Set – and Type 4 outcome ambiguity.
In scenarios where one element set is bounded and one is unbounded, outcome ambiguity
will fall somewhere between low (Type 1) and high (Type 4). However, the scenario
where the Knowledge Usage Set is unbounded and the Recipient Action Set is bounded
(Type 2) is not the same as the scenario where the Knowledge Usage Set is bounded and
the Recipient Action Set is unbounded (Type 3)– the contributors of their uncertainties
are quite different.
25
Type 2 outcome ambiguity is characterized by an unbounded Knowledge Usage Set and a
bounded Recipient Action Set, and the eventual Outcome Set is unbounded. However,
because the actions of the knowledge recipient are considered to be understood, outcomes
associated with negative serendipity due to opportunistic action on the part of the
recipient, can be eliminated.
Alternatively, Type 3 outcome ambiguity is characterized by the opposite scenario. In a
Type 3 scenario, the knowledge in question is proven (i.e., the Knowledge Usage Set is
bounded), but the actions of the recipient are unknown (i.e., the Recipient Action Set is
unbounded) and the eventual Outcome Set is unbounded because outcomes associated
with opportunistic behavior cannot be eliminated.
The concept of outcome ambiguity is particularly relevant when more than one firm is
involved in a relationship such as is true in a multi-firm network environment. For
example, in a multi-organizational network, the knowledge source would be required to
consider the Recipient Action Set for multiple knowledge recipients, increasing the
complications related to the decision to share. This is the rationale for why Type 3
outcome ambiguity contributes to a higher level of knowledge transfer difficulty than
Type 2 outcome ambiguity – the uncertainties related to the actions of the knowledge
recipient(s) make the eventual Outcome Set less stable than uncertainties related to the
Knowledge Usage Set. The issue of network size is paradoxical – as the size of the
network increases, the potential base of accessible knowledge increases. However, the
decision to share knowledge becomes more complex, because the knowledge source must
consider more Recipient Action Sets, translating into greater outcome ambiguity and
26
greater knowledge transfer difficulty. However, inter-organizational networks can
mitigate the uncertainties related to initially unbounded Recipient Action Sets through
governance policies and controls and potentially contributing positively to the issues of
partner protectiveness, duration and trust. As will be seen below, some network forms
can effectuate trust or eliminate the need of trust and, thus can help reduce such
complexity, reduce outcome ambiguity and reduce knowledge transfer difficulty.
The outcome ambiguity matrix in Figure 2, can be further understood through a brief
examination of game theoretic concepts3. It is important to note that in Type 1 and in
Type 2 outcome ambiguity, the recipient action sets are considered to be bounded
because a formal (e.g., a strong centralized governance structure) or an informal (e.g.,
limited competition among members, objective interdependence, an economically
dominant entity acting as a strong governance authority) governing mechanism exists.
Where this is the case, the entities can be considered to be operating within the context of
co-operative or non-zero sum “games” where accurate communication among “players”
is critical (Davis, 1997). Under these circumstances, in theory, entities are looking for
the largest possible payoff (value in the outcome set) for both players. However, in Type
3 and Type 4 outcome ambiguity (higher states of outcome ambiguity), the recipient
action sets are not bounded, in part because of the lack of a centralized governing
structure to punish for non-network oriented or opportunistic behavior. Where this is the
case, the entities can be considered to be operating within the context of a non-cooperative or zero-sum game, where communication is less relevant (Davis, 1997). Under
these circumstances, a “win” for one entity is perceived as a “loss” for the other entity.
3
For a more detailed treatment of game theoretic concepts, see Davis, 1997 or Kuhn, 1997.
27
As a result, the outcome set is analyzed differently by the decision makers operating
under Types 3 and 4 versus under Types 1 and 2 outcome ambiguity.
In a network environment, the game theoretic concepts move from 2-person “games” to
n-person “games”, where potential complications such as collusion, auctions, power,
symmetry of information all need to be considered – making the issues related to
outcome ambiguity particularly complex for the networked knowledge source.
Hypothesis 3 – Outcome Ambiguity will have a positive relationship with interorganizational network knowledge transfer difficulty.
In this section, I have provided a description of two well-established factors of
knowledge transfer difficulty – absorptive capacity and causal ambiguity. Within the
discussion of each of these factors, I provided additional exploration of the sub-factors of
common language, common process, common problem solving and common knowledge
as contributors to absorptive capacity and exploration of “internal” causal ambiguity of
inputs and factors and “external” causal ambiguity of the drivers of competitive
advantage as sub-factors of causal ambiguity. I then introduced and developed the factor
of outcome ambiguity and how its presence would be expected to affect knowledge
transfer difficulty. Although this factor has not been explicitly researched in previous
studies examining knowledge transfer difficulty, the previous work of Szulanski (1996)
and work by researchers studying the effects of trust (e.g., Gefen, 2002; Ford, 2002;
Lewis and Weigert, 1985), and perceived environmental uncertainty (Milliken, 1987)
have provided a logical basis for the development of the factor and its expected
28
relationship with knowledge transfer difficulty. Outcome ambiguity was also broken into
two sub-factors – the proveness of the knowledge in question and the expectations of the
knowledge source of the actions of the knowledge recipient. For each of these factors, a
relationship with knowledge transfer difficulty was hypothesized. These hypotheses are
represented as Part 1 of the specified model of this dissertation in Figure 3.
I acknowledge that direct control variables do exist that would be expected to have a
direct and invariant impact on the efficacy of knowledge transfer. Two of these factors,
knowledge type and technology, are briefly discussed below, but will not be included in
the eventual empirical testing.
Knowledge Type Winter (1987) identified four dimensions that describe different types
of knowledge. These dimensions include codification (explicit vs. tacit), complexity
(simple vs. complex), observability (observable vs. unobservable) and degree to which
the knowledge is embedded in a process (process independent vs. process dependent).
Knowledge that is “easy to transfer” can be characterized across these four dimensions,
as explicit, simple, observable and process independent. Alternatively, knowledge which
is considered “difficult to transfer” can be characterized across the same dimensions as
tacit, complex, not observable and process dependent. This general taxonomy is
consistent with that described by von Hippel (1994), Nonaka (1991) and Polanyi (1962).
29
Technology Although a highly specific discussion of technology is outside of the scope
of this dissertation, an exploration of knowledge transfer and its determinants would be
incomplete without some organized attention to technology.
The factors of knowledge transfer – absorptive capacity, causal ambiguity and outcome
ambiguity – characterize the “environment” to facilitate the transfer of knowledge.
Technology represents an enabler of knowledge transfer. Specifically, Information and
Communication Technologies (ICTs) enable the collection, collation, storage and
dissemination of data, and thereby assisting the creation and transfer of knowledge in a
way that could never have existed prior to the advent of desktop computing (Roberts,
2000). Some common examples of ICTs include e-mail, teleconferencing, CAD/CAM,
and Groupware. Carlsson (2002) highlights that although knowledge transfer is not
contingent upon the use of ICTs, their adoption and use is recognized as an important
enabler to the knowledge transfer process. The extent to which technology is adopted
within an organization, and by extension within a network of organizations, is contingent
on many factors. Eder and Igbaria (2001) identified earliness of adoption, and organized
top management support as the key factors for diffusion and infusion of intranet
technology adoption. Roberts (2000) found that adoption in ICTs can contribute to an
increased level of trust – an important component of outcome ambiguity –
“The development of trust between agents from different cultural bases will
necessitate a higher investment to build up shared experience than that which
would be required between agents that share a common…base”. (2000:434)
2.3 Inter-organizational Network Types
30
In this section, I review the literature examining inter-organizational network types, and
describe the four network types that provide the foundation for Part II of the specified
model for this dissertation.
A basic “network” can be defined as “nothing more (or less) than a system…consisting of
objects and connections” (Casti, 1995) – generally referred to as ‘nodes’ and ‘linkages’
in social network theory. When addressing the concepts of networks in general, a
fundamental distinction should be made between durable, permanent networks and
temporary networks (Westlund, 1999). My concern is with the former case and
conceptually I am more aligned with the description put forth by Johnson (1995) – that
networks represent a form of infrastructure and can thus,
“…be considered to be a relatively static backcloth which supports and
constrains a relatively dynamic traffic of system activity”. (emphasis in the
original)
Researchers have studied inter-organizational networks from different vantage points.
Thorelli (1986) and Almeida et al. (2002) studied membership in a network as a strategy
between complete organizational self-sufficiency with no inter-organizational
transactions on one end and a complete outsourcing strategy with exclusively marketbased transactions on the other end. Allee (2002) proposed the concept of a “value
network” as the basis for understanding the activities related to the creation of intangibles
such as knowledge. Carlsson developed a generalized framework of networks organized
for the purposes of strategic knowledge management (2002).
31
My work has been much enriched by these studies. However, although these studies
developed and defined different types of networks, they did not examine the concept of
knowledge transfers within these network types. Therefore, the question regarding how
knowledge transfer difficulty, and its associated factors, varies with network type remains
unanswered.
I approached network study from a slightly different perspective – using the established
foundational theories of Transaction Cost Economics, the Knowledge Based View of the
Firm and Social Network Theory, I differentiate network types using three established
network characteristics; Governance Structure, Intensity of Competition and Scope of
Operations. I then use these characteristics to create an abstract representation of interorganizational network types and then explain the different networks’ existence in
practice.
Governance Structure
From his seminal work in Transaction Cost Economics, Williamson (1973, 1975)
explained that the choice of governance structure for organizations was guided by the
need to minimize transaction cost through consideration of issues related to bounded
rationality, the risks associated with opportunistic behavior, information impactedness
and general uncertainty. As highlighted above, researchers have looked to networks as
an organizing principle residing between pure market transactions and complete
organizational self-sufficiency. However, once within the network, the question of
governance structure remains…specifically to what extent network governance is
32
centralized or decentralized. The two basic organizational structures of centralization and
decentralization have been studied extensively (Galbraith and Merrill, 1991; Adler, 2001;
Van den Bosch, 1991; Volberda et al., 1998). Williamson provides insight into these two
structural forms using the lens of Transaction Cost Economics. I extend the principles of
his work into the inter-organizational network domain.
Williamson describes a hierarchical governance structure as providing the authority to
address issues related to opportunistic behavior, information impactedness and bounded
rationality. A (formal or informal) hierarchical authority would also have the ability to
mandate standardization of operations, language, policies, etc. Alternatively, a
decentralized governance structure is described as one of peer group associations, without
subordination, involving collective and usually co-operative activities. Williamson
highlights that this governance structure is deficient in its ability to address opportunism,
and free rider abuses. However, recent researchers have found a decentralized structure
to be particularly well-adapted to facilitate innovation and new knowledge creation
(where the former structure has been found to better facilitate the diffusion and
implementation of existing knowledge) (Galbraith and Merrill, 1991; Adler, 2001; Van
den Bosch, 1991; Volberda et al., 1998).
Intensity of Competition
Within the context of social network theory, an important component of network
structure that has been found to have significant impact on the transfer of knowledge is
the ties or linkages among network entities (Uzzi and Lancaster, 2003; Dacin et al., 1999;
33
Granovetter, 1985). The linkages that exist among network entities have been described
as being ‘embedded’ (integrated) or at ‘arm’s length’ (Dacin et al. 1999). Integrated ties
“are considered to create behavioral expectations that…shift the logic of opportunism to
a logic of trustful co-operative behavior in a way that creates a…basis for knowledge
transfer…” (Uzzi and Lancaster, 2003:384).
By contrast, linkages at ‘arm’s length’ are
“…cool, impersonal, atomistic…motivated by instrumental profit seeking” (Uzzi
and Lancaster, 2003:384).
Although it may initially appear counterintuitive that organizations voluntarily join
networks while maintaining ‘arm’s length” ties, consider the VISA network. Individual
banks are fierce competitors, yet collectively benefit from the functionality of global
credit card acceptance afforded by the VISA network – their relationships are “cool and
impersonal”, with linkages created for the purposes of decreased transaction costs. In
addition, Powell et al. (1996) found that as the technological sophistication of an industry
increases, the intensity and number of competitive alliances also increases – although
relationships are again, “cool and impersonal”, they come together to reduce the costs
associated with R&D –
“When there is a regime of rapid technological development, research
breakthroughs are so broadly distributed that no single firm has all the internal
capabilities necessary for success…Firms thus turn to collaboration to acquire
resources and skills they cannot produce internally, when the hazards of
cooperation can be held to a tolerable level.” (1996:117)
Using a more commonly accepted description of these integrated and arm’s length
linkages, I will refer to this network characteristic as “intensity of competition” among
34
the network members, with low intensity of competition equating to integrated linkages
and high intensity of competition equating to arm’s length linkages4.
Scope
Another concept from social network theory involves the similarity of network nodes,
where “nodes” are defined as the end points or members of a network. Westlund (1999)
refers to the concept of similarity as the,
“…homogeneity and heterogeneity in the nodes of the respective networks – which
constitutes one component of the totality of …underlying transaction costs.” (1999:93)
This concept of similarity among network members is also addressed by Simonin (1999)
in his examination of knowledge transfer between strategic alliance partners. Simonin
examined the concept of operational similarity as a factor to explain the extent to which
knowledge transfers between alliance partners. For the purposes of this dissertation, I
will refer to this concept as “scope”, where broad scope networks will have operationally
dissimilar (heterogeneous) nodes (members) while narrow scope networks will have
operationally similar (homogenous) nodes (members).
Clearly, there are other network characteristics that I could have selected to differentiate
among network types. I have specifically chosen these three, in large part because they
have been shown in previous research to have an impact on inter-organizational
knowledge transfer (Simonin, 1999; Westlund, 1999).
4
I am aware that the linkages within a network may exhibit variance and all linkages may not necessarily
be similar. However, for this paper, the unit of analysis is the network, and for the purposes of theory
development, I will use the description to characterize the majority of the linkages among members.
35
Using a positivist theory approach, I will now consider four types of inter-organizational
network that are the most prevalent in practice, as evidenced through the discussion and
examples provided below.
Three of the four network types examined in this dissertation have already received some
organized research attention. These types include the franchise network (Argote, 1999;
Darr et al., 1995; Thorelli, 1986), the value chain network (Thorelli, 1986; Dyer, 1997;
Li, 2002), and the innovation network (Harris, 2000; Harrisson, 2002; Powell et al.,
1996). A fourth network type, the co-opetive network form, has had the least amount of
formal treatment in the literature. The term “co-opetive” has been used to describe a
situation where traditional competitors have agreed to cooperate to achieve a common
objective (Brandenburger and Nalebuff, 1996; Loebecke et al., 1999; Shapiro and Varian,
1999). Using this accepted notion of “co-opetive”, I extend this concept to define a coopetive network as some formalized arrangement of N competitors, collaborating to
achieve some common objective.
I will provide a brief description of each network type and explain the unique space each
type occupies within the abstract model (Figure 4).
Franchise Network
The general concept of a “franchise” network is well-accepted and understood in research
and in practice, with common examples in the restaurant and hotel industries (e.g.,
McDonalds and Holiday Inns). A franchise network is generally considered to have a
36
strong centralized governance structure. This is attributable to the presence of a “head
office” that would have some degree of authority, including the ability to punish for noncompliance, establish branding usage policies, enforce image and quality controls,
standardize the customer experience, etc. By definition, this network form would be
characterized as having narrow scope – the network members would be expected to
engage in similar, often standardized operations, using similar language, similar methods
of problem solving, etc. Competition is generally low among members of franchise
networks, because of collective brand identity and economic interdependence (Volberda,
1998).
Innovation Network
Innovation networks can generally be described as alliances that have been formed to
mitigate research costs, the risks associated with innovative research, or to ensure that an
organization is exposed to the most recent discoveries in a rapidly changing environment.
Therefore, the primary motivation for participation in an innovation network is access to,
or the creation of, new knowledge. Firms within industries that are characterized by
competence-destroying technical change, where innovations can and do significantly
diminish the effectiveness of existing competencies, are particularly dependent upon
injections of the type of external knowledge obtained through networks (Powell et al.,
1996). Because of the emphasis on knowledge creation, this network form is likely to
have a decentralized /non-hierarchical governance structure (Galbraith and Merrill, 1991;
Adler, 2001; Van den Bosch, 1991; Volberda et al., 1998). For example, the MAOSCO
(Multiple Application Operating System Company) consortium includes banks, plastic
card manufacturers, and chip manufacturers, supporting the emergence of smart cards in
37
the U.S. and abroad – a high scope network. Although MAOSCO operates as a forum for
discussion, exchange of documentation and development of new technology, it has
almost no authority over its members regarding how they operate, how they interact, or
how they eventually implement products or services using the agreed upon technology
standards. Competition within an innovation network may change based upon scientific
discovery, product introduction, patent status, regulation, etc. However, as the
technological sophistication of the industry increases, the intensity and number of interfirm alliances would be expected to increase (Powell et al., 1996).
Value Chain Network
The value chain network form has been studied in at least three configurations, including
the dyadic (i.e., the manufacturer/supplier relationship), one-to-N relationships (i.e., the
manufacturer with multiple suppliers) and N-to-N relationships (i.e., multiple
manufacturers and multiple suppliers). Firms engage in the value chain network form in
large part to achieve both lower transaction costs and knowledge symmetry. For
example, Covisint is an N-to-N network of automotive manufacturers (e.g., Toyota,
Nissan, Daimler Chrysler, Ford), parts manufacturers (e.g., Delphi, Johnson Controls,
BASF), and industry technology suppliers (e.g., CommerceOne, Oracle, Sun,) who
aggregate orders and leverage common business and manufacturing processes to lower
their overall costs of production. When relevant, firms also participate in value chain
networks to move from a position of asymmetric knowledge to symmetric knowledge.
Using the example highlighted above, for the development and production of its
strategically critical Windows operating system, Microsoft requires some (but not a
complete) knowledge of the circuitry of the processing chip from Intel. That is, the
38
knowledge is asymmetric between the two organizations. To alleviate this asymmetry
and facilitate production of its operating system, Microsoft would form a value chain
network with Intel and other strategically important suppliers. However, Microsoft
would not participate in a network if one of the two following scenarios were true. First,
if Microsoft needed Intel’s knowledge of the microchip development process, the
technology, and specifications to develop their operating system (i.e., if a significant or
even complete overlap of knowledge was required), then Microsoft would likely acquire
Intel or replicate the capabilities within its own boundaries. Second, if Microsoft was
able to produce Windows with only a cursory understanding of the processing chip, then
they would trade with Intel and other suppliers through standard market interfaces. As
described by Thorelli (1986) and others, joining a network provides for an option
between the two extremes of “buy” versus “trade”.
Although the value chain network form is characterized by multiple dyadic contracts, the
network generally has a weak, or non-existent official central authority. The unofficial,
albeit influential, authority in this network form will often reside with the most
economically dominant member of the chain. For example, Wal-Mart is well known to
have multiple manufacturer and supplier relationships. Although the value chain
networks that Wal-Mart is embedded within generally have no official central authority,
Wal-Mart will often dictate standards and policies as the economically dominant member
of the chain. As with the innovation network, the participating organizations within this
network form exhibit broad operational scope and members generally engage in very
different types of business processes and often have different knowledge bases, use very
39
different descriptive languages, and experience different types of problem solving
environments. Participants within a value chain network do not generally compete - a
manufacturer would not be considered to be in competition with its suppliers.5
Co-opetive Network Form
Although the co-opetive network as a concept is generally lacking in the literature, many
well-known examples can be cited in practice. One example of a well-known co-opetive
network is VISA International. Highly competitive banks, engaged in very similar
operations (indicating a low scope network), join the VISA payment network, in part,
because it is economically infeasible for any single bank to develop a global transaction
processing system that would facilitate credit card transactions at any of 22 million
merchant locations around the world. And although VISA provides forums for
discussion and provides discounts for members through the aggregation of orders to
suppliers, ultimately VISA is a weakly centralized organization with limited authority to
punish members for opportunistic behavior. Organizations are motivated to participate in
networks of competitors, like VISA, to access existing and/or newly created knowledge,
which may vary based upon industry, pace of technological change, regulatory
environment, etc. Carlsson (2002) found that when risk of opportunism is high, the
governance structure of the network needs to be strong. However, often co-opetive
networks are networks of peers with limited central governance outside of very limited
parameters, as shown in the VISA example.
Li (2002) and Almeida et al. (2002) described a concept of “horizontal” value chain competition (i.e., when multiple
suppliers are engaged in a common network with a single manufacturer and are required to share their knowledge
amongst their competitors to support the development of a new product or process at the manufacturer). I consider this
concept to be a special type of co-opetive network.
5
40
2.4 Inter-Organizational Network Types and Factors of Knowledge Transfer
Difficulty – Specified Model Part II
Van den Bosch et al. (1999) studied the effects of intra-organizational forms on only one
factor of knowledge transfer – absorptive capacity. Using a similar framework, I have
developed each relationship between each network type and each factor of knowledge
transfer difficulty. The summary of these relationships comprises Part II of the specified
model and can be seen in Figure 5.
The influence of the franchise network form
Absorptive Capacity
Four commonalities, individually as well as in combination, have been found to
contribute to absorptive capacity – commonalities of knowledge, process, problem
solving and language (Lane and Lubatkin, 1998; Cohen and Levinthal, 1990). When
organizations are engaged in similar processes, it is logical to conclude that they have
developed a similar base of knowledge related to these processes and that the
organizations have developed a similar language to describe their tasks. Shared language
is important to absorptive capacity, in part because it facilitates deeper and more
meaningful communication than would otherwise be possible (Cohen and Levinthal,
1990).
Recall that two of the characteristics of the franchise network are narrow operational
scope and a hierarchical central governance structure. I argue that this narrow scope
41
provides a fertile environment for the four commonalities above, while a strong
hierarchical governance structure has the ability to mandate standards governing, for
example, service and quality. These standards would logically lead to the commonalities
of process, knowledge and language identified above – and a high state of absorptive
capacity.
Hypothesis 4a - The franchise network form will be associated with a high state of
inter-organizational absorptive capacity.
Causal Ambiguity
Where organizations are engaged in similar processes, they would be expected to have a
common understanding of the inputs and causal factors contributing to particular outcomes.
The common processes which exist in a franchise network would be expected to support a
common knowledge of inputs and causal factors, both before and after outcomes associated
with their use are known – thereby creating a low state of causal ambiguity. A related
characteristic of causal ambiguity identified by Mosakowski (1997) is task complexity - the
more complex tasks become, the more difficult it becomes to identify the specific cause
and effect that each input or factor has on related outcomes. Where this complexity can be
mitigated, causal ambiguity is reduced. Simon (1962) determined that a strong,
centralized/hierarchical governance structure can mitigate task complexity through
specialization of labor and standardization. Given the expected hierarchical central
governance structure of the franchise network, complexity of task is expected to be low.
Hypothesis 4b - The franchise network form will be associated with a low state of interorganizational causal ambiguity.
42
Outcome Ambiguity
Recall that one of the characteristics of a franchise network is a motivation to participate
based upon access to pre-existing knowledge, such as operational guidelines (e.g.,
handbooks, manuals) or employee training programs – proven knowledge and a bounded
knowledge set. As a result, only outcome ambiguity Types 1 and 3 would be considered
to be relevant for this network form (Figure 2). Another characteristic of a franchise
network is low intensity of competition amongst the network members. Franchisees are
generally stakeholders within a larger entity – they are economically interdependent.
Adler (2001) and Kogut and Zander (1996) refer to this interdependence as “shared
destiny”. Shared destiny would help to mitigate actions related to opportunistic behavior,
and contribute to a bounded Recipient Action Set. Another characteristic of a franchise
network is limited organizational scope, evidenced in part by a commonality of
operational processes, again, decreasing the uncertainty related to the knowledge in
question, and contributing to a bounded Knowledge Usage Set. The fourth characteristic
of a franchise network is strong central governance. A hierarchical governance structure
would include an authority for punishment associated with opportunistic behavior
amongst the franchises. Assuming this threat of punishment is severe enough to prevent
defection, trust (or at least trust-like behaviors) could be mandated. As a result, the
Recipient Action Set would again, be considered to be bounded.
Where the knowledge in question is proven, and the actions of the recipients can be
anticipated, the Knowledge Usage Set and the Recipient Action Sets, respectively, can be
43
bounded. And, where these element sets are bounded, the Outcome Set is bounded,
leading to the least amount of outcome ambiguity.
Hypothesis 4c - The franchise network form will be associated with a low state of interorganizational outcome ambiguity.
To summarize briefly, the franchise network is hypothesized to be associated with: (i) a
high state of absorptive capacity, (ii) a low state of causal ambiguity, and a (iii) low state
of outcome ambiguity. Therefore,
Hypothesis 4d – The franchise network type will be associated with a low state of
knowledge transfer difficulty.
The Influence of the Innovation Network Type
Absorptive Capacity
The innovation network is characterized by a broad scope, and therefore the network
would not be expected to exhibit commonalities of process or problem solving, two
contributors to absorptive capacity. However, given that organizations generally join
innovation networks for the purposes of gaining access to new knowledge, they would be
expected to come to the network with some common or “base” or knowledge. This is
true because if organizations did not have a base knowledge of the topic in question,
transfer would be almost impossible. For example, if a network was developed to
research a particular form of cancer, and a strong base knowledge of biochemistry was
required, an organization with no previous experience in the area of biochemistry would
have a limited ability to absorb or contribute to the knowledge exchange. In addition,
within the context of this base knowledge, it would be logical to conclude that a common
44
language is used. For example, the Human Genome Project, incorporates government
(e.g., U.S Department of Energy) quasi-government (e.g., National Institutes of Health)
private (e.g., Wellcome, IBM) and public (e.g., MIT, Baylor College of Medicine,
Washington University) institutions. Each organizational entity approaches the project
with some common working knowledge of, for example, genetics, which is then used to
develop new knowledge related to Gene Sequencing, Bioinformatics, and other topics.
Although these respective organizations may use completely different vernaculars within
their respective operating environments, within the context of the Human Genome
Project, it would be expected that in the pursuit of new knowledge, the organizations
would engage in a common language. Consequently, this network type exhibits only two
of the four commonalities needed for absorptive capacity. Therefore,
Hypothesis 5a - The innovation network type will be associated with a low state of interorganizational absorptive capacity.
Causal Ambiguity
This network type is most common in environments characterized by rapid and turbulent
change, where organizations are working to mitigate risk or diffuse the cost of research.
In her causal ambiguity framework, Mosakowski (1997) indicated that when an operating
environment is unstable and unpredictable, past experiences provide less advantage in
understanding the ex-ante causal factors in a new scenario. This is true, in part, because
rapidly changing environments do not enable knowability of causal factors or inputs
before an outcome. As a result, causal ambiguity amongst members in this network
would be expected to be high.
45
Scenarios characterized by complexity and ill-structured problems are also considered to
have high causal ambiguity (Mosakowski, 1997). This is logical – if a particular process
or product has many interdependent components, identifying or isolating the impact of
each one on the eventual outcome would be difficult, if not impossible. Simon (1962)
suggests that more hierarchical structure helps to mitigate this complexity, consistent
with TCE. However, this structure is generally not present in an innovative network, in
part because it has been shown to negatively impact new knowledge creation and
innovation (Galbraith and Merrill, 1991; Adler, 2001; Van den Bosch et al, 1991;
Volberda, 1998; Teece, 1998).
Hypothesis 5b - The innovation network type will be associated with a high state of
inter-organizational causal ambiguity.
Outcome Ambiguity
The innovation network is characterized by a motivation to access newly created
knowledge, significant operational scope and a decentralized governance structure. The
intensity of competition for this network type is difficult to discern and would be
expected to vary based upon scientific discovery, product introduction, patent status or
regulatory environment.
Newly created knowledge is, by definition, unproven. And, the range of applications of
unproven knowledge is less certain and less bounded than the range of proven
knowledge. As a result, the innovation network type would be expected to contribute to
an unbounded Knowledge Usage Set. Consequently, only outcome ambiguity Types 2
and 4 would be considered to be relevant for this network type (Figure 2). The second
46
characteristic of an innovation network, significant organizational distance, is evidenced
by the fact that members of this network type engage in very different types of operations
and different types of industries, contributing to an increase in the uncertainty related to
how other organizations use knowledge, thereby increasing the elements in the
Knowledge Usage Set. The third characteristic of this network type is a decentralized
governance structure. Recall from Section 2.3, that a decentralized governance structure
is more conducive to innovation and knowledge creation than to knowledge transfer, and
the lack of an official authority to issue punishment for opportunistic behavior, can lead
to a lack of trust among members and an unbounded Recipient Action Set. Finally, the
intensity of competition among members is difficult to discern in this network type – it
would be expected to vary based upon context, represented industries and domains of
operation.
The types of outcome ambiguity that would be relevant for this network type are Types 2
and 4 – both characterized by unbounded Knowledge Usage Sets. However, the extent to
which the Recipient Action Set can be considered to be bounded is contingent upon the
components of the relationships among the network members, such as partner
protectiveness, duration of relationships and trust, which are determined by the policies
and directives of the network.
Hypothesis 5c(1) – When the Recipient Action Set is bounded, the innovation network
type will be associated with a low-medium state of outcome ambiguity (Type 2).
Hypothesis 5c(2) – When the Recipient Action Set is unbounded, the innovation network
type will be associated with a high state of outcome ambiguity (Type 4).
47
Because knowledge transfer difficulty is positively correlated with causal ambiguity and
outcome ambiguity, I would expect some degree of knowledge transfer difficulty to be
associated with this type. Initially, this expectation may violate conventional wisdom – I
stated earlier that the primary motivation to participate in this network type was access to
new knowledge. However, a further consideration of knowledge creation versus
knowledge transfer may explain the (possibly) perceived violation. This network type is
not based on the need to transfer an existing knowledge asset within the network, but
rather on the need to develop new, unproven knowledge. Therefore, it is not surprising
that the characteristics of an innovation network – some competition, broad scope, low
centralization of authority and decision making – contribute to a high degree of
knowledge transfer difficulty. However, these same characteristics may be expected to
facilitate new knowledge creation.
In summary, this network form is hypothesized to be associated with: (i) a low state of
absorptive capacity, (ii) a high state of causal ambiguity, and (iii) a medium or high state
of outcome ambiguity. Therefore,
Hypothesis 5d – The innovative network is expected to be associated with a high state of
knowledge transfer difficulty.
The Influence of the Value Chain Network Type
Absorptive Capacity
The value chain network is characterized by a broad scope (e.g., manufacturers,
suppliers, distributors, etc), which leads to different knowledge bases, different languages
48
and different methods of problem solving. Although some commonalities would be
necessary at the point of interaction, they would be minimized in the interest of costeffective transactions – recall the Microsoft/Intel discussion in Section 2.3.1. These
commonalities have been identified above as being necessary for absorptive capacity.
And, where they do not exist, as in a value chain network, absorptive capacity would be
expected to be low.
Hypothesis 6a - The value chain network type will be associated with a low state of
inter-organizational absorptive capacity.
Causal Ambiguity
The differing processes present in a value chain network utilize different inputs and
factors to create different outcomes – if the outcomes were the same, there would be no
need for participation. And, value chain networks generally exist in relatively stable
manufacturing or retail environments, characterized by incremental rather than
fundamental change – generally associated with low causal ambiguity (Mosakowski,
1997).
Value chain networks do not generally have a single formal centralized authority to
govern the interactions among the network members. Again, a strong hierarchical
governance structure would mitigate task complexity and non-standardized transactions,
and where this task complexity is addressed, causal ambiguity is expected to be low. A
unique characteristic of this network type is that the structure, by definition, mitigates
task complexity through specialization of labor – each stage in the vertical chain is
49
accomplished in a different (but contributing) process. As a result, the causal ambiguity
related to task complexity would be low.
Hypothesis 6b - The value chain network type will be associated with a low state of
inter-organizational causal ambiguity.
Outcome Ambiguity
As stated above, when the primary motivation for network participation is related to
access to existing knowledge, the expected Knowledge Usage Set is bounded. Therefore,
only outcome ambiguity Types 1 and 3 are relevant for this network type (Figure 2). In
the value chain network, members are typically accessing demand forecasts, inventory
levels, product development specifications, etc – proven knowledge. In addition,
accessing existing knowledge through constant relationships represents a method to
control costs. New relationships have associated costs – search costs, contract costs, set
up costs, and costs associated with initial production inefficiencies. Relationships of long
durations, as is expected in a network environment versus a dyadic environment, drive
down these costs, or avoid them altogether, thereby contributing to a decreased risk of
opportunistic behavior and a decreased Recipient Action Set.
As with the innovation network type, the operational scope of the network members is
significant; if members were all engaged in similar operations, there would be no need to
engage in a value chain network. However, the existing knowledge shared among
suppliers, manufacturers, distributors, and retailers is usually highly specific to the
sequential process in question. As a result, the Knowledge Usage Set would again, be
expected to be bounded. The third characteristic of this network type, low intensity of
competition, is logical – manufacturers are generally not considered to be in competition
50
with their suppliers. Consequently, the knowledge source in a value chain should
perceive a limited Recipient Act Set, with little or no expectation of opportunistic
behavior. Finally, although value chain networks are generally characterized by little
formal governance structure, an unofficial, albeit influential, central “governance” may
exist through the most economically dominant member of the network. Another
substantiation of this unofficial governance is demonstrated through an attribute of this
network type not found in other types – the process stream. A manufacturer is
“downstream” from the supplier. The manufacturer requires knowledge regarding the
supplier’s product that is not reciprocated. As a result, the manufacturer would be
expected to have more knowledge about the process than the supplier. This asymmetry
contributes to the dominance of the manufacturer and the manufacturer’s power to
penalize for opportunism. In addition, in a typical “1-to-N” value chain network type (1
manufacturer and multiple suppliers), the manufacturer would have economic dominance
and could effectively penalize inappropriate supplier behavior through shifting of
business from one supplier to another.
As with the franchise type, where the knowledge in question is proven, and the actions of
the recipients can be anticipated, the Knowledge Usage Set and the Recipient Action
Sets, respectively, can be bounded. And, where these element sets are bounded, the
Outcome Set is bounded, leading to the lowest type of outcome ambiguity.
Hypothesis 6c - The value chain network type will be associated with a low state of interorganizational outcome ambiguity.
51
The hypothesized factor influences for this network type are logical. Within a value
chain, the effective utilization of the product by the recipient is not dependent upon the
recipient possessing the knowledge embedded within the actual product (Grant, 1996),
indicative of Grant’s (1997) sequencing concept described in Section 2.1. Reiterating the
example used earlier, Microsoft does not need to posses the knowledge related to the
actual development and production of Intel’s chips, but only needs to understand how
Intel’s chip(s) support(s) the Windows operating system. If Microsoft did require the
knowledge embedded in the process, they would bring Intel within the boundaries of their
firm, as suggested by Transaction Cost Economics. Therefore, although this network is
associated with a low state of absorptive capacity, given the presence of “sequencing” as
a basis for knowledge transfer, absorptive capacity is expected to have less significance
in this network setting. As a result, the low states of causal and outcome ambiguity
would be expected to play a more significant role in this network type. Therefore –
Hypothesis 6d – The value chain network will be associated with a low state of
knowledge transfer difficulty.
The influence of the co-opetive network type
Absorptive Capacity
As with the franchise network type, the co-opetive type is characterized by narrow scope.
Again narrow scope provides a fertile environment for the four commonalities that are
contributors to absorptive capacity – language, process, base knowledge and problem
52
solving. Because the members of this network would be expected to share these
commononalities, the type is expected to exhibit high absorptive capacity.
Hypothesis 7a - The co-opetive network type will be associated with a high state of interorganizational absorptive capacity.
Causal Ambiguity
Although the narrow scope of this network type would be expected to reduce the causal
ambiguity associated with inputs and causal factors and converge to Low Causal
Ambiguity, the significant risk of opportunism, as a function of intense competition, may
override this. Specifically, in a network environment where the organizations all engage
in the same processes, it is through causal ambiguity that they can develop some level of
competitive advantage. A high state of causal ambiguity therefore is the goal that each
entity in the network strives to achieve. High causal ambiguity creates a barrier to
imitation and represents an opportune area for investment (Reed and DeFillippi, 2001).
The paradox of causal ambiguity is that the very inputs that enable a competitive
advantage may be the most undervalued by a market –
“It might be argued that…inputs are undervalued because competitors fail to
recognize them…” (Lippman and Rumelt, 1982:419)
Hypothesis 7b. - The co-opetive network type will be associated with a high state of
inter-organizational causal ambiguity.
Outcome Ambiguity
I described the co-opetive network type above as characterized by a limited or narrow
scope, significant intensity of competition and a weak governance structure. However,
53
the knowledge in question may be proven or unproven depending upon the objectives of
the network.
The limited operational scope of this network type would be expected to contribute to a
bounded Knowledge Usage Set. However, in this network type, the relevant knowledge
in question may or may not be proven. Using the VISA network as an example, member
banks may be engaged in a research initiative regarding security of transactions on the
Internet. In this scenario, the knowledge in question is “unproven” – as findings are
made known, no bank would have had previous experience with the new knowledge.
Alternatively, banks within the VISA provide knowledge on a regular basis regarding
fraud activity and fraud reduction practices – “proven knowledge” with which most
banks would have had previous experience. As a result, the Knowledge Usage Set for
this network type cannot be determined as bounded or unbounded.
Given that a co-opetive network is comprised of competitive members, the intensity of
competition would be high. In addition, the network type is generally a configuration of
“peers”, with no subordination or hierarchical structure. As a result, the expected
Recipient Action Sets would be unbounded, due to the high risk of opportunistic
behavior.
As a result, the expected outcome ambiguity for this network would either be
Type 3 or Type 4 (Figure 2).
In the horizontal value chain, which I explained to be a special case of the co-opetive
network type (Section 2.3), Lee and Whang (2000) found that organizations were
concerned about the confidentiality of what was shared vertically in the presence of
horizontal competition. Li (2002) describes this as the “leakage effect”. In this scenario,
54
the motivations of both the intended and the unintended recipients must be considered –
contributing to an unbounded Recipient Action Set.
Hypothesis 7c(1) – When the Knowledge Usage Set is bounded, the co-opetive network
type will be associated with a medium-high state outcome ambiguity (Type 3).
Hypothesis 7c(2) - When the Knowledge Usage Set is unbounded, the co-opetive network
type will be associated with a high state outcome ambiguity (Type 4).
Like the franchise network type, the high state of inter-organizational absorptive capacity
creates a rich environment for knowledge transfer to occur. However, unlike the
franchise network type, the co-opetive network type contributes to a high state of interorganizational causal and outcome ambiguity, fueled by competition and the risk of
opportunism, and the lack of punishment for opportunistic behavior. Consequently,
knowledge would be expected to transfer with difficulty.
Hypothesis 7d – The co-opetive network form will be associated with a high state of
knowledge transfer difficulty.
2.5 Chapter Summary
The guiding research questions of this dissertation address the relative effects of
absorptive capacity, causal ambiguity and outcome ambiguity on knowledge transfer
difficulty and how these factors vary with inter-organizational network type. This
chapter developed the two parts of the model that I used to evaluate these research
questions.
55
Transaction Cost Economics and the Knowledge Based View of the Firm, together
represent the theoretical basis that I used to develop the model and nineteen hypotheses
detailed in this Section.
Several factors have been shown to influence knowledge transfer. Based upon the KBV,
and the work of Cohen and Levinthal (1990), Szulanski (1996), and others, the first factor
identified was absorptive capacity. The definition developed by Cohen and Levinthal
(Section 2.2.1) has become the predominant working definition across almost all studies
involving absorptive capacity, and has been shown consistently to be one of the most
commonly cited influencing factor of knowledge transfer difficulty. The second factor of
knowledge transfer was causal ambiguity. The influence of this factor is based, in part,
on the concept of information impactedness from the TCE, as well as work by
Mosakowski (1997), Szulanski (1996) and others. The third factor, outcome ambiguity,
represents a unique contribution of this dissertation (in the context of knowledge
transfer). The extent to which the Knowledge Usage Set and the Recipient Action Sets
can be bound, the state of outcome ambiguity can be determined. The final two factors,
knowledge type and technology, are recognized as important, but considered to be
invariant with inter-organizational network type and will not be included in the empirical
study.
I then reviewed the various organizational types, based upon the characteristics of
governance structure, intensity of competition and scope of each network type. Finally,
56
using the three factors and the four organizational types, I stated my specified model in
two parts, including the nineteen hypotheses.
The remainder of this dissertation is organized as follows. The next section will provide
detailed overview of the research methodology. Section 4 will provide the pilot and field
study results, including hypothesis testing results. The results will be discussed in
Section 5. And finally, the concluding section, Section 6, includes the implications of the
study results for practitioners and researchers.
57
3.0 METHODOLOGY
3.1 Construct Operationalization Review and Approach
The hypotheses developed in this dissertation focus directly on the relative effects of the
factors of knowledge transfer difficulty – absorptive capacity, causal ambiguity, and
outcome ambiguity. The first two factors, along with knowledge transfer, have been
operationalized in many different ways through many different studies. Although the
operationalization of the constructs of absorptive capacity and causal ambiguity in the
specific context of knowledge transfer difficulty within inter-organizational networks has
never been developed, I benefited from understanding how these constructs were
operationalized in different intra-organizational contexts, and some dyadic interorganizational contexts. Following the three stage process of survey instrument
development developed by Churchill (1979) – see Figure 6 – the first stage involves the
specification and development of measurement items based upon a thorough literature
review and utilization of previously validated items, where possible. Using this guide, a
brief summarization of previous operationalizations is outlined below.
3.1.1 Knowledge Transfer Difficulty
The variable “knowledge transfer difficulty” has received a great deal of research
attention. As a result, it has been operationalized in many ways. The two most common
methods of operationalization include different measurement items within questionnaires
and analysis of operational learning curves using secondary data.
58
In his widely cited work on knowledge “stickiness”, Szulanski (1996), using a survey
instrument, developed eight measurement items divided across three “technical success”
indicators of an intra-organizational project to determine the extent to which knowledge
transferred within an organization. These indicators, or constructs included time
expended, budget and overall client satisfaction. Similarly, Hansen (2002) used product
completion time as the construct of knowledge transfer in his survey of multi-unit
organizations. Birkinshaw et al. (2002) used survey data to determine if the
characteristics of knowledge, including its transferability, can be used to predict
organizational structure. Finally, Tsai (2001) used several survey questions which loaded
onto the two factors of “innovation” and “performance” to measure knowledge transfer.
A listing of measurement items from these studies can be found in Table 3 (with the
exception of the Tsai (2001) study which did not report the actual survey questions).
A second method of operationalization of knowledge transfer difficulty is analysis of an
organization’s “learning curve” to determine if knowledge has been incorporated into the
established processes.
The study by Darr et al. (1995), which examined knowledge
transfer among pizza franchises, operationalized the variable through analysis of
secondary performance data. Specifically, the cumulative number of units produced was
used as a proxy variable for knowledge acquisition. Lapre and Van Wassenhove (2001)
also used the learning curve for their analysis of knowledge transfer among production
lines in a Belgian manufacturing plant.
59
Using a combination of these approaches, I operationalized knowledge transfer difficulty,
the dependent variable of this dissertation, through constructs and measurement items,
based upon measures developed for previous studies. I developed 8 measurement items
for the pilot study to evaluate knowledge transfer difficulty (See Appendix II for the final
list of measurement items for the field study). I also measured performance directly
using secondary data provided by my data sources (described in the next section).
Similar to the analysis of learning curves as an indicator of knowledge transfer, I
recorded the sales per FTE (full time employee) for branches in the SunTrust franchise
network and the net worth-to-total assets for the credit union members of CUNA as a
performance guage. Because these metrics are ratios, they are effectively “normalized”
for comparison of variance within the two networks.
3.1.2 Absorptive Capacity
Of the factors of knowledge transfer difficulty, absorptive capacity has received the
greatest amount of formal research attention. In their seminal work on absorptive
capacity, Cohen and Levinthal (1990) reported R&D intensity within the context of a
survey as a proxy for absorptive capacity. They reasoned, that “…a firm’s ability to
exploit external knowledge is often generated as a by-product of its R&D…”. They
specifically identified R&D intensity as R&D expenditures as a percentage of business
unit sales. Similarly, George et al. (2001) used R&D spending in their study of biopharmaceutical firms as a measure of absorptive capacity. Although R&D spending has
been used as a surrogate measure of absorptive capacity, this assumes that there is no
differentiation between the two as a cause and an effect. I view R&D spending as a
60
control decision that can be set as high or low, and, therefore may not be an indication of
absorptive capacity. That is, R&D spending may lead to absorptive capacity, but may not
necessarily indicate the degree of “current” actual absorptive capacity.
Absorptive capacity is one of the nine independent variables Szulanski (1996) used in his
study of intra-organizational knowledge trasnfer. The constructs that he used to
operationalize the variable focused on commonalities of language and knowledge base.
Similarly, Lane and Lubatkin (1998) used questions addressing commonalities of
operations, knowledge base and of problem solving in their measurement of absorptive
capacity. In their survey of IT-related absorptive capacity, Boynton et al. (1994) asked
questions related to managerial IT knowledge and IT management process effectiveness.
Again, where available, a list of survey questions is included in Table 3.
In an effort to measure this latent variable, I defined the four sub-constructs to be the four
commonalities of language, base knowledge, processes and problem solving, based, in
large part on the studies of Szulanski (1996) and Lane and Lubatkin (1998). I developed
16 measurement items (Appendix II) to evaluate these four sub constructs.
3.1.3 Causal Ambiguity
Three prominent studies were found that operationalized the variable causal ambiguity.
Szulanski (1996) used causal ambiguity as one of his nine independent variables.
Szulanki focused on “fundamentally irreducible” causal ambiguity –
61
“When the precise reasons for success or failure cannot be determined, even ex post,
causal ambiguity is present and it is impossible to produce an unambiguous list of the
factors of production”.
He developed 7 items for measurement of this concept (Table 3).
Using survey data to assess causal ambiguity in hospitals and the textile industry,
Wilcox-King and Zeithaml (2001) focused on two constructs of causal ambiguity. The
first, “Linkage Causal Ambiguity”, is based upon ambiguity regarding the link between a
competency and its competitive advantage. The second, “Characteristic Causal
Ambiguity”, is based upon ambiguity regarding the competency in question. Where
available, the specific questions that have been used for the construct are included in
Table 3.
Mosakowski (1997) used a combination of illustrative examples, a secondary case study
and analysis of secondary data to examine the causal ambiguity variable. In the context
of her analysis, she measures causal ambiguity using a monotonically decreasing function
of a firm’s age – the natural logarithm of firm age is used to represent a function in which
the rate of causal ambiguity decreases over time.
Although Mosakowski did measure this variable directly using secondary data, I agree
with the position taken by Szulanski (1996) and Wilcox-King and Zeithaml (2001) that
the variable is latent and cannot be measured directly with validity using a secondary data
source. A total of 7 measurement items were developed to measure these two constructs
62
of causal ambiguity for the pilot survey. For an explanation of the measurement items,
see Appendix II.
3.1.4 Outcome Ambiguity
In this dissertation, I define outcome ambiguity as “the inability by the knowledge source
to identify the possible outcomes associated with knowledge transfer”. I distinguish
between two sources of outcome ambiguity, which comprise its two formative sub
constructs. The first is a manipulated measurement to measure the provenness of the
knowledge in question, or the boundedness of the Knowledge Usage Set. Where
possible, these questions were based upon Szulanski (1996). The second source of
outcome ambiguity is the relationship between the source and the recipient. The actual
measurement items were developed based upon the work of Simonin (1999) from his
exploration of the role of ambiguity on knowledge transfer and Dyer (1997) from his
article examining trust between suppliers and manufacturers. A total of 12 measurement
items were developed for outcome ambiguity (Appendix II).
3.2 Study Approach
In an effort to test the complete model, including Parts I and II, specified in Section 2.5, I
examined two inter-organizational network types – the franchise network and the coopetive network. These types were chosen because of their respective unique roles in the
study of inter-organizational networks. The franchise type is appropriate to include
because it has received a significant amount of previous research attention. Therefore,
the findings associated with this network type should be more confirmatory. The specific
63
franchise studied was the SunTrust Atlanta Branch network. This network included 165
branch members. Consistent with the definition developed in Section 2.3.1 of a franchise
network, the branches are engaged in the transfer of primarily proven knowledge, they
exhibit low operational scope, and are part of a highly structure centralized hierarchy6.
The second type, the co-opetive network, is particularly appropriate to include because it
has received the least amount of previous research attention. Therefore, results from this
network type could have the greatest opportunity to make contributions to the existing
literature. The specific co-opetive network studied was the Credit Union National
Association (CUNA). CUNA members participate in the network for the purpose of
lowering transaction costs, in part by accessing existing knowledge at CUNA regarding
operational best practices, marketing materials, etc. From the CUNA website:
“CUNA (Credit Union National Association), based in Washington, D.C., and Madison,
Wisconsin, is the premier national trade association serving America's credit unions. The
not-for-profit trade group is governed by volunteer directors who are elected by their
credit union peers….CUNA provides many services to credit unions, including
representation, information, public relations, continuing professional education, and
business development. “ (www.cuna.org/cuna/index.html)
Although CUNA credit union members are part of a single network that affords them
access to existing knowledge (where the emphasis is on knowledge transfer rather than
knowledge creation), CUNA members are considered to be in intense competition with
each other, with a high risk of opportunistic behavior. Mark Condon, Senior Vice
6
I am conscious of the fact that arguments could be made that bank branches are extensions of the same
company and they don’t voluntarily choose to participate. However, my experience, which is consistent
with my discussions with SunTrust, is that the bank branches operate independently and their degree of
“active” participation in the branch network functions and communications vary significantly. As a result,
there is justification to believe that the branches will interact in a manner that would be consistent with an
inter-organizational franchise network.
64
President of Association Services and Research for CUNA, stated that he would
characterize the intensity of competition within CUNA as “a 7 or higher” on a scale of 110, with 10 representing the most intense degree of competition (see Appendix 1 for a
record of study interviews). CUNA is also representative of a co-opetive network, in
part, because the members engage in similar operations – creating very low operational
scope. Finally, CUNA has a hierarchical and formalized, albeit loose, governance
structure, where CUNA provides forums for discussion and communication, but holds
little authority to punish for opportunistic behavior.
3.3 Data Collection
Primary data was required to answer the guiding research questions posed by this
dissertation. Specifically, I tested Hypotheses 1, 2, 3 – the relative effects of the factors
of knowledge transfer difficulty – and Hypotheses 4a, 4b, 4c, 4d and 7a, 7b, 7c, and 7d –
the expected states of the factors of knowledge transfer difficulty for a franchise network
and a co-opetive network, respectively.
3.3.1 Sample Description
Credit unions within the CUNA network range in size from a few million dollars in assets
to the Navy Federal Credit Union with over $16 billion in assets. After several
interviews with CEOs of credit unions of varying sizes (Appendix I), it became clear that
abilities to transfer knowledge as well as its importance to the operations of the credit
union vary with the size of the operation. Therefore, I controlled for size by only
surveying CEOs from credit unions with assets between $5 million and $50 million.
65
This segment had approximately 600 credit unions. Because CUNA had limited
confidence with the email addresses on file, surveys were mailed.
The 165 SunTrust branches in the Atlanta Region represented the sample for the
franchise network type.
3.4 Data Analysis Methodology
Spender and Grant (1996) highlight the frustrations of empirical researchers studying the
factors related to knowledge transfer –
“…there is a growing realization that the variables which are most theoretically
interesting are those which are least identifiable and measurable”.
Similarly, Mowery and Oxley (1996) further highlight that empirical research on
knowledge transfer has suffered because of widespread reliance on anecdotes and
assertions rather than on primary data and statistical analysis.
One of the contributions of this paper is a departure from these anecdotes and assertions
through an organized attempt to provide quantitative evidence for the specified model
developed in Section 2.5 using Partial Least Squares (PLS).
The PLS technique was selected over co-variance based Structural Equation Modeling,
for two primary reasons - its ability to accommodate smaller sample sizes and its unique
ability to accommodate formative variables in the structural model.
66
PLS does not require parametric assumptions. As a result, it is especially suited for the
analysis of small data set as well as data that does not necessarily exhibit multivariate
normality, as required by covariance-based SEM (Chin, 1998). This characteristic of
PLS is in contrast to covariance-based SEM, which requires a sample of at least 150
(Bollen, 1989), because of the sensitivity of χ2 to sample size (Bollen, 1989; Hair et al.,
1998).
Second, PLS supports both formative as well as reflective constructs. Formative
variables, as the name implies, “cause” or form the latent construct. In this dissertation,
the constructs of absorptive capacity, causal ambiguity and outcome ambiguity were
formative. Reflective variables, “reflect” the latent construct and as a representation of
the construct, should be unidimensional and correlated (Gerbing and Anderson, 1988). In
this study, the sub constructs (e.g., common language for absorptive capacity) were
reflective. Chin (1998) highlights that, where possible, reflective variables are preferable
to formative variables when the objective is related to theory testing rather than theory
development. Covariance-based SEM does not readily support formative constructs.
Gefen et al. (2000) provide an excellent comparison of the two techniques as well as
regression analysis, including guidelines for researchers.
A brief overview of these comparisons is highlighted below.
First, the overall objective of PLS and regression is to demonstrate high R2 values and
significant t-values, thus rejecting the null hypothesis of no effect (Thompson et al.,
1995). On the other hand, the objective of SEM is to demonstrate that the null
67
hypotheses (the assumed model with all paths) is insignificant given the sample data,
potentially making the hurdle for “success” more difficult. In addition, the goodness of
fit tests for SEM, such as χ2, test the restrictions implied by a model…including paths not
explicitly stated by the analyst. These “implicit” paths are not tested nor identified in
PLS or regression.
Another important difference between SEM and PLS is that covariance-based SEM,
unlike PLS, enable an assessment of unidimensionality of the measurement items. This
is the degree to which items load only on their respective constructs without
demonstrating “parallel correlational pattern(s)” (Segars and Grover, 1998). In factor
analysis terms, unidimensionality means that the items reflecting a single factor have
only that one shared underlying factor among them. Although cross loadings can be
examined, unidimensionality cannot be assessed using the PLS or the traditional
techniques of factor analysis or Cronbach’s alpha. (Gerbing and Anderson, 1988).
A third difference between SEM and PLS involves the algorithms used. Covariancebased SEM uses maximum likelihood based model fitting to compare the covariance
structure fit of the model to a best possible covariance structure (hence the use of χ2).
The variance fit indices provide the researcher with evidence regarding how closely the
proposed model fits the data versus the best possible covariance structure. Alternatively,
PLS is designed to explain variance in the dependent variable. Consequently, PLS is
more appropriate for predictive applications and theory development. Using ordinary
least squares as the estimation technique, PLS generates an iterative set of factor analyses
68
combined with path analyses until the differences among the R2 values converge
(Thompson et al., 1995).
The software used in this analysis included SAS v.8.2 and PLSGraph 3.0 (Chin, 2001).
69
4.
4.1
QUANTITATIVE RESULTS AND ANALYSIS
Pilot Study
The model depicted in Figure 5 includes constructs representing the independent
variables of absorptive capacity, causal ambiguity, outcome ambiguity, and one construct
representing the model’s dependent variable – knowledge transfer. The three mediating
independent variables – absorptive capacity, causal ambiguity and outcome ambiguity –
as developed in Section 2.2.1, represent formative constructs with multiple subconstructs. The specific operationalization of these constructs and the respective subconstructs was outlined in Section 3.1.
Following Churchill’s established model of survey instrument development, interviews
were conducted with sample respondents and practitioners in knowledge management
(Appendix 1). The specific impacts to the survey instrument from each interview can be
seen in Table 5.
The pilot survey was sent to 100 credit unions managers across the United States in May,
2003 (Appendix IV). The members of the CUNA network were selected for the pilot
study because of the relatively large population size – approximately 550 members. On
the other hand, SunTrust branch managers were not included as part of the pilot survey
because of the relatively small population size – 165. In an effort to capture any
differences that may exist between credit unions that perceived their operation to be
highly integrated within the CUNA (Credit Union National Association) versus those
credit unions that perceived their operation to be loosely (or not at all) affiliated with
70
CUNA (even though their credit union is a registered member of CUNA), I included a
pre-screening question:
As a member of CUNA (the Credit Union National Association), your credit union may
receive, as well as provide, information on a variety of topics, including operations,
marketing, compliance, pending legislation, etc. You may also take advantage of other
opportunities to interact with CUNA and/or other credit union members of CUNA.
Considering your position as a credit union member of CUNA, which statement most
closely reflects your views:

A. Our credit union is very integrated within CUNA. We place a high value on
our membership.

B. Our credit union has limited interaction within CUNA. We don’t really
identify with the CUNA network.
The responses of credit union managers who answered “B”, were treated as a proxy
“control” group in the field study as non-networked or independent operations.
The survey included a total of 50 statements (two statements were not applicable for
those respondents answering “B” to the pre-screen question). Respondents were asked to
rate each statement on a typical Likert scale of 1-7 in terms of how representative each
statement was to their credit union. A score of “1” indicated that the statement was not
representative of their credit union and a score of “7” indicated that the statement was
completely representative of their credit union. An example statement:
Our credit union has policies (either formal or informal) to restrict the sharing of
knowledge with other credit unions.
1
2
3
4
71
5
6
7
The pilot survey generated a total of 39 usable returned surveys, including 30 surveys
that came from “integrated” credit unions (answer “A” to the pre-screen question) and 9
surveys that came from “independent” credit unions (answer “B” to the pre-screen
question). These responses were then analyzed to determine scale reliability.
Each measurement item (survey question) was intended to be reflective of the respective
(sub) constructs. Research literature provides a variety of metrics to determine the extent
to which measurement items truly represent a reliable measurement of a larger construct.
Again, following the methodology outlined by Churchill (1979), in an effort to determine
if the items used in this pilot reliably measure the (sub) constructs in this study, three
scale reliability metrics were utilized – Average Variance Extracted (AVE), Fornell and
Larcker Rho (or Construct Reliability Metric) and the traditional Cronbach Alpha metric
(Hair, et al., 1998; Nunnally and Bernstein, 1994). The first two are calculated using the
PLS loadings, while the third is calculated using the inter-item correlations.
The AVE measure reflects the overall amount of variance in the indicators accounted for
by the construct through the ratio of the sum of captured variance and the measurement
error. Higher variance extracted values occur when the indicators are truly representative
of the latent construct. Guidelines suggest that the average variance extracted should
exceed .50 for a construct (Hair et al., 1998). The Fornell and Larcker Rho value is also
used to examine the shared variance among measurements for a construct. It is often
used in conjunction with the AVE, since it does not measure the amount of variance that
is captured due to measurement error, where AVE does (Fornell and Larcker, 1981). The
expectation is a result above .70. Cronbach’s alpha is the most commonly applied
72
measure of scale reliability. The generally agreed upon lower limit is .70 (Hair, et al.,
1998).
Although most of the (sub) constructs included in the pilot survey were found to be
reliable (based upon the three metrics above), several (sub) constructs generated
unacceptable scores and represented opportunities for refinement of the survey for the
field study. The complete unaltered results (no items dropped) for the pilot study can be
seen in Table 6.
After reviewing the results from the pilot study, changes were made in an effort to
increase the validity scores where they were unacceptably low. Most changes were made
after interviews with representatives from the two respective networks. Details of all
changes made can be seen in Table 5. Examples include:
1. Rephrasing of statements. In the pilot survey instrument, the first four statements
utilized the phrase “operational processes”. Cookie Yoder, CEO of the Pittsburgh
Federal Credit Union, provided feedback that although all areas of process were
included parenthetically (e.g., marketing, risk, compliance, etc.), respondents may
have biased their responses to reflect knowledge transfer in only the specific
“operations” areas, using a more restricted definition of “operations”. As a result
of this conversation, the phrasing was changed to only utilize the term
“processes” with “operations” used within the parentheses with the other areas
such as marketing and compliance.
2. Addition of statements. Kirk Watkins, VP of Atlanta Region Management for
Branch Banking, stated that because of the existence of an “artificial” competition
73
among branch managers (see Interview 4S in Appendix 1 for detail), managers
may take it upon themselves to create (formal or informal) policies to restrict
knowledge transfer. This conversation resulted in the addition of two statements
that eventually became statements 39 and 41 in the field study survey.
4.2
Field Study Results
4.2.1 Description of Populations and Responses
Following changes to the pilot survey, the field survey was mailed to the two populations.
The final field survey and example cover letters are included as Appendix IV.
The first population represented the co-opetive network described in Section 2.3. This
population included CEOs/Presidents of credit unions within the Credit Union National
Association (CUNA), a non-profit network which provides credit unions with centralized
support such as education for employees, a data and information repository, public
relations, lobbying and legislative support, product development support, etc.7 CUNA
has approximately 10,000 credit union members (because their membership numbers
change quarterly, approximate numbers are provided). Credit unions can be roughly
categorized into three groups:

Those with a “low income” designation that provide financial services to
individuals who would otherwise not qualify for traditional banking
relationships. Approximately 1,000 credit unions have this designation;
7
http://www.CUNA.org
74

Those with an “associated” designation that are affiliated with a corporation, a
religious organization, the military, a university, etc. These credit unions
generally have an effective monopoly on their constituents and experience little
competition with other credit unions. Approximately 7,000 credit unions have
this designation;

The remaining credit unions are locally chartered, in non-low income areas and
with no affiliation. Their clientele generally qualify for traditional banking
services, but they have elected to utilize the services of a credit union because of
favorable rates, service, product offerings, location, etc. These credit unions are
generally found in smaller communities across the country that may not have
many traditional banking outlets.
Because of the specific nature of the first two groups, I limited the population to credit
unions falling into the third category. Given the subjects of interest (e.g., common
operating processes, the factors generating competitive advantage, use of a common
language, etc), the third group appeared to represent a population that would be more
diverse, making the results from the third group not only more generalizable, but also
more representative of a co-opetive network than results from the first two groups.
Credit unions within this third category ranged in size from a few million dollars in assets
under management to billions of dollars in assets under management. Given the study
objectives, Mark Condon, Senior Vice President for Services, recommended a target of
credit unions with between $5 million and $50 million in assets under management. This
range was chosen, because, in his opinion, as the credit unions become larger (e.g.,
75
greater than $100 million), their interaction within CUNA and with other credit unions
decreases. As a result, the final population selected for study was credit unions within
this range of assets under management. This provided a population of approximately 550
credit unions.
The second population represented the franchise network described in Section 2.3. This
population also came from the financial services industry. SunTrust Bank in Atlanta, GA
provided me with access to all 165 of their Atlanta Region branch managers, which
report to a single region manager. These branch managers were considered to comprise a
franchise network.
Field study surveys were mailed during the second week of July, 2003. By the second
week of September, 2003 a total of 101 useable surveys were returned by credit union
managers, for a total response rate of 18.3%, slightly less than the expected 20% (Straub,
1998). These responses included 68 “integrated” credit unions and 33 “independent”
credit unions. A total of 70 useable surveys were returned by SunTrust branch managers
for a total response rate of 43%. Descriptive statistics on survey responses can be found
in Table 7a, 7b and 7c.
4.2.2 Assessment of Field Study Scale Reliabilities
Following from Churchill’s (1979) methodology, assessing the validity of the proposed
(sub) constructs, including the convergent and divergent validities of the measures,
necessitated the implementation of two validation techniques. This is due to the
reflective and formative nature of the different constructs.
76
4.2.2.1 Validation of Reflective Measures
Within the specified model (Figure 5), eight sub constructs and one construct were
proposed to exist. Each of these (sub) constructs had reflective measures. As with the
pilot data, assessment of instrument validity within PLS was accomplished through the
measurement model, where the relationship between the observed variables and the latent
(sub) constructs is specified (Igbaria et al., 1995). In assessing the field study
measurement model, Cronbach’ alpha, the Fornell and Larcker Rho and the Average
Variance Extracted (AVE) were again used to determine reliability.
As can be seen in Table 8, all (sub) constructs, met or surpassed the minimum
requirements for the three respective metrics. Indicators were retained in any given scale
where the loadings exceeded a value of .70, where it is argued that 50% of the variance is
explained by the construct (Fornell, 1982). The convergent validity of the construct as a
whole is in turn increased through the inclusion of only the items with loadings in excess
of .70.
In addition to evaluation of the measurement model across all respondent data, the
respective measurement models were also evaluated for each of the two network samples.
Using the same retained measurement items as for the larger model, all minimum validity
values were achieved (see Table 8).
The discriminant validity, or the extent to which indicators differentiate among (sub)
constructs, was also explored. The square root of the AVE of the measures and the
correlations among the measures were examined in an effort to establish discriminant
77
validity (Chin, 1998; Klein, 2002). Where the square root of the AVE of a measure
exceeded the correlations between the measure and all other measures, the discriminant
validity of the measure is assumed to be adequate. In all cases the square roots of the
AVEs well exceeded the intercorrelations of the other constructs, indicating that no
discriminant validity issues were present. The results of this analysis can be seen in
Table 9.
4.2.2.2 Validation of Formative Measures
As detailed in the section above, AVE represents a metric through which to assess the
overall variance in the indicators captured by the latent construct (Table 8) as well as the
discriminant validity or differentiation among the constructs (Table 9). AVE analysis
assumes that all measures are reflective and therefore cannot be used to test formative
constructs. An alternate measure of convergent and divergent validity includes the usage
of a variation of the Campbell and Fiske (1959) multitrait-multimethod analysis (Klein,
2002). In testing formative constructs, it is proposed that formative items be correlated
with a “global item that summarizes the essence of the construct” formula employed by
Diamantopolous (2001).
Although the multitrait-multimethod analysis for measuring constructs does not directly
apply here, two comparative analyses were performed. First, the correlations between the
reflective items and each respective sub construct were compared. Second, the
correlations between the formative sub construct and each construct were compared.
This was done through the derivation of a weighted score, based upon PLS generated
weights, for all items believed to measure the same construct.
78
In assessing convergent validity, the expectation is that measures believed to be part of
the same (sub) construct will correlate at a significant level with one another (Campbell
and Fiske, 1959). Using the PLS weighted composite scores for each construct and the
original items, the significance of the individual measure correlations with the weighted
composite sub constructs, was analyzed.
In assessing discriminant validity, the expectation is that the inter-item and item-toconstruct correlations should correlate at a higher level with each other and their
composite construct score than with the measures of other constructs and or other
composite scores (Klein, 2002).
As can be seen in Table 10a, all measures that are part of the same sub construct correlate
highly with one another, indicating that no convergent validity issues exist. In addition,
all item-to-construct correlations are higher among construct items than between items
with other constructs.
4.2.3 Hypothesis Testing
The specified model in Figure 5, was developed in two parts. Hypotheses 1, 2 and 3 were
developed in Section 2.2.1 as Part I of the model, to explain the relative effects of
absorptive capacity, causal ambiguity and outcome ambiguity with knowledge transfer
difficulty. Hypotheses 4a, 4b, 4c, 4d and 7a, 7b, 7c, and 7d were developed as Part II of
the model to explain the relationship between the two network types and absorptive
capacity, causal ambiguity, outcome ambiguity and knowledge transfer difficulty.
Hypothesis testing was completed through a PLS model for the first set of hypotheses and
79
through PLS-derived composite scores using ANOVA for the second set of hypotheses.
Discussion and interpretation of these results will follow in Section 5.
4.2.3.1 Hypothesis Testing – Part I
The PLS model, which incorporated all respondents (n=171), evaluated the explanatory
effects of the three factors of knowledge transfer difficulty. The PLS model generated an
R2 value of .199, with absorptive capacity providing most of the explained variance in
knowledge transfer difficulty. The first hypothesis stated that absorptive capacity would
have a negative relationship with inter-organizational network knowledge transfer
difficulty. This hypothesis was found to be supported (path coefficient of -.417 and tstatistic of -4.425) and represents further confirmation of the significance of absorptive
capacity as a factor of knowledge transfer difficulty (Szulanski, 1996). Specifically, as
absorptive capacity – defined here as a formative construct of common language,
common processes, common problem solving and common knowledge – increases, the
difficulties associated with the transfer of inter-organizational knowledge decreases.
The PLS model was run for each of the two networks individually as well (the overall R2
values were .502 for the franchise network model and .223 for the co-opetive network
model). The findings were similar – absorptive capacity was found to have a significant
negative relationship with knowledge transfer difficulty for both network types.
The second hypothesis stated that causal ambiguity would have a positive relationship
with inter-organizational network knowledge transfer difficulty. Although the path
coefficient for this variable was positive (.035), the result was not significant (t statistic =
80
.313). As a result, the hypothesis was not supported. The construct of causal ambiguity
is formative, comprised of the two sub-constructs of causal factors (understanding of the
inputs and factors that generate outcomes) and competitive advantage (understanding of
the factors that create a competitive advantage). I thought that perhaps one of the sub
constructs would be found to have a significant relationship with knowledge transfer
difficulty. This was not the case – even at the sub construct level, there were no
significant relationships. In addition, the hypothesis was evaluated for the two networks,
and was still found to be insignificant (t-statistics of .261 for the franchise network and
.403 for the co-opetive network).
The third hypothesis stated that outcome ambiguity would have a negative relationship
with inter-organizational network knowledge transfer difficulty. As with causal
ambiguity, the relationship, although positive, was not significant (t-statistic of .209).
The relationship was also tested at the network level. It was found to be highly
significant for the franchise network (path coefficient of .399 and t-statistic of 1.833), but
insignificant for the co-opetive network (path coefficient of .018 and t-statistic of .068).
From Part I of the specified model, I found that absorptive capacity was a significant
contributor to knowledge transfer difficulty in both network types – consistent with
previous research findings. Causal ambiguity was not found to be a significant factor in
any of the tested settings. Since causal ambiguity was not tested previously in a multiorganizational context, it may be an indication that the factor is not meaningful outside of
the intra-organizational setting, which raises an important question for future research.
Finally, outcome ambiguity was determined to be significant for the franchise network
81
setting, but not for the co-opetive or global setting. This finding may indicate that the
source-recipient relationship is a stronger contributor to knowledge transfer difficulty in a
franchise network type than in a co-opetive network type. These results are summarized
in Table 11, and a detailed discussion and interpretation of results can be found in Section
5.
4.2.3.2 Hypothesis Testing – Part II
The proposed hypotheses outlined in Section 2.5 were evaluated through numerous
ANOVAs, using Tukey’s post hoc pair-wise comparison test (Neter, et al., 1996), which
are summarized in Table 12.
An issue that was determined only after the data was collected, was the (in retrospect)
problematic wording of the hypotheses. For example, Hypothesis 7a states:
The franchise network type will be associated with a high state of inter-organizational
absorptive capacity.
The adjective “high” was not defined within the hypothesis statement. As a result, it is
difficult to determine if a generated score is “high” or “low” in an absolute sense.
Therefore, the PLS derived scores were used to test the hypotheses in two ways. First,
the scores for each construct were compared between the two network forms. If one
hypothesis stated that the construct state was expected to be “high” and the other
hypothesis stated that the same construct for the second network type was expected to be
“low”, the two scores were tested for a significant difference in the predicted direction.
However, if both hypotheses stated that the construct state was expected to be “high”, and
82
the two scores indicate no statistical difference, this cannot be considered to be a
conclusion that the hypotheses were supported.8
The second evaluation of the hypotheses involved the utilization of the scores for the
non-networked collection of credit unions. This group was used as a control group. The
scores for the two networks were compared to the scores for the control group across the
four constructs. Based upon the discussion in Sections 1 and 2, both networks would be
expected to perform “better” than the control group (higher absorptive capacity scores,
lower causal ambiguity scores, lower outcome ambiguity scores and higher knowledge
transfer scores). The results for these comparisons are reported below and also
summarized in Table 12.
Hypotheses 4a and 7a addressed the construct of absorptive capacity. Both network
forms were predicted to be associated with high states of absorptive capacity. The
franchise network generated a composite score of 11.45 and the co-opetive network type
generated a composite score of 12.31. The co-opetive score was significantly higher at
the p<.05 level. This result is contrary to what was expected.
An examination at the sub construct level provided insight and explanation into these
results. The co-opetive network was found to generate a significantly higher score for
the sub-construct of absorptive capacity – common language, relative to the franchise
network (p<.10). The non-aligned entities were also found to have a higher score than
the franchise network on this construct (p<.05). Because the co-opetive network and the
8
Therefore, the results for hypotheses 4a and 7a (which both state that the network type will be associated
with a “high” state of absorptive capacity) will be reported, but no specific conclusions will be drawn
regarding hypothesis support.
83
control group were both found to generate higher scores than the franchise network, the
franchise network was viewed as the source of the contrary result. One possible rationale
for this result is the two different branch categories within the selected franchise network.
Specifically, within the SunTrust branch network, there are in-store branches (bank
branches within grocery stores) and retail branches (traditional stand alone bank
buildings). Further analysis at the branch category level indicated that the retail branches
generated a significantly higher score on this sub construct than did in-store branches
(p<.01). This differential may indicate that the terminology used within this franchise
network may not represent a truly “common language” between the two branch
categories.
The second absorptive capacity sub-construct – common process – provided additional
insight into the hypothesis results. The franchise network generated significantly higher
scores on this sub-construct than both the co-opetive network and the control group
(p<.01). This may be attributable to differences in governance structures. Specifically,
the franchise network has a more centralized governance structure with greater authority
to standardize processes. As a result, greater commonality of process for the franchise
network is not unexpected.
The third absorptive capacity sub construct – problem solving – did not produce any
differences between the two networks. This result was expected. The only difference
occurred between the franchise network type and the control group (p<.05).
84
The final absorptive capacity sub construct – common knowledge – also did not produce
any difference between the two networks. And, both networks were found to have
significantly higher scores than the control group (p<.05).
The absorptive capacity construct is formative of the four sub-constructs – common
language, common process, common problem solving and common knowledge. The
source of the difference in results for hypotheses 4a and 7a relative to expectation came
from the sub-construct of common language. The fact that the co-opetive network
generated a higher score, is attributed, in large part to the differences between the two
branch categories within the franchise network – in-store and retail.
Hypotheses 1b and 4b addressed the construct of causal ambiguity. The franchise
network type was predicted to be associated with a low state of causal ambiguity while
the co-opetive type was predicted to be associated with a high state of causal ambiguity.
Again, the construct level results did not support the hypotheses. The co-opetive network
type generated a composite score of 3.95, which was not significantly different from the
composite score of 3.55 generated by the franchise network type. However, the franchise
type was significantly lower than the control group (p<.01). Again, the comparative
results for the sub-constructs were examined to better understand these results.
The first sub-construct examined for causal ambiguity was common factors. Recall from
Section 2.2.1, that this sub-construct described the knowledge of the inputs or factors that
contribute to a particular outcome. On this sub-construct, a low score is indicative of low
ambiguity regarding these factors. As expected, the franchise network generated a
composite score that was significantly lower than the composite score for the co-opetive
85
network ( p<.01), and both networks generated composite scores that were lower than the
score for the non-networked entities ( p<.01 and p<.10, for the franchise and co-opetive
networks, respectively).
The second sub-construct for causal ambiguity was the ambiguity related to competitive
advantage – knowledge regarding what makes an entity a better or worse performer.
Again, a low score indicates low ambiguity. The franchise network and the co-opetive
network did not generate statistically different composite scores. And although the
franchise network did generate a lower score than the control group (p<.05), there was no
difference between the co-opetive network and the control group. The fact that the score
for the competitive advantage sub-construct for the franchise network was not lower (as
was expected) than the score for the co-opetive network, may be attributable to the
“artificial” competition explained by the SunTrust VP (Interview 4S in Appendix 1).
Because bank branch managers are effectively put into competition with each other,
branches may experience the causal ambiguity described by Lippman and Rumelt (1982)
– bank branch managers may attempt to prevent imitation of their capabilities by their
competitors (other bank branch managers) by purposefully making outputs causally
ambiguous. In short, “artificial” competition may be the source of “artificial” causal
ambiguity, which may lead to decreased knowledge transfer within the network.
Hypotheses 4c and 7c addressed the construct of outcome ambiguity. The franchise
network type was predicted to be associated with a low state of outcome ambiguity while
the co-opetive type was predicted to be associated with a high state of outcome
ambiguity. Results supported these hypotheses. The composite score for the franchise
86
network was lower than the composite score for the co-opetive network, significant at
p<.05. In addition, because both network types are expected to be superior to the non
network control group for each factor and, like causal ambiguity, outcome ambiguity has
a negative relationship with knowledge transfer difficulty, both network types would be
expected to generate composite scores lower than the control group. And this is in fact
what occurred (p<.01 for the franchise network and p<.10 for the co-opetive network).
Finally, Hypotheses 4d and 7d addressed the dependent construct of knowledge transfer
difficulty. Specifically, the hypotheses were stated as: The franchise (co-opetive)
network type will be associated with a low (high) state of knowledge transfer difficulty.
The ANOVA results, indicated that the two different network types did, in fact,
experience modest differences for this dependent variable (p<.10). However,
surprisingly, the non-networked entities did not score significantly differently from either
of the two network groups.
These results from Part II of the specified model, provide support for the second research
question of this dissertation – the factors of knowledge transfer difficulty do appear to
vary with inter-organizational network type.
Although absorptive capacity was hypothesized to be high for both network types, the
results indicate that absorptive capacity was found to be higher for the co-opetive
network type relative to the franchise network type. This was attributed to the fact that
the co-opetive network type generated a stronger composite score on the sub-factor of
common language. The fact that common language was the only sub factor where the co-
87
opetive type was higher, provides initial evidence for the strength of this sub construct
within this industrial setting – a commonality of language may be the strongest
contributor to absorptive capacity in a Financial Services context.
Causal ambiguity was hypothesized to be low for the franchise network and high for the
co-opetive network. Again, this hypothesis was not supported – the network types were
not found to generate a different composite score on this construct. Differences were
found between the two networks for the sub construct of causal factors (internal causal
ambiguity), but for the sub construct of competitive advantage (external or “inter” causal
ambiguity). Because the larger construct was not found to differ between the two
network types, the differences found for causal factors appear to have been negated by
the lack of difference for competitive advantage – indicating that the latter sub construct
may be the more significant in this setting.
The new construct of outcome ambiguity was found to differ between the two network
types – providing support for the hypothesis that network type is associated with different
states of outcome ambiguity. As a result, the construct is not only introduced to exist, but
is also found to vary significantly in different network settings.
Finally, knowledge transfer was found to differ between the two network types, providing
initial evidence that network type does matter for inter-organizational knowledge
transfer.
A summary of these results can be found in Table 12, and a detailed discussion of these
results can be found in Section 5.
88
4.2.5 Analysis of Non Response Bias and Specification Bias
Although the response rates for the two populations were acceptable – 43% for the
SunTrust branch managers and 18% for the CUNA CEOs/Presidents – a majority of
organizations that received the survey did not respond. This raises the question “Are the
respondents different from the non-respondents?” Clearly, if the answer is “yes”, the
study results would have limited credibility.
In an effort to determine if non-response bias was present, the key performance indicator
for each population was examined between the respondents and non-respondents for each
population. For the SunTrust branch managers, the number of sales per FTE in the
branch was identified through interviews with the Atlanta Region management as the
primary key performance indicator. For the credit union CEOs, the financial ratio Net
Worth : Total Assets was identified as the key performance indicator.
Among the branch managers who responded to the survey, the sales per FTE were 1.59,
while the sales per FTE for the branch managers who did not respond were 1.71. This
difference was not found to be statistically significant (t=1.28 and p=.21). Among the
credit union CEOs who responded to the survey, the Net Worth: Total Assets ratio was
10.49, while the same ratio was 11.17 for non-respondents. Again, this difference was
not found to be statistically significant (t=1.30 and p=.19). Consequently, non-response
bias was not considered to be present.
In Section 3.3.2, the validity issue common methods bias was recognized as a possible
challenge of this field study. Because the same individual (the branch manager for the
89
SunTrust surveys and the Credit Union CEO for the CUNA surveys) was providing
responses for both the independent variables and the dependent variable, findings could
be biased. Therefore some external validation of response was needed. The network
members’ key performance indicator was used for external validation, specifically of the
dependent variable knowledge transfer. For the SunTrust bank branches, the Knowledge
Transfer variable demonstrated a correlation of .3089, with a p value of .0183 –
indicating that those who responded that they participated in knowledge transfer within
their network demonstrated higher performance. For the credit unions, the Knowledge
Transfer variable demonstrated a correlation of .2899, with a p value of .0173 – similarly
indicating that those who responded that they participated in knowledge transfer within
their network demonstrated higher performance. Consequently, common method bias is
not considered to be present.
90
5. DISCUSSION AND INTERPRETATION OF RESULTS
This dissertation examined the relative effects of absorptive capacity, causal ambiguity
and outcome ambiguity on knowledge transfer difficulty and of network type on these
same factors. Through the course of the study, two network types were empirically
examined and compared to a “control” group of non-networked entities.
The specific contributions of this dissertation for each identified factor of knowledge
transfer difficulty are explored and interpreted below.
5.1 Absorptive Capacity
As stated early in this dissertation, absorptive capacity is well established in the
knowledge management literature as a contributor to knowledge transfer. Cohen and
Levinthal (1990) developed the most widely utilized definition of the concept as the
ability of a firm to first recognize the value of new information and then apply it to
commercial ends. Using this definition, researchers have established the link between
absorptive capacity and knowledge transfer (Szulanski, 1996; Van den Bosch, 1999;
Boynton, 1994; George et al., 2001).
The findings from Part I of the specified model related to absorptive capacity were
consistent with prior theoretical work – increased absorptive capacity decreases the
difficulties associated with knowledge transfer. This finding was consistent for each of
the two network types studied.
91
The state of absorptive capacity was found to be higher for the co-opetive network
relative to the franchise network. This difference was unexpected. Evaluation of the
contributing sub factors provided insight into this finding.
Common language was higher for the co-opetive network. Simon (1981) identified
“intensity of interaction” as a guide for organizational structure, where “intense”
interactions (defined as depth and frequency) facilitate common or shared language.
Grant (1996) and others have referenced this work as a theoretical basis for
organizational structure – individuals who are “reciprocally interdependent” (Grant,
1996) will be closer organizationally than individuals who relate through “pooled
interdependence” – a looser affiliation based upon common resources. Given the ability
of the franchise network to institute standards through its hierarchical structure, it would
have been expected to have exhibited a greater common language among members than
the more loosely governed co-opetive network. This research found the opposite to be
true. Consideration of network structure suggests two possible explanations. First, the
greater degree of competition that exists within the co-opetive network, relative to the
franchise network, may have contributed to a new level of intensity that facilitates greater
shared language9. Second, it may be that Simon’s explanation of intensity of interaction
as a basis for shared language is only a guide in an intra-organizational environment, and
is less relevant when the environment includes other organizations.
Simon (1981) also argued that hierarchy was an effective mechanism for coordination
and standardization of process. Therefore, the finding that the more hierarchical network
9
Although some competition was found to exist in the franchise network studied, the level of competition
was lower than the level of competition which was found to exist in the co-opetive network studied.
92
type – the franchise network – exhibited greater common process than the less
hierarchical network type, was not a surprise. The other two sub factors of absorptive
capacity – common knowledge and common problem solving – were not found to vary
between the two network types.
Given the fact that common language was the only sub factor of absorptive capacity
found to be higher for the co-opetive network and the overall factor of absorptive
capacity was found to be higher for the co-opetive network, it would be logical to
conclude that common language has a stronger contribution to absorptive capacity than
do any of the other sub factors identified by Cohen and Levinthal (1990) and Lane and
Lubatkin (1998). This finding provides insight into the respective contributions of each
element of the concept of absorptive capacity and provides initial evidence suggesting
that common language, at least within the context studied, appears to be the primary
contributor to absorptive capacity. This finding is logical within the context of Financial
Services. Unlike other industrial settings like manufacturing, retail, pharmaceutical or
biotech, there is no tangible “product” in a financial services environment. As a result,
there is nothing to complement the utilization of language in the transfer of knowledge;
there is nothing to “point to” or to provide as a further explanation of a concept.
Therefore, it would be logical that in a service based industrial setting like Financial
Services or Management Consulting, commonality of language could play a particularly
significant role in the development of absorptive capacity. This differentiation has not
been made in previous literature.
93
5.2 Causal Ambiguity
Most researchers have approached causal ambiguity from one of two perspectives. The
first perspective is of an “intra” ambiguity related to the inputs and factors that generate
an outcome within a single firm (Mosakowski, 1997; Barney, 1999; Szulanski, 1996).
The second perspective is that of an “inter” ambiguity related to one firm’s inability to
decipher the inputs and factors that another firm utilizes to generate a specific outcome,
defined as the competencies which create a competitive advantage (Wilcox-King and
Zeithaml, 2001; Reed and DeFillipi, 1990; Lippman and Rumelt, 1982). It should also be
noted that with the exception of the work of Wilcox-King and Zeithaml (2001) and
Szulanski (1996), causal ambiguity has either been studied through the use of secondary
data or developed using theoretical arguments.
In this dissertation, causal ambiguity was not found to be a significant variable in
knowledge transfer difficulty. Since my measurement of the variable was based upon a
new formative construct that incorporates both the inter and intra aspects of the concept,
it could be argued that the lack of significance comes from the development of the
formative construct. However, the two components of the construct were relatively well
established and when each was tested as a separate independent variable, neither was
found to have a significant contribution to knowledge transfer difficulty. This naturally
raises the question – why? In previous studies, causal ambiguity had either been studied
from an intra-firm perspective or in a dyadic context. This study looked at causal
ambiguity in an inter-organizational context. At least within the context studied, it
94
appears that the specific uncertainty related to the causal factors and inputs into a process
to create an outcome, may be less relevant to knowledge transfer than previously thought.
When the level of causal ambiguity was compared between the two networks, there was
no difference found between the networks for the “external” or competitive advantage
aspect of causal ambiguity. This was unexpected. In previously studied settings, the
alliance members (or intra-organizational departments) were “voluntarily” interacting.
As a result, their barriers, or partner protectiveness policies (Simonin, 1999) were
reduced or eliminated, resulting in a decreased need to create competitive advantages. In
the co-opetive network it was expected that these barriers would be “up” and a perceived
need to create competitive advantage would exist, in large part because of the presence of
competition. However, this would not have been expected in the franchise network,
which is characterized by limited competition. A follow up interview with a regional
Vice President at SunTrust (Appendix I) provided insight into this finding. Specifically,
it was determined that although SunTrust exhibits all of the characteristics of a franchise
network, the regional management purposefully creates an “artificial” level of
competition among the branches, as a strategy to improve overall branch network
performance. This artificial competition results, in part through the creation of “zero sum
games” among the branch managers – creating winners at the expense of losers in sales
contests and promotions. The result is that this franchise network may have behaved
more like a co-opetive network type along the competition continuum, providing some
explanation as to why this aspect of causal ambiguity was not found to differ between the
two networks.
95
5.3 Outcome Ambiguity
Unlike absorptive capacity and causal ambiguity, outcome ambiguity represents a new
factor of knowledge transfer difficulty – and a contribution of this dissertation. The
development of the construct closes the gap in the existing research related to causal
ambiguity and perceived environmental uncertainty as it relates to inter-organizational
knowledge transfer.
Mosakowski (1997) effectively makes the point that the concept of causal ambiguity does
not address the ambiguity related to the actions of others (an outcome). Milliken’s work
on perceived environmental uncertainty (1987), although enlightening and insightful,
does not provide (nor was intended to provide) a sufficient amount of specificity in
defining the perceived uncertainty related to inter-organizational interactions. Her
typology focused on the broader macro-environment without attention to how different
constituents within that environment affect uncertainty or how uncertainty affects
knowledge transfer. Outcome ambiguity was developed to fill the gap related to interorganizational knowledge transfer between causal ambiguity and perceived
environmental uncertainty.
Outcome ambiguity was found to be a significant factor of knowledge transfer difficulty
in the franchise network but not in the co-opetive network. This result may indicate that
when the network in question is a franchise type, the source-recipient relationship is
highly significant to the transfer of knowledge, and is not taken for granted.
Specifically, recall from the discussion of outcome ambiguity in Section 2 that based
upon the work of Simonin (1999) and Szulanski (1996), there were two components to
96
the source-recipient relationship – partner protectiveness and trust. In a co-opetive
network, where the members have accepted competition as a presence within the network
environment, the issues of partner protectiveness and trust may be less relevant for
knowledge transfer – they may be self-censoring the types of knowledge shared.
However, in a franchise environment where competition is not expected to exist, and
therefore no censoring occurs, these issues become more relevant, especially when, as
was true in this case, some competition, and therefore some self-censoring, within a
franchise network was found to be present. Therefore, the findings of Szulanski and
Simonin regarding the role of partner protectiveness and trust may have changed if they
had studied different inter-organizational settings. The finding that outcome ambiguity
was found to be higher for the co-opetive network relative to the franchise network
provides a complement to Milliken’s work (1987) in perceived environmental uncertainty
through the establishment of a specific type of state uncertainty that firms must consider
when evaluating the outcomes associated with knowledge transfer – the actions of their
network partners.
97
6.0 CONCLUSIONS
6.1 Study Limitations
The current study examined the states of the factors determined to influence interorganizational knowledge transfer difficulty and inter-organizational network types.
Given the depth and breadth of the specified model and supporting theory, a single
empirical study, almost by definition will have to accept limitations for manageability.
First, in Part I of the model, the unit of analysis was the organizational entities. The
models were evaluated using an N of 171, a sufficient sample to draw statistical
conclusions. In Part II of the model, the unit of analysis was the network. Clearly size
and complexity issues exist when attempting to study multiple networks. In this
dissertation, two networks were studied. Although an N of 2 presents some limitations
regarding generalizations, useful conclusions can be drawn as is the case in “focal firm”
studies.
A second point of limitation involves the industry of focus. The advantage to restricting
this analysis to a single industry – in this case the Financial Services Industry – is the fact
that most variation due to industry specifics is mitigated. For example, a comparison of
one network type in manufacturing to a second network type in consumer retail, could be
criticized for variation being attributable to specific industry characteristics. However,
the disadvantage to restricting this analysis to a single industry is the potential lack of
generalizability to networks in other industries.
98
A third limitation of this study involves the control group. Comparing networks to a
group of similar organizations, which are part of the same industry, but not part of a
network, is a challenging task. The “independent” credit unions, although part of CUNA,
indicated through the survey that did not perceive themselves to be part of the network
and therefore appeared to represent a control group. However, several of the ANOVA
results indicated no significant differences between the CUNA “integrated” credit unions
and the “independent” credit unions, where differences between the SunTrust network
and the independent credit unions were detected, suggesting that the “independent” credit
unions may have been too similar to the “integrated” credit unions to represent a distinct
and separate control group.
6.2 Study Implications
6.2.1 Implications for Practitioners
For managers currently operating within a network of entities, as more firms increasingly
do, this dissertation has several implications for practice, including both descriptions of
phenomena as well as prescriptions for improvement.
In this dissertation, the factor of absorptive capacity was developed as a formative
construct, comprised of commonalities of language, process, problem solving and
common knowledge. This factor was found to be a highly significant contributor to
knowledge transfer difficulty, regardless of the network in question. As a result,
managers should look to invest in creating these commonalities to decrease the
difficulties associated with knowledge transfer difficulty. As the example of the MARS
99
Climate Orbiter example cited in the beginning of this dissertation highlights, failure to
invest in these commonalities can lead to difficulties associated with knowledge transfer.
The factor of outcome ambiguity was developed as a measure of the knowledge sourcerecipient relationship. This factor was found to be a highly significant for the franchise
network and insignificant for the co-opetive network. The explanation put forth for this
result was based on the acceptance of competition within the two networks and the extent
to which self-censoring occurs. In the franchise network, since competition is not
expected to exist, censoring of what is shared is not expected to exist. However, because
censoring does not take place, the concepts of partner protectiveness and trust become
more important to the transfer of knowledge, especially when some competition is
present, as was evidenced in this study. Therefore, for managers operating within a
franchise network, developing trust and decreasing the presence of “partner
protectiveness” among entities is critical for knowledge transfer. This might be achieved
through the centralized authority increasing punishment for opportunistic or “nonpartner”-oriented behaviors.
The general findings of the states of the respective factors (and their respective subfactors) associated with each network type provides managers operating both within and
outside of a network with information regarding strengths and weaknesses of the two
network types. For example, the co-opetive network was found to be associated with a
higher state of causal ambiguity related to causal factors and a similar state of causal
ambiguity related to competitive advantage relative to the franchise network. This
100
dissertation effectively provides managers with an outline of the expected states of each
factor and sub-factor for consideration.
6.2 Conclusions and Implications for Further Research
Overall, some differences were found to exist for each factor of knowledge transfer
difficulty between the two network types. This finding is an extension of previous
research in inter-organizational learning, where networks and multi-organizational
alliances have been found to be superior to independent operations for knowledge
transfer, but how these networks and alliances vary in this capacity has not previously
been explored. Previous research has examined knowledge transfer and its contributing
factors in intra-organizational or dyadic contexts. However, it should be noted that a
multi-organizational network is very different from a dyad. Unlike a dyadic relationship,
networks can take on a life of their own that supersedes the presence of any individual
member – if one bank pulled out of VISA, the network would continue. Simmel (1950),
who studied social relationships, found that social triads (and relationships involving
more than three entities) had fundamentally different characteristics than did dyads. First
there is no majority in a dyadic relationship – there is no peer pressure to conform. In
any group of 3 or more, an individual organization can be pressured by the others to
suppress their individual interests for the interests of the larger group. Second, individual
organizations have more bargaining power in a dyad. This is not only true because of
percentages, but if one member withdraws from a dyad the dyad disappears – this is not
true in a network. Finally, third parties represent alternative and moderating perspectives
when disagreements arise.
101
As a result of these differences, multi-organizational networks are more complex to study
than intra-organizational or dyadic settings. Therefore the initial confirmation of the
differences of factors of knowledge transfer difficulty in these more complex
environments is in itself a significant contribution of this dissertation.
The foundational theories of this dissertation – the KBV and TCE – both provide insight
into how firms organize within their own boundaries. The KBV provides for different
organizational approaches to maximize efficiencies related to knowledge utilization and
transfer, while the TCE addresses the minimization of costs related to the transactions
necessary for business processes, in part through organizational structure. Organization
within a firm’s boundaries would logically affect factors of knowledge transfer, as
examined in previous research (e.g., Galbraith and Merrill, 1991; Adler, 2001; Van den
Bosch, 1991; Volberda et al., 1998). However, the finding that factors of knowledge
transfer vary with inter-organizational network type now provides initial evidence that
how a firm organizes outside of its own boundaries influences these same factors.
In addition to this general finding, which has broad implications for the application of
TCE and KBV in inter-organizational knowledge transfer, this dissertation provides
specific contributions to the KBV.
First, the KBV seeks to explain the existence of the firm in terms of its superior
efficiency in the transference of knowledge relative to market-based interactions. This
dissertation seeks to explain the existence of the network in terms of its superior
efficiency of the transference of knowledge relative to both the firm and market-based
interactions. As stated earlier, firms experience bounded rationality regarding knowledge
102
– no firm can know all that is knowable economically. As was evidenced through the
work of Doz, Prahalad and Hamel (1998), Madhavan (1998), Gulati and Gargiulo (1999)
and others, firms join networks in part because networks represent a significant conduit
for incremental knowledge. Argote (1999), Darr et al. (1995), Powell et al. (1996) have
demonstrated that networks are superior to independent firms for the purposes of
knowledge transfer.
This dissertation takes this discussion one step further and provides empirical evidence
for the position that different network types experience knowledge transfer differently.
Second, the KBV identifies four mechanisms for transferring knowledge within the
boundaries of the firm – rules and directives, sequencing, routines and through group
problem solving. These mechanisms were empirically found to exist in some form in the
two networks studied, and logical arguments were developed for their existence within
the two networks that were not studied empirically, but discussed earlier in this
dissertation. For example, the franchise network type is considered to have a centralized
governance structure and transfers knowledge through rules and directives, and this was
supported through results from the SunTrust branch network. Similarly, the co-opetive
network was found to transfer knowledge similarly through rules and directives. The
value chain network type was proposed to have a weak formal governance structure and
was discussed to transfer knowledge through sequencing. The innovation network type
was defined as having a weak governance structure and, given the focus on creation of
new knowledge, was argued to transfer primarily through the communication-intensive
mechanism of “group problem solving”.
103
These mechanisms of transfer are argued to exist differently across the different network
types primarily because of the network governance structures. As discussed in Section
2.3 and illustrated in Figure 4, the different network types are also defined based upon the
operational scope and intensity of competition. Based upon these three characteristics,
the different states of the three factors of knowledge transfer difficulty were argued, and
tested empirically for two network types. This discussion extends the position of the
KBV regarding the mechanisms of knowledge transfer.
The KBV also makes a case for the importance of common knowledge. Specifically,
Grant (1996) defines common knowledge in a very general sense that includes common
language. Using the definition of absorptive capacity, this dissertation makes the
argument for differing relative contributions of common knowledge, common language
and common process and common problem solving.
These findings also raise an important question – could the differences found here to exist
in knowledge transfer and its associated factors between the two network types have been
attributed to the influence of the specific individual network characteristics? For
example, the franchise network was identified as an inter-organizational network type
characterized by a strong, centralized hierarchy, limited competition among members and
limited operational scope. The aggregated “bundle” of these characteristics was then
labeled a “franchise network”. Where differences were found to exist, were they a
function of the network type – defined by the bundle of characteristics? Or were they a
function of one particular characteristic of the network, such as competition?
104
In this dissertation, the extent to which each network studied existed along the three
continuums of the three characteristics identified in Figure 4 was a result of a relatively
subjective judgment, albeit grounded in practical examples. To the extent that a
network’s presence along each of these continuums could be quantified, the question of
whether differences are attributed to network type or network characteristic could be
explored more fully.
A second research implication made by this study is the development of and initial
empirical evidence for the existence of the outcome ambiguity factor of knowledge
transfer difficulty. Based upon the research of Szulanski (1996), Mosakowski (1997) and
others, the factor of outcome ambiguity was operationalized and the measurement items
were validated through the PLS measurement model. This construct was then tested and
found to be a significant explanatory variable of knowledge transfer difficulty in the
franchise network – the most frequently studied network type. In addition, the state of
outcome ambiguity was found to vary with network type – highly significant for the
franchise network and not significant for the co-opetive network. These findings,
separately and in combination, provide researchers with a new path of emphasis in
developing a deeper and more refined understanding of inter-organizational knowledge
transfer.
The third implication of this dissertation for researchers is made through the links in Part
II of the model in Figure 5 – the links between network type and the factors of knowledge
transfer difficulty. As stated in the beginning of this study, several studies have
demonstrated, primarily through case studies, that networks are superior to non-aligned
105
entities for the purposes of knowledge transfer (Argote, 1999; Darr et al., 1995; Powell et
al., 1996). However, no study appears to make the assertion that different types of
networks may experience knowledge transfer differently, and its associated factors
differently. This theoretical link may have implications for the Knowledge Based View
of the Firm.
Specifically, Grant (1997) explains the role of strategic alliances in the KBV, from the
perspective of resource (knowledge) acquisition and utilization efficiencies within the
boundaries of the firm versus outside of the firm – analogous to the foundations of TCE.
However, the KBV is currently void of any specificity regarding the general forms that
these alliances assume and how these forms then affect knowledge transfer. This study
may provide a basis to frame this specificity, through the differences that were found to
exist between the two network types studied.
The fourth research implication of this study comes from the overall results of the
measurement model described in Section 4.2.2. Most of the empirical work in the area of
knowledge management has been done utilizing secondary data or case studies…in large
part because of the difficulty of measuring the variables of interest. The strength of the
measurement model in this dissertation provides a contribution for future researchers
interested in studying the factors of knowledge transfer by providing validated scales.
In addition to the implications for the KBV, this dissertation provides for new research
opportunities. One opportunity is the further development, refinement and testing of the
outcome ambiguity factor. In this study, the initial factor was developed using two
formative sub-constructs – provenness of knowledge and the source-recipient
106
relationship. The provenness of knowledge was eventually discarded as a contributing
sub-construct because it did not demonstrate divergent validity (the sub –construct was
found to load on absorptive capacity and causal ambiguity). Given the theoretical logic
of its existence, further research and refinement may find that this sub-construct should
be retained.
The two network types that were identified in this dissertation, and for which hypotheses
were created, but were not tested empirically represent a third and obvious opportunity
for further research. The value chain and innovation networks have been studied
extensively, including research related to knowledge transfer difficulty. However, as
with the franchise and co-opetive network types, no research to date has looked at how
these network types experience knowledge transfer differently.
107
Table 1: Summarization of Knowledge Transfer Difficulty Studies
Study
Szulanski (1996)
von Hippel (1994)
Tsai (2001)
Lapre (2001)
Hansen (2002)
Birkinshaw (2002)
Darr et al. (1995)
Definition of Knowledge Transfer Difficulty
Unit of Analysis
"The transfer of best practice inside the firm…connotates
the firm's replication of an internal practice that is
performed in a superior way in some part of the
organization and is deemed superior to internal alternate
practices and known alternatives outside of the
firm...Cost could be a poor descriptor of
difficulty...Eventfulness or the need for ad hoc solutions
could be translated into an outcome-based descriptor of
stickiness..."
“…when information is costly to acquire, transfer and
use…it is ‘sticky’”.
No explicit definition provided…however, references
made to Szulanski (1996) and von Hippel (1994).
Firm (intraorganizational)
through surveys of
individuals
Knowledge Transfer Difficulty is defined as a
generalized "Learning Rate"; the logarithm of unit costs,
which decreases with the log of cumulative number of
units produced at a uniform rate.
Knowledge transfer is a function of (in)direct relations or
path lengths….including the "distance" or number of
intermediaries between the knowledge recipient and the
source…and the number of direct connections (although
not explicitly stated, a measurement of network
centrality)
Knowledge transfer is evaluated in terms of observability
and system embeddedness.
Firm (learning
curves across
different plants
studied)
Organizational unit
or project team
within a multi-unit
corporation.
"By (transfer of learning) we mean whether organizations
learn from the experience of other organizations".
The Franchise
Unit…through
learning curve
analysis
108
Innovation within
the Firm
Organizational unit,
within a multi-unit
corporation
Organization (focus
on org. structure)
Table 2: Summarization of Causal Ambiguity Studies
Study
Szulanski (1996)
Wilcox-King (2001)
Barney (1999)
Mosakowski (1997)
Definition of Causal Ambiguity
Unit of
Analysis
"When the precise reasons for success or failure
cannot be determined even ex post, causal
ambiguity is present and it is impossible to produce
an unambiguous list of the factors of
production…embodied in highly tacit human skills,
causal ambiguity results from imperfectly
understood features of the new context of
application”.
"Causal Ambiguity, which is ambiguity about the
link between firm resources and sustained
competitive advantage, protects resources from
competitive imitation". The author studies two
forms – Linkage Ambiguity is ambiguity among
decision makers about the link between competency
and competitive advantage, Characteristic
Ambiguity focuses on the characteristics of
competency that can be simultaneous sources of
advantage and ambiguity...it focuses on the resource
itself (e.g., tacitness)
Firm (intraorganizational)
through
surveys of
individuals
"…Sometimes it is not clear which actions a firm
should take to create a particular capability…Causal
Ambiguity about how to create capabilities exists
whenever there are multiple competing hypotheses
about how to create those capabilities and when
these hypotheses cannot be tested. These conditions
are likely when the sources of a firm's capabilities
are taken for granted, unspoken and tacit...”
…"the number of distributions that are not ruled out
by one's knowledge of the situation, with each
performance distribution resulting from both a
unique, stochastic causal structure and a unique
specification of inputs…I do not focus on a decision
maker's ability to predict a specific future
performance outcome given the assumption of a
causal structure with stochastic outcomes..."
N/A
109
Firm (intraorganizational)
through
surveys of
individuals
Organizational
Unit
Table 3: Summarization of Measurement Items from Previous Studies
Construct
Study
Relevant Survey Questions
Knowledge
Transfer
(Difficulty)
Szulanski (1996)
1. Stickiness or KTD was measured with a set of 8 items corresponding to the
so-called technical success indicators of a project - on time, on budget and
satisfaction. Deviation in time was measured from (i) the start of the project,
(ii) the first day the practice became operational and (iii) achievement of
satisfactory performance.
2. Two items measured departure of actual cost from expected cost for both the
source and the recipient
3. Three items measured recipient's satisfaction. One item measured adjustment
in the recipient's expectation after gaining experience with the practice in
question.
4. Two items measured whether the recipient was satisfied with the quality of
the practice in question and with the quality of the knowledge transfer.
Hansen (2002)
The KT variable includes the transfer of software and hardware that were
defined as follows in the survey: Software included firmware and flow and
structure charts; hardware included electronic, electrical, and mechanical parts.
Project managers were asked: "Of all the software (hardware) that was needed
in the product, what were the following breakdowns?" They were given 100
points to allocate across 6 categories...(i) already developed ware from own
division (ii) New ware developed in own division, (iii) already developed ware
from other divisions, (iv) new ware developed in other divisions, (v) already
developed ware from outside of the company, (vi) new ware developed outside
the company.
Dyer (1997)
1. To what extent has x provided technical, engineering, or other assistance in
the past which has allowed you to make changes in your manufacturing process
which have allowed you to lower manufacturing costs?
2. To what extent has x provided assistance to help you reduce defects and
increase the overall reliability and quality of the products you sell to x?
3. To what extent has x provided assistance in developing a "Just-In-Time"
inventory management system designed to lower inventory costs and/or make
delivery more efficient?
Absorptive
Capacity
Birkinshaw (2002)
"In the last 5 years, has your unit actively transferred locally developed
technological know-how to other manufacturing or RandD units in your
company?"
Szulanski (1996)
1. Members of the recipient organization have a common language to deal with
the practice in question
2. The recipient had a vision of what it was trying to achieve through the
knowledge transfer
3. The recipient had information on the state-of-the-art of the practice in
question
4. The recipient had a clear division of roles and responsibilities to implement
the practice in question
5. The recipient had the necessary skills to implement the practice in question
6. The recipient had the technical competence to absorb the practice in question
7. The recipient had the managerial competence to absorb the practice in
110
question
8. It is well known who can best exploit new information about the practice in
question within the recipient organization
9. It is well known who can help solve problems associated with the practice in
question
Causal
Ambiguity
Szulanski (1996)
1. The limits of the practice in question are fully specified
2. With the practice in question, we know why a given action results in a given
outcome.
3. When a problem surfaced with the practice in question, the precise reasons
for failure could not be articulated even after the event.
4. There is a precise list of the skills, resources, and pre-requisites necessary for
successfully performing the practice in question.
5. It is well known how the components of that list interact to produce the
output from the practice in question
6. Operating procedures for the practice in question are available.
7. Existing work manuals and operating procedures describe precisely what
people working in the practice actually do.
Outcome
Ambiguity
Szulanski (1996)
1. We had solid proof that the practice in question was really helpful
2. The practice in question contributes significantly to the competitive
advantage of the company.
3. For the success of the company, the practice is: Critical, Very important,
Fairly Important, Fairly Unimportant, Not Important at all.
Dyer (1997)
1. To what extent do you trust that confidential/proprietary information shared
with x will be kept strictly confidential?
2. To what extent do you trust x personnel to deal with you fairly?
3. If given the chance, to what extent will this customer take unfair advantage
of your business unit?
4. To what extent has x developed a reputation for fairness and trustworthiness
among the supplier community?
111
Table 4: Interviewees Contact Information
1. Washington Dender, SVP Strategy. SunTrust Bank. 303 Peachtree Street, eighth
floor, Atlanta, GA 30302. (404) 575-2476.
2. Will Hileman, SVP Retail Banking. SunTrust Bank. 303 Peachtree Street, Box
4418, Atlanta, GA 30302. (404) 230-5022.
3. Mark Condon, SVP Association Services. CUNA. 5710 Mineral Point Road,
Madison WI, 608-231-4078.
4. Eric Anderson – Zych, CEO, Genessee Co-Op Federal Credit Union. 741South
Avenue, Rochester, NY, 14620. 585-461-2230 ext. 18.
5. Cookie Yoder, CEO, Pittsburgh Federal Credit Union. 412-381-6363.
6. Kevin Bartlett, knowledge manager for global retail practice at Accenture.
312693 0563.
7. Kirk Watkins, Vice President for Branch Operations at SunTrust Bank. 303
Peachtree Street, eighth floor, Atlanta, GA 30302. (404) 588-7156. Discussions
took place with Kirk throughout the Field Study.
112
Table 5: Changes Made to Survey Instrument
Between Pilot Survey and Field Survey
Pilot Survey
Field Survey
Q1. CUNA (or another
credit union member of
CUNA) has provided
knowledge of
operational processes
(including marketing,
risk or compliance, etc)
to our credit union that
has enabled us to
improve our
performance.
Q2. Our credit union
has provided
knowledge of
operational processes
(including marketing,
risk or compliance, etc)
to CUNA or to another
credit union member of
CUNA that enabled
performance
improvement.
Q3. Our credit union
has received
knowledge of an
operational process
(including marketing,
risk or compliance, etc)
that was developed at
another credit union
member of CUNA.
Q4. Other credit union
members within CUNA
have received
knowledge of an
operational process
(including marketing,
risk or compliance, etc)
that was developed at
our credit union.
Q5. Our credit union
has
trialed/implemented
concepts that came
from CUNA or from
another credit union
member within CUNA.
Q1. CUNA (or another
credit union member of
CUNA) has provided
knowledge of processes
(including operations,
marketing, risk or
compliance, etc) to our
credit union that has
enabled us to improve
our performance.
Q2. Our credit union
has provided
knowledge of processes
(including operations,
marketing, risk or
compliance, etc) to
CUNA or to another
credit union member of
CUNA that enabled
performance
improvement.
Q3. Our credit union
has received
knowledge of a process
(including operations,
marketing, risk or
compliance, etc) that
was developed at
another credit union
member of CUNA.
Q4. Other credit union
members within CUNA
have received
knowledge of an
process (including
operations, marketing,
risk or compliance, etc)
that was developed at
our credit union.
Q5. Our credit union
has trialed concepts
that came from CUNA
or from another credit
union member within
CUNA.
113
Construct Effected
Source of Change*
Knowledge Transfer
Difficulty
Interview 2C – Cookie
Yoder indicated that
the term “operational
processes” may be too
restrictive.
Knowledge Transfer
Difficulty
Interview 2C – Cookie
Yoder indicated that
the term “operational
processes” may be too
restrictive.
Knowledge Transfer
Difficulty
Interview 2C – Cookie
Yoder indicated that
the term “operational
processes” may be too
restrictive.
Knowledge Transfer
Difficulty
Interview 2C – Cookie
Yoder indicated that
the term “operational
processes” may be too
restrictive.
Knowledge Transfer
Difficulty
Interview 3S – Kirk
Watkins stated that in
some cases a concept
could be trailed, but
not
implemented…even
though the knowledge
“transferred”.
N/A
N/A
Q8. When we place a
request for information
to CUNA, the recipient
of our request generally
“gets it right” the first
time.
Q22. Our credit union
understands how other
credit union members
of CUNA increase their
membership.
Q23. Our credit union
understands how other
credit union members
of CUNA retain good
personnel.
Q24. Our credit union
understands how other
credit union members
of CUNA achieve high
CAML ratings.
Q27. Our credit union
employees understand
what differentiates our
credit union from other
credit unions.
N/A
Q6. Our credit union
has implemented
concepts that came
from CUNA or from
another credit union
member within CUNA.
Knowledge Transfer
Difficulty
Q7. Our credit union
actively seeks out other
credit union members
of CUNA for
collaboration.
Q8. Other credit union
members of CUNA
look to our credit union
for collaboration.
Q11. When we place a
request to CUNA, the
recipient who will
respond to our request,
generally “gets it right”
the first time.
Q25. Employees of our
credit union could
create a specific list of
the factors that do or do
not contribute to an
increase in
membership.
Q26. Employees of our
credit union could
create a specific list of
the factors that do or do
not contribute to the
retention of good
personnel.
Q27. Employees of our
credit union could
create a specific list of
the factors that do or do
not contribute to high
CAML ratings.
Q30. Employees of our
credit union could
create a list of factors
that differentiate our
credit union from other
credit unions
Q31. Employees of our
credit union could
create a list of the
factors that do or do
not contribute to the
financial success of
Knowledge Transfer
Difficulty
114
Interview 3S – Kirk
Watkins stated that in
some cases a concept
could be trialed, but
not
implemented…even
though the knowledge
“transferred”.
Simonin (1999)
Knowledge Transfer
Difficulty
Simonin (1999)
Absorptive Capacity
Interview 5C – Mark
Condon of CUNA
Causal Ambiguity
Change was made by
the author to increase
specificity and clarity.
Causal Ambiguity
Change was made by
the author to increase
specificity and clarity.
Causal Ambiguity
Change was made by
the author to increase
specificity and clarity.
Causal Ambiguity
Change was made by
the author to increase
specificity and clarity.
Causal Ambiguity
Change was made by
the author to increase
specificity and clarity.
other credit unions.
Q37. When credit
Outcome Ambiguity
Interview 3C –
union members of
comments provided by
CUNA interact, there is
credit union CEO Eric
an understanding that
Zych-Anderson.
the interaction will be
based in fairness and
trust.
N/A
Q43. We believe that
Outcome Ambiugity
Interview 3S – Kirk
there are credit union
Watkins of SunTrust
members of CUNA
who would refuse to
share knowledge of a
process with other
credit union members
of CUNA
Note – although all examples used here come from the CUNA surveys, similar changes were made to
the SunTrust surveys.
N/A
*See Appendix 1
115
Table 6: Pilot Study Scale Reliabilities
Item
Construct
(Sub-Construct)
Std Loading
AVE
Fornell and
Larcker Rho
Cronbach Alpha
Q1
Q2
Q3
Q4
Q5
Q6
Knowledge Transfer
Knowledge Transfer
Knowledge Transfer
Knowledge Transfer
Knowledge Transfer
Absorptive Capacity
(Common Language)
Absorptive Capacity
(Common Language)
Absorptive Capacity
(Common Language)
Absorptive Capacity
(Common Language)
Absorptive Capacity
(Common Language)
Absorptive Capacity
(Common Language)
Absorptive Capacity
(Common Language)
Absorptive Capacity
(Common Process)
Absorptive Capacity
(Common Process)
Absorptive Capacity
(Common Process)
Absorptive Capacity
(Problem Solving)
Absorptive Capacity
(Problem Solving)
Absorptive Capacity
(Problem Solving)
Absorptive Capacity
(Common Knowledge)
Absorptive Capacity
(Common Knowledge)
Absorptive Capacity
(Common Knowledge)
Causal Ambiguity
(Causal Factors)
Causal Ambiguity
(Causal Factors)
Causal Ambiguity
(Causal Factors)
Causal Ambiguity
(Causal Factors)
0.622
0.841
0.785
0.684
0.858
0.583
0.873
0.825
0.256
0.578
0.729
0.629
0.832
0.830
0.824
0.933
0.919
0.560
0.753
0.467
0.487
0.768
0.739
Q7
Q8
Q9
Q10
Q11
Q12
Q13
Q14
Q15
Q16
Q17
Q18
Q19
Q20
Q21
Q22
Q23
Q24
Q25
0.848
0.149
0.927
0.065
0.089
0.352
0.241
0.971
0.707
0.668
0.915
0.842
0.962
0.939
0.872
0.197
0.938
0.858
0.437
0.374
116
Item
Construct
(Sub-Construct)
Q26
Causal Ambiguity
(Competitive Advantage)
Causal Ambiguity
(Competitive Advantage)
Outcome Ambiguity
(Knowledge Provenness)
Outcome Ambiguity
(Knowledge Provenness)
Outcome Ambiguity
(Knowledge Provenness)
Outcome Ambiguity
(Source/Recipient Relations)
Outcome Ambiguity
(Source/Recipient Relations)
Outcome Ambiguity
(Source/Recipient Relations)
Outcome Ambiguity
(Source/Recipient Relations)
Outcome Ambiguity
(Source/Recipient Relations)
Outcome Ambiguity
(Source/Recipient Relations)
Outcome Ambiguity
(Source/Recipient Relations)
Q27
Q28
Q29
Q30
Q31
Q32
Q33
Q34
Q35
Q36
Q37
Std Loading
AVE
Fornell and
Larcker Rho
Cronbach Alpha
0.461
0.496
0.34
0.718
0.865
0.804
0.229
0.561
0.714
.071
.958
0.884
0.931
0.711
0.019
0.054
0.549
0.063
0.762
0.488
0.690
117
Table 7a: Descriptive Statistics of Field Study Respondents
(SunTrust Branch Managers N=70)
Percentage
44.93
Frequency
31
$5M - $20M
37.68
26
$20M - $50M
10.14
7
Type of Branch
Over $50M
In Store (in a Publix)
7.25
45.71
5
32
Age of Branch
Retail Outlet
Less than 1 Year
54.29
5.80
38
4
1-3 Years
13.04
9
3-7 Years
26.09
18
Over 7 Years
Less than 1 Year
55.07
23.19
38
16
1-3 Years
30.43
21
3-7 Years
34.78
24
Over 7 Years
11.59
8
Size of Branch
Branch Manager
Experience*
Category
Under $5 M
*Question not included in CUNA surveys as a request from CUNA Management
118
Table 7b: Descriptive Statistics of Field Study Respondents
(“Integrated” Credit Union Presidents/CEOs N=68)
Size of Credit
Union
Membership in
CUNA
Category of
Membership
Category
Under $10 M
Percentage
20.59
Frequency
14
$10M - $50M
76.47
52
$50M - $100M
2.94
2
Over $100M
Less than 1 Year
0
0
0
0
1-5 Years
0
0
6-10 Years
1
1.47
Over 11 Years
Low Income
67
1.47
98.53
1
Company/Association
Affiliated
4.41
3
Community
Chartered
94.12
64
119
Table 7c: Descriptive Statistics of Field Study Respondents
(“Independent” Credit Union Presidents/CEOs N=33)
Size of Credit
Union
Membership in
CUNA
Category of
Membership
Category
Under $10 M
Percentage
31.25
Frequency
10
$10M - $50M
65.63
21
$50M - $100M
3.13
1
Over $100M
Less than 1 Year
0
0
0
0
1-5 Years
0
0
6-10 Years
0
0
33
3.13
100.00
1
0
0
96.88
31
Over 11 Years
Low Income
Company/Association
Affiliated
Community
Chartered
120
Table 8: Field Study Scale Reliabilities
Construct
(Sub Construct)
All Respondents
<Retained Items>
(N=171)
Co-opetive Network
(CUNA)
Franchise Network
(SunTrust)
(N=68)
(N=70)
AVE
FandL
Rho
Alpha
AVE
FandL
Rho
Alpha
AVE
FandL
Rho
Alpha
Knowledge Transfer
0.554
0.878
.838
.524
0.868
0.827
0.677
0.890
0.849
<5>
Absorptive Capacity
(Common Language)
0.861
0.925
0.851
0.823
0.902
0.827
0.880
0.936
0.873
<3>
Absorptive Capacity
(Common Process)
0.741
0.896
0.829
0.739
0.894
0.842
0.729
0.888
0.810
<3>
Absorptive Capacity
(Problem Solving)
0.829
0.936
0.898
0.863
0.950
0.923
0.772
0.910
0.862
0.761
0.858
0.792
0.821
0.873
0.863
0.874
0.887
0.815
<3>
Causal Ambiguity
(Causal Factors)
0.611
0.825
0.713
0.496
.735
.701
.651
.847
.700
<3>
Causal Ambiguity
(Comp. Advantage)
.756
.851
.768
.509
.561
.716
.780
.863
.752
0.743
0.896
0.807
0.750
.900
.796
.791
.920
.874
0.625
0.847
0.735
0.523
.658
.760
.745
.636
.749
<3>
Absorptive Capacity
(Common
Knowledge)
<3>
Outcome Ambiguity
(Provenness of
Knowledge)
<3>
Outcome Ambiguity
(Source/Recipient)
<3>
121
Table 9: Intercorrelations for Field Study, with SQRT of AVEs in Diagonal
Know
Trans
fer
Know
Transfer
ACCommon
Language
ACCommon
Process
ACCommon
Knowledge
AC-Problem
Solving
CA-Comp
Adv.
CACommon
Factors
OA-Proven
Knowledge
OA-SourceRecipient
ACCommon
Language
ACCommon
Process
ACCommon
Knowledge
ACProblem
Solving
CA –
Comp
Advant.
CACommon
Factors
OA-Proven
Knowledge
OASourceRecipient
.744
.396
.928
.323
.138
.861
.359
.509
.440
.873
.312
.108
.578
.474
.910
.248
.157
.371
.340
.333
.869
.204
.104
.463
.382
.444
.613
.781
.420
.505
.429
.458
.368
.438
.342
.861
.175
.017
.136
.193
.115
.050
.171
.051
122
.807
Table 10a: Item –to – Sub Construct Correlations
Question
Knowledge
Transfer
Difficulty
Q1
Q2
Q3
Q4
Q6
Q7
Q9
Q10
Q16
Q17
Q18
Q19
Q20
Q21
Q22
Q23
Q24
Q25
Q26
Q28
Q29
Q30
Q31
Q32
Q33
Q34
Q41
Q42
Q43
.682
.784
.819
.707
.749
.732
ACCL
ACCP
ACPS
ACCK
CACF
CACA
OAPK
OASR
.965
.889
.861
.867
.859
.925
.929
.878
.911
.930
.648
.752
.724
.836
.834
.826
.811
.872
.884
.789
.659
.658
.958
123
Table 10b: Subconstruct-to-Construct Correlation Matrix
ACCL
AC-CL
AC-CP
AC-PS
AC-CK
Absorptive
Capacity
CA-CF
CA-CA
Causal
Ambiguity
OA-PK
OA-SR
Outcome
Ambiguity
ACCP
1
.1390
1
.1055 .5753
.4582 .4627
.8687 .5405
AC-PS
1
.4952
.4877
ACCK
Absorptive
Capacity
1
.7513
1
CACF
CACA
Causal
Ambiguity
.1086 .4695 .45586 .4449
.1259 .3979 .3661 .3828
.1314 .4047 .3734 .3898
.3744
.3598
.3635
1
.6387
1
.6585 .9996
1
.4998 .4291
.0204 .1324
.0669 .1504
.6300
.1538
.1997
.3501 .4325
.1692 .0907
.1931 .1205
.4326
.0919
.1215
.3598
.1184
.1427
.4357
.2735
.3038
124
OAPK
OASR
1
.0419
1
.1147 .9973
Outcome
Ambiguity
1
Table 11: Hypothesis Testing and Network Results – Part I of Specified Model
Network
Hypothesis
PLS Path
Coefficient
PLS tstatistic
Supported?
Global
Franchise
PLS R2
.199
H1: Absorptive Capacity will
have a negative relationship with
knowledge transfer difficulty
H2: Causal Ambiguity will have a
positive relationship with
knowledge transfer difficulty
-.417
-4.425
Yes (p<.01)
.035
.313
H3: Outcome Ambiguity will have
a negative relationship with
knowledge transfer difficulty
.041
.209
No –
insignificant
result
No –
insignificant
result
Variable
Absorptive Capacity
-.695
-4.467
Causal Ambiguity
.043
.261
Outcome Ambiguity
.399
1.833
.502
Significant
(p<.01)
Not
Significant
Significant
(p<.01)
CoOpetive
.223
Absorptive Capacity
-.369
-2.157
Causal Ambiguity
.066
.403
Outcome Ambiguity
.018
.068
125
Significant
(p<.01)
Not
Significant
Not
Significant
Table 12: Hypothesis Testing Results – Part II of Specified Model
(Using PLS-Weighted Scores)
Hypothesis
Franchise
Score
H4a – The franchise type will be
associated with a high state of
absorptive capacity
H7a – The co-opetive type will be
associated with a high state of
absorptive capacity
Both forms expected to be higher
than control.
11.45
H4b – The franchise type will be
associated with a low state of
causal ambiguity
H7b – The co-opetive type will be
associated with a high state of
causal ambiguity
Both forms expected to be lower
than control.
3.55
H4c – The franchise type will be
associated with a low state of
outcome ambiguity
H7c – The co-opetive type will be
associated with a high state of
outcome ambiguity
Both forms expected to be lower
than the control.
2.81
H4d – The franchise type will be
associated with a low state of
knowledge transfer difficulty
H7d – The co-opetive type will be
associated with a high state of
knowledge transfer difficulty
Both forms expected to be lower
than the control.
4.04
Coopetive
Score
Control
Score
Supported?*
Significance
+
12.31
+
11.33
Y/N
Co-opetive
higher than
control at p<.05
N
3.95
N
4.48
3.56
4.39
4.50
4.40
Y/N
Y
Franchise lower
than control at
p<.05
p<.05
Y
p<.05
Y/Y
Y
Franchise lower
than control at
p<.01/Coopetive lower
than control at
p<.10.
p<.10
Y
p<.10
N
*Significance reported for hypotheses represents whether there is a statistically different
score between populations. P values below .10, indicate that there is. +Not testable
126
Table 13: Regression Model Results by Population
Network
Global
Franchise
CoOpetive
IVs
Stand.
β
tStatistic
VIF
Sig
F
Sig
Adj.
R2
Absorp. Capacity
.333
5.30
1.175
<.0001
14.56
<.0001
.228
Causal Ambiguity
-.115
-1.25
1.153
.2127
Outcome Ambiguity
Absorp. Capacity
-.066
.526
-1.03
7.47
1.027
1.285
.3056
<.0001
23.03
<.0001
.516
Causal Ambiguity
-.052
-.43
1.232
.666
Outcome Ambiguity
Absorp. Capacity
-.175
.289
-1.89
2.33
1.106
1.285
.0643
.024
3.89
.0146
.148
Causal Ambiguity
-.161
-.99
1.284
.328
Outcome Ambiguity
-.087
-.74
1.097
.462
127
Figure 1: Causal Ambiguity
Inputs
Outcomes
X1
X2
X3
.
.
.
Xn
Outcome1
Outcome2
Outcome3
.
.
.
Outcomen
Causal Factors
Factor1
Factor2
Factor3
.
.
.
Factorn
128
Figure 2: Outcome Ambiguity Framework
Known Recipient Actions
Bounded Recipient Action Set: [RA1, RA2, RA3]
Proven Knowledge
Bounded Knowledge Usage Set: [KU1, KU2, KU3]
Unproven Knowledge
Unbounded Knowledge Usage Set: [KU1, KU2, KU3…KU∞]
Type 1
(Low Outcome Ambiguity)
Outcome Set: [O1, O2, O3]
Type 2
(Medium Outcome Ambiguity)
Outcome Set: [O1, O2, O3…O∞]
129
Unknown Recipient Actions
Bounded Recipient Action Set: [RA1, RA2, RA3…RA∞]
Type 3
(Medium Outcome Ambiguity)
Outcome Set: [O1, O2, O3…O∞]
Type 4
(High Outcome Ambiguity)
Outcome Set: [O1, O2, O3…O∞]
Figure 3: Knowledge Transfer Factors
Part I of Specified Model
Absorptive
Capacity
Hypothesis 1:
Causal
Ambiguity
-
Hypothesis 2:
Hypothesis 3:
+
+
Outcome
Ambiguity
Legend:
- = Negative Relationship
+ = Positive Relationship
130
Difficulty of
Knowledge
Transfer
Figure 4: Differentiation Among Network Types
131
Figure 5: Specified Model – Parts I and II
Network Form
Knowledge Transfer
Difficulty Factor
Absorptive
Capacity
Franchise
Innovation
Difficulty of
Knowledge
Transfer
Causal
Ambiguity
Value
Chain
Outcome
Ambiguity
Co-opetive
Part II
Part I
132
Figure 6: Construct Development Methodology
(adapted from Churchill, 1979)
Specify Constructs (content validity)
• Literature Review
• Interviews with Practitioners and Experts
Generate Item Pool (content validity)
• Literature Review
• Existing Scales
• Items suggested by Practitioners and Experts
Purify Measures (construct validity)
• Establish item Correlation
•
Establish item Behavior
133
References
Adler, P. 2001. Market, hierarchy and trust: The knowledge economy and the future of capitalism.
Organization Science, 12(2): 215-234.
Allee, V. 2002. Value networks and evolving business models for the knowledge economy. In
Holsapple, C.W. (Ed), Handbook on Knowledge Management: 605-621 Berlin: Springer Verlag.
Almeida, P., Song, J., and Grant, R. 2002. Are firms superior to alliances and markets? An empirical
test of cross-border knowledge building. Organization Science, 13(2): 147-162.
Argote, L. 1999. Organizational Learning. Creating, Retaining and Transferring Knowledge.
Norwell, MA: Kluwer Academic Publishers.
Babbie, E. The Practice of Social Research. Stamford, CT: Wadsworth/Thompson Learning.
Barney, J. B. 1986. Strategic factor markets: Expectations, luck, and business strategy. Management
Science, 32 (9): 1231-1241.
Barney, J. B. 1991. Firm resources and sustained competitive advantage. Journal of Management,
17(1): 99-120.
Barney, J. B. 1999. How a Firm’s Capabilities Affect Boundary Decisions. Sloan Management
Review. Spring: 137-195.
Baum, J.A.C. and Ingram, P. 1998. Survival-enhancing learning in the Manhattan hotel industry 18981980. Management Science, 44 (7): 996-1016.
Beer, S. 1969. The aborting corporate plan: A cybernetic account of the interface between planning
and action, In Jantsch, E. (Ed) Perspectives of Planning: 397-422. Paris.
Birkinshaw, J., Nobel, R., and Ridderstale, J. 2002. Knowledge as a contingency variable: Do the
characteristics of knowledge predict organization structure? Organization Science, 13(3): 274-290.
Boerner, C. and Macher, J. 2003. Transaction Cost Economics: An Assessment of Empirical
Research in the Social Sciences. Working Paper.
Bollen, K. 1989. A new incremental fit index for general structural equation modeling. Sociological
Research Methods and Research, 17(3): 303-317.
Boynton, A., Zmud, R., Jacobs, G. 1994. The Influence of IT Management Practice on IT Use in
Large Organizations. MIS Quarterly, 18(3): 299-318.
Brandenburger, A.M. and Nalebuff, B.J. 1996. Co-opetition. New York: Doubleday.
Burt, R.S. 1992. Structural Holes: The social structure of competition. Cambridge, MA: Harvard
University Press.
Carlsson, S. A. 2002. Strategic knowledge managing within the context of networks. In Holsapple,
C.W., (Ed) The Handbook on Knowledge Management: 623-650 Berlin: Springer Verlag.
Casti, J. 1995. The theory of networks. In Batten, D., Casti, J., and Thord, R. (Eds) Networks in
action. Communication, Economics and Human Knowledge Berlin: Springer Verlag.
134
Churchill, G. 1979. A Paradigm for developing better measures of marketing constructs. Journal of
Marketing Research, 16(1) 64-74.
Chin, W. 1998. The partial least squares approach to structural equation modeling. In Marcoulides,
G, (Ed) Modern Methods For Business Research: 295-336 Mahwah, NJ: Lawrence Erlbaum
Associates.
Coase, R. 1937. The nature of the firm. Economica, 4(4): 386-405.
Cohen, W. M. and Levinthal, D. A. 1990. Absorptive capacity: A new perspective on learning and
innovation. Administrative Science Quarterly, 35(1): 128-153.
Cook, T. and Campbell, D. 1979. Quasi-Experimentation: Design and analysis issues for field
settings. Boston : Houghton Mifflin.
D’Aveni, R.A. 1994. Hypercompetition. New York: Free Press.
Dacin, M., Ventresca, M., and Beal, B. 1999. The embeddedness of organizations: dialogue and
directions. Journal of Management, 25(3): 317-356.
Darr, E. P., Argote, L. and Epple, D. 1995. The Acquisition, transfer and depreciation of knowledge in
service organizations: Productivity in franchises. Management Science, 41(11): 1750-1762.
Demsetz, H. 1991. The Theory of the Firm Revisited. In Williamson, O.E. and Winter, S. (Eds) The
Nature of the Firm: 159-178 New York: Oxford University Press.
Diamantopoulos, A. and Winklhofer, H. 2001. Index construction with formative indicators: An
alternative to scale development. Journal of Marketing Research, 38.
Drucker, P. 1995. The Post-Capitalist Society. Oxford, UK: Butterworth-Heinemann.
Dyer, J. H. 1997. Effective interfirm collaboration: How firms minimize transaction costs and
maximize transaction value. Strategic Management Journal, 18(7): 535-556.
Dyer, J. H. and Singh, H. 1998. The Relational View: Cooperative Strategy and Sources of
Interorganizational Competitive Advantage. Academy of Management Review, 23 (4): 660-679.
Eder, L. and Igbaria, M. 2001. Determinants of Intranet Diffusion and infusion. The International
Journal of Management Science, 29: 233-242.
Ford, D. P. 2002. Trust and knowledge management: The seeds of success. In Holsapple, C.W., (Ed)
The Handbook on Knowledge Management: 545-575 Berlin: Springer Verlag.
Fornell, C. 1982. A second generation of multivariate analysis, I and II: Methods. New York, New
York: Praeger Special Studies.
Fornell, C. and Larcker, D. 1981. Structural equation modeling with unobservable variables and
measurement error. Journal of Marketing Research, 18: 39-50.
Galbraith, C.S. and Merrill, G.B. 1991. The effect of compensation program and structure on SBU
competitive strategy: A study of technology-intensive firms. Strategic Management Journal, 12 (5):
353-370.
Gefen, D. 2002. Nurturing clients’ trust to encourage engagement success during the customization of
ERP systems. Omega, 30(4): 287-300.
135
Gefen, D., Straub, D., and Boudreau, M-C. 2000. Structural equation modeling and regression:
Guidelines for research practice. Communications of the Association for Information Systems, 4(7).
George, G., Zahra, S., Wheatley, K., Khan, R. 2001. The effects of alliance portfolio characteristics
and absorptive capacity on performance. A study of biotechnology firms. Journal of High Technology
Management Research, 12: 205-226.
Gerbing, D. and Anderson, J. 1988. An updated paradigm for scale development incorporating
unidimensionality and its assessment. Journal of Marketing Research, 25(2): 186-193.
Granovetter, M. 1985. Economic action and social structure: The problem of embeddedness. The
American Journal of Psychology, 91(3), 481-510.
Grant, R. M. 1996. Prospering in Dynamically-Competitive Environments: Organizational Capability
as Knowledge Integration. Organization Science, 7(4): 375-387.
Grant, R. M. 1997. The knowledge-based view of the firm; Implications for management practice.
Long Range Planning, 30(3): 450-454.
Gulati, R. and Gargiulo, M. 1999. Where do interorganizational networks come from? American
Journal of Sociology, 104 (5): 1439-1493.
Gulati, R., Dialdin, D., and Wang, L. 2002. Organizational Networks. In Baum, J. (Ed) Companion
to Organizations: 281-303. Basil Blackwell.
Hair, J., Anderson, R., Thatham, R., and Black, W. 1998. Multivariate Analysis Data Analysis.
Upper Saddle River, NJ: Prentice Hall.
Hamel, G. 1991. Competition for competence and inter-partner sharing within international strategic
alliances. Strategic Management Journal, 12(4): 83-104.
Hamel, G., Doz, Y. L., and Prahalad C.K. 1989. Collaborate with your competitors – and win.
Harvard Business Review, 67(1): 133-139.
Hansen, M. 2002. Knowledge Networks: Explaining effective knowledge sharing in multi-unit
companies. Organization Science, 13(3): 232-248.
Harris, L., Coles, A.M., and Dickson, K. 2000. Building innovation networks: Issues of strategy and
expertise. Technology Analysis and Strategic Management, 12(2): 229-238.
Harrisson, D, and Laberge, M. 2002. Innovation, identities, and resistance: The social construction of
an innovation network. Journal of Management Studies, 39(4): 497-518.
Heiman, B. and Nickerson, J. 2002. Towards Reconciling Transaction Cost Economics and the
Knowledge-based View of the Firm: The Context of Interfirm Collaborations. International Journal of
Economics and Business, 9(1): 97-116.
Ho, J., Keller, R., and Keltyka, P. 2002. Effects of Probabilistic and Outcome Ambiguity on
Managerial Choices. Journal of Risk and Uncertainty, 24(1): 47-74.
Hosmer, L. T. 1995. Trust: The connecting link between organizational theory and philosophical
ethics. Academy of Management Review, 20(2): 379-404.
Huber, G. P. 1991. Organizational learning: The contributing processes and the literatures.
Organization Science, 2 (1): 88-116.
136
Hutchins, E. 1991. Organizing Work by Adaption. Organization Science, 2 (1): 14-38.
Ingram, P. and Simons, T. 2002. The Transfer of Experience in Groups of Organizations:
Implications for Performance and Competition. Management Science, 48(12): 1517-1534.
Ivory.com Retrieved 29, January, 2003 http://www.ivory.com/history.htm.
Johnson, J. 1995. The multidimensional networks of complex systems. In Batten, D., Casti, J., and
Thord, R. (Eds) Networks in action. Communication, Economics and Human Knowledge Berlin:
Springer Verlag.
Johnston, R. and Lawrence, P. R. 1988. Beyond vertical integration – The rise of value-adding
partnership. Harvard Business Review, 66 (4): 94-102.
King, A. W., and Zeithaml, C. 2001. Competencies and firm performance: Examining the causal
ambiguity paradox. Strategic Management Journal, 22(1): 75-99.
Klein, R. 2002. The effects of symmetric information sharing practices in e-business client/vendor
relationships. Unpublished Dissertation. Georgia State University.
Kogut, B. and Zander, U. 1992. Knowledge of the firm, combinative capabilities and the replication of
technology. Organization Science, 3(3): 383-398.
Kogut, B. and Zander, U. 1996. What firms do? Coordination, identity and learning. Organization
Science, 7(5): 502-519.
Korczynski, M. 1996. The low trust route to economic development: Interfirm relations in the UK
engineering construction industry in the 1980s and 1990s. Journal of Management Studies, 33(6):
787-808.
Krackhardt, D. 1998. Simmelian Ties. In Kramer R. and Neale, M (Eds), Power and Influence in
Organizations: 21-35. Sage: Thousand Oaks, CA.
Lane, P. J. and Lubatkin, M. 1998. Relative absorptive capacity and inter-organizational learning.
Strategic Management Journal, 19 (5): 461-477.
Lane, P, Koka, B., and Pathak, S. 2002. A Thematic Analysis an Critical Assessment of Absorptive
Capacity Research. Academy of Management Proceedings, 2002.
Lapre, M. and Wassenhove, L. 2001. Creating and Transferring Knowledge for Productivity
Improvement in Factories. Management Science, 47(10): 1311-1325.
Lee, H and Whang, S. 2002. The impact of the secondary market on the supply chain. Management
Science, 48(6): 719-732.
Lewis, D. J. and Wiegert, A. 1985. Trust as a social reality. Social Forces, 63(4): 967-985.
Li, L. 2002. Information sharing in a supply chain with horizontal competition. Management
Science, 48(9): 1196-1212.
Lim, K. 2000. The many faces of absorptive capacity: Spillovers of copper interconnect technology
for semiconductor chips. Paper presented at the 2000 Academy of Management Conference.
Lippman, S.A. and Rumelt, R.P. 1982. Uncertain imitability: An analysis of interfirm differences in
efficiency under competition. Bell Journal of Economics, 13(2): 418-439.
137
Loebecke, C., van Fenema, P. C., and Powell, P. 1999. Co-opetition and knowledge transfer. The
DATABASE for Advances in Information Systems, 30(2): 14-25.
Madhavan, R., Koka, B. R., and Prescott, J. E. 1998. Networks in transition: How industry events
(re)shape interfirm relationships. Strategic Management Journal, 19(5): 439-459.
Marakas, G. 1999. Decision Support Systems in the Twenty-First Century. Upper Saddle River, NJ:
Prentice Hall.
March, J. G. and Simon, H. A. 1958. Organizations. New York:Wiley.
Matthews, R. 1986. The economics of institutions and the sources of growth. Economic Journal,
96(384): 903-918.
Mayer, R., Davis, J., and Schoorman, F. 1995. An integrative model of organizational trust. Academy
of Management Review, 20(3): 709-734.
Mosakowski, E. 1997. Strategy making under causal ambiguity: Conceptual issues and empirical
evidence. Organization Science, 8(4): 414-442.
Mowery, D., and Oxley, J. 1996. Strategic alliances and interfirm knowledge transfer. Strategic
Management Journal, 17: 77-92.
Neter, J., Kutner, M., Nachtsheim, C., and Wasserman, W. 1996. Applied linear statistical models.
New York, New York: McGraw Hill.
Nonaka, I. 1991. The knowledge creating company. Harvard Business Review, 69(6): 96-105.
Nonaka, I., and T. H. 1995. The Knowledge creating company. Oxford, UK: Oxford University Press.
Nonaka, I. and Teece, D. J. 2001. Managing industrial knowledge: Creation, transfer and utilization.
London: Sage.
Nunnally, J. and Bernstein, I. 1994. Psychomteric Theory. New York: Mcgraw Hill.
Oxley, J. 1997. Appropriability Hazards and Governance in Strategic Alliances: A Transaction Cost
Approach. Journal of Law Economics and Organization. 13(2): 387-409.
Polanyi, M. 1962. Personal knowledge: Toward a post-critical philosophy. New York: Harper
Touchbooks.
Postrel, S. Islands of Shared Knowledge: Specialization and Mutual Understanding in ProblemSolving Teams. Organization Science: A Journal of the Institute of Management Sciences , 13(3): 303321.
Powell, W., Koput, K., and Smith-Doerr, L. 1996. Inter-organizational collaboration and the locus of
innovation: Networks of learning in biotechnology. Administrative Science Quarterly, 41(1): 116146.
Prahalad, C.K. and Hamel, G. 1990. The core competence of the corporation. Harvard Business
Review, 68(3): 79-92.
Reed, R. and Defillippi, R. J. 1990. Causal ambiguity, barriers to imitation, and sustainable
competitive advantage. Academy of Management Review, 15 (1): 88-102.
138
Rigdon, E. 1998. Structural Equation Modeling. In Marcoulides, G. (Ed) Modern Methods for
Business Research: 251-272. Mahwah, N.J.: Lawrence Erlbaum.
Roberts, J. 2000. From Know-how to Show-how? Questioning the Role of Information and
Communication Technologies in Knowledge Transfer. Technology Analysis and Strategic
Management, 12(4): 429-443.
Segars, A. and Grover, V. 1998. Strategic information systems planning success: An investigation of
the construct and its measurement. MIS Quarterly, 22(2): 139-163.
Shapiro, C. and Varian, H. R. 1999. Information rules: A strategic guide to the network economy.
Boston, MA: HBS Press.
Simmel, G. 1950. Individual and society. In K.H. Wolff (Ed.) The sociology of Georg Simmel. New
York: Free Press.
Simonin, B. 1999. Ambiguity and the process of knowledge transfer in strategic alliances. Strategic
Management Journal, 20(7): 595-624.
Spender, J.C. 1996. Making knowledge the basis of a dynamic theory of the firm. Strategic
Management Journal, 17 (Winter Special Issue): 45-62.
Straub, D. 1989. Validating Instruments in MIS Research. MIS Quarterly, 13(2): 146-170.
Stuart, T. 2000. Interorganizational Alliances and the Performance of Firms: A Study of Growth and
Innovation Rates in a High Technology Industry. Strategic Management Journal, 21(8): 791-812.
Szulanski, G. 1996. Exploring internal stickiness: Impediments to the transfer of best practice within
the firm. Strategic Management Journal, 17(Winter Special Issue): 27-43.
Teece, D. J. and Pisano, G. 1994. The dynamic capabilities of firms: An introduction. Industrial and
Corporate Change, 3(3): 537-556.
Teece, D. J. 1996. Firm organization, industrial structure, and technological innovation. Journal of
Economic Behavior and Organization, 31(2): 192-225.
Teece, D. J. 1998. Capturing value from knowledge assets: The new economy, markets for know-how
and intangible assets. California Management Review, 40(3): 55-79.
Teece, D. J. 2001. Strategies for managing knowledge assets: The role of firm structure and industrial
context. In Nonaka, I., and Teece, D. J. (Eds) Managing industrial knowledge: Creation, transfer
and utilization: 125-144. London, UK: Sage.
Thompson, J. 1967. Organizations in action. New York: McGraw Hill.
Thorelli, H. B. 1986. Networks: Between markets and hierarchies. Strategic Management Journal,
7(1): 37-52.
Tsai, W. 2001. Knowledge Transfer in Intraorganizational Networks: Effects of Network Position and
Absorptive Capacity on Business Unit Innovation and Performance. Academy of Management
Journal. 44(5) 996-1004.
Tushman, M. and Anderson, P. 1986. Technological discontinuities and organizational environments.
Administrative Science Quarterly, 31(3): 439-468.
139
Uzzi, B. 1996. The sources and consequences of embeddedness for the economic performance of
organizations: The network effect. American Sociological Review, 61(4): 674-699.
Uzzi, B. and Lancaster, R. 2003. Relational embededdness and learning: The case of bank loan
managers and their clients. Management Science, 49(4): 383-400.
Van den Bosch, F., Volberda, H. W. and de Boer, M. 1991. Coevolution of firm absorptive capacity
and knowledge environment: Organizational forms and combinative capabilities. Organization
Science, 10(5): 551-568.
Van de Ven, A. 1976. On the nature, formation and maintenance of relationships among
organizations. Academy of Management Review, 1:24-36.
Volberda, H.W. 1998. Building the flexible Firm: How to remain competitive. Oxford, UK: Oxford
University Press.
von Hippel, E. 1994. “Sticky information” and the locus of problem solving: Implications for
innovation. Management Science, 40(4): 429-438.
Walker, W. 1995. Technological innovation, corporate RandD alliances and organizational
learning. Unpublished doctoral dissertation, RAND, Santa Monica, CA.
Westlund, H. 1999. An interaction-cost perspective on networks and territory. The Annals of
Regional Science, 33: 93-121.
Wernerfelt, B. 1984. A resource-based view of the firm. Strategic Management Journal, 5(2): 171181.
Wilcox-King, A. and Zeithaml, C. 2001. Competencies and Firm Performance: Examining the Causal
Ambiguity Paradox. Strategic Management Journal, 22(1), 75-99.
Williamson, O. E. 1975. Markets and hierarchies: Analysis and antitrust implications. New York:
Free Press.
Williamson, O.E. 1973. Markets and hierarchies: Some elementary considerations. American
Economic Association, 63(2): 316-325.
Winter, S. 1987. Knowledge and competence as strategic assets. In Teece, D. (Ed), The Competitive
Challenge. Cambridge, MA: Ballinger.
Zucker, L. 1986. Production of trust: Institutional sources of economic structure, 1840-1920. In
Straw, B. and Cummings, L., (Eds) Research in Organizational Behavior: 53-111. Greenwich CT:
JAI Press.
140
Appendix I: Interview Records
CUNA INTERVIEWS
Interview 1C
Date:
Contact:
Duration:
April 2, 2003
Mark Condon, SVP of Association Services at CUNA
45 minutes
The purpose of the call was to:
1. Familiarize Mark with the study. Although Mark and I had
communicated several times via email, this represented our first live
conversation;
2. Determine if, based upon the definition in the document, if CUNA was a
“co-opetive” network;
3. Obtain thoughts and guidance from Mark regarding the proposed pilot
(CUNA members would be solicited for the pilot, because of the large
number of members – 10,000).
Mark was able to “confirm” that CUNA was a co-opetive network and appropriate for
inclusion in this study. Specifically, he pointed to the following facts:



All credit unions (of the same general size) have approximately the same
operations and do approximately the same thing – they would be considered to
have “low scope”. Mark stated that CUs of different sizes may have different
experiences (see Interview 4 notes). As a result, a clarifier was inserted into the
survey questions regarding similarities of operations –
o “Our day-to-day operations are similar to most other credit union
members (of similar size) within CUNA.”
The CUNA organization in Madison, WI represents a repository for information
and a general consortium for discussion, education, training and exchange. But,
ultimately, they cannot standardize language or operations, or have the authority
to punish – low governmental authority.
Credit unions that are not chartered for low income or have a corporate
affiliation, are generally in competition with each other, especially the smaller
credit unions.
Mark stated that knowledge transfer is viewed as one of the advantages of membership in
CUNA. The sharing of experiences and operational best practices are encourages
through regular meetings at the local, regional and national levels. In addition, members
are encouraged to regularly access the CUNA website to obtain information regarding
best practices, financial information, etc. They also use a LISTSERV process to post
questions and responses to share information.
141
Regarding the interactions among members, Mark indicated that trust and willingness to
share information would be expected to vary based upon topic. For example, there is
high willingness to share information regarding bankruptcies/risk/disaster recovery…but
low willingness to share information regarding marketing and HR practices.
Mark identified the specific factors of importance for performance at a credit union.
These include:
 Increasing membership (of customers)
 Retaining personnel (decreasing turnover)
 Achieving high CAML ratings (capital adequacy, asset quality, earnings,
liabilities)
These factors were used to develop the measurement items related to causal ambiguity –
common factors.
Interview 2C
Date:
Contact:
Duration:
April 10, 2003
Cookie Yoder, CEO of Pittsburgh Federal Credit Union
30 minutes
As the CEO of a member of CUNA, Cookie provided an alternative perspective to Mark.
Cookie was very active in the committees and boards within CUNA and therefore was
more likely to engage in knowledge sharing than perhaps other CU CEOs. It is also
important to note that the Pittsburgh Credit Union is relatively small operation ($6M in
assets) and is within Mark’s recommended target. Cookie emphasized the need for CUs
to share their learnings and best practices related to specific programs and regulations,
such as the requirements associated with the Patriot Act and the recent alliance between
CUNA and the Fannie Mae corporation. This identification of experiences with specific
programs is similar to the statement made by Mark regarding the circumstances under
which CUs would be expected to share. I shared the draft pre-pilot instrument with her
and she helped to refine the wording of the pilot survey, to ensure that language that
would be familiar to credit union managers was utilized.
Interview 3C
Date:
Contact:
Duration:
April 14, 2003
Eric Anderson-Zych, CEO of Genessee Credit Union in Rochester NY
30 minutes
Like Cookie, Eric is the CEO of a member credit union of CUNA. Unlike Cookie’s
credit union which was a community chartered organization, Eric’s credit union is a “low
income” designated credit union, operating for the purposes of economic development in
142
a disadvantaged area of Rochester. After reviewing the study objectives with Eric, he
stated that he did not believe that CUNA treated all CUs equally. Specifically, Eric
stated that he believed that larger CUs received more services for lower fees (on a
percentage basis) than did smaller CUs like his. He was also of the perspective that the
managers of larger credit unions had a better understanding of the industry than did the
smaller credit unions…although the larger credit unions (those affiliated with large
employee unions, the military, universities, etc) are the least likely to be engaged in a
competitive environment. His comments eventually lead to the creation of several
questions including:
- “CUNA has developed a reputation for ensuring fairness and trust among its members”.
- “All credit union members of CUNA have access to the same basic knowledge of the
credit union industry”
- “CUNA makes the same basic knowledge available to all credit union members”.
Interview 4C
Date:
Contact:
Duration:
May 19, 2003
Mark Condon, SVP of Association Services at CUNA
20 minutes
This follow up interview with Mark was intended to define the sample for the field study.
In Mark’s opinion, credit unions with different charters and of different sizes may not
have the same operations and overall experiences, and as a result may experience
different levels of sharing and of competition. For example, in his opinion, CUs with
over $500M in assets do not have a similar operation to CUs with $5M in assets. In
addition, CUs with “low income” designations (those credit unions which have a federal
charter to support the banking needs of communities that are considered to be
“disadvantaged”) would be expected to have experiences very different from other credit
unions. In addition, these credit unions, because of their charter, encounter effectively no
competition. Also, CUs with “affiliated” designations (those credit unions that are
affiliated with a military entity, a university, an employee union, etc) would also not be
expected to experience much competition. Overall, it was of Mark’s opinion that the
appropriate target for this study was community chartered credit unions with assets
between $5M and $50M.
Interview 5C
Date:
Contact:
Duration:
June 9, 2003
Mark Condon, SVP of Association Services at CUNA
20 minutes
The initial results of the pilot study were shared with Mark. Mark provided feedback on
comments to improve the wording of the questions.
143
SUNTRUST INTERVIEWS
Interview 1S
Date:
Contact:
Banking
Duration:
March 28, 2003
Washington Dender, SVP of Strategy and Will Hileman, SVP Retail
60 minutes
As the initial contact with SunTrust, the purpose of this discussion was to explain the
research project and gain access to the bank for the project. Will stated that based upon
the definition provided of a franchise network in the document, he believed that the
branches were indeed a “franchise”. His justification was the low scope and centralized
governance structure of the network. He did state, however, that SunTrust management
created an “artificial competition” among branch managers in an effort to increase
performance. However, neither he nor anyone in the organization fully understood how
this “artificial” competition impacted the extent to which knowledge transferred (or not)
among the branch managers.
The result of this conversation was two fold – as the “gatekeepers” Wash and Will
provided “approval” for inclusion of SunTrust in the field study and provided for access
to the branch managers. In addition, Wash and Will stated the importance of measuring
the two-way perceptions of the source-recipient relations…how the source perceives the
recipient and vice versa amongst the branch managers.
Interview 2S
Date:
April 22, 2003
Contact:
Joy Wildermorgan, SVP of Atlanta Region and Kirk Watkins, VP of
Atlanta Region
Duration:
30 minutes
Joy and Kirk would become the primary contacts for SunTrust’s inclusion in the field
study. Joy was particularly interested in the efficiency of communication between the
branch mangers and the Atlanta Regional Headquarters and how this efficiency (or lack
of) eventually impacted knowledge transfer between and among branch managers. This
discussion lead to the development of the following questions:
“When we communicate with SunTrust Atlanta Region Management, we believe that
what we say is generally understood"
“When we place a request for information to SunTrust Atlanta Region Management, we
never have to re-state or explain our request”
144
“When we place a request for information to SunTrust Atlanta Region Management, the
recipient who will respond to our request generally “gets it right” the first time”.
(Although these points were used in the pilot survey with CUNA members, this
conversation was the basis for the development of the questions).
Interview 3S
Date:
Contact:
Duration:
May 26, 2003
Kirk Watkins, VP of Atlanta Region
20 minutes
The pilot survey and results (which included only CUNA respondents) were shared with
Kirk. His input was requested to ensure that the wording was appropriate for the final
field study survey for the branch managers.
Interview 4S
Date:
October 17, 2003
Contact:
Kirk Watkins, VP of Atlanta Region
Duration:
60 minutes
I requested this discussion to get clarification from SunTrust on the “competition” that
existed among branch managers.
Kirk explained that there are four types of reward and recognition programs at SunTrust:
1. Every employee associated with Branch Banking has an individual performancebased incentive that is based on absolute figures and involves no comparisons
with other’s performance.
2. At an Area level (the Atlanta Region has 4 areas with approximately 70 people in
an area), employees must achieve a customer satisfaction score of 80 or greater to
receive an Area level incentive.
3. The “Edgemaster” Award is awarded to the top 5% of quarterly performers in
each position within SunTrust bank. In addition to the public recognition, the
recipient receives approximately $1000.
4. “LeaderBoard” Rankings is a recognition only (no financial incentives) “Board”
that is published on the Company intranet and in regular paper-based
communications sent to the entire company.
Although the LeaderBoard is only a medium for communication of relative Area
performance, Kirk stated that being on the top or the bottom in an incentive for
performance. He acknowledges that it creates competition among the branch
managers…in some cases less to be at the top and more to not be at the bottom. “We use
competition as a strategic tool to drive performance”.
145
In a similar ranking scheme, the branch managers are ranked on an annual basis from 1165. From this ranking, the bottom 30 are then placed on a disciplinary action plan,
where the penalty for non-improvement is dismissal. Kirk stated that this too creates
inherent pressure and competition amongst the managers to not fall into this bottom band.
“There is a great sense of pressure on the branch managers to perform well”.
KNOWLEDGE MANAGER INTERVIEW
Interview 1K
Date:
Contact:
Duration:
April 11, 2003
Kevin Bartlett, Manager of the Accenture “Knoweldge Exchange”
30 minutes
The Knowledge Exchange is the facility at Accenture that employees can use to gather
information from other Accenture projects as well as from outside sources.
Main points from discussion with Kevin –
 Accenture is struggling with the most effective way to measure the economic
impact of the knowledge exchange. The primary metric they use today is
“opportunity cost”…for example…if the KX completed 1500 hours worth of
research, that 1500 hours became client billable. If the average client billing rate
is $300/hour, then that translates to $450,000.
 When asked why people don’t share within the knowledge exchange, the primary
reason was cultural – knowledge hoarding to create a perception of power.
Although Kevin said that this is slowly changing.
 Culturally, there is a perception that everything must be perfect before it can go
out…people don’t share work in progress. As a result, people are less willing to
share negative knowledge...although Kevin stated that negative knowledge is
sometimes more useful than “positive” knowledge.
 When things go right…it is generally for two reasons:
o Top management support makes knowledge sharing a priority and
engrains it as important.
o Each project group will assign someone who is strong in KM…as well as
possessing a base knowledge of the topics on the project – this may be
representative of the “trans-specialist” described by David Teece.
 There is generally a culture of fairness and willingness to share…and an
assumption that Accenture will punish for opportunistic behavior.
 Kevin expressed a concern that the KX was contributing to a “dumbing down” of
the consultants…because people just look for an answer…they don’t want to take
the time to think.
146
Appendix II: Explanation of Field Study Measurement Items
Question
Variable
Source (if applicable)
Q1.
Q2.
Q3.
Q4.
Q5.
Q6.
Q7.
Q8.
Knowledge Transfer
Knowledge Transfer
Knowledge Transfer
Knowledge Transfer
Knowledge Transfer
Knowledge Transfer
Knowledge Transfer
Knowledge Transfer
Dyer (1997)/Birkinshaw (2002)
Dyer (1997)/Birkinshaw (2002)
Dyer (1997)/Birkinshaw (2002)
Dyer (1997)/Birkinshaw (2002)
Hansen (2002)
Hansen (2002)
Simonin (1999)
Simonin (1999)
Q9.
Q10.
Q11.
Q12.
Q13.
Q14.
Q15.
Q16.
Q17.
Q18.
Q19.
Q20.
Q21.
Q22.
Q23.
Q24.
Absorptive Capacity – Common Language
Absorptive Capacity – Common Language
Absorptive Capacity – Common Language
Absorptive Capacity – Common Language
Absorptive Capacity – Common Language
Absorptive Capacity – Common Language
Absorptive Capacity – Common Language
Absorptive Capacity – Common Process
Absorptive Capacity – Common Process
Absorptive Capacity – Common Process
Absorptive Capacity – Problem Solving
Absorptive Capacity – Problem Solving
Absorptive Capacity – Problem Solving
Absorptive Capacity – Common Knowledge
Absorptive Capacity – Common Knowledge
Absorptive Capacity – Common Knowledge
Szulanski (1996)
Szulanski (1996)
Szulanski (1996)
Interview
Interview
Szulanski (1996)
Szulanski (1996)
Szulanski (1996)
Szulanski (1996)
Szulanski (1996)
Szulanski (1996)
Interview
Interview
Interview
Szulanski (1996)
Szulanski (1996)
Q25.
Q26.
Q27.
Q28.
Q29.
Q30.
Q31.
Causal Ambiguity – Causal Factors
Causal Ambiguity – Causal Factors
Causal Ambiguity – Causal Factors
Causal Ambiguity – Causal Factors
Causal Ambiguity – Competitive Advantage
Causal Ambiguity – Competitive Advantage
Causal Ambiguity – Competitive Advantage
Szulanski (1996)
Szulanski (1996)
Szulanski (1996)
Szulanski (1996)
King and Zeithaml (2000)
King and Zeithaml (2000)
King and Zeithaml (2000)
Q32.
Q33.
Q34.
Q35.
Q36.
Q37.
Q38.
Q39
Q40.
Q41.
Q42.
Q43.
Outcome Ambiguity – Proven Knowledge
Outcome Ambiguity – Proven Knowledge
Outcome Ambiguity – Proven Knowledge
Outcome Ambiguity – Source Recipient Relations
Outcome Ambiguity – Source Recipient Relations
Outcome Ambiguity – Source Recipient Relations
Outcome Ambiguity – Source Recipient Relations
Outcome Ambiguity – Source Recipient Relations
Outcome Ambiguity – Source Recipient Relations
Outcome Ambiguity – Source Recipient Relations
Outcome Ambiguity – Source Recipient Relations
Outcome Ambiguity – Source Recipient Relations
Based on Dyer (1994)
Based on Dyer (1994)
Based on Simonin (1999)
Based on Simonin (1999)
Based on Simonin (1999)
Based on Dyer (1994)
Based on Dyer (1994)
Based on Dyer (1994)
Based on Dyer (1994)
Based on Dyer (1994)
Based on Dyer (1994)
Based on Dyer (1994)
147
APPENDIX III – PILOT SURVEY
As a member of CUNA (the Credit Union National Association), your credit union may
receive, as well as provide, information on a variety of topics, including operations,
marketing, compliance, pending legislation, etc. You may also take advantage of other
opportunities to interact with CUNA and/or other credit union members of CUNA.
Considering your position as a credit union member of CUNA, which statement most
closely reflects your views:
 A. Our credit union is very integrated within CUNA. We place a high value on
our membership.
 B. Our credit union has limited interaction within CUNA. We don’t really
identify with the CUNA network.
If you checked “A”, please answer questions 1A-50A only (first attachment). If you
checked “B”, please skip to the second attachment and answer questions 1B-48B only
(second attachment).
This research study is not sponsored nor directly affiliated with CUNA.
148
“A” Questions
Please indicate how representative each statement is of your credit union.
Not Representative
1
2
3
4
5
Very Representative
6
7
Q1A. CUNA (or another credit union member of CUNA) has provided knowledge of operational processes
(including marketing or risk or compliance, etc) to our credit union that has enabled us to improve our
performance.
1
2
3
4
5
6
7
Q2A. Our credit union has provided knowledge of operational processes (including marketing or risk or
compliance, etc) to CUNA or to another credit union member of CUNA that enabled performance
improvement.
1
2
3
4
5
6
7
Q3A. Our credit union has received knowledge of an operational process (including marketing or risk or
compliance, etc) that was developed at another credit union member of CUNA.
1
2
3
4
5
6
7
Q4A. Other credit unions within CUNA have received knowledge of an operational process (including
marketing or risk or compliance, etc) that was developed at our credit union.
1
2
3
4
5
6
7
Q5A. Our credit union has trialed/implemented concepts that came from CUNA or from another credit
union member within CUNA.
1
2
3
4
5
6
7
Q6A. When we communicate with CUNA, I believe that what we say is generally understood.
1
2
3
4
5
6
7
Q7A. When we place a request for information to CUNA, we never have to re-state or explain our request.
1
2
3
4
5
6
7
Q8A. When we place a request for information to CUNA, the recipient of our request generally “gets it
right” the first time.
1
2
3
4
5
6
7
149
Q9A. Our credit union uses similar terminology to that used by other credit union members of CUNA to
describe issues, concepts, processes, etc.
1
2
3
4
5
6
7
Q10A. Other credit union members of CUNA generally use terms that our credit union understands.
1
2
3
4
5
6
7
Q11A. When our credit union receives documents or other communications from CUNA, we always
understand all terminology.
1
2
3
4
5
6
7
Q12A. We rarely require any terminology clarification when reading CUNA documents.
1
2
3
4
5
6
7
Q13A. Our day-to-day operations are similar to most other credit union members (of similar size) within
CUNA.
1
2
3
4
5
6
7
Q14A. Considering only the operations (ignoring culture, personalities, pay scale, etc), it would be easy for
an employee from our credit union to transition into another credit union within CUNA.
1
2
3
4
5
6
7
Q15A. Our credit union understands the basic operations of most other credit union members (of similar
size) within CUNA.
1
2
3
4
5
6
7
Q16A. The way that managers approach problems at our credit union is similar to the way that managers
approach problems at other credit unions within CUNA.
1
2
3
4
5
6
7
Q17A. Our credit union resolves issues in the same way that other credit unions within CUNA resolve
issues.
1
2
3
4
5
6
7
Q18A. The problem solving skills of employees at other credit union members of CUNA are similar to the
problem solving skills of employees at our credit union.
1
2
3
4
5
6
7
150
Q19A. All credit union members of CUNA have access to the same basic knowledge of the credit union
industry.
1
2
3
4
5
6
7
Q20A. CUNA makes the same basic knowledge available to all credit union members.
1
2
3
4
5
6
7
Q21A. All managers at credit union members of CUNA have the same basic knowledge of the industry.
1
2
3
4
5
6
7
Q22A. Our credit union understands how other credit union members of CUNA increase their membership.
1
2
3
4
5
6
7
Q23A. Our credit union understands how other credit union members of CUNA retain good personnel.
1
2
3
4
5
6
7
Q24A. Our credit union understands how other credit union members of CUNA achieve high CAMEL
ratings.
1
2
3
4
5
6
7
Q25A. Senior Management of all credit union members of CUNA would agree about the factors which
create financial success at a credit union.
1
2
3
4
5
6
7
Q26A. Our credit union understands what makes our credit union a better/worse performer relative to other
credit union members of CUNA.
1
2
3
4
5
6
7
Q27A. Our credit union personnel understand what differentiates our credit union from other credit unions.
1
2
3
4
5
6
7
Q28A. When CUNA or other credit unions share knowledge of processes or practices with our credit
union, they have a good idea of how we will use it.
1
2
3
4
5
6
7
Q29A.When we share knowledge of a process or practice with CUNA or with other credit union members
within CUNA, we generally understand how that knowledge will be used.
1
2
3
4
5
6
7
Q30A. When CUNA or other credit unions share knowledge of processes or practices with our credit
union, they already have experience with how that knowledge contributes to performance.
1
2
3
4
5
6
7
151
Q31A. Our credit union trusts other credit union members of CUNA to deal with us fairly.
1
2
3
4
5
6
7
Q32A. Given the opportunity, we believe that other credit union members of CUNA would take unfair
advantage of our credit union.
1
2
3
4
5
6
7
Q33A. CUNA has developed a reputation for ensuring fairness and trust among its credit union members.
1
2
3
4
5
6
7
Q34A. Our credit union has policies (either formal or informal) to restrict the sharing of knowledge with
other credit unions.
1
2
3
4
5
6
7
Q35A. Other credit unions within CUNA are protective of their knowledge.
1
2
3
4
5
6
7
Q36A. Our credit union has had a direct experience with another credit union member of CUNA refusing
to share knowledge of a process.
1
2
3
4
5
6
7
Q37A. Our credit union is in competition with other credit union members of CUNA.
1
2
3
4
5
6
7
Q38A. The majority of the knowledge that our credit union receives from CUNA is in paper or electronic
form (rather than knowledge that has to be demonstrated).
1
2
3
4
5
6
7
Q39A. My credit union has worked directly with CUNA or another credit union member of CUNA to
demonstrate a process or concept that could not be easily documented.
1
2
3
4
5
6
7
Q40A. Employees within our credit union feel comfortable using a computer to communicate with CUNA
and other credit union members within CUNA.
1
2
3
4
5
6
7
Q41A. Employees within our credit union feel comfortable accessing the CUNA website.
1
2
3
4
5
6
152
7
Q42A. Computers enhance the productivity of employees at our credit union.
1
2
3
4
5
6
7
Q43A. Considering the transfer of knowledge among credit union members within CUNA, would you say
that your credit union:
Shares more knowledge than we receive
Receives more knowledge than we share
Shares and Receives knowledge equally
Q44A. What is the approximate size of your organization?
Under $10million
Over $10 million,
But under $50 million.
Over $50 Million,
But under $100 Million
Over $100 Million
Q45A. Approximately how long has your organization been a member of CUNA?
Less than 1 year
1-5 years
6-10 years
11+ years
Q46A. How would you characterize your credit union?
 Low Income Designation
 Community Chartered
 Company/Association Affiliated (including military, educators, etc.)
Q47A. Has your credit union received knowledge of the following topics from within CUNA? (check all
that apply):
 Providing services to small business (Please provide an example if possible)
 The requirements and opportunities related to the Patriot Act (Please provide an example if
possible)
 New mortgage product options related to the Freddie Mac/CUNA alliance (Please provide an
example if possible)
 Other – Please Explain
 Other – Please Explain
 Other – Please Explain
Q48A. Has your credit union shared knowledge of the following topics (check all that apply):
 Providing services to small business
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 The requirements and opportunities related to the Patriot Act
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
153
 New mortgage product options related to the Freddie Mac/CUNA alliance
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 Other
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 Other
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 Other
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
49A.Does your credit union generally share knowledge through any of the following channels? (check all
that apply):






CUNA Councils
COBWEB
CU Exchange
National Credit Union Roundtable
Other
Our credit union does not generally share knowledge within CUNA
50A. What is your primary reason for participating within CUNA?
154
“B” Questions
Please indicate how representative each statement is of your credit union.
Not Representative
1
2
3
4
5
Very Representative
6
7
Q1B. Our credit union has received knowledge of operational processes (including marketing or risk or
compliance, etc) from another organization that has enabled us to improve our performance.
1
2
3
4
5
6
7
Q2B. Our credit union has provided knowledge of operational processes (including marketing or risk or
compliance, etc) another organization that enabled performance improvement.
1
2
3
4
5
6
7
Q3B. Our credit union has received knowledge of an operational process (including marketing or risk or
compliance, etc) that was developed at another organization.
1
2
3
4
5
6
7
Q4B. Other organizations have received knowledge of an operational process (including marketing or risk
or compliance, etc) that was developed at our credit union.
1
2
3
4
5
6
7
Q5B. Our credit union has trialed/implemented concepts that came from another organization.
1
2
3
4
5
6
7
Q6B. When our credit union communicates with other organizations in our industry, I believe that what we
say is generally understood.
1
2
3
4
5
6
7
Q7B. When I place a request for information to another organization in our industry, I never have to restate or explain my request.
1
2
3
4
5
6
7
Q8B. When I place a request for information to another organization in our industry, the recipient of my
request generally “gets it right” the first time.
1
2
3
4
5
6
7
155
Q9B. Our credit union uses similar terminology to that used by other organizations in our industry to
describe issues, concepts, processes, etc.
1
2
3
4
5
6
7
Q10B. Other organizations in our industry generally use terms that our credit union understands.
1
2
3
4
5
6
7
Q11B. When my credit union receives documents or other communications from other organizations in our
industry, we always understand all terminology.
1
2
3
4
5
6
7
Q12B. We rarely require any terminology clarification when reading documents that come from other
organizations in our industry.
1
2
3
4
5
6
7
Q13B. Our day-to-day operations are similar to most other organizations in our industry (of similar size).
1
2
3
4
5
6
7
Q14B. Considering only the operations (ignoring culture, personalities, pay scale, etc), it would be easy for
an employee from our credit union to transition into another organization in our industry.
1
2
3
4
5
6
7
Q15B. Our credit union understands the basic operations (including marketing or risk or compliance, etc)
of most other organizations in our industry.
1
2
3
4
5
6
7
Q16B. The way that managers approach problems at our credit union is similar to the way that managers
approach problems at other organizations in our industry.
1
2
3
4
5
6
7
Q17B. Our credit union resolves issues in the same way that other organizations within our industry resolve
issues.
1
2
3
4
5
6
7
Q18B. The problem solving skills of employees at other organizations within our industry are similar to the
problem solving skills of employees at our credit union.
1
2
3
4
5
6
7
156
Q19B. All organizations within our industry have access to the same basic knowledge of the credit union
industry.
1
2
3
4
5
6
7
Q20B. All managers at organizations within our industry have the same basic knowledge of the credit
union industry.
1
2
3
4
5
6
7
Q21B. Our credit union understands how other organizations within our industry increase their customer
base.
1
2
3
4
5
6
7
Q22B. Our credit union understands how other organizations within our industry retain good personnel.
1
2
3
4
5
6
7
Q23B. Our credit union understands how other organizations within our industry achieve financial success.
1
2
3
4
5
6
7
Q24B. Senior Management of all organizations within our industry would agree about the factors which
create financial success.
1
2
3
4
5
6
7
Q25B. Our credit union understands what makes our credit union a better/worse performer relative to other
organizations within our industry.
1
2
3
4
5
6
7
Q26B. Our credit union personnel understand what differentiates our credit union from other organizations
within our industry.
1
2
3
4
5
6
7
Q27B. When organizations within our industry share knowledge of processes or practices with our credit
union, they have a good idea of how we will use it.
1
2
3
4
5
6
7
Q28B.When we share knowledge of a process or practice (including marketing or risk or compliance, etc)
with organizations within our industry, we generally understand how that knowledge will be used.
1
2
3
4
5
6
7
157
Q29B. When organizations within our industry share knowledge of processes or practices (including
marketing or risk or compliance, etc) with our credit union, they already have experience with how that
knowledge contributes to performance.
1
2
3
4
5
6
7
Q30B. Our credit union trusts other organizations within our industry to deal with us fairly.
1
2
3
4
5
6
7
Q31B. Given the opportunity, we believe that other organizations within our industry would take unfair
advantage of our credit union.
1
2
3
4
5
6
7
Q32B. Our credit union has policies (either formal or informal) to restrict the sharing of knowledge with
other organizations within our industry.
1
2
3
4
5
6
7
Q33B. Other organizations within our industry are protective of their knowledge.
1
2
3
4
5
6
7
Q34B. Our credit union has had a direct experience with another organization within our industry refusing
to share knowledge of a process (including marketing or risk or compliance, etc).
1
2
3
4
5
6
7
Q35B. Our credit union is in competition with other credit union members of CUNA.
1
2
3
4
5
6
7
Q36B. The majority of the knowledge that our credit union receives from organizations within our industry
is in paper or electronic form (rather than knowledge that has to be demonstrated).
1
2
3
4
5
6
7
Q37B. My credit union has worked directly with other organizations within our industry to demonstrate a
process or concept that could not be easily documented.
1
2
3
4
5
6
7
Q38B. Employees within our credit union feel comfortable using a computer to communicate with
organizations within our industry.
1
2
3
4
5
6
7
158
Q39B. Employees within our credit union feel comfortable accessing the Internet.
1
2
3
4
5
6
7
Q40B. Computers enhance the productivity of employees at our credit union.
1
2
3
4
5
7
6
Q41B. Considering the transfer of knowledge, would you say that your credit union:
Shares more knowledge than we receive
Receives more knowledge than we share
Shares and Receives knowledge equally
Q42B. What is the approximate size of your organization (assets under management)?
Under $10million
Over $10 million,
But under $50 million.
Over $50 Million,
But under $100 Million
Over $100 Million
Q43B. How would you characterize your credit union?
 Low Income Designation
 Community Chartered
 Company/Association Affiliated (including military, educators, etc.)
Q44B. Has your credit union received knowledge of the following topics from within CUNA? (check all
that apply):
 Providing services to small business (Please provide an example if possible)
 The requirements and opportunities related to the Patriot Act (Please provide an example if
possible)
 New mortgage product options related to the Freddie Mac/CUNA alliance (Please provide an
example if possible)
 Other – Please Explain
 Other – Please Explain
 Other – Please Explain
Q45B. Has your credit union shared knowledge of the following topics (check all that apply):
 Providing services to small business
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 The requirements and opportunities related to the Patriot Act
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 New mortgage product options related to the Freddie Mac/CUNA alliance
o Shared only with personal contacts
159
o
 Other
o
o
 Other
o
o
 Other
o
o
Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
Shared only with personal contacts
Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
Shared only with personal contacts
Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
Shared only with personal contacts
Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
46B.Does your credit union generally share knowledge through any of the following channels? (check all
that apply):






CUNA Councils
COBWEB
CU Exchange
National Credit Union Roundtable
Other
Our credit union does not generally share knowledge within CUNA
47B. What is your primary reason for membership within CUNA?
160
APPENDIX IV: FIELD STUDY COVER LETTERS AND SURVEYS
June 9, 2003
Branch Manager
ABC Suntrust Branch
123 Main Street
Atlanta, GA 30303
Dear Manager,
As a doctoral student at Georgia State University, I am currently engaged in a research
initiative examining the factors which influence knowledge sharing among organizations
embedded within a network – such as the branch network of Suntrust Bank. When
completed, the results of this research initiative will be used by both university
researchers as well as individuals in the business community to improve the flow of
knowledge both within and across firm boundaries.
I have contacted you, because as a Suntrust branch manager, your real world experiences
are extremely valuable and relevant to this research. Your responses will be aggregated
with other Suntrust branch managers; individual responses will remain anonymous.
This research initiative is neither funded nor sponsored by Suntrust Bank.
I have enclosed a brief survey that should take less than 15 minutes to complete. After
you have completed all questions, please return it to me using the enclosed postage-paid
envelope.
If you would like a copy of the final results, and/or would like to discuss this survey or
your responses, please enclose your business card and I will contact you.
Your time is very much appreciated. In advance, thank you.
Jennifer Priestley, Ph.D. Candidate
www.dsc.gsu.edu/dscjlp
Georgia State University
35 Broad Street, Ste. 826
Atlanta, GA 30303
161
Thank you for agreeing to participate in this research initiative. Your real world input is
extremely valuable. This survey should take no more than 15 minutes to complete. Please return
your responses using the enclosed postage-paid return envelope. Your responses will remain
anonymous and will be completely confidential.
Please indicate how relevant each statement is to your experiences as a SunTrust Bank Branch
Manager, by circling the appropriate number:
Not Representative
1
2
3
4
5
Very Representative
6
7
Q1. SunTrust Atlanta Region Management has provided knowledge of processes (including marketing or
operations or risk, etc) to our branch that has enabled us to improve our performance.
1
2
3
4
5
6
7
Q2. Our branch has provided knowledge of processes (including marketing or operations or risk, etc) to
SunTrust Atlanta Region Management or to another SunTrust branch that enabled performance
improvement.
1
2
3
4
5
6
7
Q3. Our branch has received knowledge of processes (including marketing or operations or risk, etc) that
was developed at another SunTrust branch.
1
2
3
4
5
6
7
Q4. Other SunTrust branches have received knowledge of processes (including marketing or operations or
risk, etc) that was developed at our branch.
1
2
3
4
5
6
7
Q5. Our branch has piloted concepts that came from SunTrust Atlanta Region Management or from another
SunTrust branch.
1
2
3
4
5
6
7
Q6. Our branch has implemented concepts that came from SunTrust Atlanta Region Management or from
another SunTrust branch.
1
2
3
4
5
6
7
Q7. Our branch actively seeks out other SunTrust branches for potential collaboration or assistance.
1
2
3
4
5
162
6
7
Not Representative
1
2
3
4
5
Very Representative
6
7
Q8. Other SunTrust branches look to our branch for potential collaboration or assistance.
1
2
3
4
5
6
7
Q9. When we communicate with SunTrust Atlanta Region Management, we believe that what we say is
generally understood.
1
2
3
4
5
6
7
Q10. When we place a request for information to SunTrust Atlanta Region Management, we never have to
re-state or explain our request.
1
2
3
4
5
6
7
Q11. When we place a request for information to SunTrust Atlanta Region Management, the recipient who
will respond to our request generally “gets it right” the first time.
1
2
3
4
5
6
7
Q12. We use similar terminology to that used by other branch employees to describe issues, concepts,
processes, etc.
1
2
3
4
5
6
7
6
7
Q13. Other branch employees generally use terms that we understand.
1
2
3
4
5
Q14. When our branch receives documents or other communications from SunTrust Atlanta Region
Management, we always understand all terminology.
1
2
3
4
5
6
7
Q15. We rarely require any terminology clarification when reading SunTrust documents or when speaking
with SunTrust Atlanta Regional Management.
1
2
3
4
5
6
7
Q16. What we do on a day-to-day basis is similar to what most other SunTrust branches do on a day-to-day
basis.
1
2
3
4
5
6
7
163
Not Representative
1
2
3
4
5
Very Representative
6
7
Q17. It would be easy for an employee from our branch to transition into another SunTrust branch.
1
2
3
4
5
6
7
Q18. Employees at our branch understand the basic processes of most other SunTrust branches.
1
2
3
4
5
6
7
Q19. The way that our branch employees approach problems at our branch is similar to the way that other
branch employees approach problems at other SunTrust branches.
1
2
3
4
5
6
7
Q20. Our branch employees resolve issues in the same way that other SunTrust branch employees resolve
issues.
1
2
3
4
5
6
7
Q21. The problem solving skills of employees at other SunTrust branches are similar to the problem
solving skills of employees at our branch.
1
2
3
4
5
6
7
Q22. All SunTrust branch managers have access to the same basic knowledge of the banking industry.
1
2
3
4
5
6
7
Q23. SunTrust Atlanta Region Management makes the same basic knowledge available to all branch
managers.
1
2
3
4
5
6
7
Q24. All SunTrust branch managers have the same basic knowledge of the banking industry.
1
2
3
4
5
6
7
Q25. Our branch employees could create a specific list of the factors that do or do not contribute to an
increase in the customer base.
1
2
3
4
5
6
7
Q26. Our branch employees could create a specific list of the factors that do or do not contribute to the
retention of good personnel.
1
2
3
4
5
164
6
7
Not Representative
1
2
3
4
5
Very Representative
6
7
Q27. Our branch employees could create a specific list of the factors that do or do not contribute to high
sales per FTE.
1
2
3
4
5
6
7
Q28. SunTrust branch managers would agree about the factors which generate sales production at a branch.
1
2
3
4
5
6
7
Q29. Our branch employees understand what makes our branch a better/worse performer relative to other
SunTrust branches.
1
2
3
4
5
6
7
Q30. Our branch employees could create a list of the factors that differentiate our SunTrust branch from
other SunTrust branches.
1
2
3
4
5
6
7
Q31. Our branch employees could create a list of the factors that do or do not contribute to sales production
at other SunTrust branches.
1
2
3
4
5
6
7
Q32. When other SunTrust branches share knowledge of processes or practices with our branch, they have
a good idea of how we will use it.
1
2
3
4
5
6
7
Q33.When our branch shares knowledge of a process or practice with SunTrust Atlanta Region
Management or with another SunTrust branch, we generally understand how that knowledge will be used.
1
2
3
4
5
6
7
Q34. When other SunTrust branches share knowledge of processes or practices with our branch, they
already have experience with how that knowledge contributes to performance.
1
2
3
4
5
6
7
Q35. Employees of our branch trust employees of other SunTrust branches to deal with us fairly.
1
2
3
4
5
6
7
Q36. Given the opportunity, we believe that other SunTrust branches would take unfair advantage of our
branch.
1
2
3
4
5
6
7
165
Not Representative
1
2
3
4
5
Very Representative
6
7
Q37. Other SunTrust branch managers believe that our branch would take unfair advantage of them.
1
2
3
4
5
6
7
Q38. SunTrust Atlanta Region Management has developed a reputation for ensuring fairness and trust
among its branch managers.
1
2
3
4
5
6
7
Q39. Our branch restricts the sharing of knowledge (either formally or informally) with other SunTrust
branches.
1
2
3
4
5
6
7
Q40. Other SunTrust branches are protective of their knowledge.
1
2
3
4
5
6
7
Q41. We believe that other SunTrust branches have policies (either formally or informally) to restrict the
sharing of knowledge with our branch.
1
2
3
4
5
6
7
Q42. There have been times where employees of our branch (including myself) have not shared knowledge
with employees of another SunTrust branch because we see our branch in competition with other branches.
1
2
3
4
5
6
7
Q43. We believe that there are employees of other SunTrust branches that would refuse to share knowledge
of a process (including marketing or operations or risk, etc.) with employees of other SunTrust branches.
1
2
3
4
5
6
7
Q44. The majority of the knowledge that our branch receives from SunTrust Atlanta Region Management
is in the form of manuals or guidelines or other similar documentation.
1
2
3
4
5
6
7
Q45. SunTrust Atlanta Region Management or employees of other SunTrust branches rarely need to come
to our branch to discuss face-to-face or to demonstrate a process in person.
1
2
3
4
5
6
7
166
Not Representative
1
2
3
4
5
Very Representative
6
7
Q46. When our branch receives information from SunTrust Atlanta Region Management, face-to-face
meetings would enhance our understanding.
1
2
3
4
5
6
7
Q48. Employees at our branch are skilled at using email.
1
2
3
4
5
6
7
Q49. Employees at our branch are skilled at accessing the SunTrust intranet website.
1
2
3
4
5
Q50. Computers enhance the productivity of employees at our branch.
1
2
3
4
5
6
7
6
7
Q51. Our branch has access to a database of “Best Practices” for operations, marketing, etc. that we can
access on an as-needed basis.
1
2
3
4
5
6
7
Q52. Considering the transfer of knowledge among SunTrust branches, would you say that your branch:
Shares more knowledge than we receive
Receives more knowledge than we share
Shares and Receives knowledge equally
Q53. Considering the market that your branch serves, would you say that your market is more similar to or
more different than the markets served by other SunTrust branches:
More similar than different
More different than similar
Q54. Please indicate how you perceive the amount of information that you receive from each area:
Not Enough
Just Right
Operations
Marketing
Sales
167
Too Much
Q55. When I receive information from SunTrust Atlanta Regional Management:
 I generally don’t understand how it should be used in practice
 I generally understand how it should be used in practice
Q56. In the last twelve months, has your branch received knowledge of the following from SunTrust
Atlanta Regional Management or from another branch?:
 The Branch Operating Model
 Best Practices for prospecting business banking customers
 Other
 Other
Q57. In the last twelve months, has your branch shared knowledge of the following?:
 The Branch Operating Model
o With Personal Contacts at other SunTrust branches only
o With SunTrust Atlanta Regional Management
 Best Practices for prospecting business banking customers
o With Personal Contacts at other SunTrust branches only
o With SunTrust Atlanta Regional Management
 Other
o
o
With Personal Contacts at other SunTrust branches only
With SunTrust Atlanta Regional Management
 Other
o
o
With Personal Contacts at other SunTrust branches only
With SunTrust Atlanta Regional Management
Q58. What is the approximate size of your branch (in terms of loans outstanding)?
Under $5million
Over $5 million,
But under $20 million.
Over $20 Million,
But under $50 Million
Q59. Is your branch:
 An In Store branch
 A Retail branch
168
Over $50 Million
Q60. Our branch has been in operation for:




<1 year
over 1, but less than 3 years
over 3, but less than 7 years
over 7 years
Q61. I have been a branch manager for:




<1 year
over 1 year, but less than 3 years
over 3 years, but less than 7 years
over 7 years
Thank you for your time. Your input is greatly appreciated.
Please return your responses in the enclosed postage paid return envelope.
169
Jerry Ober
Boone County Community Credit Union
817 8Th St
Boone, IA 50036-50036
Dear Mr. Ober,
At Georgia State University, I am currently engaged in a research initiative examining the
factors which influence knowledge sharing among organizations embedded within a
network – such as the network of credit unions represented by CUNA. When completed,
the results of this research initiative will be used by individuals in the business
community, as well as by academics, to improve the flow of knowledge both within and
across firm boundaries.
I have contacted you, because as the manager of a Credit Union, your real world
experiences are extremely valuable and relevant to this research. Your responses will
remain completely confidential and anonymous.
I have enclosed a brief survey that should take less than 15 minutes to complete. After
you have completed all questions, please return it to me using the enclosed postage-paid
envelope.
If you would like a copy of the final results, and/or would like to discuss this survey or
your responses, please enclose your business card and I will contact you.
Your time is very much appreciated. In advance, thank you.
Jennifer Priestley, Ph.D. Candidate
www.dsc.gsu.edu/dscjlp
Georgia State University
35 Broad Street, Ste. 826
Atlanta, GA 30303
170
Thank you for agreeing to participate in this research initiative. Your real world input is
extremely valuable. This survey should take no more than 15 minutes to complete.
Please return your responses using the enclosed postage-paid return envelope. Your
responses will remain anonymous and will be completely confidential.
As a member of CUNA (the Credit Union National Association), your credit union may
receive, as well as provide, information on a variety of topics, including operations,
marketing, compliance, pending legislation, etc. You may also take advantage of other
opportunities to interact with CUNA and/or other credit union members of CUNA.
Considering your position as a credit union member of CUNA, which statement most
closely reflects your views:
 A. Our credit union is very integrated within CUNA. We place a high value on
our membership.
 B. Our credit union has limited interaction within CUNA. We don’t really
identify with the CUNA network.
If you checked “A”, please complete the Blue Survey. If you checked “B”, please
complete the Yellow Survey. Return only the completed portion using the enclosed
postage paid envelope.
This research study is neither sponsored nor funded by CUNA.
171
Please indicate how representative each statement is of your credit union.
Not Representative
1
2
3
4
5
Very Representative
6
7
Q1A. CUNA (or another credit union member of CUNA) has provided knowledge of processes (including
operations, marketing, risk or compliance, etc) to our credit union that has enabled us to improve our
performance.
1
2
3
4
5
6
7
Q2A. Our credit union has provided knowledge of processes (including operations, marketing, risk or
compliance, etc) to CUNA or to another credit union member of CUNA that enabled performance
improvement.
1
2
3
4
5
6
7
Q3A. Our credit union has received knowledge of a process (including operations, marketing, risk or
compliance, etc) that was developed at another credit union member of CUNA.
1
2
3
4
5
6
7
Q4A. Other credit unions within CUNA have received knowledge of a process (including operations,
marketing, risk or compliance, etc) that was developed at our credit union.
1
2
3
4
5
6
7
Q5A. Our credit union has trialed concepts that came from CUNA or from another credit union member
within CUNA.
1
2
3
4
5
6
7
Q6A. Our credit union has implemented concepts that came from CUNA or from another credit union
member within CUNA.
1
2
3
4
5
6
7
Q7A. Our credit union actively seeks out other credit union member of CUNA for collaboration.
1
2
3
4
5
6
7
Q8A. Other credit union members of CUNA look to our credit union for collaboration.
1
2
3
4
5
6
7
Q9A. When we communicate with CUNA what we say is generally understood.
1
2
3
4
5
6
7
Q10A. When we place a request for information to CUNA, we never have to re-state or explain our request.
1
2
3
4
5
6
7
172
Not Representative
1
2
3
4
5
Very Representative
6
7
Q11A. When we place a request for information to CUNA, the recipient who will respond to our request
generally “gets it right” the first time.
1
2
3
4
5
6
7
Q12A. Our credit union uses similar terminology to that used by other credit union members of CUNA to
describe issues, concepts, processes, etc.
1
2
3
4
5
6
7
Q13A. Other credit union members of CUNA generally use terms that our credit union understands.
1
2
3
4
5
6
7
Q14A. When our credit union receives documents or other communications from CUNA, we always
understand all terminology.
1
2
3
4
5
6
7
Q15A. We rarely require any terminology clarification when reading CUNA documents.
1
2
3
4
5
6
7
Q16A. Our day-to-day operations are similar to most other credit union members (of similar size) within
CUNA.
1
2
3
4
5
6
7
Q17A. Considering only the operations (ignoring culture, personalities, pay scale, etc), it would be easy for
an employee from our credit union to transition into another credit union within CUNA.
1
2
3
4
5
6
7
Q18A. Our credit union understands the basic operations of most other credit union members (of similar
size) within CUNA.
1
2
3
4
5
6
7
Q19A. The way that managers approach problems at our credit union is similar to the way that managers
approach problems at other credit unions within CUNA.
1
2
3
4
5
6
7
Q20A. Our credit union resolves issues in the same way that other credit unions within CUNA resolve
issues.
1
2
3
4
5
6
7
Q21A. The problem solving skills of employees at other credit union members of CUNA are similar to the
problem solving skills of employees at our credit union.
1
2
3
4
5
6
7
173
Not Representative
1
2
3
4
5
Very Representative
6
7
Q22A. All credit union members of CUNA have access to the same basic knowledge of the credit union
industry.
1
2
3
4
5
6
7
Q23A. CUNA makes the same basic knowledge available to all credit union members.
1
2
3
4
5
6
7
Q24A. Our credit union believes that managers at other credit union members of CUNA have the same
basic working knowledge of the industry.
1
2
3
4
5
6
7
Q25A. Employees of our credit union could create a specific list of the factors that do or do not contribute
to an increase in membership.
1
2
3
4
5
6
7
Q26A. Employees of our credit union could create a specific list of the factors that do or do not contribute
to the retention of good personnel.
1
2
3
4
5
6
7
Q27A. Employees of our credit union could create a specific list of the factors that do or do not contribute
to high CAMEL ratings.
1
2
3
4
5
6
7
Q28A. I believe that management of all credit union members of CUNA would probably agree about the
factors which create financial success at a credit union.
1
2
3
4
5
6
7
Q29A. Our credit union understands what makes our credit union a better/worse performer relative to other
credit union members of CUNA.
1
2
3
4
5
6
7
Q30A. Employees at our credit union could create a list of the factors that differentiate our credit union
from other credit unions.
1
2
3
4
5
6
7
Q31A. Employees at our credit union could create a list of the factors that do or do not contribute to the
financial successes of other credit unions.
Q32A. When CUNA or other credit unions share knowledge of processes or practices with our credit
union, they have a good idea of how we will use it.
1
2
3
4
5
6
7
174
Not Representative
1
2
3
4
5
Very Representative
6
7
Q33A.When we share knowledge of a process or practice with CUNA or with other credit union members
within CUNA, we generally understand how that knowledge will be used.
1
2
3
4
5
6
7
Q34A. When CUNA or other credit unions share knowledge of processes or practices with our credit
union, they already have experience with how that knowledge contributes to performance.
1
2
3
4
5
6
7
Q35A. Our credit union trusts other credit unions to deal with us fairly.
1
2
3
4
5
6
7
Q36A. We believe that other credit unions would take unfair advantage of knowledge shared by our credit
union.
1
2
3
4
5
6
7
Q37A. When credit union members of CUNA interact, there is an understanding that the interaction will be
based in fairness and trust.
1
2
3
4
5
6
7
Q38A. CUNA has developed a reputation for ensuring fairness and trust among its credit union members.
1
2
3
4
5
6
7
Q39A. Our credit union has policies (either formal or informal) to restrict the sharing of knowledge with
other credit unions.
1
2
3
4
5
6
7
Q40A. Other credit unions within CUNA are protective of their knowledge.
1
2
3
4
5
6
7
Q41A. We believe that other credit unions have policies (either formal or informal) to restrict the sharing of
knowledge with our credit union.
1
2
3
4
5
6
7
Q42A. Our credit union has had experience with another credit union member of CUNA refusing to share
knowledge.
1
2
3
4
5
6
7
Q43A. We believe that there are credit union members of CUNA who would refuse to share knowledge of
a process (including operations, marketing, risk or compliance, etc.) with other credit union members of
CUNA.
1
2
3
4
5
6
7
175
Not Representative
1
2
3
4
5
Very Representative
6
7
Q44A. Our credit union is in competition with other credit union members of CUNA.
1
2
3
4
5
6
7
Q45A. The majority of the knowledge that our credit union receives from CUNA is in the form of manuals
or guidelines or other similar documents.
1
2
3
4
5
6
7
Q46A. Representatives of CUNA or other credit unions rarely need to come to our credit union to discuss
face-to-face or demonstrate a process (including operations, marketing, risk or compliance).
1
2
3
4
5
6
7
Q47A.W hen we receive knowledge from CUNA or from other credit unions, face-to-face meetings would
enhance our understanding.
1
2
3
4
5
6
7
Q48A. Employees within our credit union are skilled at using email.
1
2
3
4
5
6
7
Q49A. Employees within our credit union are skilled at accessing the CUNA website.
1
2
3
4
5
6
7
Q50A. Computers enhance the productivity of employees at our credit union.
1
2
3
4
5
7
6
Q51A. Our credit union has, or has access to, a database of “Best Practices” for operations, marketing, risk,
compliance, etc. that we can access on an as-needed basis.
1
2
3
4
5
6
7
Q52A. Considering the transfer of knowledge among credit union members within CUNA, would you say
that your credit union:
Shares more knowledge than you receive
Receives more knowledge than you share
Shares and Receives knowledge equally
Q53A. What is the approximate size of your organization?
Under $10million
Over $10 million,
But under $50 million.
Over $50 Million,
But under $100 Million
176
Over $100 Million
Q54A. Approximately how long has your organization been a member of CUNA?
Less than 1 year
1-5 years
6-10 years
11+ years
Q55A. How would you characterize your credit union?
 Low Income Designation
 Community Chartered
 Company/Association Affiliated (including military, educators, etc.)
Q56A. Has your credit union received knowledge of the following topics in the last year from within
CUNA? (check all that apply):
 Providing services to small business (Please provide an example if possible)
 The requirements and opportunities related to the Patriot Act (Please provide an example if
possible)
 New mortgage product options related to the Freddie Mac/CUNA alliance (Please provide an
example if possible)
 Other – Please Explain
 Other – Please Explain
 Other – Please Explain
Q57A. Has your credit union shared knowledge of the following topics in the last year? (check all that
apply):
 Providing services to small business
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 The requirements and opportunities related to the Patriot Act
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 New mortgage product options related to the Freddie Mac/CUNA alliance
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 Other
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 Other
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 Other
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
177
58A.Does your credit union generally share knowledge through any of the following channels? (check all
that apply):






CUNA Councils
COBWEB
CU Exchange
National Credit Union Roundtable
Other
Our credit union does not generally share knowledge within CUNA
59A. What is your primary reason for participating within CUNA?
178
Please indicate how representative each statement is of your credit union.
Not Representative
1
2
3
4
5
Very Representative
6
7
Q1B. Another credit union has provided knowledge of processes (including operations, marketing, risk or
compliance, etc) to our credit union that has enabled us to improve our performance.
1
2
3
4
5
6
7
Q2B. Our credit union has provided knowledge of processes (including operations, marketing, risk or
compliance, etc) to another credit union that enabled performance improvement.
1
2
3
4
5
6
7
Q3B. Our credit union has received knowledge of a process (including operations, marketing, risk or
compliance, etc) that was developed at another credit union.
1
2
3
4
5
6
7
Q4B. Other credit unions have received knowledge of a process (including operations, marketing, risk or
compliance, etc) that was developed at our credit union.
1
2
3
4
5
6
7
Q5B. Our credit union has trialed concepts that came from another credit union.
1
2
3
4
5
6
7
Q6B. Our credit union has implemented concepts that came from another credit union.
1
2
3
4
5
6
7
Q7B. Our credit union actively seeks out other credit unions for collaboration.
1
2
3
4
5
6
7
Q8B. Other credit unions look to our credit union for collaboration.
1
2
3
4
5
6
7
Q9B. When we communicate with other credit unions, what we say is generally understood.
1
2
3
4
5
6
7
Q10B. When we place a request for information to another credit union, we never have to re-state or
explain our request.
1
2
3
4
5
6
7
Q11B. When we place a request for information to another credit union, the recipient who will respond to
our request generally “gets it right” the first time.
1
2
3
4
5
6
7
179
Not Representative
1
2
3
4
5
Very Representative
6
7
Q12B. Our credit union uses similar terminology to that used by other credit unions to describe issues,
concepts, processes, etc.
1
2
3
4
5
6
7
Q13B. Other credit unions generally use terms that our credit union understands.
1
2
3
4
5
6
7
Q14B. When our credit union receives documents or other communications from other credit unions, we
always understand all terminology.
1
2
3
4
5
6
7
Q15B. We rarely require any terminology clarification when reading documents that come from other
credit unions.
1
2
3
4
5
6
7
Q16B. Our day-to-day operations are similar to most other credit unions (of similar size).
1
2
3
4
5
6
7
Q17B. Considering only the operations (ignoring culture, personalities, pay scale, etc), it would be easy for
an employee from our credit union to transition into employment at another credit union.
1
2
3
4
5
6
7
Q18B. Our credit union understands the basic operations of most other credit unions.
1
2
3
4
5
6
7
Q19B. The way that managers approach problems at our credit union is similar to the way that managers
approach problems at other credit unions.
1
2
3
4
5
6
7
Q20B. Our credit union resolves issues in the same way that other credit unions resolve issues.
1
2
3
4
5
6
7
Q21B. The problem solving skills of employees at other credit unions are similar to the problem solving
skills of employees at our credit union.
1
2
3
4
5
6
7
Q22B. All credit unions have access to the same basic knowledge of the credit union industry.
1
2
3
4
5
6
7
180
Not Representative
1
2
3
4
5
Very Representative
6
7
Q23B. Our credit union believes that managers at other credit unions have the same basic working
knowledge of the industry.
1
2
3
4
5
6
7
Q24B. Employees of our credit union could create a specific list of the factors that do or do not contribute
to an increase in membership.
1
2
3
4
5
6
7
Q25B. Employees of our credit union could create a specific list of the factors that do or do not contribute
to the retention of good personnel.
1
2
3
4
5
6
7
Q26B. Employees of our credit union could create a specific list of the factors that do or do not contribute
to high CAMEL ratings.
1
2
3
4
5
6
7
Q27B. I believe that management of all credit union members of CUNA would probably agree about the
factors which create financial success at a credit union.
1
2
3
4
5
6
7
Q28B. Our credit union understands what makes our credit union a better/worse performer relative to other
credit unions.
1
2
3
4
5
6
7
Q29B. Employees at our credit union could create a list of the factors that differentiate our credit union
from other credit unions.
1
2
3
4
5
6
7
Q30B. Employees at our credit union could create a list of the factors that do or do not contribute to the
financial successes of other credit unions.
Q31B. When other organizations share knowledge of processes or practices with our credit union, they
have a good idea of how we will use it.
1
2
3
4
5
6
7
Q32B.When we share knowledge of a process or practice with other organizations, we generally
understand how that knowledge will be used.
1
2
3
4
5
6
7
181
Not Representative
1
2
3
4
5
Very Representative
6
7
Q33B. When other credit unions share knowledge of processes or practices with our credit union, they
already have experience with how that knowledge contributes to performance.
1
2
3
4
5
6
7
Q34B. Our credit union trusts other credit unions to deal with us fairly.
1
2
3
4
5
6
7
Q35B. We DO NOT believe that other organizations would take unfair advantage of any knowledge shared
by our credit union.
1
2
3
4
5
6
7
Q36B. When credit unions interact, there is an understanding that the interaction will be based in fairness
and trust.
1
2
3
4
5
6
7
Q37B. Our credit union has policies (either formal or informal) to restrict the sharing of knowledge with
other credit unions.
1
2
3
4
5
6
7
Q38B. Other credit unions are protective of their knowledge.
1
2
3
4
5
6
7
Q39B. We believe that other credit unions have policies (either formal or informal) to restrict the sharing of
knowledge with our credit union.
1
2
3
4
5
6
7
Q40B. Our credit union has had experience with another credit union refusing to share knowledge.
1
2
3
4
5
6
7
Q41B. We believe that there are credit unions that would refuse to share knowledge of a process (including
operations, marketing, risk or compliance, etc.) with other credit unions.
1
2
3
4
5
6
7
Q42B. Our credit union is in competition with other credit unions.
1
2
3
4
5
182
6
7
Not Representative
1
2
3
4
5
Very Representative
6
7
Q43B. The majority of the knowledge that our credit union receives from other organizations is in the form
of manuals or guidelines or other similar documents.
1
2
3
4
5
6
7
Q44B. Representatives of other credit unions or other organizations rarely need to come to our credit union
to discuss face-to-face or to demonstrate a process (including operations, marketing, risk or compliance).
1
2
3
4
5
6
7
Q45B. Employees within our credit union are skilled at using email.
1
2
3
4
5
6
7
Q46B. Employees within our credit union are skilled at accessing the internet.
1
2
3
4
5
6
7
Q47B. Computers enhance the productivity of employees at our credit union.
1
2
3
4
5
6
7
Q48B. Our credit union has, or has access to, a database of “Best Practices” for operations, marketing, risk
compliance, etc. that we can access on an as-needed basis.
1
2
3
4
5
6
7
Q49B. Considering the transfer of knowledge among credit unions, would you say that your credit union:
Shares more knowledge than you receive
Receives more knowledge than you share
Shares and Receives knowledge equally
Q50B. What is the approximate size of your organization?
Under $10million
Over $10 million,
But under $50 million.
Over $50 Million,
But under $100 Million
Over $100 Million
Q51B. Approximately how long has your organization been a member of CUNA?
Less than 1 year
1-5 years
6-10 years
183
11+ years
Q52B. How would you characterize your credit union?
 Low Income Designation
 Community Chartered
 Company/Association Affiliated (including military, educators, etc.)
Q53B. Has your credit union received knowledge of the following topics in the last year from within
CUNA? (check all that apply):
 Providing services to small business (Please provide an example if possible)
 The requirements and opportunities related to the Patriot Act (Please provide an example if
possible)
 New mortgage product options related to the Freddie Mac/CUNA alliance (Please provide an
example if possible)
 Other – Please Explain
 Other – Please Explain
 Other – Please Explain
Q54B. Has your credit union shared knowledge of the following topics in the last year? (check all that
apply):
 Providing services to small business
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 The requirements and opportunities related to the Patriot Act
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 New mortgage product options related to the Freddie Mac/CUNA alliance
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 Other
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 Other
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
 Other
o Shared only with personal contacts
o Shared with the larger CUNA network, through a listserv, a conference or other wide
spread communication channel
184
55B.Does your credit union generally share knowledge through any of the following channels? (check all
that apply):






CUNA Councils
COBWEB
CU Exchange
National Credit Union Roundtable
Other
Our credit union does not generally share knowledge within CUNA
56B. What is your primary reason for participating within CUNA?
185