Download - Eapen Krishnan Paper
-
8/13/2019 Eapen Krishnan Paper
1/51
0
Transferring tacit know-how through licensing or joint venture: Is
opportunism really redundant?
Alex EapenThe University of Sydney Business School
Discipline of International BusinessNSW 2006, Australia
Tel: + 61-2-903 66435Email:[email protected]
Rekha KrishnanSimon Fraser University
Department of International Business8888 University Drive
Burnaby, BC V5A 1S6, CanadaTel: +1-604-291-3047
Email:[email protected]
Version: Mar, 2012
We thank Aks Zaheer, Anand Nandkumar, Anita McGahan, Anoop Madhok, HarbirSingh, Kannan Srikanth, Klaus Meyer, Peter Liesch, Rajiv Krishnan, Sarah Kaplan,and seminar participants at the Indian School of Business and the University ofSydney, for helpful comments on earlier drafts of this paper.
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]:[email protected] -
8/13/2019 Eapen Krishnan Paper
2/51
1
Transferring tacit know-how through licensing or joint venture: Is opportunism
really redundant?
Abstract
Transaction cost economics scholars argue that firms exist due to failure of markets,
a main stimulus of which is opportunism. On the contrary, proponents of the
knowledge based view claim that the opportunism assumption is redundant in
explaining why firms exist. A key feature of this debate, however, is that despite
competing causal mechanisms, predictions of both theories are identical. We see this
lack of predictive uniqueness as the prime bottleneck in resolving the question of
whether opportunism matters, and contribute to the debate in two ways: (1) we
exploit the contingency logics inherent in knowledge based and transaction cost
approaches to build distinctive predictions from both theories, and (2) we
empirically test their competing logics, bringing empirical evidence to the debate
which has this far been largely on conceptual turf. Our results, based on an analysis
of tacit know-how transfer in strategic alliances between foreign and local firms in
India, and subject to certain caveats we discuss in the paper, indicate that
opportunism matters.
(165 words)
-
8/13/2019 Eapen Krishnan Paper
3/51
2
INTRODUCTION
A key debate in the theory of the firm is on the role of opportunism. Some scholars, primarily
from a transaction cost economics (TCE) standpoint, argue that firms exist due to market
failure, a main stimulus of which is opportunism (Hennart, 1982; McFetridge, 1995;
Williamson, 1985; 1991; 1993). Others maintain that the opportunism assumption is
redundant, that is, the existence of the firm can be explained without the notion of
opportunism (Conner & Prahalad, 1996; Ghoshal & Moran, 1996; Kogut & Zander, 1993;
Madhok, 2006). Yet another group suggests that opportunism may be relevant, but not to the
encompassing extent that TCE portrays (Madhok, 2006), nor always directed by guileful self-
interest seeking (Verbeke & Greidanus, 2009). Conceptual disagreement aside, empirical
evidence on whether opportunism is critical to explaining the boundaries of the firm is also
very scant. Even within the huge body of empirical research on TCE (David & Han, 2004;
Macher & Richman, 2008), scholars usually assume but omit testing the underlying
opportunism assumption (Tsang, 2006; Verbeke & Griedanus, 2009).
A specific context in which this debate has gained significant play in recent years is
the international strategy literature on the implications of ownership advantages of firms on
their boundaries. Examining how firms transfer their knowledge-based advantage into new
countries, Kogut & Zander (1993) suggest that firms, as social communities with superior
cooperation and coordination mechanisms, are simply better vehicles to transfer know-how.
Transferring knowledge across borders, particularly when it is tacit, is hence more likely to
occur within the firm than on the market. This is a key prediction of the knowledge-based
view (KBV) of the multinational enterprise. Transaction cost theorists agree that tacit
knowledge is more efficiently transferred within the firm, but emphasize that this is because
contracting for tacit knowledge on the market is rendered inefficient by opportunism-related
hurdles (McFetridge, 1995; Hennart, 1982). Within firms, on the other hand, members have
-
8/13/2019 Eapen Krishnan Paper
4/51
3
fewer incentives to behave opportunistically (Williamson, 1985). Thus, unlike KBV, the
transaction cost theory of the multinational enterprise turns on the notion of opportunism.
This debate in the international strategy literature has witnessed passionate arguments from all
sides (Kogut & Zander, 1993; McFetridge, 1995; Love, 1995; Verbeke, 2003) and little sign
of being settled (e.g., Fransson, Hakanson, & Liesch, 2011).
We seek to advance the literature surrounding the opportunism debate in both
conceptual and empirical ways. Our starting point is the observation that despite
fundamentally different views from KBV and TCE on the relevance of opportunism, both
sides arrive at the same prediction that tacit know-how is more efficiently transferred
internally than on the market. In other words, while the causal mechanisms differ, predictions
from transaction cost and knowledge-based views are identical. This lack of predictive
uniqueness poses a key conceptual bottleneck in this debate1
We seek to get around these conceptual and empirical issues in this paper.
Conceptually, we exploit the contingency logics inherent in KBV and TCE to build distinctive
predictions from both theories. Contingency propositions also better represent the causal
mechanisms underpinning both theories (cf. Rajan & Zingales, 1998). Our theoretical strategy
involves identifying situations where opportunism is likely to be a concern, and then
. A second critical gap, given that
the debate has been largely on conceptual turf (e.g., Conner & Prahalad, 1996; Foss, 1996a;
1996b; Kogut & Zander, 1992; Fransson et al., 2011), is the lack of empirical evidence on
whether or not opportunism matters. There is indeed abundant empirical research on KBV and
TCE, but as Verbeke & Greidanus (2009) rightly say, very few studies assess directly
whetheropportunism is actually instrumental [or not] to governance choices (p.1476). The
question of whether opportunism is critical to understanding the boundaries of the firm, yet, is
ultimately an empirical one (Tsang, 2006).
1 The predictive similarity of TCE and KBV in spite of their different theoretical mechanisms has beenhighlighted by scholars in other contexts as well (e.g., Poppo & Zenger, 1998; Sampson, 2004).
-
8/13/2019 Eapen Krishnan Paper
5/51
4
examining whether the likelihood of internal transfer of tacit know-how increases in those
circumstances. In other words, we test whether the potential for opportunism makes any
difference to how tacit know-how is transferred. We expect tacit know-how, in general, to be
transferred internally, but if the effect of tacitness on transfer through hierarchical means
increases with the level of opportunism faced by transferor and recipient, this could be the
smoking gun to suggest that opportunism matters. On the contrary, if the effect of tacitness
on the mode of transfer is insensitive to the potential for opportunism, then perhaps as KBV
suggests, opportunism is not so crucial for the theory. So the difference in the magnitude of
the effect of tacitness under conditions of low and high opportunism serves as our litmus test
for whether opportunism matters. To obtain a complete picture, we also frame a contingency
argument for KBV. We argue that while the likelihood of internal transfer increases with the
tacitness of know-how, this effect is weaker when the transferor and unaffiliated recipient
have similar routines. That is, when routines are similar, transferring tacit know-how should
be easier, even to unaffiliated firms. Any evidence for this would corroborate the KBV view
that ease of transfer (and not opportunism) is what drives firms proclivity to internal transfer.
Empirically, we test these notions using unique data from a survey of technology
transfer arrangements between foreign and domestic firms in India. We surveyed licensing
contracts, which are closer to a market transaction, as well as joint ventures, which have
overtones of hierarchy or internal transfer (Gulati & Singh, 1998; Hennart, 1988; Osborn &
Baughn, 1990; Oxley, 1997; Phene & Tallman, 2012). We collected detailed micro-level data
on the nature of know-how transferred such as its tacitness, importance to the recipient, and
the extent of specific investments required to be made by the recipient for its implementation.
As we argue below, the latter two variables are reasonably good indicators of potential
opportunism. We also measured the similarity between transferor and recipient in this case,
between foreign and domestic partners in their routines and culture. Employing logistic
-
8/13/2019 Eapen Krishnan Paper
6/51
5
regression analyses, we find that tacit know-how is more likely transferred through joint
venture than through licensing. This is consistent with prior literature. However, we also find
that the effect of tacitness on the choice of joint venture is stronger under conditions of high
potential opportunism, i.e., when the know-how being transferred is central to the recipient, or
when the recipient needs to make specific investments. But we do not find the effect of
tacitness varying with whether or not the recipient has routines similar to that of the
transferor. Subject to certain caveats we discuss later on, we interpret these results to mean
two things firstly, that opportunism is relevant to understanding the boundaries of the firm,
and secondly, that opportunism is the likely causal mechanism behind the often observed
empirical link between tacitness of know-how and its internal transfer. These conceptual and
empirical observations, as we discuss later on in the paper, have fairly significant implications
for the debate over opportunism and the theory of the firm, and for issues related to the
question of what firms really do (cf. Kogut & Zander, 1992).
BACKGROUND
The assumption of opportunism has received a significant amount of attention and
commentary in the organization, strategy, and international business literatures. Not all of it
has been positive. While there is seldom any disagreement over the need to carefully identify
the nature of human beings that underlies their behavior (Simon, 1985; Verbeke & Greidanus,
2009; Williamson, 1993), scholars have attacked opportunism, inter alia, as being an under-
socialized and narrow view of human behavior (Donaldson, 1990; Granovetter, 1985), of little
normative value (Ghoshal & Moran, 1996), and irrelevant to understanding why firms exist
(Conner & Prahalad, 1996; Hodgson, 1998; Kogut & Zander, 1992; 1996; Love, 1995). This
critical commentary has been reciprocated in equal measure with extensions (Rugman &
Verbeke, 2003), elaboration (Wathne & Heide, 2000), and reconceptualization of the
opportunism construct (e.g., Verbeke & Greidanus, 2009).
-
8/13/2019 Eapen Krishnan Paper
7/51
6
In the international strategy literature, the study of how tacitness of know-how
influences the boundaries of the MNE has turned out to be a duelling ground for this debate.
By questioning the need for the opportunism assumption, Kogut and Zander (1993) reshaped
the way scholars understood the MNE and its boundaries (Tallman, 2003; Verbeke, 2003).
Until Kogut and Zander (1993), the existence of the MNE was largely attributed to the failure
of markets for critical intermediates such as knowledge, raw materials, and distribution
(Hennart, 2001). According to proponents of this approach, whether the cross-border transfer
of a firm's ownership advantage, e.g., its proprietary knowledge, occurs on the market or
through hierarchical modes such as joint ventures or wholly-owned subsidiaries depends on
whether or not there is a viable market for that advantage (Buckley & Casson, 1976; Hennart,
1982; Rugman, 1981). In some instances, markets fail for reasons related to opportunism
(Williamson, 1985), necessitating the internalization of that transfer. Kogut and Zander (1993)
took issue with this explanation and argued that it is overdetermined with its reliance on
opportunism. In building their case, they introduced the concept of tacitness of knowledge to
the literature which until then considered knowledge as a public good that is easy to transfer
but hard to protect. By bringing tacitness of knowledge into the picture, they established that
the transfer of tacit knowledge, due to the difficulty in describing and teaching it, may not be
easy even among willing parties. However, because routines, values, and assumptions are
shared within the firm, it turns out less costly to transfer tacit knowledge internally. Hence,
according to Kogut and Zander (1993), the existence of the MNE, or more precisely, the
transfer of tacit knowledge across borders through hierarchical modes can be attributed
simply to the efficiency of firms in transferring knowledge internally. The notion of
opportunism is redundant.
Kogut and Zanders thesis received a lot of support (e.g. Love, 1995; Tallman, 2003),
but also provoked a series of critiques (Foss, 1996a; 1996b; McFetridge, 1995), rebuttals
-
8/13/2019 Eapen Krishnan Paper
8/51
7
(Kogut & Zander, 1995), and reflections (Kogut & Zander, 2003). Like Kogut and Zander,
Love (1995) too asserts that opportunism is not necessary to explain the existence of the firm.
But unlike them, Love (1995) maintains that firms exist because of market failure, but a
different class of it that comes not from opportunism but from genuine unbridgeable
differences in routines. Madhoks (1996) argument is similar and largely consistent with
Kogut & Zanders. From a transaction cost perspective, on the other hand, McFetridge (1995)
argued that the reason firms transfer tacit knowledge internally is because the potential for
opportunism is subdued when transactions are organized through hierarchical modes. Kogut
and Zanders finding that MNEs transfer tacit knowledge internally rather than externally,
thus, is entirely consistent with transaction cost reasoning that is based on opportunism
(McFetridge, 1995). Several other scholars too have weighed in on this issue of whether
opportunism matters (e.g., Foss, 1996a; 1996b; Ghoshal & Moran, 1996), and the debate
continues (e.g., Fransson et al., 2011), but largely on conceptual terrain. Empirical evidence
for whether or not opportunism matters is rare, not only in the TCE literature that rests on the
notion of opportunism (Tsang, 2006), but also in the KBV literature that refutes it.
THEORY
Theoretical Strategy
As we have indicated, a key bottleneck in the opportunism debate as it has played out in the
international strategy literature is that predictions from TCE and KBV regarding the transfer
of tacit know-how are identical. Hence, what would help make progress, in our view, is a way
to draw and empirically test distinctive predictions based on the 'with opportunism' and
'without opportunism' logics of both theories. As we demonstrate below, this can be done by
means of contingency arguments. Furthermore, contingency arguments are native to the key
logics of both KBV and TCE, so portraying them as such is really just more precise
articulation.
-
8/13/2019 Eapen Krishnan Paper
9/51
8
Take the KBV for example. Kogut and Zander (1993) argue that firms are social
communities that specialize in the creation and internal transfer of knowledge. Firms are
simply better at transferring tacit knowledge. And this is largely because they are a
...repository of social knowledge that structures cooperative action... (p.627). In other
words, firms naturally beget the cooperation and coordination necessary to transfer tacit
knowledge. The reverse argument then is also true; the key hurdle in transferring knowledge
to unaffiliated firms is that transferor and recipient lack a common social fabric. When
unaffiliated firms lack the same routines, structures, and identity as the transferor, extra effort
needs to go into teaching the recipient, and transfer becomes costly. Herein lies the key
contingency in the knowledge based view. Tacitness of know-how, per se, is not the real
barrier to its transfer. Tacit knowledge can be transferred (it is transferred within the firm,
after all), but less easily so, when the recipient lacks similar routines. Kogut and Zander do
emphasize that differential routines between transferor and recipient are what matter (p.630),
a point that Love (1995) too nicely summarizes: ...the boundaries of the firm are
determined...by differential 'embedded capabilities' that exist between creators of knowledge
and its users (p.400). By embedded capabilities, Love essentially means routines. Although
there is a bit of confusion over which side of the debate Teece (1986) stands (cf., Love, 1995:
402), his statement that trying to transact tacit knowledge may ...fail if it must take place
between unaffiliated enterprises with different internal cultures (and) codes of
communication... (Teece, 1986: 29) also attests to the point we are making. Transferring tacit
knowledge is problematic only to the extent that the recipient lacks the appropriate routines.
This contingency argument is inherent in the knowledge based view of the firm.
Let us move on to TCE. An important rebuttal from TCE to Kogut and Zander's
thesis is that the relationship between tacitness and internal transfer can easily be explained by
transaction cost theory (McFetridge, 1995). Firms are indeed relatively efficient in
-
8/13/2019 Eapen Krishnan Paper
10/51
9
transferring tacit knowledge but this is because the market for tacit knowledge fails. And
according to TCE theorists, opportunism is what drives this failure. The key implication of
this argument is that contracting for tacit knowledge on the market becomes a problem only to
the extent that conditions of opportunism exist. One such condition, for example, is when
transaction specific investments are needed. Transaction specific investments and the resulting
lock-in problems are a breeding ground for opportunism related contractual problems
(Williamson, 1985). So tacitness could pose a problem to contractual exchange, but more so
in the presence of transaction specific investments. McFetridge (1995: 410) clarifies this and
says: if there were no transaction-specific investments and hence no lock-in, (tacitness) need
not burden market and unified governance differentially. It is also widely acknowledged in
the TCE literature that uncertainty coupled with asset specificity begets contractual problems
(Leiblein & Miller, 2003; Williamson, 1985). In our context of know-how transfer,
uncertainty arises from the difficulty in defining and specifying tacit knowledge. And
whenever assets are specific in nontrivial degree, [an increased] degree of uncertainty makes
it more imperative that the parties devise a machinery to 'work things out' (Williamson,
1985: 60). In other words, the combination of uncertainty and asset specificity is what
necessitates hierarchical solutions. So the kernel of transaction cost theory too is a
contingency argument. Tacitness matters only to the extent that transferor and recipient face
circumstances that beget opportunism concerns.
Phrasing our KBV and TCE predictions as contingency statements, thus, is consistent
with the conceptual core of both theories. There are also empirical benefits of employing
contingency arguments in the KBV-TCE debate. They bring sharper focus to the terms of
disagreement by better capturing the underlying causal mechanisms of both theories (cf.
Rajan & Zingales, 1998), and in turn, allow us to test the competing logics. The first set of
contingency arguments we propose below are on how the effect of tacitness varies with the
-
8/13/2019 Eapen Krishnan Paper
11/51
10
potential for opportunism, and the second, on how it varies with similarity of routines and
culture between transferor and recipient.
Hypotheses Development
As a baseline, and as TCE and KBV scholars in the international strategy field have
suggested, we maintain that tacit know-how is more likely to be transferred through
hierarchical modes than on the market (Kogut & Zander, 1993; McFetridge, 1995). So our
baseline hypothesis is that the more tacit the know-how, the more likely it will be transferred
through hierarchical modes than on the market. However, in order to accurately portray and
test TCE and KBV predictions relating to how tacit know-how is transferred, in the following
sections we frame propositions as contingency statements.
Tacitness, opportunism, and internal transfer. Contracting on the market works well
when the buyer and seller have a fairly good understanding of what is being exchanged
(Arrow, 1962; Barzel, 1989; Mayer & Nickerson, 2005). And for this reason, transacting any
kind of knowledge on the market is inherently difficult because the buyer does not completely
know what he is buying (Arrow, 1962; Hennart, 1982) and the seller cannot describe it to the
buyer because in doing so he reveals it free of charge. This problem only gets compounded
when the knowledge is tacit as well. This information asymmetry between buyer and seller is
the first contributor to market failure for knowledge. The second, is opportunism. When it is
difficult to clearly specify the details of the knowledge being transferred, the seller has
sufficient room to behave opportunistically. And so, the buyer maybe concerned, ex-ante, that
the seller is exploiting the information asymmetry and quoting a higher price, or that, ex-post,
the seller will not deliver what was promised. With room for such opportunistic manoeuvres,
it is unlikely that the buyer will consent to a market-based contractual transfer of the
knowledge. The safeguards and opportunism-mitigation inherent in hierarchy (Hennart, 1988;
Oxley, 1997; Sampson, 2004) become necessary. The seller too may have similar opportunism
-
8/13/2019 Eapen Krishnan Paper
12/51
11
concerns, but mainly around whether or not he/she will be able to fully appropriate rents on
the knowledge (Oxley, 1997). If the buyer uses the knowledge ex-post in ways not covered by
the contract, the seller may lose access to rents that are legally due to him/her. Tacit
knowledge, by definition, is not patentable, and so the ability to prevent such
misappropriation is weak. The market for tacit know-how thus fails under conditions of
opportunism. And given the opportunism-mitigating properties of hierarchy (Hennart, 1988;
Oxley, 1997; Sampson, 2004), such know-how is likely to be transferred through hierarchical
modes.
However, the threat of opportunism is not uniform across all transactions. Some
conditions beget greater opportunism concerns than others. We focus on two such conditions
when the know-how is of central importance to the recipient, and when it requires specific
investments by the recipient.
When the know-how being transferred is central to the recipient, this raises the
sceptre of opportunism for both the transferor and recipient. Recall that from the recipient's
perspective, the key concern is that the transferor may misrepresent the value of the know-
how, or not deliver as promised. These are always concerns for the recipient when contracting
for know-how, but arguably, more of a concern when the know-how being purchased is
critical to its operations. The stakes are higher since the recipient has more to lose from the
poor performance of know-how related to its core operations. Consequently, he/she will be
particularly wary about the transferor being opportunistic and misrepresenting the value or
defaulting on the delivery of the know-how. From the transferor's side too, the stakes are high.
Everything else constant, the potential for the recipient to misappropriate know-how that is
central to its operations is higher because the recipient may have better ability and incentive to
do so. It is likely that recipients are more capable of misusing technologies in the areas of
operation they are most familiar with. And the gains from doing so are higher. Thus from both
-
8/13/2019 Eapen Krishnan Paper
13/51
12
the recipient's and transferor's perspectives, concerns over opportunism are likely to be higher
when the know-how is central to the recipient.
Putting these arguments together, it becomes clear that transferring tacit know-how
on the market is likely to fail, but more so when it is central to the recipient. In such cases,
both parties are likely to take recourse to transfer through more hierarchical modes, where the
incentives to act opportunistically are drastically reduced (Hennart, 1993; McFetridge, 1995;
Williamson, 1985). It is perhaps useful to reiterate our key point that from the TCE point of
view, the market failure for tacit know-how, and the need to replace markets with internal
transfer, is not due to tacitness per se, but because of associated opportunism concerns. By
specifying the following contingency hypothesis we directly capture this underlying causal
mechanism.
Hypothesis 1: Tacit know-how is more likely to be transferred through hierarchical modes
(e.g., through joint ventures) than on the market (e.g., through licensing), but more so when
the know-how is central to the recipient's operations.
As we briefly alluded to earlier, another situation likely to give rise to opportunism
concerns is when the recipient is required to make specific investments to implement the
know-how (Shelanski & Klein, 1995). The problems associated with asset specificity and how
it yields room for opportunism concerns have received a lot of commentary in the TCE
literature (for the sake of brevity we do not repeat them here). And as mentioned earlier, the
combination of asset specificity with uncertainty raises significant transaction costs
(Williamson, 1985).
Again, as discussed above, a key concern for the recipient in attempting to contract
for tacit know-how is that the transferor may change terms of the contract ex-post and not
deliver as promised. This is always a concern but a greater one when the recipient has
invested heavily in specific investments. The cost to the recipient of the transferor
opportunistically reneging on the contract is immensely higher. Thus:
-
8/13/2019 Eapen Krishnan Paper
14/51
13
Hypothesis 2: Tacit know-how is more likely to be transferred through hierarchical modes
(e.g., through joint ventures) than on the market (e.g., through licensing), but more so when
the recipient needs to make specific investments to implement the know-how.
Tacitness, similarity of routines and culture, and internal transfer. The knowledge-
based view of the MNE contends that it is not opportunism but the difficulty in effectively
coordinating with third parties that prevents firms from transferring tacit knowledge on the
market. If that is indeed the case, then firms should be more likely to transfer tacit knowledge
externally when routines and organizational culture of the recipient are similar to their own 2
Tacit knowledge is not easy to describe and makes systematic transfer cumbersome
(Nonaka, 1994). Codifying tacit knowledge in order to make it teachable requires extensive
communication and coordination between source and recipient. In the context of knowledge
transfer, prior research suggests that similarity in management styles between the source and
recipient results in similarity of work expectations, operating procedures and interpretation of
strategic issues (Olk, 1997; Park & Ungson, 1997; 2001; Simonin, 1999). Compared to their
culturally distant counterparts, organizations that share similar expectations and
.
Culture is typically considered a set of cognitions shared by members of a social unit [such
as an organization] (OReilly et al., 1991:491; Scott, 2001; Smircich, 1983). Such shared
cognition, which involves translating events consistently and developing shared meanings and
conceptual schemes (Daft & Weick, 1984; Klimoski & Mohammed, 1994), guide the behavior
of organizational members (Rousseau, 1990; Schein, 1985) and allows them to develop
common interpretations (Klimoski & Mohammed, 1994). Scholars argue that shared mental
models held by organizational members result in effective coordination (Klimoski &
Mohammed, 1994; Srikanth & Puranam, 2011).
2The need for articulating the knowledge based view as a contingency argument is evident in commentary by
Fransson et al. (2011: 430) that ...it cannot be assumed that firms are always better attransferring...knowledge in house than through market transactions,...there may be conditions, as yet
unspecified, when this might be the case. The only disagreement we have is that the conditions are notunspecified. The key condition under which firms are better at transferring knowledge, as we argue, is when
potential recipients have a culture and routines dissimilar routines to those of the transferor.
-
8/13/2019 Eapen Krishnan Paper
15/51
14
interpretations need to expend less effort and energy in devising compatible inter-
organizational routines. This in turn facilitates better communication between them (Park &
Ungson, 1997; 2001) and reduces knowledge exchange costs (Jensen & Szulanski, 2004).
Therefore, when an unaffiliated knowledge recipient shares similar values and cognitions with
the transferor, that similarity facilitates better communication and effective coordination. In
turn, this should encourage the transfer of tacit knowledge on the market rather than through a
more hierarchical mode, such as joint venture.
Second, because tacit knowledge is firmly embedded in context, it is important to
understand the context in which knowledge is stored and retrieved (Nonaka, 1994). In
knowledge-exchange, scholars have argued that an organizations culture represents its history
and is therefore considered one of the facilities where the organizations memory is retained
for future reference (Klimoski & Mohammed, 1994; Walsh & Ungson, 1991). Thus, an
organizations culture, besides providing a framework for information processing, enables it
to store knowledge (Brief & Downey, 1983; Busenitz & Lau, 1996; Gray, Bougen, &
Donnellon, 1985; Harris, 1994; Shaw, 1990). Accordingly, organizations that are culturally
similar tend to organize and store knowledge in a mutually understandable manner (Schneider
& Angelmar, 1993) as well as retrieve knowledge from their memory in similar ways
(Klimoski & Mohammed, 1994). Their framework for decision-making, problem-solving,
information and knowledge processing tends to be comparable as well (Mitchell et al., 2000).
Thus, a recipient that is similar to the transferor in terms of organizational culture is more
likely to understand the context in which the transferors knowledge is embedded, which
makes the transfer of tacit knowledge easier.
Taken together, these arguments suggest that the relative efficiency of internal
transfer of tacit knowledge reduces when the knowledge recipient shares the same (or similar)
routine and culture as the transferor. Hence:
-
8/13/2019 Eapen Krishnan Paper
16/51
15
Hypothesis 3: Tacit know-how is more likely to be transferred through hierarchical modes
(e.g., through joint ventures) than on the market (e.g., through licensing), but less so when the
transferor and recipient are similar in their organizational routines and culture.
METHODS
Empirical Context
We test these hypotheses using data from a survey of technology recipients in joint
ventures and technology licensing arrangements between foreign and domestic firms in India.
Our choice of the survey method and the target population of licensing and joint ventures (as
opposed to, say, wholly-owned subsidiaries), as well as our decision to survey the recipient
and not the transferor were deliberate. Scholars in the past have relied on secondary data to
study licensing and joint ventures (Gulati, 1995; Oxley, 1997). But with secondary data, as
those very scholars admit, ...it is very difficult to obtain data on aspects of technology
transferred...such as tacitness or age... (Oxley, 1997: 394). Hence, to get at the attributes and
details of the technology transferred, a survey is more appropriate. Our decision to survey the
technology recipient is rather unique but necessary in our case for two reasons. First, some of
the moderating variables we propose in our hypotheses, e.g., centrality of know-how to the
recipient and the extent of specific investments it needed to make, are obviously best
measured by surveying the recipient. Secondly, and perhaps more importantly, it seems to us
appropriate to measure tacitness of the technology being transferred through the eyes of the
recipient. Whether a technology is perceived as tacit or not, depends on the experience and
repertoire of capabilities of the recipient. What might seem simple or codifiable technology to
the transferor may appear complex and unclear to a recipient who lacks the relevant
experience or capabilities. So the recipient's perception of tacitness is at least equally (and
perhaps even more) relevant to the choice of mode of transfer. Surveying the recipient
automatically precludes foreign wholly-owned subsidiaries from our sampling frame because
in wholly-owned operations there is no unaffiliated recipient. Our empirical focus is thus on
-
8/13/2019 Eapen Krishnan Paper
17/51
16
licensing and joint ventures and on the choice between them, considering licensing closer to a
market contract and joint ventures closer to a hierarchical solution to knowledge transfer
(Gulati & Singh, 1998; Hennart, 1988; Oxley, 1997; Phene & Tallman, 2012).
Surveying only licensing and joint ventures, if anything, makes our empirical tests
more conservative. Compared to the differences between licensing and wholly-owned
subsidiaries, the differences between licensing and joint venture are less distinctive. They do
for sure differ in transaction cost mitigating and knowledge transfer properties (Oxley, 1997),
but these differences are not as stark as between licensing and wholly-owned subsidiaries.
Also, given that most tacit know-how is perhaps transferred through wholly-owned
subsidiaries, which we do not observe, the variation in tacitness in our dataset is likely to be
limited. All these factors militate against us in finding any significant results. Focussing on
the choice between licensing and joint ventures thus only raises the bar for us, making it even
harder to find any significant results.
Data
We used Capitaline, a secondary database, and member lists of various international chambers
of commerce in India to identify a sample of 700 international licensing and joint venture
agreements operating in India. Following Parkhe (1993) and Simonin (1999), we aimed to
target respondents highly knowledgeable about the relationships. In our case these were
mainly managing directors and chief executive officers. We obtained the names of these target
respondents from Capitaline and chambers of commerce data.
We designed our questionnaire and implemented our survey according to Dillmans
(2000) 'Tailored Design Method', which suggests several ways to encourage response. Our
measurement items were generated through a review of prior literature. We used university
faculty and doctoral students to assess the content of the items and to ascertain whether the
items tapped into the conceptual domain of the focal construct (DeVellis, 1991). This yielded
-
8/13/2019 Eapen Krishnan Paper
18/51
17
a set of fine-tuned questionnaire items that were used in personal interviews and early pre-
tests with managing directors of Indian firms involved in international licensing and joint
venture agreements. These steps allowed us to check for any item ambiguity.
The first wave of questionnaires was sent to managing directors and senior
executives of the 700 Indian firms we identified that had licensing and joint venture
agreements with foreign firms. This was followed, four weeks later, by a second wave of
survey mailings. Of the 700 managing directors and senior executives that received
questionnaires, 126 responded, yielding an 18% response rate. This is comparable to surveys
of such relationships in other emerging economies: e.g., 14.4% for China (Isobe, Makino &
Montgomery, 2000), and 19% for Mexico (Robins, Tallman & Lindquist, 2002). All the
responses to our survey came from individuals directly responsible for the relationships: 80
came from chairpersons and managing directors of the alliances, 30 from presidents, vice
presidents and general managers, and 16 from full-time directors. Nearly 75% of the
respondents had been with the firm for more than five years, and of these almost 25% for
more than 20 years. The foreign partners of Indian firms we surveyed were spread over 21
countries. All relationships in our sample are dyadic, and belong to industries in the
manufacturing sector where licensing and joint ventures are more prevalent (e.g., Parkhe,
1993; Simonin, 1999). Tests of proportions show that the distribution of our responses
according to their two-digit manufacturing SIC is not statistically different from that reported
in two landmark studies by Harrigan (1988) and Ghemawat, Porter & Rawlison (1986). The
top four industries in our sample rank in the same order as in these studies, and as others have
also found (e.g. Parkhe, 1993), are relatively high-tech (industrial machinery and equipment,
chemicals and allied products, electrical and electronic equipment, and transportation
equipment).
Some of our key measures were collected using the same survey instrument and from
-
8/13/2019 Eapen Krishnan Paper
19/51
18
a single respondent. Hence we undertook multiple procedural and statistical remedies to
address the potential concern of common method bias and single informant bias. Specifically,
we undertook procedural remedies such as protecting respondent anonymity, scale reordering,
and reducing item ambiguity, as well as statistical remedies like Harmans (1967) one factor
test. Based on these, we are confident that common method or single respondent bias is not a
serious problem in our study.
We checked the potential for non-response bias by comparing the characteristics of
our respondents to those of the targeted population sample. T-tests for the size of the firms (p
= 0.284) and age of the Indian firm (p = 0.344) revealed no significant differences between
respondent and non-respondent groups. In line with Mohr & Spekman (1994) and Poppo &
Zenger (2002), we also tested for non-response bias by comparing early and late respondents.
Armstrong & Overton (1977) argue that late respondents are more representative of non-
respondents. We found no significant difference between early and late respondents on
characteristics such as number of employees of the Indian partner (p = 0.18) and relationship
duration (p = 0.29).
Dependent Variable
We used a dummy variable to represent the mode of know-how transfer. This variable, JV,
took the value 1 when know-how was transferred via joint venture, and 0 when it was through
a licensing agreement. We carefully considered using on a more nuanced version of this
variable based on the fact that an ideal test of transaction cost theory requires a dependent
variable that captures finer aspects of safeguards built into the contract (Malhotra &
Lumineau, 2011; Mesquita & Brush, 2008). Hierarchy is not the only solution to transaction
costs; other safeguards such as contracts of longer duration (Crocker & Masten, 1988;
Joskow, 1985; 1987), exclusivity (Gallick, 1984), or even relational assets (Madhok, 2006)
could meet the same opportunism-mitigation purposes. So ideally, the extent of contractual
-
8/13/2019 Eapen Krishnan Paper
20/51
19
safeguards, irrespective of it source, should be our dependent variable.
However, we stuck with the more conventional market hierarchy dichotomy measure
for three reasons. Firstly, we needed our measure to be relevant to the debate we are
contributing to. In other words, we needed a dependent variable that would be flexible enough
to reflect both KBV and TCE mechanisms and thus appeal to both camps. While finer details
of the contract would help in better capturing opportunism-mitigation elements of the
governance structure, it is unclear how it would better represent the commonality of social
fabric that, according to KBV, eases the cost of transfer. Secondly, the debate at the center of
our work has primarily been around why firms are more efficient carriers of tacit know-how,
or alternatively, why tacit know-how is transferred internally as opposed to on the market. For
the sake of relevance to this question, we persisted with our dichotomous variable; it picks up
whether the know-how transfer was within-firm or to an unaffiliated party. Thirdly, despite
recent advances, the terrain of contract research where scholars micro-analyze contracts to
capture their finer transaction cost mitigation characteristics still has teething problems.
These include issues like which clauses of the contract to look at, whether sets of clauses still
serve opportunism mitigation when they interact with each other, and indeed the definition of
a contractual safeguard (Mayer, 2009). As a simple example, while Mesquita & Brush (2008)
measure contract completeness with the number of contingency clauses included in the
contract, Crocker & Reynolds (1993) consider those with fewer contingency clauses more
complete because such contracts potentially cover broader areas of dispute. As TCE scholars
duly acknowledge, contract research is an area where the enterprise as a whole needs to
improve (Mayer, 2009). Without specific guidance from the literature on specific contract-
based measures, and not to distract from the main goal of the paper, we decided to stick with
our parsimonious measure. Our measure is not without precedent, though. Several other
scholars both from KBV and TCE approaches have dichotomized the governance choice
-
8/13/2019 Eapen Krishnan Paper
21/51
20
into market or hierarchy (Colombo, 2003; Kogut & Zander, 1993; Mayer & Salomon, 2006;
Nickerson & Silverman, 2003; Osborn & Baughn, 1990; Phene & Tallman, 2012; Sampson,
2004)
Independent Variables
Our main independent variable is tacitness, and our moderator variables fall under two
groups: the potential for opportunism and the commonality of social fabric across transferor
and recipient.As we explain below, we conceptualized tacitness as tacitness of the know-how
packagebeing transferred. We also used two indicators of potential opportunism - centrality
of know-how to the recipient(for hypothesis 1) and size of specific investment(for hypothesis
2). We captured the commonality in social fabric across transferor and recipient using a
variable similarity of routines and culture (hypothesis 3).
Tacitness. In measuring tacitness, we focussed on the tacitness of the know-how
package being transferred. The know-how package refers to the entire package being
transferred, and includes not only the core technology, but also the surrounding supporting
skills. The know-how package thus varies with the range of supporting skills needed from the
foreign partner by the recipient. Capable recipients need fewer skills and vice-versa. So to
measure tacitness of the know-how package, we used a five-item measure that captured the
recipients need for supporting skills to implement the core technology. Specifically, based on
prior conceptual descriptions (Baranson, 1969; Hall and Johnson, 1970; Odagiri, et al., 2010),
we operationalized this variable by asking recipients whether or not the marketing and
production skills they possessed matched those required to produce and sell the product for
which know-how is being sourced, and whether they required production process, marketing,
management, and other technical assistance to implement the technology contributed by the
transferor. The key insight underpinning this measure is that as the need for tacit skills such as
marketing and management increases, so does the overall tacitness of the know-how package.
-
8/13/2019 Eapen Krishnan Paper
22/51
21
This measure had satisfactory reliability. An important consideration when using
binary items in a scale, as we do, is that the traditional Cronbach alpha can be negatively
biased; an ordinal alpha, based on tetrachoric or polychoric correlations, is more appropriate
(Zumbo, Gadermann, & Zeisser, 2007). For our five items, the ordinal alpha was 0.68. We
also conducted factor analysis to assess construct validity. Following an orthogonal rotation,
we extracted one factor that had an eigenvalue greater than 1 and accounted for 85% of total
variance. The last three out of the five items loaded well onto this one factor with loadings
greater than 0.65 and the 0.40 cut-off normally prescribed (Ford, MacCallum, & Tait, 1986).
For our main analyses we therefore chose to retain only the last three items in the scale.
Redoing factor analysis with only the last three items returned factor loadings that were
higher than 0.71. Also, the ordinal reliability alpha improved significantly from 0.68 to 0.79.
To be sure though, we ran our regressions with both the three-item and the five-item index
and obtained substantially similar results.
Moderators. In capturing the potential for opportunism, we resorted to unobtrusive
measures (Webb & Weick, 1979). As research on personality traits has shown, using direct
survey questions to capture negative constructs such as narcissm (Chatterjee & Hambrick,
2007), racism (Newman & Krzystofiak, 1979), or in our context, opportunism, is likely to
elicit low response rates and social desirability bias. Hence, rather than trying to measure
opportunism itself as some scholars have done (Anderson, 1988; Sako & Helper, 1998), we
used unobtrusive indicators of opportunism (Gulati & Singh, 1998; Klein et al., 1978; Oxley,
1997).
We measured centrality of know-how to the recipient using a dummy variable. We
asked the respondent who in our case was also the recipient whether the know-how was
central to its operations, and coded the variable 1 if the answer was in the affirmative, and 0
otherwise.
-
8/13/2019 Eapen Krishnan Paper
23/51
22
To measure size of specific investment, we asked the recipient to indicate the size of
investment that had to be made specifically to implement the know-how, and was less relevant
to other uses. This variable was coded in five bands: 1 for specific investments of less than
US$10 million, 2 for investments between US$ 10 and 45 million, 3 for investments between
US$45 and 110 million, 4 for investments between US$110 and 220 million, and 5 for those
above US$ 220 million. Our contention is that higher values on this variable indicate greater
levels of asset specificity (Dyer, 1996). This measure meets the three guidelines Mayer (2009)
offers to increase construct validity of empirical measures of transaction-specific investments.
That is, our item captures the specificity of the investment, the size (and not only the
presence) of the investment, and we obtain this information from the firm making the
investment and not from just any of the firms involved in the transaction.
To measure similarity of routines and culture, we used a two-item measure to capture
the extent to which there was overlap in the transferor's and recipient's organizational routines
and culture. Specifically, we used five point likert type items to ask the respondents to what
extent their business practices and operational mechanisms, and organizational culture and
management style were similar to that of their foreign partner. This scale had an ordinal alpha
of 0.80.
Control Variables
To control for alternative explanations for the relationship between our dependent
and independent variables, we included controls for industry, age of the technology, cultural
distance between home country of the foreign firm and India, and prior export experience of
the foreign firm in India.
Tacitness of a technology could be correlated with its age. As time wears on, details
of the know-how become more established and well known. And it has been shown that older
technologies are more likely to be transferred on the market (Davidson & McFetridge, 1985).
-
8/13/2019 Eapen Krishnan Paper
24/51
23
To rule out any correlation we observe between tacitness and choice of transfer mode being
due to the common correlation with age of the technology, we specifically controlled for it.
Age of technology was coded 1 when the respondent indicated that the core technology
transferred (the technology stricto sensu) had been introduced in the transferors home
country within the past year, 2 when it was two to three years old, 3 when it was three to five
years old and 4 when its first introduction was more than five years ago.
Cultural distance between the foreign and Indian firm could be correlated with both
the latter's perception of tacitness and the mode of transfer. Whether a given technology
appears tacit or not to a recipient could be partly a function of the cultural distance between
the sender and recipient, and cultural distance has been shown to influence the choice between
licensing and joint venture (Garca-Canal, Valds-Llaneza, & Snchez-Lorda, 2008). Hence
we included cultural distance, measured using the Kogut & Singh (1988) index, as a control.
A technology transferor who has been exporting to India in the past may be more
willing to make greater resource commitments, i.e. to choose an EJV over a licensing
agreement. To control for this, we included Prior Export, a dummy variable that was coded 1
if the foreign firm had exported its products to India prior to the present collaboration and 0
otherwise.
Finally, to control for possible industry effects, we included industry dummies for
chemicals, electronics, industrial machinery and transportation equipment industries. We are
unable to include dummies for all industries in our data because the number of observations
per industry is not large enough to justify that.
RESULTS
We report descriptive statistics and pair-wise correlations in table 1. Given the binary nature
of our dependent variable, we used logistic regression to estimate the coefficients in our
model. The results of our conventional logistic regression analyses are in table 2. To test our
-
8/13/2019 Eapen Krishnan Paper
25/51
24
hypotheses, we include in our models interaction terms between tacitness and each of our
moderator variables. To attenuate problems related to multicollinearity, we mean-centered
these variables before multiplying them to create the interaction terms. Model 1 includes only
control variables, models 2 and 3 include the main independent variables, and models 4, 5,
and 6 include interaction terms one at a time. Model 7 is the full model that includes all three
sets of variables controls, independent variables, and the interaction terms. The main
variables of interest, given our hypotheses, are the interaction terms. Looking at the
coefficient estimates and their standard errors in model 7, we find support for hypotheses 1
and 2, but not for hypotheses 3.
--------------Please insert tables 1 and 2 about here-----------
However, interpreting results in logistic regression models is notoriously difficult
(Hoetkar, 2007). Unlike in linear regressions, estimates of coefficients in non-linear models,
such as the logistic one, do not represent the actual marginal effect of independent variables.
In the logistic case, the effect of each variable is non linear, and depends on its own values as
well as on the values of other independent variables in the model. Interpreting interaction
terms is even less straightforward. The coefficients of interaction terms are not equal to their
marginal effect (Ai & Norton, 2003; Zelner, 2009) and conventional tests of statistical
significance are wrong (Zelner, 2009).
To make interpretation easy and to draw valid conclusions from logistic regressions,
Zelner (2009) recommends computing differences in predicted probabilities () at different
values of the independent variables. For example, we could calculate the predicted probability
of joint venture () while holding tacitness at its minimum value and other independent
variables at some theoretically interesting level. We could then repeat this, this time holding
tacitness at its maximum value. The difference between these two predicted probabilities ()
i.e., at low and high levels of tacitness - indicates the 'effect' of tacitness on the probability
-
8/13/2019 Eapen Krishnan Paper
26/51
25
to joint venture. To get a sense of the magnitude of interaction effects, we could calculate
at different values of the moderating variable. That is, we could calculate and compare the
effect of tacitness () when the know-how is central to the recipient as well as when it is not.
This double difference (), the difference in the effect of tacitness () when know-how
is central to the recipient and when it is not, represents the interaction effect. If we find that
the effect of tacitness is different when the know-how transferred is central to the recipient
and when it is not, this suggests an interaction effect. And if we find that the effect of tacitness
on joint venture is stronger when know-how is central to the recipient, this supports
hypothesis 1.
While examining differences in predicted probabilities () and double differences
() gives us a sense of the magnitude of the effects we are interested in, we still do not
know if these magnitudes are statistically significant. To calculate statistical significance,
Zelner (2009) recommends a simulation-based method drawn from the political science
discipline (King, Tomz, & Wittenberg, 2000). Essentially here, confidence intervals for
coefficients, differences in predicted probabilities, and double differences are computed by
drawing several values of these estimates from a simulated distribution. Using these
confidence intervals, we can make inferences about statistical significance. We implemented
this procedure by using the clarify suite of programs written for Stata (Tomz, Wittenberg,
& King, 2001). The results of these analyses are in tables 3 and 4.
---------------Please insert tables 3 and 4 about here-------------
Table 3 shows the effect of tacitness () at minimum and maximum values of other
variables. The predicted probability of joint venture at low levels of tacitness and when the
know-how is central to the recipient is 0.30 (row 1(b), column 1 in table 3). This increases to
0.96 in cases where the know-how transferred is highly tacit (row 1(b), column 2 in table 3).
The effect of tacitness, when know-how is central to the recipient, is thus 0.66. In a similar
-
8/13/2019 Eapen Krishnan Paper
27/51
26
way, tacitness increases the probability of joint venture by 0.15 when the size of specific
investment variable is held at its minimum (row 2(a); column 3), and by 0.96 when the size of
specific investment is very high (row 2(b); column 3). We also compute the confidence
intervals for these effects. Essentially, if the confidence interval contains 0, is insignificant,
and vice-versa. An interesting result that is not revealed in the conventional logistic regression
results is that the effect of tacitness varies in magnitude across values of other variables. Also,
the effect of tacitness is significant only at higher levels of the other variables. This is
suggestive of interaction effects.
Based on Zelners (2009) recommendation, we also plot these results in figures 1, 2,
and 3. Figure 1 depicts the estimated effect of tacitness () along with its 95 percent
confidence intervals at various (mean-centered) values of centrality of know-how to the
recipient. The line in the plot that is bounded by confidence intervals represents the effect of
tacitness () at various levels of centrality of know-how to the recipient. The narrow
vertical rectangles denote the confidence intervals associated with these estimates. Figures 2
and 3 show the plots for size of specific investmentand similar routinesrespectively.
-----------Please insert figures 1, 2, and 3 about here-----------
The statistical test of the interaction hypotheses is essentially whether the effect of
tacitness at low and high values of the moderator variables are significantly different. This is
what we represent in table 4. As per Zelner's recommendation we compute double differences
(). These are the differences between the effects of tacitness at lowest and highest values
of the moderating variables. We also construct confidence intervals that allow us to run
statistical significance tests.
Table 4 shows that there are large differences in the effect of tacitness at various
values of the moderator variables. We find that tacitness has a much stronger effect when
know-how is central to the recipient and when the need for specific investment is high. These
-
8/13/2019 Eapen Krishnan Paper
28/51
27
two results are consistent with transaction cost theory. Contrary to the prediction from the
knowledge based view though tacitness seems to lead to the choice of joint venture when
transferor and recipient have similar routines.
Sensitivity Analysis
We tested if our results were sensitive to alternative measures of tacitness. We recoded
tacitness of the know-how packageas a count measure that took values from 1 to 5 depending
on the number of types of supporting skills required. We also ran our analysis with another
measure of tacitness tacitness of the core technology a set of three items that tapped into
the extent to which the technology being transferred could be codified and taught, again
measured from the recipient's perspective. Specifically, the five point likert type items
captured the extent to which the technology transferred could be described in a manual, the
extent to which the recipient can learn to manufacture the product or implement the
manufacturing process by looking at a set of blueprints, and the extent to which
manufacturing the product and/or implementing the process requires the on-site guidance of
employees of the know-how transferring firm. The ordinal alpha for this measure was 0.75,
and all three items loaded well onto one factor. Our results are robust to the use of these two
different measures of tacitness.
A possible threat to internal validity in our study lies in our use of single-item
measures for centrality of know-how to the recipient and size of specific investment. One
argument, based on the literature on (psychometric) measurement is that single-item measures
are prone to measurement error. Measurement error in single-item scales could arise out of not
being able to capture the multiple dimensions of a construct, and by leaving respondents with
greater ambiguity to interpret the item in their own way. Moreover, one cannot compute a
reliability statistic such as the cronbach or ordinal alpha in a single-item measure (Hoeppner
et al., 2011). Using multiple items on the other hand generally allows for averaging out
-
8/13/2019 Eapen Krishnan Paper
29/51
28
random measurement error, and better means of capturing a complex construct. In our study,
however, measurement error is unlikely to be a significant problem given that our single-item
variables centrality of know-how to the recipientand size of specific investment capture
objective and not perceptual data. We are indeed interested in the potential for opportunism in
each transaction, but we do not directly measure the perception of opportunism (e.g.,
Anderson, 1988; Sako & Helper, 1998). Instead, for reasons mentioned before and as several
others have done (e.g., Gulati & Singh, 1998; Klein et al., 1978; Oxley, 1997 etc.), we infer
the potential for opportunism from certain objective characteristics of the transaction. And for
such objective characteristics, single-item measures are usually sufficient (e.g., Sako &
Helper, 1998: 396). Measurement error in our single-item measures, thus, is likely to be small.
Nevertheless, we did some additional analysis to assess the resilience of our results to
potential measurement error. Specifically, we employed regression calibration (Carroll &
Stefanski, 1990; Hardin, Schmiediche, & Carroll, 2003) to check the sensitivity of our
estimation to measurement error. Measurement error in independent variables generally poses
two problems for regression estimates: (a) it reduces the power of the test, making it more
difficult to detect significant effects (Anderson, 1988: 255; Carroll et al., 2006), and (b)
induces bias in the coefficient estimates (Carroll, et al., 2006; Hardin et al., 2003). The first is
likely not a problem for our study because for the interaction terms that include our two
single-item variables centrality of know-how to the recipient and size of specific investment,
we do find significant effects. In other words, we get significant effects even when
measurement error should have made finding support for our TCE hypotheses more difficult.
The latter problem of potentially biased coefficients, on the other hand, is something we
would like to rule out. To do this, we estimated two sequences of regression calibration
models, one for centrality of know-how to the recipient and the other for size of specific
investment. In each sequence we progressively attributed a larger proportion of the total
-
8/13/2019 Eapen Krishnan Paper
30/51
29
variance in the (single-item) variable and its interaction term to measurement error. For
example, in the sequence of regression calibration models for size of specific investment, we
initially assumed that all the variance in size of specific investment and its interaction term is
due to actual variance in the true measures (i.e., zero measurement error). In subsequent re-
runs of the model, we progressively increased the proportion of total variance due to
measurement error to five, ten, fifteen, twenty, and twenty-five percent. Regression
calibration modelling takes these estimates of error variance into account, and produces
unbiased coefficients (Hardin & Carroll, 2003; Hardin, et al., 2003). We find that as we
assume greater levels of measurement error, the unbiased coefficient estimate of the
interaction terms are much larger than what we report in our main results table. In other
words, if measurement error were present in our data, what we will have, more likely than
not, is attenuation bias; in the other words, the coefficients we report here are smaller than
what they ought to be. This obviously is not very disconcerting. It is perhaps useful to also
note that the interaction term with size of specific investment remains significant in the
regression calibration models. The interaction term with centrality of know-how to the
recipient has the correct sign, but is significant only in a one-tailed test. This is not cause for
excessive concern, however, given the inherent tendency in regression calibration models to
trade off variability of the coefficient estimates in favour of unbiasedness (Carroll et al.,
2006).
Finally, while our results show that firms use structural solutions, such as equity joint
ventures, to mitigate opportunism concerns in the transfer tacit knowledge, there is a rich line
of research on how firms can handle opportunism but through informal solutions, such as
trust-based governance (Madhok, 1995; Carson et al., 2003; Ghoshal & Moran, 1996). Gulati
(1995) shows that partners formed non-equity alliances instead of equity alliances when they
had prior ties, which he argued indicates the presence of prior trust between partners. We
-
8/13/2019 Eapen Krishnan Paper
31/51
30
tested if our results held even after controlling for prior familiarity between partners. We
included in our regressions a dummy variable that indicated whether or not the recipient and
transferor had a licensing or joint venture relationship before the focal one. And the results
were not different from what we report here.
DISCUSSION
Our study seeks to contribute to the KBV-TCE debate on the relevance of
opportunism for the theory of the firm. Both TCE and KBV predict that tacit know-how is
more likely to be transferred through hierarchical modes than on the market, but TCE
attributes this to market failure arising from opportunism concerns and KBV, to the difficulty
in transferring tacit know-how owing to the absence of similar routines. We see this as the key
bottleneck in the debate this far. While both theories disagree on the relevance of
opportunism, they arrive at identical predictions for how tacit know-how is transferred. This
lack of predictive uniqueness has also hindered empirical tests of the competing causal
mechanisms.
In this paper, we sought to move the KBV-TCE debate forward by crafting and
testing contingency arguments from both theories. We examined whether opportunism matters
to how tacit know-how is transferred. In testing contingency arguments from TCE, we
predicted that the likelihood of tacit know-how being transferred through hierarchical modes
will be stronger when opportunism inducing conditions are prevalenti.e., when know-how
is core to the recipient firm in the host country and when the specific investments made by the
recipient firm is high. Similarly, in testing contingency arguments from KBV, we predicted
that the likelihood of tacit know-how being transferred through hierarchical modes will be
weaker when the organizational culture and routines of the transferor and recipient are similar.
Irrespective of our empirical findings, this theoretical approach confers several
advantages. Framing arguments from KBV and TCE as contingency statements brings
-
8/13/2019 Eapen Krishnan Paper
32/51
31
distinctiveness to the predictions of both theories, and in the process, greater clarity to the
debate. Contingency arguments also allowed us to tease out and isolate the different causal
mechanisms in KBV and TCE and as a result, we conduct more direct tests of TCE and KBV
positions (Tsang, 2006).
We considered several candidate moderators of the relationship between tacitness
and transfer mode to use in our study, but not all of them were useful for our purpose, viz.,
separating TCE and KBV predictions. On the one hand, many potential moderators that signal
opportunism concerns can also be argued to signal transfer costs, but more importantly, the
moderators that will do the job for us are those that deliver distinctive predictions for TCE
and KBV. This is what has primarily guided our choice of moderators. From a TCE point of
view we argue that when know-how is core to the recipient, it is likely to trigger opportunism
concerns among both partieswith the transferor being concerned about potential knowledge
leakage and the recipient, about misrepresentation of knowledge. Our second moderator,
specific investment by the recipient, closely mirrors the TCE emphasis on asset specificity.
These two, in our view, are distinctively TCE hypotheses. They are useful for our purpose
because it is difficult to come up with knowledge-based explanations that result in the same
predictions. Similarly, our third moderator similarity in routines and culture closely relates
to the conceptual mechanisms of the knowledge-based view. It is difficult to draw an
opportunism-based explanation for why similarity in routines should make the transfer of tacit
know-how any easier.
Besides the novelty of our theoretical approach that is based on contingency
arguments, the fact that we bring some empirical evidence to bear on this debate is also
valuable because the literature surrounding the debate has so far been largely on conceptual
turf alone. Except for Kogut and Zanders (1993) article, all the other articles critiquing or
defending their contention (e.g., McFetridge, 1995; Madhok, 1996; Love, 1995; Fransson et
-
8/13/2019 Eapen Krishnan Paper
33/51
32
al., 2011) have been conceptual. Even Kogut and Zanders (1993) article, although empirical
in nature, was not able to partition out the causal mechanisms underlying KBV. As Kogut &
Zander (1993: 633) themselves indicate, ...this test is by no means fully specified. In
particular, we do not look at the capabilities of the recipient. Our paper has taken an
empirical step forward in this direction.
Implications
The key implication of our study is that opportunism matters. Several aspects of our
findings lend robust credence to this conclusion. Firstly, we set up two separate tests
hypotheses 1 and 2 for whether opportunism matters, and across both, we find consistent
results that it does. The likelihood of tacit know-how being transferred through hierarchy is
significantly greater, both in terms of magnitude and statistical significance, when the know-
how is core to the recipient and when the recipient has made transaction specific investments.
So opportunism necessitates recourse to hierarchy whenever the transferor seeks to transfer
tacit know-how. Secondly, the robustness of this conclusion also comes from the fact that our
results hold even when characteristics of our empirical test militate against finding any
significant results. As discussed earlier, our test, based on only licensing and joint ventures, is
very conservative. Thirdly, we do not find evidence that that transfers will be via licensing
when transferor and recipient share the same routines and culture. In fact we find the opposite,
although this result is not statistically significant. That is, we find that the likelihood of joint
venture is higher when tacit know-how is transferred to recipients with similar routines and
culture as the transferor. If we assume for a moment that similar routines reflect the ability of
the recipient to compete back against the transferor, this result may also be considered
additional support for TCE. When routines are similar, the transferor is likely to want better
avenues for monitoring the use of know-how. And this is better afforded through JVs than
licensing.
-
8/13/2019 Eapen Krishnan Paper
34/51
33
The second key implication of our results is that the causal mechanism underlying
the often observed link between tacitness and hierarchical transfer modes is opportunism and
not ease of transfer due to similar routines. As we have mentioned, an important point of
contention in the KBV-TCE debate is that the link between tacitness and hierarchy is equally
amenable to both KBV and TCE explanations. The way we set up our tests, we were able to
directly test which of these two mechanisms is in play. There were three occasions in our
empirical analysis for the KBV argument to receive support and show through. Had the effect
of knowledge tacitness on transfer mode choice been insensitive to opportunism concerns (in
hypotheses 1 and 2), that would have been sufficient to support the KBV contention.
However, our results support the contention that opportunism matters. We also examined
separate contingencies for KBV that transferring tacit knowledge hierarchically might be less
likely if organizational culture and routines are similar. Finding this would have lent support
to KBV that it is transfer costs and not opportunism that guides transfer mode choice. Our
result for this prediction too was not significant. Across these three sets of results, we think it
is fairly robust conclusion that, at least in our empirical context, opportunism and not
similarity of routines is the relevant causal mechanism in play.
A third important implication of our study is that it speaks to the question of what do
firms do? This is a significant issue for the theory of the firm in general (Conner, 1991;
Conner & Prahalad, 1996; Grant, 1996; Kogut & Zander, 1992; 1996; Foss, 1996a; 1996b),
but also for the study of joint ventures (Gulati & Singh, 1998; Hennart, 1988). The dominant
view until recent times has been that the raison d'etre for firms is their ability to curb
opportunism. By virtue of better monitoring and incentive-alignment mechanisms, hierarchy
reduces the payoffs from opportunism for the members of a firm (Kale & Puranam, 2004;
Wathne & Heide, 2000; Williamson, 1985). To this end, Williamson (1985:66) says:
...market contracting would be ubiquitous in the face of nonopportunism. Departing from
-
8/13/2019 Eapen Krishnan Paper
35/51
34
this emphasis on opportunism and transaction cost mitigation, other scholars have suggested
that what firms offer, over and above markets, is superior coordination abilities (Grant, 1996),
the origins of which are in the social community of voluntaristic action inherent in firms
(Kogut & Zander, 1992: 384). A similar opportunism-mitigation versus coordination argument
has played out in the strategic alliance literature as well. One view in this literature is that
joint ventures mainly serve an opportunism reduction purpose, i.e., since partners in a joint
venture share from the gains of the venture, their incentives are aligned and they are less
likely to act opportunistically (Hennart, 1988; Kale & Puranam, 2004). An alternative view is
that joint ventures are chosen because they facilitate better coordination between partners
(Gulati & Singh, 1998; Kogut & Zander, 1993; Madhok, 1995; 2006; Phene & Tallman,
2012). In recent times, however, scholars have converged on the notion that hierarchical
governance mechanisms simultaneously serve both opportunism-mitigation and coordination
functions (Kale & Puranam, 2004; Mesquita & Brush, 2008). In fact, TCE scholars in strategy
and international business have not only highlighted the importance of coordination
(Williamson, 1985; 1993), but even extended the transaction cost construct to include
communication and coordination costs as well (Rugman & Verbeke, 2003; Phene & Tallman,
2012).
What still distinguishes TCE and other perspectives on the question of what firms do,
however, is its view on where the coordination benefits of hierarchy come from. To TCE
proponents, coordination comes from reduced incentives of members of a firm to act
opportunistically; ex-post maladaptation problems are less likely to arise in firms due to the
absence of opportunism (Williamson, 1993: 97). The knowledge based view on the other hand
ascribes coordination benefits to higher-order organizing principles (Kogut & Zander, 1992),
a shared identity among members of a firm, and common procedural rules (Kogut & Zander,
1996). Somewhat similarly, Ghoshal & Moran (1996) suggest that coordination arises from a
-
8/13/2019 Eapen Krishnan Paper
36/51
35
sense of shared purpose. Contrasting the TCE view with theirs, they remark: the advantage
of organizations over markets may not lie in overcoming human pathologies through
hierarchy, but in leveraging the human ability to...cooperate (p. 42).
So where do cooperation and coordination in firms really come from? Our view, which
we think is reflected also in our empirical results, rests on the following three observations.
Firstly, hierarchy need not automatically or always confer coordination benefits (Sako &
Helper, 1998:391). Coordination, even within firms, is difficult unless specific mechanisms
are in place (Bechky, 2003; Hansen, 1999; Mors, 2010). Hierarchy and superior coordination
are thus not synonymous3
3 To be fair, Kogut & Zander do allude somewhat briefly to the difficulty in transferring know-
how between departments (1992: 389) and to the fact that firms differ in what they can do well(1996: 515). But their mainstream emphasis is on the coordination-enabling features of the firmsthrough shared identities and higher order organizing principles.
. Secondly, there is no reason why coordination, with some effort,
cannot be effected on the market as well. If this statement is true, then hierarchy may not be
essential to generating coordination benefits. Foss (1996a) makes this very point: ...the gains
from...being embedded in higher order organizing principles could be realized over the
market. In other words, agents could simply meet under the same factory roof, own their
own pieces of physical capital equipment or rent it to each other, and develop value-
enhancing organizing principles among themselves, or in other ways integrate their
specialized knowledge. Firms would not be necessary (Foss & Klein, 2008; italics added).
This view resonates well with Mesquita & Brush's (2008) empirical results that contracts,
with some embellishments, can facilitate hierarchy-like coordination between alliance
partners while at the same time, they do not adequately safeguard from opportunism (p.803).
Srikanth & Puranam (2011) find similar results showing that fairly intensive coordination can
be effected even in arms-length (offshoring) contracts. Moreover, investigating the kinds of
coordination mechanisms within and across firms, Srikanth & Puranam (2010) conclude that
there is no unique coordination mechanism within firms that cannot be implemented in
-
8/13/2019 Eapen Krishnan Paper
37/51
36
contracts. Extrapolating from these results, one could argue that hierarchy is critical to
curtailing opportunism but not to ensuring coordination. Thirdly, the argument that
coordination arises from high-order organizing principles need not be inconsistent with an
opportunism-based argument. Transaction cost theorists would suggest that ....the absence of
opportunism that hierarchy may help to create stimulates the emergence of trust, cooperation,
information exchange, and various sets of commitments, which is precisely what [is] packed
into the concept of higher order organizing principles (Foss, 1996a: 473). Simply put, higher
order coordination routines generally seen within firms could very well be because of the
reduced payoffs from being opportunistic for members of the firm, and in that respect
consistent with transaction cost theory.
Taken together, these three observations suggest that the while the need for coordination
is important in alliance governance, its role in driving the preference for hierarchy over
market is not unequivocally clear. Coordination-benefits do not seem unique to hierarchy
since firms can still suffer coordination problems, and effective coordination can occur on the
market (Foss, 1996a; Srikanth & Puranam, 2011). What hierarchy really does is opportunism
mitigation. And we think our results reflect this. In conditions were coordination is likely to
be difficult, such as when routines and culture of partners are dissimilar, we do not find
managers choosing joint venture over licensing. But on the other hand, the managers we
surveyed chose joint ventures when opportunism concerns were high. So, what do firms do?
They may very well facilitate coordination, but we think this is likely only because, as a first
step, they mitigate opportunism. To be clear, we certainly do believe that coordination is
crucial to transferring tacit knowledge (Nonaka, 1994), but not that this is what fundamentally
distinguishes hierarchy and markets.
There are, however, three important caveats to our story. Firstly, as we have said, our
tests are very conservative. And despite that we find results supporting TCE. This strengthens
-
8/13/2019 Eapen Krishnan Paper
38/51
37
our confidence in the TCE results, but at the same time suggests that the insignificant findings
for KBV do not indicate the theory is irrelevant or wrong. Our lack of support for KBV could
simply be because our tests are too conservative to pick up those effects. Secondly, scholars
have suggested that opportunism need not always be guileful. For example, Verbeke &
Greidanus (2009) propose the concept of bounded reliability, taking into account that all
failed commitments need not always be motivated by deceit. Our empirical analysis
unfortunately does not distinguish between opportunism that is and isnt deceitful. And in that
sense, Verbeke & Greidanuss (2009) concept of bounded reliability could very well be an
alternative explanation to the conclusions we draw from our results. Thirdly, while we control
for industry characteristics, we do not thoroughly explore whether our results are industry-
dependent. It is plausible that the relevance of opportunism could vary by industry. For
example, in biotechnology alliances, knowledge transfer concerns rooted in coordination
concerns and not opportunism could be more relevant (Phene & Tallman, 2012).
Nevertheless, and despite these limitations, our study makes a worthwhile contribution to the
opportunism debate. Conceptually, we derive distinctive predictions that capture the
mechanisms underpinning KBV and TCE views, and empirically, we bring a set of results to
bear on the question of whether or not opportunism matters. So, is opportunism really
redundant in explaining the boundaries of the firm? Based on our research and subject to the
caveats we have just mentioned, we think not.
REFERENCES
Ai, C.R., & Norton, E.C. 2003. Interaction terms in logit and probit models. EconomicsLetters,80(1):123129.
Anderson, E. 1988. Transaction costs as determinants of opportunism in integrated andindependent sales forces. Journal of Economic Behavior and Organization, 9(3): 247-264.
Armstrong, S.J., & Overton, T.S. 1977. Estimating non-response bias in mail surveys.Journal of Marketing Research, 14: 396-402.
Arrow, K., 1962. Economic welfare and the allocation of resources for invention. In Arrow, K.
(ed.) The rate and direction of economic activity.Princeton: Princeton University Press.Baranson, J. 1969.Industrial technologies for developing economies.NY: Praeger.
-
8/13/2019 Eapen Krishnan Paper
39/51
38
Barzel, Y. 1989.Economic analysis of property rights.NY: Cambridge University Press.Bechky, B.A. 2003. Sharing meaning across occupational communities: The transformation of
understanding on a production floor. Organization Science, 14(3): 312- 330.Brief, A., & Downey, H.K. 1983. Cognitive and organizational structures: A conceptual
analysis of implicit organizing theories.Human Relations. 36 (12): 1065-1090.
Buckley, P., J & Casson, M., 1976. The future of the multinational enterprise, London:Macmillan.
Busenitz, L.W., & Lau, C.M. 1996. A cross-cultural cognitive model of new venture creation.Entrepreneurship Theory and Practice. Summer: 25-39.
Carroll, R.J., & Stefanski, L.A. 1990. Approximate quasilikelihood estimation in models withsurrogate predictors.Journal of the American Statistical Association, 85: 652-663.
Carroll, R.J., Ruppert, D., Stefanski, L.A., & Crainiceanu, C.M. 2006. Measurement error innonlinear models: A modern perspective.London: Chapman & Hall.
Carson, S.J.