Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Empirical essays on relationships between alliance experience and firm capability development
(USC Thesis Other)
Empirical essays on relationships between alliance experience and firm capability development
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
EMPIRICAL ESSAYS ON RELATIONSHIPS BETWEEN
ALLIANCE EXPERIENCE AND FIRM CAPABILITY
DEVELOPMENT
by
Rui Wu
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BUSINESS ADMINISTRATION)
December 2010
Copyright 2010 Rui Wu
ii
ACKNOWLEDGEMENTS
This dissertation will not be made possible without warm encouragements
and constructive inputs from many people. I would like firstly thank the guidance
from my dissertation chair, Professor Kyle Mayer, who has pushed me toward
theoretical rigorousness and a dedicated researcher. Professors Nandini Rajagopalan
and Mark Kennedy have always been supportive and encouraging, with their
generously invested time and efforts throughout my dissertation process. To me, they
are great examples of passionate and caring academic experts. At the end of this
journey, I must thank Professor Paul Adler, who admitted me into the doctoral
program and opened the door for me toward management research.
Throughout the journey of completing this dissertation, I was constantly
encouraged, intellectually and emotionally, by my dearest friends, Hu Bin, Professor
Huang Zhi (HKUST), Dr. Li Tong (Milken Institute), Professor Shen Changyu
(Indiana University), Professor Yu Tieying (Boston College), and Wang Haofei.
Professor Libby Weber (UC Irvine), my dear colleague and friend throughout the
doctoral program, has always been with me at my toughest times. Last but probably
the most important, the eternal love, trust, and supports from my family.
iii
TABLE OF CONTENTS
Acknowledgements ........................................................................................................ ii
List of Tables................................................................................................................. vi
List of Figures .............................................................................................................. vii
Abstract ....................................................................................................................... viii
Chapter 1: Introduction ...................................................................................................1
1.1 Alliance Experience ...........................................................................................1
1.2 Theoretical Perspectives in Alliance Research ..................................................3
1.2.1 Transaction Cost Economics ......................................................................3
1.2.2 The Capability Perspective .........................................................................7
1.2.3 Organizational Learning Perspective ........................................................12
1.3 Motivations and Key Findings .........................................................................16
1.3.1 Essay One .................................................................................................16
1.3.2 Essay Two .................................................................................................19
1.4 Dissertation Outline .........................................................................................22
Chapter 2: Literature Review ........................................................................................24
2.1 Transaction Cost Economics (TCE) ................................................................24
2.1.1 Overview ...................................................................................................24
2.1.2 Behavioral Assumptions ...........................................................................25
2.1.3 Transaction Costs ......................................................................................27
2.1.4 Governance Structures ..............................................................................29
2.1.5 Attributes of Transactions .........................................................................32
2.1.6 Discriminating Alignment ........................................................................36
2.1.7 Empirical TCE Research Related to Strategic Alliances ..........................38
2.1.8 Contributions of Dissertation ....................................................................49
2.2 The Capability Perspective .............................................................................52
2.2.1 The Resource-Based View (RBV) ............................................................52
2.2.2 The Knowledge-Based View (KBV) ........................................................56
2.2.3 The Capability Perspective .......................................................................57
2.2.4 The Capability Perspective in Strategic Alliance Research .....................60
2.2.5 Contributions of Dissertation ....................................................................64
2.3 Organizational Learning .................................................................................67
2.3.1 Overview ...................................................................................................67
2.3.2 Learning Processes ...................................................................................69
2.3.3 Learning Effects ........................................................................................73
2.3.4 Learning Theories in Strategic Alliance Research ...................................80
iv
2.3.5 Contributions of Dissertation ....................................................................83
Chapter 3: Industrial Context – The Software Industry ................................................86
3.1 Industry Background .......................................................................................86
3.1.1 Overview ...................................................................................................86
3.1.2 Major Market Segments ...........................................................................90
3.1.3 Companies in the Data Used in this Dissertation .....................................91
3.2 Strategic Alliances in the Software Industry ...................................................94
3.2.1 Alliance Relationships ..............................................................................95
3.2.2 Alliance Types ..........................................................................................99
3.2.3 Trends of Alliances .................................................................................101
3.2.4 Examples of Alliance Experience ...........................................................106
3.3 Conclusions ....................................................................................................109
Chapter 4: Essay One —Effects of Governance Experience on Alliance
Organizational Structure Design .................................................................................112
4.1 Introduction ....................................................................................................112
4.2 Literature Review ..........................................................................................117
4.2.1 Transaction Cost Theory and Governance..............................................117
4.2.2 Firm Capabilities and Governance .........................................................119
4.2.3 Firm Learning and Governance ..............................................................123
4.3 Development of Hypotheses ..........................................................................125
4.3.1 Alliance Governance Forms ...................................................................125
4.3.2 Contractual Hazards ................................................................................128
4.3.3 Experience-Based Capabilities ...............................................................132
4.3.4 Governance Diversity of Alliance Experience .......................................137
4.4 Data and Methods ..........................................................................................149
4.4.1 Sample ....................................................................................................149
4.4.2 Measures .................................................................................................152
4.4.2.1 Dependent Variable: Governance Forms .........................................152
4.4.2.2 Independent Variables .....................................................................154
4.4.2.3 Control Variables .............................................................................158
4.5 Results ............................................................................................................162
4.6 Discussion and Conclusions ..........................................................................168
4.6.1 Summary of Findings .............................................................................168
4.6.2 Implications ............................................................................................170
4.6.3 Contributions ..........................................................................................172
4.6.4 Limitations and Future Research ............................................................174
Chapter 5: Essay Two —Effects of Alliance Experience on Stock Market Value
Creation .......................................................................................................................177
5.1 Introduction ....................................................................................................177
v
5.2 Literature Review ..........................................................................................182
5.2.1 Organizational Learning and Firm Experience .......................................182
5.2.2 Resource-Based View and Alliance Capabilities ...................................185
5.3 Development of Hypotheses ..........................................................................189
5.3.1 Dimensions of Alliance Governance Experience ...................................189
5.3.2 Effects of Governing Capabilities on Value Creation ............................191
5.3.3 Effects of Selection Capabilities on Value Creation ..............................195
5.3.4 Boundary Conditions of Experience Effects ..........................................199
5.4 Data and Methods ..........................................................................................206
5.4.1 Sample ....................................................................................................206
5.4.2 Measures .................................................................................................210
5.4.2.1 Dependent Variables: Abnormal Stock Market Returns .................210
5.4.2.2 Independent Variables: Alliance Experience ...................................212
5.4.2.3 Control Variables .............................................................................214
5.5 Results ............................................................................................................220
5.6 Discussion and Conclusions ..........................................................................233
5.6.1 Summary of Findings .............................................................................233
5.6.2 Implications ............................................................................................235
5.6.3 Contributions ..........................................................................................237
5.6.4 Limitations and Future Research ............................................................238
Chapter 6: Conclusions ...............................................................................................241
6.1 Review of Motivations ..................................................................................241
6.2 Summary of Findings ....................................................................................242
6.3 Contributions to Theories and Practice ..........................................................244
6.3.1 Theoretical Contributions and Implications ...........................................245
6.3.2 Empirical Contributions and Managerial Implications ...........................248
6.4 Limitations and Directions for Future Research ............................................250
References ...................................................................................................................255
Appendix 1: Standard Industrial Classification Definitions .......................................271
Appendix 2: List of Companies in Empirical Sample ................................................272
vi
LIST OF TABLES
Table 2.1: Attributes of Governance Structures ............................................................32
Table 3.1: Worldwide Packaged Software Revenue Forecast, by Region ....................90
Table 3.2: Examples of Software Companies: Microsoft Corp vs. ANSYS Inc ........108
Table 4.1: Comparison of Alliance Governance Structures........................................128
Table 4.2: Summary Statistics ....................................................................................160
Table 4.3: Ordered Probit Model Regression Results .................................................166
Table 5.1: Summary Statistics of the Regression Sample...........................................218
Table 5.2: Even Study Results of Abnormal Returns of Alliance Announcements ...219
Table 5.3: Abnormal Returns by Focal Alliance Governance Structures ...................220
Table 5.4: Effects of Alliance Experience on Abnormal Stock Returns.....................224
Table 5.5: Split Sample Regressions by Alliance Governance Form .........................228
Table 5.6: Split Sample Regressions by Focal Firm Performance..............................231
vii
LIST OF FIGURES
Figure 1.1: Dimensions of Alliance Experience ...........................................................21
Figure 2.1: Williamson ’s Conception of Exchange Governance Structures ................29
Figure 2.2: Levels of Administrative Control and Incentive Intensity in Governance
Forms ............................................................................................................................30
Figure 2.3: Governance Choices as a Function of Asset Specificity ............................36
Figure 3.1: Software Industry Market Value ................................................................88
Figure 3.2: Annual Growth Rate of Software Market Value ........................................89
Figure 3.3: Average New Alliances and Average Portfolio Size per Firm .................103
Figure 3.4: Alliance Governance Diveristy by Experience Quantiles ........................105
Figure 4.1: A Model of Alliance Governance Decisions ............................................148
Figure 5.1: Dimensions of Alliance Experience .........................................................191
Figure 5.2: Effects of Alliance Experience on Market Value Creation ......................198
Figure 5.3: A Model of Experience Effects on Market Value Creation .....................206
viii
ABSTRACT
In the context of interfirm alliances, this dissertation analyzes partners ’ alliance
experience as a multi-dimensional construct, and examines the effects of experience
dimensions on governance decisions and on market value creations. This dissertation
focuses on the governance aspect of experience, or the extent to which a firm has
managed focused or diverse alliance governance structures. A firm ’s experience of
prior alliances can be characterized by the depth in a specific governance form and
the breadth of diverse governance forms. In-depth experience creates governing
capabilities that are specific to a focal structure and result in exploitation of the same
structure. Diverse governance experience broadens the range of alliance-related
knowledge, and lead to better informed governance decisions by the creation of
selection capabilities. In the first empirical essay, I examine a model that integrates
both contractual hazards and experience-based capabilities to predict governance
decisions. The second essay takes a further step by examining how the stock market
responds to experience factors when evaluating events of new alliance formation. In
a sample of alliances formed by US software companies, I find strong empirical
evidence for the argument of multi-dimensional experience in affecting strategic
decisions and value creations.
1
CHAPTER 1: INTRODUCTION
1.1 Strategic Alliance
A strategic alliance is a form of voluntary interfirm relationships whereby
two or more firms agree to pool respective resources to pursue specific market
opportunities (Gulati, 1995). Taking the form of an interfirm contract, a strategic
alliance is neither a pure market transaction nor pure hierarchy. Rather, it is a hybrid
arrangement that represents an intermediate organizational form between the two
(Hennart, 1988). While strategic partners may have common interests in alliances
that lead to resource (1995), they may also seek to use alliances to further interests
that are not shared. Because such pursuit of private interests sometimes leads to non-
cooperative behaviors that damage or end alliance relationships (Khanna, Gulati, &
Nohria, 1998), alliances are commonly characterized as featuring tension between
cooperation and competition (Zeng & Chen, 2003).
Given this tension, early alliance research considered relationship duration or
stability as indicators for alliance success (Doz, 1996; Parkhe, 1993a). But recent
studies examine alliance as a learning process where partners develop both common
and private knowledge (Khanna, 1998). Allied partners learn to cooperate, to
contract, and to manage relationships over time (Mayer & Argyres, 2004).
Throughout such learning processes, knowledge and capabilities accrue in successful
as well as failed relationships (Arino & de la Torre, 1998). A parallel stream of
2
research considers alliance as an option for future expansion or withdrawal from a
focal relationship, whose value unravels over time (Folta, 1998; Arend & Seale,
2005; Kumar, 2005). From this perspective, an option to exit from alliances creates
value by granting strategic flexibility to participating firms (Olk & Young, 1997;
Reuer & Zollo, 2005).
Academic research of the past thirty years has deepened understanding about
strategic alliances by covering a wide variety of topics including alliance formation
incentives (e.g. Ahuja, 2000; Rothaermeli & Boeker, 2008), interfirm governance
design (e.g. Oxley, 1997; Mayer & Salomon, 2006), and collaboration benefits (e.g.
Kale, Dyer, & Singh, 2002; Hoang & Rothaermel, 2005). In addition, practitioner-
oriented research on strategic alliance has also expanded significantly since the early
1980s (Bamford & Ernst, 2002; Wassmer, 2010). Firms repeatedly engage in
alliances with the same or different partners to leverage endowments with external
resources and to develop capabilities to manage cooperation (Bleeke & Ernst, 1995;
Goerzen, 2007).
As a result, prior relationships evolve into a portfolio of alliances over time,
from which firms establish capabilities and draw inferences to manage future
alliance relationships (Bamford & Ernst, 2002; Wassmer, 2010). The recent
emergence of alliance portfolio research takes the firm‘s view and analyzes how a
firm should structure and manage a portfolio of alliances. This research stream
examines topics of managerial incentives to construct bigger portfolios (Reuer &
3
Ragozzino, 2006), inter-temporal configuration of portfolios (Hoffman, 2007), and
management of network relationships and partners‘ bargaining power (Koka &
Prescott, 2007; Lavie, 2007; Lavie & Miller). The general argument is that firms can
effectively configure their portfolio of alliances in terms of the compositions of
network/partner characteristics and structural forms to achieve value creation and
enhanced performance levels (Sarkar et al., 2009). In this dissertation, I combine the
research on individual alliance choice with the portfolio analysis by examining how
the firm‘s portfolio of past alliances influences future alliance choices.
The following sections summarize three key theoretical perspectives that
formulate the basis for this dissertation, and describe motivations and key arguments
in the three essays. A brief outline then follows.
1.2 Theoretical Perspectives in Alliance Research
Three theoretical perspectives have formed the basis for most alliance
research: transaction cost economics, capability perspective, and organizational
learning theory. This section briefly summarizes each theory and connects to this
dissertation. The second chapter, Literature Review, provides more detailed
discussion of each.
1.2.1 Transaction Cost Economics
Transaction cost economics (TCE) theorizes that interfirm transactions
characterized by different attributes should be discriminatively aligned with different
4
governance structures, namely market, hierarchy, and hybrid (Williamson, 1985).
Transactions vary in uncertainty, frequency, and asset specificity, the last being the
most significant attribute that drives governance choice. Specific assets are non-re-
deployable investments made to a transaction, that is, they have little value in
alternative investment opportunities. Economic actors are assumed to be boundedly
rational and at risk of making opportunistic moves. Interfirm contracts are
unavoidably incomplete in coping with unforeseeable and unforeseen uncertainties
given bounded rationality; and values of specific assets will suffer when partners
behave opportunistically (that is, self-interest seeking with guile).
TCE considers adaptation as the central economic problem. Market,
hierarchy, and hybrid are generic governance modes that vary in attributes of
incentive intensity, administrative control, contract law regime, and adaption types
(Williamson, 1991). These attributes determine that comparative costs in adjusting to
disturbance in a transaction also vary across governance modes. In market, partners
have low bilateral dependence, and have high-power incentive to autonomously
adapt under price mechanism. In hierarchy, cooperated adaption is more cost-
effective because high-power incentive is traded off by reduced opportunism under
administrative control. Hierarchy is costly to set up, but it has ―adaptive advantage‖
resolving consequential disturbance (i.e. deviation from efficiency) than autonomous
adaptation in market when bilateral dependency between transacting parties
5
―progressively builds up‖ (Williamson, 1991: p. 279). Hybrids, such as franchise and
alliance, falls between the two end forms.
Differences among transaction attributes and governance forms determine
comparative advantages of each governance form in economizing transaction costs,
including searching, negotiating, and enforcement costs. Thus, TCE theorizes the
boundary of the firm as discriminating alignment: At low level of asset specificity,
market (or ―buy‖) is the most cost-effective form to organize transaction; as asset
specificity increases, hybrid replaces market in minimizing transaction costs; when
transacting assets are highly specific, hierarchy (or ―make‖) becomes the dominant
form in economizing transaction costs.
Early research on alliances was challenged by a wide variety of taxonomy on
interfirm alliance, such as joint venture (Hennart, 1988), research consortium (Olk &
Young, 1997), and keiretsu (Dyer, 1996). Oxley (1997) classified alliance forms in a
similar vein to the market-hierarchy continuum according to governance attributes of
incentive intensity, administrative control, and contract law. In all these studies,
scholars have commonly viewed alliance as a form of hybrid governance and
analyzed alliance governance decisions based on TCE logics. In fact, Oxley and
Silverman (2008) observed that TCE perspective ―has brought significant discipline‖
to studies on alliance organizational structures (p.212).
Focusing on the transaction or type of transaction (of which an alliance is one)
as the unit of analysis, TCE views governance form as efficient in economizing
6
transaction costs using comparative statics. This method has been criticized as overly
static (Langlois, 1992). Indeed, Williamson (1999) reckoned that TCE can benefit
from more dynamic constructions, for instance ―pre-existing strengths and
weaknesses‖ (p. 1104) from previous experience. This dissertation responds to the
call for integrating the learning perspective into TCE by examining experience
characteristics derived from prior alliances. Specifically, the assumption that
economic actors are farsighted is challenged: firms will have the capacity to
recognize contractual hazards and investment opportunities after they develop certain
knowledge from prior experience. Foresight is important in evaluating governance
forms, but has rarely been tested in empirical work. Unpacking firm experience
complements TCE in a dynamic view. As Williamson (1999) has put it, ―the
requisite recognition [of hazards and opportunities] will come as a product of
experience.‖ (p. 1104)
Bringing experience characteristics into TCE can also make contribution to
the emerging research of alliance portfolios. The portfolio research has primarily
focused on direct impacts from alliance experience on firm outcomes (Wassmer,
2010), for example, the composition of partner characteristics on focal firm
performance (Lavie, 2007; Lavie & Miller, 2008). However, such a direct link
ignores the fact that previous experience first enters into future governance decisions,
and then affects firm benefits from any alliance relationship. By analyzing
complementarity between TCE and experience, this dissertation suggests that the
7
capabilities to identify contractual hazards are developed from prior experience;
these identified hazards subsequently determine alliance governance decisions.
Therefore, alliance governance is driven by intertwined factors between transaction-
level hazards and firm-level experience factors.
1.2.2 The Capability Perspective
The capability perspective has been a major theory that complements TCE in
advancing the theory of firm (Conner & Prahalad, 1996; Grant, 1996; Williamson,
1999; Argyres & Zenger, 2010). Its general thesis on firm boundary is that activities
should be brought internally if a firm has specific expertise in these activities, and
that market transactions are preferred when the firm lacks relevant expertise
(Madhok, 1996). In this view, economic actors are assumed to have bounded
rationality as in TCE, but the opportunism assumption, albeit bearing debates, is
largely ignored. Instead, this theory has an internal focus. While TCE proposes
economizing transaction cost as the source for superior performance, the capability
perspective considers the firm as a bundle of resources, whose uniqueness and
complementarity create sustained competitive advantages (Barney, 1991; Mahoney
& Pandian, 1992). Specific elements of a resource bundle may be available on
strategic factor market, but value creation is driven by resource combination and
organization that are unique to a firm (Montgomery & Wernerfelt, 1988; Rumelt,
1984). Being firm-specific and causally ambiguous, resource combination and
8
organization represent a firm‘s core competence, thus supporting sustainability by
protecting firms from imitation or replication (Prahalad & Hamel, 1990).
The capability perspective is similar to TCE in several aspects. First, both
build on the assumption of bounded rationality. At the transaction level, bounded
rationality renders all contracts incomplete. At the firm level, bounded rationality
makes inter-temporal learning important in building organizational capabilities over
time. Second, both theories recognize that rents are created by isolation mechanisms.
In TCE, transacting parties become increasingly dependent on each other when a
large numbered ex ante bidding turns into a small numbered ex post bargaining
(Williamson, 1985). And asset specificity in this ex post bargaining incurs different
levels of costs in different governance structures. In capability view, causal
ambiguity in complementary resources poses as a high barrier to keep outsiders from
perfect imitation (Mahoney & Pandian, 1992). In addition, both theories analyze
comparative advantages between market and hierarchy.
Given these similarities, Argyres and Zenger (2008) challenge the separation
of the capability view and TCE. They suggest that the simple comparative capability
view ignores intertwined dynamics between firm capability and transaction cost
factors. Specifically, the distribution of firm capabilities at any point of time is a
result of prior governance decisions to carry activities inside or outside of firm
boundaries. Further, portfolios of capabilities change over time as firms buy and sell
strategic factors on market. They offer an integrative view in that the capability view
9
identifies complementary resource and assets, and that TCE provides guidelines for
governing identified resource and assets. This proposition echoes Williamson‘s
(1999) call for more integrated perspective between transaction level and firm level
factors in firm boundary research.
The capability view has been applied in alliance studies to analyze
motivations for alliance formation and partner selection. Scholars show that these
decisions are driven by firms‘ seeking of complementary resources in forms of
market position (Ahuja, 2000), technological capability (Mowery, Oxley, &
Silverman, 1998), and social status (Chung, Singh, & Lee, 2000). For instance,
Mowery et al. (1998) showed that partner knowledge converges after alliance
formation using a sample of patent citation data: prior to alliance formation, partners
have a low level of technology overlap, but this overlap increases significantly after
alliances are implemented. More generally, heterogeneity in firm resources and
assets represents opportunities to integrate and leverage external resources through
alliance relationships (Eisenhardt & Schoonhoven, 1996; Hitt, et al., 2000; Park &
Zhou, 2005).
However, the capability view has received critiques on its key constructs.
Comparative advantages between market and hierarchy proposed in this theory have
been criticized for its efficiency criterion. Because capability theory deals with
dynamic efficiency (that is, continuously improving efficiency over time),
organizational capabilities are path-dependent and evolving (Teece, Pisano, & Shuen,
10
1997). Thus there is no equilibrium state to which efficiency of different governance
forms can be benchmarked. Similar to this efficiency problem, the definition of
resource is criticized for being tautological because resources are defined as anything
that creates value for the firm (Preim & Butler, 2001). In sum, there exists a real
challenge in trying to establish criteria for judging efficiency in the dynamic
framework of the capability perspective.
In addition, alliance research using the capability view has primarily focused
the difference between no alliance (i.e. market) and alliance, that is, the motivation
for alliance formation (e.g. Baum et al., 2000; Gulati & Westphal, 1999). More
recent studies have examined the design of alliance governance as driven by
partners‘ technological capabilities (Colombo, 2003) or by relational capital (Reuer
& Arino, 2007; Hagedoorn & Hesen, 2007). However, it has yet to theorize the
comparative advantages among governance forms, that is, which governance
structure is comparatively more efficient than others. This may be attributed to the
capability logic that firm is more efficient than market in coordinating collective
learning (Foss, 1996). When comparing learning benefits across organizational forms,
however, the comparison should fall among comparable learning alternatives or
alternative alliance governance modes. Williamson (1999) commented that the
comparison should be ―between (at least) two firms (supplier and buyer) and one
firm (produce internally)‖ instead of between market and internal learning (p. 1097,
italics original).
11
To understand the effect of alliance experience on firm performance, Kale,
Dyer, and Singh (2002) conducted a study to examine the relationship between
experience-based alliance capabilities and firm performance. Using a survey from
alliance managers, they found that firms who built a dedicated alliance function from
prior alliance experience would achieve greater stock market returns and managerial
satisfactions. This study proposes the concept of alliance capability, which is a firm-
level capability derived from experience. This dissertation takes a similar vein,
suggesting that alliance experience creates capabilities that influence governance
design and enhance firm performance.
Williamson (1999) also noted that the capability view need to ―define and
dimensionalize‖ resource. Dimensions of capabilities can be examined jointly with
transaction hazards to enrich both theories. In fact, scholars have shown that in
interfirm relationships, technological capability (Mayer & Salomon, 2006),
contracting capability (Argyres & Mayer, 2007; Mayer & Argyres, 2004), and
managerial capability (Mesquita, Anand, & Brush, 2008) can complement TCE in
predicting governance decisions and can be developed over time. This dissertation
joins these studies and analyzes from a different angle that when firms have
sufficient capability to foresee contractual hazards. Firms derive selection
capabilities—capability to make rational comparison among different forms—not
only through more alliance relationships, but also through diverse experience in
governance forms. This argument challenges the generic hypothesis of ―the more
12
resource the better‖ in the capability perspective, and responds to Williamson‘s
(1999) call that whether ―more resources are really better than less should be judged
comparatively‖ (p. 1098).
1.2.3 Organizational Learning Perspective
Learning is a characteristic of adaptive organization. Actors are assumed to
be boundedly rational but able to recognize changes from external and internal
environment and adapt accordingly (Levinthal & March, 1981). With cognitive
limitations, organizations continuously adapt to environment changes to survive and
grow (Huber, 1991). Huber (199) has noted that the learning literature covered
diverse but relatively less integrated topics. However, this literature can be generally
classified into two broad topics: how firms learn and their learning outcomes.
In terms of learning processes, firms can learn from own experience or from
vicarious observations (Baum, Li, & Usher, 2000; Shaver, Mitchell, & Young, 1997).
Experiential learning improves internal operation efficiency (e.g. Argote, Beckman,
& Epple, 1990) and develops path-dependent capabilities that are difficult to imitate
or replicate by outsiders (Eisenhardt & Martin, 2000). Meanwhile, the path-
dependency in experiential learning may generate biased knowledge that prevents a
firm from further absorbing new information (Huber, 1991). Vicarious learning on
the other hand is more cost-efficient and time-efficient by skipping the process of
internal development (Haunschild & Miner, 1997). However, this learning process
13
encounters the problem that certain experience or knowledge in one organization
may not be applicable to another (Beckman & Haunschild, 2002).
Firms can also learn from exploitative or explorative knowledge. Learning-
by-doing is one form of exploitation, in that firms repeat certain production activities
and learn from highly focused knowledge to exploit knowledge in subsequent
productions (Argote et al., 1990). The benefits of experiential learning primarily
include in-depth knowledge and thus operational efficiency. Exploration represents
learning about knowledge in distant areas from a firm‘s current knowledge set
(March, 1991). Distant knowledge creates the capacity to continuously absorb new
knowledge to enhance firm competencies (Cohen & Levinthal, 1990). In addition,
exploitative learning creates short-term benefits that facilitate product development,
and explorative learning encourages long-term benefits of creative and innovative
outcomes (Lavie & Rosenkopf, 2006; Rothaermel & Deeds, 2004). Throughout the
learning literature, adaptive learning is identified as a key source for organizational
capabilities, and is closely related to the capability view (Zollo & Winter, 2002).
Given that individuals are boundedly rational, learning scholars recognize
that learning at the individual level encounters myopia and biases, and that these
limitations are generalized to the organizational level (Levinthal & March, 1993).
Organizations, bounded by cognitive limitations, tend to simplify and specialize in
their learning processes. Thus myopia arises when long run goals, big pictures, and
prior failures are ignored. When firms repeatedly draw inferences from experience,
14
learning in one domain (specific to this experience) is enhanced, but in the long run
there is the risk of decayed adaptive capabilities in other domains, or ―competency
trap‖ (Barnett & Hansen, 1996). With focused experience in a domain, firms are
prone to generalize domain-specific knowledge to other domains. The mis-
application lowers the chance of success in new domains. Biased learning thus leads
to a history of biased decision making (Levinthal & March, 1993).
Recognizing these limitations in learning, March (1991) proposes that
learning which is cross-domain or explorative generates value. Because learning
focused in one area (exploitation) can refine knowledge quickly, it is beneficial in
the short run but may miss new ideas important in the longer run. Explorative
learning, on the other hand, crosses boundaries to incorporate new ideas in a way
that fosters creativity and innovation, especially over the longer run. This argument
has been supported by studies on new product development (Rothaermel & Deeds,
2004) and innovation (McGrath, 2001). Capabilities to balance exploration and
exploitation in organizational learning thus determine sustainability of competitive
advantage.
The learning perspective adopted in alliance research has focused primarily
on benefits from alliance relationships. A common theme is learning-by-doing: firms
accumulate alliance-relevant knowledge from prior alliance experience, and become
more effective and efficient in managing and benefiting from future alliances (Anand
& Khanna, 2000; Kale, Singh, & Perlmutter, 2000; Sampson, 2005). Extending the
15
learning perspective to the dyadic level, scholars analyze a tension between
cooperation and competition in alliances (Das & Teng, 2000). Partners have private
interests in appropriating value from a relationship, but also share common interests
to promote alliance success, that is, expand the size of the pie (Khanna et al, 1998).
From the information sharing view, openness and information sharing are critical for
effective learning between partners; but these factors simultaneously incur leakage of
proprietary assets and knowledge (Oxley, 2003). Unfortunately, such tension and its
dynamics have not received strong empirical support.
Organizational learning studies suffer from a similar critique as does the
capability view in that the definition of learning is vague: if any behavioral change in
an organization is considered learning, then the effect of learning is difficult to
ascertain. Further, while research has shown that learning-by-doing enhances
organizational efficiency and promote competitive advantage (e.g. Argote et al. 1990;
Tsang, 2002), we have limited knowledge about the boundary conditions of learning.
For example, when can firms recognize changes in environment? And when will
learning become more or less effective?
This dissertation attempts to address these questions. I focus on the effects of
variation in a firm‘s experience in diverse governance forms. Governance experience
represents a form of explorative learning that spans across various adaptation types.
Diverse governance experience enables a firm to readily adapt to contingencies that
may require different autonomy/cooperation. Thus governance diversity in prior
16
alliances forms a channel to learn about what to plan for and what not to plan for,
and develop firm capabilities for future relationships (Argyres & Mayer, 2007). In
addition, governance experience represents a new form of experience that has yet to
be examined in the learning perspective.
This dissertation combines above three theoretical perspectives in three
individual essays. The following section briefly summarizes motivations and
findings in the essays.
1.3 Motivations and Key Findings
This dissertation focuses on governance structures in prior alliance
experience (briefed as governance experience). The following essays analyze two
research questions about governance experience. The first question is how the
experience characteristics and contractual hazards individually and collectively
influence alliance governance choices. Second, how the combination of experience
depth and breadth influences market reactions to new alliance formation. These
questions are examined in two empirical essays, which are briefly summarized below.
1.3.1 Essay One
TCE suggests that alliance governance structures should be aligned with
levels of contractual hazards in a relationship in order to minimize transaction costs
(Oxley, 1997). However, for each firm, it makes governance decisions by
simultaneously examining transactional hazards and its experience-based capabilities
17
with respect to the types of governance modes. All else equal, capabilities constrain
or enable a firm‘s choice of TCE-predicted governance forms. In particular, this
essay examines prior experience in similarly or differently structured alliances as
generating capabilities to manage interfirm relationships. These capabilities affect
both the seriousness of potential hazards in a new alliance and a firm‘s ability to
foresee the hazards.
Essay One focuses on experience-based capabilities as a key factor that
determines alliance governance. Alliance governance forms, similarly arranged
along a continuum of the hierarchical power as the market-hierarchy continuum,
vary in their attributes of incentive intensity, administrative control, and applicable
contract laws. Following the TCE logics, alliances that involve appropriation risk,
measurement difficulty, and task complexity are under greater hazards of losing the
value of partners‘ specific assets invested in a focal relationship. Thus, these hazards
increase the likelihood that a firm will choose more hierarchical governance forms.
Meanwhile, firms also consider their history of alliance forms when making new
governance decisions. Given the value created from past experience in managing
specific governance forms, firms are likely to make path-dependent choices to
exploit the value embedded in experience with specific governance forms.
Alliance experience may consist of focused or diverse governance forms.
When a firm engages in diverse alliance structures, or has governance diversity in
experience, it is likely to have developed knowledge about different collaborative
18
norms and potential hazards governed by different organizational structures. I
propose a moderating effect of governance diversity on TCE predictions. An implicit
assumption in TCE is that firms always have the capability to see contractual hazards
and know how to mitigate the hazards by structural arrangements. However, such
capabilities are not necessarily inherent to firm endowments. Instead, it takes
experience to develop the capabilities to recognize and manage contractual hazards. I
argue that diverse governance experience represents a form of explorative learning,
which allows a firm to identify possibly alliance-related risks. Thus, the governance
diversity positively moderates the relationship between contractual hazards and
governance choices.
Governance diversity also expands a firm‘s awareness of alternative
governance forms. With focused governance experience, the firm develops
specialized knowledge and skills that pertain to the focal governance form.
Consequently, the firm is more likely to exploit such knowledge in similarly
structured alliance governance. However, focused and exploitative experience
confines the range of knowledge to manage alternative structures. For example, a
firm may choose a structure that it is most familiar with based on its experience,
even though this structure is misaligned with contractual hazards, such as excessive
bureaucracy and uncontrolled opportunism. Therefore, governance diversity weakens
the path-dependency in alliance governance decisions.
19
In sum, governance decisions are determined by experientially developed
capabilities, in addition to contractual hazards or prior alliance forms. This essay
focuses on the foresight of managers as a result of alliance experience in different
governance structures. Using a sample of interfirm alliances formed by US software
companies, I find strong supports for the above arguments. Specifically, alliance-
level contractual hazards and firm-level alliance experience influence governance
decisions. More interestingly, the diversity of governance experience strengthens two
out of three measures of hazards—measurement difficulty and alliance task
complexity, but not appropriability risk. As diversity increases, the propensity of
choosing similar governance forms as in prior experience is weakened.
1.3.2 Essay Two
Governance experience affects not only governance design, but also market
reactions upon the announcement of a new alliance. Prior alliance research has
shown a generally positive effect of alliance experience on market value creation
(Anand & Khanna, 2000; Kale, Dyer, & Singh, 2002). However, measuring
experience by a sum of total alliances glosses over two important dimensions:
breadth and depth. Depth refers to a firm‘s experience in the governance mode of a
focal (in this essay, the newly announced) alliance; breadth refers to the diversity of
governance forms with which the firm has prior experience. Combining these
dimensions creates a two-by-two matrix, in which each cell represents a type of
experience portfolio and draws different market responses
20
The following figure describes the two-by-two matrix to be examined in this
essay. Quadrants I and II illustrate a typical argument in alliance experience research:
more prior alliances draws more positive market reactions when a firm announces
new alliance formation (e.g. Anand & Khanna, 2000a; Arend, 2004; Kale et al.,
2002). Instead of general alliance experience, here I focus on past alliances governed
by the same structure as in the newly announced alliance. Structure-specific
governance experience fosters firms‘ governing capability, which is specific to
norms and routines in the focal structure. Governing capabilities are more relevant
and directly applicable to a new alliance with the same structure. Thus it is predicted
that the market will positively evaluate this capability.
Taking into account the governance diversity in experience, as discussed in
Essay One, firms will develop capabilities to identify potential hazards, or the
selection capability. Selection capabilities enable a firm to be more aware of
structural alternatives, and make more appropriate governance decisions. With better
informed governance design, the effect of governing capability is strengthened.
Therefore, the increase from Quadrants III to IV demonstrates a steeper slope than
that from I to II. Furthermore, the gap between Quadrants I and III is smaller than
that between II and IV, because governance diversity contributes more to
experienced firms than to inexperienced ones.
21
Figure 1.1 Dimensions of Alliance Experience
Experience in-type: experience in a
focal alliance governance form
Low High
Governance diversity
in experience
Low I: -- II +
High III: - IV: +++
Furthermore, the market perceptions of experience characteristics are
bounded by alliance- and firm-level factors. The degree of market value creation
varies by the governance structure of a focal alliance, because prior experience
matters to a greater degree in collaborative than in non-collaborative relationship,
and to a lower degree when inter-partner dynamics in the post-formation stage is
much more important in determining value creation, e.g. in joint ventures. In
addition, a focal firm‘s prior performance can draw different market attentions to its
new alliance announcements. Lower performance signals the possibility that a firm
lacks necessary capabilities to manage new relationships, thus experience matters
less for low-performing firms. On the other hand, the market gives the benefit of
doubts to high-performing firms, and thus pays less attention to such firms‘ prior
alliance experience. Thus, experience factors are predicted to demonstrate stronger
effects on value creation when both alliance governance and firm performance are at
an intermediate level.
The above predictions are tested using an event study method, where
abnormal stock market returns indicate market value creation of a new alliance.
22
Drawing from a similar sample as in Essay One, I find empirical supports for the
positive effect of governing capability and the positive moderating effect of
governance diversity on market valuations. The boundary conditions of alliance
governance and firm performance are tested by splitting the sample according to
these variables. In sum, I find that the market attribute different values to dimensions
of experience by differentiating the breadth and depth of alliance governance
experience.
1.4 Dissertation Outline
This dissertation is organized into six chapters:
Chapter One, the current chapter, provides a general overview of this dissertation.
Chapter Two reviews main theoretical perspectives and their empirical research,
with a focus on strategic alliance research.
Chapter Three describes the empirical context: the software design industry, with
its industrial trends, alliance activities, and managerial insights.
Chapter Four is the first empirical essay. This essay will develop and test
hypotheses that experience characteristics influence alliance governance
decisions via path-dependency and bounded foresight.
Chapter Five is the second empirical essay. This essay tests how market reacts to
new alliance formation by firms with different experience characteristics.
23
Chapter Six is the conclusion chapter. It summarizes the main findings of the
three essays. Contributions, implications, limitations, and future research will
also be discussed.
24
CHAPTER 2: LITERATURE REVIEW
Three theoretical perspectives are particularly pertinent to strategic alliance
research: transaction cost economics, the capability perspective, and organizational
learning. This chapter reviews each theory respectively. While each perspective has a
large number of empirical studies, this section reviews those that are specifically
applicable to interfirm relationships or strategic alliances. Contributions of this
dissertation to each perspective are summarized at the end of each section.
2.1 Transaction Cost Economics (TCE)
TCE is a major perspective of organizational economics theorizing the
boundary of firms. This section reviews its assumptions, main arguments, and
implications. Research on strategic alliances that draws from TCE is then
summarized. This section concludes with the contributions of this dissertation to
TCE research.
2.1.1 Overview
The main thesis of TCE is that transactions that vary in attributes should be
governed by organizational forms that vary in costs and competences (Williamson,
1991). In a world with positive transaction costs an uncertainty, transactions need to
be governed to execute effectively. TCE views the firm as an organizational structure
that governs transactions between economic organizations, who have bounded
25
rationality and opportunism. Given the focus on transactions, a transaction is the unit
of analysis. As developed by Williamson (1985, 1991, 1996), transaction costs
primarily arise from asset specificity, and are exacerbated by frequency and
uncertainty. Governance structures including market, hybrid, and hierarchy, vary in
attributes of incentive intensity, administrative controls, and applicable contract laws
(Williamson, 1991). These attributes determine that structures differ in their
performance of coordinated and autonomous adaptation, which is considered to be
―the central economic problem‖ by Williamson (1991: p. 277). Instead of analyzing
individual governance forms, TCE compares their relative advantages in
economizing transaction costs.
2.1.2 Behavioral Assumptions
Economic actors are assumed to be boundedly rational and at risk of
behaving opportunistically. Bounded rationality is taken from Herbert Simon‘s idea
that individuals want to be rational but are limited in their ability to do so
(Williamson, 1985: p.72). Lacking full rationality, economic actors are incapable of
writing complete contracts that are complex (Williamson, 1999). Given
incompleteness, contractual hazards are always present since contracts cannot
specify all possible contingencies. However, boundedly rational actors are assumed
to have foresights within their range of rationality. That is, they look ahead and plan
for relevant issues that they can foresee.
26
Another behavioral assumption is opportunism. Some actors are assumed to
engage in ―self interest seeking with guile‖ (Williamson, 1985: p.72). A central
aspect of opportunism, however, is that while only a certain percentage of the
population is assumed be opportunistic (likely a small percentage), the fact that
actors cannot tell with certainty whether their exchange partner is one of those who
is prone to opportunism drives them to create safeguards when selecting and
designing transaction governance. Thus there is always a probability of inefficient
consequences that are driven by opportunistic behaviors, such as adverse selection,
moral hazard, and opportunism. Given opportunism, contracts need to be governed
and to include credible commitments and safeguards to be effectively executed.
These behavioral assumptions are fundamental to TCE, because TCE views
economic organizations as means to economize transaction costs in bounded
rationality and mitigate contractual hazards that arise from the potential for
opportunistic behavior (Williamson, 1999: p. 1090).
In addition to behavioral assumptions, TCE also assumes the presence of
some level of uncertainty in external environment, such as market fluctuations in
demand and supply, technological change, and political institutions. Because of
environmental uncertainty, disagreements may arise from exchanges, as the future
situations cannot be outlined in a contract. Similar to neoclassical economics, TCE
also embraces the assumption that economic actors always seek to maximize
efficiency in decision making.
27
The combination of the above assumptions provides a scenario in which
contractual hazards tend to be present, to varying degrees, in most exchanges. First,
transacting partiers all strive to maximize efficiency in a transaction. Second, the
parties may be opportunistic, so there is a probability that one party may take
advantage of the other during the exchange. Finally, it is impossible to specify all
contingencies ex ante given firms‘ bounded rationality and environmental
uncertainty. In a world with these assumptions, TCE compares and predicts efficient
forms of governance for transactions under various degrees of hazards.
2.1.3 Transaction Costs
The foundation of TCE emerged from Coase‘s attempt to address the origin
of the firm (Coase, 1937). At the time, the field of economics primarily focused on
markets and, when it addressed them at all, treated the firm as a black box. Coase
was not satisfied with this approach, however. His investigation into why firms co-
existed with markets led him to the conclusion that economic exchanges are not free,
but in fact generate costs (p. 390). Based on this idea, Coase concluded that the firm
and the market are two alternatives to coordinate exchanges. When the costs of
determining the price of goods and services in the market were high, firms formed to
internalize such transactions and reduced these transaction costs significantly,
because contracting for each exchange would not have to be negotiated (p. 391).
Coase also examined why the market existed if firms were able to reduce
transaction costs that are normally incurred in that production coordination option.
28
He suggested three primary reasons: 1) as firm size increases, the costs of organizing
more transactions within the firm increase (p. 394); 2) entrepreneurs may become
less efficient at resource allocation as the number of transactions over which he or
she has authority increases (p. 395); and 3) supply prices of production factors may
increase as firm size increases. As a result, Coase concluded that a firm will expand
until its transaction costs are equal to those in the market.
In considering Coase‘s elucidation of the cost of exchanges, the question of
what exactly constitutes these transaction costs arises. North and Thomas (1973)
proposed three types (p. 93): search costs, negotiation costs and enforcement costs.
Search costs are those associated with finding the exchange partner, while
negotiation costs are those incurred while defining terms of the exchange and writing
the contract. Finally, enforcement costs include all fees related to enforcing the
exchange, including monitoring and litigation costs.
Williamson expanded on Coase‘s idea of the firm (also termed as hierarchy
or fiat) and the market as distinct alternatives for organizing transactions (1991). He
suggested that these two organizational forms were at opposite ends of a single
continuum (see Figure 2.1 below) with various types of hybrid contracts in between
these two extremes that could either more closely resemble a firm or a market
structure depending on their contents. Hybrid covers a wide range of interfirm
relationships, including franchise agreement, long-term supply contract, interfirm
alliances, and business consortium (Williamson, 1985). The Hybrid Contracts box in
29
the figure below was absent from Coase‘s initial conception of governance forms for
exchanges.
Figure 2.1Williamson‘s Conception of Exchange Governance Structures
2.1.4 Governance Structures
Governance forms of market, hybrid, and hierarchy differ in several key
attributes: incentive intensity, administrative control, and contract law regime
(Williamson, 1991). First, different contract laws support different governance
forms. In market, transactions are spot contracts, and disputes between parties are
heard and settled by the courts. Because parties in hybrid contracts have greater
bilateral dependency, the costs of court settlement and terminating a contract
prematurely are likely to outweigh the benefits of continued relationship. Thus
disputes often end up in arbitration than in the court to preserve the contract. Finally,
the court refuses to hear cases that arise within a firm‘s boundary. Thus internal
disputes often go through forbearance (i.e., decided by management within the firm
with the CEO, as overseen by the Board of Directors, as the ultimate authority)
instead of courts.
The tradeoff between incentive intensity and administrative controls also vary
by governance structures. The market provides strongest incentive through the price
mechanism for economic actors to adjust behaviors. Such a high-powered incentive
Hierarchy Market Hybrid Contracts
30
is traded off with enhanced administrative controls in hierarchy. Within a firm, fiat
power controls behaviors and efforts, which can lead to problems of politics,
inefficient use of resources, and subgoal pursuit. Meanwhile, internal actors are now
controlled by hierarchical structures rather than price mechanisms. So they have
lowered incentive to behave as acutely as they would to price signals in a market
relationship. Between market and hierarchy, hybrid contracts contain some degrees
of high-powered incentives and some degrees of administrative controls. Thus hybrid
falls on the intermediate range between the two polar forms. The tradeoff between
incentive and administrative power are illustrated in Figure 2.2 below.
Figure 2.2 Levels of Administrative Control and Incentive Intensity in
Governance Forms
The tradeoff between administrative controls and incentive intensity results in
different performance of coordinated and autonomous adaptation. Williamson (1991)
Market
Administrative Control
Incentive Intensity High
High
Low
Low
Hierarchy
Hybrid
31
views adaptation as the central economic problem for economic organizations. He
considers governance forms possessing differential adaptability through autonomy
and cooperation when responding to disturbances. In autonomous adaptation,
exchange partners adjust toward efficiency upon price signals. They are less
dependent on the counterparty‘s behavior or on coordinated efforts. Such
adjustments have high-powered incentive and low bureaucratic costs from
coordination. Thus the market is optimally suited for autonomous adaptation as each
firm optimizes in response to changes in prices, while hierarchy is least suited to
autonomous adaptation as the bureaucracy designed to create administrative controls
and lower incentives weakens the ability to respond quickly in these situations.
On the other end, partners in coordinated adaptation toward efficiency need
to organize bilateral efforts. Given bilateral dependency, self-interested bargaining
between partners is costly and inefficient. In this case, coordinated adaptation under
internal hierarchy is more cost-efficient to unify partner actions under mutual
dependence. However, coordinated adaptation comes at the cost of increased
bureaucracy and reduced incentive intensity. Inside a hierarchy, taking actions
require additional administrative costs, and the actions are not so tightly linked to
consequences as in market. Attributes of governance structures are summarized
below in Table 2.1.
32
Table 2.1: Attributes of Governance Structures (adapted from Williamson,
1991, p. 281)
Incentive
Intensity
Administrative
Control
Contract Law Adaptation
Performance
Market High Low Classical contract law Autonomous
Hybrid Medium Medium Neoclassical contract law Intermediate
Hierarchy Low High Forbearance Coordinated
2.1.5 Attributes of Transactions
Given the assumptions discussed in 2.1.2, TCE‘s primary goal is to minimize
transaction cost through appropriate governance forms of economic exchanges. The
novelty of this perspective lies in its focus on the transaction as the unit of analysis.
In contrast, organizational theory primarily concentrates on the organization, while
economics is mainly preoccupied with markets. Transactions differ in three primary
attributes: frequency, uncertainty, and asset specificity. Transaction costs increase
when a transaction takes place more frequent, encounters greater uncertainty or
disturbance, and contains more specific assets.
Frequency refers to how often an exchange needs to recur to complete a
transaction. Given incomplete contracts under the above assumptions, transaction
costs accrue each time when an exchange repeats, and the cumulative costs of
recurring transactions will be high. Williamson notes that frequent transactions
provide an incentive for internal organization because ―the cost of specialized
governance structure will be easier to recover for large transactions of a recurring
33
kind‖ (1985: p. 60). Thus the more frequent an exchange takes place, the greater
benefits to organize an exchange internally to minimize transaction costs.
Uncertainty means unanticipated changes or disturbances surrounding a
transaction. It can stem from either environment, or partner behaviors, or both. As
discussed in above assumptions, environmental uncertainty arises when there are
disturbances in market demand/supply, technological change, and political
environment. Behavior uncertainty revolves around partner efforts in a transaction.
When efforts are not directly observable or the outcome of a transaction is difficult
to measure, there is a probability that a partner may shirk in efforts or seek self-
interests at the cost of the other party. Behavioral hazards are shown in the forms of
adverse selection, moral hazard, and opportunism (Williamson, 1985). Uncertainty is
often associated with relationship specific investments that are made to support a
transaction. Some governance forms minimize transaction cost more efficiently than
others only when contractual hazards brought on by transaction-specific investments
are at risk (Shelanski & Klein, 1995). However, uncertainty without specific
investments does not prove one governance structure having comparative advantages
than others.
While frequency and uncertainty have been developed as key transaction
attributes that can influence governance choice, they primarily operate in conjunction
with the final attribute of transactions, asset specificity, which is the main
determinant of the choice of governance form. Asset specificity describes the degree
34
to which transaction-specific assets can be redeployed outside of the focal
transaction. Asset specificity increases as productive values of investments in a
transaction reduces in best alternative means.
Williamson (1991) classifies six types of asset specificity: site, physical asset,
human asset, brand name capital, dedicated asset, and temporal specificity (p. 281).
Site specificity is high when the physical location of an asset makes its use
implausible for another purpose. For example, building oil pipelines near drilling
facilities is one form of site-specific assets. Physical specificity refers to the physical
characteristics of an asset that has little value in alternative purposes, such as
specialized machine tools. Human-asset specificity is high when knowledge specific
to a person has little use for different purposes, such as a salesperson‘s working
relationship with his/her own organization. Brand name capital can be specific to a
brand that may lose its value in alternative brands or markets. Dedicated assets
include investments in transactions that cannot be readily used for other purposes.
Finally, temporal specificity pertains to timely responsiveness, which is at risk of
losing the time value in later actions.
Absent of specific assets, the identity of a transacting partner matters little
and a firm can switch to alternative candidates on the market. The switching cost is
low because potential partners are largely identical and because previous investments
can maintain their values in new transactions. When investments become more
specific to a transaction, bilateral dependency between transacting parties increases
35
when an exchange is repeated, since partner identity now become important. The
large pool of candidates in ex ante bidding shrinks to one or a few partners who are
the winners in bidding, or a small-number bargaining situation ex post. This shift is
termed the ―fundamental transformation‖ (Williamson, 1985: p. 61), where once
transaction-specific investments are made, a large number of identical and
anonymous exchange partners in ex ante bidding will reduce to ex post bargaining
where there are only a few idiosyncratic partners when this transaction repeats. This
shift does not occur, however, when transaction-specific investments are not made
by the winning supplier(s). In this case, the large number of potential partners
remains possible.
The conditions necessary for the fundamental transformation lead to a major
implication for TCE. When transacting over assets with varying levels of asset
specificity, governance costs differ across government mechanisms. In a small-
numbered ex post bargaining situation, coordinated adaption within hierarchical
structures is more cost-effective in minimizing transaction costs. Hierarchy uses fiat
power to monitor opportunism and hazards, which are driven by transaction-specific
investments that are always facing risks from environmental and behavioral
uncertainties. Internal deviating behaviors from efficiency are detected and penalized
by administrative power at a lower cost than by the market mechanism. This
implication is mapped in Figure 2.3:
36
Figure 2.3: Governance Choices as a Function of Asset Specificity
(reproduced from Williamson, 1991: p. 284)
2.1.6 Discriminating Alignment
The discriminating alignment argument is built upon the components
discussed above. TCE studies how transacting partners protect themselves from the
hazards associated with exchange relationships. In a complex world, all transactions
need to be governed because contracts are unavoidably incomplete. Because of this
incompleteness, firms who invest in transaction-specific assets are exposed to a
hazard that, under disturbances, their partners may try to expropriate the rents
generated by the specific assets. There is a variety of governance structures that fall
into market, hybrid, or hierarchy may be employed to protect transacting parties
from expropriation. These governance structures differ in several key attributes. And
37
their relative appropriateness (i.e. transaction cost minimization) depends on
characteristics of the focal transaction. As summarized by Shelanski and Klein
(1995), TCE ―tries to explain how trading partners choose, from the set of feasible
institutional alternatives, the arrangement that offers protection for their relationship-
specific investments at the lowest total cost‖ (p.337).
When contractual hazards are low and relationship-specific assets are absent,
the market is the least costly form to govern transactions. The market provides strong
incentives to adapt autonomously according to price information, and incurs low
bureaucratic costs. Once specialized assets are made, partners are exposed to greater
contractual hazards that the values of their investments are subject to expropriation.
More coordinated forms of governance become desirable, because the specialized
assets raise bilateral dependency between partners. Along the continuum of
governance forms, the fully integrated firm lies at the opposite end to market. In this
form, transacting partners are under unified hierarchical control. Hierarchy offers
greater protection for specific assets and relatively efficient mechanisms for
coordinated adaption when changes take place. Within a firm, authority replaces
market as the motivating and monitoring mechanism. Thus, as external trading
partners become internal actors, there is a weaker incentive that they will maximize
desired efforts (and profits) than they will in market mechanisms. Thus additional
bureaucratic costs accrue. Between the polar forms of market and hierarchy, there is
38
a variety of hybrid forms, which entail different degrees of tradeoff between high-
powered incentives and administrative controls.
As discussed above, these governance structures are compared in relative
rather than absolute terms. Rather than analyzing individual types of governance
forms, TCE takes the view of comparative analysis, and theorizes discriminating
alignment on the basis of comparative advantages. Whereas transaction costs are
positively related to contractual hazards across all governance forms (see Figure 2.3),
it is governance forms‘ relative ability in minimizing transaction costs that
determines their efficiency. The TCE perspective has motivated a large number of
empirical studies, particularly examining the decision of firm boundaries (ref).
Scholars have also begun to examine the performance implications of structural
alignment, and shown that misaligned transactional forms results in reduced
managerial satisfaction (Poppo & Zenger, 1998) and technological performance
(Leiblein et al., 2002).
Instead of an extensive review on all empirical TCE studies, the next section
reviews empirical literature inspired by TCE in the areas of strategic alliances, a
form of interfirm relationships that are particularly relevant to this dissertation.
2.1.7 Empirical TCE Research Related to Strategic Alliance
Strategic alliance is one particular type of hybrid governance structures that
fall between spot market purchases and integrated firms. Hybrid structures cover a
variety of relationships, e.g. long-term supply contracts, franchise agreements,
39
business consortia, etc. Among these forms, strategic alliance is often defined as a
form of voluntary interfirm cooperative relationships where two or more parties pool
resources to pursue specific market opportunities (Gulati, 1995; Parkhe, 1993). The
pooled resources reflect that partners make relationship-specific investments, and
that they are mutually dependent to achieve the success of alliances. Therefore, a
main issue for strategic alliance research is how to design governance structures to
protect relationship-specific assets and to achieve alliance successes. TCE is
particularly pertinent in this area.
Scholars have examined three major types of contractual hazards that are
consequential to the specific investments in a relationship. Early research (e.g. Gulati
& Singh, 1998; Oxley, 1997) has established that the appropriability risk is of
particular concerns in interfirm alliances. Alliance partners‘ collaborative outcomes
are subject to weak property rights, because the ownership of alliance outcomes is
not clearly defined on joint research inputs. This appropriation concern can reach a
very high level that alliance governance offers insufficient control power. In this case,
firms choose to reduce collaborative activities to curb the risk of value mis-
appropriation (Oxley & Sampson, 2004).
The second type of contractual hazards in interfirm relationships is
measurement costs. When partner inputs and outcomes are difficult to measure,
partners may have the incentive to shirk from a desirable level of efforts. Alliance
outcomes can be difficult to calibrate by the collaborative nature or by the product
40
features, such as the cost reduction levels of customized software (Mayer &
Nickerson, 2005). Therefore, to prevent from inferior performance, partners strive to
craft more complete contractual clauses to ensure proper efforts dedicated into a
relationship (Argyres, Bercovitz, & Mayer, 2007; Ryall & Sampson, 2009).
Finally, alliances can encounter hazards from complicated collaborative task
and from partner interdependence. The net benefits from an alliance are more
dependent on the coordination efforts when the relationship involves complex
collaborations or partner interactions (Gulati & Singh, 1998). High interdependence
between partners imposes greater risks that invested assets may lose their values
(Mayer & Nickerson, 2005). When partners are highly inter-dependent on each
other‘s effort, more hierarchical control is predicted to provide greater safeguard to
achieve the goal of collaboration (Poppo & Zenger, 2002).
To apply TCE analysis to alliance governance, it would be ideal to rank
alliance governance forms along a continuum similar to the market- hierarchy
continuum. However, this rank is much more challenging to design than that of
market-hierarchy. The diversity in alliance activities is highly demanding for micro-
analytic data on alliance contracts to compare governance instruments such as
incentive intensity, administrative control, adaptability, and contract laws. Alliances
focus on different main collaborative activities, for example, a marketing service
agreement differs from a software development agreement in idiosyncratic
governance features. Further, many alliances cover broad scopes, involving multiple
41
activities simultaneously (Khanna, 1998; Oxley & Sampson, 2004). This fact adds
complexity because it is not apparent how to rank the combinations of governance
instruments in various activities (Oxley & Silverman, 2008).
Given the difficulty in ranking alliance governance forms, early studies
mostly measure the dependent variable as a dichotomy between equity versus non-
equity forms (Gulati, 1995; Pisano, 1989, 1990). Pisano (1989) applies TCE logics in
the biotechnology sector and finds that equity linkages are preferred to pure
contractual agreements in collaborative alliances under high contractual hazards.
Using a sample of 195 collaborative projects, he shows that equity relationship is
preferred when an alliance involves R&D components, carries on multiple projects
simultaneously, and when there are few potential collaborators. Extending on the last
factor, he later shows in another study that, in a sample of 92 biotech projects made
by major pharmaceutical companies, the small-number bargaining problem
motivates firms to internalize R&D projects instead of choosing external contracts
(Pisano, 1990).
Gulati (1995) also contrasts the governance decisions of equity versus non-
equity governance forms. He uses cross-industry and cross-country data covering
2400 announced alliance over twenty years, and shows that both contractual hazards
and interfirm ties influence alliance governance choices. While this study proposes
the importance of prior inter-partner ties from the social network perspective, it also
provides strong empirical evidence that the discriminating alignment of TCE is
42
robust across industries, countries, and time. Alliances are shown to be managed
more likely by equity-based governance if they involve collaborative R&D
components, cross-border participants, and complexity of coordinating three or more
partners.
The dichotomy between equity and non-equity alliances is a rough
classification that resembles the make-or-buy decision with respect to financial
commitments. In a related study as his 1995 piece, Gulati recognizes this problem
and refines his dependent variable of governance forms into joint venture, minority
equity, and contractual alliances (Gulati & Singh, 1998). However, alliances often
incorporate collaborative efforts that are not captured by financial investments. For
example, an equity relationship may involve limited partner interactions beyond
signing the contract (e.g. minority equity in OEM partners), and a non-equity
relationship may require different degrees of interactions. Focusing on the degree of
partner incentive alignment, Oxley (1997) proposed that alliance agreements can be
clustered in discrete forms from the most market-style mode to the most firm-style
mode: unilateral contractual agreements (e.g. unilateral licensing, long-term supply
contracts, outsourced R&D contracts), bilateral contractual agreements (e.g. bilateral
licensing, joint design, co-marketing, co-development agreements), and equity
alliances (e.g. equity joint ventures). This market-hierarchy continuum has become a
standard classification in later alliance research (e.g. Colombo, 2003; Santoro &
McGill, 2005) in addition to the equity/non-equity distinction.
43
Appropriability hazard is defined as the ability to maintain control over
proprietary knowledge that generates values in cooperative relationships (Grant,
1996; Oxley, 1997). Appropriability hazards are more likely under weak property
right regime. Because values created by an alliance are made possible based on
pooled partner resources, the appropriation of such values is difficult to define and to
control. With weak property rights, partners have trouble in adequately specifying
payoff-relevant activities, monitoring the execution of prescribed activities, and
enforcing contracts through the courts (Oxley, 1997). Another source of
appropriability is the tension between cooperation and competition (Khanna et al,
1998). Partners need to cooperate to create value, but they also compete to
appropriate values accrued to their focal alliance (Das & Teng, 2000). This tension
can be exacerbated when partners are competitors outside a cooperative relationship,
such as competing for inputs in the resource market and for market shares in the
output market. Using 165 technological alliances, Oxley shows that more
hierarchical governance decisions are determined by greater appropriability hazards,
which are operationalized as transaction types (design, production, or mixed),
technological range, geographic scope, and the number of partners in an alliance.
More recent alliance studies on the governance decision start to combine
TCE with the capability perspective. The argument here is that firm-level capabilities
or experience influences firms‘ decisions on governance forms. For example, Oxley
and Sampson (2004) suggest that governance structures per se, even as hierarchical
44
as equity joint ventures, may provide insufficient protections from contractual
hazards when extensive knowledge sharing is necessary. That is, governance
decisions are complemented by the scope of alliance activities to accommodate a
firm‘s capabilities in managing interfirm knowledge flows. In this case, firms may
choose to limit alliance scopes, or the number of activities involved, to those that can
be successfully achieved under limited knowledge sharing. Their arguments are
supported in a sample of international R&D alliances in the telecommunication
industry.
Drawing from the capability view, Mayer and Salomon (2006) propose that a
firm‘s technological capabilities can drive governance decisions and mitigate
impacts of contractual hazards. Technological capabilities determine how efficient it
is for a firm to internally develop skills and capabilities that it lacks. Therefore, firms
with strong technological capabilities that are relevant to a focal transaction are more
likely to organize this transaction with internal forms due to cost-efficiency concerns.
Further, technological capabilities strengthen a firm‘s capability to govern
transactions, such as assessing partner ability and efforts, identifying appropriability
risk, monitoring cooperative behaviors, and even writing more complete contracts. In
a sample of 405 IT contracts, they find evidence that weak technological capabilities
increase the likelihood of subcontracting, and that strong capabilities mitigate the
effect of hold-up hazards on choices of internal forms.
45
Aggregating from the transaction level to the firm level, Aggarwarl and Hsu
(2009) examine characteristics of a firm‘s portfolio of cooperative alliances on the
portfolio-level alliance governance forms. They argue that a firm‘s knowledge (i.e.
patent) pool and prior partnering experience represent its appropriation environment
from which contractual hazards may raise concerns in collaboration. When a firm
has a large patent pool and a higher level of collaborative experience, the firm is
likely to have technological and collaborative capabilities to manage R&D alliances.
Similar to the capability argument made by Mayer and Salomon (2006), they suggest
that such capabilities reduce the need to control hazards with more hierarchical
governance. As a result, the firm‘s alliance portfolio will contain a smaller
percentage of equity alliances. In addition, prior experience in either equity alliance
or licensing agreements generates path-dependent governance decisions, that is, there
will be more subsequent equity or licensing agreements, respectively, in a firm‘s
alliance portfolio (as percentages). This study uses a panel data of biotechnology
startups and finds general supports.
The above studies collectively demonstrate that firm-level capabilities help
discerning contractual hazards and potential risks, thus influence governance
decisions. While these studies show that firms indeed choose governance forms to
protect transactions from contractual hazards, they provide little evidence for the
consequences of (mis-)aligned governance forms. TCE suggests that transactions
that are managed by misaligned governance are deviations from the cost-efficient
46
equilibrium and result in inferior outcomes, thus choosing the right governance form
is first-order economizing of costs. TCE scholars thus analyze the outcomes of
alignment problems. With respect to contract negotiation, Reuer and Arino (2002)
examine post-formation contractual renegotiations from a survey of 91 Spanish
manufacturing alliances. They use two-stage estimations, first on governance form
decision, and then on the probability of renegotiation. Their results support that
alliance agreements that are initially misfit with asset specificity are more likely to
be renegotiated in post-formation stages.
With respect to firm performance, Silverman and Nickerson (2003) use a
large firm-year panel-data in the for-hire trucking industry, and show that firms who
have experienced misaligned structures in prior transactions realize lower
profitability than firms with better aligned transaction experience. Combining with
the organizational change perspective, they also find that firms with misalignment
make effort to adapt or ―re-align‖ toward equilibrium. They examine costs of
adaptation, and find the rate of realignment is constrained by a firm‘s specific assets,
union presence, employee replacement, and firm history.
Sampson (2004) uses a sample of 464 R&D alliances in the telecom
equipment industry to examine alliance-level misalignment and aggregated firm-
level performance implications. While alliance governance serves to mitigate moral
hazard concerns, governance forms that are misfit with hazard levels may limit
collaborative benefits. Two specific types of problems are pertinent to misalignment:
47
excessive bureaucracy and uncontrolled opportunism (that is, insufficient
governance). Sampson shows that, while misalignment is harmful in general, firms‘
innovation performance measured as patenting activities suffer to a greater extent
under excessive bureaucratic costs than under insufficient governance control. This
finding is reasonable for innovation performance because excessive bureaucratic
structures are likely to stifle innovative potentials, a problem absent from
uncontrolled opportunism.
These alignment studies provide important empirical supports for the TCE
argument that structural alignment is a first-order economizing problem. In
collection with studies on alliance governance design, TCE contributes to strategic
alliance research from three aspects. First, alliances are an important hybrid form of
organizational structures that govern interfirm relationships. The choices of alliance
governance represent a key strategic decision that directs collaborative relationships
(Pisano, 1989, 1990). Alliance governance forms can be characterized along a
continuum varying in the degree of equity involvement (Gulati & Singh, 1998) and
the degree of partner incentive to collaborate (Oxley, 1997). Among the pool of
alliance governance structures, firms choose the appropriate modes to protect their
relationships from moral hazards that arise in post-formation dynamics.
Second, moral hazards in alliances can arise in different types, such as
appropriation of alliance-generated values (Aggarwal & Hsu, 2009), complexity or
scope of alliance activities (Oxley & Sampson, 2004), and observability of partner
48
incentive, efforts, and outcomes (Mayer & Nickerson, 2005). Grounded in TCE
framework, these hazards are shown to increase the likelihood that transactions are
governed by internal contracts or equity-based agreements, rather than by external
subcontracting or contractual agreements.
In addition, interfirm relationships that are governed by misaligned
governance forms are harmful to firm performance (Silverman & Nickerson, 2003).
The liability of misalignment lies in inefficient transaction costs that are driven by
either excessive bureaucracy or excessive opportunism (Sampson, 2004). Firms
recognize the costs of misalignment, and adjust toward efficient forms that minimize
transaction costs (Reuer & Arino, 2002). Whereas such adaptive processes are
constrained by various factors (Silverman & Nickerson, 2003), the adaption per se
implies that the TCE prediction of discriminating alignment represents first-order
economizing strategies.
Finally, a more interesting contribution is the integration of TCE with the
capabilities perspective. The above empirical studies demonstrate, to various degrees,
that firm-level capabilities affect alliance governance decisions, for example,
managerial capabilities from prior interfirm experience (Reuer & Arino, 2002;
Sampson, 2004), technological capabilities in specific areas (Mayer & Salomon,
2006), and governance capabilities from combined technical and alliance
experiences (Aggarwarl & Hsu, 2009). These studies respond to Williamson‘s (1999)
call for more integrated theories between TCE and the capabilities perspective to
49
further understand the variance of governance forms. Joining in this integrative
research stream, this dissertation examines the collective effects of contractual
hazards and alliance experience. The next section summarizes contributions of this
dissertation.
2.1.8 Contributions of Dissertation
The first empirical essay in this dissertation draws from TCE and the
capability perspective, and analyzes how firm-level alliance experience influences
alliance-level governance decisions. Drawing from prior research on alliance
governance forms, this essay suggests that contractual hazards and alliance
experience affect governance decisions individually and collectively. Prior alliance
research that tests the main effect of experience often captures experience by
counting the total number of past relationships, but has found inconsistent effects of
experience on governance design (Oxley, 1997; Reuer & Arino, 2002; Sampson,
2004). This inconsistency may be due to the coarse measure of the sum of past deals.
The sum neglects specific aspects of experience which firms draw inference from
when choosing governance forms. To further understand experience characteristics,
this essay focuses on a firm‘s governance experience, or experience in alliance
governance forms, and analyzes two orthogonal dimensions: experience in specific
governance forms (depth) and the diversity across different governance forms
(breadth).
50
On one hand, in-depth experience in any specific governance form may
create path-dependency that leads to choices of similar or even same governance
(Teece et al. 1997). For example, extensive experience with equity alliances is likely
to develop knowledge and routines that are specific to equity structures, but not to
non-equity agreements. Firms that are specialized in equity alliances may find it less
costly to manage a new relationship with the same structure again (Aggarwarl & Hsu,
2009). On the other hand, firms may participate in differently structured alliances
and develop diverse governance experience. Since experience generates firm
capabilities, the interesting question to ask is: how do capabilities derived from the
diversity of governance forms affect a firm‘s governance decisions that are
determined by contractual hazards and firm history?
This essay argues that the diversity of governance experience reflects a firm‘s
capability of adapting to various contractual disturbances. Because experience helps
firm to learn to manage relationships (Mayer & Argyres, 2004), diverse experience
with governance forms expands firm knowledge about collaborative forms, or
alternative governance structures. The implication of this expanded knowledge is
that diverse knowledge weakens the reliance on historical governance decisions.
That is, the path-dependency of governance design is weakened by the diversity of a
firm‘s alliance governance experience. More importantly, governance diversity
creates knowledge about how alternative governance forms may or may not work.
Firms with diverse governance experience thus can perceive a broader range of
51
interfirm opportunities, can evaluate investment opportunities with greater accuracy,
and can choose governance forms that are better aligned with perceived contractual
hazards. In sum, governance-related knowledge that is sufficiently diverse generates
capabilities of foresights and enables more informed alliance governance choices.
This essay contributes to TCE by analyzing the role of experience in building
firms‘ capabilities to foresee contractual hazards. An implicit assumption in TCE is
that managers are capable to foresee hazards in exchanges and to adapt different
structural arrangements to control these hazards. However, Williamson (1999)
recognizes that TCE ―assumes that economic actors have the capacity to look ahead
and recognize contractual hazards and investment opportunities. Often, however, the
requisite recognition will come as a product of experience.‖ (p. 1004) Indeed,
Argyres and Liebeskind (1999) propose the concept of ―governance inseparability‖,
suggesting that there are interdependencies between prior contractual commitments
and present governance decisions. The interdependencies are the results of prior
formal and informal commitments that are made by a firm.
Responding to this recognition, this essay provides empirical evidence that
foresight, instead of being a natural ability to all actors, are indeed a result of
exploratory experience from diverse governance forms. Thus the ability to foresee
opportunities and contractual hazards vary across firms. Absent of diversity,
governance decisions are likely to be biased by historic decisions, which may or may
not be consistent with TCE-style structural alignment. The path-dependency in
52
governance decisions also explains the observation that some transactions may be
organized under misaligned governance structures whereas others not, even with the
same level of contractual hazards (Reuer & Arino, 2002; Sampson, 2004).
This essay also shows that path-dependency is weakened by diverse
experience. This finding suggests that there can be different types of capabilities
derived from alliance experience. To further understand firm experience and
experience-based capabilities, the next two sections reviews the capability
perspective and the learning theory, respectively.
2.2 The Capability Perspective
The capability perspective considers organizational level capabilities as the
key strategic resource determining competitive advantages. Because this perspective
is closely related to the resource-based view, this section begins by reviewing the
arguments of the resource-based view, including its main assumptions. Next, the
knowledge-based view is briefly summarized. The final part reviews the argument of
the capability view that focuses on firm capabilities as a particular form of resources.
Empirical research that applies the capability perspective then follows.
2.2.1 The Resource-Based View (RBV)
The resource-based view (RBV) has emerged as an important theory
explaining persistent firm-level performance differences, that is, why some firms
consistently outperform others. RBV suggests that sustained competitive advantages
53
are driven by firms‘ internal resources, such as physical inputs, managerial expertise,
and technological knowledge (Penrose, 1959; Wernerfelt, 1984). For example,
Wernerfelt (1984) argues that resource scarcity in product or input markets leads to
persistent superior performance. Barney (1986) suggests that imperfections in the
market for strategic factors determine that firms who have access to strategic
resources will have higher levels of economic performance. He later (1991)
summarizes the characteristics of resources that can generate sustainable above-
average returns: valuable, rare, inimitable, and non-substitutable (VRIN).
RBV shares two behavioral assumptions with TCE: firms are assumed to be
profit-maximizing and boundedly rational (Rumelt, 1984). Therefore, managers lack
the knowledge and skills to plan for future contingencies. While TCE analyzes a
transaction as a unit of analysis, RBV considers a firm as the unit of analysis, and
examines firms‘ resources or resource bundles as determinants for competitive
performance. In addition to these assumptions, RBV is similar to TCE in that they
both consider imperfect competition as a source for competitive advantage. That is, it
is the frictions in the market that enable some firms to outperform others. RBV
differs from TCE in that it tackles firm-level sources of competitive advantage from
internal factors (i.e. resource owned by firm), while TCE proposes to minimize
transaction costs under contractual hazards at the transaction level. Overall, RBV
contributes to strategy research with its focus on resource and capabilities of the firm.
54
Two assumptions about resource underlie the RBV framework: resources are
heterogeneous and are imperfectly mobile across firms. Heterogeneity suggests that
competing firms possess different bundles of resources. Firm resources differ in
initial resource endowments as well as in subsequent efforts and effectiveness to
acquire, integrate, and leverage resources. Immobility suggests that resources cannot
travel across firm boundaries smoothly and effectively. The main reason is that
resources are embedded in institutional, social, and organizational contexts, thus
values of resources to one firm may be lost to another. For example, standard forms
of knowledge may generate benefits to various degrees when deployed in firms with
different tacit knowledge. Because resources are heterogeneous, those who can
capitalize on superior resource tend to outperform rivals. When such resources are
imperfectly mobile, performance gaps will persist over time.
The key argument of RBV is that firms gain competitive advantages if they
can acquire and organize ―a bundle of unique resources and relationships‖ (Rumelt,
1984, p.558). Rumelt (1984) views a resource bundle as containing firm resources,
activities, and assets that are complement to each other and are combined to generate
economic rents. Resource complementarity (or ―super-additive productivity‖ among
input factors, as noted by Montgomery and Wernerfelt, 1988) means that the
combination of resources creates greater benefits than the sum of productivities of
each individual resource. While resource elements may be observed or imitated, it is
55
the capabilities that a firm uses to combine and deploy these resources that determine
its competitive advantages.
Moreover, firm resources and capabilities impose causal ambiguity to
outsiders (Dierickx & Cool, 1989). Resource bundles and internal capabilities are
often the outcomes of firm-specific history and are embedded in institutional and
social environments. Thus outsiders are unlikely to comprehend the specific linkage
between observable inputs or strategies and performance. Even if the linkage is
perfectly understood and replicated, it will be costly to replicate the entire
development path of a target firm. Therefore, when valuable, unique, and non-
substitutable resource bundles of a firm collectively create temporary competitive
advantages, the firm can enjoy sustainability of competitive advantages when its
resources are inimitable due to causal ambiguity.
RBV considers the resource bundle covering a variety of firm-level elements:
physical resources, operational activities, managerial skills, and other assets (Barney,
1986; Penrose, 1959; Wernerfelt, 1984). However, this definition of resource brings
the problem of tautology: it could be argued that anything that creates value can be
defined as resource (Priem & Butler, 2001). Relatively specific types of resources
are identified from the overarching pool of resources, particularly in the forms of
firm knowledge and capabilities. The knowledge-based view (KBV) follows a
similar vein as RBV, proposing that firm-specific knowledge is a form of valuable,
56
unique, and inimitable resource that determines competitive advantage (Foss, 1996;
Grant, 1996).
2.2.2 The Knowledge-Based View (KBV)
The knowledge-based view (KBV) applies RBV logic of resources and
focuses on knowledge as a particular form of firm resource. This perspective
proposes the conditions under which it is optimal to coordinate resources within a
firm instead of acquiring from the market (e.g. Conner & Prahalad, 1996; Grant,
1996). Tacit knowledge is considered the key resource that determines the efficiency
to organize transactions through the market or within the firm, given that it is
―sticky‖ and not easily transferrable across organizational boundaries (Grant, 1996).
Knowledge can be expressed in forms of codified and tacit. While codified
knowledge can be replicated and transferred across firm boundaries, tacit knowledge
such as specialized skills and know-how is accumulated from experience, embedded
in organizational memory, and is highly path-dependent. For instance, Foss (1996)
argues that the firm is a function of repository of constantly evolving tacit knowledge.
Given the stickiness of tacit knowledge, internal organizational structures are more
effective to manage knowledge flows, such as assimilating and distributing to
individuals or groups, than market structures.
KBV develops a theory of the firm on the basis of knowledge acquisition.
Focusing on knowledge flows within and between firms, Kogut and Zander (1992,
1995) suggest that the firm is superior to the market in creating and transferring tacit
57
knowledge, because the firm has a unique stock of routines and capabilities to
assimilate, utilize, and distribute knowledge. When knowledge is tacit in nature,
firms are more efficient than the market in activities of sharing, transferring, and
recombining knowledge among internal individuals and groups. Organizations with
specialized knowledge outperform market, because they leverage knowledge with
superior effectiveness and efficiency. For example, Grant (1996) proposes the role of
organizations as ―an integrator of knowledge‖ (p.377).
There has been some empirical evidence for KBV‘s organizational form
argument. For example, Almeida, Song, and Grant (2002) use patent citations of
semiconductor firms to show that multinational firms are superior to alliances and
markets in transferring technological knowledge cross-border. In the same industrial
setting, Macher (2006) shows that, with respect to problem solving related to
knowledge development, integrated firms generate faster and higher quality problem
solving than specialized firms when the problem solving is complex and mis-
structured. These studies present evidence that firms have greater effectiveness in
coordinating knowledge flows and create competitive advantages.
2.2.3 The Capability Perspective
The capability perspective expands on KBV‘s focus on knowledge to a
broader range of firm capabilities, and suggests that capabilities represent alternative
factors than transaction cost explanations in determining firm boundaries (e.g. Dyer
& Singh, 1998; Foss, 1996; Kogut & Zander, 1992; Madhok, 1996). For example,
58
Madhok (1996) argues that organizational capabilities provide complementary
explanation to the TCE logic of market failures. Instead of risks of opportunism,
Madhok suggests that capabilities under bounded rationality determine the advantage
(or disadvantages) of firms. Capabilities are examined in the forms of identifying,
acquiring, integrating, and developing unique and complementary resources, and are
proposed to be strategic determinants for generating quasi-rents.
In understanding firm boundary decisions, the capability-based view suggests
that the level of firm capabilities relevant to certain activities determine whether
these activities are organized by the firm or by the market. Grant (1996) argues that
knowledge acquisition requires more specialization than its application. When the
knowledge to be acquired is sticky across firm boundaries and is subject to
expropriation, coordination through firm structures is more effective than the market.
Thus internal knowledge and capabilities determine the means to acquire knowledge:
activities are more likely to be organized internally if a firm has superior capabilities
than outside agents. External or market structures are more likely when the firm
lacks the expertise in target activities (Kogut & Zander, 1996).
Williamson (1999) recognizes that organizational capabilities can
complement TCE research to further understand transaction governance decisions.
He suggests to ―define and dimensionalize‖ capabilities to examine the
organizational forms of transactions. Responding to his call, scholars have brought
firm capability factors into strategic decisions. Capabilities are often measured as a
59
firm‘s technological expertise and are shown to affect governance decisions (e.g.
Argyres, 1996; Leiblein & Miller, 2003; Mayer & Salomon, 2006) and partner
selections (e.g. Hoetker, 2005; Mowery, Oxley, and Silverman, 1998; Hitt et al.,
2000). For example, Leiblein and Miller (2003) show that firm-level productive
capabilities are more likely to internalize activities that are complementary to their
features than firms that lack these capabilities. Hoetker (2005) analyzes supplier
selections under various degrees of uncertainty. He shows that technical capabilities
are the primary driver for partner selection. However, the significance of technical
capabilities on the selection decision diminishes as uncertainty increases.
Current research has generally compared the relative levels of capabilities.
This ―comparative capability‖ view is re-examined by Argyres and Zenger (2010).
They argue that the distribution of firm capabilities is likely to be a result of previous
transaction cost considerations. The choice between internal development and
external acquisition is driven by transaction costs that are specific to the target, such
as searching, negotiation, and enforcing costs in market exchanges. Thus, today‘s
capabilities may well be the outcome of yesterday‘s effort in minimizing transaction
costs, and the transaction costs pertaining to a specific capability are likely to be
determined by relative capabilities between exchange partners. At odds with the
capability perspective, firms may choose to make when capability is lacking and may
choose to buy when capability is present. Here the key determinant factor is that
firms seek to own capabilities ―that are uniquely complementary to their bundles of
60
resources‖ (p. 17). This study suggests that transaction costs concerns and
organizational capabilities are intertwined processes.
In sum, the capability perspective draws from RBV and KBV logics that
uniquely complementary resources and firm-specific knowledge create a competitive
edge. Knowledge and capabilities are key competitive resources that are difficult to
imitate, costly to transfer, and highly path-dependent. Transactions are more
efficiently organized by the firm, or ―make,‖ when the firm has greater expertise in
relevant activities. The market is the more efficient choice when the firm lacks
relevant capabilities.
2.2.4 The Capabilities Perspective in Strategic Alliance Research
In alliance research, the capability perspective, along with RBV and KBV,
has been applied to topics of alliance formation, partner selection, governance design,
and performance implications. Alliance formation is considered a result of seeking
uniquely complementary resource bundles. For example, Eisenhardt & Schoonhoven
(1996) suggest that firms are motivated to acquire complementary resources from
alliance partners when they are in vulnerable positions in terms of industry
competition, technological development, and social status. Similarly, Ahuja (2000)
shows that firms that are rich in technical, commercial, and social capital stocks are
often sought after as ideal alliance partners. Park and Zhou (2005) present in a game
theoretic model that competitors form alliances when they foresee values from
resource heterogeneity and competitive dynamics.
61
These studies on alliance formation also show that firms with rich and
complementary resources are often pursued as ideal partners. However,
complementary capabilities are useful for potential partners when they are neither
too distant nor too close. Mowery et al. (1998) use patent citation data to capture
partners‘ technology overlaps, and show that allied partners have significant greater
overlap than non-partners. Whereas technological overlaps between two firms
increase the likelihood to form an alliance, this positive effect decreases at very high
levels of overlap. This finding supports the argument of resource complementarity as
well as the idea of absorptive capacity in the learning perspective. Li and Rowley
(2002) analyze the choice of prior alliance partners in a study about the syndication
formation in the US investment banking industry. They show that, beyond inertia,
prior experience and performance represent a form of dyadic capability that firms
can draw from when selecting partners. To further understand the choice between
new and existing partners, Beckman, Haunschild, and Phillips (2004) show that in
new alliance decisions, new partners are chosen for exploration and existing partners
are chose for exploitation purposes.
Alliance governance structures have been mostly analyzed by the capability
perspective in joint with TCE. TCE focuses on transaction at a micro-analytic level,
where governance decisions are driven by the alignment between transaction
attributes (uncertainty, frequency, and asset specificity) and governance attributes
(incentive intensity, administrative control, and contract laws). However, as pointed
62
out by Williamson (1991), previous adaptation experience affects how a firm will
adapt to disturbances in the future, thus capabilities and experience influence
governance decisions. Subsequent research responds to this call.
At the inter-firm level, Dyer and Singh (1998) analyze how inter-
organizational relationships create competitive advantage in a conceptual study.
They identify four key elements from the capability perspective and the governance
perspective (TCE) as important sources for competitive advantage. From TCE,
relation-specific assets and effective governance provide the structural foundation to
effectively manage and safeguard inter-organizational ties. From the capability
perspective, inter-firm knowledge sharing routines and complementary resources and
capabilities facilitate new knowledge creation and diffusion among partners. When
the four elements are present in a relationship, the partners will receive improved
collaborative performance.
Examining the impact of firm capabilities on governance structures, Mayer
and Salomon (2006) show that greater technological expertise increases the
likelihood that a transaction is managed by internal mechanisms than by
subcontracting. Moreover, technological capabilities enable greater capabilities in
governing a relationship and in designing more complete contracts, thus reduce the
impact of contractual hazards on governance decisions. Reuer, Zollo, and Singh
(2002) present an interesting contrast between technological capabilities and
relational capabilities. They find in a sample of 105 high-tech alliances that, at the
63
firm-level, technological capabilities reduce the likelihood of ex post governance
change in terms of contractual alterations and board changes. Interestingly, at the
dyad-level, prior ties between the same partners increase the chance of these ex post
changes.
At the firm-level, Aggarwal and Hsu (2009) examine the characteristics of a
firm‘s alliance portfolio over time in a panel data. Drawing from the capability view,
they find that a firm on average adopts governance forms in which they have greater
experience, thus greater governance capabilities. Experience in equity alliances
increases the percentage of equity alliances in a firm‘s alliance portfolio, and
decreases the percentage of licensing agreements, and vice versa. Their study
provides evidence for path-dependent capability development in governance
decisions.
Performance implications are examined from the view that complementary
resources strengthen a firm‘s capabilities to manage uncertainties and future
interfirm ties. For example, Baum, Calabrese, and Silverman (2000) study the
Canadian biotech startups‘ alliances, and show that forming alliances brings in
complementary resources to a startup and enhances its early performance,
particularly in the form of innovative performance. Sarkar, Echambadi, and Harrison
(2001) show that a firm‘s proactiveness in forming alliances indicates its capabilities
in identifying and responding to opportunities to acquire strategic resources. In a
sample of 182 firms, proactiveness is shown to create greater value for a focal firm.
64
More recently, Mesquita, Anand, and Brush (2008) use a survey data from 253
vertical alliances to test the different effects of the resource-based and the relational
views. They find that RBV explain performance gains from the general or average
partner level (i.e. across all partners), and that there is additional performance edge
pertaining to specific partner when the newly forged capability lies in the dyad level.
From a learning perspective, cooperative experience is considered to generate
firm capabilities to manage future alliances. This stream of research is reviewed in
the next section of the organizational learning perspective.
2.2.5 Contributions of Dissertation
The determinants of the capability perspective include resource, routines, and
capabilities. However, all these concepts are elastic constructs to measure
(Williamson, 1999). This dissertation attempts to capture capabilities in the form of
alliance experience, and to characterize dimensions of experience-based capabilities
that affecting governance decisions and market value creations.
Alliance experience is accumulated over time, and represents a path through
which firms develop a repository of knowledge, skills, and routines to coordinate
interfirm relationships. Nelson & Winter (1982) propose that experience generates
routines that are unique to a firm and capable of governing future organizational
activities. Thus experience from historic activities of the firm represents a type of
capabilities that are highly firm-specific and can never be perfectly replicated. While
65
some capabilities may be acquired from the market, experience-based capabilities are
path-dependent and can only be derived internally over time.
The first empirical essay (Chapter 4) directly tests the capability argument of
path-dependency, that is, firms are likely to choose governance forms that are
consistent with prior choices. Next, Chapter 4 proposes that the diversity of
experience fosters the capability to foresee contractual hazards and reduces the
influence of past governance decisions, which contributes to the capability
perspective by explicitly examines the breadth of experience as a source of
governing capability. Moreover, the finding that experience diversity weakens path-
dependency addresses Williamson‘s (1999) call that dimensions of capabilities can
impose various impacts on governance decisions.
My second empirical essay (Chapter 5) analyzes whether and how the market
perceives experience characteristics as indicators for capabilities. The level of a
firm‘s experience in a specific governance form represents its governing capability
specific to this form, and the firm‘s diversity of experience across governance forms
indicates selection capabilities among alternatives. In-depth experience in a new
alliance‘s governance form signals to the market that the focal firm is likely to have
sufficient expertise to manage this type of governance structure. This signaling effect
becomes stronger when the firm also has diverse governance experience, because
diversity as selection capability is perceived to lead to more appropriate governance
design. Chapter 5 presents that dimensions of experience, as indicators of different
66
types of capabilities, generate different impacts on the values that accrue to an
alliance.
Meanwhile, the extent to which governance and selection capabilities are
beneficial to firm valuation depends on the specific governance mode of the
announced alliance and a focal firm‘s performance level. Specifically, experience
factors have stronger salience to investors when an alliance is organized by
governance with intermediate levels of collaboration and hierarchical structure, and
when the firm is a mid-level performer. One implication is that the market perceives
these alliance and firms as having greater uncertainty than their respective end forms,
thus puts greater weights on alliance experience when evaluating a new relationship.
Scholars have attempted to understand sources and development processes of
capabilities in a dynamic framework. Drawing from the capability literature (e.g.
Eisenhardt & Martin, 2000; Helfat & Peteraf, 2003; Teece, Pisano, & Shuen, 1997),
this dissertation shares the view that capabilities are path-dependent, specific to firm
resources, and guided by learning mechanisms. In a longitudinal case study to
understand the learning process, Mayer and Argyres (2004) provides an interesting
examination of interfirm contracts as a repository of knowledge, and find that firms
gradually build capabilities to coordinate transactions and to write more complex
contracts as they learn from contingencies in a relationship. A dynamic view of
capabilities development is necessarily intertwined with the process of organizational
learning, which is reviewed in the next section.
67
2.3 Organizational Learning
Similar to TCE‘s view of organizations, the organization learning perspective
also considers organizations as boundedly rational and capable of adapting to
environmental changes and informational inflows. In fact, Williamson (1999)
recognizes that learning is important to organizational survival and successes
because organizations develop knowledge and capabilities to manage internal and
external disturbances. This section overviews the learning perspective, elaborates
several key areas of this literature, examines empirical research that applies learning
theories to strategic alliances, and concludes with the contributions of this
dissertation to organizational learning research.
2.3.1 Overview
The organizational learning perspective considers organizations (or specific
in this dissertation, firms) as adaptive actors to informational and environmental
changes. Learning has been broadly defined as cognitive and/or behavioral changes
of organizations, in the sense that organizations can change in contents or levels of
knowledge, and their realized or potential activities (Huber, 1991). March and Simon
(1958) first thoroughly analyzed organizational learning, suggesting that the learning
process in organizations is similar to that in individuals and includes stages of
acquiring, storing, assimilating, and implementing knowledge and information. Later,
learning was further characterized as routine-based, history-dependent, and target-
68
oriented (Levitt & March, 1988). Learning is generally considered to happen when
organizations encode inferences from history into routines to guide their behaviors.
The goal of organizational learning focuses on improving organizational
performance, in particular survival and growth (Argote, 1999).
Despite that the learning perspective was started early on by March and
Simon (1958), this perspective has yet to see an integrative framework (Huber, 1991).
Lacking a unified theoretical perspective, research in organizational learning has
mostly identified a variety of learning topics. Regarding the process of learning, or
how organizations learn, scholars suggest that organizations can learn from own
experience or observing others‘ experience (Levitt & March, 1988), from exploiting
a focal type of activity or exploring different tasks (March, 1991), and from internal
knowledge transfer among units or external knowledge acquisition between
organizations (Darr, Argote, & Epple, 1990). In terms of the outcomes of learning,
organizations are shown to improve operational efficiency by learning-by-doing
(Argote et al., 1990), to augment competitiveness by building absorptive capacity
(Cohen & Levinthal, 1990), and to enhance innovation capabilities by explorative
learning (McGrath, 2001). These topics are reviewed in the following sections.
Another component in learning research is the content of learning, or what to
learn. But content is relatively understudied in comparison to the above two topics.
Some scholars generally define the content of learning either as behaviors versus
cognition development (Fiol & Lyles, 1985) or as history versus current competition
69
(Ingram & Baum, 1997). However, the definition of content is often viewed as an
element in learning sources or processes. Thus the following section focuses on
processes and outcomes of learning. Empirical studies in each are also summarized.
2.3.2 Learning Processes
Huber (1991) proposes five sources from which organizations can replicate
activities and develop knowledge: congenital, experiential, vicarious, grafting, and
searching. Out of the five sources, experiential and vicarious learning have received
most academic attention. Experiential learning refers to the process that
organizations look back into their history of behaviors and derive knowledge and
routines from own experience. One form of experiential learning is learning-by-
doing: the more an organization repeats some activity, the more efficient this
organization will be when repeating it again. In a series of empirical studies by
Argote and her colleagues (Argote et al., 1990; Epple et al., 1991; Darr et al., 1995),
they show that production costs follow a learning curve through organizational
learning-by-doing.
In experiential learning, knowledge and capabilities can accrue from both
successful and unsuccessful experience. Organizations tend to establish routines and
formalize knowledge following prior successful or attractive experience, because
such experience creates salience in organizational memory (March, Sproull, &
Tamuz, 1991). However, excessive attention on successful experience may lead to
learning myopia, or ignorance of relatively novel, small, and even unsuccessful
70
experience (Levinthal & March, 1993). On the other hand, failures as salient events
can also impact organizational memory, and leads to cognitive and behavioral
adjustments (Levitt & March, 1988). In a case study that illustrates learning from
failures, Arino & de la Torre (1998) trace the interactions between two partners in an
international joint venture, and find that when negative shocks happen, partners
adjust to restore mutual understanding and agreement by renegotiating contract terms
or modify individual behaviors. When such positive feedback is absent, the
relationship deteriorates gradually to dissolution.
Besides experiential learning, vicarious learning suggests that organizations
can learn by observing and imitating behaviors from others. The sources of learning
may be imitating others‘ behaviors or performance outcomes, or accessing similar
information and resources from related organizations (Huber, 1991). Empirically,
Shaver, Mitchell, and Yeung (1997) examine vicarious learning as a form of positive
information externality, and show that learning from previous entrant moves in
foreign direct investment (FDI) enhances an observer‘s FDI survival rate. Beckman
and Haunschild (2002) test vicarious learning in the form that firms sample diverse
experience from their network partners. They show that learning from similar
network resources augments a focal firm‘s acquisition premiums.
Both experiential and vicarious learning processes have pros and cons. As
discussed above, experiential learning occurs internally, thus learning outcomes will
fit the organizational context and create useful internal knowledge, regardless of
71
successful or failed experience. In addition, experiential learning as a highly path-
dependent process is unlikely to be replicated by outsiders. However, path-dependent
learning may backfire because repeating a narrow range of activities can limit an
organization‘s capabilities to explore new knowledge and to create innovative
potentials (Cohen & Levinthal, 1990). Sampson (2005) shows that, in the telecom
equipment industry, too much experiential learning limits innovation outcomes from
collaborative relationships.
Vicarious learning saves an organization‘s costs and time from internal
development, because others‘ practices prove the validity and value of a learning
target. However, vicarious learning may encounter the problem of applicability.
Implementing new activities requires changes in current operational routines; and
such changes can be costly depending on organizational contexts. Further, learning
from others‘ successes may suffer the problem of survival bias; that is, backward
inducing others‘ activities based on their visible successes whereas their failures are
not observable. Haunschild and Miner (1997) use a sample of 539 acquisition deals
to show that when firms imitate others, the salience and outcomes of others‘
experience are asymmetric in successful versus failed experience.
Even if vicarious learning interprets target organizations‘ activities correctly,
replication may not create equal values for the focal one. Therefore, organizations
practice both experiential and vicarious learning simultaneously to adjust to new
knowledge. In a study of spatial moves by Ontario nursing home chains, Baum, Li,
72
and Usher (2000) show that experiential learning leads chains to replicate themselves,
and that vicarious learning leads chains to imitate visible and comparable
competitors.
Another type of learning processes can be categorized along a dimension of
how distant the target knowledge locates, or exploitation and exploration in learning
(March, 1991). March (1991) defines ―distant‖ knowledge as the learning target falls
in areas that are beyond the focal organization‘s core knowledge set. Exploitation
learning means that learning benefits accrue when organizations exploit their
existing knowledge and skills more efficiently, often through repetition. Exploitative
learning develops greater in-depth knowledge and improves performance in terms of
efficiency. In this sense, learning-by-doing is a form of exploitative learning, in
which repeated operations reduce production costs and improve operational
efficiency.
Explorative learning suggests a learning process that organizations access and
integrate information or knowledge from distant areas, for the purpose of expanding
a current knowledge pool. If new information can be integrated effectively with
current knowledge, the organization will have expanded knowledge range and
become more capable in producing creative outcomes. Importantly, the success of
explorative learning depends on whether there exist synergies between current and
distant knowledge. March (1991) argues that, in an adaptive process, exploitation is
refined more rapidly than exploration, thus is likely to be more effective in the short
73
run than in the long run. On the other hand, exploration may not receive immediate
performance benefits in the short run, as it directs resources to areas that may not be
the current focus of an organization. But exploration provides long-run benefits by
expanding organizational knowledge, and creating absorptive capacity for future
learning (Cohen & Levinthal, 1990, reviewed below).
Because knowledge can be developed internally or acquired externally,
organizations can learn from internal development or external transfer. Epple et al.
(1991) suggests that, within an organization, skills and knowledge developed from
learning-by-doing in one shift can be transferred to another. Moreover, such
internally developed knowledge can be completely transferred across shifts if it is
completely embodied in production technologies. In a study of pizza franchise stores,
Darr et al. (1995) find that multi-store franchisees manage knowledge transfer
among their own stores, but not to stores owned by other franchisees. Examining
learning from external knowledge transfer, Tsang (2002) shows that overseeing
efforts and management involvement represent two key channels for firms to acquire
knowledge from their alliance partners. Examining the flow of knowledge or skills,
this learning process is often connected with the knowledge-based view as discussed
above.
2.3.3 Learning Effects
Experiential and vicarious learning effects are reflected in multiple types of
organizational performance, such as productivity (e.g. Argote et al., 1990), strategic
74
decision (Baum et al., 2000), and survival (Kalnins, Swaminathan, & Mitchell, 2006).
Studies of experiential learning suggest unit-level productivity improves because
organizational units will become more skillful in the process and reduce average
costs of production as a production process repeats (Epple et al., 1991; Darr et al.,
1995). Vicarious information from other players in a market is beneficial because a
focal organization interprets market information from others‘ moves. As a
consequence, the organization can make better informed decisions about entry/exit,
location choices (Baum et al., 2000), and can improve its likelihood of survival
(Kalnins et al., 2006). In addition, the learning effect depreciates over time, and the
marginal returns of learning diminish as experience accumulates (Argote et al., 1990).
Exploitation and exploration learning effects have received extensive
academic efforts. In general, exploitation as a form of experiential learning enhances
operational efficiencies, for example, learning-by-doing. Exploration broadens
knowledge range, and is likely to benefit organizations in innovation purposes and in
longer runs. Learning scholars recognize that the effectiveness of learning in local
domains/short runs often hampers learning in distant domains/longer runs, and
results in learning myopia (Levinthal and March, 1993). Levinthal and March (1993)
observe two mechanisms that generate myopia: simplification and specialization.
The former indicates that organizations have a propensity to simplify experience and
to restrict sense-making to limited spatial and temporal range of actions. The latter
indicates that a learning process tends to focus on a narrow range of competences.
75
The costs of these learning mechanisms result in three types of myopia: ignore long
run; ignore the big picture; and overlook failures in history. Therefore, firms tend to
settle in domains where they possess strong competence, and to exploit knowledge
specific to these domains. When exploiting successful experience in these domains,
the successful experience will be generalized to other domains, where the likelihood
of success may be lower than expected. Overall, exploitative learning encounters the
risk of biased learning, and may generate a biased history of decision making.
Exploration benefits organizations in a different way than exploitation.
Exploration expands knowledge range, thus increases the likelihood of creative and
innovative outcomes. With respect to innovation, exploitative learning concentrates
on existing resources, and enhances domain-specific competencies. Explorative
learning spans across domains, thus likely to enhance performance in distant
knowledge domains (Hayward, 2002). With respect to product development,
exploitation focuses on local or existing products. Exploitative activities increase
efficiency in refining current products, and are likely to generate process innovation
that facilitates product development. On the other hand, explorative learning reaches
a wider range of knowledge, integrates different resources, and triggers more product
innovation. Rothaermel & Deeds (2004) find that new product development in the
biotechnology industry is more dependent on explorative effort at the beginning, and
on exploitation in later stages once the new product is going through daily operations.
76
To further understand how learning affects innovative outcomes, Cohen and
Levinthal (1990) propose an important concept of absorptive capacity. Absorptive
capacity describes an organization‘s abilities to recognize the value of external
information, assimilate it, and apply it to commercial ends. This ability is a function
of a firm‘s prior knowledge, and is critical to the firm‘s innovation performance.
Absorptive capacity draws from the cognitive psychology theory that individual
learning is easier when there is an existing knowledge base. At the organizational
level, prior knowledge and heuristics increase a firm‘s ability to assimilate new
knowledge and capabilities into memory and recall it. The broader a firm‘s prior
knowledge base is, the more efficient the firm is in making novel linkages and
assimilating new information when it comes up. Therefore, greater absorptive
capacity increases a firm‘s innovative capabilities.
While possessing a critical knowledge base is important for continuous
learning, the other key part is the ability to know where to look for resources and
knowledge. This ability requires a firm to be aware of learning possibilities,
including internal capabilities and external resources. In this sense, a diverse
knowledge set can be instrumental. However, diverse knowledge comes at the cost
of specialized knowledge, that is, a trade-off between diversity and commonality of
knowledge in an organization. Learning-by-doing in one domain increases
specialized expertise, but meanwhile reduces experimentation of alternative ideas.
Dynamically, firm investments in R&D represent the building of absorptive capacity,
77
some realized and some potential (Zahra & George, 2002). While R&D investments
may take long time to exemplify any effect, such investments build a firm‘s potential
absorptive capacity that deals with unforeseen changes. Continuous investment in
absorptive capacity is important, because the cost to catch up can be high once new
ideas are too distant from a firm‘s knowledge base to appraise or access. Thus a lack
of absorptive capacity locks a firm out from new development (Cohen & Levinthal,
1990).
While the above discuss focuses on absorbing external information,
absorptive capacity also notes the ability of transferring knowledge internally. When
organizational members have sufficient absorptive capacity, external information can
be distributed and employed more efficiently. In this sense, absorptive capacity
covers both experiential and vicarious learning. A balance between external and
internal absorptive capacity is instrumental: acquired information must have certain
similarity with current knowledge for ease of absorption and transfer, and must have
certain diversity for potential synergies from knowledge combination. External
resources that are too similar to a firm‘s own lack combinative benefits, and those
that are too different create problems for absorption.
Absorptive capacity is closely related to explorative learning. Stronger
absorptive capacity enables effect explorative learning in diverse domains. And
explorative learning expands a firm‘s knowledge base, which facilitate absorbing
new knowledge. Using a survey of 108 multinational enterprise (MNE) subunits,
78
Luo and Peng (1999) operationalize exploration as experience diversity and
exploitation as experience intensity. Both intensity and diversity of MNE value chain
experience create knowledge and enhance subunit performance. Over time, the
exploitation effect diminishes but exploration remains positive impacts on
performance. Further, explorative learning is more beneficial, as it provides
knowledge-wise cushions for external shocks, when there are various forms of
environmental uncertainties. In a later study, Lavie and Rosenkopf (2006) show that,
while organizational inertia drives exploitation in a firm‘s alliance formation
decisions, absorptive capacity increases the firm‘s exploration in alliance decisions.
The learning effects of internal and external knowledge transfer are mostly
intertwined with other aspects of learning. Internal development is a form of
experiential learning from history, and can take the form of either exploitation or
exploration. External knowledge acquisition represents an organization‘s absorptive
capacity, in that external knowledge is constructively only when the focal
organization has sufficient prior knowledge. External knowledge may be acquired by
mandatory transfers among units (e.g. Darr et al., 1995), by vicarious observation
and imitation (e.g. Haunschild & Miner, 1997), by merger and acquisitions (e.g.
Zollo & Singh, 2004), or by voluntary inter-organizational knowledge transfer (e.g.
Tsang, 2002). The last one represents a key motive for interfirm relationships, as
reviewed above in the capability perspective.
79
Notably in inter-organizational learning, there is a dilemma of knowledge
sharing: partners need to share knowledge with each other to collaborate on a mutual
understanding, but exposing knowledge to partners imposes a risk of knowledge
appropriation (Kale et al., 2000). Firms may ―race to learn‖ to acquire knowledge
that is optimal for their private interests, but at the cost of partners‘ loss. If one
partner quickly learns what it wants to know and leaves a focal relationship, its
partner may have provided proprietary knowledge without any learning benefits.
Therefore, alliance partners must balance risks and benefits of learning (Das & Teng,
2000). For example, one way to balance risks and benefits of knowledge sharing in
alliances is to limit the scope of activities involved in alliance (Oxley & Sampson,
2004). The next section reviews alliance research that employs the learning
perspective.
Overall, the learning theory suggests that firms can and do learn from
experience. A learning process contains both cognitive and behavioral adjustments.
And learning can take place in different dimensions: from own experience or from
others‘ moves, from focused experience or from diverse types of activities, and from
internal development or external acquisitions. In general, effective learning consists
of knowledge assimilation, integration, and distribution, and enhances organization
performance and innovative outcomes.
80
2.3.4 Learning Theories in Strategic Alliance Research
Scholars using the learning perspective in alliance research often adopt a
learning-curve perspective that predicts positive returns to experience. For example,
Anand and Khanna (2000a) suggest that firms draw from alliance experience to
create market values. They use an event study methodology to show that the stock
market attributes positive abnormal returns to alliance announcements made by
experienced firms. Zollo et al. (2002) propose dyadic routines as a result of interfirm
ties, and such routines can be exploited in future collaborations. Using a sample of
145 biotech alliances, they find that partner-specific experience, or a firm‘s
experience with a specific partner, improves the alliance performance between the
same dyad.
As suggested by Zollo et al. (2002), alliance experience creates capabilities to
manage existing or new relationships. Thus inter-organizational learning is also
connected with the capability view in empirical studies. For example, Kale et al.
(2002) suggests ―alliance capabilities‖ as outcomes of alliance experience. They also
show that, when alliance experience and capabilities are embodied in a dedicated
alliance function, alliances managed by this dedicated function often receive greater
market abnormal returns and overall long-term successes. While indirectly in the
learning perspective, Gulati‘s (1995) study on prior inter-partner ties lends supports
to a dyadic learning perspective. He shows that, between a pair of partners, their
previous equity alliances reduce the likelihood that they will enter into another
81
equity relationship. From the learning perspective, prior ties are a learning process
through which partners develop knowledge about each other to facilitate future
relationships.
Consistent with the learning-by-doing research (Argote et al, 1990), learning
effects from alliance experience are shown to create benefits at a diminishing rate.
Hoang and Rothaermel (2005) use project-level drug development alliances to show
that both general alliance experience and partner-specific experience of biotech firms,
but not of their partnering pharmaceutical firms, positively improves the
performance of joint projects. Sampson (2005) tests the curvilinear effect of alliance
experience in the telecom equipment industry, and shows that collaborative returns
from R&D alliances actually diminish at a high level of experience.
In an effort to examine how experience affects learning effectiveness, Tsang
(2002) surveyed over 150 Asian companies about their international joint venture
experience. He finds that knowledge is effectively transferred between partners if
there are overseeing efforts and management involvement in place. Given the inter-
firm knowledge flows, partners improve own managerial skills through learning-by-
doing and through observing their partners‘ behaviors. Interestingly, he also shows
that learning mainly revolves around key joint ventures that are either large in size or
strategically significant, whereas smaller or more novel ventures are often left on the
side. This finding provides evidence for possible learning myopia, as proposed by
Levinthal and March (1993).
82
Testing the idea of absorptive capacity, Lavie and Miller (2008) examine
alliance experience as a portfolio. They argue that, as a firm participates in more
internationalized alliances, its financial performance will improve because latent
national differences are perceived to create synergies. But the positive effect turns
negative if a portfolio is too internationalized, due to challenges on collaborative
routines and liabilities on cross-national differences. In a related study, Lavie and
Rosenkopf (2006) illustrate that absorptive capacity and organizational inertia are
two opposite factors that driving explorative and exploitative learning differently in
different domains. Another large-sample study on exploration-exploitation is
conducted by Rothaermel and Deeds (2004) in the biotech industry. They propose a
path of product development as the outcome of exploration and exploitation: because
exploration expands the range of knowledge and capabilities, explorative alliances
helps the process of creating new products. Once a new product is designed,
exploitative alliances are more instrumental in developing the product and leading to
the market stage.
Research on interfirm contracts suggest that firms learn to design complex
contracts over time as they cope with contingencies that may affect contract
specifications. As discussed above, Arino and de la Torre (1998) show how partners
mutually adapt to post-formation disturbances by renegotiating contractual terms or
adjust unilateral behaviors. Mayer and Argyres (2004) use a longitudinal case study
between two firms to show that, as the relationship evolves, the partners change their
83
contract structures over time. As such changes cannot be fully explained by changes
in relationship-specific assets, they attribute the changes as a result of mutual
learning. Interestingly, they also find that such learning is local and incremental. This
observation provides an indicator for the possibility that organizations may not be
far-sighted. In a large sample study, Argyres et al. (2007) argue that contractual
provisions in the forms of contingency planning and task description represent inter-
partner learning. They show that repeated partnerships lead to greater effort in
specifying more contingency planning in subsequent contracts.
In sum, the above empirical studies collectively support the learning effect in
interfirm relationships. Alliance experience as a learning channel provides the
benefits of experiential, vicarious, exploitative, explorative, internal, and external
learning. Firms can benefit from prior knowledge and capabilities to more effectively
design interfirm contracts, manage collaborative relationships, promote innovation
and synergies, and enhance financial performance.
2.3.5 Contributions of Dissertation
Learning studies examining alliance experience have primarily portrayed a
direct relationship between prior alliances and firm-level or alliance-level outcomes.
This link assumes that capabilities generated from experience directly improve
organizational performance. However, the benefits of capabilities are manifested
through implementations of new relationships, which are organized under
governance structures that are more informed due to experience-based knowledge.
84
This dissertation contributes to the learning literature by proposing that diverse
experience across governance forms, as a form of exploration, creates two important
benefits: reducing the chance of focal or ―biased‖ learning (Levinthal & March,
1993), and enhancing exploitative learning. These two benefits are examined in
Chapters 4 and 5.
The first empirical essay (Chapter 4) examines how governance diversity
affects alliance governance decisions. As a firm participates in various governance
forms, it develops capabilities to recognize structural alternatives (e.g. licensing,
non-equity, and equity agreements) and assess their respective risks and benefits
with greater accuracy. The accuracy means that the firm can make decisions that are
better aligned not only with contractual hazards in an alliance, but also with its own
capabilities to manage the alliance. Diverse governance represents a form of
explorative learning, which is likely to reduce myopia and localness in decision
making. Because firms have cognitive limitations (e.g. bounded rationality) and may
not necessarily have foresights, governance diversity improves their ability to see
clearer investment opportunities and adopt governance forms that are better aligned
with contractual hazards.
Chapter 4 responds to the call for integrating the learning perspective into
TCE research. In fact, Williamson (1999) reckons the role of learning in influencing
governance decisions because the capability to recognize key factors ―will come as a
product of experience‖ (p.1004). It also contributes to the learning literature by
85
examining governance experience as a form of learning. Prior studies of exploration
often focus on different contents, such as technological areas (Rosenkopf & Nerkar,
2001; Rothaermel & Deeds, 2004) and value chain activities (Lavie & Rosenkopf,
2006). This study proposes exploration in inter-organizational governance structures
as a new process of learning.
Coming from the market perspective, the second empirical essay (Chapter 5)
examines the role of alliance experience in affecting market valuation of new
alliance formation. Alliance experience is characterized into dimensions of depth and
breadth. While prior research generally suggests exploitative benefits of alliance
experience (e.g. Anand & Khanna, 2000; Sampson, 2005), this essay suggests that
broader governance experience, or exploration, can augment the benefits that accrue
to exploitation. Moreover, the benefits of integrated (i.e. interacted) exploration and
exploitation learning vary by the degree of a focal alliance‘s collaboration and by a
focal firm‘s prior performance. This study contributes to the learning literature by
testing boundary conditions for the benefits of explorative and exploitative learning.
Before presenting the two essays, the next chapter briefly introduces the
industry context in which the empirical tests are conducted.
86
CHAPTER 3: INDUSTRIAL CONTEXT – THE SOFTWARE INDUSTRY
The software industry serves as the empirical context to test the hypotheses
developed in this dissertation. The software industry has been a leading
technological force in the evolving high-technology society. Software is needed to
run the computers and networks that support the storage, management, and flow of
information in today‘s global economy. Software enables numerous products and
services, from vastly complex enterprise systems to consumer home entertainment
systems. With the proliferation of software in the economy, this sector has also
witnessed a proliferation of interfirm relationships in the past thirty years. This
chapter provides an overview of this industrial setting, reviews academic research
that draws from this context, and the appropriateness of software design industry in
testing the ideas proposed in this dissertation.
3.1 Industry Background
3.1.1 Overview
The software industry has evolved along with the computer hardware
business. In the 1960s, the first software programs to direct the operations of the
mainframe computers in use at that time were written. Software was later developed
to work with minicomputers, workstations, personal and portable computers, and
computer networks, as each of these platforms became popular. Much of the
87
software industry has entered a more mature phase of its life cycle, with overall
growth influenced by the global economy. Broadly speaking, demand continues to
rise throughout the world but at a slower pace than in the ferment period of the late
1990s. Consumption is greater in industrial countries, but there are faster growth
rates in developing markets
1
.
Since its birth in the 1960s, Software has gradually taking on an increasingly
important role in the world economy. Software in various areas provides
functionality for personal computers, business systems, network servers, internet
applications, telecommunication services, etc. The industry experienced double-digit
growth of the 1990s, but suffered a major crash in the technology market in 2000.
The crash was followed by further market shrinkage after September 11
th
2001,
when companies began to cut down on computer-related expenditures. However, as
information technology (IT) becomes an inseparable part of modern business and
lifestyle, the software industry showed its flexibility and recovered from the crash
within in a couple of year. The industry regained its market values since 2001.
1
Source: S&P NetAdvantage, Industry Surveys, April, 2010
88
Figure 3.1 Software Industry Market Value ($billion), in 2006 dollars
(Source: Datamonitor Industry Market Research, Industry Overiew 2004-2006)
The double-digit growth rate in the 1990s dropped to 8.5% in 2000, 6.2% in
2001, and -0.4% in 2002. International Data Corporation (IDC), an IT market
research firm, pointed out that 2002 is the first year ever the software industry posted
a decline. The decline was considered mostly due to weak enterprise IT spending and
lackluster results in most consumer markets. As market values grew, the growth rate
recovered in 2003 and became stable thereafter at 4-5%.
89
Figure 3.2 Annual Growth Rate of Software Market Value, in %
(Source: Datamonitor Industry Market Research, Industry Overiew 2004-2006)
The US software industry has dominated the world market. According to the
industry analysis published by Business & Company Resource Center, the US
software companies have 55.3% of the global software market in 2003, but its share
declined to 38.7% in 2006. Europe is catching up fast (from 24% in 2003 to 36% in
2006), as well as Asia-Pacific (12.2% in 2003 to 21.8% in 2006). IDC has taken into
consideration the 2008 economy downturn and forecast a steady growth path for the
software industry worldwide, reaching a total market size of $360 billion and a
growth rate of 6.9% by the year of 2013.
90
Table 3.1: Worldwide Packaged Software Revenue Forecast, by Region
Annual Revenue ($millions) CAGR
2008 2009 2010 2011 2012 2013 2008-13
Americas
144,588 147,891 154,092 163,043 173,769 185,646 5.1%
Europe, Middle East &
Africa
98,014 99,921 103,769 109,526 116,474 124,514 4.9%
Asia-Pacific
40,480 38,656 40,487 43,191 46,314 49,763 4.2%
Total
283,081 286,468 298,347 315,760 336,557 359,923 4.9%
Growth
7.0% 1.2% 4.1% 5.8% 6.6% 6.9% …
CAGR: Compound Annual Growth Rate.
Source: IDC‘s Forecast Summary, published June 2009.
Despite of the market crash in 2000 and the financial crisis in 2008, industry
analysts have agreed that this industry is a growth sector and remains very
competitive
2
. As internet, personal computers (PCs), and smart electronic devices
become a necessity, the market of software is expected to continue to grow. Given
the size, importance, and the stable growth of this industry since early 2001, this
dissertation focuses on alliance activities by the US software companies in the period
of 2001-2005.
3.1.2 Major Market Segments
The software industry serves three general segments: applications software,
application development and deployment software, and system infrastructure
software
3
. Application software comprises programs that perform specific functions,
such as word processing, spreadsheets, and desktop publishing. Application
development and deployment includes information and data management software,
2
Source: http://www.valueline.com/Stocks/Industry_Report.aspx?id=7178
3
Source: IDC, http://www.idc.com/research/reshome.jsp
91
application design and construction tools, application life-cycle management,
application development platforms, and middleware. These two segments are
sometimes collectively defined as applications software. Major application
companies include Activision, Autodesk, Cerner, Citrix, and so forth.
System infrastructure software comprises operating systems, operating
system enhancements, and database management systems (DBMS). Different
operating systems are required in different classes of computers, e.g. personal
computers (PCs), mainframes, servers, and so forth. According to Datamonitor,
system software has consistently accounted for approximately 55.5% of the US
software market‘s value from 2000 to 2006, and the application segments has
accounted for about 44.5% of total market value. System software companies include
key players such as Microsoft, Oracle, Symantec, etc.
3.1.3 Companies in the Data Used in this Dissertation
The major players in the software industry include familiar names such as
Adobe, Apple, Google, Yahoo!, IBM, Microsoft, Oracle, and so forth. However,
multidivisional companies with a software branch may or may not identify
themselves as primarily a software company, for reasons of market positions or
revenue shares. They report primary SIC in related but different areas, for example,
Apple Inc, IBM Corp, and HP define themselves as in the area of electronic
computers, thus fall into SIC 3571. Because their major strengths and market shares
92
are in other sectors such as computer equipments and hardware, these companies are
not included in the current sample.
This dissertation identifies companies according to their reported four-digit
primary Standard Industrial Classification (SIC) codes in CRSP. That is, a firm is
included in the sample if its primary SIC falls into 7371, 7372, 7373, or 7374. Note
that this definition excludes companies in the industry of information retrieval (SIC
7375, e.g. Google Inc.). This identification scheme is consistent with those used in
prior strategy research (e.g. Lavie, 2007). Further, alliances announcements of these
qualified companies are collected from Securities Data Corporation (SDC) Joint
Venture/Alliances database and Lexis-Nexis news reports. Given that alliance
experience is the key concept of interest here, firms that have announced less than
two alliances (including the current one) in the sample period are considered to have
too few observations to depict experience, and are thus dropped.
One caveat of the sample is that public companies that have multiple
alliances may differ significantly from smaller rivals who may or may not have rich
alliance experience. However, the analysis of this dissertation justifies the sample
selection method for internal and external reasons. Internally, visible and prominent
companies with alliance experiences are often considered attractive alliance
candidates, because they are likely to possess a large pool of physical, technical,
social capital (Ahuja, 2000; Gulati & Higgins, 2003; Stuart, 2000). Therefore,
93
companies with greater visibility and publicity are likely to attract more alliance
partners and develop more meaningful knowledge from alliance experience.
Externally, to understand how experience influences market reactions to new
alliance announcements, it is important that firms have a sufficient level of alliance
experience as signals to effect market reactions. That is, alliance experience may be
too nuanced a factor for the market to discern if a focal firm has too few alliances.
However, there is no threshold for how many past alliances are needed to create a
―sufficient‖ level of experience. Given the semi-efficient market assumption, the
market takes into account all historic information. Thus, if a firm has previously
announced at least one alliance, in addition to the currently announced one, this past
alliance will factor into market reactions when investors evaluate the firms‘ new
alliance formation. Based on the internal and external reasons, it is thus appropriate
to focus on public software companies with at least two prior alliances.
The final list of companies includes 154 companies that are presented in
Appendix 2. The 154 companies conduct business activities of designing, developing,
maintaining, consulting, and vending software. Some companies are identified by
S&P‘s classification of system versus application software. However, the
identification is incomplete here. One reason is that S&P only classifies major
players with significant visibility (e.g. size). Thus some firms were added or dropped
annually since 2001. For instance, Informatica and i2Technolgy were not in the list
before 2004. Whereas application software and system software are categorically
94
different, some companies may shift their focuses from one to anther or work on
both. For example, Adobe Systems Inc was classified as a system software company
prior to October 2006, but as an application company thereafter. This number of
companies drops to 147 in the second essay, due to a lack of stock market data for
the event study methodology.
3.2 Strategic Alliances in the Software Industry
According to interviewees who have had experience in both technical and
managerial functions in major software companies, strategic alliances are a prevalent
strategy commonly adopted by almost all companies. Whereas publicly listed
companies have been more conservative in using the term ―alliance‖ since the late
1990s, many interfirm agreements and projects are de facto strategic alliances. In
practice, it is usually the business development group within a company that reaches
out to the partners to handle the initial contacts and agreements. Larger companies
may have an alliance manager who takes over after the deal is done. Once a
relationship is in place, the product teams usually direct the business develop groups
on partner relations and provide technical details.
As alliance experience accumulates, companies and alliance managers learn
to manage these relationships through designing contracts and coordinating partner
efforts. For example, one interviewee from a major software design company
considers that ―pricing and discounts are the main component that we try to keep
95
equitable across different partners that might be competing for the same deal. So
from that standpoint the prior alliance pricing is very relevant.‖ In terms of
relationships between partners, a sales director from a large IT service company said
that ―if you agreed to something in the past that came back to bite you, then the prior
experience comes into play and you will change that term going forward. You
always learn from past mistakes.‖
However signing a careful designed agreement is only the first step. A
company may need to deploy a team of technicians and managers to its partner‘s site
to ―walk with the partner through implementation‖. In such cases, the company‘s
overall alliance experience, as well as the team members‘ skills, expertise, and even
personal ties, comes into play to facilitate the relationship-building process beyond
signing contracts. A software engineer noted in the interview that ―an agreement
doesn‘t mean success. It takes a lot of time to ‗crawl-walk-run‘ with a partner.‖
These general descriptions pertain to firm-level and alliance-level operations.
The rest of this section overviews interfirm relationships, alliance types, and inter-
temporal trends from the industry perspective.
3.2.1 Alliance Relationships
From the managerial perspective, a strategic alliance is considered a
significant relationship that formed for the purpose of mutual benefits. Steve
Steinhilber, Cisco‘s vice president of strategic alliances, describes alliances as ―a
relationship between two or more organizations that through the combination of
96
resources can create significant and sustainable value for everyone involved‖
(―Strategic Alliances: The Value of Partners‖, Cisco System, Inc., 2009). As the
industry advances, interfirm alliances have become a key strategy to integrate
external resources, to build network reputations, and to boost market shares (Stuart,
2000). In fact, the percentage of publicly traded software companies that engage in
alliances has increased from 32% to 95% during the 1990s (Lavie, 2007).
Firms engaged in alliance relationships may encounter the need for
cooperation as well as the pressure for competition. Thus alliances often face the
tension between these forces known as ―coopetition‖ (Khanna et al., 1998). An
illustrative example is the alliance between Adobe Systems Inc. and Apple Inc.
These two companies have been long-time technology partners, with a relationship
dating back to the 1980s. This alliance has fostered a strong and mutually beneficial
relationship, especially in advancing developments in professional desktop
publishing. However, their relationship has been marked by cooperation and
competition for some time. Most recently, Apple announced that it would exclude
Adobe‘s Flash from both Apple‘s hugely successful iPhone and its recently launched
iPad. Steve Jobs, Apple‘s iconic CEO, is advocating HTML5 (Hyper Text Markup
Language Version 5) as the new standard over Flash, Adobe‘s nearly ubiquitous
video player. When the Adobe-Apple relationship is going south, interestingly,
Google Inc. has quickly formed an alliance with Adobe right after Apple‘s
announcement. The example shows that the dynamics of cooperation-competition
97
tension can change rapidly in high-tech industries, and that alliance relationships are
prevalent among companies.
Interviews with industrial experts suggest that alliances are considered highly
important in creating and maintaining a competitive edge, in terms of technological
visions and network reputations. An anonymous senior manager in a large software
company mentioned during the interview that almost all software companies have to
have some alliance relationships to reach diverse resources, for example, to reach
resources up or down in a value chain. Because alliance announcements that going
into news press are often seen as strong signals to the market , however, many
companies are careful to use the terms of ―strategic alliance‖ or ―partnership‖ due to
legal concerns. Meanwhile, the interviewees also point out that, particularly during
the 1990s, there were generally positive market reactions to alliance announcements.
Thus many alliances were formed in a rush with weak strategic scrutiny, and
performed poorly as the relationships unraveled.
An interviewee specialized in software engineering estimated that less than
40% of alliances ever worked out. From his technical background, he attributed the
reasons to changes in various elements. Market players may change in terms of
consumer characteristics and competitor identities, such as shifting from established
large companies to startups. Another reason is the competitive dynamics, such as
how major players react to the introduction of new technologies and products. The
replacement of management teams at the senior executive level may bring changes to
98
a firm‘s alliance strategy, for example, strengthening relationships with independent
service providers (ISPs), which might have been ignored by previous executives. In
addition, a focal firm‘s strategic visions affect what type of alliance it will engage in,
such as that internet services and open source software have become commonly
collaborative focus among software partners.
Despite frequent turnovers of alliance relationships, the interviewees
generally agree that firms derive useful knowledge from one alliance to another.
There are several aspects companies look for when they seek a new partnered
solution. First, the combined relationship should provide broad technical and
industrial expertise as well as a long list of customers that highlight the depth of
experience. Second, access to capital is important to support financing and
purchasing options. In addition, in-depth joint technology integration offers inter-
operability and more potential for future technological development. Integration is
considered to provide customers with material benefits. Experience in prior alliances
can help in identifying and evaluating these factors. However, as pointed out by a
director-level interviewee, ―because each alliance or partner can be very different
from another, it takes time to understand the partner's organization and how best to
effectively work and communicate with them.‖ Alliance experience thus helps a
company to ―read‖ partners.
99
3.2.2 Alliance Types
There has yet to be a common terminology of types of alliances. Academic
research has classified alliances in different ways (Oxley & Silverman, 2008). With
respect to formation motives, alliances may be formed for the purposes of combining
partner outputs (i.e. co-specialization alliance), acquiring partner knowledge and
capabilities (i.e. learning alliance), and increasing scale or scope in a product line
(co-option alliance). With respect to alliance forms, the hierarchical control differs
between joint ventures and contractual agreements (Gulati & Singh, 1998). And
contractual agreements can be further categorized by activity types, such as licensing,
manufacturing, marketing, R&D, and esoteric relationships (e.g. buy-back options).
Further, an alliance may bear purposes of creating value (i.e. collaboration alliance),
appropriating values among partners (competition alliance), or both (co-opetition)
(Khanna et al., 1998).
In the software industry, companies often define alliance types by their
functionality, or the areas in which partners provide complementary resources. These
areas often include technology, marketing, system integration, and consulting
services, and so forth. Technology services provide technical solutions and product
development; marketing services help partners to reach a broader customer base,
conducting activities such as joint marketing, promotion campaigns, and
wholesale/retail services. Integration services combine and smooth operations
100
between different systems, platforms, and interfaces. Consulting services offer
solutions and trainings to specific industries, geographical regions, and business lines.
Another type specific to software vendors is distribution alliance. Software
vendors enter distribution relationships with major PC manufacturers to bundle their
software with the original equipment. These relationships, while less profitable than
selling through retail channels, help vendors gain market share and ―mind share‖ (i.e.
consumer awareness). In addition, they often result in the sale of future product
upgrades and help software vendors sell new and different products. While large
companies may have their own distribution teams, other vendors would prefer to
team up with partners who have expertise in distribution channels.
It is challenging to clearly define alliance types given the various dimensions
summarized above. Alliances may offer specialized services in any specific type or a
mix of activities. For example, the alliance between Cisco and Accenture, called the
Accenture Cisco Business Group (ACBG), combines their expertise in technological
and consulting services. Managed by a joint team from both partners, ACBG
collaboratively designs, promotes, and delivers solutions to enable communication-
enabled business processes (CEBP). An alliance between Cisco and Microsoft, on
the other hand, focuses exclusively on providing interoperability to customers who
use both of their applications
4
.
4
Source: Cisco Solution Overview, 2009 (www.cisco.com)
101
Another challenge on defining alliance types is that some activities are
difficult to categorize. For example, some software vendors, especially those with
products that work on midrange computers, act like computer resellers. In this regard,
they combine their software with another vendor‘s hardware and resell the whole
package as an integrated system. Because alliance activities are often self-reported,
this type of relationships can be classified as marketing, distribution, system
integration, or all of the above. Collectively, these challenges suggest that alliance
partners collaborate on diverse areas of activities. A company can be highly involved
in alliance relationships that cover a wide range of activities.
3.2.3 Trends of Alliances
During the period of 2001-2005, software companies were recovering from
the technology market crash in 2000, and had become more cautious in forming (or
announcing) new alliance relationships than they were pre-2000. On average,
software companies in the current sample announced about 3.2 new alliances per
year. Their alliance portfolios, including discounted historic relationships, are
averaged to about 12.4 alliances per firm. Figure 3.3 shows that there is a general
declining trend during this period: software companies have been reducing alliance
activities in the post-bubble era, possibly to cope with the aftermath in the financial
and technology markets.
Lavie (2007) presents data of US software companies and shows that the
average size of alliance portfolios has reached more than 30 per firm in 2001. Taking
102
into account of different discount factors and different data samples, the current data
still suggests a significant shift from the increasing trend of alliance formation from
the late 1990s to the decreasing trend in the early 2000s. Lavie‘s study illustrates that
the average portfolio size per firm had increased from less than 5 in 1990 to above
30 in 2001, a trend opposite to the one shown in Figure 3.3. Compared to Lavie‘s
data, the average portfolio size during the early 2000s (12.4 per firm) is comparable
to the level of 1996-1997, when fervent investment and collaboration activities about
IT companies had only begun to prevail the software industry.
Standard & Poor published an industry survey in 2006, suggesting that firms
have become more wary in forming new alliances, because ―…publicly traded
software companies are cognizant of the wariness their investors and customers have
towards large software deals following disappointing experiences in the late 1990s.
In those cases, cost synergies were achieved; however, significant headcount
reductions, customer attrition, and confusing product roadmaps were frequent side
effects‖ (p.2, April 27, 2006). Therefore, software companies are more cautious and
selective when considering new alliances. On the other hand, such wariness suggests
that alliances formed during this period (2001-2005) were more likely to be built
upon strong strategic rationality than those that were formed in the pre-2001 era.
103
Figure 3.3: Average New Alliances and Average Portfolio Size per Firm
The newly formed alliances can be organized by different governance
structures. Oxley (1997) analyzes appropriability risks in alliances and identifies a
three-category continuum of alliance governance from least to most hierarchical
modes: unilateral contract, bilateral contract, and equity jont venture. However,
further reading into alliance announcement reports reveals that within the second
class there are bilateral relationships with and without collaborative work. Absent of
collaboration, partner resources are maintained within firm boundaries and are
subject to relativley strong property regime. But when partners collaborate by
pooling resources together, they face a higher level of appropriabilty risk because the
pool resource generate values that are subject to weak property regime (Grant, 1996).
Firms in a collaborative relationship are able to exert more control over their partners
than they are in a non-collaborative relationship. Therfore, in this dissertation the
bilateral contract class is further divided into to subgroups: bilateral relationship with
versus without collaborative work.
104
The diversity of governance experience is calculated as a Herfindal index
across the four classes of governance forms. The maximum possible value for this
measure is 0.75 by definition. Figure 3.4 presents the annualized average diversity.
The full sample average diversity slightly increases from 0.35 in 2001 to 0.41 in
2005. Meanwhile, firms with a high level of alliance experience may make different
governance decisions than those with little expeirence in alliance activities.
Therefore, the full sample is divided into five quantiles according to a firm‘s
experience level, that is, by firms‘ total numbers of alliances in 2001.
There are some interesting observations from Figure 3.4. First, more
experienced firms maintained a highly stable pattern of governance diversity over
time (around 0.45). Second, less experienced firms, in particular those below the
average experience level, often start with focused governance forms (i.e. low
diversity). They subsequently extend to different modes and raise govenrance
diveresity toward the level of more experienced rivals. The bottom 10% firms
increased governance diversity by 74%, from 0.23 to 0.4, during the five-year period.
Additionally, the rate of diversity increase diminishes among firms in higher quantile
groups. These observations suggest that, firms tend to adopt more diverse
organzational structures as they accumulate alliance relationships. However,
experienced firms do not necessarily keep diversifying governance forms. At least in
this sample, software companies with a high level of alliance experience have
maintained an almost unchanged diversity level.
105
Figure 3.4: Alliance Governance Diveristy by Experience Quantiles
The trend of governance diversity may be driven by several factors.
Technologically, the popularity of internet has led to a movement away from
proprietary software to more open-source software. Meanwhile, the internet enables
a new delivery mechanism, software-as-a-service (or ―SaaS‖), to deliver application
services over internet as needed. These changes have affected the old business model,
when customers pay significant amount of up-front fees to a software vendor for
installation, maintainance, and technical supports. Under the SaaS model, customers
only need to pay a monthly subscription fee per user to access application services,
with limited implementation and without the need to install and maintain software
in-house. While experienced firms can manage such changes by exploiting prior
(relatively) diverse experience, inexperienced firms will resort to new and different
modes of collaboration that they may not have done before.
S&P‘s 2006 survey of the software industry estimates that worldwide internet
users exploded since the early 2000 and exceeded one billion in 2005 (p.11).
106
Internet applications, particularly online banking, financial services, and travel
services were the key drivers for this growth. Strategic collaborations in these types
of services thus may drive to different governance forms than previous ones. This
change may be particularly strong for less experienced companies, as they seek new
opportunities to grow.
In addition, software companies become more wary in making commitments
to new relationships during this period. Returns on investments are more stringently
evaluated before strategic decisions are made. Companies with a lot of prior alliances
are more likely to stick to previously used governance modes, for the purposes to
keep costs down and to avoid risks associated with new governance forms. Prior
experience provides a cushion to changes brought about by internet evolution.
3.2.4 Examples of Alliance Experience
To illustrate the different paths of alliance experience, this section contrasts
two companies, one from the top 10% companies that have most alliance experience,
and the other from the bottom 10% of alliance experience. Their comparison is
presented in Table 3.2 below.
In the software industry, Microsoft Corp. (Nasdaq: MSFT) has been the
largest player from the sample period. Its total revenue has increased from $25.3
billion in 2001 to $41.1 billion for the 12 months ended December 31, 2005
5
. With
its size and resources, Microsoft has long been active in developing alliance
5
Source: Microsoft Annual Report, 2005 (www.microsoft.com/msft/ar05)
107
relationships. It had engaged in approximately 60 publicly revealed alliances per
year throughout the 1990s. Even after the market shock in the late 1990s, Microsoft
still managed to form 157 alliances from 2001 to 2005, an average of about 30 deals
per year. Table 3.2 shows that Microsoft has reduced its new alliance formation,
along with a slight decline in its governance diversity. During this period, Microsoft
formed fewer alliances in the class of bilateral relationships (both collaboration and
non-collaboration forms), and its equity joint venture formation has dropped from 6
in 2001 to 2 in 2005.
Among the relatively less experienced companies, ANSYS Inc. (Nasdaq:
ANSS) is one that designs and markets application software for applications of
engineering simulation, technological design, and academic learning. Its total
revenue grows from $84.8 million in 2001 to $158 in 2005, and its return-on-assets
(ROA) grows from 13% to 16%. ANSYS had built strategic alliances gradually
during the five-year period at a constant pace, but it had engaged in alliances that are
organized under more diverse governance forms since 2003. Specifically, ANSYS
moved from bilateral collaborations to licensing agreements and bilateral non-
collaborative relationships, switched from US partners to foreign partners in India,
Germany, and UK, and broadened the scope of activities by adding system
integration and distribution services to their alliances. In its 2005 annual report,
ANSYS recognized that ―a portion of the growth came from strategic alliances and
108
acquisitions that have helped ANSYS to build its capabilities to meet customer
needs‖ (p.2)
6
.
Table 3.2: Examples of Software Companies: Microsoft Corp vs. ANSYS Inc
Microsoft Corp (top 10%) ANSYS Inc (bottom 10%)
Year Number of
new alliances
Governance
diversity
Number of
new alliances
Governance
diversity
2001 45 0.487 2 0
2002 31 0.477 3 0
2003 31 0.462 2 0.211
2004 21 0.443 2 0.233
2005 29 0.427 2 0.395
Total=157 Avg=0.463 Total=11 Avg=0.163
Table 3.2 presents a generally declining but still positive trend in new
alliance formation activities, as shown in Figures 3.3. Meanwhile, companies on
average organize alliances under governance forms that differ from previous ones,
thus witness a trend of growing governance diversity during this period. However,
the trend of diversity varies by the level of alliance experience. Firms with rich
experience have maintained a relatively stable portfolio of alliance by keeping their
governance diversity largely unchanged. On the other hand, firms with a low
experience level seek opportunities to expand their businesses, and have increased
the diversity of their alliance governance forms. One caveat here is that alliance
experience has a notable correlation with firm size (0.66 in the sample). This is
reasonable as large firms often possess abundant resources and thus are more likely
6
Source:
http://www.annualreports.com/HostedData/AnnualReports/PDFArchive/anss2005.pdf
109
to engage in alliance relationships. Therefore, large companies often have a high
level of alliance experience, and tend to maintain a stable pattern of alliance
characteristics.
3.3 Conclusions
This chapter summarizes the software industry and the alliance activities
within this industry. The software industry has experienced market crash in the late
1990s, and has revitalized in growth rates during the early 2000s. The proliferation
of internet usage and information technological developments have again intensified
competition and created new business opportunities. After the crash, companies have
become more wary in investment decisions and in strategic relationships. Alliances
were formed less frequently, probably by going through more stringent scrutiny and
investment criteria. However, an average firm still manages to forge more than 10
alliances per year (see Figure 3.3). Active alliance formation makes the software
industry an ideal setting to examine consequences of alliance experience.
Experience is generally recognized as instrumental for business development
in the software industry. Interviews with industrial experts reveal that prior
experience can generate useful knowledge to write more complete contracts and to
facilitate new cooperation. Prior alliances may result in positive or negative
outcomes that are unobservable to outsiders. But experience overall helps learning,
because experienced companies can be more prepared for future relationships. In fact,
110
negative performance may be a better lesson. An interviewee with over 15 years of
experience in both engineering and marketing pointed out that learning from
experience can be more salient when a relationship is going south: ―If you are burned,
you will learn to change internal processes and rules about how you are going to
manage (future) partnerships. People tend to learn more when something hurts than
doesn‘t.‖
Focusing on the software industry, the sample contains companies that share
similar industrial backgrounds. These companies are likely to face similar
environmental factors (e.g. regulation policies, technological developments, etc.)
when forming new alliances. Resources and capabilities that are deemed valuable
tend to be consistent across firms in the same industry. Multi-industry studies may
suffer from the problem that inputs or variables are labeled the same thing but are
really different, or vice versa (Shelanski & Klein, 1995).
Whereas the single-industry sample may not yield implications for different
industries, the theories developed in this dissertation can be generalized to industrial
settings where there are complex products, rapid technological changes, frequent
entries, and relatively low customer switching costs. In addition, software companies
are representative of technological industries in general. The findings can be
applicable to industries of telecommunications (e.g. Sampson, 2005), semi-
conductors (e.g. Stuart, 2000), biotechnologies (e.g. Baum et al., 2000; Rothaermel
& Deeds, 2004), chemicals (e.g. Ahuja, 2000), and the like. Thus, the results and
111
implications to be presented in the next two essays are at least applicable to these
sectors.
112
CHAPTER 4: ESSAY 1 – EFFECTS OF GOVERNANCE EXPERIENCE ON
ALLIANCE ORGANIZATIONAL STRUCTURE DESIGN
4.1 Introduction
How to govern interfirm relationships is a key strategic decision, and has
received extensive academic interests (e.g. Dyer & Singh, 1998; Gulati, 1995;
Leiblein & Miller, 2003; Mayer & Salomon, 2006; Oxley, 1997; Oxley & Sampson,
2004). The dominant theory addressing governance decisions has been the
transaction cost economics (TCE). TCE proposes a discriminating alignment
between transaction costs and governance structures. Specifically, interfirm
exchanges vary in their degrees of uncertainty, frequency, and asset specificity, thus
they also vary in ex post contractual hazards. For the purpose of economizing
transaction costs, exchanges should be governed by organizational structures that
differ in attributes of incentive intensity, administrative controls, and applicable
contract laws (Williamson, 1991). Therefore, transactions that involve greater
contractual hazards thus should be governed by more hierarchical structures. In
addition, governance structures that are improperly aligned with contractual hazards
will result in inferior performance of the focal transaction and of the exchange
partners (Mayer & Nickerson, 2005; Sampson, 2005; Silverman & Nickerson, 2003).
Whereas contractual hazards are shown to determine governance decisions,
they are transaction-level factors that are assumed to be independent of exchange
113
partners‘ capabilities. The capability view represents another main perspective
theorizing governance decisions. Originated from the resource-based view (RBV),
the capability perspective suggests that assets, resources, and knowledge of
transacting parties determine how capable a firm is in managing ex post uncertainty.
Specifically, firms with stronger technological capabilities are more likely to
internalize a production that draws on related technologies, for purposes of
protecting core competencies and exploiting values of unique capabilities (Argyres,
1996; Madhok, 1996). This view highlights the importance of heterogeneous firm
resources and capabilities that influence the considerations of contractual hazards.
Empirical research has provided strong and consistent support for the theorized
relationship between firm-specific technological capabilities and the governance
forms for exchange relationships (Leiblein & Miller, 2003; Mayer & Salomon, 2006).
Indeed, Williamson (1999) has called for more research in analyzing how existing
firm capabilities affect governance forms.
This essay helps advance the integration between TCE and the capability
perspective by addressing capabilities that are derived from governance-related
experience. In the context of strategic alliances, I draw on the organizational learning
perspective that suggests that prior alliance experience represents a source of
capability development (e.g. Rothaermel & Deeds, 2004; Sampson, 2005; Zollo,
Reuer, & Singh, 2002). While the learning theories do not directly address
governance decisions, I propose that by focusing on a firm‘s experience in
114
governance structures we can develop a more integrative picture on the relationship
between contractual hazards and experience-based firm capabilities. Specifically, the
governance experience incorporates two orthogonal dimensions that importantly
affect firm-specific capabilities: experience in specific governance forms and
experience across diverse structures.
Drawing from the capability view, I argue that in-depth experience in any
governance form generates exploitable knowledge and capabilities that are specific
to this structure and are applicable in future alliances. Firms thus have a tendency to
exploit experience-based, governance-specific capabilities and demonstrate a path-
dependent pattern in their governance decisions. Experience in diverse governance
forms, on the other hand, broadens the range of a firm‘s knowledge with respect to
different organizational structures. Diverse governance experience thus suggests that
a firm has experience in adapting to various degrees and types of ex post hazards.
Therefore, experience in diverse governance forms enables the firm to identify and
assess contractual hazards with greater insights, and makes contractual hazards more
salient in governance decisions. In addition, the diversity of governance experience
implies explorative learning across various structures. Diverse knowledge provides a
broad reference base when making governance decisions, thus reduces the reliance
on past governance choices. In sum, this essay shows that alliance governance
structures are importantly determined by factors of governance experience beyond
those of contractual hazards.
115
Empirically, I test the hypotheses in a sample of 1700 alliances formed by
154 focal companies in the US software industry between 2001 and 2005. Software
companies represent an ideal context to study the experience effects, given their
active and frequent formations of interfirm relationships. Based on archival data of
alliance announcements, governance forms are classified into four classes in a way
that is consistent with prior alliance governance studies (e.g. Oxley, 1997).
Experience variables are coded from each firm‘s historic alliances traced back to
1990. Overall, the data provides strong support for the predicted effects of alliance
experience. In addition to supporting TCE predictions of contractual hazards, the
empirical tests also support the effect that the diversity of governance experience
expands firm knowledge in coordinating variously structured relationships, and
makes firms more capable in foreseeing contractual hazards. Experience in specific
governance modes creates path-dependency in subsequent governance decisions, and
the path-dependent decisions are weakened when the firm has broad experience
across governance forms.
While the idea of capabilities complementing TCE logic is not new (e.g.
Leiblein & Miller, 2003; Mayer & Salomon, 2006), this essay makes several
contributions to the integration between contractual hazards and firm capabilities.
First, this essay suggests that firms can develop capabilities that are pertinent to
alliance governance forms by managing differently structured relationships. Such
experience-based capabilities, with multiple dimensions, influence firms‘ willingness
116
and ability to choose governance forms. These capabilities complement the focus of
current strategy literature on technical capabilities. Second, this essay contributes to
the transaction cost theory by illustrating possible cognitive limitations of economic
actors. Firms may not naturally possess the ability to foresee ex post contractual
hazards. In fact, such foreseeing capabilities are often the learning result from prior
experience in various governance forms. This essay also bridge theories in the
strategy literature regarding the links between organizational learning and
governance. While empirical studies have shown that learning enhances firm
performance (e.g. Argote et al., 1990; Sampson, 2005), there has been little effort
examining whether and how learning may affect governance decisions. Focusing on
learning about governance structures, this essay suggests that both exploitation and
exploration exist in governance experience: the former generates governance-
specific capabilities to be exploited in new alliances, and the latter strengthens
learning in diverse structures and leads to better informed governance decisions.
The rest of this chapter is organized as follows. First, I review relevant
theoretical perspectives: TCE, RBV, and organizational learning. Drawing from
these theories, hypotheses are developed for empirical tests. Then I describe the data,
methods, and variables used to measure key constructs. Following the presentation of
empirical results, I discuss the conclusions, contributions, and limitations of this
essay.
117
4.2 Literature Review
4.2.1 Transaction Cost Theory and Governance
The transaction cost theory is a fundamental perspective explaining the
decision of governance structures. TCE differs from the neoclassical economics
analysis on ex ante technological factors and incentive alignment, and asserts that
firm boundary decisions are also influenced by the efficiency characteristics of
different organizational structures. Efficiency is assumed to be inversely related to
transaction costs in the forms of searching for an exchange partner, negotiating the
contract, and monitoring the implementation of contract (North & Thomas, 1973).
Among the three types of transaction costs, TCE emphasizes the importance of costs
in monitoring or enforcing an exchange ex post. Market exchanges are generally
assumed to be more efficient than internal structures (or hierarchy) due to specialized
productions. However, the costs of market exchanges may exceed those of hierarchy
in certain situations. Thus TCE focuses on identifying exchange characteristics that
fit best to organizational structures of market and hierarchy.
The inefficiency of market is considered to arise from the small number
bargaining situation. While such situation may exist ex ante, TCE focuses on
exchange characteristics that may lead to ex post small-numbered bargaining.
Economic actors are assumed to be boundedly rational and at risk of behaving
opportunistically. Given asymmetric information between exchange partners and
uncertainties inherent in the exchange and environment, all contracts are inevitably
118
incomplete in specifying desired partner behaviors to cope with unforeseen and
unforeseeable contingencies (Williamson, 1991). Therefore, contracts are subject to
renegotiation and to opportunistic behaviors that may forfeit values invested in an
exchange. Further, TCE suggests that the risk of opportunism is greater when an
exchange involves investments of specific assets made by one or both parties.
Specific assets can create quasi-rents in an exchange, and are subject to partner hold-
ups (Klein, Crawford, & Alchian, 1978).
Williamson (1991) argues that different organizational structures provide
different solutions to contractual problems. Specifically, market, hybrid, and
hierarchy governance forms are arranged along a continuum and suggested to
provide different levels of incentive intensity and administrative controls. The
applicable contract laws also differ across these forms. When specific assets are
subject to severe contractual hazards, hierarchical governance is most beneficial
because it aligns partner incentives via administrative structures. Coordinated
adaptation within a hierarchy is more effective in such situations. On the other end of
the continuum, market exchanges are more effective when contractual hazards are
low and inconsequential. The price mechanism provides high-powered incentives for
exchange partners to react to contingencies, involves little bureaucratic costs in
administrative structures, and enables partners to adapt autonomously. Hybrid
governance contains some characteristics from both end forms, and provides
intermediate levels of incentives and controls between market and hierarchy.
119
TCE provides important insights regarding the most efficient means to
govern a transaction. However, the development of TCE has largely focused on
transaction characteristics as independent of firm-level characteristics, for instance,
the portfolio of a firm‘s transactions and other firm-specific assets and capabilities
that may affect the governance mode of a focal transaction (Leiblein & Miller, 2003).
Absent of firm characteristics, TCE would predict that, in equilibrium, all firms will
choose the same governance structure to manage transactions with similar attributes.
However, firms differ in their boundary decisions even when facing the same set of
contractual hazards. In fact, ―misaligned‖ governance, or governance forms that are
mismatched to the level of transactional hazards, remain a significant portion of
practical governance decisions (e.g. Sampson, 2004; Silverman & Nickerson, 2003).
Recent research has shown that firm experience and capabilities enter into the
decision of governance structures (e.g. Mayer & Salomon, 2006; Leiblein & Miller,
2003). Thus, the optimal governance decisions is likely to be contingent on both
attributes of contractual hazards at the transaction-level and strengths and
weaknesses to manage certain governance form at the firm-level.
4.2.2 Firm Capabilities and Governance
The resource-based view (RBV) provides an alternative theory to analyze the
effects of firm-level capabilities on transaction governance decisions. RBV is built
upon two basic conditions. First, firms possess heterogeneous resources and
capabilities (Wernerfelt, 1984). Second, these different resources and capabilities are
120
limited in supply and are costly to imitate (Dierickx & Cool, 1989; Barney, 1991).
As a result of these two conditions, firms‘ governance decisions may be driven by
their efforts to leverage and/or protect their unique capabilities (e.g. Madhok, 1996).
While TCE proposes a discriminating alignment between the degree of contractual
hazards and governance structures (Williamson, 1991), RBV emphasizes the
opportunity to create competitive advantages by exploiting unique firm-level
resources and capabilities.
A firm‘s governance choice may be influenced by its specific technical
capabilities. For example, Argyres (1996) describes in a case study that a
manufacturing firm chose to outsource a key production component because the firm
felt that it could not support related production activities. Internal governance can
organize production activities more efficiently, because a hierarchical structure
assimilates and deploys relevant technological knowledge with greater cost-
efficiency (Kogut & Zander, 1992). Recent empirical studies support the capability
effect on the probability that a firm internalizes a transaction when strengths in
related technologies are present. Leiblein and Miller (2003) show that technological
capabilities in related value chain activities increase the likelihood of vertical
integration when a firm has more experience using the relevant technology. Mayer
and Salomon (2006) find that internal governance forms are preferred when a
transaction draws upon technologies in which a focal firm has strong capabilities.
121
The common argument is that firms with strong technical capabilities are more
efficient to organize a transaction than the market exchanges.
Beyond technologies, firm capabilities may also influence governance
decisions by prior committed relationships. TCE assumes that firms may renegotiate
contractual provisions effortlessly as they adapt to emerging contingencies. However,
prior governance commitments are likely to create switching costs among alternative
governance forms, and may lock a firm into a particular governance choice. For
example, Argyres and Liebeskind (1999) introduce the concept of governance
inseparability, suggesting that there are interdependencies between prior contractual
commitments and present governance decisions. The interdependencies are the
results of prior formal and informal commitments that are made by a firm. Such
commitments evolve into a portfolio of relationships that makes some governance
structures easier and others more difficult to implement.
TCE assumes that the specialized investments needed to support a transaction
have yet to be made before a transaction is implemented. However, a firm‘s historic
governance decisions suggest the opposite. Specifically, prior governance decisions
represent a form of governance-specific investments. Williamson (1999) recognizes
that pre-existing investments may tilt the alignment calculus ―in favor of the
[governance] form that possesses such specialized, under-utilized capacity‖ (p. 1103).
Historic decisions can create implementation obstacles that constrain the feasible
range and types of mechanisms. Therefore, a firm‘s portfolio of historic governance
122
decisions can create path-dependency, and influences the firm‘s incentives to engage
in subsequent governance structures.
The path-dependency in governance decisions is consistent with the technical
capability argument. A firm learns from prior governance experience to manage
governance modes, as learning takes place with respect to the nature of the
specialized investments in a particular governance form. Over time, a firm derives
capabilities to effectively support the focal type of transactions. Thus learning from
experience shapes a firm‘s capabilities that can be exploited in subsequent
governance decisions. In a study on path-dependent capabilities, Aggarwal and Hsu
(2009) use firm-level panel data to show that a firm on average adopts governance
forms in which they have greater experience than alternative forms.
To explore the firm-level antecedents of governance decisions, it is critical to
identify specific aspects of firm-level experiential capabilities. Experience varies
across firms, and generates firm-specific capabilities that lead to different choices of
governance forms under the same set of contractual hazards. Firms may exploit such
capabilities through subsequent governance decisions. The capability factors from
the RBV perspective relax the constraint of homogeneous production functions
across firms in the TCE framework, and contribute to traditional TCE-dominated
studies of firm boundaries.
123
4.2.3 Firm Learning and Governance
The learning perspective analyzes learning processes and outcomes, and
suggests that learning in general creates useful knowledge and routines, which can be
deployed to manage contingencies and to enhance organizational survival and
performance. Learning is considered to arise from organizational experience. Firm-
level knowledge and routines accrue through experiential learning and/or vicarious
learning (Huber, 1991). The range of learning can cover organizational activities that
are focused, local, and exploitative or those that are diverse, distant, and explorative
(March, 1991). As an inter-temporal process, learning is by definition path-
dependent and firm-specific. Over time, learning in focused and repetitive activities
can create operational efficiency as well as rigidity in firm knowledge and in
adaptive behaviors (Levinthal & March, 1993). Thus, distant or diverse learning is
important in expanding a firm‘s capacity to continuously absorb new information
and knowledge to maintain its competence (Cohen & Levinthal, 1990).
Learning research on firm-level experience has mostly focused on
performance outcomes instead of structural decisions. The direct link between
learning and performance assumes that capabilities generated from experience
directly improve organizational performance. However, capabilities are beneficial
only when they are exploited in subsequent contractual relationships, which are
organized under governance structures that are more informed due to experience-
based knowledge. Scholars have recently begun to address the relationship between
124
learning and contractual agreements. For example, Arino and de la Torre (1998) find
that alliance partners mutually adapt to ex post disturbances by renegotiating
contractual clauses. Mayer and Argyres (2004) use a longitudinal case study to show
that, as an interfirm relationship evolves, the partners change contractual provisions
over time. In a large sample study on alliances, Anand and Khanna (2000) report that
there are path-dependent governance decisions, particularly the governance form of
equity joint venture. While supporting the learning effects on governance design,
these studies remain as the few exceptions that analyze this relationship.
It is important to unpack the dimensions of experience in order to further
explore the effects of learning on governance structures. Path-dependent governance
decisions represent one form of learning effect, which suggests that a firm becomes
more capable and efficient in managing a particular governance form over time, or
learning-by-doing (e.g. Argote et al, 1990). While path-dependency implies a firm‘s
in-depth experience in a single governance form, a contemporaneous dimension
central to this dissertation is the breadth of experience across diverse governance
forms. The learning perspective suggests that explorative learning expands a firm‘s
range of knowledge and routines to cope with contingencies and to absorb new
information. Therefore, diverse experience in various governance structures
represents a form of explorative learning from variously structured interfirm
relationships, and may affect subsequent choices of governance modes. In sum, a
firm‘s experience should be examined beyond the repetitive measure to advance the
125
theoretical connection between firm-level learning and interfirm-level governance
decisions.
4.3 Development of Hypotheses
In this section, an empirically testable model is developed on the ground of
transaction cost, RBV (in particular, capabilities), and the learning perspectives. First,
alternatives of alliance governance forms are defined as an extension of current
alliance research. Next, Hypotheses 1 and 2 briefly describe the base model of the
impacts from transaction costs and path-dependent capabilities. The model is then
extended to take into account the diversity of governance experience. Specifically,
Hypotheses 3 and 4 suggest that firm-level experience in diverse governance
alternatives reduces a firm‘s dependency on its historical decisions, and strengthens
its capabilities to foresee contractual hazards.
4.3.1 Alliance Governance Forms
It would be ideal to rank alliance governance forms along a continuum that is
similar to the market-hybrid-hierarchy continuum by Williamson (1991). However,
this rank is much more challenging to design than that of market-hierarchy. Alliances
incorporate different collaborative activities, for example, a marketing service
agreement differs from a software development agreement in idiosyncratic
governance features. Many alliances cover broad scopes, and involve multiple
activities simultaneously (Khanna, 1998; Oxley & Sampson, 2004). Alliance
126
activities can be highly diverse. Diverse activities are demanding for micro-analytic
data to compare governance instruments of incentive intensity, administrative control,
adaptability, and contract laws. This fact adds complexity because it is not apparent
how to rank the combinations of governance instruments in various activities (Oxley
& Silverman, 2008).
Given difficulties in ranking alliance governance forms, early studies often
measure the dependent variable as a dichotomy between equity and non-equity forms.
Pisano (1989) shows that equity relationship is preferred when contractual hazards
are high, for instance, when an alliance involves R&D components, carries on
multiple simultaneous projects, or has few potential candidates. In a related study,
Pisano (1990) shows that the small-number bargaining situation motivates firms to
internalize R&D projects instead of choosing external contracts. Gulati (1995)
contrasts choices of equity versus non-equity governance forms. Alliances are shown
to be managed more likely by equity-based governance if they involve collaborative
R&D components, cross-border participants, and complexity of coordinating three or
more partners.
The dichotomy between equity and non-equity alliances only roughly
resembles the make-or-buy decision with respect to financial commitments. Gulati
and Singh (1998) recognize this problem and classify governance forms into three
types of joint venture, minority equity, and contractual alliances. However, alliances
often incorporate collaborative efforts that are captured beyond financial investments.
127
For example, an equity relationship may involve limited partner interactions beyond
signing the contract (e.g. minority equity in OEM partners), and a non-equity
relationship may require different degrees of interactions.
Oxley (1997) draws from Williamson‘s (1991) comparative governance
forms and examines the level of appropriability risks in technological alliances. She
proposes that alliance agreements can be clustered in discrete forms. Three forms are
defined: unilateral contractual agreements (e.g. unilateral licensing, long-term supply
contracts, outsourced R&D contracts), bilateral contractual agreements (e.g. bilateral
licensing, co-marketing, co-development agreements), and equity alliances (e.g. joint
ventures). The three-mode continuum has been adopted in later alliance research (e.g.
Colombo, 2003; Santoro & McGill, 2005) in addition to the equity/non-equity
distinction.
This essay builds upon Oxley‘s classification. Whereas unilateral agreement
and joint venture are clearly defined, the bilateral category can cover a variety of
collaborative relationships. The key determinant for the degree of appropriation is
whether there are values created by collaborative efforts. Values created in a bilateral
relationship can be delineated by firm boundaries in alliances with collaborative
effort than those that are without. Therefore, the bilateral category is divided into two
classes. Following Williamson (1991), alliance governance forms differ in the
following key attributes in Table 4.1. These governance forms are ordered from least
to most hierarchical from left to right.
128
Table 4.1 Comparison of Alliance Governance Structures, adapted from Williamson
(1991) and Oxley (1997)
Unilateral
Contract
Bilateral
Contract w/o
Collaboration
Bilateral
Contract w/
Collaboration
Equity Joint
Venture (JV)
Incentive Intensity High power
price
mechanism
High in
collaboration
Low in
collaboration
Low within
JV
Administrative
control
None Low Medium High
Contract law Litigation Arbitration and
litigation
Arbitration and
forbearance
Forbearance
within JV
Adaptation Autonomous Relatively
autonomous
Relatively
coordinated
Highly
coordinated
Diverse experience: Generates capabilities to adapt to contingencies in relationships that contain
various incentive intensity, administrative power, and dispute resolutions.
The following three sections examine the choice of governance structures
among the four types identified in Table 4.1. Unilateral agreements are seen as the
most market-type structure, and equity joint ventures are considered to resemble a
firm structure, or the most hierarchy-type mode. Between the two extreme forms,
bilateral contracts that involve collaborative work are considered to impose greater
administrative control and lower-powered incentives on alliance partners than
bilateral contracts without collaborative efforts.
4.3.2 Contractual Hazards
Contractual hazards are present in interfirm relationships when alliance
partners invest specific assets in a relationship. Investments can be specific in the
forms of physical, human, and relational assets (Dyer & Singh, 1998). Specific assets
are at risk of being lost because transactions are subject to uncertainty and partners
129
may behave opportunistically. TCE addresses the level of contractual hazards in
interfirm exchanges as the key driver of integration (Williamson, 1985). The extent
of such hazards is shaped by the degree of appropriability risk (Oxley, 1997),
measurement difficulty (Poppo & Zenger, 1998; Williamson, 1991), or task
complexity (Khanna, 1998).
Appropriability refers to the contractual risk that valuable intellectual
property may be appropriated by a partner (Gulati & Singh, 1998; Oxley, 1997).
When firms engage in interfirm relationships, they exchange proprietary information
and technology. The information-sharing leads to the possibility that one partner will
expropriate the valuable and proprietary knowledge of the other. Exchange contracts
may specify details to prevent the risk of appropriation, but all contracts are
inevitably complete (Williamson, 1985). Further, partner efforts in alliances generate
collaborative values that may be appropriated by a partner, because the created
values are subject to weak property right regime between the partners. Alliance
partners therefore have an incentive to use greater administrative control to prevent
the appropriability risk.
Measurement difficulty refers to the risk that the outcome of a transaction is
difficult to observe or measure. When the quality of output is difficult to gauge ex
post, it is problematic to reward or penalize a partner‘s accurate efforts. Thus
partners have the incentive to shirk in their investments and efforts. The problem of
observability and measurement can be more consequential in alliances, where the
130
outcomes are dependent on bilateral and collaborative efforts. Given that certain
outputs are difficult to specify, the hierarchical structure has an advantage over the
market by ensuring inputs rather than evaluating outputs. Therefore, when the
measurement difficulty is great, an interfirm relationship is managed more
effectively by hierarchical alliance governance than by market-type structures
(Mayer & Nickerson, 2005; Poppo & Zenger, 1998).
Task complexity refers to the degree that firms in an exchange needs to
coordinate in order to complete the exchange. When the task to be coordinated in an
interfirm relationship is broad or complicated, there is increased difficulty in
specifying contractual terms to guide the exchange. More incomplete contracts
impose greater risks on the values to be created in a relationship. Alliance partners
may cooperate in a relationship but compete in areas outside the relationship
(Khanna, 1998). In alliances covering complex activities, there are greater needs for
partners to coordinate efforts in managing and preventing conflicts between
cooperation and competition. When the complexity of a relationship increases, the
market structure becomes less effective in coordinating diverse activities via
autonomous adaptation. The coordinated adaption in hierarchical structures provides
greater control power over diverse activities. Further, when a governance form is
insufficient to manage complex alliance activities, partners may choose to reduce the
scope of activities to accomplish the goal of an alliance (Oxley & Sampson, 2004).
131
The key implication of these contractual hazards is that firms should use
alliance governance forms with greater hierarchical power when the risks of
appropriability, measurement difficulty, and task complexity are high and
consequential to a relationship. When these risks are low and inconsequential,
market-type governance forms can coordinate alliances more efficiently.
Hierarchical structures are costly to set up, but the set-up costs are traded off by
effective safeguards and internal coordination. The potential of misappropriation
under these contractual hazards raises the cost of market-type alliance structures by
making contractual negotiations more contentious and cost-inefficient, because
partners have to redeploy resources from other efficiency-improving assets to
negotiating and designing more contractual safeguards. Overall, the costs to organize
alliances under hierarchical structures are lower than under market-type structures
when the above discussed contractual hazards are high.
Hypothesis 1a: Appropriability risk in an alliance increases the likelihood
that firms choosing more hierarchical governance forms.
Hypothesis 1b: Measurement difficulty in an alliance increases the likelihood
that firms choose more hierarchical governance forms.
Hypothesis 1c: Complexity of alliance task increases the likelihood that firms
choose more hierarchical governance forms.
132
4.3.3 Experience-Based Capabilities
While TCE describes market and hierarchy as differing in their efficiencies to
respond to transaction costs, it is relatively silent on the influences of firm-specific
capabilities. Indeed, Williamson (1999: 1104) suggests that a key question to
advance TCE is to move beyond the comparative advantages of the best generic
governance structure available to organize transaction X, and to understand how
firms with pre-existing strengths and weaknesses should organize transaction X.
Examining firm capabilities provides insights for the more particular question of
firm-level characteristics on governance decisions.
According to the resource-based view, unique and valuable firm capabilities
are critical to a firm‘s performance (Wernerfelt, 1984). Empirical research has
supported this view by showing that technological, managerial, and marketing
capabilities enhance alliance partners‘ performance (e.g. Baum et al., 2000; Kale et
al., 2002; Kale & Singh, 2007; Sarkar et al., 2009). Capabilities can be derived from
prior relevant experience. For example, past alliances foster capabilities to manage
situations of ambiguity more readily in subsequent collaborations (Kale & Singh,
2007; Sampson, 2005), and inter-temporal cooperation generates capabilities to
design more effective and complete contracts (Argyres & Mayer, 2007; Reuer et al.,
2002).
The effects of firm capabilities on the governance of interfirm relationships
are examined in two primary streams. One draws directly from RBV, arguing that
133
capabilities from prior collaborative relationships enables a firm to achieve better
performance due to enhanced managerial capabilities and mutual learning (e.g. Kale
et al., 2002; Sampson, 2005). The other combines TCE with technological
capabilities, suggesting that governance decisions are influenced by firm-level
technical strengths (e.g. Hoetker, 2005; Leiblein & Miller, 2003; Mayer & Salomon,
2006). Between these two streams of research, it remains relatively under-studied
that how managerial capabilities from prior relationships influence the choices of
alliance governance forms.
Regarding the governance decisions, the RBV suggests that valuable and
difficult-to-imitate capabilities can determine how a firm prefers to organize an
interfirm relationship. A key source of these capabilities is the firm‘s prior
experience in related organizational structures. Whereas some capabilities may be
acquired from the strategic factor market (Barney, 1986), experience-based
capabilities are path-dependent and causally ambiguous to outsiders. Over time,
firms develop a set of stable, detailed, and predictable patterns of behaviors in
production activities (Nelson & Winter, 1982). The values of these routines are
exploited in subsequent activities. Firm capabilities are embedded in the routines
derived from past experience, and represent resources that are firm-specific, history-
dependent, and can never be perfectly replicated by a different firm. Similarly, a firm
accumulates alliance experience, from which it creates a set of routines to manage
future interfirm relationships (Kale et al, 2002). Firms with such capabilities will be
134
more likely to exploit in successive activities. Therefore, experientially derived
capabilities create path-dependent decisions.
A particular form of path-dependent capabilities is those that arise from a
firm‘s previously adopted governance structures. From the capability view, prior
experience with a specific governance form creates routines and capabilities that are
pertinent to this particular form. That is, capabilities to coordinate interfirm
relationships are developed experientially at the firm level. Prior alliances provide
learning opportunities that enhance a firm‘s capabilities to collaborate and adapt to
contingencies. Meanwhile, because the learning process occurs under a chosen
governance structure, capabilities that are developed are likely to be structure-
specific. For example, capabilities to manage different licensees are different than
those to manage daily operations in joint ventures. When a firm has efficiency in
managing one organizational form but not in others, it may choose to use the most
familiar structure in other relationships for the benefits of exploiting relevant
capabilities. Governance structure-specific capabilities also create organizational
rigidities (March, 1991), and limit a firm‘s willingness and capabilities to choose an
alternative structure. Therefore, even though firms strive to choose the most cost-
efficient governance form matching contractual hazards in an alliance, they are
simultaneously confined by experience-based strengths and weaknesses in managing
a TCE-predicted form. Path-dependent capabilities thus explain the observation that
135
some firms choose to govern alliances with structures that are misaligned with
transactional characteristics.
Activities that are carried out under a certain structure evolve into routines,
which determine how a firm will adapt to contingencies in the future. Experience in
non-equity alliances may create a set of knowledge and routines that are specific to
non-equity relationships. As presented in Table 4.1, more hierarchical governance
structures involve more coordinated adaptation that manage partner incentives,
disputes, and cooperation. Firms with experience in hierarchical structures thus are
more likely to exploit such equity-relationship-based knowledge by choosing similar
structures again. On the other hand, experience in market-type, less hierarchical
structures suggests that a focal firm may have capabilities in adapting to
contingencies in a more autonomous way, and that the firm is likely to exploit
capabilities that are specific to these governance forms.
Note that experience in related governance structures may create capabilities
that are useful to each other. For instance, the same set of routines to coordinate
between partners can be applied to manage bilateral relationships with and without
collaboration. However, experience that is specific to one governance form may not
be equally applicable in alternative governance forms. First, experiential capabilities
can be highly specific in terms of contract design. For example, joint venture
contracts contain much more detailed, partner-specific clauses than licensing
agreements (Mayer & Teece, 2008). While licensing relationships often rely on
136
boilerplate contracts (Anand & Khanna, 2000b), joint venture contracts are
idiosyncratic to each partner, and are complex in terms of operations, liabilities, and
financial terms. In addition, experience in a governance mode affects a firm‘s
specific capabilities, including the abilities to design systems to manage relationships
outside of the hierarchy and to monitor and enforce partners‘ compliance to the
contract (Doz & Hamel, 1998). Firms will resort to routines that they are most
familiar with, and these routines are based on particular governance modes in prior
experience. Therefore, experience in specific governance forms is likely to lead a
firm to make path-dependent governance decisions, even when externalities among
alternative governance modes are present.
Supporting the argument of path-dependent capabilities, Leblein and Miller
(2003) find that prior outsourcing relationships reduce the likelihood of vertical
integration. Anand and Khanna (2000a) find a momentum that firms will follow
prior joint venture experience in subsequent alliances. More recently, Aggarwal and
Hsu (2009) study firm-level alliance portfolios, and show a positive correlation
between a firm‘s current alliance governance form and the forms in which the firm
has greater experience. Further, interviews with practitioners validate the path-
dependency of capabilities from experience. According to the interviewed managers,
prior experience creates useful knowledge for future relationships. One particular
form of useful knowledge is that partners can design more polished and effective
contracts. A serial entrepreneur of software startups believed that his companies
137
gradually learn from writing alliance deals to create boilerplate contracts.
Interviewees in large companies mentioned that alliance agreements become more
thorough and careful over time, for instances, in financial clauses, legal
contingencies, and end-user supports. The standardizing process of alliances evolves
into a template of agreements, which can be fit onto new relationships. In sum, firms
are likely to resort to contract design knowledge from previous alliances, and design
similar governance structures in a new alliance. The main reason being that
exploiting knowledge created in previous relationships and is embedded in
contractual agreements.
Hypothesis 2: Experience in a specific governance mode increases the
likelihood that firms choose the same governance modes in subsequent
alliances.
4.3.4 Governance Diversity of Alliance Experience
Recent literature has begun to theorize about the complementarity between
firm-specific technological capabilities and the TCE predictions on make-or-buy
decisions (e.g. Argyres, 1996; Hoetker, 2005; Leiblein & Miller, 2003; Mayer &
Salomon, 2006). The main effect from the capability view suggests that firms prefer
to exploit the values of its superior technological capabilities by internalizing
productions that employ relevant technologies (Argyres, 1996; Leiblein & Miller,
2003). Combining capabilities with contractual hazards, Mayer and Salomon (2006)
suggest that technological capabilities enable a firm to safeguard interfirm exchanges
138
with more complete contracts and more ready adaptations, and thus mitigate the
salience of contractual hazards on governance decisions.
In addition to technological capabilities, another important type is the ability
to manage a relationship. The alliance literature has viewed a firm‘s prior experience
as a key source to build managerial capabilities and enhance a focal firm‘s
performance. By participating in new alliance relationships, firms learn about
valuable know-how from partners and about means to protect their core
competencies from partner misappropriation (Kale et al, 2000). Firms with
collaborative experience are more capable to coordinate partner efforts and adapt to
contingencies that may arise during the implementation of a relationship (Sampson,
2005). As a result, alliance experience creates knowledge about collaborations, and
is beneficial to a firm‘s performance in terms of post-formation adjustments (Reuer
et al., 2002), firm survival (Baum et al., 2000), financial returns (Anand & Khanna,
2000a), innovation outcomes (Sampson, 2005), and managerial satisfactions (Kale et
al., 2002).
In the current alliance research, a firm‘s alliance experience is often captured
by a cumulative count of prior interfirm deals. The implicit assumption is that
experiential learning gradually generates in-depth know-how and capabilities to
manage interfirm relationships. However, such measurement of experience glosses
over specific dimensions of experience that may generate different impacts on firm
capabilities. Unpacking dimensions of experience thus may contribute to the research
139
on experience-based capabilities. Specifically, this essay draws from the learning
perspective and suggests that alliance experience can be defined in two major
dimensions: the degree of experience in a specific governance form and the diversity
of experience across different governance forms. That is, the depth and breadth of
alliance experience.
While the RBV recognizes values in firm-specific resources and capabilities,
this perspective rarely extends beyond the cumulative degree of a certain type of
resource or capability. The learning perspective addresses this gap by proposing two
types of learning processes: exploitation and exploration (March, 1991). Exploitative
learning benefits a firm by improving efficiency in repetitive activities (Argote et al.,
1990). Viewing alliance experience as a sum, prior alliance studies have commonly
agreed that alliance experience allows a firm to become better at managing interfirm
relationships (e.g. Anand & Khanna, 2000a; Sampson, 2005). While this research
examines an alliance as the focal activity that is being exploited, the governance
forms of alliances are different, and thus can be classified into exploitative and
explorative dimensions.
A firm‘s experience in a particular governance form represents a form of
exploitative learning, which is specific to the structures, routines, and interactive
patterns of the focal governance. The discussion of Hypothesis 2 suggests that
experience in any governance form can create exploitable knowledge, which is
specifically useful for future alliances under the same governance structure. The
140
path-dependency in governance decisions indicates a process of exploitative learning.
Thus, governance decisions are likely to be gravitated toward the form in which a
firm has greater experience. Firms benefit from governance choices that are similar
to their previous ones by gaining greater efficiency from learning-by-doing and from
exploiting governance structure-specific capabilities.
A more interesting dimension of alliance experience is the degree to which a
firm has participated in diverse governance structures, or the diversity of governance
experience. Diverse governance experience represents a form of explorative learning
across structures of unilateral agreements, bilateral agreements with or without
collaborative work, and equity joint ventures. Alliance governance forms differ in
the levels of incentive intensity, administrative control, and contract laws that are
applicable to disputes, as presented in Table 4.1. Firms with experience in all of
these forms are likely to have encountered various types and degrees of ex post
adaptation that require different coordination.
A firm invests increasing efforts to align partner incentives when the firm
moves from unilateral contracts to joint ventures. The market-type unilateral
agreements, including unilateral licensing, outsourcing, or long-term supply
contracts, resemble the ―buy‖ decision, in which partners react to the price
mechanism autonomously. Contractual hazards are relatively uniform, such as
violations in transferring or applying property right or copyright, and in license
payment issues. The contracts of unilateral agreements tend to be boilerplate. When
141
there are disputes, partners resort to litigation since the boundaries of property rights
are relatively confined within each firm.
The price-based incentive is weakened as the choice of governance moves
away from unilateral relationships toward joint ventures, where partners pool their
resources into a new business entity with its own operations. High-powered
incentives are then substituted by administrative structures. The most hierarchy-type
governance involves a joint board of directors, who exert administrative power to
mitigate contractual risks that may arise from insufficient partner efforts, inter-
partner conflicts, and value misappropriation. Joint venture contracts specify
financial, managerial, and technical details of collaborations that to be implemented
on a daily basis, thus are highly specific to each partner. Joint venture contracts also
contain details about dispute resolution, which often goes through internal
forbearance at the first place.
Between the two extreme forms, bilateral contracts have intermediate levels
of incentive intensity and administrative controls. Depending on the extent to which
knowledge and information are shared with a partner, firms in bilateral relationships
face different degrees of appropriation risk. While there is no formal administrative
structure, partners affect each other‘s behaviors, because the benefits from a focal
alliance depend on collaborative efforts to different extents. When contingencies
arise in a relationship, partners adapt in a semi-coordinated way to adjust toward
142
efficiency. The degree of coordination is contingent on the presence of collaborative
efforts.
With experience in diverse governance forms, a firm can derive collaborative
capabilities to align partner incentives, to coordinate adaptation, and to resolve
disputes. Experience in each governance form generates structure-specific
capabilities that may not be applicable to other structures, but this structural rigidity
is mitigated by the diversity of governance experience. High diversity creates the
benefit of explorative learning to absorb new information and knowledge (Cohen &
Levinthal, 1990). Diverse governance experience expands a firm‘s knowledge range
to learn about inter-partner dynamics and potential hazards, thus enhances the firm‘s
capabilities to assess potential benefits and hazards of a new relationship.
Appropriability risk stems from knowledge sharing and collective value
creating among partners. With experience in different alliance structures, a firm
derives more insights about to what extent knowledge leakage and value
appropriation can happen under different governance forms. However, the effect of
diversity on the appropriability risk is ambiguous. On one hand, diverse governance
experience allows a firm to better foresee the possibilities of appropriation inherent
in an alliance. Thus, the firm will consider appropriability as a more imminent risk
that should be governed by more hierarchical structures. On the other hand, alliances
that contain high appropriability risks are likely to bare the goal of innovation from
pooled resources. In such relationships, firms are able to foresee appropriability risks
143
based on diverse governance experience, but are still willing to commit proprietary
resources under the recognized risks, for the purpose of sharing knowledge to create
innovative values. In this case, the diversity of governance experience may not affect
the salience of appropriation hazards to firms with diverse governance experience.
Furthermore, the risk that values created in an alliance may be mis-appropriated by a
partner is highly partner-specific. Thus, diverse knowledge about different
governance forms may shed little light on a firm‘s perception of appropriation
hazards with a new partner or in a new alliance. Therefore, on average, governance
diversity is expected to influence the effect of appropriability risk to an insignificant
extent.
The diversity of governance experience facilitates a firm to read and interpret
partner behaviors under different structures. With focused experience in a particular
form, a firm may lack capabilities to foresee the likelihood of partner shirking, and to
identify the occurrence and consequences of inferior output quality. For example,
shirking is less of an issue in licensing agreements, but can destroy values to be
generated in a joint venture. Firms with focused experience in licensing contracts but
not in joint ventures are less likely to perceive shirking as a significant threat to
alliance outcomes. When choosing an alliance governance form, firms with focused
governance experience will choose to focus on other types of contractual hazards
instead of measurement difficulty due to their insufficient ability to assess hazards in
partner efforts and output quality. On the other hand, diversity of governance
144
experience enables the firms to anticipate risks that are related to measurement
problems, and to put more emphasis on the hazard of measurement difficulty.
Similarly, the complexity of alliance tasks is perceived as a greater concern
when a firm has experienced diverse governance forms. Complex tasks require a
firm to coordinate multiple types of productive activities and/or multiple partners.
There is a greater probability that contractual hazards can arise in either productions,
or partner efforts, or both. Focused experience in one governance form limits a
firm‘s knowledge range to recognize potential hazards embedded in complex
alliance tasks. Because the firm is incapable foreseeing contingencies and hazards
embedded in complex tasks, it is less likely to assign weights to complexity when
evaluating ex post hazards. However, differently structured relationships broaden the
firm‘s ability to identify possible contingencies in a relationship, and make task
complexity a more salient determinant in governance decisions. Thus, the cognitive
limit from focused governance experience is weakened when the firm has experience
in various governance forms.
In sum, the learning benefits associated with diverse experience across
governance forms enable firms to better foresee, identify, and adapt to the particular
hazards associated with a new cooperative relationship. Diverse experience helps a
firm to recognize the risk of appropriability, but the firm may choose to leave the
risk open for the tradeoff with innovative outcomes. When a firm is capable to
foresee contractual hazards that arise from measurement difficulty and task
145
complexity, it will make more informed decisions to align the hazards with
governance structures.
Hypothesis 3a: Diversity of governance experience has no impact on the
effect of appropriability risk on a firm ’s alliance governance decision.
Hypothesis 3b: Diversity of governance experience strengthens the effect of
measurement difficulty on a firm’s alliance governance decision.
Hypothesis 3c: Diversity of governance experience strengthens the effect of
task complexity on a firm ’s alliance governance decision.
When firms learn from experience, they have the tendency to simplify the
contents of learning and to specialize in most familiar activities (Levinthal & March,
1993). Given the limitations in learning process, learning outcomes are often local
and narrow, and leads to biased learning and myopia in subsequent activities. In fact,
Mayer and Argyres (2004) present in a longitudinal case study that exchange
partners learn to improve contract design over time, but the learning effects are
mostly incremental and local. Local learning creates the benefits of exploitative
learning, but also creates rigidity and limitations on the range of knowledge.
Similarly, focused experience in a particular governance form implies a firm‘s local
learning specific to this governance form. When the firm becomes more capable in
managing the focal form by repeatedly applying it in alliance relationships, its
knowledge is gravitated toward this focal form but ignorant of other forms. As a
146
result, focused governance experience makes a firm less competent in recognizing,
assessing, and managing alternative forms.
The limitations in focused governance experience will be reduced when the
firm adopts more diverse governance forms. Diverse governance experience, or
governance diversity, represents a type of explorative learning across governance
modes. Because governance structures differ in their attributes shown in Table 4.1,
high governance diversity suggests that a firm has managed adaptations that are
either autonomous or coordinated by managing different degrees of incentives and
administrative controls. Thus, the firm is likely to have created adaptive capability to
disturbances in a relationship. With governance diversity, a firm derives knowledge
about how different governance forms may or may not work. Broad knowledge
about alternative governance forms reduces the firm‘s path-dependency on any
specific form. That is, the influence of historic decisions dwindles as governance
diversity increases. When past decisions impose minimal effects on the current
choice of governance, firms can see clearer investment opportunities and choose
governance forms that are better aligned with contractual hazards.
Explorative learning also provides the benefit of absorptive capacity (Cohen
& Levinthal, 1990), which enables a firm to constantly absorb new information and
technologies and avoid being locked-out from new developments. Absorptive
capacity has mostly been analyzed with respect to technological capabilities (Zahra
& George, 2002), but it can also be applied to examine firm‘s managerial capabilities
147
from governance decisions. Firms with a certain level of governance diversity
absorptive capacity are aware of the effectiveness of different structures, and thus are
more capable to understand, design, and implement governance forms that are novel
or more effective. Experienced firms in various governance forms may be more
capable to select better partners, understand how to organize relationship more
effectively, and better anticipate and respond to technological, relational, or market
contingencies. Governance diversity enables a firm to bring in decisions factors from
explorative experience, to rely less on its past decisions, and to make better informed
decisions about governance forms. In sum, the diversity of governance experience as
a form of explorative learning allows a firm to compare a more complete set of
alternative governance forms. The knowledge about all possible alternatives reduces
the dependency on past governance decisions.
Hypothesis 4: Diversity of governance experience weakens the effect of path-
dependency on alliance governance decisions.
The preceding hypotheses describe an empirical model of firms‘ governance
decisions. This model is depicted in Figure 4.1 to be tested in the empirical sections.
The top part of the figure depicts the traditional TCE approach to alliance
governance decisions. TCE focuses on ex post contractual hazards that may arise
from opportunism inherent in contracts, and offers optimal choices of governance
form for transactions with a specific set of attributes. The rest of the figure relaxed
the constraint of generic firm capabilities, and unpacks alliance experience at the
148
firm level. Drawing from the resource-based view and learning perspective, the
model suggests that the depth of experience in specific governance forms creates
path-dependent governance decisions. As alliance experience accumulates, firms
also may experience diverse governance forms. The governance diversity as a form
of explorative learning benefits governance decisions by allowing firms to identify
contractual hazards with greater accuracy and salience, and to choose governance
forms under less influences from historic decisions. In sum, the presence of firm-
specific capabilities derived from experience suggests an alternative set of factors
that accompany the transaction cost considerations in predicting the choice of
alliance governance.
Figure 4.1: A Model of Alliance Governance Decisions
H1a, H1b, H1c: +
H2
Contractual Hazards:
Appropriability Risk
Measurement Difficulty
Task Complexity
Alliance Experience:
Prior Alliances in Each
Governance Mode
Hierarchical
Alliance
Governance
Alliance Experience:
Diversity of
Governance Forms
H4: -
H3a:0; H3b:+; H3c:+
149
4.4 Data and Methods
4.4.1 Sample
Industries characterized by frequent alliance formation are ideal to test the
above hypotheses. This essay focuses on the US software industry, technically
identified by four-digit SIC codes 7371-7374 (Lavie, 2007). Computer software has
been a leading force in technological advancements during the past three decades.
This industry has witnessed constantly technological updates and innovations,
arguably one of the fastest changing industries. In such dynamic environment, it is
crucial for software companies to speed up capability development by leveraging and
integrating external resources through collaborative relationships. Indeed the
software industry has witnessed a proliferation of interfirm collaborations since the
1990s. A recent study by Schilling (2008) shows that the software industry is among
the most active industrial sectors in which alliances are frequently formed
7
.
Some alliances in the late 1990s might have been formed as a fad or fashion
to catch up the fervent market before it crashed, and were grounded on weak
strategic rationales. This study thus examines alliances that were formed after the
market crash, between the years of 2001 and 2005 (inclusive). This is a period during
which irrationality in alliance formation has diminished significantly. While software
companies have become more cautious in scrutinizing and engaging in new alliances
7
Her study suggests that other industries primarily consist of biotechnology, pharmaceutical,
telecommunications, airlines, and miscellaneous manufacturing.
150
(see Chapter 3 for alliance activities during the sample period), the industry as a
whole remain more active in alliance activities than most other industries. Frequent
alliance activities by software companies offer variance, reliability, and
meaningfulness for the above hypotheses developed.
I examine a list of software companies and their announced alliances between
2001 and 2005. The sample contains 154 US software companies identified by their
reported primary 4-digit SICs (7371-7374), publicly listed on major stock markets
(NYSE, AMEX, and NASDAQ). This sample is representative because US software
companies have been dominating software businesses worldwide. As shown in
Chapter 3, the US companies have long been taking up over 50% of the global
market values. A caveat in the sample is that publicly listed companies are often
larger and more resource abundant than private companies. However public
companies provide two key advantages to test the hypotheses. First, large companies
are more likely than smaller ones to have accumulated a meaningful amount of past
alliances from which useful inferences can be applied to future governance decisions.
Second, interviews with industry experts suggest that alliance agreements may be
implicit or informal among private companies. Some companies are hesitant to
classify relationships as alliances for publicity or legal concerns. Therefore, it can be
difficult or even infeasible to identify a private company‘s alliance portfolio. Public
companies thus are relatively more visible in their alliance activities, and represent
an ideal group for this study.
151
The list of companies is presented in Appendix 2. Their newly formed
alliances during 2001-2005 are collected from two sources. First, Securities Data
Corporation‘s (SDC) Joint Venture/Alliances database provides basic alliance
information. SDC has been arguably the most complete archival database for
strategic alliances that are formed by public companies (Sampson, 2005; Schilling,
2008). Whereas SDC has certain completeness in its methodology of data collection
(Lavie, 2007), it remains the most comprehensive data source for alliance research
(Schilling, 2008). To overcome SDC‘s insufficient data collection, I follow Lavie‘s
(2007) method by obtaining alliance announcements from Lexisi-Nexis Academics
alliance-relevant keyword search
8
. Each data source contributes about 50%
observations to the final sample. Both sources rely on public information of alliances,
including news wires, SEC filings, analyst conference, and announcements on
company websites.
SDC describes alliance characteristics using multiple variables, such as main
value-chain activities and participants‘ background. Unfortunately, there are no
documentations about SDC‘s coding schemes that can be directly applied to code
alliances from other sources. To ensure that Lexis-Nexis announcements are coded
in a consistent way with SDC, I first identify SDC coding pattern by going through
its alliance records, and then code Lexis-Nexis records following the identified
8
Key words include a wide range of terms, e.g. strategic alliance, strategic partnership,
collaboration, joint agreement, etc. In some announcements partners simply describe
alliance activities without a label of alliance. I started by using highly inclusive terms,
which result in about 3200 announcements. Reading through each announcement reduces
the total number of collaborative alliances to 1700.
152
pattern from SDC. The overlapped alliance reports are particularly instrumental in
creating this pattern. As SDC reports its news sources, I searched on Lexis-Nexis for
the each piece of announcement from the same data source. Then I compared
overlapped announcements to verify the coding scheme with SDC information. This
coding process was simultaneously conducted by three research assistants; and final
coding results were compared and compiled. In case of contradicting information, we
compared information from alternative sources of Google Finance and Hoover‘s
business profiles. Finally, the 154 focal companies yielded a sample of 1700 alliance
announcements.
4.4.2 Measures
4.4.2.1 Dependent Variable: Governance Forms
Alliance governance modes are classified into four categories along a
continuum of the degree of control over collaboration (see Table 4.1). This
classification is built upon Oxley‘s (1997) three-form continuum along the degree of
hierarchical control, namely unilateral contract, bilateral contract, and equity joint
venture. Unilateral contracts primarily consist of unilateral licensing, outsourcing
agreements, and long-term supply contracts. On the opposite end of the continuum is
equity joint venture (JV), a separate business entity built upon resources from both
parent firms. Partners manage a JV through its own board of directors as well as
parent firms‘ management. With own administrative systems, JV collaborations
resemble most closely the hierarchy structure, and are coordinated on a daily basis
153
by the JV‘s own operational routines. JV partners have the highest level of
administrative control over collaborative behaviors.
The bilateral category in Oxley‘s (1997) classification may cover alliances
involving different levels of joint efforts. For example, firms in bilateral licensing
agreements and original equipment manufacturing (OEM) agreements often perform
their responsibilities separately. A partner has little influence on the other beyond
financial investments or payment schemes. Assets invested in such alliances are
protected within each partner‘s boundary, and are subject to relatively less
appropriation risk. However, alliances of joint marketing, R&D, and system
integrations require more frequent and interactive efforts between partners. Firms in
these alliances need to pool resources into collaboration to materialize the values of
alliance. With pooled resources and collective alliance values, the risk of partner
misappropriation is greater in collaborative alliances than bilateral but non-
collaborative alliances. Thus the bilateral contracts are further divided into two
subgroups: with versus without collaborative work.
SDC reports whether an alliance is joint venture or not, but provides no
further governance information. The other three classes of governance forms thus are
defined based on the text information in each announcement. For the publicity
purpose, firms indeed report their respective responsibilities in alliance
announcements. Thus it is relatively clear to differentiate unilateral from bilateral
relationships. With respect to bilateral relationships, announcements may not cover
154
so many details as to identify the degree of collaborative work. According to field
interviews and prior research (e.g. Anand & Khanna, 2000a), however, alliance
announcements serve the purpose of boosting market values by positive firm news.
As a result, companies that announce their alliances prefer to reveal collaborative
factors whenever possible. Given this incentive, public announcements of bilateral
alliances in this sample are classified into with- and without-collaboration groups
with reasonable accuracy.
The dependent variable of governance decisions thus contains four ordinal
values in the following order: 1 if unilateral contract (e.g. unilateral licensing, long-
term supply contracts, unilateral retail & distribution); 2 if bilateral contract without
collaborative work (e.g. cross-licensing, bilateral distribution, OEM contract); 3 if
bilateral contract with collaborative work (e.g. joint product development, system
integration, joint marketing campaign); and 4 if equity joint venture. Given that this
classification may be too nuanced, I further re-grouped the four classes into a
dichotomous variable: 0 if no collaboration (i.e. governance structure 1 and 2 above),
1 if collaboration presents (i.e. governance structure 3 and 4 above). The
dichotomous variable and the ordinal variable yield highly similar regression results.
Thus I report results below using only the ordinal dependent variable.
4.4.2.2 Independent Variables
Hypotheses 1a, 1b, and 1c suggest that exchange hazards in terms of
appropriability risk, measurement difficulty, and task complexity increase the
155
likelihood that a firm will govern a new alliance with more hierarchical structures.
Appropriability risks arise from two key factors in alliances: R&D and technology
transfer. R&D activities create innovative outcomes with uncertain values. Firms
collaborate on R&D activities are subject to the risk that collective know-how in
R&D outputs may be appropriated by a partner. When technologies are transferred
between partners, the receiver may breach the sender‘s benefits by reverse-
engineering or deploying transferred technologies inappropriately. While these
factors are difficult to quantify, I measure appropriation hazards by their presence:
appropriability risk takes the value of 0 if an alliance involves neither R&D nor
technology transfer, 1 if either factor is present, and 2 if both are present.
Measurement difficulties stem from two sources: software development and
cross-border participants. In high-technology industries, it is often costly to verify
system functionality when new software is developed. For example, firms have
difficulties identifying whether software developed under a contractual agreement
can effectively reduce the costs of system upgrade to a pre-specified level (Mayer &
Nickerson, 2005). In the sample, some alliances involve software development
where one partner customizes software functions to meet the other‘s need in time-
saving, cost reduction, and efficiency improvement. However, these targets are often
not directly observable once the new software is deployed. In addition, when
partners are in different countries, it is costly and time-consuming to calibrate their
counter-parties invested efforts invested and final outputs (Oxley, 1997; Sampson,
156
2004). Therefore, the variable of measurement difficulty takes the value of 0 if an
alliance involves neither software development nor cross-border participants, 1 if
either factor exists, and 2 if both are present.
The complexity of alliance tasks can be characterized by the scope of alliance
activities and the number of partners involved. The scope of alliance activities
describes tasks in which partners need to collaborate, i.e. different value-chain
activities necessary for the focal alliance. A broader scope of activities indicates that
partners need to collaborate in different functional areas (Oxley & Sampson, 2004).
This scope is measured by counting the number of value-chain activities specified in
an alliance announcement
9
. Most alliances in the current sample have no more than a
couple of activities: the scope measure has an average of 1.61 and a standard
deviation of 0.67. In addition, alliances with three or more partners, such as research
consortia, require greater coordination efforts among the parties. Lacking efforts or
deviating behaviors from any partner may result in inferior alliance outcomes. Thus I
use a dummy variable to capture multi-partner alliances: 1 if there are three or more
partners and 0 if two partners are involved. The majority of alliances in the sample
are dyadic, with an average of 0.05 and standard deviation of 0.21. Subsequently,
task complexity is defined as 0 if an alliance covers only one activity and involves
two partners, 1 if either activity scope is greater than one or there are three or more
9
This variable has a maximum value of 5, including activities of (1) new product
development, (2) software and system integration, (3) marketing and distribution, (4)
wholesale and retail, and (5) licensing.
157
partners, and 2 if more than one activity and more than two partners are involved in
the focal alliance.
Hypotheses 2-4 suggest that the two dimensions, depth and breadth, of
alliance governance experience affect governance decisions in different ways.
Therefore, this essay depicts alliance experience by the experience in each
governance form and the governance diversity. Using a similar method as the
alliance experience literature (Hoang & Rothaermel, 2005; Sampson, 2005),
governance-specific experience is measured by a discounted sum of alliances that
fall into each governance type (i.e. governance 1 through 4), and are formed prior to
the date of a focal alliance announcement. The history of each firm‘s alliances was
traced back up to 1990 using SDC data. Historic alliances are discounted using a
factor of 0.2, that is, the value of past experience depreciates 20% each year. While
this factor is consistent with other alliance research (Lavie, 2007), alternative
discount factors of 0.1 and 0.3 are also tested. Different discount factors show no
significant impacts on estimated results. The discussion of Hypothesis 2 suggests
that experience in less hierarchical structures (e.g. structures of 1s and 2s) increases
the likelihood of choosing these structures again, and decreases the likelihood of
choosing structures of 3s and 4s for a new alliance, and vice versa. The overall
alliance experience, measured as the sum of alliances across different governance
structures, is included as a control for general alliance experience.
158
Governance diversity describes the degree to which prior alliances have been
managed under different organizational structures. I measure diversity using a
Herfindal index based on the four governance modes in Table 4.1. A Herfindal
measure is defined as
4
1
2
) / ( 1
i
i i
g g H , where g indicates the number of
alliances that falls into a particular governance form. The calculation is based on the
proportion of alliances in each governance mode out of a firm‘s entire alliance
portfolio. In this sample, the diversity has a maximum of 0.75 (i.e., 1-4*(1/4)
2
).
Experience variables are all annualized instead of counting up to alliances that are
immediately preceding a new announcement. The assumption here is that alliances
formed immediately before a new announcement may pose little impact on the
current one, because alliance experience takes time to be assimilated into the pool of
firm resource. Meanwhile, annualized variables are legitimate based on an
assumption that all alliances that are formed last year are likely to exert similar
impacts on new alliances in the current year.
4.4.2.3 Control Variables
In analyzing firm-level experience, a firm‘s resources enter into the decisions
of choosing governance modes. Resource-abundant firms are likely to have greater
freedom in choosing how to leverage internal resource and capabilities. Therefore, I
include firm size (the log number of employees divided by 100), slack (cash in $mil),
and return on assets in the previous year as controls for firm resources. These
variables are collected from Compustat‘s annual database. In addition, product
159
market strategies may influence firm decisions to participate in differently structured
relationships. Therefore, I calculate a Herfindal index of product market diversity for
each firm using CRSP‘s four-digit SIC-based segment data. In this calculation,
product diversity is a conservative measure of its true value, because companies are
required to report segments that contribute more than 10% of their total revenue.
Year dummies are also included but omitted from the result table. Robust standard
errors are clustered on firm.
Table 4.2 presents summary statistics for the above described variables.
Correlations that are significant at 10% level are labeled by asterisks. Because some
observations (i.e. alliances) have missing values in control variables, these
observations are dropped in the regressions. The resulting sample contains 150
companies and their 1511 alliance announcements. High pair-wise correlations
appear among total alliance experience, firm size, and slack. This is expected
because large firms, given their visibility and resource abundance, are more likely to
be pursued as attractive alliance partners. Therefore, large companies tend to
accumulate a large pool of alliances. When including total alliance experience, I drop
variables of size, slack, or both in alternative regressions to detect whether the high
correlations make the regression model unstable. All results remain the same except
that the product diversity coefficient loses its significance when the size and slack
variables are both dropped.
160
Table 4.2: Summary Statistics
Obs Mean Std Dev Min Max
1. Governance Structure (1-4) 1511 2.467 0.825 1 4
2. Firm Size (employees) 1511 0.503 1.965 -4.423 4.111
3. Firm Slack ($millions) 1511 0.858 2.328 0.00004 15.982
4. Return on Assets 1511 -0.030 0.318 -3.095 0.425
5. Product Diversity 1511 0.240 0.275 0 0.766
6. Appropriability Risk 1511 0.286 0.494 0 2
7. Measurement Difficulty 1511 0.593 0.669 0 2
8. Task Complexity 1511 0.558 0.544 0 2
9. Governance Diversity 1511 0.439 0.169 0 0.745
10. Log (total alliances) 1511 2.832 1.298 0 6.245
Pair-wise Correlations 1 2 3 4 5
1. Governance Structure (1-4) 1
2. Firm Size (log of employees) 0.1870* 1
3. Firm Slack (cash) 0.1813* 0.5745* 1
4. Return on Assets 0.0652* 0.3999* 0.2154* 1
5. Product Diversity 0.0435* 0.5582* 0.4498* 0.3187* 1
6. Appropriability Risk 0.1448* -0.0513* -0.0288 -0.022 -0.037
7. Measurement Difficulty 0.3019* 0.1291* 0.1782* 0.0505* 0.0253
8. Task Complexity 0.3522* 0.0697* 0.0415 -0.017 0.0042
9. Governance Diversity -0.3079* 0.0560* 0.0336 0.0583* 0.0734*
10. Log (total alliances) 0.1704* 0.7163* 0.6616* 0.2839* 0.4241*
Pair-wise Correlations: continued 6 7 8 9
6. Appropriability Risk 1
7. Measurement Difficulty -0.012 1
8. Task Complexity 0.0194 0.2161* 1
9. Governance Diversity -0.0074 -0.0906* -0.0662* 1
10. Log (total alliances) 0.012 0.2070* 0.0565* 0.1847*
161
Given that the dependent variable contains four ordinal values, an ordered
probit model is appropriate for empirical tests. There are three cutoff points to be
estimated as constants:
2 1 0
, , ,
According to the hypotheses,
1
b and
3
b are expected to be positive. Variables
of specific governance experience have different signs in the vector of
2
b , and the
signs in
4
b is opposite to those in
2
b due to the predicted negative moderating effects.
Unobserved firm factors can influence alliance governance decisions. Fixed firm
effects lead to a significant loss in the degree of freedom, thus all models are run
with variances clustered on each firm to obtain robust standard errors. Random firm
effects were tested in robustness checks. While not reported, year dummies of 2002-
2005 are included to account for year-specific fixed effects, and the year 2001 is
omitted as the default group. There are no significant year effects in the full sample.
) ( 1 ) | 4 Pr(
) ( ) ( ) | 3 Pr(
) ( ) ( ) | 2 Pr(
) ( ) | 1 Pr(
2
1 2
0 1
0
i i i
i i i i
i i i i
i i i
X X Form
X X X Form
X X X Form
X X Form
ijt it
t i
t i ijt
t i
ijt
ijt
Controls b
Diversity Governance Experience b
Diversity Governance lHazards Contractua b
Experience b
lHazards Contractua b a
g g Governance
5
1 , 4
1 , 3
1 , 2
1
] * [
] * [
) 4 , 3 , 2 , 1 ; Pr(
162
4.5 Results
The estimation results are presented in Table 4.3. Model 1 tests Hypotheses
1a, 1b, and 1c. The hypotheses suggest that hierarchical governance forms are more
efficient in minimizing transaction costs when contractual hazards are great.
Appropriability, measurement, and complexity problems at the alliance level are all
shown to increase the probability that a firm will choose more hierarchical forms,
and their effects are significant at the 1% level. Considering alliance characteristics,
task complexity is a more important determinant than the other two types of hazards.
The economic significance of complex alliance activities is about 70% higher than
those of appropriability and measurement difficulty. Therefore Hypotheses 1a, 1b,
and 1c are supported. Moreover, when ignoring firm-level alliance capabilities,
alliances‘ task complexity is a more salient concern to firms‘ governance decisions.
In addition, firm-level resources positively influence the likelihood of hierarchical
governance forms. This is reasonable as hierarchical structures are more costly to set
up.
Model 2 adds alliance experience measures of governance diversity and
overall alliance experience. The number of total alliances takes the log form because
the linear distribution of alliance experience in the sample is highly skewed. Overall
alliance experience is included to draw a comparative inference with prior research
on alliance experience, because prior alliances are often considered as positively
related to firm- and alliance-level outcomes (e.g. Reuer et al., 2002; Sampson, 2005).
Interestingly, total past alliances have no significant impact on the governance
163
decision in new alliances. Given that contractual hazards now have slightly larger
economic significance, this finding suggests that general alliance experience may
enable firms to be more cautious about potential risks, rather than affecting
governance choices directly. The governance diversity variable reduces the
likelihood of hierarchical governance forms with strong economic and statistical
significance. One explanation is that firms benefit from diverse governance
experience because of the explorative learning: when they are aware of more
alternatives, they have greater freedom to choose, and thus become less likely to
focus on any specific governance forms. According to the definition of the dependent
variable, firms are thus less likely to choose more hierarchical governance forms.
Model 3 breaks down total alliance experience by governance forms. At the
firm level, more experience in unilateral contracts reduces the probability that a firm
will choose structures of bilateral collaboration or joint venture. The experience in
bilateral, non-collaborative alliances has a similar impact but to a lesser degree. On
the other end, joint venture experience makes a firm more inclined to choose joint
venture again. Bilateral collaborations also generate similar path-dependent decisions,
but their impacts are at a much lower degree than experience in the other three
governance types. Therefore, Hypothesis 2 finds strong supports. While contractual
hazards remain their significance in driving toward hierarchical structures, their
relative effect sizes are different than Model 2. When examining the experience
effect of each alliance governance form, appropriability risk becomes more
important a determinant.
164
Model 4 includes interaction terms to test the moderating effects of
governance diversity. Governance diversity is predicted to create benefits of
explorative learning and enable greater ability to foresee contractual hazards.
Specifically, diverse governance experience provides the knowledge of possible
difficulties that may arise from calibrating partner inputs, measuring output quality,
coordinating multiple alliance activities, and managing complex relationships.
Hypotheses 3b and 3c are supported, as diversity significantly strengthens the effects
of measurement difficulty and task complexity on governance decisions. Hypothesis
3a suggests that firms may willingly expose themselves to certain appropriation
hazards for the goals of innovative and creative outcomes, even though they are able
to identify such risks. As a result, the diversity of governance experience may have
little influence how appropriation risk is perceived in a specific alliance. The
insignificant coefficient on the interaction between diversity and appropriability risk
thus supports Hypothesis 3a.
Note that the main effects of contractual hazards decrease when interaction
variables are included. The coefficient of task complexity has dropped from 0.7
(significant at 1% level) to 0.352 (significant at 10% level). Measurement difficulty
shows no significant effect different than 0, a substantial decrease from 0.456
(significant at 1% level) in Model 3. These changes suggest that, while governance
decisions are importantly influenced by contractual hazards, the degree of such
influence is importantly bounded by a firm‘s capability in assessing potential hazards.
When firm-specific managerial capabilities are taken into account, it is contractual
165
hazards under certain level of capabilities, rather than the hazards per se, that driving
alliance governance decisions. Consistent with Hypothesis 3a, appropriability risk is
often voluntarily accepted by a firm, thus its size and significance are relatively
independent of firm capabilities.
Finally, Model 5 adds interactions between governance diversity and
governance-specific experience. Hypothesis 4 argues that diverse experience
broadens a firm‘s knowledge about alternative governance forms, and thus reduces
the propensity of making path-dependent decisions. The interactive variables
between governance diversity and experience in the first two governance types yield
positive coefficients; because the main effects of unilateral and bilateral, non-
collaborative contracts are negative, the positive coefficients support the
hypothesized negative moderating effects. However, path-dependency in
collaborative bilateral contracts and joint ventures are not influenced by diverse
governance experience. One possible reason is that more hierarchical governance
structures are more difficult to negotiate, design, and implement. Thus, decisions to
participate in these relationships are more likely to be stringently scrutinized rather
than following prior decisions. Indeed, in Model 5 the coefficients of collaborative
governance experience are at a much smaller size than those of non-collaborative
experience, suggesting that path-dependency is stronger in non-collaborative
relationships than in collaborative ones. In sum, Hypothesis 4 finds partial supports
in that diverse experience mitigate the path-dependent governance decisions to the
extent of less hierarchical, less collaborative alliances.
166
In Model 5, the only significant contractual hazard variable is that of
appropriability risk. Firms are willing to bear appropriation hazards regardless of
their capabilities in recognizing such risks. Meanwhile, the interacted variables
between diversity and contractual hazards remain robust after adding additional
experience interactions. In contrary to contractual hazards, the path-dependency in
governance choices remains stable and even stronger when the moderating effects of
diversity are concerned. In fact, coefficients of each governance experience variables
have gained considerable economic size, except for joint venture experience. This is
because governance diversity as a negative moderator takes away opposing effects
that were mixed in alliance experience in Model 4.
Table 4.3: Ordered Probit Model Regression Results
VARIABLES Model 1 Model 2 Model 3 Model 4 Model 5
Cut 1 -0.364*** -1.326*** -1.240*** -1.730*** -1.914***
(0.111) (0.225) (0.184) (0.260) (0.269)
Cut 2 0.272** -0.632*** -0.530*** -1.013*** -1.166***
(0.123) (0.198) (0.168) (0.255) (0.260)
Cut 3 3.093*** 2.334*** 2.507*** 1.996*** 1.910***
(0.144) (0.222) (0.162) (0.253) (0.258)
Firm Size 0.0941* 0.0772 0.0543 0.0577 0.0187
(0.049) (0.054) (0.041) (0.040) (0.031)
Firm Slack 0.0608** 0.0507* -0.00443 -0.00783 0.0329*
(0.030) (0.027) (0.015) (0.015) (0.020)
Return on Assets 0.0905 0.136 0.223 0.233 0.181
(0.179) (0.158) (0.180) (0.180) (0.148)
Product Diversity -0.371* -0.331* -0.071 -0.0748 0.00172
(0.216) (0.177) (0.166) (0.163) (0.144)
Appropriability Risk (H1a) 0.398*** 0.401*** 0.465*** 0.339** 0.351**
(0.094) (0.088) (0.094) (0.153) (0.139)
Measurement Difficulty (H1b) 0.463*** 0.424*** 0.456*** -0.0331 -0.069
(0.062) (0.064) (0.066) (0.138) (0.135)
167
Table 4.3: Continued
VARIABLES Model 1 Model 2 Model 3 Model 4 Model 5
Task Complexity (H1c) 0.712*** 0.726*** 0.700*** 0.352* 0.192
(0.074) (0.075) (0.075) (0.180) (0.164)
Governance Diversity
-2.431*** -1.704*** -2.751*** -3.250***
(0.307) (0.337) (0.522) (0.519)
Log (total alliances)
0.0749
(0.066)
Gov 1: Unilateral (H2)
-0.0397** -0.0412** -0.540***
(0.017) (0.016) (0.085)
Gov 2: Bilateral, without Collab (H2)
-0.0302** -0.0283* -0.339***
(0.014) (0.015) (0.108)
Gov 3: Bilateral, with Collab (H2)
0.00632* 0.00731** 0.0466***
(0.004) (0.004) (0.014)
Gov 4: Equity JV (H2)
0.0454* 0.0400* -0.00538
(0.025) (0.024) (0.177)
Diversity*Appropriability Risk (H3a)
0.24 0.267
(0.340) (0.312)
Diversity*Measurement Difficulty (H3b)
1.106*** 1.133***
(0.329) (0.330)
Diversity*Task Complexity (H3c)
0.755* 1.094***
(0.390) (0.359)
Diversity*Gov 1 (H4)
0.867***
(0.151)
Diversity*Gov 2 (H4)
0.439**
(0.175)
Diversity*Gov 3 (H4)
-0.00667
(0.030)
Diversity*Gov 4 (H4)
-0.056
(0.288)
Year Dummies Yes Yes Yes Yes Yes
Observations 1511 1511 1511 1511 1511
Chi Square 402.9 488.6 636.9 694.7 1024
Log Likelihood -1325 -1254 -1228 -1219 -1187
Pseudo R-Square 0.132 0.178 0.195 0.201 0.222
Firm Clusters 150 150 150 150 150
Robust standard errors in parentheses *** p<0.01, ** p<0.05, * p<0.1
168
Several robustness checks are conducted based on Table 4.3. First, the
present definition of dependent variable indicates that the four classes of governance
structures are ranked along a continuum of control power over collaboration.
However, these structures may not be ordered as depicted here. Thus I test the same
models above using multinomial logit regressions, because multinomial models do
not assume any order among the discrete values of a dependent variable. Second, the
four classes of governance modes were re-grouped into a dichotomous variable, 1 if
there is collaboration and 0 otherwise, to test whether the classification of four
governance structures contains unnecessary nuanced differentiation. Further, random
effects of firm and year are included, instead of the fixed effects and clustered
variance in Table 4.3. All these robustness tests were run following the order of
Models 1 through 5. The alternative regressions produce results that are highly
robust and similar to those presented above.
4.6 Discussion and Conclusions
4.6.1 Summary of Findings
In this study, I attempt to develop a deeper understanding of how firm-
specific, experientially derived capabilities shape governance decisions in
conjunction with contractual hazards. Williamson (1999) proposes to move beyond
the focus on transaction level hazards to take into account heterogeneous firm
capabilities. Specifically, the question on governance decisions should be addressed
as ―How should firm A—which has pre-existing strengths and weaknesses (core
169
competencies and disabilities—organize X?‖ (p. 1103). Recent strategy literature has
responded to this call by presenting that firm-level technological capabilities enter
into assessments of contractual hazards when firms make governance decisions (e.g.
Hoetker, 2005; Leiblein & Miller, 2003; Mayer & Salomon, 2006).
The findings in this essay support the contingent view of contractual hazards
and firm capabilities. Consistent with the TCE logic, I found that contractual hazards
motivate firms to choose more hierarchical governance forms in a new alliance.
Following the argument of path-dependent capabilities in the RBV perspective, I
found that firms prefer to choose governance forms in which they have accumulated
a high level of experience. Path-dependent governance decisions provide the
opportunity to exploit governance structure-specific knowledge and capabilities
developed from prior experience. When choosing governance forms, firms recognize
both contractual hazards and own capabilities built from prior experience.
More interesting findings are presented when examining the construct of
governance diversity of alliance experience. Experience in diverse governance
structures represents a form of explorative learning about how different structures
may function in adapting to various contractual hazards. With respect to contractual
hazards, diverse knowledge allows a firm to be more capable to identify the hazards
in measurement and complexity of an alliance, thus assessing contractual hazards as
more salient concerns. Interestingly, appropriability risk appears to affect governance
decisions independent of diverse governance experience. This evidence implies that
diverse experience may shed no light on the assessment of appropriation hazards that
170
are highly specific to a new relationship, because of the idiosyncrasy of the alliance
or the firm‘s incentive to bear such risk.
With respect to path-dependent capabilities, diverse governance experience
broadens the range of knowledge that is related to governance forms, and weakens
the influence of historic governance decisions on a current alliance. Thus governance
diversity is predicted to negatively moderate the effects of prior governance-specific
experience. However, diversity is shown to mitigate the impact of past governance
decisions only to the extent of alliances without collaborative efforts. Collaborative
relationships are much more costly to negotiate and set up. Effects of collaborative
governance experience thus are less likely to be affected by governance diversity.
4.6.2 Implications
By focusing on experientially derived capabilities, this essay provides several
implications. First, the findings highlight the need for further integration between
TCE and the capability perspective by analyzing the managerial capabilities from
cooperative experience. While technological capabilities in R&D, marketing, or
production foster firm capabilities to govern transactions due to in-depth expertise
(Leiblein & Miller, 2003; Mayer & Salomon, 2006), a more direct source of
governance capabilities is prior experience in managing different alliance structures.
Scholars need to better understand to what extent governance capabilities are derived
from the managerial aspect, or the past experience with specific governance forms.
Second, this essay proposes a more thorough understanding of possible
dimensions of firm experience. Experience contains multiple aspects beyond a
171
simple count of the total number of prior relationships. And these aspects create
learning benefits in different ways. In-depth experience creates values that can be
exploited through repetitions. When alliance governance structures are relatively less
costly to set up, firms may simply follow prior governance decisions in a new
relationship. However, exploiting prior knowledge limits a firm in learning about
alternative governance structures. It is important to derive a certain level of diversity
across different governance forms. Diverse experience broadens the range of
knowledge and capabilities to make better informed governance decisions according
to contractual hazards, and to rely to a lesser extent on historic governance choices.
Further, the varying size and significance of the three contractual hazard
variables suggest that we should not treat all contractual hazards as equal in
governance decisions. While the effects of some hazards (i.e. measurement and
complexity) are bounded substantially by experience-based capabilities, others (i.e.
appropriability) are relatively independent of firm-level capabilities because of their
specificity to an alliance or to a partner. Therefore, future research should examine
whether and how different types of firm capabilities affect different types of
contractual hazards.
In sum, the findings in this essay highlight alliance governance decisions as
contingent outcomes on both transaction-level contractual hazards and firm-level
experience-based capabilities. Alliance experience can be conceptualized in distinct
dimensions, from which firms can draw different inferences in subsequent
governance decisions. Firms derive in-depth, exploitative knowledge from focused
172
experience in any particular governance form. At the same time, experience across
various governance forms generates diverse, explorative knowledge, enhancing a
firm‘s capabilities to identify contractual hazards and possible alternative structures.
4.6.3 Contributions
This essay joins recent academic effort in integrating the TCE framework
with firm capabilities. Empirical research has widely supported the TCE logic of
discriminating alignment between contractual hazards and governance structures (e.g.
Leiblein et al., 2002; Oxley, 1997; Poppo & Zenger, 1998; Sampson, 2004), and
research drawing from the capability perspective has suggested firm capabilities as
an important determinant for governance decisions (e.g. Hoetker, 2005; Leiblein &
Miller, 2003; Mayer & Salomon, 2006). While the idea of capabilities
complementing TCE logic is not new, this essay makes several contributions to this
integrated stream of research. First, firms can develop capabilities that are pertinent
to alliance governance forms by managing differently structured relationships. This
type of capabilities is parallel to but different from technical expertise, and also
affects subsequent governance decisions. Indeed, Mayer and Salomon (2006) point
out that other types of capabilities, beyond technological expertise, deserve more
academic research. Managerial or governance capabilities arise from the experience
of negotiating, designing, and managing different alliances. In-depth experience
enhances the efficiency of utilizing the same governance form again, and broad
experience enables better informed governance decisions based on knowledge about
diverse alliance structures.
173
Second, this essay contributes to the transaction cost theory by illustrating
possible cognitive limitations of economic actors. TCE assumes that economic actors
have the capability to look ahead and recognize contractual hazards and investment
opportunities. However, such foreseeing capabilities are often the result of
organizational learning from prior experience. In fact, Williamson (1999) has
pointed out that learning through experience is ―more ambitious than merely trial-
and-error learning, but is less ambitious than the idea of farsighted contracting‖
(p.1104). The findings in this essay provide empirical evidence for Williamson‘s
insight. Specifically, economic actors are capable in identifying certain levels of
contractual hazards, but they become more cautious and more alert about the hazards
when they have experience in diverse organizational structures. Given that learning
often occurs locally and incrementally (Levinthal & March, 1993), firms may be
myopic in recognizing hazards with limited, governance-specific experience. But this
cognitive limitation is mitigated when firms explore governance alternatives.
Therefore, firms are not necessarily farsighted about ex post contractual hazards. The
degree to which a governance structure is properly aligned with a transaction‘s
attributes importantly depends on a firm‘s learned capabilities.
The findings also bridge theories in the strategy literature regarding the links
between organizational learning and governance. Empirical learning research has
shown that exploitative learning creates benefits of increased efficiency (e.g. Anand
& Khanna, 2000a; Argote et al., 1990; Rothaermel & Deeds, 2004), and explorative
learning contributes to innovative performance (e.g. Hoang & Rothaermel, 2005;
174
Sampson, 2005). However, there has been little effort examining the effect of
learning on governance decisions. While Argyres and Liebeskind (1999) propose the
interesting concept of governance inseparability, empirical studies supporting their
arguments are still lacking. This essay suggests that both exploitation and
exploration exist in learning from governance experience: the former generates
governance-specific capabilities to be exploited in new alliances, and the latter
strengthens learning in diverse structures and leads to better informed decisions.
Overall, learning takes place not only in technical expertise, but also in managerial
and governance experience. Future research should tease out more dimensions of
experience to analyze how different aspects of a firm‘s history can influence its
strategic decisions and performance.
4.6.4 Limitations and Future Research
Several caveats are drawn from this essay. First, this study uses a single
industry as the empirical context. This focus may limit the extent to which the
findings can be generalized into other industrial settings. However, the single-
industry sample provides comparable industrial background for all newly formed
alliances. Resources and capabilities that are deemed valuable tend to be consistent
across firms in the same industry. In addition, contractual hazards can be defined
more consistently for firms in the same industry. Indeed, multi-industry studies may
suffer from the problem that inputs or variables share the same label of variables, but
are effectually different constructs (Shelanski & Klein, 1995).
175
Second, the measure of governance experience only contains the four
structures classified in Table 4.1. In practice, alliance governance may be further
classified into more nuanced types with respect to the degrees of partner
collaboration and hierarchical power. Viewing governance diversity as explorative
learning, there is a possibility that too diverse governance experience may make
explorative learning ineffective, that is, ―too much of a good thing‖. For example,
Sampson (2005) shows that extensive alliance experience has a curvilinear effect on
a firm‘s innovative performance. With the limited number of governance alternatives,
the diversity measure in this study lacks sufficient variance to identify possible
curvilinear effects on governance decisions. However, to the extent that alliances can
be generally assigned into one of the four types identified in this study, I found no
significant effect of the quadratic form of governance diversity.
Third, the sample of alliances contains no indicators of alliance-level
performance. The success of prior alliances may create a different set of knowledge
about how to govern a new alliance than failed experience. Ideally, performance
information can offer more insights about the learning mechanism from experience.
However, alliance performance is often difficult to measure. Allied firms may have
different and highly subjective criteria for evaluating the success of an alliance.
Instead of directly measuring prior alliance performance, this essay assumes that
firms derive knowledge and capabilities from both successes and failures. Prior
alliance research has shown that learning indeed occurs from failures (Arino & de la
176
Torre, 1998). Interviews with industrial experts also validate that unpleasant or
unsuccessful experience often provides more effective learning in a corrective way.
Finally, repeated interfirm ties within a dyad may also influence the choice of
alliance governance (Gulati, 1995; Zollo et al., 2002). Repeated relationships
increase the familiarity between partners, create relational capital, and reduce the
likelihood that the same partners will choose a hierarchical form (Dyer & Singh,
1998; Kale et al., 2000; Ryall & Sampson, 2009). While dyadic characteristics
indeed affect governance decisions, this essay focuses on the experience effects at
the firm level, instead of the dyad level. That is, a firm on average learns about how
to design and manage focused or diverse governance forms regardless of with the
same or different partners. Given that partner-specific learning can create different
effects on performance than general learning (Hoang & Rothaermel, 2005), future
research should examine partner-specific experience as a parallel dimension of
experience to that of depth-breadth in experience.
Whereas the aforementioned limitations exist, this essay provides valuable
implications for both research and practice. Given these findings, further research on
more detailed experience-based capabilities will surely advance our understanding
about governance structures in alliances and in general interfirm relationships.
In addition to the effects on governance decisions, specific dimensions of
alliance experience may draw different market attentions when new alliances are
announced. The next essay examines how stock market takes into account a firm‘s
detailed alliance experience upon the news of alliance formation.
177
CHAPTER 5: ESSAY 2 – EFFECTS OF ALLIANCE EXPERIENCE
ON STOCK MARKET VALUE CREATION
5.1 Introduction
Strategic alliances are interfirm relationships that represent an important
organizational form for governing collaborations (e.g. Ahuja, 2000; Parkhe, 1993b;
Reuer, Zollo, & Singh, 2002). From collaborative relationships, firms build
capabilities and enhance performance by coordinating partner efforts, leveraging
collective resources, and creating proprietary knowledge. Empirical alliance research
has suggested that, on average, alliance experience creates market value: firms with
more prior alliances accrue greater positive abnormal returns when forming new
alliances (e.g. Anand & Khanna, 2000a; Gulati & Singh, 1998; Kale, Dyer & Singh,
2002; Oxley, Sampson, & Silverman, 2009). While the magnitude of value creation
varies with partner capabilities and alliance activities, the overall effect remains
positive. These studies mostly draw from the organizational learning and resource-
based theories, and suggest that the primary benefits of alliance experience consist of
operational efficiencies from learning-by-doing (Sampson, 2005) and coordinative
capabilities from managing inter-partner dynamics (Doz, 1996; Reuer et al., 2002).
Research on alliances has examined alliance experience primarily as a
cumulative sum of a firm‘s prior relationships. The implicit assumption in this
approach is that learning is largely uniform across different alliances. This view of
experience is inherited from the learning-by-doing literature that views a firm‘s total
178
past productions as experience (e.g. Argote et al., 1990). However, while internal
productions are relatively homogeneous, interfirm alliances can greatly differ from
one another. Thus the assumption of uniform learning may not hold when we
examine aspects of a firm‘s of alliance experience. For example, two firms with the
same number of prior alliancesmay have both accrued learning benefits from
experience, but their learned knowledge and skills are likely to differ if one relied on
a single alliance structure while the other managed diverse governance forms. Indeed,
learning processes of exploitation and exploration are shown to affect firms‘
performance in different ways (Rothaermel & Deeds, 2004). Therefore, it is
important to advance our understanding about learning mechanisms by unpacking
the construct of alliance experience into different dimensions.
In this essay, I draw from the learning perspective and the resource-based
view to examine how the market takes into account characteristics of firm-level
experience and capabilities. Two types of capabilities are derived from alliance
experience. First, experience in a particular governance form indicates a firm‘s
capability to manage the focal structure, thus creates structure-specific governing
capabilities. As a firm accumulates alliance experience, it also develops a certain
level of diversity with respect to alliance governance forms. The diversity of
experience expands the range of knowledge about different organizational structures,
and enables a firm to be less constrained by its history and to have greater foresight
in making subsequent decisions. Therefore, diverse experience enables firms to
choose more appropriate governance forms, and builds their selection capabilities as
179
well. When a new alliance is organized under the structure in which a focal firm has
great governing capability, the market will consider this alliance as more likely to
succeed and hence react more positively. The effect of governing capability is
strengthened when the firm has strong selection capabilities, because the alliance is
more likely to be coordinated under the most appropriate structure.
The degree of market value creation is also bounded by other factors that the
market will use as reference points. First, alliance organizational structures often
attract different degrees of market attention. For example, joint ventures are
considered as highly significant events because of the partners‘ physical, financial,
and human resource commitments, and thus invite closer attention than licensing
agreements do (Anand & Khanna, 2000a). Thus the experience factors may be
perceived to influence value creation to different degrees in alliances governed by
different structures. In addition, a focal firm‘s prior performance may signal to the
market a firm‘s overall capabilities (Fombrun & Shanley, 1990; Gulati & Higgins,
2003). Performance signals to the market the possibility that a firm can manage
resources to create value. While a new alliance may impose risks in creating value,
the performance level implies firm capabilities in managing resources in general.
Therefore, the predicted effects of alliance experience on market valuation are
bounded by the current alliance‘s governance form and the focal firm‘s prior
performance.
Empirically, I test the predicted experience effects using an event-study
methodology. Drawing from a sample similar to that in Chapter 4, the sample
180
consists of 1445 alliances announced by 147 focal software companies between 2001
and 2005. Value creation, measured by abnormal stock market returns, is shown to
be positively affected by a firm‘s experience. While prior research demonstrates
experience in general creates value (e.g. Anand & Khanna, 2000a), this essay
highlights the effect of experience in-type, or prior alliances that are governed by the
same type of governance structure. Experience across diverse governance forms
builds capabilities to select more appropriate governance, and thus enhances the
value creation driven by experience in-type. In addition, the degree of value creation
appears to be stronger for firms with an intermediate level of prior performance,
because the market perceives low-performing firms as incapable and gives high-
performing ones the benefit of doubt in managing new alliances. The experience
effects are stronger for alliances governed by an intermediate level of collaboration,
because alliance experience matters most when a new alliance has no formal
administrative control (such as in a joint venture) and collaborative work is
necessary for the alliance outcome.
While the idea that experience creates value is not new in alliance research
(e.g. Anand & Khanna, 2000a; Hoang & Rothaermel, 2004; Sampson, 2005), this
essay makes several contributions by integrating arguments from firm capabilities
and learning benefits. Following Chapter 4, this essay applies the governance
dimensions of experience. Distinguishing between the depth and breadth of extends
the concept of learning beyond repetition of organizational activities, which has been
a common approach in the learning literature (Argote et al, 1990; Vanneste &
181
Puranam, 2008). As a result, different types of capabilities can be derived from
different aspects of experience. Specifically, when entering into new alliances, firms
resort to prior experience to select a proper governance mode ex ante and to govern
the relationship ex post. Further, this essay highlights the boundary conditions of
experience at the firm level and at the alliance level. The learning literature has
generally supported a positive relationship between experiential learning and
performance (e.g. Anand & Khanna, 2000a; Hoang & Rothaermel, 2005; Kale et al.,
2000). The findings in this essay suggest that the benefits of learning can differ by
firm and alliance characteristics. By addressing the characteristics of alliance
experience, this essay joins the emerging alliance portfolio research (Hoffman, 2007;
Lavie, 2007; Lavie & Miller, 2008; Wassmer, 2010) to depict important dimensions
in alliance portfolios that influence firm outcomes.
The rest of this chapter is organized as follows. First, the theories of
organizational learning and of resource-based view are reviewed. Next, I develop
hypotheses about two key aspects of alliance experience, namely experience in
specific governance type and diversity across different types. The following section
describes the data, methods, and measurements used in empirical tests. After the
presentation of empirical results, I discuss the conclusions, contributions, and
limitations of this study.
182
5.2 Literature Review
5.2.1 Organizational Learning and Firm Experience
The organizational learning perspective views organizations as adaptive
actors in response to informational and environmental changes. Learning has been
broadly defined as cognitive and/or behavioral changes, manifested as organizations‘
changing knowledge and in realized or potential activities (Huber, 1991). March and
Simon (1958) first thoroughly analyzed organizational learning, suggesting a
considerable similarity between organizational learning and individual learning
particularly in the sequence of acquiring, storing, assimilating, and implementing
knowledge and information. Later, Levitt and March (1988) characterized learning
as routine-based, history-dependent, and target-oriented processes. In general,
learning is considered to happen when organizations encode inferences from history
into routines of behaviors. The goal of organizational learning focuses on improving
organizational performance, such as survival and growth (Argote, 1999).
Despite an early start (March and Simon, 1958), the learning perspective has
yet to evolve into an integrative theory (Huber, 1991). Research on organizational
learning has covered a wide range of topics with respect to learning processes,
contents, outcomes, and limitations. However, these topics have yet to be integrated
into a uniform framework that can explain or predict strategic decisions, such as the
predictions developed in TCE and RBV. Regarding the process of learning, scholars
suggest that organizations can learn experientially or vicariously (Levitt & March,
1988), exploitatively or exploratively (March, 1991), and internally or externally
183
(Darr et al., 1990). Learning often leads to the outcomes of improved production
efficiency from learning-by-doing (Argote et al., 1990), enhanced technological
competitiveness from building absorptive capacity (Cohen & Levinthal, 1990), and
augmented innovation capabilities from explorative learning (McGrath, 2001).
Recent research on interfirm relationships has demonstrated that learning from
cooperative experience can enhance the chance of alliance successes (Kale et al.,
2000; Zollo et al., 2002) and the participants‘ contract design capabilities (Argyres &
Mayer, 2007; Mayer & Argyres, 2004) as well as innovative performance
(Rothaermel & Deeds, 2004; Sampson, 2005).
Notwithstanding insufficient theoretical integrations, the learning perspective
provides key insights to understand the effects of experience. In general,
organizational experience represents the venue in which experiential learning may
occur. Experience helps to create capabilities, knowledge, and routines that are firm-
specific. Given the organizational context, experiential learning is by nature unique
and path-dependent. The specific capabilities, knowledge, and routines, as unique
and inimitable resources, are instrumental for a firm to adapt to contingencies (Kale
& Singh, 2007; Zollo et al., 2002). Further, experiential learning can cover
organizational activities that are focused, local, and exploitative or those that are
diverse, distant, and explorative (March, 1991). Over time, focused and repetitive
experience can create operational efficiency as well as bring rigidity to firm
knowledge and adaptive behaviors (Levinthal & March, 1993). Thus, distant or
diverse learning is important in expanding the capacity to continuously absorb new
184
information and knowledge to maintain organizational competence (Cohen &
Levinthal, 1990).
Empirical research has primarily established the benefits of repetitive
activities as exploitative learning, but has placed little emphasis on the explorative
aspect of learning. Scholars often build upon the learning-by-doing argument
(Argote et al., 1990; Epple et al., 1991) and measure the frequency of a certain type
of organizational activity. However, repetitive learning processes may constrain a
firm‘s long-run development by confining its knowledge range. Indeed, some studies
have recognized the limitations of learning-by-doing in the form of myopia (e.g.
Tsang, 2002; Mayer & Argyres, 2004). For sustainable development, firms will need
to learn from diverse activities to counter the limitations of local and incremental
learning. While there is not much research on the simultaneous effect of exploitation
and exploration, one exception is the study by Rothaermel and Deeds (2004) that
shows that the process of new product development is benefited from explorative
alliances at the beginning, and benefited from exploitative alliance in later stages.
Thus, the learning theory can be enriched by advancing simultaneous but possibly
different effects from focused versus diverse learning from experience.
Another area to be further explored is how learning occurs in different
organizational structures. The learning research has mostly focused on experience in
the content of activities, for instance, production and development (e.g. Argote et al.,
1990), contractual provisions (e.g. Mayer & Argyres, 2004), and overall alliances
(e.g. Sampson, 2005). However, learning occurs not only in the specific activities
185
conducted, but also in how such activities are organized or governed. That is, the
organizational structures under which activities are managed. The structural aspect
of experience represents a parallel dimension to repetitive activities, and may
influence a firm‘s learning processes and outcomes. There have been few studies
examining how the structural aspect affects learningthe only exception to the best of
my knowledge is the study by Anand and Khanna (2000a), who found that
experience effect varies by alliance governance forms: prior equity joint ventures
provide greater market value to a focal firm than prior licensing agreements. I
discuss in the hypothesis development how this essay contributes to the structural
emphasis on experience.
5.2.2 Resource-Based View and Alliance Capabilities
Organizational learning is often associated with the process of building
organizational capabilities. This connection is drawn because organizations can
derive context-specific, path-dependent capabilities from experiential learning
(Eisenhardt & Martin, 2000; Zollo & Winter, 2002). The argument that firm
capabilities determine competitive performance finds its origins in the resource-
based view (RBV). RBV is built upon two basic conditions. First, firms possess
heterogeneous resources and capabilities (Wernerfelt, 1984). Second, these different
resources and capabilities are limited in supply and are costly to imitate (Dierickx &
Cool, 1989; Barney, 1991). As a result of these conditions, firms that possess and
exploit their uniquely bundled resources and capabilities will enjoy sustained
competitive advantages relative to their competitors (Rumelt, 1984). In sum, RBV
186
emphasizes the opportunity to create competitive advantages by exploiting unique
and causally ambiguous resources and capabilities.
RBV considers the bundle of resources and capabilities as covering a broad
range of firm-level elements: physical resources, operational activities, managerial
skills, tacit know-how, etc. (Barney, 1986; Penrose, 1959; Wernerfelt, 1984).
However, this depiction of resource brings the problem of tautology: it could be
argued that anything that creates value can be defined as a resource (Priem & Butler,
2001). Relatively specific types of resources are then identified, particularly firm-
specific knowledge and capabilities. The knowledge-based view (KBV) follows
RBV by proposing that unique and inimitable knowledge is a key resource that
determines competitive advantage (Foss, 1996; Grant, 1996). Given that tacit know-
how is difficult to articulate and transfer, KBV also proposes that internal
organization structures are superior to the market mechanism in coordinating the
flows of knowledge (Kogut & Zander, 1992).
The capability perspective extends the concept of unique bundles of
resources to firm capabilities as the determinant for organizational decisions and
performance. For example, Madhok (1996) argues that organizational capabilities
provide a complementary explanation to the transaction cost theory of market
failures. He suggests that it is capabilities under bounded rationality, rather than the
risks of opportunism, that determine the advantages (or disadvantages) of firms
relative to the market. Capabilities are the result of learning through a dynamic
process, and can exist in the process of identifying, acquiring, integrating, and
187
developing unique and complementary resources (Eisenhardt & Martin, 2000). For
instance, firms may have capabilities embedded in technological expertise or
managerial skills. Firms can exploit these capabilities as unique and valuable
strategic resources to obtain quasi-rents.
Performance implications of RBV are drawn from the argument that
complementary resources strengthen a firm‘s capabilities to manage uncertainties
and interfirm ties. Baum, Calabrese, and Silverman (2000) study the Canadian
biotechnology startups‘ alliances, and show that forming interfirm relationships
brings in complementary resources to a startup and enhances its innovative
performance. Sarkar, Echambadi, and Harrison (2001) use a firm‘s proactiveness as
an indicator of its capabilities in identifying and responding to opportunities, and
show that proactiveness results in greater value for the firm. Mesquita, Anand, and
Brush (2008) recently use survey data to test the resource-based arguments. They
find that RBV explains average performance advantages in general partner
experience, and that certain partners confer additional performance benefits when the
newly developed capabilities are dyad-specific.
Empirical research on alliances has examined experiential learning as an
important source of capabilities. Anand and Khanna (2000a) argue that firms build
capabilities to manage alliance by learning from accumulated alliance experience.
Zollo and Winter (2002) examine capabilities as dynamically developed routines,
and argue that capabilities are developed through routinization, which accumulates
and codifies experience into operational routines. Sampson (2005) shows that prior
188
alliance experience increase a firm‘s capabilities to manage ambiguous situations
more readily, and thus enhances partners‘ collaborative benefits. At the firm-level,
Aggarwal and Hsu (2009) examine characteristics of a firm‘s alliance portfolio over
time. They suggest that capabilities specific to a governance form will be exploited
in subsequent alliances that reuse the same governance, and show that experience in
equity alliances increases the percentage of equity alliances in a firm‘s alliance
portfolio, and decreases the percentage of licensing agreements. This study provides
evidence for path-dependency in firm-level capability developments.
While these studies have established benefits from experience-based
capabilities, they commonly view experience as a cumulative measure for repeated
organizational activities (e.g. alliances). However, firms may draw inferences from
different aspects in experience. In fact, interviews with industry experts reveal that
firms can acquire a range of useful skills from a single alliance, such as contractual
provisions, dispute resolutions, and informal relationship maintenance. To further
explore how capabilities are built from experience, it is critical to examine specific
aspects of experience. This essay focuses on the governance experience a firm has
accumulated from prior alliances, and suggests that experience not only varies in
degree, but also in type. Therefore, different capabilities can emerge from the
different dimensions of experience.
189
5.3 Development of Hypotheses
5.3.1 Dimensions of Alliance Governance Experience
Two dimensions of alliance governance experience are of particular interest
in this essay: experience in specific alliance governance and experience across
different governance forms. In the previous chapter, I showed that experience in
specific governance forms generate path-dependent governance decisions, and that
the path-dependency is present in each of the four governance types. Here I examine
the market value a firm accrues when it announces a new alliance. When assessing
the value creation potential of a new alliance, one of the most visible indicators to
outsiders is the partners‘ experience in managing the particular governance structure,
or experience in-type. With prior governance experience, a firm is perceived to have
certain knowledge in implementing and governing the structure. Experience in other
governance types is likely to attract relatively less attention than experience in-type.
Thus, when examining alliance announcements, I focus on the level of governance
experience in-type, that is, the same as focal alliance governance, rather than on
experience in each governance form.
Experience across different governance forms, or governance diversity,
indicates the extent to which a firm has organized prior alliances using diverse or
focused structures. Following the discussion on governance forms in Chapter 4,
alliance governance forms are classified into four primary types: unilateral, bilateral
without collaboration, bilateral with collaboration, and joint venture. When firms
accumulate alliance experience, they are likely to encounter relationships with
190
different types and levels of hazards, such as risks in relationships, partners, or
environments (Luo, 2005). As established by the transaction cost theory, firms will
adopt more hierarchical governance structures to minimize transaction costs when
contractual hazards increase (Williamson, 1991). Therefore, when experience
accumulates, there is a greater possibility that a firm has encountered contractual
hazards that call for diverse alliance governance structures. Note that this is not to
say that less experienced firms will necessarily have low governance diversity. In
addition, firms may choose governance structures that are misaligned with
contractual hazards (e.g. Silverman & Nickerson, 2003; Sampson, 2004). Thus,
governance diversity is positively but partially correlated with the level of general
alliance experience.
Either dimension of alliance experience can be generally depicted as low or
high. Juxtaposing these dimensions yields a two-by-two matrix as presented in Table
5.1, where each quadrant represents a type of experience portfolio with respect to
alliance governance. Firms that fall into Quadrant I specialize in a focused
governance form, and are leaping into a new structure in the currently announced
alliance. The market may perceive this type of firms as lacking necessary experience
and capabilities to create value in the focal alliance. Firms in Quadrant II are
specialists in the focal governance form. They are capable to exploit their structure-
specific knowledge in a new alliance structured in the same way. With expertise in
the governance form, these firms can create exploitative values in the new alliance.
191
The first two quadrants are concerned with alliance experience that covers
focused governance experience. In Quadrant III, firms have general knowledge about
alternative governance forms but lack experience in the form of a focal alliance. The
market may consider this combination of governance experience as value-creating,
because there can be generalizeable alliance capabilities across governance structures,
in particular leading to more appropriate structural decisions. However, the degree to
which this type of experience creates market value is uncertain. This is discussed in
the following sections. Finally, Quadrant IV includes firms with both experience in-
type and governance diversity. Such firms are generalists in terms of governance
experience, and are likely to have strong capabilities in designing and managing
alliances. Their alliance formation activities will be assessed by the market as having
the greatest potential to succeed. In the next section, I develop hypotheses based on
the four quadrants highlighted in Figure 5.1.
Figure 5.1: Dimensions of Alliance Experience
Experience in-type: experience in a
focal alliance governance form
Low High
Governance diversity
in experience
Low I: -- II +
High III: - IV: +++
5.3.2 Effects of Governing Capabilities on Value Creation
While prior studies have mostly examined overall experience (e.g. Kale et al.,
2002; Reuer et al., 2002; Sampson, 2005), this essay focuses on experience in the
192
same organizational structure as the focal alliance, or experience in-type. This
concept of experience in-type is a more appropriate measure than overall experience
when studying the market reactions to alliance events. Whether a firm is sufficiently
capable to manage a new type of alliance depends on the firm‘s prior experience.
Whereas general alliance experience develops knowledge about collaborations, a
firm is likely to encounter challenges and risks when managing a governance
structure outside of its experience range. With experience in the relevant governance
structure, a firm is perceived to have certain level of knowledge or skills to cope with
the same governance structure again.
Experience in-type indicates the capability to manage a focal governance
form. With increasing experience in a particular form, a firm will gradually develop
capabilities to design and govern this structure and coordinate partner interactions
under this structure. Regarding the capability to design alliance contracts, experience
in licensing agreements creates the knowledge and capabilities to design
standardized boilerplate contracts, and experience in joint ventures enables a firm to
consider more thoroughly about contingencies to be included in a contract (Anand &
Khanna, 2000b). Further, alliance contracts are shown to contain different
contractual provisions as a result of asset specificity (Reuer & Arino, 2007). Because
asset specificity determines the choice of alliance governance, experience in specific
governance type will focus a firm‘s attention and efforts toward these provisions, and
creates capabilities to contract on and cope with corresponding activities.
193
In addition to contract design capability, experience in-type also provides the
opportunity to develop capabilities to manage interfirm relationships. Once partners
commit to a relationship, they will coordinate the relationship using both formal
contractual terms and informal routines. Experience in managing norms and routines
that are specific to certain organizational structure evolves into a firm‘s tacit know-
how and capabilities (Nelson & Winter, 1982).With respect to alliances, firms with
alliance experience will develop from routines the capabilities to coordinate partner
efforts under specific organizational arrangement (Zollo et al., 2002). The degree of
collaborative work and hierarchical controls between partners increases from
unilateral agreements to joint ventures, as discussed in Table 4.1 in Chapter 4. As
experience accumulates, there are different emphases on coordination in different
alliance governance forms. Firms with profound experience in unilateral agreements
will become capable in managing high-powered incentives and autonomous
adaptation, and firms with intense experience in equity joint ventures will grow skills
in internal resource deployment and coordinated adaptation.
When firms develop structure-specific capabilities of contract design and
relationship management, these capabilities may be valuable, or have positive
externalities, to other related structures. For example, experientially developed
capabilities from non-collaborative bilateral alliances can be partially applied to
bilateral relationships with collaborations, given that these forms both involve
bilateral efforts. However, there are limitations on how these capabilities are
applicable across structures, because of structure-specific characteristics. Alliances
194
with collaborative efforts are likely to encounter a greater probability of
appropriation hazards and dispute resolutions. When all partners devote resources
and efforts to an alliance, their collaborative outcomes are dependent on all partners‘
proprietary assets, thus imposing difficulties on appropriating value from the alliance.
Therefore, in-depth experience in non-collaborative alliances may create very limited
capabilities to control appropriability risk or to settle inter-partner conflicts. In sum,
experience in-type signals a firm‘s capabilities in governing a particular structure,
but relatively little in governing alternative structures.
Experience in-type can take on different measures. While Chapter 4 captures
experience as a sum of prior deals, the market may not evaluate a firm‘s alliance
experience in specific governance forms as a linear factor that continuously affects
capability building. In fact, market investors are more likely to have a general or
coarse idea about whether the firm has a high or low level of experience with respect
to the new alliance‘s governance form. In addition, alliance governance structures
are more complicated to identify beyond the distinction between licensing and joint
venture (Anand & Khanna, 2000a; Oxley & Silverman, 2008). Capturing detailed
alliance governance experience is challenging to both academics (Oxley, 1997) and
to managers. In practice, firms rarely report alliance governance forms in publicly
released announcements, except for the relatively well defined boundaries of joint
venture and licensing relationship. Therefore, this essay proposes that the most
visible experience factor to market investors is whether experience in-type matches a
firm‘s highest level of governance experience.
195
In sum, when the governance of a new alliance falls into a firm‘s highest
experience in alliance governance, the firm is perceived by investors as having
strong governing capabilities to implement and safeguard the alliance. In Figure 5.1,
the implication is that market value creations are greater for firms with greater
experience in-type: for those with focused governance experience, Quadrant II is
higher than Quadrant I; for those with diverse governance experience, Quadrant IV is
higher than Quadrant III.
Hypothesis 1: An alliance creates more market value when it is governed by
the structure in which a focal firm has the highest level of experience.
5.3.3 Effects of Selection Capabilities on Value Creation
Hypothesis 1 suggests that experience in-type creates market values because
governing capabilities enable firms to manage a focal governance form more
effectively. However, the effects of governing capabilities may be confined by the
probability that an alliance governance form is a misaligned choice with transaction
hazards (Silverman & Nickerson, 2003; Sampson, 2004). The downside of possible
misalignment can be mitigated by the simultaneous dimension of governance
diversity in alliance experience.
Firms are likely to develop experience in diverse governance structures when
accumulating alliance experience. Governance diversity suggests that a firm has
managed interfirm relationships with various structural characteristics (see Table
4.1), and that the firm has developed knowledge about different levels of partner
incentives, administrative control, and coordinated adaptation (Oxley, 1997). Diverse
196
experience provides the benefits of explorative learning with respect to varying
contractual arrangements and organizational characteristics. When a firm has diverse
governance experience, it can draw inferences from a wide range of knowledge
about how governance structures and alliance relationship may or may not work.
Such knowledge creates a firm‘s selection capability among alternative governance
forms. Selection capabilities lead firms to make better informed governance
decisions by providing the ability to foresee possible ex post hazards in a new
relationship. As a result, firms with selection capabilities can choose alliance
governance that is better aligned with contractual hazards. As discussed in Chapter 4,
the diversity of governance experience reduces a firm‘s dependency on its historic
governance decisions, and enhances the salience of contractual hazards.
The experience effect on value creation is thus enhanced by selection
capabilities embedded in diverse governance experience. At a low level of
governance diversity, experience in-type positively affects market valuation due to
the exploitation benefits because firms are employing a structure in which they have
in-depth expertise. However, exploitation does not necessarily imply an appropriate
governance choice. At a high level of governance diversity, firms possess strong
selection capability. Their governance decisions in new alliances are less influenced
by path-dependent decisions or by cognitive limitations with respect to ex post
hazards, and are more likely to be properly aligned with contractual risks. Thus,
governance diversity combined with governing capabilities generates the benefits of
197
both exploitation and of selecting the right governance form, and subsequently a
steeper slope of segment III-IV in Figure 5.2.
In sum, when investors take into account of firms‘ selection capabilities, they
will view a focal alliance as having greater likelihood of success and value creation
because of increased likelihood of structural alignment. The gap between Quadrants
III and I is at a smaller scale than that between IV and II, primarily due to the added
benefits of selection capabilities, in addition to the benefits of exploiting structure-
specific governing capabilities. Therefore, the positive relationship between
experience in-type and value creation becomes stronger for the alliances whose
partners have diverse governance experience.
Hypothesis 2: Diversity of governance experience positively affects the
relationships between experience in-type and market value creation.
The hypothesized effects are illustrated in Figure 5.2 below. Hypothesis 1
suggests a positive relationship between governance experience in-type and new
alliance market value creation. Thus the slopes of segments I-II and III-IV are
positive. Hypothesis 2 suggests that the slope of I-II is flatter than that of III-IV,
because diverse experience creates selection capabilities that result in better
informed governance decisions.
198
Figure 5.2: Effects of Alliance Experience on Market Value Creation
Note that Figures 5.1 and 5.2 suggest that the experience portfolio in
Quadrant III creates less value than the portfolio in Quadrant II. This relative effect
suggests that the market will assign more weight to past experience in-type than to
the possibility of a better informed (or aligned) governance form. Two primary
reasons underlie this argument. First, experience in-type indicates past events that
have already happened. Firms can draw inferences from hindsight to facilitate the
management of future alliances. However, the governance choice of a new alliance is
yet to prove its appropriateness through implementation. In the absence of a proven
track record (i.e. in-depth experience) in a particular structure, the market will be
uncertain about how well a focal firm can manage a relatively new structure. Thus,
investors will hesitate to assign positive returns to the allied firms.
Second, new alliance relationships face uncertainty from the partners, the
project, and the environment. Even if firms can resort to capabilities from experience
in alternative governance forms, there is the problem that other structure-specific
Governance
experience in-type
IV
II
III
Market
value
creation
I
199
capabilities may only be limitedly applicable to the focal alliance. The market is thus
likely to perceive a greater probability of alliance success in a specialist (i.e.
Quadrant II) than a generalist (i.e. Quadrant III) when the new alliance governance
form falls into the regime of the specialist‘s governance experience. Given these two
reasons, I conclude that more value will be assigned to specialists with low diversity
in governance experience than to generalists with high diversity.
5.3.4 Boundary Conditions of Experience Effects
Alliance announcements are significant public information. Senior managers
interviewed for this study concur that, when a firm announces a new alliance
relationship, the news is often made for the purpose of positive market returns,
because stock market generally perceives such news as significant events and
appreciate interfirm collaborations. In the market‘s perception, a key factor is what
governance structure is adopted to manage the new alliance. Intuitively, equity joint
ventures build new business entities, and involve much more resource commitments
than non-equity relationships. JV participants need to reallocate internal resources of
financial, physical, and human capital to the focal JV, and to conduct coordinative
efforts on a daily basis. With the demand for resource inputs and managerial efforts,
JVs are likely to attract greater market attention than non-equity alliances. Indeed,
prior alliance studies on market valuation have found that governance forms affect
stock market responses to various degrees (e.g. Anand & Khanna, 2000a; Arend,
2004). Given these findings, it is likely that the effects of experience on market
valuations are bounded by specific governance forms of a focal alliance.
200
The effects of alliance experience are stronger in collaborative relationships.
Relative to non-collaborative ones, collaborative alliances have two unique
characteristics: collaborative efforts and value appropriation. Collaborations demand
greater efforts in coordinating partner activities to achieve alliance goals, such as
pooling resources into a joint development process and enforcing the collective
efforts from partners. Thus, capabilities of coordination can be developed from prior
alliance experience (Reuer et al., 2002; Zollo et al., 2002). In particular, the
experience in managing the focal alliance governance is instrumental. With extensive
knowledge in bilateral collaborations, for example, a firm has the capabilities to
predict contingencies that may arise from collaboration. The firm also may have
developed routines, such as management processes of dispute reporting and
mediation (Sampson, 2005), to work out technical or managerial problems with
alliance partners. On the other hand, experience in non-collaborative experience
focuses on individual rather than collective efforts, and is less likely to generate
capabilities that can be applied to solving inter-partner problems.
In addition, because collective outputs are built upon pooled partners‘ inputs,
alliance outcomes are subject to a weak property regime, where the risk of
misappropriation arises (Oxley, 1997). This risk can be accounted for by a firm‘s
governing and selection capabilities. With the selection capability from diverse
alliance experience, firms can choose better aligned alliance structures to control the
appropriability hazards. With the governing capability, firms have more structure-
specific insights about what could go wrong and how to adapt to contingencies. Such
201
capabilities are important because contingencies can be consequential to the
performance of a relationship. In less collaborative relationships, alliance outputs are
protected by relatively stronger property regime, because resources that contribute to
alliance performance are maintained within each partner‘s boundary. In sum, alliance
experience factors become more important determinants to create values when focal
alliances are governed by more collaborative structures.
Equity joint ventures involve greater collaborative requirements than non-
equity forms. However, the experience effects on value creations are not necessarily
stronger in joint ventures. A joint venture relationship, as a newly established
business entity, has complex collaboration details because the partners hold longer-
lasting goals, highly interactive norms, and much more significant resource
commitments (Hennart, 1988). Public announcements, while being a significant
event, only represent an initial step of complicated JV relationships. Initial JV
conditions often quickly give way to evolved conditions as JV partners develop an
understanding of each other (Inkpen and Currall, 2004). Collaborative processes are
then constantly adjusted during the JV implementation, which is a learning process
between partners.
Given the dynamics of JV structures, selection and governing capabilities
from prior experience help the implementation and performance of JVs to a limited
extent. The selection capability is most relevant to alliance structure design ex ante,
and is less helpful once a relationship has taken place. While the governing
capability may create collaborative routines that may be generalizeable across
202
governance forms, its applicability is decreased by the idiosyncrasies among JV
partners. JVs represent a significant but relatively infrequent form of alliance
relationships. For a focal firm, each of its JV partners is idiosyncratic and each
relationship requires unique design and relationship management. As a result,
selection and governing capabilities play a smaller role in affecting JV‘s value
creation.
To summarize, experience factors are more instrumental to collaborative
alliances than to non-collaborative forms, because collaborative outcomes are
facilitated by selection and governing capabilities derived from prior alliance
experience. However, alliance experience that pertains to general relationships (i.e.
experience in-type and governance diversity in this essay) will become less
beneficial when a focal alliance is highly dependent on dyadic relationships in the
post-formation stage, which requires significant inter-partner learning and
cooperation. From the market perspective, investors will perceive alliance experience
effects as stronger in collaborative alliances. When evaluating the value-creating
potential of a JV, investors are likely to pay more attentions to post-formation
dynamics than to a current announcement.
Hypothesis 3: The positive effect of alliance experience on value creation will
be stronger when a focal alliance is organized by a governance form that
involves an intermediate level of collaboration.
At the firm level, prior performance reflects a firm‘s overall capabilities in
managing resources and creating values. Because the success of a new alliance
203
depends on participants‘ resources and capabilities (Ahuja, 2000), it is likely that the
values to be created in the alliance will contingent on partner‘s previous performance.
A low level of performance signals that the firm is challenged in its capabilities to
capitalize on resources. The insufficiency of capabilities is further challenged when
the firm needs to coordinate interfirm alliances. Alliance partners often need to spare
internal resources to coordinate joint efforts. The above discussion suggests that in-
depth and/or diverse governance experience implies a firm‘s governing and selection
capabilities with respect to alliance governance. Thus, firms with these experience
characteristics are capable to recognize the importance of relevant alliance activities.
However, low performance indicates a lack of competence to implement necessary
activities. Lacking capabilities, a firm is perceived to have performance challenges
and thus a lower likelihood to design and/or to manage a new relationship, even
when it has pertinent experience-based capabilities.
As the level of performance increases, the focal firm is perceived to be
developing more firm resource and capabilities to coordinate productions and/or
relationships. The growing resources allow the firm to apply its capabilities
developed from experience to facilitate new alliances. Thus, only when these internal
resources exceed a certain level do firms have the freedom to apply experience-based
capabilities to new alliances. On the contrary, the market views interfirm alliances
for low-performing firms as overstretching the firms‘ resources. Therefore, with an
increasing level of performance, firms can create greater values by announcing new
204
relationships. Experience characteristics become more salient factors in affecting
market valuation as performance improves.
However, while performance strengthens the relationship between alliance
experience and market valuations to a new alliance, this relationship is not linear.
The market is likely to give top-performing firms the benefits of doubt upon the
events of alliance formation. In a semi-efficient market, all historic and public
information are taken into account when responding to public events, such as
alliance formation (Kale et al., 2002). Firms with a very high level of performance
are generally considered as capable of handling intra- and inter-firm activities. Their
prior alliance experience characteristics are relatively neglected at the presence of
superior performance. To the market, high performance in the past renders alliance
experience factors less salient in determining value creation. Thus, a firm‘s high
performance beyond a certain level makes its alliance experience relatively irrelevant
to the market assessments of its new alliances. In sum, the effects of alliance
experience are stronger for firms with an intermediate level of performance than
those with either low or high levels of performance.
Hypothesis 4: The positive effect of alliance experience on value creation will
be stronger when a focal firm has an intermediate level of performance in the
previous year.
While new alliance announcements are shown to create market values in
general (e.g. Anand & Khanna, 2000a; Oxley et al., 2009), this essay highlights the
importance of prior alliance experience in affecting this valuation. The effects
205
predicted in Hypotheses 1 through 4 are summarized in the model depicted in Figure
5.3. First, governance experience that is consistent with a new alliance‘s governance
form generates a firm‘s governing capability, and accrues positive market returns to
the focal firm. This relationship is strengthened by diverse governance experience,
which formulates a firm‘s selection capability toward more appropriate governance
decisions. Further, the effects of experience are bounded by alliance- and firm-level
factors. When a focal alliance is governed by an intermediate level of collaboration
and hierarchical control, the experience factors will demonstrate stronger effects than
in the cases of non-collaborative or joint venture relationships. When a focal firm has
an intermediate level of prior performance, its alliance experience will more strongly
affect market valuations than firms with low or high levels of performance. The
discussions of Hypotheses 3 and 4 also suggest that these inverted-U shape effects
are driven by different reasons for below-median than above-median cases.
To test stock market responses, I use an event-study methodology combined
with linear regressions. The following section describes the sample, variables, and
methods. Empirical results and discussions then follow.
206
Figure 5.3: A Model of Experience Effects on Market Value Creation
5.4 Data and Methods
5.4.1 Sample
This essay draws from the same sample as in Chapter 4, focusing on the US
software industry technically identified by a firm‘s primary four-digit SIC in 7371-
7374. To briefly recap, the computer software industry has been a leading force in
technological advancements during the past three decades. This industry has
witnessed constantly technological updates and innovations, arguably one of the
fastest changing industries. In such dynamic environment, it is crucial for software
companies to speed up capability development by leveraging and integrating external
resources through collaborative relationships. Indeed the software industry has
witnessed a proliferation of interfirm collaborations since the 1990s. A recent study
on strategic alliance activities conducted by Schilling (2008) shows that the software
Stronger effects when:
Intermediate
collaboration and
hierarchical control
(H3)
Intermediate level
of prior performance
(H4)
Governance
experience in-type:
Governing capability
Governance diversity:
Selection capability
Market
value
creation
H1: +
H2: +
207
industry is among the most active industrial sectors in which alliances are frequently
formed
10
.
This study also focuses on the period of 2001 to 2005, because this is a
period during which irrationality in alliance formation has diminished significantly
after the technology market crash in 1999. After 2001, interfirm relationships are
scrutinized under much more stringent criteria before formation or public
announcement, due to constrained financial conditions of public companies and
uncertain market trends (see Chapter 3 for discussions on alliance activities during
this period). While software companies have significant cautiousness toward new
alliances, the industry as a whole still remains more active in alliance formation than
most other industries (Lavie, 2007). Frequent alliance activities by software
companies offer variance, reliability, and meaningfulness for empirical tests of the
above hypotheses.
The sample contains 147 US publicly listed software companies identified by
their primary 4-digit SICs (7371-7374). This sample is representative because US
software companies have been dominating software businesses worldwide. As
shown in Chapter 3, the US companies have long been taking up over 50% of the
global market values. A caveat in the sample is that publicly listed companies are
often larger and more resource abundant than private companies. However public
companies provide two key advantages to test the hypotheses. First, large companies
10
Other industries listed in this study primarily consist of biotechnology, pharmaceutical,
telecommunications, airlines, and miscellaneous manufacturing.
208
are more likely than smaller ones to have accumulated a meaningful amount of past
alliances from which useful inferences can be applied to future governance decisions.
Second, interviews with industry experts suggest that alliance agreements may be
implicit or informal among private companies. Some companies are hesitant to
classify relationships as alliances for publicity or legal concerns. Therefore, it can be
difficult or even infeasible to identify a private company‘s alliance portfolio. Public
companies thus are relatively more visible in their alliance activities, and represent
an ideal group for this study.
The list of companies is presented in Appendix 2. Firms labeled with
asterisks are those that are excluded from this sample due to missing data. Their
alliance formation news was collected from two key sources. First, Securities Data
Corporation‘s (SDC) Joint Venture/Alliances database provides basic alliance
information. SDC has been arguably the most complete archival database for
strategic alliances that are formed by public companies (Sampson, 2005; Schilling,
2008). Whereas SDC has incompleteness in its methodology of data collection
(Lavie, 2007), it remains the most comprehensive data source for alliance research
(Schilling, 2008). To overcome SDC‘s insufficient data collection, I follow Lavie‘s
(2007) method by obtaining alliance announcements from Lexisi-Nexis Academics
alliance-relevant keyword search
11
. Each data source contributes about 50% of all
11
Key words include a wide range of terms, such as strategic alliance, strategic partnership,
collaboration, joint agreement, etc. In some announcements partners simply describe alliance
activities without a labeling term of alliance. I started by using highly inclusive terms, which
result in about 3200 announcements. Reading through each announcement reduces the total
209
observations to the final sample. Both sources rely on public information of alliances,
including news wires, SEC filings, analyst conference, and announcements on
company websites.
SDC describes alliance characteristics using multiple variables, such as main
value-chain activities and participants‘ background. Unfortunately, there are no
available documentations about SDC‘s coding schemes that can be directly applied
to code alliances from other sources. To ensure that Lexis-Nexis announcements are
coded in a consistent way with SDC, I first identified a coding pattern by going
through SDC‘s alliance records. Then I coded Lexis-Nexis records following the
identified pattern from SDC. The overlapped alliance reports are particularly
instrumental in creating this pattern. As SDC reports its news sources, I searched on
Lexis-Nexis for the each piece of announcement from the same data source. Then I
compared overlapped announcements to verify the coding scheme with SDC
information. This coding process was simultaneously conducted by three research
assistants; and coding results were frequently compared and compiled across all
coding efforts. In case of contradicting information, we compared information with
searches on Google Finance and Hoover‘s Business Profiles.
Market responses are captured by stock market abnormal returns based on
daily price data. The following section describes the data and method to calculate
abnormal returns. The regression sample in Chapter 4 has a list of 150 companies
number of collaborative alliances to 1700. Further financial data matching reduces the
current sample size to 1445.
210
and their 1511 alliances. However, not all firms have sufficient daily stock
information for calculating market returns, for reasons of very recent IPO or too
scattered price information. Therefore, I dropped firms who lack stock information
along with all their alliances. The sample for regressions in this essay now contains
145 focal companies yielded a sample of 1445 alliance announcements, with a
comparable average to that in Chapter 4 of about 10 alliances per firm.
5.4.2 Measures
5.4.2.1 Dependent Variables: Abnormal Stock Market Returns
Market valuations are calculated using data from the Center for Research in
Security Prices (CRSP). Daily stock prices surrounding alliance announcement dates
are collected. Stock market abnormal returns around event dates are key measures in
studies that examine value creation effects upon the events of alliance formation
(Das et al, 1998; Anand & Khanna, 2000a; Oxley et al., 2009). Following prior
studies, I use a standard event-study approach in the empirical analysis. This
approach involves the following steps. First, estimate a market model of each firm‘s
stock returns during an estimation window prior to the event date (t=0). Following
prior research (e.g. Anand & Khanna, 2000a; Oxley et al., 2009), I use an estimation
window of 240 days (that is, -250<=t<=-21). The following equation is used to
estimate a return function for each firm i :
it mt i i it
r r
211
where
it
r denotes the daily return for firm i on day t ,
mt
r represents the market
return on the same day, commonly captured by the daily returns of the Standard &
Poor (S&P) 500 Composite Index
12
. Coefficients
i
and
i
are firm-specific
parameters; and
it
indicates the error term that follows an independent and identical
normal distribution (i.i.d.).
Next, I use the estimated coefficients for each company i from the above
model,
i
ˆ and
i
ˆ
, to obtain predicted daily returns for firm i over a target period, or
event window, that is, returns during the days immediately surrounding the news
date of an alliance announcement:
mt i i it
r r
ˆ
ˆ ˆ
Here
it
r ˆ denotes the predicted daily return for firm i on day t . Abnormal returns (ARs)
for firm i are thus calculated as
it it
r r ˆ for each announcement on t . Cumulative
abnormal returns (CARs) for each firm are computed by summing up ARs over
event windows. In this study, I use a two-day window [-1, 0] and a three-day
window [-1, +1] as in prior alliance studies (Anand & Khanna, 2000a; Arend, 2004;
Oxley et al. 2009). These studies also suggested wider event windows of [-3, +3]
(that is, CAR 7-day) and [-10, +10] (that is, CAR 20-day). Thus I tested these
alternative event windows following this norm. However, wider event windows yield
12
Alternative market returns, such as weighted or un-weighted, were also tested for
robustness checks. As these alternative measures yield almost identical results, I stick to the
S&P index through all tests in this essay.
212
no significant effects of the predicted model. While not unexpected, this finding is
discussed in the following sections.
Finally, this procedure yields three dependent variables that are reported in
the empirical tests: CAR2, cumulative abnormal returns in a two-day window [-1, 0];
and CAR3, cumulative abnormal returns in a three-day window [-1, +1]. These
variables link to prior alliance research on market valuations as responses to alliance
formation. As linear continuous variables, linear regression models with fixed effects
are used for testing the experience effects on abnormal market returns.
5.4.2.2. Independent Variables: Alliance Experience
Hypotheses 1 and 2 propose the role of experience in-type and governance
diversity of alliance experience. Chapter 4 measures a firm‘s governance-specific
experience by a discounted sum of alliances that fall into each of the four governance
types. Historic alliances are traced back to 1990 and annually discounted by 0.2 to
present years. Experience in-type is thus measured as a firm‘s past alliances that are
governed by the same structure as that of a current alliance. If firm A has managed X
unilateral alliances, Y collaborative alliance, and Z joint ventures, its experience in-
type is equal to X if its new alliance is governed by a unilateral contract, and so on.
However, the market may not differentiate firm experience to such a detailed
degree as to the specific amount of past alliances. Instead, the level of high or low in
alliance experience is more visible to investors. Therefore, I measure experience in-
type using a dummy variable: 1 if a newly formed alliance is governed by the
governance form in which a focal firm has the highest level of experience and 0
213
otherwise. In the above example, if X is greater than Y and Z and if the new alliance
is a unilateral agreement, then experience in-type takes the value of 1. If X is smaller
than Y or Z, then the variable is equal to 0. I also use an alternative measure to
capture experience in-type. This measure takes three values: 2 if the focal alliance is
managed by the form in which a firm has most experience; 0 if this alliance is
managed by the form in which the firm has least experience; and 1 otherwise.
Governance diversity is measured in the same way as was in Chapter 4. It
describes the degree to which prior alliances have been managed under different
organizational structures. A Herfindal index, defined as
4
1
2
) / ( 1
i
i i
g g H
, is used
to capture diversity. The calculation is based on the proportion of alliances in each
governance mode out of a firm‘s entire alliance portfolio. Given the four identified
governance types, the diversity has a maximum of 0.75 (i.e. 1-4*(1/4)
2
). Experience
variables are all annualized instead of counting up to the date immediately preceding
a new announcement. The assumption is that alliances formed immediately before a
new announcement may pose little impact on the current one, because alliance
experience takes time to be assimilated into firm knowledge and resources.
Meanwhile, annualized variables are legitimate based on the assumption that all
alliances formed in the previous year are likely to exert similar impacts on new
alliances in the current year.
Hypotheses 3 and 4 suggest the governance of a focal alliance and the focal
firm performance both affect the extent to which experience can influence market
214
valuations. To test these moderating effects, I divide the sample in two different
ways. First, the sample is split on the basis of focal alliance governance: non-
collaborative (i.e. unilateral and bilateral contracts without collaboration),
collaborative, and joint venture. One caveat is that there is comparatively a small
number of joint ventures (N=33). This sample size is too small to yield consistent or
convergent regression estimates. Therefore, I test the effect of joint ventures by
grouping JVs with collaborative alliances. According to Hypothesis 3, the minimal
effect of joint venture will mitigate the positive effect of collaborative bilateral
contracts. Thus it is expected that the grouped sample will yield no effects of
experience variables. Next, all the alliances are mapped to their corresponding focal
firm‘s return on assets (ROA) in the year immediately preceding the announcement
date. Annual ROAs are obtained from CRSP annual data archive. The sample then is
split into three groups: observations with ROAs at the bottom 30%, those at the
bottom 30%, and the middle 40%.
5.4.2.3 Control Variables
The hazards embedded in an alliance affect a firm‘s decision of governance
forms, which can subsequently affect the performance of value creation of a focal
alliance. Because partners report alliance activities and partner information in their
public announcements, market investors will also take into account of the hazards in
their assessments about the value creation potentials of a new alliance. While the
market in general considers alliance formation as a positive event (e.g. Oxley et al.,
2009), the performance of an alliance is contingent on contractual hazards that may
215
hurt values to be created in the alliance. Indeed, market perceptions are shown to
reflect the success of alliance operations in post-formation stages (Kale et al., 2002).
Therefore, the three key contractual hazards proposed in Chapter 4 are controlled for:
appropriability risks, measurement difficulties, and alliance task complexity (see
Chapter 4 for detailed descriptions about operationalization).
Firm resources enter into market valuations. First, prior research shows a
firm‘s overall (discounted) alliance experience affect market responses (e.g. Das et
al., 1998; Anand & Khanna, 2000a; Oxley et al., 2009). Thus I control for the
discounted sum of a firm‘s all prior alliances as a general indicator of alliance
experience. Next, the discussion of Hypothesis 4 suggests that resource-abundant
firms are likely to have greater freedom in employing experientially developed
capabilities in new alliances. Therefore, I follow the measures in Chapter 4 to control
for firm resources of size (the log number of employees divided by 100), slack (cash
in $mil), and product market diversity. These data are collected from Compustat‘s
annual database and CRSP‘s four-digit SIC-based segment data. Note that the
Herfindal index of product diversity is a conservative measure of its true value,
because companies are required to report segments that contribute more than 10% of
their total revenues. Segments that contribute less than 10% are omitted and result in
a smaller diversity level than the true level. Finally, year dummies are also included.
Robust standard errors are clustered on each firm.
Table 5.1 presents summary statistics for the above described variables.
Correlations that are significant at 10% level are labeled by asterisks. The sample
216
contains 145 companies and their 1445 alliance announcements. The dependent
variables of CAR2 and CAR3 are at the level of less than 1%. Thus it is expected
that estimated coefficients are at a small scale. The dummy variable ―Experience In-
Type: 0/1‖ indicates whether the focal alliance is governed by a form in which a firm
has most experience. This is the key explanatory variable that measures governing
capabilities. Its mean of 0.69 implies that, on average, about 70% of all alliances are
managed by the governance form that a focal firm is most familiar with. This
observation also supports the path-dependency argument in governance decisions in
Chapter 4. The variable for total prior alliances is measured as a discounted sum, and
has a skewed distribution. I tested its log form in robustness checks, and found no
significant different estimate results.
The majority of pair-wise correlations are at a very low level. The previous
discussion suggests that governance diversity is likely to be developed when a firm
has accumulated a certain level of alliance experience. This is shown by the positive
correlation between diversity and total past alliances. Whereas significant at the 10%
level, their pair-wise correlations is only 0.065. High pair-wise correlations only
appear among firm size, product diversity, and total alliances. These correlations are
expected, because large firms have more resources to diversify into different product
segments, and to be pursued as attractive alliance partners. Therefore, large
companies tend to develop a diversified product portfolio and to accumulate a large
pool of alliances. I drop the three variables separately and collectively in alternative
regressions to detect whether the high correlations make the regression model
217
unstable. The results remain robust. By definition, the pair of CAR variables and the
pair of experience in-type variables are highly correlated. They are included in
alternative regressions.
218
Table 5.1: Summary Statistics of the Regression Sample
Variable Obs Mean Std. Dev. Min Max
1. CAR2 [-1, 0] 1445 0.007 0.073 -0.573 0.531
2. CAR3 [-1, +1] 1445 0.009 0.087 -0.457 0.694
3. Firm Size (employees) 1445 0.508 1.975 -4.423 4.111
4. Return on Assets 1445 -0.027 0.312 -3.095 0.425
5. Product Diversity 1445 0.242 0.275 0 0.766
6. Appropriability Risk 1445 0.287 0.494 0 2
7. Measurement Difficulty 1445 0.594 0.670 0 2
8. Task Complexity 1445 0.558 0.545 0 2
9. Total Prior Alliances 1445 51.772 110.529 1 515.4
10. Governance Diversity 1445 0.439 0.170 0 0.745
11. Experience In-Type, 0/1 1445 0.690 0.463 0 1
12. Experience In-Type, 0/1/2 1445 1.674 0.502 0 2
Pairwise Correlations 1 2 3 4 5 6 7 8 9 10 11
1. CAR2 [-1, 0] 1
2. CAR3 [-1, +1] 0.837* 1
3. Firm Size (employees) -0.079* -0.087* 1
4. Return on Assets -0.083* -0.080* 0.404* 1
5. Product Diversity -0.053* -0.059* 0.561* 0.327* 1
6. Appropriability Risk -0.005 -0.015 -0.053* -0.035 -0.039 1
7. Measurement Difficulty 0.008 -0.004 0.124* 0.039 0.020 -0.018 1
8. Task Complexity 0.010 -0.023 0.074* -0.016 0.005 0.024 0.213* 1
9. Total Prior Alliances -0.019 -0.026 0.594* 0.242* 0.334* -0.040 0.200* 0.087* 1
10. Governance Diversity -0.010 -0.007 0.055* 0.065* 0.076* -0.006 -0.094* -0.067* 0.065* 1
11. Experience In-Type, 0/1 0.050* 0.059* 0.070* 0.017 0.013 0.102* 0.141* 0.116* 0.093* -0.339* 1
12. Experience In-Type, 0/1/2 0.046* 0.054* 0.022 0.014 -0.003 0.096* 0.130* 0.109* 0.051* -0.320* 0.970*
219
Table 5.2 presents daily abnormal returns for the full sample from 10 days
prior to and 9 days after an announcement date (Day 0). The returns are calculated
using S&P 500 Composite Index as market returns and firms‘ holding period returns
with equal weights. The significance of daily abnormal returns (AR) only appears on
announcement days and one day immediately after.
Table 5.2: Even Study Results of Abnormal Returns of Alliance Announcements
Notes: Average daily excess returns and cumulative excess returns for a sample of 1445
alliance announcements. Focal firms are traded on NYSE, AMEX, or NASDAQ with
available CRSP returns data during 2001-05. Excess returns are the residuals from a market
model used to predict firm returns. The announcement day is defined as Day 0.
Firm Return=Holding Period Return; Mkt Ret = S&P Composite Index
Event Day Daily Excess Ret Robust Std Err Cum Excess Ret Robust Std Err
-10 -0.00061 0.00136 -0.00061 0.00136
-9 0.00345** 0.00135 0.00285 0.00193
-8 0.00018 0.00130 0.00303 0.00221
-7 0.00146 0.00123 0.00449* 0.00250
-6 0.00066 0.00110 0.00515* 0.00265
-5 0.00105 0.00122 0.00620** 0.00291
-4 0.00171 0.00123 0.00791** 0.00315
-3 0.00056 0.00115 0.00847** 0.00327
-2 0.00160 0.00114 0.01007*** 0.00342
-1 0.00180 0.00120 0.01187*** 0.00352
0 0.00584*** 0.00133 0.01771*** 0.00382
1 0.00242** 0.00114 0.02013*** 0.00395
2 -0.00059 0.00117 0.01954*** 0.00408
3 0.00037 0.00114 0.01990*** 0.00428
4 -0.00104 0.00119 0.01887*** 0.00445
5 -0.00226** 0.00105 0.01661*** 0.00458
6 -0.00078 0.00112 0.01583*** 0.00473
7 -0.00078 0.00115 0.01505*** 0.00491
8 -0.00106 0.00117 0.01400*** 0.00502
9 0.00202* 0.00113 0.01601*** 0.00517
220
Table 5.3 summarizes market value creations (AR, CAR2, CAR3, and CAR7)
by the four alliance governance modes. Consistent with the literature, equity joint
ventures on average create much higher abnormal returns than the other three modes
(Anand & Khanna, 2000a). Whereas the other three categories receive different level
of returns, the differences are at a smaller scale than the gap with JVs.
Table 5.3: Abnormal Returns by Focal Alliance Governance Structures
Unilateral
Bilateral, No
Collaboration
Bilateral,
Collaboration
Equity Joint
Venture
AR on Day=0 0.0069 0.0041 0.0055 0.0191
S. D. 0.0648 0.0468 0.0512 0.0613
CAR2 [-1, 0] 0.0026 0.0057 0.0093 0.0187
S. D. 0.0777 0.0714 0.0690 0.0662
CAR3 [-1, +1] 0.0107 0.0050 0.0110 0.0182
S. D. 0.0973 0.0883 0.0796 0.0544
CAR7 [-3, +3] 0.0059 0.0081 0.0142 0.0306
S. D. 0.1377 0.1233 0.1107 0.0832
5.5 Results
Linear regression results for Hypotheses 1 and 2 are presented in Table 5.4.
Fixed-effects of year are included in all models, but omitted here for presentation
purposes. To avoid losing the degree of freedom, standard errors are clustered on
each firm, instead of using firm-fixed effects. Columns 1 through 4 present the
results from regressions that use CAR2 (i.e. two-day event window) as the dependent
variable. Column 1 tests experience in-type using a dummy variable, which is equal
to 1 if the focal alliance is governed by a firm‘s most skilled firm. As predicted in
221
Hypothesis 1, experience in-type positively enhances abnormal returns. Given that
firm size is highly correlated with product diversity and total alliance experience, I
estimate the model‘s post-estimation variance-inflation factor (VIF) for possible
multicollinearity among independent variables. The mean VIF of this model is 1.37;
and as expected, the highest VIF appears in firm size, at the level of 2.16. Both VIFs
are well below the commonly accepted threshold of 10. Thus the model is not subject
to multicollinearity problem.
Column 2 adds the interaction between experience in-type and governance
diversity, testing whether experience effects on value creation will be strengthened
by selection capabilities. While the interaction shows positive effect, this effect is not
statistically significant. Thus Column 2 provides no support for Hypothesis 2. The
mean VIF of this model is 4.25, suggesting minimal threats of multicollinearity.
Columns 3 and 4 replicate the regressions in Columns 1 and 2, except for a different
measure of experience in-type: this variable contains three categorical values, and
captures whether an alliance is governed by a structure in which the firm has most or
least experience. Both Hypotheses 1 and 2 are supported when experience is
measured in this way. That is, investors appreciate a firm‘s in-depth experience in a
focal alliance‘s governance form; and diverse governance experience provides
selection capabilities to make more appropriate governance decisions and
strengthens the effect of in-depth experience. The VIFs for Columns 3 and 4 are 1.36
and 6.87, reflecting a low risk of multicollinearity.
222
Among the control variables, firm size measured as log of every 100
employees is shown to negatively affect abnormal returns. This finding is consistent
with prior alliance research that suggests large firms may accrue less market value
than smaller partners (e.g. Das et al., 1998). In addition, total alliance experience
measured as a count variable positively affects value creation. This finding provides
support for empirical research on alliance experience (e.g. Kale et al., 2002; Reuer et
al., 2002; Sampson, 2005), which has established that firms develop coordinative
capabilities to safeguard subsequent alliances toward successes. Finally, contractual
hazard variables show no significant effects across all models, indicating that
alliance-level risks, whereas significant to partners‘ governance decisions, play a
minimal role on affecting investors‘ perceptions about an alliance.
Columns 5 to 8 use CAR3 as the dependent variables and replicate the
regressions in Columns 1 to 4. The estimated coefficients are largely unchanged,
except that experience in-type and firm size now find slightly stronger effects on
abnormal returns. Note that total alliance experience loses statistical significance
when considering three-day abnormal returns. The reason is likely to be that alliance
experience in general only affects market perception when the news is released.
General experience effects diminish rapidly once the new is past. Finally, the VIFs in
Columns 5 through 8 remain the same as in the first four columns.
Small R-squares in Table 5.4 are expected in event-studies on stock market
abnormal returns. Because market returns are at a much smaller scale than the other
variables, such as firm size and alliance experience, the interpretive percentages are
223
often at the range of 1-3% (Oxley et al., 2009). In addition, adding an interaction
term between experience in-type and governance diversity improves the model fit by
about 5-6%. This change suggests that the inclusion of a moderating effect lends
additional explanatory power.
224
Table 5.4: Effects of Alliance Experience on Abnormal Stock Returns
DV = CAR2 [-1, 0] DV = CAR3 [-1, +1]
VARIABLES Exp-in-type: 0/1 Exp-in-type: 0/1/2 Exp-in-type: 0/1 Exp-in-type: 0/1/2
1 2 3 4 5 6 7 8
Log (Firm Size) -0.0032** -0.0033*** -0.0031** -0.0032** -0.0041** -0.0041** -0.0039** -0.004**
(0.0012) (0.0012) (0.0012) (0.0012) (0.0017) (0.0016) (0.0017) (0.0016)
Return on Assets -0.0124 -0.0123 -0.0127 -0.0126 -0.0124 -0.0122 -0.0127 -0.0126
(0.0080) (0.0080) (0.0080) (0.0080) (0.0080) (0.0080) (0.0080) (0.0080)
Product Diversity 0.0007 0.0004 0.0004 0.0002 -0.0006 -0.0009 -0.001 -0.0013
(0.0075) (0.0074) (0.0075) (0.0074) (0.0098) (0.0097) (0.0098) (0.0098)
Appropriability Risk -0.0029 -0.0029 -0.0027 -0.0027 -0.0052 -0.0052 -0.0049 -0.0049
(0.0036) (0.0036) (0.0036) (0.0036) (0.0041) (0.0041) (0.0041) (0.0041)
Measurement Difficulty 0.0007 0.0007 0.0008 0.0008 0.0000 0.0000 0.0001 0.0002
(0.0027) (0.0027) (0.0027) (0.0027) (0.0033) (0.0033) (0.0033) (0.0033)
Task Complexity 0.00050 0.00076 0.00057 0.00085 -0.00454 -0.00425 -0.00442 -0.00407
(0.0031) (0.0031) (0.0031) (0.0031) (0.0039) (0.0039) (0.0039) (0.0039)
Total Prior Alliances 2.21e-05* 2.09e-05* 2.34e-05** 2.20e-05* 2.24E-05 2.11E-05 2.46E-05 2.27E-05
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Alliance Governance Diversity 0.0069 -0.0247 0.0052 -0.0654 0.0115 -0.0246 0.0087 -0.0812*
(0.0126) (0.0233) (0.0127) (0.0414) (0.0145) (0.0228) (0.0145) (0.0453)
Experience In-Type: Max (0/1) 0.0098*** -0.0091
0.0152*** -0.0064
(0.0035) (0.0110)
(0.0043) (0.0142)
Experience In-Type*Diversity
0.0373
0.0426
(0.0229)
(0.0271)
Experience In-Type: Min/Max (0/1/2)
0.0078** -0.0114
0.0120*** -0.0125
(0.0031) (0.0102)
(0.0038) (0.0133)
Experience In-Type*Diversity
0.0385*
0.0490*
(0.0214)
(0.0258)
225
Table 5.4: Continued
DV = CAR2 [-1, 0] DV = CAR3 [-1, +1]
VARIABLES Exp-in-type: 0/1 Exp-in-type: 0/1/2 Exp-in-type: 0/1 Exp-in-type: 0/1/2
1 2 3 4 5 6 7 8
Constant 0.0003 0.0168 -0.0056 0.0304 0.0030 0.0219 -0.0058 0.04
(0.0084) (0.0137) (0.0093) (0.0219) (0.0097) (0.0142) (0.0112) (0.0254)
Year Dummies Y Y Y Y Y Y Y Y
Observations 1445 1445 1445 1445 1445 1445 1445 1445
R-squared 0.016 0.017 0.015 0.016 0.021 0.022 0.02 0.021
Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1
226
Hypothesis 3 suggests that the focal alliance‘s governance structure will
influence the extent to which a firm‘s alliance experience creates market value. This
hypothesis is tested by dividing the full sample according to alliance governance
forms. Table 5.5 presents regression results that use CAR2 as the dependent
variables. Columns 1 and 2 are from the sample of non-collaborative alliances, or
alliances that are governed by either unilateral contracts or bilateral, non-
collaboration agreements. The only significant coefficients are governance diversity
and the constant term. This finding implies that in the sample of non-collaborative
relationships, diverse governance experience may not contribute to value creation.
Focused experience, presumably in the focal alliance‘s governance form, is a more
positive signal to the market. In addition, there is little collinearity problem, given
that the post-estimation analysis shows VIFs of 1.27 and 4.05 for Columns 1 and 2,
respectively.
Columns 3 and 4 cover the sample of bilateral relationships that are
collaborative but are not joint ventures. The main effects of experience in-type and
governance diversity are both positive and significant. These coefficients suggest
that the predicted benefits of governing and selection capabilities are much stronger
in collaborative relationships than in non-collaborative ones. Columns 4 includes the
interaction term, and presents that governance diversity strengthens the effects of
experience in-type on abnormal returns, primarily from more appropriate choices of
governance, with economic and statistical significance.
227
Because the sample of JVs are too small to produce consistent estimate
(N=33), Columns 5 and 6 pool the sample of collaborative non-JV alliances with
joint ventures in regressions. Hypothesis 3 suggests a curvilinear effect of
governance forms; that is, alliances that are governed by structures of intermediate
collaboration and hierarchical control will show the strongest effects of experience.
According to this argument, the pooled sample is expected to yield either small or
insignificant coefficients than the second class (i.e. Columns 5 and 6). The reason is
that experience should play a smaller role in affecting market valuations to a new JV,
whose value creation is more dependent on post-formation dynamics. While the
main effect of diversity reduces slightly, the effect of experience in-type decreases
from 0.0213 to 0.0128, a significant 48% drop. This decrease from adding merely 33
observations suggests that the group of JVs considerably reduces the experience
effects. In addition, the interaction between these two experience measures becomes
insignificant in Column 6. These findings highlight that experience factors contribute
to a lower degree of value creation as far as JVs are considered.
Collectively, the results in Table 5.5 provide supports for Hypothesis 3.
Experience effects are shown to be stronger when a new alliance is managed by
collaborative, non-JV structures. The requirement of collaboration obviates the
importance of prior interfirm experience. On the other end, the long-term, high-
commitment characteristics of equity JVs render the investors to put less weights on
partner-level experience; instead, post-formation cooperating behaviors are more
closely monitored to determine value creation. In addition, the model fits best in the
228
sample of collaborative, non-JV alliances, with R-square above 4%. This observation
implies that the predicted experience effects provide the strongest explanatory power
for alliances that are managed by this structure.
Table 5.5: Split Sample Regressions by Alliance Governance Form;
DV=CAR2, standard errors clustered on each firm
Gov=Non-Collab Gov=Collaborative Gov=Collaborative
Alliance Alliance Equity JV
VARIABLES 1 2 3 4 5 6
Log (Firm Size) -0.003 -0.0027 -0.0045** -0.0043** -0.0039** -0.0039**
(0.0023) (0.0023) (0.0018) (0.0018) (0.0018) (0.0018)
Return on Assets 0.0007 0.0008 -0.0199 -0.0182 -0.022 -0.0216
(0.0101) (0.0100) (0.0143) (0.0143) (0.0137) (0.0137)
Product Diversity -0.001 -0.0018 0.0069 0.0049 0.0028 0.0022
(0.0112) (0.0113) (0.0093) (0.0091) (0.0092) (0.0091)
Appropriability Risk -0.0029 -0.0026 -0.0030 -0.0032 -0.0022 -0.0023
(0.0088) (0.0088) (0.0046) (0.0046) (0.0046) (0.0047)
Measurement Difficulty 0.0072 0.0075 -0.0034 -0.0032 -0.0022 -0.0022
(0.0070) (0.0071) (0.0029) (0.0028) (0.0028) (0.0027)
Task Complexity -0.0082 -0.0081 0.0009 0.0010 0.0033 0.0034
(0.0051) (0.0051) (0.0042) (0.0042) (0.0041) (0.0040)
Total Prior Alliances 2.74E-05 2.89E-05 2.76e-05* 2.49E-05 2.69e-05* 2.56E-05
(0.0000) (0.0000) (0.0000) (0.0000) (0.0000) (0.0000)
Alliance Governance Diversity -0.0448* -0.0246 0.0326** -0.116 0.0320** -0.0148
(0.0265) (0.0292) (0.0161) (0.0861) (0.0157) (0.0862)
Experience In-Type: Max (0/1) 0.0077 0.0321 0.0213*** -0.0643 0.0128* -0.0135
(0.0062) (0.0320) (0.0072) (0.0486) (0.0066) (0.0442)
Experience In-Type*Diversity
-0.0459
0.152*
0.0494
(0.0562)
(0.0856)
(0.0875)
Constant 0.0244* 0.0139 -0.0165 0.0691 -0.0099 0.0157
(0.0146) (0.0156) (0.0117) (0.0506) (0.0107) (0.0467)
Year Dummies Y Y Y Y Y Y
Observations 524 524 888 888 921 921
R-squared 0.032 0.033 0.04 0.042 0.038 0.038
Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1
229
Hypothesis 4 proposes a similarly inverted-U shaped moderating effect as in
Hypothesis 3 based on focal firms‘ performance. The argument is that the
relationship between experience and abnormal returns will appear to be stronger for
mid-level performers than for either top- or low-level ones. The results are presented
in Table 5.6. The full sample is divided into three groups by a focal firm‘s ROA in
the previous year. Columns 1 and 2 are estimated in the sample of bottom 30%,
Columns 3 and 4 are for the middle 30%, and the final two columns are for the top
30% performers.
The first two columns show no significant effect of experience variables.
Interestingly, governance diversity for low-performing firms does not send negative
signals to the market. While Hypothesis 4 proposes that governance diversity is an
indicator of selection capability, in the low-performing sample this indicator may
also suggest that a focal firm is simply experimenting with alternative structures,
rather than truly learning from experience and developing capabilities. With a lack of
negative significance, Columns 1 and 2 suggest that investors do not perceive
diversity as a bad thing. It is possible that the market is tolerant to experimenting
governance forms, or that investors do not necessarily differentiate diverse
governance experience for low-performing firms.
Columns 5 and 6 present largely similar results as in the first two columns.
For top-performing firms, the market is likely to lend some benefits of doubt to their
new alliances. Therefore, experience is not so salient a determinant for market
valuation once the market has taken into account of a focal firm‘s performance. Note
230
that the signs of governance diversity are opposite for top- versus low-performing
firms. Whereas lacking significance, this finding implies that the two subsamples
receive similar market valuations for different reasons.
Comparing the top and bottom subsamples with mid-performers, I find
supports for Hypothesis 4. Columns 3 and 4 present the results for mid-level
performers. For alliances that are managed by mid-level performers, alliance
experience in the focal governance type significantly enhances a focal firm‘s market
returns. This experience effect is strongly augmented by the diversity of governance
experience. That is, the selection capability enables a firm to better leverage
governing capabilities in attracting positive market responses.
An interesting effect is shown in appropriability risk. For mid-level
performers, the risk of misappropriation is detrimental to value creation. This finding
highlights that the market is more cautious to appropriability risk when a focal firm
has an intermediate level performance. The reason is likely to be that mid-performers
are at risk to join either the low- or high-performing group. The outcome of a focal
alliance may exert considerable impacts on the firm‘s performance trajectory.
Therefore, the market evaluates inter-partner appropriation as a key threat to alliance
outcomes.
In sum, Hypothesis 4 is supported by the split sample analysis according to
the level of firm performance. Experience factors are more salient determinants for
market value creation when a firm‘s performance is neither low nor high. To the
market, experience plays a more visible role upon new alliance formation when there
231
is uncertainty about the focal firm‘s performance. Again, R-squares for the mid-level
sample are at the highest 5% level, suggesting that the hypotheses developed above
fit best to middle level performers than to the two end groups.
Table 5.6: Split Sample Regressions by Focal Firm Performance; DV=CAR2,
standard errors clustered on each firm
VARIABLES ROA, bottom 30% ROA, middle 40% ROA, top 30%
1 2 3 4 5 6
Log (Firm Size) 0.0003 0.0002 -0.0073** -0.0073** -0.0006 -0.0006
(0.0049) (0.0049) (0.0033) (0.0032) (0.0019) (0.0019)
Return on Assets -0.0030 -0.0031 0.111 0.112* -0.003 -0.0034
(0.0098) (0.0099) (0.0680) (0.0675) (0.0283) (0.0284)
Product Diversity 0.0078 0.0083 -0.0032 -0.0045 0.0018 0.0018
(0.0248) (0.0247) (0.0085) (0.0082) (0.0123) (0.0122)
Appropriability Risk 0.0066 0.0065 -0.0101* -0.0103* -0.0053 -0.0056
(0.0101) (0.0102) (0.0053) (0.0052) (0.0035) (0.0034)
Measurement Difficulty -0.0078 -0.0078 0.0034 0.0037 0.0026 0.0026
(0.0068) (0.0068) (0.0038) (0.0038) (0.0032) (0.0032)
Task Complexity 0.0086 0.0090 -0.0035 -0.0034 -0.0029 -0.0031
(0.0086) (0.0083) (0.0043) (0.0043) (0.0037) (0.0037)
Total Prior Alliances 0.0001 0.0001 0.0002 0.0002 0.0000 0.0000
(0.0005) (0.0005) (0.0001) (0.0001) (0.0000) (0.0000)
Alliance Governance Diversity 0.0173 -0.00951 0.016 -0.038 -0.023 0.0164
(0.0312) (0.0530) (0.0148) (0.0299) (0.0196) (0.0534)
Experience In-Type: Max (0/1) 0.0151 -0.0001 0.0121*** -0.0217 0.00133 0.0242
(0.0092) (0.0220) (0.0043) (0.0153) (0.0045) (0.0291)
Experience In-Type*Diversity
0.0314
0.0647**
-0.0454
(0.0493)
(0.0281)
(0.0529)
Constant -0.0014 0.0117 -0.014 0.0164 0.0156 -0.0044
(0.0163) (0.0256) (0.0129) (0.0195) (0.0157) (0.0326)
Year Dummies Y Y Y Y Y Y
Observations 430 430 615 615 400 400
R-squared 0.012 0.012 0.048 0.053 0.021 0.023
Robust standard errors in parentheses; *** p<0.01, ** p<0.05, * p<0.1
Several robustness checks are conducted following the above analysis. To
ensure the above estimations are not sensitive to variable measures, I re-estimate
232
above models using alternative measures. In calculating the dependent variables, I
use equal- and value-weighted market returns instead of S&P 500 Composite Index
to obtain abnormal returns. I also use firm returns that include and exclude dividends
in conjunction with equal-weighted and value-weighted market returns include and
exclude dividends, respectively. The results are largely stable to those presented in
the tables above.
Abnormal returns across different event windows may demonstrate a degree
of mean aversion in abnormal returns. In fact, Table 5.2 shows a rapid decline in
abnormal returns after Day 1. Following the norms in the current literature (e.g.
Anand & Khanna, 2000a; Oxley et al., 2009), I calculated cumulative abnormal
returns for alternative lengths of event windows during the periods of 7-day [-3, +3]
and 20-day [-10, +9]. The effects of experience variables diminish to insignificant
levels. Such diminished effects reflect that the market rapidly absorbs the news of
alliance formation and factors this information into a firm‘s market value three days
after the announcement date. Further, firm random effects are tested instead of
standard error clustering on each firm. While this method decreases the degree of
freedom in regressions, coefficients remain mainly unchanged.
233
5.6 Discussion and Conclusions
5.6.1 Summary of Findings
This essay attempts to develop a more thorough understanding about the
relationship between firm-level alliance experience and market reactions. From the
resource-based view, firm experience is a key source for capability development
(Eisenhardt & Martin, 2000). Experiential learning develops organizational routines
and skills, whose values are to be exploited in future activities (Nelson & Winter,
1982). From the learning perspective, prior experience is a process of learning-by-
doing as firms accumulate knowledge to adapt more effectively to contingencies
(Argote, 1999). Combining these perspectives, recent research on alliance experience
has addressed the question of whether experience creates value to participating firms
(e.g. Anand & Khanna, 2000a; Oxley et al., 2009; Sampson, 2005).
The findings in this essay support the argument that alliance experience
creates values. I drew the distinction between explorative and explorative learning
processes by examining the two structural dimensions of firm-level experience. The
experience in a specific governance form indicates in-depth knowledge about how to
design, govern, and coordinate relationships under this particular structure. The
structure-specific experience fosters governing capabilities. Experience of managing
different governance structures represents a form of explorative learning, and
broadens the knowledge about how alternative structures may or may not work.
Diverse governance experience thus generates selection capabilities, which help in
choosing more appropriate governance forms.
234
Consistent with the current alliance research (e.g. Arend, 2004; Das et al.,
1998; Sampson, 2005), I found that alliance experience creates benefits to the
partnering firms. Abnormal stock returns are accrued to a firm when it has
established a track record of alliance relationships. In particular, when a new alliance
is governed by the structure in which the firm has most experience, the market reacts
positively because of a greater expected probability of alliance success. More
interesting findings emerge from the effect of governance diversity. This diversity
signals to the market that a firm has more knowledge about alternative structures
than firms with focused governance experience. As shown in Chapter 4, focused
governance experience may lead to path-dependent governance decisions, which
may be at odds with the discriminating alignment proposed by TCE. Explorative
learning allows a firm to draw from experience in different governance forms and to
make better informed governance decisions. Therefore, experience is perceived as
more valuable when the firm has the capability to choose an appropriate form for its
new alliances.
Further, investor assessments on alliance events vary under different
conditions. At the alliance level, experience is a more important determinant when
the focal alliance is a collaborative and non-JV relationship. Non-collaborative
relationships are relatively simple structures that involve less partner interactions and
collective efforts, and rely on prior alliance experience to a lower degree. JV
relationships on the other hand are highly complex and idiosyncratic; thus the value
creation hinges to a greater extent on post-formation dynamics beyond the
235
announced event. It is the intermediate level collaborative structure that is most
dependent on past experience to create values. Therefore, the boundary condition of
alliance governance forms has a curvilinear effect on experience factors.
A similar effect appears when examining how firm performance affects the
salience of experience. At the firm level, experience appears to be more importantly
affecting value creation when the focal firm has an intermediate level of performance
in the previous year. Low-performance signals to the market that a firm is likely to
have insufficient capabilities to manage new inter-firm relationships. The
performance information is more salient indicator, and leads investors to ignore the
firm‘s alliance experience. High-performing firms can enjoy the benefit of doubt,
since investors would generally consider them as capable in managing relationships.
Performance information outweighs experience factors when evaluating the potential
of a newly formed alliance. Mid-level performers, given the risk that their alliances
may drag them into either top- or bottom-performing groups, are scrutinized more
closely in terms of alliance experience characteristics.
5.6.2 Implications
The findings in this study provide several implications for alliance practices.
First, alliance experience is a multidimensional concept, thus the learning theory and
RBV can be enriched by examining different dimensions of experience. When
cumulative experience creates benefits to allied firms (e.g. Anand & Khanna, 2000a),
there can be different impacts from different aspects of experience. In particular, the
depth and breadth of experience with alliance governance create different types of
236
capabilities that can create values in new relationships. Firms may adopt different
alliance structures to build governing or selection capabilities.
Second, this essay highlights the need for examining governance forms as a
channel of learning. Not only do firms learn from functional activities such as R&D,
manufacturing, or marketing, they also develop valuable knowledge by designing
and managing organizational structures with different attributes. Designing alliance
governance is an important capability that can be derived from writing contract for
interfirm relationships (Argyres & Mayer, 2007). Learning about a certain form is
beneficial when the same form is used again, but may not be so helpful as to manage
other governance forms. Thus, alliance managers can develop more expertise in
managing interfirm relationships by adopting agreements that govern and monitor
partner activities with different means.
Further, while announcing new alliances generally enhances firm valuation
(Oxley et al., 2009), the degree of such benefits is contingent upon how the new
alliance is governed and how the focal firm has performed in the past. When forming
new alliances, managers resort to prior experience to design alliance structure. Their
experience may contribute to various degrees of market returns depending on the
governance structure of new alliances. They should also take their firm performance
into account when expecting market responses. Firms with an intermediate level of
performance are likely to draw more attention to their alliance experience. These
boundary conditions affect how a firm‘s portfolio of alliances can affect market
returns upon the event of alliance formation.
237
5.6.3 Contributions
Understanding experience in the aspects of depth and breadth extends the
learning concept beyond repetition of organizational activities, or learning-by-doing
in the learning literature (Argote et al, 1990; Vanneste & Puranam, 2008). Alliance
research has commonly treated experientially learning as a sum of total past alliances
(e.g. Reuer et al., 2002). However, this sum measure provides little detail about the
learning process through complicated alliance relationships. This essay unpacks the
wholesome measure of learning into orthogonal dimension of depth and breadth.
Different types of learning benefits are accrued from either of these dimensions.
Given a certain governance form, experience in the same form provides the benefits
of exploitative learning. When entering into new alliances, experience across
different governance forms provides explorative benefits and enables a firm to select
a proper governance mode.
Second, this essay focuses on the governance aspect of experience, and
suggests different types of capabilities from experience. Previous research has
considered ―alliance capabilities‖ as a general outcome by managing multiple
interfirm relationships (Anand & Khanna, 2000a; Kale et al., 2002; Kale & Singh,
2007). However, there is little research on what consists of such capabilities. This
essay argues that, because alliances are complex relationships, there should be
different types of capabilities generated from alliance experience. The exploitative
learning process creates the capability to govern a specific structure, and the
explorative learning process provides more insights to choose a proper governance
238
form. These dimensions collectively enable a firm to leverage prior experience to
create values from new alliances.
This essay also highlights the boundary conditions of experience at the firm
level and at the alliance level. The learning literature has generally supported a
positive relationship between experiential learning and performance (e.g. Anand &
Khanna, 2000a; Hoang & Rothaermel, 2005; Kale et al., 2000). The findings in this
essay suggest that the benefits of learning can differ by firm performance and by
alliance governance characteristics. Further, both RBV and learning theories have
yet to examine the structural aspect of experience, that is, whether and how
experience in specific governance can generate firm-level benefits. To my
knowledge, this study makes an early step connecting firm experience from
governance structures with market value creation.
5.6.4 Limitations and Future Research
Several caveats are worth noting in this essay. Focusing on a single industry,
the empirical findings may not be generalizeable to other industrial settings. The
tradeoff is that business activities and alliances formed within the same industrial
context are relatively comparable than those that are across various industries. The
sample of US software companies is representative of the software industry and
high-technology firms at large. US companies have been dominating in the global
software industry, accounting for over 50% of total sales for the past two decades.
Software has been a leading force driving technological development in diverse
industrial areas, such as internet services, computer manufacturing, communication
239
technology, etc. Thus the context here may provide helpful implications for other
related industries.
Second, the classification of alliance governance is subject to more stringent
scrutiny. The current four-class alliance governance is built upon the classification of
alliance governance forms in prior alliance governance research (Gulati & Singh,
1998; Oxley, 1997). But as pointed out by Oxley (1997), identifying alliance
governance forms is much more challenging than the make-or-buy decisions. There
are no specific guidelines or requirements about whether and how firms should
report alliance activities. While the complexity of alliance activities poses further
difficulties, allied firms do not necessarily specify the alliance structure (except for
JVs) using a consistent terminology. Survey (e.g. Kale et al., 2002) or archival
databases (e.g. Sampson, 2005) are often idiosyncratic with respect to the target
sample or industries, and have limited details to delineate one structure from another
except for JVs. This essay makes the effort to first identify public alliance
announcements following Lavie‘s (2007) method, and then combine the results with
SDC coding scheme and Oxley‘s (1997) alliance classification. The data sources and
methods are arguably the most commonly acknowledged in the alliance research
(Schilling, 2008). Future research is in need for more thorough and systematic
examinations of alliance governance forms.
In addition, the event-study method demonstrates value creation only to the
extent of the short-term event windows. Alliances as on-going relationships may last
over months, years, or even decades. There are a lot of contingencies that can emerge
240
in post-formation stages. Thus event-studies can only reflect the market responses to
the news rather than to long-term alliance performance (Oxley et al., 2009). In fact,
whereas most alliance announcements are positive signals to the market (e.g. Anand
& Khanna, 2000a; Arend, 2004), many alliances fail during implementation.
Interviewed managers concur that, at least in the software industry, about 40 to 50%
of publicly announced alliances have gone inactive or dysfunctional after one or two
years. However, the short-term market returns may reflect long-term alliance
outcomes to some extent in an efficient market. For example, Kale et al. (2002) use
survey data to show that stock market-based measures of alliance success (i.e.
abnormal returns) are positively correlated with managerial assessments. Therefore,
event-studies lend some support for value creation in the post-formation dynamics of
alliances.
Future research on alliance experience can advance the understanding about
how firms derive capabilities from specific aspects of experience. This dissertation
focuses on the breadth and depth dimensions with respect to experience with
governance structures. With measures about performance of prior alliances, we can
develop a more complete picture of firm learning. For example, the effectiveness of
learning can be affected by successful versus failed relationships; governance
experience may contribute to value creation given different performance records.
The market takes into account of past information, including alliance outcomes. Thus,
the performance factor can provide further explanatory power for market responses
to the news of alliance formation.
241
CHAPTER 6: CONCLUSIONS
6.1 Review of Motivations
Although classic transaction cost economics (TCE) does not directly address
the effects of alliance governance structures on firm capabilities, much has been
done to enrich this perspective. The resource-based view, with a focus on capabilities,
has established that the heterogeneity in firm resources represents opportunities to
integrate and leverage external resources through interfirm relationships (Eisenhardt
& Schoonhoven, 1996; Hitt, et al., 2000; Park & Zhou, 2005). The RBV-based
analysis of capabilities thus complements the TCE logicfor governance decisions.
In particular, firm-level technological capabilities have been examined as
main drivers for the governance decision of interfirm exchanges (e.g. Argyres, 1996;
Hoetker, 2005; Leiblein & Miller, 2003; Mayer & Salomon, 2006). This dissertation
extends this stream of research by adding experience-based alliance capabilities.
Specifically, I propose that firms derive capabilities from prior alliance experience.
Different characteristics of alliance experience generate different types of
capabilities and hence, firms can exploit such capabilities when they choose
subsequent alliance governance structure. These capabilities also influence the
market valuation of firms because investors consider these capabilities as significant
sources of value creation.
The central motivation behind this dissertation is to analyze the role of
alliance experience. As interfirm alliances become increasingly more frequent, firms
242
should constantly step back from forging new relationships and ask themselves:
What can we learn from prior alliances to make future ones more effective and more
value-enhancing? In order to answer this question, the two empirical essays examine
orthogonal dimensions of governance experience: in-depth experience with specific
structures and broad experience across governance forms. These dimensions reflect
two types of underlying firm capabilities: the governing capability in the former and
the selection capability in the latter. Firms consider these capabilities as they make
future governance decisions. Market investors also make inferences about a firm‘s
capabilities based on the firm‘s observed history of alliances. Therefore, the
characteristics of governance experience contribute to both firm-level structural
decisions and alliance-level market value creation. In sum, the two empirical essays
highlight the importance of experience-based capabilities with respect to designing
and managing alliances. Experience as a multi-dimensional construct deserves
further academic research.
6.2 Summary of Findings
While the alliance literature has examined firm-level alliance experience as a
determinant for capability development (e.g. Kale et al., 2002; Kale & Singh, 2007;
Reuer et al., 2002), there has been rare effort in dimensionalizing alliance experience
and its consequences. The two empirical essays attempt to fill this gap by analyzing
dimensions of alliance experience and their consequences on governance decisions
and market value creation. The focus is the governance aspect of experience, that is,
243
the degree to which a firm has experience in specific or diverse governance
structures in prior alliances.
The first essay examines alliance governance decisions as the outcome of
both alliance-level contractual hazards and firm-level experientially developed
capabilities. More hierarchical governance forms, for example, joint ventures relative
to licensing agreements, can better align partner incentives and coordinate adaptation
in the post-formation stages under high contractual hazards, specifically the risks of
appropriability, measurement, and task complexity in a given alliance. Meanwhile,
experience in specific governance forms creates structure-specific knowledge and
capabilities. As a result, firms are likely to exploit such capabilities by managing
similarly structured relationships, thus make path-dependent governance decisions.
A more interesting dimension of governance experience is the degree to
which a firm has managed diverse governance structures. Governance diversity as
explorative learning helps firms to be more informed about designs and post-
formation dynamics in different structures. Firms can then build capabilities from
diverse experience to foresee potential hazards in new relationships and to choose
governance that is better aligned with contractual hazards. Experience in diverse
governance forms also allows the firm to have necessary knowledge to choose
alternative structures more freely, instead of being confined by a history of focused
governance decisions.
The second essay examines stock market reactions to firms‘ experience
factors upon the event of new alliance formation. The stock market is shown to
244
prefer specialized to generalized governance experience: Firms with more experience
in the focal alliance‘s governance receive greater abnormal returns. This appreciation
of experience is strengthened under governance diversity, because diverse
experience fosters the selection capability to make better informed governance
decisions. Finally, the experience effects are bounded by the structure of a focal
alliance and by the performance of a focal firm. Experience is shown to be most
beneficial when a new alliance is governed by intermediate collaborative structures,
and when a focal firm has an intermediate level of prior performance.
The models proposed in the two essays have found strong supports from a
sample of publicly announced alliances in the US software industry during 2001-
2005. Companies in the software industry are active in technological development,
market entry, and alliance formations. The findings offer implications that can be
generalized to industrial settings with similar characteristics.
6.3 Contributions to Theories and Practice
This dissertation suggests that, given the history of an industry and/or the
history of a firm in the industry, it might be infeasible to develop a universal logic
for normative decisions about alliance governance modes. Rather, predictions on
governance decisions will change as industry structures evolve with the capabilities
of firms that specialize in different types of work or alliance activities, and with
variations in firm-specific capabilities based on their prior experience. Firm-level
experience accumulated over time thus represents a key determinant for governance
245
decisions, in addition to the classic transaction-level contractual hazards. Bringing
experience and capabilities into the TCE framework contributes to the research on
governance design that integrates contractual hazards and firms‘ technological
capabilities (Hoetker, 2005; Leiblein & Miller, 2003; Mayer & Salomon, 2006).
Learning theory can also be enriched by incorporating governance structures as a
channel of learning.
6.3.1 Theoretical Contributions and Implications
While TCE assumes economic actors are boundedly rational, it does not
explicitly identify the underlying reasons for this outcome. Understanding
experience-based capabilities extends the TCE research by presenting a possible
explanation for bounded rationality. Firms may choose governance structures that are
similar to prior decisions for reasons of value exploitation or cognitive limitations.
With a history of focused governance experience, firms gravitate toward the focused
form and become more narrowly bounded when assessing contractual hazards. When
governance experience broadens to incorporate more diverse structures, the liability
of focused experience is alleviated. Indeed, Williamson (1999) recognized that firms
enter into exchanges with pre-existing capabilities, and that it is important to
understand the learning process through which firms reach certain capabilities. This
dissertation responds to his call for firm-level learning processes. Explorative
learning among different organizational structures broadens the range of a firm‘s
knowledge, and hence, reduces its bounded rationality when evaluating contractual
risks embedded in a specific exchange. Therefore, the pre-existing capabilities that
246
affect governance decisions can be traced back to prior experience with governance
modes.
An interesting implication is that firms may balance their concerns about
contractual hazards and experience-based capabilities when making governance
decisions. For example, when TCE predicts that an interfirm relationship should be
governed by a hierarchical structure, such as a joint venture, to minimize transaction
costs, the participating firms may not have sufficient capabilities to do so. Lacking
experience in highly collaborative and hierarchical forms may lead the participants to
organize new relationships with a structure that they are most familiar with and thus
are most capable of managing. Such capability-based choices may or may not be
consistent with TCE predictions. Therefore, it is expected that alliance governance
may be aligned with either contractual hazards, or experiential capabilities, or a
combination of both sets of factors.
The findings also suggest that value creation occurs not only in exploiting
experiential learning in general, but also in better informed governance decisions.
Alliance research often adopts the learning theory and suggests that the overall
volume of alliance activities improves firm and alliance performance (e.g. Hoang &
Rothaermel, 2005; Kale et al., 2002; Sampson, 2005; Zollo et al., 2002). However,
this approach neglects the role of governance forms in creating benefits for the
participating firms. This dissertation proposes that learning benefits also accrue when
a firm has selection capabilities derived from diverse governance experience. The
mechanism is that governance diversity enables more appropriate structural decisions
247
by helping to identify relevant contractual hazards. Thus, this dissertation also
contributes to the strategy literature by showing how learning processes can be
augmented by characteristics of governance experience.
The empirical findings of the dissertation collectively suggest that there are
mutual benefits between learning and governance decisions: TCE obtains greater
explanatory power when taking firm learning into account; and learning theories are
more complete when considering different governance forms as contributing to
learning processes. Therefore, the effects of accumulated experience can be better
understood by integrating the arguments from both theories. In addition, the RBV
helps bridging TCE and the learning arguments in the context of alliance experience.
Specifically, the experiential learning process contains various elements that can
build different types of capabilities. Experience-based capabilities thus represent a
form of non-imitable resource due to path-dependent organizational history. And,
firm-specific capabilities determine the extent to which a firm is capable of
foreseeing ex-post contractual hazards and make better informed governance
decisions.
Further, the capability view and the learning theories have yet to examine
governance forms as a source of capability building. This dissertation attempts to fill
this gap by analyzing capabilities that emerge from experience in focused versus
diverse governance structures. The key argument is that structures can determine
how learning occurs, that is, under what structural characteristics, e.g. partner
interaction norms, dispute resolutions, and output divisions, a relationship is
248
coordinated and implemented. In sum, strategy theories can be advanced by
integrating firm capabilities into governance decisions, and by bringing governance
structures into the learning process.
6.3.2 Empirical Contributions and Managerial Implications
While capabilities contribute to normative decisions about alliance
governance and to firms‘ market value creation, managers should pay attention to the
dynamic process through which the capabilities are developed. The two empirical
essays collectively suggest that, while the governing capability developed from
focused experience is value-creating, it most likely imposes cognitive constraints
when deciding the optimal form of alliance governance. Therefore, managers should
be aware that focused governance experience may lead to the dependence of the firm
on a narrow range of knowledge and capabilities, which may limit the firm‘s
capabilities to choose an appropriately aligned governance form with transactional
hazards. Misaligned governance structures may take the form of uncontrolled
opportunism or excessive bureaucracy (Samspon, 2004), and are shown to result in
inferior firm performance (Silverman & Nickerson, 2003).
Managers may want to pay more attention to how a firm‘s alliance portfolio
is constructed. Recent research on alliance portfolio has examined the proportion of a
given characteristic among alliance partners, for example, the percentage of
prominent partners in a firm‘s social network (Lavie, 2007) and the ratio of
international partners in a portfolio (Lavie & Miller, 2008), determine a focal firm‘s
market performance. Differing from these studies, this dissertation contributes to the
249
portfolio research by focusing on the effects of the sequence of alliance portfolio
construction. Specifically, the two-by-two matrix in Figure 5.1 shows that being a
generalist at a low level of relevant governing capabilities makes a firm worse off
than being a specialist who keeps adding similarly structured alliances. Moreover,
the market takes into consideration a focal firm‘s past performance and its focal
alliance‘s governance form. Intermediate levels of collaborative structure and prior
performance can be more value-creating than either high or low levels of these
characteristics. Therefore, firms should be aware of possible market perceptions
when building or expanding their alliance portfolios.
In addition, firms can reap financial benefits from prior experience when
forming new alliances. In the assessment of alliance formation, the market seems to
prefer specialists (i.e. focused governance experience) to generalists when the firms
engage in new alliances and use their specialized governance forms. Everything else
equal, specialists in any governance structure are perceived as having sufficient
knowledge and skills to govern the focal form more effectively than generalists.
Thus firms may create additional value by developing governing capabilities that are
specific to certain structures. In general, the decision of how to govern an interfirm
relationship should be scrutinized according to two major types of concerns: the risk
of losing values of alliance-specific investments and the capabilities of managing a
particular structure.
250
6.4 Limitations and Directions for Future Research
Besides the specific limitations noted in each empirical essay, this
dissertation as a whole should benefit from future research that can address several
key limitations. First, while this dissertation developed a more nuanced theoretical
framework by invoking both the learning theory and TCE logics, the normative
question that still remains to be answered is: what is the right form of governance?
Given the path-dependent tendency in governance decisions, there is a chance that a
firm‘s history may predict governance choices that are at odds with what TCE would
predict. However, it is not obvious why this discrepancy emerges. It can be that
because the firm has insufficient capabilities to identify contractual hazards, it cannot
align hazards with appropriate governance forms. Alternatively, it can be that the
firm recognizes the hazards and the TCE-predicted form, but still chooses the
historically preferred form to fit with its alliance capabilities. In this case, the firm
undertakes misaligned governance according to TCE, but properly aligned with its
own capabilities to manage alliances. Thus it remains ambiguous which choice is a
better one for a focal firm.
Therefore, an important extension to this dissertation that can be addressed
by future research is the examination of ―right‖ form of governance structures. TCE
scholars have suggested that misaligned governance structures with contractual
hazards are detrimental to firm performance (Sampson, 2004; Silverman &
Nickerson, 2003). Given this downside, firms are expected to choose the properly
aligned governance forms in an equilibrium state. However, the percentage of
251
misaligned transactions remains high (in the range of 40%) in these studies. This
high ratio is explained by prior research as the result of firms‘ bounded rationality.
Instead, this dissertation suggests that experiential capabilities can explain why firms
may deviate from a TCE-optimal choice that minimizes transaction costs. Therefore,
structural misalignment in the view of TCE may in effect be a proper alignment with
firm capabilities.
Subsequent research should also examine the conditions under which each
type of alignment is a better choice. For example, Adobe has committed to long-term
alliances governed by non-equity collaborative structures (e.g. its decade-long
relationship with Apple). With this type of governance experience, will Adobe
benefit more from future alliances governed by TCE-predicted or path-dependent
structures, if different? To address this question, it is necessary to use the approach
of matched sample analysis to compare actual performance with counterfactuals that
are aligned with different factors (i.e. contractual hazards versus prior experience).
Further, the optimal type of alignment is likely to vary with the dynamics of an
industry and with the evolution of firm experience. Hence, it is important to examine
multiple industries with different dynamics and longer time periods that enable us to
trace evolutionary patterns at both the firm and industry levels.
Second, while the empirical tests in the dissertation included the period
between 2001 and 2005, this period is still relatively short to capture the dynamics in
the software industry. In particular, due to the technology market crash from the late
1990s to early 2000s, the underlying rationale for why firms entered into interfirm
252
alliances may have changed over the time period examined. For example, software
companies in the 1990s were highly active in forming interfirm ties. During this
period there might well be weak strategic scrutiny of the governance mode choices.
From 2002 to 2005, however, firms became much more stringent in evaluating
partnerships, and avoided strong resource commitments such as joint ventures. Thus,
it is reasonable to expect that experience factors will influence governance decisions
and stock market returns differentially across different time periods. Further research
that tests these empirical relationships during different time periods will help in
assessing the generalizability of the findings from this dissertation.
Third, the sample used in this dissertation was drawn from archival data that
are built upon public reports. These data sources have an inherent deficiency in
measuring alliance performance: alliance activities and target outcomes are reported
idiosyncratically, because there is no requirement for uniformity on the information
provided in alliance announcements. In the absence of a performance measure, the
assumption this dissertation relies on is that governance decisions and market
valuations are affected by prior successes and failures in a relatively homogeneous
fashion. However, this assumption may be violated particularly when investors have
strong impressions about prior alliance performance. For example, the equity
alliance between Oracle and PeopleSoft from the late 1990s turned into a two-year
battle of hostile acquisition since 2003. The market observes this outcome and will
take it into account when evaluating Oracle‘s new interfirm alliances. Similarly, in
the pharmaceutical industry, large companies acquire smaller alliance partners. The
253
market subsequently factors this information into the assessment of the
pharmaceutical companies‘ new alliance formation. Therefore, while it is difficult to
measure alliance performance consistently across deals, prior outcomes do influence
investor perceptions, and can subsequently affect the organizational structures the
focal firms will adopt in the future. However, the focus of this dissertation is
capabilities derived from managing focused or diverse governance modes. Thus, the
assumption of minimal performance impacts is valid to the extent that learning and
capability building occur regardless of how successful past alliances are.
Therefore, another interesting area following this dissertation would be the
performance implications of experience. Experience as a learning process tends to be
simplified and specialized to accommodate firms‘ cognitive capabilities (Levinthal &
March, 1993). There exists the tendency that firms attribute successful experience to
their right decisions. As a result, firms would exploit more frequently governance
forms that have led to successes than those with unsuccessful experience. On the
other hand, alliance managers have acknowledged that they tend to adjust behaviors
to a greater extent when ―something from the past comes back to bite you‖
(interviews with senior managers). This is the argument made by the research on
learning from failures (Arino & de la Torre, 1998). In either direction, the
performance of prior alliance experience can have a distinct effect on strategic
decisions, over and above the effects of the depth-breadth characteristics of
governance experience. Future research can identify the outcomes of prior alliance
254
experience, and then examine when and to what extent prior performance matters in
subsequent decisions of governance modes.
To summarize, this dissertation emphasizes the role of experience in affecting
strategic decisions and value creation. When firms engage in an increasing number
of interfirm alliances, it is important for them to reflect carefully on what they can
learn from different aspects of prior experience, so that they can make their next
steps more productive. Assessments of economic benefits and risk factors in any
strategic choice are inevitably accompanied by an evaluation of firm-level resources
and capabilities at the given time. Revisiting one‘s past is one fruitful source to build
and leverage experience-based capabilities to move forward. More than two
thousand years ago, Confucius captured the value of experience succinctly: ―Study
the past if you would define the future.‖
255
REFERENCES
Aggarwal, V. A., & Hsu, D. H. (2009). Modes of Cooperative R&D
Commercialization by Start-Ups. Strategic Management Journal, 30(8), 835-
864.
Ahuja, G. (2000). The Duality of Collaboration: Inducements and Opportunities in
the Formation of Interfirm Linkages. Strategic Management Journal, 21(3),
317-343.
Akerlof, G. A. (1970). The Market for "Lemons": Quality Uncertainty and the
Market Mechanism. The Quarterly Journal of Economics, 84(3), 488-500.
Almeida, P., Song, J., & Grant, R. M. (2002). Are Firms Superior to Alliances and
Markets? An Empirical Test of Cross-Border Knowledge Building.
Organization Science, 13(2), 147-161.
Anand, B. N., & Khanna, T. (2000a). Do Firms Learn to Create Value? The Case of
Alliances. Strategic Management Journal, 21(3), 295-315.
Anand, B. N., & Khanna, T. (2000b). The Structure of Licensing Contracts. Journal
of Industrial Economics, 48(1), 103-135.
Arend, R. J., & Seale, D. A. (2005). Modeling Alliance Activity: An Iterated
Prisoners' Dilemma with Exit Option. Strategic Management Journal, 26(11),
1057-1074.
Argote, L. (1999). Micro Underpinnings of Organizational Learning. In
Organizational Learning: Creating, Retaining, and Transferring Knowledge.
Argote, L., Beckman, S. L., & Epple, D. (1990). The Persistence and Transfer of
Learning in Industrial Settings. Management Science, 36(2), 140-154.
256
Argyres, N. (1996). Evidence on the Role of Firm Capabilities in Vertical Integration
Decisions. Strategic Management Journal, 17, 129-150.
Argyres, N., & Liebeskind, J. P. (1999). Contractual Commitments, Bargaining
Power, and Governance Inseparability: Incorporating History into
Transaction Cost Theory. Academy of Management Review, 2(1), 49-63.
Argyres, N., & Mayer, K. J. (2007). Contract Design as a Firm Capability: An
Integration of Learning and Transaction Cost Perspectives. Academy of
Management Review, 32(4), 1060-1077.
Argyres, N. S., Bercovitz, J., & Mayer, K. J. (2007). Complementarity and Evolution
of Contractual Provisions: An Empirical Study of IT Services Contracts.
Organization Science, 18(1), 3-19.
Argyres, N. S., & Zenger, T. (2010). Capabilities, Transaction Costs, and Firm
Boundaries: A Dynamic Perspective and Integration. SSRN Working Paper.
Arino, A., & de la Torre, J. (1998). Learning from Failure: Towards an Evolutionary
Model of Collaborative Ventures. Organization Science, 9(3), 306-325.
Bamford, J., & Ernst, D. (2002). Managing An Alliance Portfolio. McKinsey
Quarterly, 3, 29-39.
Barnett, W. P., & Hansen, M. T. (1996). The Red Queen in Organizational Evolution.
Strategic Management Journal, 17(7), 139.
Barney, J. (1991). Firm Resources And Sustained Competitive Advantage. Journal
of Management, 17(1), 99-120.
Barney, J. B. (1986). Strategic Factor Markets: Expectations, Luck, and Business
Strategy. Management Science, 32(10), 1231-1241.
257
Baum, J. A. C., Li, S. X., & Usher, J. M. (2000). Making the Next Move: How
Experiential and Vicarious Learning Shape the Locations of Chains'
Acquisitions. Administrative Science Quarterly, 45(4), 766-801.
Beckman, C. M., Haunschild, P. R., & Phillips, D. J. (2004). Friends or Strangers?
Firm-Specific Uncertainty, Market Uncertainty, and Network Partner
Selection. Organization Science, 15(3), 259-275.
Bleeke, J., & Ernst, D. (1995). Is Your Strategic Alliance Really A Sale. Harvard
Business Review, 73(1), 97-105.
Chung, S. A., Singh, H., & Lee, K. (2000). Complementarity, Status Similarity and
Social Capital as Drivers of Alliance Formation. Strategic Management
Journal, 21, 1-22.
Coase, R. (1937). The Nature of the Firm. Economica, 4, 386-405.
Cohen, W. M., & Levinthal, D. A. (1990). Absorptive Capacity: A New Perspective
on Learning and Innovation. Administrative Science Quarterly, 15, 128-152.
Colombo, M. G. (2003). Alliance Form: A Test of the Contractual and Competence
Perspectives. Strategic Management Journal, 24(12), 1209-1229.
Conner, K. R., & Prahalad, C. K. (1996). A Resource-based Theory of the Firm:
Knowledge Versus Opportunism. Organization Science, 7(5), 477-501.
Darr, E. D., Argote, L., & Epple, D. (1995). The Acquisition, Transfer, and
Depreciation of Knowledge in Service Organizations: Productivity in
Franchises. Management Science, 41(11, Focused Issue on Service
Management), 1750-1762.
Das, S., Sen, P. K., & Sengupta, S. (1998). Impact of Strategic Alliances on Firm
Valuation. Academy of Management Journal, 41(1), 27-41.
258
Das, T. K., & Teng, B.-S. (2000). Instabilities of Strategic Alliances: An Internal
Tensions Perspective. Organization Science, 11(1), 77-101.
Dierickx, I., & Cool, K. O. (1989). Asset Stock Accumulation and Sustainability of
Competitive Advantage. Management Science, 35(12), 1504-1511.
Doz, Y. L. (1996). The Evolution of Cooperation in Strategic Alliances: Initial
Conditions or Learning Processes? Strategic Management Journal,
17(Special Issue: Evolutionary Perspectives on Strategy), 55-83.
Doz, Y. L., & Hamel, G. (1998). The Alliance Advantage: The Art of Creating Value
through Partnership. Boston, MA: Harvard University Press.
Dyer, J. H., & Singh, H. (1998). The Relational View: Cooperative Strategy and
Sources of Interorganizational Competitive Advantage. Academy of
Management Review, 23(4), 660-679.
Eisenhardt, K. M., & Martin, J. A. (2000). Dynamic Capabilities: What Are They?
Strategic Management Journal, 21(10-11), 1105-1121.
Eisenhardt, K. M., & Schoonhoven, C. B. (1996). Resource-Based View of Strategic
Alliance Formation: Strategic and Social Effects in Entrepreneurial Firms.
Organization Science, 7(2), 136-150.
Epple, D., Argote, L., & Devadas, R. (1991). Organizational Learning Curves: A
Method for Investigating Intra-Plant Transfer of Knowledge Acquired
Through Learning by Doing. Organization Science, 2(1, Special Issue:
Organizational Learning: Papers in Honor of (and by) James G. March), 58-
70.
Fiol, C. M., & Lyles, M. A. (1985). Organizational Learning. Academy of
Management Review, 10(4), 803-813.
Folta, T. B. (1998). Governance and Uncertainty: The Trade-off between
Administrative Control and Commitment. Strategic Management Journal,
19(11), 1007-1028.
259
Fombrun, C., & Shanley, M. (1990). What's in a Name? Reputation Building and
Corporate Strategy. The Academy of Management Journal, 33(2), 233-258.
Foss, N. J. (1996). Knowledge-Based Approaches to the Theory of the Firm: Some
Critical Comments. Organization Science, 7(5), 470-476.
Gerwin, D., & Ferris, J. S. (2004). Organizing New Product Development Projects in
Strategic Alliances. Organization Science, 15(1), 22-37.
Goerzen, A. (2007). Alliance Networks and Firm Performance: The Impact of
Repeated Partnerships. Strategic Management Journal, 28(5), 487-509.
Grant, R. M. (1996). Toward A Knowledge-Based Theory of the Firm. Strategic
Management Journal, 17(Winter special issue), 109-122.
Gulati, R. (1995). Does Familiarity Breed Trust? The Implications of Repeated Ties
for Contractual Choice in Alliances. Academy of Management Journal, 38(1),
85-112.
Gulati, R., & Higgins, M. C. (2003). Which Ties Matter When? The Contingent
Effects of Interorganizational Partnerships on IPO Success. Strategic
Management Journal, 24(2), 127-144.
Gulati, R., & Nickerson, J. A. (2008). Interorganizational Trust, Governance Choice,
and Exchange Performance. Organization Science, 19(5), 688-708.
Gulati, R., & Singh, H. (1998). The Architecture of Cooperation: Managing
Coordination Costs and Appropriation Concerns in Strategic Alliances.
Administrative Science Quarterly, 43(4), 781-814.
Gulati, R., & Westphal, J. D. (1999). Cooperative or Controlling? The Effects of
CEO-Board Relations and the Content of Interlocks on the Formation of Joint
Ventures. Administrative Science Quarterly, 44(3), 473-506.
260
Hagedoorn, J., & Hesen, G. (2007). Contract Law and the Governance of Inter-firm
Technology Partnerships - An Analysis of Different Modes of Partnering and
Their Contractual Implications. Journal of Management Studies, 44(3), 342-
366.
Hayward, M. L. A. (2002). When do Firms Learn from Their Acquisition Experience?
Evidence from 1990-1995. Strategic Management Journal, 23(1), 21-39.
Hennart, J.-F. (1988). A Transaction Costs Theory of Equity Joint Ventures.
Strategic Management Journal, 9(4), 361-374.
Hitt, M. A., Dacin, M. T., Levitas, E., Arregle, J. L., & Borza, A. (2000). Partner
Selection in Emerging and Developed Market Contexts: Resource-Based and
Organizational Learning Perspectives. Academy of Management Journal,
43(3), 449-467.
Hoang, H., & Rothaermel, F. T. (2005). The Effect of General and Partner-Specific
Alliance Experience on Joint R&D Project Performance. Academy of
Management Journal, 48(2), 332-345.
Hoetker, G. (2005). How Much You Know Versus How Well I Know You:
Selecting a Supplier for a Technically Innovative Component. Strategic
Management Journal, 26(1), 75-96.
Hoffmann, W. H. (2007). Strategies for Managing a Portfolio of Alliances. Strategic
Management Journal, 28(8), 827-856.
Huber, G. P. (1991). Organizational Learning: The Contributing Processes and the
Literature. Organization Science, 2(1), 88-115.
Ingram, P., & Baum, J. A. C. (1997). Opportunity and Constraint: Organizations'
Learning from the Operating and Competitive Experience of Industries.
Strategic Management Journal, 18(6), 75-98.
Inkpen, A. C., & Currall, S. C. (2004). The Coevolution of Trust, Control, and
Learning in Joint Ventures. Organization Science, 15(5), 586-599.
261
Jacobides, M. G. (2008). How Capability Differences, Transaction Costs, and
Learning Curves Interact to Shape Vertical Scope. Organization Science,
19(2), 306-326.
Kale, P., Dyer, J. H., & Singh, H. (2002). Alliance Capability, Stock Market
Response, and Long-Term Alliance Success: The Role of The Alliance
Function. Strategic Management Journal, 23, 747-767.
Kale, P., & Singh, H. (2007). Building Firm Capabilities Through Learning: The
Role of the Alliance Learning Process in Alliance Capability and Firm-Level
Alliance Success. Strategic Management Journal, 28(10), 981-1000.
Kale, P., Singh, H., & Perlmutter, H. (2000). Learning and Protection of Proprietary
Assets in Strategic Alliances: Building Relational Capital. Strategic
Management Journal, 21(3), 217-237.
Kalnins, A., Swaminathan, A., & Mitchell, W. (2006). Turnover Events, Vicarious
Information, and The Reduced Likelihood of Outlet-Level Exit among Small
Multiunit Organizations. Organization Science, 17(1), 118-131.
Khanna, T. (1998). The Scope of Alliances. Organization Science, 9(3), 340-355.
Khanna, T., Gulati, R., & Nohria, N. (1998). The Dynamics of Learning Alliances:
Competition, Cooperation, and Relative Scope. Strategic Management
Journal, 19(3), 193-210.
Klein, B., Crawford, R. G., & Alchian, A. A. (1978). Vertical Integration,
Appropriable Rents and the Competitive Contracting Process. Journal of Law,
Economics, and Organization, 21, 297-326.
Kogut, B., & Zander, U. (1992). Knowledge of the Firm, Combinative Capabilities,
and the Replication of Technology. Organization Science, 3(3, Focused Issue:
Management of Technology), 383-397.
Kogut, B., & Zander, U. (1996). What Firms Do? Coordination, Identity, and
Learning. Organization Science, 7(5), 502.
262
Koka, B. R., & Prescott, J. E. (2008). Designing Alliance Networks: The Influence
of Network Position, Environmental Change, and Strategy on Firm
Performance. Strategic Management Journal, 29(6), 639-661.
Kumar, M. V. S. (2005). The Value from Acquiring and Divesting a Joint Venture: a
Real Options Approach. Strategic Management Journal, 26, 321-331.
Langlois, R. N. (1992). Transaction-Cost Economics in Real Time. Industrial and
Corporate Change, 1, 99-127.
Lavie, D. (2007). Alliance Portfolios and Firm Performance: A Study of Value
Creation and Appropriation in the US Software Industry. Strategic
Management Journal, 28, 1187-1212.
Lavie, D., & Miller, S. R. (2008). Alliance Portfolio Internationalization and Firm
Performance. Organization Science, 19(4), 623-646.
Lavie, D., & Rosenkopf, L. (2006). Balancing Exploration and Exploitation in
Alliance Formation. Academy of Management Journal, 49(4), 797-818.
Leiblein, M. J. (2003). The Choice of Organizational Governance Form and
Performance: Predictions from Transaction Cost, Resource-Based, and Real
Options Theories. Journal of Management, 29(6), 937-961.
Leiblein, M. J., & Miller, D. J. (2003). An Empirical Examination of Transaction-
and Firm-Level Influences on the Vertical Boundaries of the Firm. Strategic
Management Journal, 24(9), 839-859.
Leiblein, M. J., Reuer, J. J., & Dalsace, F. (2002). Do Make or Buy Decisions Matter?
The Influence of Organizational Governance on Technological Performance.
Strategic Management Journal, 23(9), 817-833.
Levinthal, D. A., & March, J. G. (1981). A Model of Adaptive Organizational Search.
Journal of Economic Behavior and Organization, 2, 307-333.
263
Levinthal, D. A., & March, J. G. (1993). The Myopia of Learning. Strategic
Management Journal, 14(8), 95.
Levitt, B., & March, J. G. (1988). Organizational Learning. Annual Review of
Sociology, 14, 319-340.
Li, S. X., & Rowley, T. J. (2002). Inertia and Evaluation Mechanisms in
Interorganizational Partner Selection: Syndicate Formation among US
Investment Banks. Academy of Management Journal, 45(6), 1104-1119.
Lorenzoni, G., & Lipparini, A. (1999). The Leveraging of Interfirm Relationships as
a Distinctive Organizational Capability: A Longitudinal Study. Strategic
Management Journal, 20(4), 317-338.
Luo, Y. D. (2005). Transactional Characteristics, Institutional Environment and Joint
Venture Contracts. Journal of International Business Studies, 36(2), 209-230.
Macher, J. T. (2006). Technological Development and the Boundaries of the Firm: A
Knowledge-Based Examination in Semiconductor Manufacturing.
Management Science, 52(6), 826-843.
Madhok, A. (1996). The Organization of Economic Activity: Transaction Costs,
Firm Capabilities, and the Nature of Governance. Organization Science, 7(5),
577-590.
Mahoney, J. T., & Pandian, J. R. (1992). The Resource-Based View within the
Conversation of Strategic Management. Strategic Management Journal,
13(5), 363-380.
March, J. G. (1991). Exploration and Exploitation in Organizational Learning.
Organization Science, 2(1, Special Issue: Organizational Learning: Papers in
Honor of (and by) James G. March), 71-87.
March, J. G., & Simon, H. A. (1958). Organizations (2nd ed.). Cambridge, Mass.,
USA: Blackwell.
264
March, J. G., Sproull, L. S., & Tamuz, M. (1991). Learning from Samples of One or
Fewer. Organization Science, 2(1, Special Issue: Organizational Learning:
Papers in Honor of (and by) James G. March), 1-13.
Mayer, K. J., & Argyres, N. S. (2004). Learning to Contract: Evidence from the
Personal Computer Industry. Organization Science, 15(4), 394-410.
Mayer, K. J., & Nickerson, J. A. (2005). Antecedents and Performance Implications
of Contracting for Knowledge Workers: Evidence from Information
Technology Services. Organization Science, 16(3), 225-242.
Mayer, K. J., & Salomon, R. M. (2006). Capabilities, Contractual Hazards, and
Governance: Integrating Resource-Based and Transaction Cost Perspectives.
Academy of Management Journal, 49(5), 942-959.
Mayer, K. J., & Teece, D. J. (2008). Unpacking Strategic Alliances: The Structure
and Purpose of Alliance versus Supplier Relationships. Journal of Economic
Behavior and Organization, 66(1), 106-127.
McGrath, R. G. (2001). Exploratory Learning, Innovative Capacity, and Managerial
Oversight. Academy of Management Journal, 44(1), 118-131.
Mesquita, L. F., Anand, J., & Brush, T. H. (2008). Comparing the Resource-Based
and Relational Views: Knowledge Transfer and Spillover in Vertical
Alliances. Strategic Management Journal, 29(9), 913-941.
Montgomery, C. A., & Wernerfelt, B. (1988). Diversification, Ricardian rents and
Tobin's q. Rand Journal of Economics, 19(4), 623-632.
Mowery, D. C., Oxley, J. E., & Silverman, B. (1998). Technological Overlap and
Interfirm Cooperation: Implications for the Resource-Based View of the Firm.
Research Policy, 27(5), 507-523.
Nelson, R. R., & Winter, S. G. (1982). An Evolutionary Theory of Economic Change.
Cambridge, MA: Belknap.
265
North, D. C., & Thomas, R. P. (1973). The Rise of the Western World: A New
Economic History. Cambridge: Cambridge University Press.
Olk, P., & Young, C. (1997). Why Members Stay in or Leave an R&D Consortium:
Performance and Conditions of Membership as Determinants of Continuity.
Strategic Management Journal, 18(11), 855-877.
Oxley, J. E. (1997). Appropriability Hazards and Governance in Strategic Alliances:
A Transaction Cost Approach. Journal of Law, Economics, & Organization,
13(2), 387-409.
Oxley, J. E. (2003). Learning versus Protection in Interfirm Alliances: A False
Dichotomy. In T. Pedersen & V. Mahnke (Eds.), Knowledge, Governance
and the MNC (pp. 108-129): Palgrave Macmillan.
Oxley, J. E., & Sampson, R. C. (2004). The Scope and Governance of International
R&D Alliances. Strategic Management Journal, 25(8-9), 723-749.
Oxley, J. E., Sampson, R. C., & Silverman, B. S. (2009). Arms Race or Detente?
How Interfirm Alliance Announcements Change the Stock Market Valuation
of Rivals. Management Science, 55(8), 1321-1337.
Oxley, J. E., & Silverman, B. S. (2008). Inter-Firm Alliances: A New Institutional
Economics Approach. In E. Brousseau & J.-M. Glachant (Eds.), New
Institutional Economics: A Guidebook (pp. 209-234). New York: Cambridge
University Press.
Park, S. H., & Zhou, D. (2005). Firm Heterogeneity and Competitive Dynamics in
Alliance Formation. Academy of Management Review, 30(3), 531-554.
Parkhe, A. (1993a). Partner Nationality and the Structure-Performance Relationship
in Strategic Alliances. Organization Science, 4(2), 301-324.
Parkhe, A. (1993b). Strategic Alliance Structuring: A Game Theoretic and
Transaction Cost Examination of Interfirm Cooperation. Academy of
Management Journal, 36(4), 794-829.
266
Penrose, E. T. (1959). The Theory of the Growth of the Firm. New York: Wiley.
Peteraf, M. A. (1993). The Cornerstones of Competitive Advantage: A Resource-
Based View. Strategic Management Journal, 14(3), 179-191.
Pisano, G. P. (1989). Using Equity Participation to Support Exchange: Evidence
from the Biotechnology Industry. Journal of Law, Economics and
Organization, 5(1).
Pisano, G. P. (1990). The R&D Boundaries of the Firm: An Empirical Analysis.
Administrative Science Quarterly, 35(1), 153-176.
Poppo, L., & Zenger, T. (1998). Testing Alternative Theories of the Firm:
Transaction Cost, Knowledge-Based, and Measurement Explanations for
Make-or-Buy Decisions in Information Services. Strategic Management
Journal, 19, 853–877.
Poppo, L., & Zenger, T. (2002). Do Formal Contracts and Relational Governance
Function as Substitutes or Complements? Strategic Management Journal,
23(8), 707-725.
Prahalad, C. K., & Hamel, G. (1990). The Core Competence of the Corporation.
Harvard Business Review(May-June), 79-90.
Priem, R. L., & Butler, J. E. (2001). Tautology in the Resource-Based View and the
Implications of Externally Determined Resource Value: Further Comments.
Academy of Management Review, 26(1), 57.
Reuer, J. J., & Arino, A. (2002). Contractual Renegotiations in Strategic Alliances.
Journal of Management, 28(1), 47-68.
Reuer, J. J., & Arino, A. (2007). Strategic Alliance Contracts: Dimensions and
Determinants of Contractual Complexity. Strategic Management Journal,
28(3), 313-330.
267
Reuer, J. J., & Ragozzino, R. (2006). Agency Hazards and Alliance Portfolios.
Strategic Management Journal, 27(1), 27-43.
Reuer, J. J., & Zollo, M. (2005). Termination Outcomes of Research Alliances.
Research Policy, 34(1), 101-115.
Reuer, J. J., Zollo, M., & Singh, H. (2002). Post-Formation Dynamics in Strategic
Alliances. Strategic Management Journal, 23, 135-151.
Robinson, D. T., & Stuart, T. E. (2007). Network Effects in the Governance of
Strategic Aalliances. Journal of Law Economics & Organization, 23(1), 242-
273.
Rothaermel, F. T., & Deeds, D. L. (2004). Exploration and Exploitation Alliances in
Biotechnology: A System of New Product Development. Strategic
Management Journal, 25(3), 201-221.
Rothaermeli, F. T., & Boeker, W. (2008). Old Technology Meets New Technology:
Complementarities, Similarities, and Alliance Formation. Strategic
Management Journal, 29(1), 47-77.
Rumelt, R. P. (1984). Toward a Strategic Theory of the Firm. In R. Lamb (Ed.),
Competitive Strategic Management (pp. 556-570). New Jersey: Prentice-Hall.
Ryall, M. D., & Sampson, R. C. (2009). Formal Contracts in the Presence of
Relational Enforcement Mechanisms: Evidence from Technology
Development Projects. Management Science, 55(6), 906-925.
Sampson, R. C. (2004). The Cost of Misaligned Governance in R&D Alliances.
Journal of Law, Economics and Organization, 20(2), 484-526.
Sampson, R. C. (2005). Experience, Learning & Collaborative Returns in R&D
Alliances. Strategic Management Journal, 26, 1009-1031.
268
Santoro, M. D., & McGill, J. P. (2005). The Effect of Uncertainty and Asset Co-
Specialization on Governance in Biotechnology Alliances. Strategic
Management Journal, 26(13), 1261-1269.
Sarkar, M. B., Aulakh, P. S., & Madhok, A. (2009). Process Capabilities and Value
Generation in Alliance Portfolios. Organization Science, 20(3), 583-600.
Sarkar, M. B., Echambadi, R. A. J., & Harrison, J. S. (2001). Alliance
Entrepreneurship and Firm Market Performance. Strategic Management
Journal, 22(6-7), 701-711.
Schilling, M. A. (2008). Understanding the Alliance Data. Strategic Management
Journal, 30(3), 233-260.
Shaver, J. M., Mitchell, W., & Yeung, B. (1997). The Effect of Own-Firm and
Other-Firm Experience on Foreign Direct Investment Survival in the United
States, 1987-1992. Strategic Management Journal, 18, 811-824.
Shelanski, H. A., & Klein, P. G. (1995). Empirical Research in Transaction Cost
Economics: A Review and Assessment. Journal of Law, Economics, &
Organization, 11(2), 335-361.
Silverman, B. S., & Nickerson, J. A. (2003). Why Firms Want to Organize
Efficiently and What Keeps Them from Doing so: Inappropriate Governance,
Performance, and Adaptation in a Deregulated Industry. Administrative
Science Quarterly, 48(3), 433-465.
Stuart, T. E. (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), 791-811.
Teece, D. J., Pisano, G., & Shuen, A. (1997). Dynamic Capabilities and Strategic
Management. Strategic Management Journal, 18(7), 509-533.
269
Vassolo, R. S., Anand, J., & Folta, T. B. (2004). Non-Additivity in Portfolios of
Exploration Activities: A Real Options-Based Analysis of Equity Alliances in
Biotechnology. Strategic Management Journal, 25(11), 1045-1061.
Wang, L., & Zajac, E. J. (2007). Alliance or Acquisition? A Dyadic Perspective on
Interfirm Resource Combinations. Strategic Management Journal, 28(13),
1291-1317.
Wassmer, U. (2010). Alliance Portfolios: A Review and Research Agenda. Journal
of Management, 36(1), 141-171.
Wernerfelt, B. (1984). A Resource-Based View of the Firm. Strategic Management
Journal, 14, 4-12.
Williamson, O. E. (1985). The Economic Institutions of Capitalism. New York: The
Free Press.
Williamson, O. E. (1991). Comparative Economic Organization: The Analysis of
Discrete Structural Alternatives. Administrative Science Quarterly, 36(2),
269-296.
Williamson, O. E. (1996). Economic Organization: The Case for Candor. Academy
of Management Review, 21(1), 48-57.
Williamson, O. E. (1999). Strategy Research: Governance and Competence
Perspectives. Strategic Management Journal, 20, 1087-1108.
Young-Ybarra, C., & Wiersema, M. (1999). Strategic Flexibility in Information
Technology Alliances: The Influence of Transaction Cost Economics and
Social Exchange Theory. Organization Science, 10(4), 439-459.
Zahra, S. A., & George, G. (2002). Absorptive Capacity: A Review,
Reconceptualization, and Extension. Academy of Management Review, 27(2),
185-203.
270
Zeng, M., & Chen, X. P. (2003). Achieving Cooperation in Multiparty Alliances: A
Social Dilemma Approach to Partnership Management. Academy of
Management Review, 28(4), 587-605.
Zollo, M., Reuer, J. J., & Singh, H. (2002). Interorganizational Routines and
Performance in Strategic Alliances. Organization Science, 13(6), 701-713.
Zollo, M., & Singh, H. (2004). Deliberate Learning in Corporate Acquisitions: Post-
Acquisition Strategies and Integration Capability in US Bank Mergers.
Strategic Management Journal, 25(13), 1233-1256.
Zollo, M., & Winter, S. G. (2002). Deliberate Learning and the Evolution of
Dynamic Capabilities. Organization Science, 13(3), 339.
271
APPENDIX 1: STANDARD INDUSTRAIL CLASSIFICATION
DEFINITIONS
7371: Computer programming services. Establishments of this industry perform a
variety of additional services, such as computer software design and analysis;
modifications of custom software; and training in the use of custom software.
7372: Prepackaged software. Establishments primarily engaged in the design,
development, and production of prepackaged computer software. Important products
of this industry include operating, utility, and applications programs. Establishments
of this industry may also provide services such as preparation of software
documentation for the user-installation of software for the user; and training the user
in the use of the software.
7373: Computer integrated system design. Establishments primarily engaged in
developing or modifying computer software and packaging or bundling the software
with purchased computer hardware (computers and computer peripheral equipment)
to create and market an integrated system for specific application. Establishments in
this industry must provide each of the following services: (1) the development or
modification of the computer software; (2) the marketing of purchased computer
hardware; and (3) involvement in all phases of systems development from design
through installation.
7374: Computer processing, data processing, and other computer related services.
Establishments primarily engaged in providing computer processing and data
preparation services. The service may consist of complete processing and
preparation of reports from data supplied by the customer or a specialized service,
such as data entry or making data processing equipment available on an hourly or
time-sharing basis.
Source: United States Department of Labor, Occupational Safety & Health Administration
(OSHA, http://www.osha.gov)
272
APPENDIX 2: LIST OF COMPANIES IN EMPIRICAL SAMPLE
ID Firm Name Primary SIC Market Segment
1 ACI Worldwide Inc 7372 Application
2 Activision Inc 7372 Application
3 Actuate Corp 7372 Application
4 Acxiom Corp 7374 IT Service
5 ADAM Inc 7372 IT Service
6 Adept Technology Inc 7372 Other
7 Adobe Systems Inc 7372 System***
8 Advent Software Inc 7372 Application
9 Affiliated Computer Services 7374 IT Service
10 Alliance Data Systems Corp 7374 Other
11 AllScripts Healthcare Solutions 7373 Other
12 Amdocs Ltd 7372 Application
13 American Software Inc 7372 Application
14 Amicas 7372 Other
15 Analysts International Corp 7371 Application
16 Ansoft Corp 7372 Application
17 Ansys Inc 7372 Application
18 Ariba Inc 7372 IT Service
19 Art Technology Group Inc (ATG) 7372 IT Service
20 Astea International Inc 7372 Other
21 AuthentiDate Holding Corp 7373 Other
22 Autodesk Inc 7372 Application
23 Automatic Data Processing 7374 Application
24 Aware Inc 7373 Other
25 Bitstream Inc 7372 Other
26 BMC Software Inc 7372 System
27 Borland Software Corp 7372 Application
28 Bottomline Technologies Inc 7372 Application
29 Bsquare Corp 7372 System
30 Computer Associate Intl (CA) Inc 7372 System
31 CACI International Inc 7373 IT Service
32 Cadence Design Systems Inc 7372 Application
33 Captaris Inc 7372 Application
273
ID Firm Name Primary SIC Market Segment
34 Cash Technologies Inc 7389** Other
35 Cerner Corp 7373 Application
36 Chordiant Software Inc 7372 Other
37 Citrix Systems Inc 7372 Application
38 Compuware Corp 7372 Application
39 Concur Tech Inc 7372 Application
40 Consulier Engineering Inc 7372 Other
41 Convera Corp 7372 Other
42 CSG Systems International Inc* 7374 Application
43 CSP Inc 7373 Application
44 CyberSource Corp 7374 IT Service
45 Datalink Corp 7373 Application
46 DataTrak International Inc 7372 Other
47 DataWatch Corp 7372 Other
48 Descartes Systems Group Inc* 7372 Application
49 Digimarc Corp 7372 System
50 DST Systems Inc 7374 IT Service
51 Dynamics Research Corp 7373 Application
52 Ebix Inc 7372 Application
53 Eclipsys Corp 7373 IT Service
54 Electronic Arts Inc 7372 Application
55 Entrust Technologies Inc 7372 Application
56 Epicor Software Corp 7372 Application
57 eResearch Technology Inc 7372 IT Service
58 Evolving Systems Inc 7371 Other
59 F5 Networks Inc 7373 IT Service
60 Fair Isaac & Co Inc 7373 Application
61 FalconStor Software Inc 7372 Application
62 Fiserv Inc* 7374 Application
63 Global Payments Inc* 7374 Other
64 GSE Systems Inc 7372 Other
65 HealthAxis Inc 7374 IT Service
66 i2 Technologies Inc 7372 Other
67 I-Many Inc 7372 IT Service
68 Informatica Corp 7372 Application
69 Innodata Isogen Corp 7374 IT Service
70 Interactive Intelligence Inc 7372 Other
274
ID Firm Name Primary SIC Market Segment
71 Interwoven Inc 7372 Application
72 Intuit Inc 7372 Application
73 JDA Software Group Inc 7372 Application
74 Keynote Systems Inc 7371 Application
75 Lionbridge Technologies Inc 7372 Application
76 LivePerson Inc 7372 IT Service
77 Logility Inc 7372 Application
78 Macrovision Solutions Corp 7389** System
79 Manhattan Associates Inc 7373 Application
80 McAfee Inc 7372 System
81 Mediware Information Systems Inc 7373 IT Service
82 Mentor Graphics Corp 7373 Application
83 Mercury Computer Systems Inc 7373 Application
84 Merge Healthcare Inc 7373 IT Service
85 Micros Systems Inc 7373 System
86 Microsoft Corp 7372 System
87 MicroStrategy Inc 7372 Other
88 Midway Games Inc 7372 Application
89 MSC Software Corp 7372 Application
90 National Instruments Corp 7372 Application
91 NaviSite Inc 7374 IT Service
92 NetScout Systems Inc 7373 Application
93 Netsol Technologies Inc 7372 Other
94 Network Engines Inc 7373 System
95 Newtek Business Services Inc 7389** Other
96 Nuance Communications Inc 7372 Other
97 On2 Technologies Inc 7372 Other
98 Online Resources Corp 7374 Application
99 Open Text Corp 7372 Application
100 OpenTV Corp 7372 System
101 Openwave Systems Inc 7372 Application
102 Opnet Technologies Inc 7372 Other
103 Oracle Corp 7372 System
104 Parametric Technology Corp 7372 Application
105 PDF Solutions Inc 7373 System
106 Peerless Systems Corp 7372 Other
107 Pegasystems Inc 7372 Application
275
ID Firm Name Primary SIC Market Segment
108 Perot Systems Corp 7373 IT Service
109 Pervasive Software Inc 7372 Application
110 Phoenix Technologies Ltd 7372 System
111 Pinnacle Data Systems 7371 IT Service
112 Progress Software Corp 7372 System
113 QAD Inc 7372 Application
114 Quadramed Corp 7372 Other
115 Quality Systems Inc 7373 System
116 Quest Software Inc 7372 System
117 Radiant Systems Inc 7373 System
118 Realnetworks Inc 7372 Application
119 Red Hat Inc 7372 System
120 Renaissance Learning Inc 7372 Application
121 S1 Corp 7373 Application
122 Saba Software Inc 7372 Application
123 Safeguard Scientifics Inc 7373 Other
124 Sapient Corp* 7372 Application
125 Scientific Learning Corp 7372 Application
126 SeaChange International Inc 7373 System
127 Secure Computing Corp 7372 Other
128 Selectica Inc 7372 Other
129 Smith Micro Software Inc 7372 Application
130 Sonic Foundry Inc 7372 Application
131 Sonic Solutions 7372 Application
132 Sonus Networks Inc 7373 Other
133 SPSS Inc 7372 Application
134 Streamline Health Solutions 7373 Other
135 Sumtotal Systems Inc* 7372 Application
136 SupportSoft Inc 7372 Application
137 Sybase Inc 7372 System
138 Symantec Corp 7372 System
139 Synopsys Inc 7372 Application
140 Take-Two Interactive Software 7372 Application
141 Technology Solutions Co 7373 IT Service
142 TeleCommunication System Inc 7373 Application
143 Tibco Software Inc 7372 Application
144 TriZetto Group Inc 7372 IT Service
276
ID Firm Name Primary SIC Market Segment
145 Tumbleweed Communications Corp 7372 Application
146 Ultimate Software Group Inc 7372 Application
147 Unisys Corp 7373 Other
148 US Dataworks Inc 7372 Other
149 VeriSign Inc 7372 Application
150 Versant Corp 7372 Other
151 Vital Images Inc 7372 Application
152 Voxware Inc* 7372 Other
153 Wind River Systems Inc 7371 System
154 Workstream Inc 7372 Other
* Firms are not included in the second essay due to data attrition.
**7389: Establishments primarily engaged in furnishing business services, not elsewhere classified
(source: http://www.osha.gov/pls/imis/sic_manual.display?id=1013&tab=description). Firms in this
section are occasionally classified as software companies. These firms are included as software
companies according to their reported major businesses in SEC filings.
*** Adobe was classified under ―system‖ prior to Oct. 2006; starting 2007 it is under ―application‖.
―Other‖ includes communication service, data management service, business/financial service, and
other unclassified segments (source: S&P industry survey and company annual reports)
Abstract (if available)
Abstract
In the context of interfirm alliances, this dissertation analyzes partners’ alliance experience as a multi-dimensional construct, and examines the effects of experience dimensions on governance decisions and on market value creations. This dissertation focuses on the governance aspect of experience, or the extent to which a firm has managed focused or diverse alliance governance structures. A firm’s experience of prior alliances can be characterized by the depth in a specific governance form and the breadth of diverse governance forms. In-depth experience creates governing capabilities that are specific to a focal structure and result in exploitation of the same structure. Diverse governance experience broadens the range of alliance-related knowledge, and lead to better informed governance decisions by the creation of selection capabilities. In the first empirical essay, I examine a model that integrates both contractual hazards and experience-based capabilities to predict governance decisions. The second essay takes a further step by examining how the stock market responds to experience factors when evaluating events of new alliance formation. In a sample of alliances formed by US software companies, I find strong empirical evidence for the argument of multi-dimensional experience in affecting strategic decisions and value creations.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Empirical essays on alliances and innovation in the biopharmaceutical industry
PDF
Three essays on young entrepreneurial firms
PDF
The role of capabilities in new alliance creation and performance: a study of the biotechnology industry
PDF
Competing across and within platforms: antecedents and consequences of market entries by mobile app developers
PDF
Three essays on distance: examing the role of institutional distance on foreign firm entry, local isomorphism strategy and subsidiary performance
PDF
How do acquirers govern the deal-making process? Three essays on U.S. mergers and acquisitions 1994 – 2017
PDF
Essays on strategic categorization
PDF
Essays on improving human interactions with humans, algorithms, and technologies for better healthcare outcomes
PDF
Shrinkage methods for big and complex data analysis
PDF
Essays on delegated portfolio management under market imperfections
Asset Metadata
Creator
Wu, Rui
(author)
Core Title
Empirical essays on relationships between alliance experience and firm capability development
School
Marshall School of Business
Degree
Doctor of Philosophy
Degree Program
Business Administration
Publication Date
08/19/2010
Defense Date
04/30/2010
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
experience,firm capability,interfirm relationship,market value creation,OAI-PMH Harvest,organizational learning,strategic alliance,strategic management,transaction cost
Place Name
USA
(countries)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Mayer, Kyle J. (
committee chair
), Fulk, Janet (
committee member
), Kennedy, Mark T. (
committee member
), Rajagopalan, Nandini (
committee member
)
Creator Email
Rui.Wu.2009@marshall.usc.edu,wur@sem.tsinghua.edu.cn
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-m3407
Unique identifier
UC195162
Identifier
etd-Wu-4024 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-393911 (legacy record id),usctheses-m3407 (legacy record id)
Legacy Identifier
etd-Wu-4024.pdf
Dmrecord
393911
Document Type
Dissertation
Rights
Wu, Rui
Type
texts
Source
University of Southern California
(contributing entity),
University of Southern California Dissertations and Theses
(collection)
Repository Name
Libraries, University of Southern California
Repository Location
Los Angeles, California
Repository Email
cisadmin@lib.usc.edu
Tags
firm capability
interfirm relationship
market value creation
organizational learning
strategic alliance
strategic management
transaction cost