University of Southern California Dissertations and Theses

Empirical essays on alliances and innovation in the biopharmaceutical industry

(USC Thesis Other)

Empirical essays on alliances and innovation in the biopharmaceutical industry

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content EMPIRICAL ESSAYS ON ALLIANCES AND INNOVATION IN THE
BIOPHARMACEUTICAL INDUSTRY

by

Luis Diestre

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BUSINESS ADMINISTRATION)

August 2009

Copyright 2009 Luis Diestre

ii

ACKNOWLEDGEMENTS
The best part of my doctoral program was not the many things that I have learnt,
but the people that I have met, with whom I have shared so many experiences these four
years at USC, and to whom I am completely indebted for life. Among all of them, I owe a
great debt to my advisor, mentor and close friend Nandini Rajagopalan. Her constant
encouragement and limitless energy, along with her zeal for perfection have been a
constant motivation all these years. I am especially grateful for her, and her family’s,
friendship and warmth. I am really going to miss our frequent and amusing meetings.
I would also like to acknowledge professors Shantanu Dutta, Kyle Mayer, and
Roger Moon for their assistance and direction in the completion of this dissertation. Their
experience, knowledge, positive and original attitude were definitely responsible for this
accomplishment. I am also infinitely grateful to many professors in the MOR department
that have offered continuous encouragement and have contributed enormously to my
intellectual and personal development over the last four years. Among them, I would like
to thank especially professor Jonathan Jaffee for his support as both, a PhD coordinator
and a friend during my years at USC. Finally I would like to thank the whole MOR
department and the PhD program at the Marshall School of Business for their assistance
and dedication.
I also want to thank all my fellow PhD students in the MOR department. I am
deeply grateful to Jade Lo, Rui Wu, Libby Webber, and Jay Chok among many others for
their sincere friendship. I also thank the many friends that I leave at USC, especially Luis
Goncalves-Pinto, Joshua Shemesh, and Florian Muenkel for sharing so may experiences

iii

and great moments these years. I sincerely believe that all these friendships will last
forever.
Finally, I am most thankful to my wife Puy Sanchez. She did not hesitate to leave
everything aside to join me on this incredible adventure, and has offered continuous love
and support even in the most difficult times. Any accomplishment in these last four years
would have been impossible without having her by my side. Finally, I hope my parents
and sister, to whom I owe so much, will feel proud of my accomplishment.

iv

Table of Contents

Acknowledgements .............................................................................................................ii

List of Tables .....................................................................................................................vi

List of Figures ...................................................................................................................vii

Abstract ............................................................................................................................viii

Chapter 1: Introduction .......................................................................................................1
1.1 Innovation in the Biopharmaceutical Industry ..............................................................1
1.2 Essay 1: The Effects of Firm Experience on New Product Developments:
Does Board Experience Matter? .........................................................................................4
1.3 Essay 2: Are all “Sharks” Dangerous? New Biotechnology Ventures and
Partner Selection in R&D Alliances ...................................................................................6

Chapter 2: Essay one: The Effects of Firm Experience on New Product
Developments: Does Board Experience Matter? ……........................................................9
2.1 Introduction …...............................................................................................................9
2.2 Theory and Hypotheses ...............................................................................................14
2.2.1 Firm Local Experience and New-Product Development .............................20
2.2.2 Firm Distal Experience and New-Product Developments ...........................22
2.2.3 Directors’ Local Experience and New-product developments ....................25
2.2.4 Directors’ Distal Experience and New-Product Developments ...................27
2.2.5 Moderating Effect of Directors’ Local Experience on Firm Local
Experience ............................................................................................................30
2.2.6 Moderating Effect of Directors’ Distal Experience on Firm Distal
Experience .............................................................................................................31
2.3 Methods …...................................................................................................................34
2.3.1 Empirical Context ........................................................................................34
2.3.2 Data and Sample …......................................................................................35
2.3.3 Measures ......................................................................................................36
2.3.4 Analysis ........................................................................................................44
2.4 Results .........................................................................................................................48
2.4.1 Firm Local Experience, Firm Distal Experience, and New-Drug
Developments: Hypotheses 1 and 23.6 Discussion ..............................................51
2.4.2 Directors’ Local Experience, Directors’ Distal Experience, and
New-Drug Developments: Hypotheses 3 and 4 ....................................................55
2.4.3 Moderating Effect of Directors’ Experience on the Relationship
Between Firm Experience and New-Drug Developments: Hypotheses
5 and 6 ...................................................................................................................55
2.4.4 Robustness tests ...........................................................................................60

v

2.5 Discussion ...................................................................................................................67
2.5.1 Implications for Theory and Practice ...........................................................69
2.5.2 Limitations and Directions for Future Research ..........................................73

Chapter 3: Essay two: Are all “Sharks” Dangerous? New Biotechnology
Ventures and Partner Selection in R&D Alliances ….......................................................76
3.1 Introduction .................................................................................................................76
3.2 Context: Biopharmaceutical Industry .........................................................................78
3.3 Theory and Hypotheses ...............................................................................................83
3.3.1 Partner Attractiveness: Knowledge Relatedness .........................................84
3.3.2 Partner Attractiveness: Development Experience …...................................86
3.3.3 The Other Side of the Coin: Appropriation Risks .......................................88
3.3.4 Partner Attractiveness vs. Appropriation Risks: Therapeutic Area
Diversity ................................................................................................................91
3.3.5 Partner Attractiveness vs. Appropriation Risks: NBF’s Technology
Breadth .................................................................................................................93
3.4 Methods and Data .......................................................................................................97
3.4.1 Data and Sample ..........................................................................................97
3.4.2 Measures ......................................................................................................99
3.4.3 Analysis ......................................................................................................107
3.5 Results …...................................................................................................................108
3.5.1 Effects of Technological Relatedness and Development Experience
on the Likelihood of Alliance Formation: Hypotheses 1 and 2 …......................111
3.5.2 The Moderating Effect of Pharma Therapeutic Area Diversity and
the NBF’s Technology Breadth on Technological Relatedness and
Development Experience: Hypotheses 3a, 3b, 4a, and 4b ..................................114
3.5.3 Robustness Test ….....................................................................................117
3.6 Discussion ….............................................................................................................123
3.6.1 Implications for Theory and Research .......................................................127
3.6.2 Limitations and Directions for Future Research ........................................129

Chapter 4: Conclusions ...................................................................................................133
4.1 Summary of Findings ................................................................................................133
4.2 New Directions for Future Research .........................................................................135

References .......................................................................................................................139

vi

List of Tables

Table 2.1: Descriptive Statistics and Correlations (Essay 1) ............................................49

Table 2.2: Negative Binomial Estimation of the Number of New-Drug
Developments (Essay 1) …...............................................................................................52

Table 2.3: Robustness Test I (Essay 1) ............................................................................63

Table 2.4: Robustness Test II (Essay 1) ...........................................................................66

Table 3.1: Descriptive Statistics and Correlations (Essay 2) ..........................................110

Table 3.2: Estimation of the Likelihood of Alliance Formation (Essay 2) .....................113

Table 3.3: Robustness Test: Selected Sample (Essay 2) .................................................122

vii

List of Figures

Figure 2.1: Sources of Experience for New-Product Developments in a given
Technological Market .......................................................................................................19

Figure 2.2: Effect of Firm Local Experience on the Number of New Drug
Developments in a Local Therapeutic Area ………….……………….............................53

Figure 2.3: Effect of Firm Distal Experience on the Number of New Drug
Developments in a Local Therapeutic Area ……….………………….............................54

Figure 2.4: Effect of Firm Local Experience on the Number of New Drug
Developments in a Local Therapeutic Area, for Different Levels of Directors’
Local Experience ..............................................................................................................57

Figure 2.5: Effect of Firm Distal Experience on the Number of New Drug
Developments in a Local Therapeutic Area, for Different Levels of Directors’
Distal Experience ..........................................................................................................…59

Figure 3.1: Likelihood of Alliance Formation ..................................................................96

viii

ABSTRACT
In this dissertation I discuss two empirical studies that examine how different
types of firms in the biopharmaceutical industry approach different types of challenges
posed by radical and uncertain technological change. In the first empirical essay I explore
the role of different sources of experience in understanding incumbent pharmaceutical
firms’ decisions to develop new drugs. In the second empirical essay I explore how
emerging new biotechnology ventures make alliance partner selection decisions as a
function of both partner attractiveness and the risks of appropriation that arise from
establishing alliances with incumbent pharmaceuticals.
In the first essay I examine the effects of different types of experience on the
number of new products that a pharmaceutical firm develops for specific therapeutic
areas. I focus on two sources of experience: a firm’s internal experience and the
experience of the members of the board of directors. First, I find that a firm’s internal
experience, which arises from prior new-product developments, has a curvilinear effect
on the extent of new-product developments for a specific therapeutic area. Second, I
provide evidence that the extent of new-product developments is explained by the
directors’ experience gained from participating in the new-product development activities
of other organizations. Finally, I provide evidence that the directors’ experience shapes
the way in which the firm’s internal experience affects the extent of new-product
developments for a specific therapeutic area. These findings suggest that directors’
experience may help firms overcome the constraints they face when trying to exploit their
internal experience through new product developments.

ix

In the second essay I explore how new biotechnology firms (NBFs) select
pharmaceutical firms as R&D allies as a function of partner attractiveness and
appropriation risks. I find that NBFs are more likely to ally with pharmaceutical firms
that have the following two capabilities: (1) the ability to understand the NBF’s
technology (technological relatedness), and (2) strong development competences. Yet, I
provide evidence showing that these positive effects of technological relatedness and a
pharmaceutical firm’s development experience on the likelihood of establishing an R&D
alliance are negatively moderated by the pharmaceutical firm’s therapeutic area diversity
and the NBF’s technology breadth. These findings suggest that NBFs see pharmaceutical
firms’ development experience and technological relatedness as increasing appropriation
risks (rather than partner attractiveness) when the NBF’s knowledge is broadly applicable
or when the pharmaceutical firm is highly diversified across many therapeutic domains.

1

CHAPTER 1
INTRODUCTION
1.1 Innovation in the Biopharmaceutical Industry
Prior research provides compelling evidence that the growth, performance and
even survival of a firm depends significantly on its ability to innovate and introduce new
products (Banbury & Mitchell, 1995; Chaney & Devinney, 1992; Rosenkopf & Nerkar,
2001). Hence, firms need to develop specific routines and innovation capabilities that will
allow them to succeed in the innovation arena. Yet, as highlighted by the literature on
innovation, technological environments are far from stable, and disruptive events in the
form of radical technological innovation might dramatically alter the competitive
dynamics, the nature of innovation, and the challenges that firms face in a given industry
(Hill & Rothaermel, 2003). This is particularly true of the biopharmaceutical industry.
In the recent past, dramatic scientific advances in molecular biology have
provided a better understanding of the ‘mechanisms of action’ of many drug compounds.
The combined effect of these advances has given rise to a radically new approach to
innovation (i.e. biotechnology) that has created entry opportunities for innovative firms
into this industry. The new biotechnology logic is grounded in a new scientific base
(molecular biology) that is significantly different from the knowledge base that was
traditionally used by pharmaceuticals (organic chemistry). Firms with innovation
capabilities based on chemical synthesis can lose a significant portion of their skills when
attempting to transition to the emerging framework of drug discovery and development
based on molecular biology (Rothaermel, 2001). Thus, the emergence of “biotechnology”

2

is widely accepted as a radical process innovation in the way drugs are discovered and
developed, and is hence viewed as competence-destroying for incumbent companies
(Stuart, Hoang, & Hybels, 1999).
As a result of this process of radical change driven by the emergence of the
biotechnology logic, two very distinct types of companies coexist in the
biopharmaceutical industry (Rothaermel & Boeker, 2008). The first group of companies
comprises the traditional pharmaceutical firm whose ability to innovate is affected by the
difficulties it faces in adapting to the new logic of biotechnology. These firms are still
able to rely on their long-standing routines and competencies to manage new drug
developments and navigate through the regulatory process, as well as on their strong
marketing skills and distribution capabilities. However, they lack the required innovation
skills for basic research and drug discovery for the new biotechnology logic, which
reduces their overall ability to keep up with the innovation pace of new entrants. The
second type of firms represents the newer incumbents into this industry—new
biotechnology firms (NBFs) —that have emerged dramatically and with a very different
set of skills and competences. These entrepreneurial firms possess strong research and
drug discovery competences suited for the new biotechnology logic, but they often lack
the required resources for the scale-dependent development stage.
Prior research and conventional wisdom suggests that when such a technological
shift is strong enough, new entrants will rise to market dominance because of the inability
of incumbents to embrace the new technology (Hill & Rothaermel, 2003). However, so
far, a more complex scenario in the biopharmaceutical industry can be observed, where

3

both types of firms have been able to coexist without the emergence of a clear winner. It
seems to be the case that each of these two types of firms has adopted a very distinct
approach to the innovation process. Each type of organization relies on a different range
of competences and skills, and more interestingly, they both also differ in terms of the
degree and nature of the challenges they face.
The goal of this dissertation is to explore the implications of the different
challenges faced by the two different types of biopharmaceutical organizations (i.e.,
traditional pharmaceutical firms and new entrepreneurial biotechnology ventures) as a
result of process of radical technological change that has taken place in this industry. If
either of these models is to survive, it needs to address a very specific and distinct set of
challenges that jeopardize the ability of the firm to keep innovating on a consistent basis.
More specifically, the biggest challenge for established pharmaceutical firms stems from
their ability to keep on innovating and developing new drugs as a way to survive in this
newer more competitive environment, and this challenge cannot be met satisfactorily if
they just continue to rely on their prior new-product development experience.
Conversely, new biotechnology ventures, to the extent that they need to rely on
incumbents’ complementary skills and resources to transform their technologies into final
marketable drugs, face a very strong risk of getting their knowledge and capabilities
appropriated by the more established and larger organizations they seek to partner with
(Alvarez & Barney, 2001). In sum, my goal is explore the way in which different types of
organizations in the biopharmaceutical industry attempt to overcome specific challenges,
by focusing on two concrete strategic activities related to the process of innovation, i.e. a

4

firm’s decision to develop new products (essay 1) and a firm’s R&D partner selection
decision (essay 2).

1.2 Essay 1: The Effects of Firm Experience on New Product Developments: Does
Board Experience Matter?
In the first essay I explore a set of challenges that pharmaceutical firms face in the
innovation process in this industry. New product development activities are costly and
involve dealing with highly uncertain outcomes, so firms that possess prior experience in
new product developments are better able to identify unsatisfied needs in a given
technological market and come up with an original product that will address those needs
in a reliable manner (Fleming, 2001). Yet, incumbent firms just relying on their internal
experience base might face very important limitations when trying to identify potentially
attractive opportunities (Fleming, 2001; Katila & Ahuja, 2002), especially under the
emergence of a new disruptive technological logic. Therefore, one of the main challenges
that incumbent pharmaceutical firms face in this industry is to overcome the limitations
that arise when trying to exploit internal knowledge for further new drug developments.
Prior studies have suggested that firms can expand the range of available
knowledge for their innovation process, and thus come up with novel original solutions
with commercial potential, by accessing experiential knowledge that resides within other
organizations. Specifically, in this first essay I argue that pharmaceutical firms might be
able to address some of the described challenges by combining internal knowledge with
knowledge that was developed from innovation activities performed outside the

5

organization’s boundaries. Prior research has shown how firms can acquire unique
expertise and capabilities from other organizations through allying (Deeds & Hill, 1996;
Rothaermel & Deeds, 2004) or simply taking over these innovative organizations (Ahuja
& Katila, 2002; Cassiman & Veugelers, 2006). However, in this essay I focus on what I
believe is a rather inexpensive alternative mechanism overlooked by prior research, i.e.
the board of directors. I expect that the experience brought by directors will allow a focal
firm to access other organizations’ experience, and thus overcome some of the constraints
that arise from relying uniquely on the firm’s own internal experience.
Recent research has come to recognize the need for a broader examination of the
role that boards of directors play in influencing firms’ strategies (Daily, Dalton, &
Cannella, 2003; Hillman & Dalziel, 2003). Several recent studies have noted that
governance research should go beyond exploring boards’ monitoring role and also
consider whether directors have the relevant experience to enable them to advise
management effectively (Carpenter & Westphal, 2001; Hillman & Dalziel, 2003).
Consistent with this claim, in the first study I propose that directors have broader and
more complex effects on firms’ innovation activities, beyond the monitoring role
emphasized in prior research (Baysinger, Kosnik, & Turk, 1991; Hoskisson, Hitt,
Johnson, & Grossman, 2002; Kor, 2006; Zahra, 1996). Hence, by adopting a broader
theoretical perspective on boards of directors, I hope to be able to expand our
understanding of the influence that boards have on firms’ innovation-related activities.

6

1.3 Essay 2: Are all “Sharks” Dangerous? New Biotechnology Ventures and Partner
Selection in R&D Alliances
In the second essay of my dissertation I adopt the perspective of the alternative
type of firm that coexists in the biopharmaceutical industry, i.e. a new biotechnology
venture. As explained before, new biotechnology firms (NBFs) have a unique and strong
advantage over incumbent pharmaceutical firms in that they possess deep research
capabilities consistent with the demands of the new biotechnology scientific logic
(Rothaermel & Boeker, 2008). Yet, new technology-based ventures usually lack the
necessary experience and resources that are required to transform their own knowledge
into final products (Rothaermel & Deeds, 2004; Teece, 1992). As described earlier, in the
biopharmaceutical industry, incumbent pharmaceutical firms are the organizations that
possess the necessary skills and resources to successfully transform any given innovation
into a final drug. Thus, for biotechnology entrepreneurial organizations, alliances with
incumbent pharmaceutical firms represent the main way of gaining access to those skills
(Rothaermel & Deeds, 2004; Ahuja, 2000).
However, new ventures entering alliances with incumbent firms face a very
critical challenge (Katila, Rosenberger, & Eisenhardt, 2008). On the one hand, they need
the resources that these established firms provide. Yet, on the other hand, these
collaborations imply putting their technology at risk of appropriation (Alvarez & Barney,
2001). This situation has been defined as the “swimming with sharks dilemma” (Katila et
al., 2008). Understanding how new ventures deal with this tension is a fundamental
question in the field of strategy. However, in spite of its relevance, partnering decisions

7

as a function of appropriation risks are still under-researched (Katila et al., 2008; Lavie,
2007).
The goal of the second essay is to focus on how NBFs deal with this unique
challenge, which is inherently distinct from the challenges that established
pharmaceutical firms face, like the ones explored in the first essay of my dissertation.
Prior research has proposed a set of different alternatives that small ventures have to deal
with appropriation risks, such as selecting equity-based alliances (Oxley, 1997), allying
with trustworthy organizations (Gulati, 1995; Li, Eden, Hitt, & Ireland, 2008), or simply
avoiding relationships with incumbent firms (Katila et al., 2008).
Yet, in this essay I claim that these alternatives are not viable options for the
specific type of firm that I explore (i.e. new biotechnology firm) and I propose an
alternative theoretical logic based on a different set of assumptions. Basically, I propose
that all pharmaceutical firms do not represent the same degree of appropriation risks. In
other words, not every incumbent firm will be perceived by each specific new venture as
equally dangerous, or as equally attractive. Hence, I theorize that new biotechnology
ventures will ultimately be more likely to establish innovation alliances with those
“sharks” that provide the greatest potential for innovation success, while representing
sufficiently low risks of appropriation. That is, I claim that NBFs will balance the
benefits and risks of partnering with specific pharmaceuticals, where both benefits and
risk arise from the specific competences and skills that these pharmaceutical firms
provide. Thus, NBFs might be willing to avoid relationships with extremely competent

8

pharmaceuticals when such organizations are very likely to use their skills and
capabilities in a selfish and opportunistic manner.
The rest of the dissertation is organized as follows. Chapter 2 presents the first
essay of my dissertation. Chapter 3 discusses the second empirical essay. Finally, a
summary and the overall conclusions from the dissertation are provided in Chapter 4.

9

CHAPTER 2
ESSAY ONE: THE EFFECTS OF FIRM EXPERIENCE ON NEW-PRODUCT
DEVELOPMENTS: DOES BOARD EXPERIENCE MATTER?
2.1 Introduction
Prior research provides compelling evidence that the growth, performance, and
even survival of a firm depends significantly on its ability to innovate and introduce new
products (Banbury & Mitchell, 1995; Chaney & Devinney, 1992; Rosenkopf & Nerkar,
2001). Yet, despite the attractiveness of this strategy, firms face very high uncertainty in
the innovation process in that many new-product development efforts do not lead to
commercial success (Griffin, 1997; Stevens & Burley, 1997). Because new-product
development activities are costly and involve dealing with uncertain outcomes, firms are
unlikely to implement these strategies unless they possess the necessary knowledge and
capabilities to both identify unsatisfied needs in a given technological market and come
up with an original product that will address those needs (Fleming, 2002; Utterback,
1994).
Prior studies have identified the different sources from which firms can obtain this
necessary knowledge. For instance, a major stream of research has explored how firms
can access useful knowledge that arises from their own internal experience—i.e., prior
innovation activities (Katila & Ahuja, 2002; King & Tucci, 2002; Macher & Boerner;
2006; Rosenkopf & Nerkar, 2001). An alternative stream of research has explored how
firms can access useful knowledge that resides inside other organizations. Thus, a firm
can acquire unique expertise and capabilities from other organizations through allying

10

(Deeds & Hill, 1996; Rothaermel & Deeds, 2004) or simply taking over these innovative
organizations (Ahuja & Katila, 2002; Cassiman & Veugelers, 2006). However, there
might be a rather inexpensive alternative mechanism overlooked by prior research that
might allow a focal firm to access other organizations’ experience that may be useful for
new-product development activities—the board of directors.
Recent research has come to recognize the need for a broader examination of the
role that boards of directors play in influencing firms’ strategies (Daily et al., 2003;
Hillman & Dalziel, 2003). Relying solely on the monitoring role of directors is unlikely
to improve our understanding of boards’ influence on different dimensions of firm
performance (Kroll, Walters, & Wright, 2008). Several recent studies have noted that
governance research should go beyond exploring boards’ monitoring role and also
consider whether directors have the relevant experience to enable them to advise
management effectively (Carpenter & Westphal, 2001; Hillman & Dalziel, 2003).
Accordingly, recent empirical work examines the influence of directors’ experience,
gained through participation on other boards, on different firm outcomes, including
acquisition performance (Kroll et al., 2008), strategic change (Carpenter & Westphal,
2001; Westphal & Fredrickson, 2001), or on adoption of different organizational forms
(Palmer, Jennings, & Zhou, 1993). However, when it comes to the link between boards of
directors and corporate innovation, to the best of my knowledge no study has explored
the advisory role of directors in innovation activities. Prior studies of innovation
outcomes have adopted an agency theory perspective and hence have examined boards as
monitoring organizations (Baysinger et al., 1991; Hoskisson et al., 2002; Kor, 2006;

11

Zahra, 1996). The characteristic most often examined in this tradition is board
independence, with the majority of the studies finding a negative relationship between the
proportion of independent directors and corporate innovation (Baysinger et al., 1991;
Kor, 2006; Zahra, 1996)—a finding that is at odds with basic agency theory predictions.
In this study, my basic argument is that directors’ experience across various
technological markets obtained from their participation in the activities of other
organizations will have an impact on a focal firm’s decision to develop new products for
specific technological markets. I believe that directors are able to develop unique
experience from their involvement on other boards, and thus in other firms’ innovation
activities. Through participating in other boards’ innovation-based strategies, directors
are likely to obtain unique knowledge about unsatisfied needs in specific technological
markets, and they can hence help the focal firm to find ways to address these needs by
developing new products targeted at that technological market. In addition, directors with
prior experience in specific technological markets might also possess first-hand
information about non-technological aspects such as distributors, competitors, or the
regulatory environment in that specific domain, which will increase the reliability and
reduce the uncertainty surrounding new-product developments for that market. In sum, I
argue that directors represent an alternative source of know-how and expertise that is
conceptually distinct from the focal firm’s internal experience, and that this alternative
source of expertise will have a substantial impact on the firm’s innovation efforts.
The reported lack of research on the advisory role of directors in innovation
activities is especially noteworthy for several reasons. First, if directors represent a viable

12

alternative mechanism through which a focal firm can access other organizations’
experiences, one could argue that they bring knowledge resources that may be
significantly less costly and more certain than the knowledge obtained from other sources
(such as acquisitions or alliances). If that is the case, firms do have at their disposal a
powerful alternative device through which they can reduce the uncertainty of new-
product development activities. Strategic alliances and corporate acquisitions might
provide access to a richer (and broader) set of capabilities that reside in other
organizations, yet these strategies are especially costly, quite uncertain, and irreversible
in many instances. Therefore, in this study, rather than proposing that the board of
directors will substitute for other sources of external knowledge, I claim that the board of
directors will be an additional source of knowledge that might also influence a firm’s
decision to start new-product developments.
A second reason I believe that exploring the advisory role of directors for
innovation activities is relevant for strategy research is the apparent “disconnect”
between practitioners’ common beliefs and current academic research. A common
understanding among firms participating in innovation-intensive industries is that
directors play a relevant role in innovation activities. For instance, it is quite common to
read press releases such as the ones provided in the next two examples:
“We are delighted that Dr. Denis Wade is joining ChemGenex, and look forward
to sharing his substantial experience,” said Dr. Greg Coller CEO of ChemGenex
Pharmaceuticals. “Through his extensive experience with Johnson & Johnson and other
companies, Denis has successfully developed and commercialized a range of products.”

13

Appointment of Dr. Denis Wade to ChemGenex Pharmaceuticals’ Board of Directors,
December 20, 2006.
Herbert J. Conrad became the sixth director on the board of Sapphire
Therapeutics in late 2005. Mr. Conrad has served as U.S. president of Roche
Pharmaceuticals as well as on the board of directors. He has also served on the boards
of Dura, UroCor, and Sicor, and as chairman of the board for GenVec and Bone Care
International. Mr. Conrad is currently on the Boards of Savient Pharmaceuticals,
Celldex Therapeutics, and Symphony Evolution. “We welcome Herb, with his more than
40 years of industry experience leading large and small pharmaceutical companies, as
an important addition to our board,” said Gary C. Cupit, PharmD, president and chief
executive officer of Sapphire.
Thus, among practitioners it is widely believed that directors are often appointed
for their ability to provide valuable strategic advice, based on their experience and prior
innovation activities in other firms of the industry. Yet, as described before, innovation
research looking at boards has adopted a rather narrow approach in that it has mainly
focused on its monitoring role. Hence, I believe that this study, in addition to developing
a theoretical perspective for the link between directors and innovation, will also serve to
resolve this apparent disconnect between practitioners’ beliefs and academic research.
In sum, in this study I aim to explore how directors’ experience independently,
and through their interaction with a firm’s internal experience, will influence the firm’s
decision to develop new products for specific technological markets. The rest of the essay
is organized as follows. First, I develop the theory and hypotheses. Second, I describe the

14

empirical context in which I study new-product development activities. Third, I describe
data, measures, analytic methods, and results. Finally, I conclude with the study’s
contributions for theory and practice, and I identify limitations of the study and discuss
directions for future research.

2.2 Theory and Hypotheses
I examine how a firm’s decision to develop new products for specific
technological markets is determined by different types of experience available to the
firm. Consistent with prior research, I rely on the assumption that experience is
developed from accumulated learning (Macher & Boerner, 2006; Nerkar & Roberts,
2004). Thus, I assume that the type of knowledge that is required for new-product
developments for specific technological markets will grow from the experience of
actually developing new products for these markets. One main reason is that prior
development experience provides the technological base from which a firm can draw for
future development projects (Katila & Ahuja, 2002; Nerkar & Roberts, 2004; Rosenkopf
& Nerkar, 2001). In addition, previous development activities provide useful information
about non-technological aspects as well (e.g., regulatory environment or competing
products in specific technological markets), which also increases the overall reliability
and reduces the uncertainty of future innovation activities.
Therefore, I claim that the experience that arises from prior development activities
will have a significant impact on firms’ decisions to develop new products for specific
technological markets. Specifically, I propose that this experience can arise from

15

different sources, depending on where the experiential learning (prior development
efforts) took place. In this study I differentiate among four types of experience available
to a focal firm by looking at two different dimensions that respond to the question of
where experiential learning occurred: an organizational dimension (prior development
efforts performed inside vs. outside the focal firm) and a technological market dimension
(prior development efforts performed in the same vs. a different technological market).
First, I argue that the nature of the knowledge that arises from prior development
efforts will be distinct depending upon the organization in which the experiential learning
took place (i.e., inside vs. outside the focal firm). Experiential learning is contingent upon
the contextual factors and the environment in which that learning takes place (March,
1991; Levitt & March, 1988). Thus, similar innovation activities performed inside two
different organizations—i.e., with different systems, organizational resources, and
corporate cultures—will lead to different learning experiences and thus to distinct stocks
of knowledge (Levitt & March, 1988; Nelson & Winter, 1982). Therefore, when facing
future development projects, a firm can obviously leverage its internal experience
obtained from its own prior product developments (Katila & Ahuja, 2002; Nerkar &
Roberts, 2004). However, a firm can also leverage other firms’ prior experience, which
arises from a different set of innovation efforts, and therefore provides a different
perspective and different solutions to similar problems. For instance, Rothaermel and
Deeds (2004) and Deeds and Hill (1996) show how firms that access alliance partners’
unique experience are more successful in new-product development activities. Similarly,
Ahuja and Katila (2001) provide evidence that technological acquisitions allow firms to

16

improve their innovation activities by accessing other organizations’ experience. In this
study, I claim that besides strategic alliances and corporate acquisitions, there is an
alternative mechanism through which firms can access other firms’ unique experience—
the board of directors.
The decision to develop a new product is of crucial strategic importance for a firm
(King & Tucci, 2002), so it is frequently brought up at the board level for discussion. The
involvement of the board of directors in new-product-development decisions is not just
limited to preliminary discussions in order to get directors’ advice and approval before
those projects are implemented; information about its progress and the challenges faced is
also likely to be shared inside the board of directors as the project evolves. I rely on prior
studies and claim that directors can gain experience from a firm’s strategic activities as
long as directors are participating in those activities (Kroll et al., 2008). Therefore, I
argue that a director will accumulate experience about new-product development
strategies when that director sits on the board of a company that performs new-product
development projects. Such a director will be exposed to—and will therefore be able to
absorb—part of that firm’s experience. Hence, if a director brings into a focal firm this
unique and distinct expertise obtained outside the focal firm’s boundaries, such a director
might have a meaningful impact on the focal firm’s innovation strategies. For these
reasons, I claim that directors’ experience obtained from their involvement in other
organizations’ innovation activities will represent a relevant source of knowledge that
will affect a firm’s new-product development activities.

17

Second, I argue that the experience on new-product developments will also be
distinct depending on the technological market in which that experiential learning was
obtained. As prior research suggests, new-product development experience is domain-
specific to the extent that its usefulness and applicability might vary across different
technological markets (Henderson & Cockburn, 1994; Macher & Boener, 2006). For
instance, Nerkar and Roberts (2004) show that the initial sales of a new product are
higher when the firm has prior technological experience in the same technological market
in which the new product is introduced (i.e., local knowledge), while technological
experience in other technological markets (i.e., distal knowledge) has no significant effect
on a new product’s initial sales. Thus, the usefulness and applicability of a given
knowledge stock seems to be contingent upon both the source (the original technological
market in which such experience was developed) and the destination (the target
technological market in which the firm plans to leverage such experience). In this study, I
follow a similar categorization and differentiate between local and distal experience with
respect to a given market. That is, local experience in a given technological market grows
from experience in new-product developments for that same technological market
(Nerkar & Roberts, 2004), whereas distal experience in a given technological market is
the accumulated experience of new-product developments in all other technological
markets beyond the focal one (Nerkar & Roberts, 2004).
In sum, in order to understand what drives a firm’s decision to develop a new
product for a specific technological market, I look at four different types of experience as
a function of the two dimensions proposed. First, I differentiate between experience that

18

grows from prior development efforts inside the focal firm and experience that grows
from prior development efforts outside the focal firm (which resides in the focal firm’s
directors). Second, I differentiate between experience that grows from prior development
efforts in the same or in a different technological market. Exploring all possible
combinations across these two dimensions leads to four specific types of experience: firm
local experience, firm distal experience, directors’ local experience, and directors’ distal
experience (see figure 2.1). Next, I explore how each of these four types of experience
will affect the extent of new-product development for a specific technological market.

19

Directors’ distal
experience
Firm distal
experience
Directors’ local
experience
Firm local
experience
External Internal
Different
Same
Technological
Market Source
Organizational
Source
Directors’ distal
experience
Firm distal
experience
Directors’ local
experience
Firm local
experience
External Internal
Different
Same
Technological
Market Source
Organizational
Source

FIGURE 2.1. Sources of Experience for New-Product Developments in a given
Technological Market

20

2.2.1 Firm Local Experience and New-Product Development
I argue that a firm that has already developed new products (innovated) for a focal
technological market (local experience) is more likely to start new product developments
for that same market for two main reasons. First, firms that repeatedly develop products
for the same technological market are able to develop efficient organizational routines
and problem-solving strategies that generate some degree of path-dependency (Katila &
Ahuja, 2002; Levitt & March, 1988). This leads to a strong inertial behavior that
increases the firm’s motivation to keep relying on these same routines, which are specific
for that focal technological market (King & Tucci, 2002). Thus, as a firm develops new
products for a local market, its inertial behavior makes it more likely that such a firm will
continue to focus on that same technological market in the near future (King &Tucci,
2002).
Second, firms that develop multiple products for the same technological market
will become increasingly competent and reliable in that task, which will enhance that
firm’s motivation to further innovate in that technological market. Firms with prior
technological experience in a given technological market are more likely to possess a
deeper understanding of these technologies and thus a greater ability to come up with
new applications or combinations of such knowledge in the form of new products (Katila
& Ahuja, 2002). In addition, experience in developing new products for these
technological markets makes the search for new solutions and its potential success more
predictable, making the whole innovation process more reliable (Fleming, 2001).
Moreover, firms that have already developed new products for a specific technological

21

market are more likely to have the required complementary assets, increasing the chances
of the market success of any new developments (Nerkar & Roberts, 2004). More
specifically, these firms are more likely to have first-hand information about consumers
and distributors in that technological market, as well as other types of complementary
assets such as reputation and legitimacy (with regulators and other key stakeholders) that
will improve the chances of success. In sum, to the extent that a firm’s local experience in
a focal technological market improves reliability, reduces uncertainty, and enhances the
effectiveness of its innovation routines, I expect such experience to have a positive
impact on the number of new-product developments in the same market.
However, I also believe that there is a crucial limitation that arises when a firm
has too much local experience (i.e., the firm has already developed multiple products for
that same technological market). Prior literature shows that experiential learning has
diminishing returns (Argote, 1999; Macher & Boerner, 2006), and hence subsequent
improvements based on similar knowledge stocks are more difficult and mainly
incremental. In other words, reliance on the same knowledge base might lead to
exhaustion of potentially attractive opportunities (Fleming, 2001; Katila & Ahuja, 2002).
Firms that repeatedly develop products for a given technological market relying on the
same knowledge base might become increasingly myopic (Levitt & March, 1988),
reducing the firm’s overall ability to identify other opportunities available in that
technological market (Katila & Ahuja, 2002).
Based on the preceding arguments, I expect that a firm’s local experience in a given
technological market will have a positive effect on the number of new products

22

developed for that focal market up to a point, after which “too much” experience might
actually have a negative effect on further developments. Once a firm has already relied
heavily on its prior local development experience, its ability to identify unsatisfied needs
in that market and its capacity to come up with new original products to satisfy those
needs will be hampered. Thus, I propose:
H1: A firm’s local experience in a particular technological market will have an
inverted U-shaped effect on the extent of new-product developments for that
market.

2.2.2 Firm Distal Experience and New-Product Developments
When looking at a firm’s distal experience (obtained from developing new
products for technological markets outside of the focal one), I will argue that initial
increases in this type of experience will reduce the extent of new-product developments
for a local technological market. I acknowledge, as proposed by prior research, that
diverse knowledge helps the firm to come up with new solutions through enhancing
recombinatory search (March, 1991; Nelson & Winter, 1982). So, through knowledge
spillovers, high levels of distal experience might actually help the firm to develop new
products that address the needs faced in the local technological market in a radically
different manner (Fleming & Sorenson, 2004; Henderson & Cockburn, 1994). Yet, I
argue that for initial increases of firm’s distal experience, these benefits will be
outweighed by costs and limitations.

23

First, as proposed in the previous section, as the firm develops products for distal
technological markets, it will develop organizational routines and problem-solving
strategies specific to those distal markets, which leads to a very strong path-dependent
and inertial behavior (Levitt & March, 1988). Therefore, such a firm will be more likely
to rely on these same routines in future innovation activities, which reduces the likelihood
that the firm will choose to start development projects for a different—local—market
(King & Tucci, 2002).
Second, applying distal experience in a local market (rather than in the original—
distal—market) leads to a less reliable and more uncertain outcome (Fleming, 2001; King
& Tucci, 2002). To the extent that these practices rely on highly idiosyncratic and tacit
knowledge specific to a particular market or technological domain (Macher & Boerner,
2006) for initial increases of distal experience, the advantages of experiential learning
will be substantially higher when applied in the same technological market (Fleming,
2001; Nerkar & Roberts, 2004). Macher and Boerner (2006) show that development
experience is technological market-specific, which means that the application of distal
experience in a focal technological market requires heavy adaptation and integration
efforts.
In other words, firms need a sufficiently deep understanding of this distal
experience (i.e., a sufficient level of absorptive capacity) in order to be able to apply such
diverse knowledge in another technological market in a successful and efficient manner
(Cohen & Levinthal, 1990). Thus, if we assume that firms have limited resources (e.g.,
time and capital), we would expect that they would rather focus on developing new

24

products for the original markets (from which distal experience was obtained), rather than
trying to apply such know-how in a different, and thus more costly and less certain,
market.
However, I argue that there will be a point after which further increases in a
firm’s distal experience might actually increase the extent of new-product developments
for a local technological market. Basically, I propose that for higher levels of distal
experience, the benefits of applying this distal experience into a new technological
market will outweigh the limitations identified before. Specifically, I argue that through
the accumulation of new-product development experience in distal technological markets,
firms are able to develop both depth and scope. Once scope (broad knowledge across
several technological markets) is supported by a deeper understanding of this distal
experience, the firm’s absorptive capacity to integrate diverse knowledge into a focal
market is enhanced (Katila & Ahuja, 2002; Macher & Boerner, 2006).
Thus, as distal experience increases, firms will face relatively lower integration
costs and, at the same time, a lower level of uncertainty when applying distal experience
in a local market. Moreover, as distal experience grows, the ability of the firm to leverage
complementary (non-specific) assets that are also of high value in the focal market
increases as well, allowing the firm to benefit from strong economies of scope
(Henderson & Cockburn, 1994; Nerkar & Roberts, 2004). For instance, firms will be able
to rely on the reputation and legitimacy created through new-product developments for
distal markets if they are able to successfully develop a new product for the local market
(Nerkar & Roberts, 2004). For all of these reasons, I claim that for sufficiently high levels

25

of distal experience, the extent of new-product development for a focal technological
market will be significantly enhanced.
In sum, I propose that firms need a sufficiently high level of distal experience
before having the ability and the incentives to successfully apply such experience in a
local (and thus different) technological market. In spite of the benefits that might accrue
from applying new distal experience in a focal technological market, initial increases in
distal experience will have a negative impact on the development of new products for a
local technological market. Yet, I believe that once firms have obtained a sufficiently
deep understanding of distal know-how, the benefits of applying distal experience in a
local technological market will outweigh the costs. Thus, I propose:
H2: A firm’s distal experience will have a U-shaped effect on the extent of new-
product developments for a particular technological market.

2.2.3 Directors’ Local Experience and New-product developments
I expect directors’ local experience to affect a firm’s development activities in
addition to the firm’s own local experience. To the extent that the firm’s and the
directors’ local experiences arise from development efforts performed in different
organizations, even though both refer to the same technological market, it is reasonable to
assume that they do not overlap completely. I basically propose that directors’ local
experience will have a positive effect on firms’ motivation to develop new products for a
local technological market to the extent that their experience will increase the reliability
and reduce the uncertainty of these projects. As proposed earlier for firm’s local

26

experience, I believe that directors’ experience about how to develop new products for
these technological markets will make the search for new profitable opportunities in a
given technological market, and their potential success, more predictable. Directors,
through the experience obtained from their involvement in other boards, might possess
unique knowledge about unsatisfied needs in that technological market, so they can help
the firm find ways to address these needs through developing new products for that
technological market. In addition, directors with prior experience in a local market might
also possess first-hand information about non-technological aspects such as distributors,
competitors, or the regulatory environment in that market, which increases the motivation
of the firm to allocate resources for innovation activities in that market.
When I theorized earlier about the effects of a firm’s local experience, I identified
the presence of an important limitation that firms suffer if they rely too much on local
experience, which led me to propose a curvilinear (inverted U-shaped) effect on new-
product developments. However, I expect that directors’ local experience will not suffer
from similar constraints. I believe that the risk of over-utilizing and exhausting all
possible opportunities of a given local knowledge stock will not be present when looking
at directors’ experience. Because directors bring experiences from other organizations’
development efforts, a higher degree of local experience does not necessarily imply that
the focal firm has already relied extensively in the past on that same stock of knowledge.
In other words, a high level of directors’ local experience does not imply that the focal
firm has already started multiple development projects for that local market based on the
same knowledge stock. Further, different directors may bring distal experience that is

27

acquired from different organizations. That is, to the extent that the experience base of the
board is not uniform, we can expect considerable heterogeneity in the local knowledge
stocks available to the focal firm. Therefore, the greater heterogeneity of directors’ local
experience relative to the firm’s local experience suggests that further increases in the
directors’ local experience do not necessarily imply that the exhaustion (inflexion) point
will be reached.
Thus, I believe that the limitations identified in the argumentation about a firm’s
local experience do not apply when looking at local experience brought by directors to a
focal firm. For this reason, I propose that directors’ local experience will have a linear
positive effect on a firm’s motivation to start new-product developments in a local
market.
H3: Directors’ local experience in a particular technological market will have a
positive effect on the extent of new-product developments for that market.

2.2.4 Directors’ Distal Experience and New-Product Developments
Similarly, I expect directors’ distal experience to affect a firm’s development
activities in addition to the firm’s own distal experience. As explained before, to the
extent that the firm’s and the directors’ experience arise from development efforts
performed in different organizations, even though they both might refer to the same
technological markets, I argue that they do not overlap completely. Following the same
logic proposed when looking at a firm’s distal experience, I argue that the potential

28

benefits of applying directors’ distal experience into a local technological market will be
outweighed by important costs.
As explained before, applying distal experience into a local market is more likely
to lead to a less reliable and more uncertain outcome (Fleming, 2001; King & Tucci,
2002). Basically, the advantages of experiential learning will be substantially higher
when distal experience is directly applied in the original technological market (Fleming,
2001; Nerkar & Roberts, 2004), to the extent that the application of distal experience into
a local technological market requires heavy adaptation and integration costs. That is,
firms require a sufficient level of absorptive capacity in order to be able to apply such
diverse knowledge in another technological market in a successful and efficient manner
(Cohen & Levinthal, 1990). Thus, assuming that firms have limited resources, I expect
that they would rather apply directors’ distal experience in the same—distal—market
from which that experience was developed, instead of trying to apply such know-how in
a different or local—and thus more costly and less certain—market. In sum, the high
integration and opportunity costs suggest that directors’ distal experience will have a
negative effect on new-product developments for a local technological market.
When I earlier conceptualized the role of a firm’s distal experience, I argued that
there will be a point after which further increases in the firm’s distal experience will have
a positive impact on new-product developments because the benefits will outweigh the
costs. However, I believe that such an inflexion point will not occur for directors’ distal
experience. Basically, I argue that increases in directors’ distal experience do not reduce
the integration costs based on the following logic. The absorptive capacity of a focal firm

29

to assimilate, integrate, and apply directors’ distal experience in the local market is not
necessarily enhanced as directors’ distal experience increases because those distal
developments were performed outside the focal firm. To the extent that the focal firm did
not participate in those development efforts, a greater amount of directors’ distal
experience is actually not associated with increases in the firm’s ability to understand this
(the directors’) distal knowledge and its potential usefulness for other technological
markets.
In addition, it is important to note that the “integration” costs of directors’ distal
experience are relatively higher that those of the firm’s distal experience. Directors’ distal
experience must go through two levels of adaptation rather than one: know-how must be
integrated from a different organizational context into the new organization’s context,
and it must also be adapted from a distal area into the local one. Thus, the challenges of
integrating directors’ distal experience are considerably more complex than the challenge
associated with understanding firm-specific distal experience. Further, as argued earlier,
there is likely to be some heterogeneity across directors’ experience, which can further
compound the complexities of leveraging such knowledge and adapting it to the focal
firm. Thus, different directors may bring distal experience that is acquired from not only
different technological markets but also different organizations, and to the extent that the
experience base of the board is not uniform we can expect considerable heterogeneity in
the distal knowledge stocks available to the focal firm.
In sum, to the extent that the integration costs are not reduced with increases in
directors’ distal experience, I expect that the negative main effect identified above will

30

prevail. For these reasons, I propose that directors’ distal experience will have a linear
negative effect on firm’s motivation to start new-product developments in a local market.
H4: Directors’ distal experience will have a negative effect on the extent of new-
product developments for a particular technological market.

2.2.5 Moderating Effect of Directors’ Local Experience on Firm Local Experience
As proposed before, I claim that these two sources of expertise (firm and
directors) in a local market are conceptually distinct to the extent that each was generated
in a different organization. In this section, I explore a subtler and more indirect manner in
which directors’ local experience might affect a focal firm’s new-product developments
(i.e., by moderating the effects of the firm’s local experience). As proposed in Hypothesis
1, I expect that a firm’s local experience will have a curvilinear effect on new-product
developments for a given focal technological market. Initially, experience from
developing new products increases reliability and certainty, driving further innovation,
but there is a threshold at which firms reach the exhaustion point and greater experience
(more products developed in the past) actually limits the ability and motivation of the
firm to keep innovating in that same local market. The question is whether directors’
local experience has an impact at all on this curvilinear relationship. Next, I claim that it
does, and I argue that the moderating effect of directors’ local experience is to delay the
inflexion point of this curvilinear effect.
As explained before, the inflexion point arises because firms with too much local
experience might have exhausted all possible combinations of their local know-how, and

31

therefore they might have already exploited the most profitable market opportunities that
their internal experience provides. Yet, to the extent that directors bring new local
experience that was generated outside the focal firm, they might rely on this distinct
expertise and combine it with the firm’s internal local experience to identify new ways to
exploit the firm’s internal know-how in that same local technological market. Consistent
with this logic, Rosenkopf and Nerkar (2001) propose that combining knowledge
generated inside the firm with knowledge from the same domain generated outside the
firm boundaries provides the company with the opportunity to identify other relevant and
original innovations. Thus, the directors’ unique experience in the focal market might be
a useful complement to a firm’s internal experience when it is facing the reported
limitations (i.e., has reached the exhaustion point). Hence, directors’ local experience will
improve the firm’s overall ability to further identify profitable opportunities in that same
local market, which will delay the inflexion point identified in Hypothesis 1. Thus, I
propose:
H5: Directors’ local experience will delay the inflexion point of the inverted U-
shaped relationship between the firm’s local experience and the extent of new-
product developments for a particular technological market.

2.2.6 Moderating Effect of Directors’ Distal Experience on Firm Distal Experience
In this section I explore an additional way through which directors’ distal
experience might affect a focal firm’s new-product developments (i.e., by moderating the
effect of the firm’s distal experience). In Hypothesis 2 I claimed that a firm’s distal

32

experience will have a curvilinear effect on new-product developments. Initially, the
costs associated with applying distal experience in a focal technological market might be
too high, for two main reasons. First, firms are more likely to rely on the same routines
that have been used in the past (i.e., specific for distal markets), so the likelihood that the
firm will choose to start development projects for the local market is reduced (King &
Tucci, 2002). Second, distal experience might be more valuable and reliable when
applied in its original technological market, to the extent that the application of distal
experience in a focal technological market requires heavy adaptation and integration
efforts.
For these two reasons, I proposed that initial increases in distal experience will
have a negative effect on new-product developments for a local market. However, I also
argued that once the firm has obtained a sufficiently deep understanding of its distal
know-how, the benefits of applying distal experience in a focal technological market will
outweigh the costs. Thus, once this threshold is reached, a firm’s distal experience will
have a positive effect on new-product developments for a local technological market. As
before, the question is whether directors’ distal experience has an impact at all on this
curvilinear relationship. Again, I claim that it does, and I argue that the moderating effect
of directors’ distal experience is to bring forward the inflexion point of this curvilinear
effect.
As explained above, once firms have obtained a sufficiently deep understanding of
their distal experience, they will face relatively lower integration costs and,
simultaneously, a lower uncertainty when applying distal experience in a local market.

33

The fact that directors have expertise in distal technological markets implies that they
might be able to improve the firm’s overall ability to understand the potential that its own
distal experience has for a focal technological market. In other words, in addition to
providing the firm with basic knowledge about specific technological markets, I claim
that directors with diverse and distal experience will also provide the firm with the ability
to integrate knowledge from different domains and adapt it to a local domain. As
suggested in the argumentation leading to Hypothesis 2, I believe that once scope (broad
knowledge across several technological markets) is supported by a deeper understanding
of these distal domains, the firm’s absorptive capacity to integrate diverse knowledge into
a focal market is enhanced (Katila & Ahuja, 2002; Macher & Boerner, 2006).
Thus, as directors provide a deeper understanding of distal knowledge, firms will
face relatively lower integration costs and, at the same time, a lower level of uncertainty
when applying distal experience in a local market.
1
Hence, I propose that firm and
directors’ distal experience complement each other, so that the firm is able to reach
sooner the threshold level at which the benefits of applying distal experience for new-
product developments in a local market outweigh the costs identified above. In other
words, the integration and adaptation costs that the firm faces will be diminished as
directors’ distal experience increases. Thus, I propose:
H6: Directors’ distal experience will move forward the inflexion point of the U-
shaped relationship between the firm’s distal experience and the extent of new-
product developments for a particular technological market.

1
The strength of this effect will of course be contingent upon the overlap between the firm’s and the
directors’ distal knowledge. I control for this in the empirical analysis.

34

2.3 Methods
2.3.1 Empirical Context
Consistent with prior research that has explored new-product developments, I
study the biopharmaceutical industry for several reasons (Macher & Boerner, 2006;
Nerkar & Roberts, 2004; Rothaermel & Deeds, 2004). First, because pharmaceutical
firms are acknowledged for their intense innovative activity in terms of new drug
developments, they provide an ideal context to explore differences in innovation
strategies in terms of the technological markets for which they decide to develop new
drugs. Moreover, because of the strong regulatory environment in this industry, I am able
to follow new development efforts of pharmaceutical firms before the products (drugs)
are finally introduced into the market. Because pharmaceutical firms need to go through
several regulatory stages required by the FDA, I am able to see what (and when) drug
candidates enter these different stages of development in a precise manner. Consequently,
I am able to capture new-product development projects in their initial stage (i.e., the pre-
clinical stage).
Second, this industry provides a very clear set of differentiable technological
markets (i.e., therapeutic areas), each referring to distinct technologies, regulatory
environments, customers, and competing drugs (Macher & Boerner, 2006; Nerkar &
Roberts, 2004). Prior research has highlighted the presence of strong economies of scope
and knowledge spillovers across different therapeutic areas in this industry (Cockburn &
Henderson, 2001; Henderson & Cockburn, 1994; Macher & Boerner, 2006). That is,
although prior development experience is particularly relevant for the same therapeutic

35

area from which that experience was obtained, it also has secondary applications for other
(distal) therapeutic domains (Henderson & Cockburn, 1994; Macher & Boerner, 2006).
Third, there seems to be an accepted understanding within this industry that
directors play a relevant role in innovation activities. As evident in the examples provided
earlier, directors seem to be appointed to pharmaceutical boards because of their
experience obtained on other boards. Thus, I believe that my study will also serve to test
this frequently claimed assumption and will explore the actual influence that directors’
experience has on pharmaceutical firms’ innovation activities.

2.3.2 Data and Sample
I rely on prior research and define pharmaceutical companies as those
organizations in the biopharmaceutical industry that market human pharmaceuticals
(Rothaermel & Boeker, 2008). Basically, pharmaceutical firms include firms such as
Merck or Pfizer that possess manufacturing, marketing, and distribution capabilities, in
addition to experience in the development process (clinical trials). I obtained data on
pharmaceuticals’ drug developments from Medtrack. This database, created by Life
Science Analytics, provides comprehensive data on private and public biomedical
companies. Specifically, the database provides firm-level information on new drug-
development activities in all 17 therapeutic areas for the 1995–2004 period.
2
In addition,
this database also provides information about each pharmaceutical firm’s alliances and

2
There are 17 therapeutic areas identified by Medtrack: autoimmune and inflammation, blood and
lymphatic system, cancer, cardiovascular and circulatory, central nervous system, dental, dermatology,
digestive system, genetic diseases, infections, kidneys and genitourinary system, metabolic and
endocrinology, musculoskeletal disorders, ophthalmology and optometry, respiratory and pulmonary
system, substance abuse, and women’s health.

36

acquisitions activities. Further, I obtained information about characteristics of the board
of directors for public pharmaceutical firms from annual proxy statements provided by
the Securities and Exchange Commission (SEC). From this source, I obtained board-
specific data (board size and number of annual meetings), director-specific data (age,
tenure, and type of directorship—independent vs. insider), as well as information about
each director’s prior directorships (participation on boards of other biopharmaceutical
public companies). Finally, I obtained patent data from the National Bureau of Economic
Research (NBER) patent database (Hall, Jaffe, & Trajtenberg, 2001) and data from
Compustat for several control variables.
Because complete data on all explanatory variables for pharmaceutical firms are
only available for public companies (especially boards of directors data), the sample for
the study was selected from the universe of (173) public companies developing and
marketing human pharmaceuticals provided in Medtrack. My final sample only contained
those public companies for which I had data on all key variables, so I ended up with 123
pharmaceutical companies for the 1994-2004 period.
3

2.3.3 Measures
Dependent variable. With this measure I wanted to capture the number of new-
product developments (new drug candidates) for specific therapeutic areas. Prior research
shows that the earliest stage at which we can identify new drug development projects is

3
I performed a set of difference-of-means tests for every available explanatory variable (e.g. firm and
directors’ experience, new-product developments, firm size, alliance and acquisitions activity) between
selected and non-selected firms and found no significant differences.

37

when a drug candidate starts pre-clinical trials (Rothaermel & Deeds, 2004). Thus, in
order to capture the commencement of new drug development activities I relied on
Medtrack data that provides information about the number of drug candidates that entered
pre-clinical tests for each firm every year. In addition, I obtained information about the
therapeutic area toward which each drug candidate is targeted. Therefore, I obtained the
total number of new-drug-development projects in each of the 17 existing therapeutic
areas by each pharmaceutical firm in every year.

Because I wanted to explore the number
of new drug developments for a specific therapeutic area, I looked at all possible
therapeutic areas for which a firm could have started development projects, which gave
me 17 observations per firm-year.
In order to look at the number of new drug developments for a specific therapeutic
area, consistent with prior research I created the following count measure (Katila &
Ahuja, 2002; Rothaermel & Boeker, 2008). DRUGS
ijt
measured the total number of
drugs that company i started developing (i.e., entered the pre-clinical stage) for
therapeutic area j in year t.
Firm’s local experience. Local experience refers to a firm’s new-drug-
development experience in a local therapeutic area. For example, if the dependent
variable in a given observation referred to new drugs developed for cancer (DRUGS
ijt

was the dependent variable, where therapeutic area j refers to cancer), the measure of
firm local experience in that observation captured the firm’s experience with cancer
(therapeutic area j). In order to capture new-drug-development experience I looked at
prior new-drug developments—that is, the number of drug candidates that entered pre-

38

clinical tests in the past. Specifically, firm’s local experience was a count of new-drug-
development projects for that therapeutic area in the previous 3-year window (Nerkar &
Roberts, 2004).
4
However, because recent experience is more valuable than older
experience, I used a depreciation factor (δ) (Macher & Boerner, 2006). Thus, my measure
of firm local experience was defined as follows:
FIRM_LOCAL_EXPERIENCE
ijt
= DRUGS
ijt
+ δ · DRUGS
ijt-1
+ δ
2
· DRUGS
ijt-2
.
I used a depreciation factor (δ) of 40 percent in the econometric analysis, but I also varied
this factor from 20 to 60 percent to test its robustness, and I observed no significant
changes in the econometric results.
Firm’s distal experience. Distal experience refers to firm experience in
therapeutic areas other than the one that is reflected in the dependent variable.
Specifically, firm’s distal experience was measured as a count of new-drug developments
for all distal (other than the one reflected in the dependent variable) therapeutic areas in
the previous 3-year window (Nerkar & Roberts, 2004)
5
, discounting older experience by
the depreciation factor (δ) (Macher & Boerner, 2006):
FIRM_DISTAL_EXPERIENCE
ijt
= ∑ DRUGS
ikt
+ δ · ∑ DRUGS
ikt-1
+ δ
2
· ∑ DRUGS
ikt-
2,

4
I could not widen this window more because my data started in 1995, and therefore I would have lost
those observations for which I did not have complete data for the whole period. However, I replicated my
tests using experience measures created with drug developments in a 4-year window rather than 3 years
(losing one year of data from my final sample) and the results were the same in terms of signs and
significance levels for all independent variables.
5
As explained before, I could not increase the number of years used to create this measure because of data
limitations. However, I replicated my tests using a 4-year window, and the results were very similar and
patterns of significance did not change.

39

where k refers to each possible therapeutic area, excluding the local therapeutic area that
is captured in the dependent variable (i.e., therapeutic area j). That is, this final measure
captures the total experience across all possible areas except the local one (which was
captured in the previous variable: firm_local_experience).
Directors’ local experience. As discussed earlier, I am interested in capturing
directors’ experience in specific markets by looking at corporate board ties (experience
developed by participating on boards of firms that started new-drug developments). Yet,
it might be the case that these directors have additional experience in specific markets
because of their executive positions in other companies (not just directorships), and
therefore by just looking at board ties I may miss another relevant source of a director’s
experience. However, it is usually the case that directors of a focal firm that hold
executive positions in other companies are also board members of these other
organizations. Thus, for these directors, by looking at board ties I am also capturing the
experience that they are able to gain from being also executives of those other firms (i.e.,
this source of expertise is still captured in my measure). Hence, my proxy for directors’
experience created by looking at board ties is still valid as long as directors also hold
directorships in their original firms in which they serve as executives. In my final
database, only 2 companies had directors that held executive positions in other firms
without being on their corporate boards. Thus, for these 2 firms, my measure of directors’
experience might not be complete, because by just looking at board ties i might be
missing the additional experience that these other directors are bringing to the focal firm
obtained from their activities in these other companies where they do not hold

40

directorships. I ran robustness tests, removing these 2 firms from the final sample, in
order to be sure that my proxy for directors’ experience is reliable, and the results were
substantially the same.
6

Thus, based on this logic, I assumed that board ties is a reliable proxy for
directors’ experience, and I followed the next three steps to create the measure of
directors’ local experience. First, I identified all the directors of a focal pharmaceutical
firm in year t, and then I created a measure of director local experience in year t for each
single director:
DIRECTOR_DRUGS
dijt
= ∑ DRUGS
cjt
.
This measure was created by examining new-drug-development projects (that
have entered pre-clinical trials), for therapeutic area j, by every biopharmaceutical
company c, where director d (who sits on the board of the focal company i) held a
directorship in year t (excluding the focal biopharmaceutical company i). Second, I
created a firm-level measure by cumulating the experience scores for all of the directors
who sat on the board of the focal company i:
BOARD_DRUGS
ijt
= ∑ DIRECTOR_DRUGS
dijt
.
This measure captured the overall experience of the focal firm’s directors (for
every d) obtained in year t. Third, in order to be consistent with the measures of firm’s
internal experience described above, I assumed that experience accumulates over time, so

6
Ideally I would want to account for the experience of these directors that work for other companies but do
not sit on their boards (my board ties measure does not capture this additional experience). One possible
way to do that is by assuming that these ties (board-executive rather board-board) are comparable; yet, it is
not clear whether the experience that these executives (without board membership) obtain in their original
firm is the same type of experience that executives obtain when they also hold a position on the board. I
have nevertheless made that assumption and treated these links as comparable (therefore including these 2
firms that have these “unique ties”), and the results are the same.

41

the firm might have access to directors’ experience obtained in previous years. Therefore,
in order to capture the full extent of directors’ local experience that the focal firm has
access to, I examined directors’ experience obtained in the 3-years window, discounting
older experience by the depreciation factor (δ):
DIRECTOR_LOCAL_EXPERIENCE
ijt
= BOARD_DRUGS
ijt
+ δ · BOARD_DRUGS
ijt-1

+ δ
2
· BOARD_DRUGS
ijt-2
.
Directors’ distal experience. This measure is a combination of the measures used
to represent directors’ local experience and firm’s distal experience. The only difference
with respect to the measure of directors’ local experience that I just described is that, in
the final step, I looked at directors’ experience in distal (rather than local) therapeutic
areas. Thus, in this case, for each firm i I measured directors’ distal experience in every
therapeutic area (except the local one) in year t:
DIRECTOR_DISTAL_EXPERIENCE
ijt
= ∑ BOARD_DRUGS
ikt
+ δ · ∑
BOARD_DRUGS
ikt-1
+ δ
2
· ∑ BOARD_DRUGS
ikt-2
,
where k refers to each possible therapeutic area, excluding the local therapeutic area that
is captured in the dependent variable (i.e., therapeutic area j). Thus, the final measure
captured the added experience of a firm’s directors across all possible areas except the
local one (which is captured in the previous variable: director_local_experience).
Control variables. I included a set of control variables in order to avoid
unobserved heterogeneity bias in my estimation. First, I followed prior research and
included a set of controls that capture firms’ resources for innovation activities: total
number of patents in the previous 5 years, R&D intensity (R&D expenditures over sales),

42

lagged R&D intensity (Rothaermel & Deeds, 2004), and firm slack (cash and short-term
investments over a firm’s assets). Second, I included the natural logarithm of firm sales to
control for firm size. Third, the actual level of firm diversification across different
therapeutic domains might impact firms’ willingness to diversify or enter other domains.
Here, I relied on prior studies in the diversification literature (e.g., Miller, 2004) and
created a measure of diversity across therapeutic areas by looking at the Herfindahl index
(sum of squared shares). The Herfindahl index increases with the level of concentration
across therapeutic domains, so in order to create a measure that increases with the level of
diversity, I subtracted the Herfindahl index value from 1. Thus, Therapeutic Area
Diversity = [1 – Σ (TA
jt
× TA
jt
)], where TA
jt
is the proportion of drugs developed in the
previous 4 years (previous to year t) for therapeutic area j. Fourth, I accounted for the
overlap between directors’ and a firm’s distal experience. If each of these two sources of
experience refers to different distal technological markets, the impact of directors’ distal
experience on the extent of new-product developments beyond the impact of firm’s distal
experience might be contingent on this overlap. Hence, I created a measure that captured
the number of distal therapeutic areas in which both the firm and directors had experience
(Distal Experience Overlap) based on the following formula:
DISTAL_EXPERIENCE_OVERLAP
jt
= Σ (OVERLAP
kt
) ,
where k refers to each possible therapeutic area, excluding the local therapeutic area that
is captured in the dependent variable (i.e., therapeutic area j). OVERLAP
kt
takes a value
of 1 if both the firm and the board had experience in therapeutic area k, and 0 otherwise.

43

Next, I included the following additional controls to capture the board’s
monitoring role that has been shown to affect firms’ innovation-related decisions
(Hoskisson et al., 2002; Zahra, 1996). First, I controlled for the proportion of outside
directors and board size (total number of directors). In addition, I included a measure of
firm leverage (ratio of debt over equity) to control for the monitoring function of
debtholders. I also included an annual measure of the total number of board meetings to
control for differences across firms in terms of directors’ opportunity to influence firms’
decisions. Also, I included measures of directors’ average tenure (average number of
years that directors have been in the focal firm’s board), directors’ tenure dispersion
(average over standard deviation), directors’ average age, and directors’ age dispersion
(average age over standard deviation). These demographic characteristics were included
to control for directors’ ability and incentives to interact and share their experience
among themselves and with the focal firm’s managers.
7

Finally, because in this study I am interested in the effect of directors’ experience
on firms’ propensity to develop new drugs for specific domains, I also controlled for
other sources of knowledge identified in the prior literature as influencing firms’
innovation activities. For instance, Rothaermel and Deeds (2004) found that firms’
exploration (drug discovery and/or research) and exploitation (development,
manufacturing, and/or marketing) alliances affect firms’ new-drug-development
activities. Therefore, I included four additional controls as follows. First, local
exploration alliances measured the number of exploration (drug discovery and/or

7
In order to control for these different dimensions of group (board of directors) diversity, I relied on prior
studies in the top-management-team literature to create such measures (Carpenter & Westphal, 2001;
Hambrick, Cho, & Chen, 1996; Kor, 2006).

44

research) alliances that the focal firm had established in the previous three years in the
focal therapeutic area, while distal exploration alliances captured the number of
exploration alliances that the focal firm had established in the previous three years in all
other (distal) therapeutic areas. Similarly, local exploitation alliances and distal
exploitation alliances measured the number of exploitation (development, manufacturing,
and/or marketing) alliances established in the previous three years in the focal and distal
therapeutic areas, respectively. All alliances measures account for all possible types of
alliances that firms established in the past (i.e. equity and non-equity relationships). In
addition, prior research has provided evidence that firms can improve their innovation
activities by acquiring other organizations (Ahuja & Katila, 2002; Cassiman &
Veugelers, 2006). Hence, I also controlled for the number of acquisitions (of other
biopharmaceutical firms) performed by each firm in the previous three years.

2.3.4 Analysis
My dependent variable was a count measure capturing the total number of drugs
that a focal firm started developing in year t for a given therapeutic area. The simplest
regression model for non-negative integer-count variables with a limited range is the
Poisson regression. However, if the assumption that the variance and the mean number of
new-drug developments are equal is not met, the Poisson regression provides inconsistent
estimates (Greene, 2003). In such cases, a negative binomial regression is the preferred
estimation technique (Greene, 2003). Recent research in the area of innovation that also
examined discrete count variables (such as forward patent citations or number of

45

alliances) has in fact relied on this more general estimation model (Arora & Gambardella,
1994; Fleming & Sorenson, 2004; Rosenkopf & Almeida, 2003; Rothaermel & Boeker,
2008). Hence, I also used the negative binomial regression model to test hypotheses
because this methodology relaxes the restrictive assumption of mean and variance
equality inherent in the Poisson model and accounts for omitted variable bias while
estimating heterogeneity.
I modeled new drug developments in year t as a function of the explanatory
variables described above. It is important to acknowledge that my experience measures
may not have an immediate impact on new drug developments. That is, my dependent
variables captured new drugs that entered the pre-clinical stage in year t, although it
might take several years from a firm’s decision to enter a given therapeutic area (the
point in time when I expect my experience measures to have a greater impact) until the
time the firm actually enters pre-clinical tests (my dependent variable measure). Prior
research in this area and industry reports suggest that this gap varies between 2 and 3
years (PhRMA, 2007). Therefore, in order to account for this effect, I used independent
and control variables’ values in year t to explain new drug developments in year t + 2. I
also tried 3-year and 4-year lag times instead of the reported 2-year lag, and the results
remained substantially the same. In sum, given these constraints, I explored new-drug
developments in different therapeutic areas in the 1997–2004 period as a function of
explanatory variables in the 1995–2002 period (2-year lag).
Finally, it is important to account for the fact that in my final sample each firm
appeared in multiple observations. Therefore, firm-specific error terms could be highly

46

correlated, which implies that the assumption of independence across error terms is likely
to be violated (Greene, 2003). To deal with this problem I followed prior research (Katila
& Ahuja, 2002) and accounted for unobserved heterogeneity using a generalized-
estimating-equations (GEE) approach (Liang & Zeger, 1986).
8
The main advantage of
GEE is that it accounts for serial dependence across observations, which relaxes the
assumption of independence across residuals (Wooldridge, 2002). It is important to note
that I also included therapeutic area and year fixed effects to control for unobserved
heterogeneity at these different levels. Therefore, I tested the following model
specification:

DRUGS
ijt+2
= β
0
+ β
1
*FIRM_LOCAL_EXPERIENCE
ijt
+ β
2
*FIRM_LOCAL_EXPERIENCE
2
ijt
+
β
3
*FIRM_DISTAL_EXPERIENCE
ijt
+ β
4
*FIRM_DISTAL_EXPERIENCE
2
ijt
+
β
5
*DIRECTOR_LOCAL_EXPERIENCE
ijt
+ β
6
*DIRECTOR_DISTAL_EXPERIENCE
ijt
+
β
7
*FIRM_LOCAL_EXPERIENCE
2
ijt
*DIRECTOR_LOCAL_EXPERIENCE
ijt
+
β
8
*FIRM_DISTAL_EXPERIENCE
2
ijt
*DIRECTOR_DISTAL_EXPERIENCE
ijt
+ β
9
*CONTROLS
ijt
+ ε
ijt
.

Based on my theoretical logic, Hypothesis 1 predicts a positive coefficient for β
1

and a negative coefficient for β
2
. In addition, Hypothesis 2 predicts a negative coefficient
for β
3
and a positive coefficient for β
4
. Moreover, Hypotheses 3 and 4 predict a positive
coefficient for β
5
and a negative coefficient for β
6
, respectively. Finally, Hypotheses 5
and 6 predict a positive coefficient for both, β
7
and β
8
, respectively.

8
I also estimated a firm fixed-effects specification and found similar results (all hypotheses received
similar support except hypothesis 5 that was not significant under this model specification). Results
available upon request.

47

Hypothesis 5 predicts that as directors’ local experience increases, the inflexion
point of the curvilinear effect of firm local experience will be shifted to the right
(delayed). Without this moderating effect, the inflexion point in this curve is determined
by the following value of firm local experience: -β
1
/β
2
. By forcing directors’ local
experience to just affect the quadratic term, I can make a clear prediction about this
variable’s effect on the position of the inflexion point.
9
In the full model, the inflexion
point’s position will be determined by: [-β
1
/ (β
2
+ (β
7
× director_local_experience))], and
it will be delayed whenever its overall value is increased. Because I expect β
1
to be
positive and β
2
to be negative, the inflexion point will increase when the denominator has
a lower absolute value—that is, when β
7
is more positive. Thus, hypothesis 5 predicts that
β
7
should have a positive sign.
In addition, Hypothesis 6 predicts that as directors’ distal experience increases,
the inflexion point of the curvilinear effect of firm’s local experience will be shifted to
the left (moved forward). Without the interaction term, the inflexion point in this curve is
determined by the following value of firm distal experience: -β
3
/β
4
. By forcing directors’
distal experience to just affect the quadratic term, as before, I can make a clear prediction
about this variable’s effect on the position of the inflexion point.
10
With the full model,
the inflexion point’s position will be determined by: [-β
3
/ (β
4
+ (β
8
×
director_distal_experience))], and it will be moved forward (to the left) whenever its

9
I also allowed directors’ knowledge to moderate both the linear and the quadratic term, and the final effect
on the position of the inflexion point was the same (i.e., it was delayed). Graphical analysis and statistical
estimations for these additional tests are available upon request.
10
As before, I also allowed directors’ knowledge to moderate both the linear and the quadratic term and the
final effect on the position of the inflexion point remained the same (i.e., it was moved forward). Graphical
analysis and statistical estimations are available upon request.

48

overall value is decreased. Because I expect β
3
to be negative and β
4
to be positive, the
inflexion point will increase when the denominator has a greater absolute value—that is,
when β
8
is more positive. Thus, Hypothesis 6 predicts that β
8
should have a positive sign.

2.4 Results
Table 2.1 displays descriptive statistics and correlations for each of the variables
described in the prior section. I find that the main independent variables (firm local
experience, firm distal experience, directors’ local experience, and directors’ distal
experience) have low correlations. For example, the correlations between firm local and
distal experience, and the correlation between directors’ local and distal experience are
the highest among these four types of experience (0.30). But no other correlation between
any measure of firm and directors’ experience is greater than 0.18. This suggests that
these four measures are capturing different types of experience. Some control variables,
however, have significantly high correlations (e.g., the correlation between firm size and
distal exploitation alliances is 0.66). I tested for potential multi-collinearity problems by
calculating the variance inflation factor (VIF) for each variable. I found that no variable
or interaction term had a VIF greater than 4.1, which is below the threshold of 10
suggested as indicative of multi-collinearity problems (Belsley, Kuh, & Welsh, 1980).

49

TABLE 2.1. Descriptive Statistics and Correlations (n = 9,095)
Mean s.d. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
1
New Drug
Developments
0.23 0.97 1.00
2
Firm Local
Experience
0.29 1.04 0.39** 1.00
3
Firm Distal
Experience
4.57 7.13 0.22** 0.30** 1.00
4
Directors’
Local
Experience
0.46 1.82 0.12** 0.18** -0.03** 1.00
5
Directors’
Distal
Experience
7.43 12.4 -0.05** -0.03** -0.01 0.30** 1.00
6 Patents 56.1 153 0.19** 0.17** 0.39** -0.02 -0.04** 1.00
7
Distal
Experience
Overlap
0.72 1.25 0.06** 0.07** 0.29** 0.20** 0.60** 0.21** 1.00
8
Exploration
Alliances
Local
0.01 0.11 0.06** 0.14** 0.02* 0.06** -0.01 0.06** 0.01 1.00
9
Exploration
Alliances
Distal
0.11 0.50 0.03* 0.03** 0.14** -0.02 -0.01 0.22** 0.08** 0.05** 1.00
10
Exploitation
Alliances
Local
0.16 0.66 0.35** 0.51** 0.19** 0.12** -0.05** 0.15** 0.04** 0.15** 0.03** 1.00
11
Exploitation
Alliances
Distal
2.48 3.93 0.18** 0.22** 0.69** -0.05** -0.07** 0.41** 0.21** 0.02* 0.16** 0.21** 1.00
12 Acquisitions 0.11 0.61 0.11** 0.19** 0.44** -0.04** -0.08** 0.02 -0.05** -0.01 -0.01 0.12** 0.34** 1.00
13
Board
Meetings
7.19 3.20 0.07** 0.10** 0.23** 0.01 0.01 0.08** -0.02* 0.01 0.04** 0.09** 0.23** 0.06** 1.00
14
Board
Independence
0.78 0.12 0.05** 0.04** 0.11** 0.07** 0.16** 0.07** 0.09** 0.01 0.01 0.05** 0.14** 0.11** 0.11** 1.00
15 Board Size 7.68 2.20 0.20** 0.22** 0.52** 0.02* 0.05** 0.51** 0.18** 0.04** 0.15** 0.18** 0.48** 0.17** 0.16** 0.22** 1.00
16
Directors’
Average
Tenure
7.96 3.55 0.04** 0.04** 0.09** 0.01 0.02 0.10** 0.06** -0.01 -0.03** 0.07** 0.18** 0.09** -0.06** -0.13** 0.06** 1.00
17
Directors’
Tenure
Dispersion
0.58 0.20 0.06** 0.10** 0.23** -0.06** -0.14** 0.14** -0.03* 0.01 0.05** 0.06** 0.17** 0.11** 0.09** -0.09** 0.17** -0.09** 1.00
18
Directors’
Average Age
58.5 4.50 0.06** 0.06** 0.14** 0.09** 0.21** 0.10** 0.17** -0.02* -0.07** 0.07** 0.20** 0.08** 0.11** 0.09** 0.10** 0.41** 0.06** 1.00
19
Directors’
Age
Dispersion
0.15 0.05 -0.09** -0.09** -0.21** 0.05** 0.12** -0.22** -0.03* -0.02 -0.06** -0.10** -0.28** -0.08** -0.05** -0.02 -0.08** -0.11** -0.04** -0.12** 1.00
20
Therapeutic
Area Diversity
0.51 0.30 0.14** 0.18** 0.41** 0.04** 0.09** 0.31** 0.27** 0.03** 0.11** 0.16** 0.43** 0.16** 0.11** 0.15** 0.41** 0.12** 0.06** 0.12** -0.19** 1.00
21 R&D Intensity 14.0 85.9 0.03* -0.03* -0.06** -0.01 -0.03* -0.04** -0.06** -0.01 -0.03** -0.03** -0.07** -0.02 -0.01 0.01 -0.10** -0.08** -0.10** -0.08** -0.07** -0.07** 1.00
22
R&D Intensity
Lagged
16.5 90.8 -0.04** -0.03** -0.07** 0.01 0.02 -0.06** -0.02 -0.01 -0.03** -0.02* -0.06** -0.01 0.03** -0.01 -0.10** -0.10** -0.09** -0.07** -0.04** -0.08** 0.16** 1.00
23 Firm Size 3.44 2.61 0.23** 0.27** 0.63** -0.05** -0.12** 0.56** 0.09** 0.04** 0.15** 0.24** 0.66** 0.27** 0.20** 0.11** 0.37** 0.37** 0.27** 0.24** -0.29** 0.49** -0.19** -0.18** 1.00
24 Firm Leverage 1.58 10.3 -0.01 0.01 0.01 -0.01 -0.03** 0.02 -0.03** 0.01 0.04** -0.01 -0.01 0.02 0.01 0.13** 0.01 0.01 -0.01 -0.01 0.04** 0.06** 0.01 0.01 -0.01 1.00
25 Firm Slack 1.41 8.67 -0.02* -0.01 -0.02 0.01 0.03** -0.04** 0.07** -0.01 -0.01 -0.02 -0.05** -0.03** -0.02 0.03* -0.01 -0.01 -0.01 -0.03** -0.02 0.01 -0.01 -0.02 -0.03** 0.01 1.00
Significance levels: ** p < 0.01, * p < 0.05

50

As noted earlier, I tested the research hypotheses by estimating a set of negative binomial
regression models. These estimations are presented in Table 2.2 and are discussed next.
In model 1 I included only the control variables. I found that firms that had established
alliances in the local therapeutic area in the past were more likely to develop more drugs
for that specific local therapeutic area (the coefficient was significant only for
exploitation alliances), which is consistent with the logic proposed in this study as well as
with prior research (Rothaermel & Deeds, 2004). Conversely, firms that in the past had
established alliances in distal therapeutic areas were less likely to develop new drugs for
a local therapeutic area (the coefficient was significant only for exploration alliances),
which is also consistent with the logic proposed in this study.
In addition, I found that directors’ average tenure had a negative impact on new-
drug developments, which is consistent with findings in the top-management-team
literature, which has found a negative relationship between managers’ tenure and R&D
investment (Kor, 2006). Moreover, I found that the more diversified a biopharmaceutical
firm is, the more likely it is to develop new drugs, a finding consistent with the idea that
diversified firms might be more able to exploit stronger economies of scope in this
industry (Macher & Boerner, 2006). I found a positive effect of board size on new-drug
developments, indicating that firms with larger boards are able to take advantage of a
broader set of skills and capabilities, which ultimately improves the ability and
motivation of the firm to start new development projects. Also, I found that greater
leverage implied a lower likelihood of developing drugs for a given therapeutic area,

51

which is consistent with the idea that debt-holders will push the firm toward lower-risk
activities to the extent that debt-holders bear a relatively greater amount of risk, while the
company reaps a greater share of the benefits in the case of a successful innovation
(Zahra, 1996). Finally, I found that larger firms start more drug developments, which is
consistent with the idea that larger pharmaceutical firms are more likely to possess
specific innovation-based resources as well as broader complementary assets (Macher &
Boerner, 2006).

2.4.1 Firm Local Experience, Firm Distal Experience, and New-Drug Developments:
Hypotheses 1 and 2
Model 2 includes the linear and quadratic terms for firm local and firm distal
experience. Overall, the incremental variance explained by Model 2 over Model 1 is
significant (χ
2
= 137.9, p < 0.001). More specifically, I found that the linear term of firm
local experience was positive and significant, whereas the quadratic term was negative
and significant (β = 0.45, p < 0.001, and β = -0.03, p < 0.01, respectively). These
coefficients suggest an inverted U-shaped effect of firm local experience on the extent of
new drug developments (see Figure 2.2). In addition, I found that the linear term of firm
distal experience was negative and significant whereas the quadratic term was positive
and significant (β = -0.03, p < 0.01 and β = 0.001, p < 0.05 respectively), suggesting a U-
shaped effect of firm distal experience on new drug developments (see Figure 2.3).
Overall, these results provide strong empirical support for both Hypothesis 1 and
Hypothesis 2.

52

TABLE 2.2. Negative Binomial Estimation of the Number of New-Drug Developments

MODEL 1 MODEL 2 MODEL 3 MODEL 4
Intercept
-5.67 ***
(0.98)
-5.37 ***
(0.96)
-5.65 ***
(0.97)
-5.87 ***
(0.94)
Firm Local Experience -
0.45 ***
(0.06)
0.44 ***
(0.06)
0.49 ***
(0.06)
Firm Local Experience
2
-
-0.03 ***
(0.01)
-0.03 **
(0.01)
-0.03 ***
(0.01)
Firm Distal Experience -
-0.03 **
(0.01)
-0.03 **
(0.01)
-0.03 **
(0.01)
Firm Distal Experience
2
-
0.001 *
(0.001)
0.001 *
(0.001)
0.002 ***
(0.001)
Directors’ Local Experience - -
0.03 *
(0.01)
0.02
(0.02)
Directors’ Distal Experience - -
-0.014 *
(0.006)
-0.016 *
(0.007)
Firm Local Experience
2
x Directors’ Local Experience - - -
0.001 *
(0.001)
Firm Distal Experience
2
x Directors’ Distal Experience - - -
0.0003 ***
(0.0001)
Controls
Patents
0.002
(0.003)
0.001
(0.003)
-0.001
(0.003)
-0.001
(0.002)
Distal Experience Overlap
-0.01
(0.05)
-0.01
(0.04)
0.05
(0.05)
-0.04
(0.06)
Exploration Alliances Local
0.03
(0.16)
-0.06
(0.13)
-0.05
(0.13)
-0.07
(0.13)
Exploration Alliances Distal
-0.24 **
(0.09)
-0.16
†
(0.09)
-0.16
†
(0.09)
-0.19 *
(0.09)
Exploitation Alliances Local
0.34 ***
(0.04)
0.21 ***
(0.04)
0.20 ***
(0.04)
0.20 ***
(0.04)
Exploitation Alliances Distal
-0.01
(0.01)
0.02
(0.01)
0.02
(0.02)
0.02
(0.01)
Acquisitions
0.06
(0.06)
-0.01
(0.13)
-0.01
(0.12)
0.09
(0.09)
Board Meetings
-0.01
(0.01)
-0.01
(0.01)
0.01
(0.01)
-0.01
(0.01)
Board Independence
-0.43
(0.39)
-0.43
(0.36)
-0.40
(0.36)
-0.19
(0.33)
Board Size
0.07 *
(0.03)
0.03
(0.02)
0.04
†
(0.02)
0.04
†
(0.02)
Directors’ Average Tenure
-0.05 **
(0.02)
-0.04 *
(0.02)
-0.05 **
(0.02)
-0.05 **
(0.02)
Directors’ Tenure Dispersion
-0.10
(0.28)
-0.19
(0.28)
-0.22
(0.27)
-0.25
(0.27)
Directors’ Average Age
0.01
(0.01)
0.01
(0.01)
0.01
(0.01)
0.02
(0.01)
Directors’ Age Dispersion
0.08
(1.03)
0.12
(0.87)
0.29
(0.88)
0.39
(0.92)
Therapeutic Area Diversity
0.70 **
(0.26)
0.80 **
(0.25)
0.79 **
(0.25)
0.85 ***
(0.25)
R&D Intensity
-0.004
(0.006)
-0.005
(0.007)
-0.004
(0.007)
-0.005
(0.006)
R&D Intensity Lagged
-0.008
(0.006)
-0.001
†
(0.001)
-0.001
†
(0.001)
-0.001
†
(0.001)
Firm Size
0.18 ***
(0.03)
0.17 ***
(0.04)
0.17 ***
(0.04)
0.17 ***
(0.04)
Firm Leverage
-0.009 ***
(0.002)
-0.009 ***
(0.002)
-0.009 ***
(0.002)
-0.010 ***
(0.002)
Firm Slack
-0.001
†

(0.001)
-0.001
†

(0.001)
-0.001
†

(0.001)
-0.001
†

(0.001)
N
-2 Log Likelihood
LR (χ
2
)
9,095
4,953.5
-
9,095
4,823.6
129.9 ***
9,095
4,812.4
11.2 **
9,095
4,779.2
33.2 ***
Significance levels: ***p < 0.001, ** p < 0.01, * p < 0.05,
†
p < 0.10.
Likelihood ratio (LR) values test for the increment in the overall model fit after including additional variables.

53

0
1
2
3
4
5
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Firm Local Experience
New Drug Developments

FIGURE 2.2. Effect of Firm Local Experience on the Number of New Drug
Developments in a Local Therapeutic Area

54

0
1
2
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31
Firm Distal Experience
New Drug Developments

FIGURE 2.3. Effect of Firm Distal Experience on the Number of New Drug
Developments in a Local Therapeutic Area

55

2.4.2 Directors’ Local Experience, Directors’ Distal Experience, and New-Drug
Developments: Hypotheses 3 and 4
Model 3 adds the variables of directors’ local and distal experience. Overall, the
incremental variance explained by Model 3 over Model 2 is significant (χ
2
= 10.6, p <
0.01). More specifically, the coefficient of directors’ local experience was positive and
significant, while the coefficient of directors’ distal experience was negative and also
significant (β = 0.03, p < 0.05 and β = -0.013, p < 0.05 respectively).
11
I interpret these
results as empirical support for both Hypothesis 3 and Hypothesis 4. When I include
these variables, the coefficients for the linear and quadratic terms of firm (local and
distal) experience remain almost identical, suggesting that these four sources of
experience capture distinct effects.

2.4.3 Moderating Effect of Directors’ Experience on the Relationship Between Firm
Experience and New-Drug Developments: Hypotheses 5 and 6
Finally, Model 4 adds the two interactions between directors’ local and distal
experience and the quadratic terms of firm local and firm distal experience, respectively.
All variables were mean-centered prior to creating the interaction term in order to avoid
collinearity problems (Aiken & West, 1991). Overall, the incremental variance explained
by Model 4 over Model 3 is significant (χ
2
= 33.0, p < 0.001). First, I explored the
moderating effect of directors’ local experience on firm local experience. I found a
positive and significant interaction between directors’ local experience and the quadratic

11
I also explored a quadratic effect of both the local and distal knowledge of directors on the number of
new drug developments, and I did not find statistical support for a curvilinear relationship in either case.

56

term of firm local experience (β = 0.001, p < 0.05). As explained before, this positive
coefficient implies that the inflexion point’s position is shifted to the right (delayed).
Figure 2.4 demonstrates the overall effect of firm local experience on new drug
developments for two different values of directors’ local experience (i.e., one standard
deviation above and below the mean value). Plotted from the estimations obtained in
Model 4, the figure clearly shows the inverted U-shaped overall relationship between
firm local experience and new-drug developments. Moreover, the figure shows the
predicted moderating effect of directors’ local experience. Specifically, when directors’
local experience is high, the inflexion point takes place at a greater value of firm local
experience (1.6 products later) than when the level of directors’ local experience is low.
To the extent that both statistical and graphical analyses are consistent with my
prediction, I interpret these results as support for Hypothesis 5.

57

0
1
2
3
4
5
6
7
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
Firm Local Experience
New Drug Developments
Low Directors' Local Experience
High Directors' Local Experience

FIGURE 2.4. Effect of Firm Local Experience on the Number of New Drug
Developments in a Local Therapeutic Area, for Different Levels of Directors’ Local
Experience

58

Similarly, I explored the moderating effect of directors’ distal experience on firm
distal experience. Again, I found a positive and significant interaction between directors’
distal experience and the quadratic term of firm distal experience (β = 0.0003, p < 0.001).
As explained before, this positive coefficient implies that the inflexion point’s position is
shifted to the left (moved forward). Figure 2.5 demonstrates the overall effect of firm
distal experience on new-drug developments for two different values of directors’ distal
experience (i.e., one standard deviation above and below the mean value). Again, plotted
from the estimations obtained in Model 4, this figure clearly shows the U-shaped
relationship between firm distal experience and new-drug developments. In addition, this
figure shows that when directors’ distal experience is high, the inflexion point takes place
at a lower value of firm distal experience (4.6 products before) than when the level of
directors’ distal experience is low. As before, because both the statistical and graphical
analyses are consistent with my prediction, I interpret these results as strong support for
Hypothesis 6.

59

0
1
2
3
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31
Firm Distal Experience
New Drug Developments
Low Directors' Distal Experience
High Directors' Distal Experience

FIGURE 2.5. Effect of Firm Distal Experience on the Number of New Drug Developments
in a Local Therapeutic Area, for Different Levels of Directors’ Distal Experience

60

2.4.4 Robustness tests
In the theoretical development, I proposed that different sources of experience
will determine a firm’s motivation to allocate resources to develop new products for
specific technological markets. Thus, my previous empirical tests relied on a very basic
assumption, which is that a greater number of products developed for a single therapeutic
domain implies a proportionally greater allocation of resources. In other words, I
assumed that a firm that developed two new drugs for a given therapeutic domain in a
given year invested double the innovation effort that a firm that had started just one
development. This assumption might not be realistic at all if innovation efforts allocated
by a focal firm are not captured in a proportional fashion by the actual number of new
drugs developed. In such a case, using a dependent variable based on a simple count of
the number of new-drug developments in a single area might provide misleading results.
Ideally I would like to possess information on the actual allocation of resources
(real innovation inputs) in a given therapeutic domain in a given year, but such data were
not available. So, in order to provide a more conservative (and robust) test, I decided to
use an alternative dependent variable—i.e., a discrete variable that took a value of 1
whenever a firm started at least one development project in a given therapeutic area in a
given year, and 0 otherwise. Prior research has already explored both count and discrete
dependent variables as a way to overcome the limitations that arise from simply relying
on a count variable (Rothaermel & Boeker, 2008). Thus, I believe that finding support for
both model specifications will increase the reliability of my theoretical logic.

61

In Table 2.3 I estimate logistic regression models with this alternative—discrete—
dependent variable (Greene, 2003). As before, it is important to account for the fact that
firm-specific error terms could be highly correlated, which implies that the assumption of
independence across error terms is likely to be violated (Greene, 2003). Therefore, I also
accounted for unobserved heterogeneity using a GEE approach (Liang & Zeger, 1986).
The first model of Table 2.3 replicates the findings provided with the count dependent
variable (Model 4 in Table 2.2), while Model 2 of Table 2.3 provides the logistic
regression models in which I estimate the likelihood of developing at least one drug for a
given therapeutic area (discrete dependent variable).
The results showed the same curvilinear effects for firm local and distal
experience found before (inverted U-shaped for firm local experience and U-shaped for
firm distal experience), providing additional support for Hypotheses 1 and 2. More
specifically, I found that the linear term of firm local experience was positive and
significant, whereas the quadratic term was negative and significant (β = 0.59, p < 0.001,
and β = -0.05, p < 0.001, respectively). The linear term of firm distal experience was
negative and significant, whereas the quadratic term was positive and significant (β = -
0.02, p < 0.10, and β = 0.002, p < 0.01, respectively). However, although the coefficient
signs were as predicted, the results did not provide significant support for the main
effects of directors’ (local and distal) experience, failing to corroborate the support for
Hypotheses 3 and 4 I found earlier with the count-dependent measure. Finally, I found
that the moderating effects proposed in Hypotheses 5 and 6 were significantly supported
for the discrete dependent measure. Specifically, as before, both interaction terms were

62

positive and significant as expected (β = 0.002, p < 0.05, and β = 0.0002, p < 0.05,
respectively).

63

TABLE 2.3. Robustness Test I

MODEL 1 MODEL 2
Intercept
-5.87 ***
(0.94)
-5.38 ***
(1.08)
Firm Local Experience
0.49 ***
(0.06)
0.59 ***
(0.08)
Firm Local Experience
2
-0.03 ***
(0.01)
-0.05 ***
(0.01)
Firm Distal Experience
-0.03 **
(0.01)
-0.02 *
(0.01)
Firm Distal Experience
2
0.002 ***
(0.001)
0.002 **
(0.001)
Directors’ Local Experience
0.02
(0.02)
0.01
(0.02)
Directors’ Distal Experience
-0.016 *
(0.007)
-0.009
(0.006)
Firm Local Experience
2
x Directors’ Local Experience
0.001 *
(0.001)
0.002 *
(0.001)
Firm Distal Experience
2
x Directors’ Distal Experience
0.0003 ***
(0.0001)
0.0002 *
(0.0001)
Controls
Patents
-0.001
(0.002)
0.002
(0.002)
Distal Experience Overlap
-0.04
(0.06)
-0.03
(0.06)
Exploration Alliances Local
-0.07
(0.13)
-0.16
(0.25)
Exploration Alliances Distal
-0.19 *
(0.09)
-0.24 *
(0.09)
Exploitation Alliances Local
0.20 ***
(0.04)
0.32 ***
(0.08)
Exploitation Alliances Distal
0.02
(0.01)
0.02
(0.01)
Acquisitions
0.09
(0.09)
0.13
(0.09)
Board Meetings
-0.01
(0.01)
0.02
(0.02)
Board Independence
-0.19
(0.33)
-0.39
(0.40)
Board Size
0.04
†
(0.02)
0.03
(0.02)
Directors’ Average Tenure
-0.05 **
(0.02)
-0.04 *
(0.02)
Directors’ Tenure Dispersion
-0.25
(0.27)
-0.27
(0.30)
Directors’ Average Age
0.02
(0.01)
0.01
(0.02)
Directors’ Age Dispersion
0.39
(0.92)
0.89
(1.03)
Therapeutic Area Diversity
0.85 ***
(0.25)
0.93 ***
(0.22)
R&D Intensity
-0.005
(0.006)
-0.001
(0.006)
R&D Intensity Lagged
-0.001
†
(0.001)
-0.001
(0.001)
Firm Size
0.17 ***
(0.04)
0.17 ***
(0.04)
Firm Leverage
-0.010 ***
(0.002)
-0.011 **
(0.004)
Firm Slack
-0.001
†

(0.001)
-0.001
(0.001)
N
-2 Log Likelihood
9,095
4,779.2
9,095
4,789.8
Significance levels: ***p < 0.001, ** p < 0.01, * p < 0.05,
†
p < 0.10.
Model 1 estimates the number of new-drug developments, while Model 2 estimates the likelihood of having at
least one new-drug development, for a specific therapeutic area in year t.

64

In addition, it is important to notice that in my sample I include observations of
firms that have no experience at all (did not develop any new drug for any market in the
past three years). In my initial estimations, I relied on the assumption that firms with no
experience are also at risk of starting new drug developments, and for this reason I also
included them in my final sample. However, it might be the case that these firms have
zero experience because they decided to rely on alternative ways of introducing new
drugs into the market that does not imply internal development, so they are not really
contemplating the option of starting new product developments anymore. If that is the
case, the inclusion of these firms in my sample might lead to misleading conclusions. In
order to be sure that my results are not driven by the erroneous inclusion of these firms, I
run additional estimations in a sub-sample that only includes those firms that have
experience measures greater than zero.
In Table 2.4 I estimate negative binomial regression models on this sub-sample.
As before, it is important to account for the fact that firm-specific error terms could be
highly correlated, which implies that the assumption of independence across error terms
is likely to be violated (Greene, 2003). Therefore, I also accounted for unobserved
heterogeneity using a GEE approach (Liang & Zeger, 1986). The first model of Table 2.4
replicates the findings provided with the full sample (Model 4 in Table 2.2), while Model
2 of Table 2.4 provides the regression models with the selected sub-sample.
The results showed the same curvilinear effects for firm local and distal
experience found before (inverted U-shaped for firm local experience and U-shaped for
firm distal experience), providing additional support for Hypotheses 1 and 2. More

65

specifically, I found that the linear term of firm local experience was positive and
significant, whereas the quadratic term was negative and significant (β = 0.45, p < 0.001,
and β = -0.04, p < 0.001, respectively). The linear term of firm distal experience was
negative and significant, whereas the quadratic term was positive and significant (β = -
0.04, p < 0.01, and β = 0.002, p < 0.001, respectively). I found statistical support also for
Hypotheses 3 and 4. Specifically, directors’ local experience had a positive and
marginally significant effect (β = 0.03, p < 0.10), while directors’ distal experience had a
negative and significant effect (β = -0.025, p < 0.001). Finally, I found that the
moderating effects proposed in Hypotheses 5 and 6 were also supported with the selected
sub-sample. Specifically, as before, both interaction terms were positive and significant
as expected (β = 0.001, p < 0.10, and β = 0.0002, p < 0.001, respectively). Overall I
interpret all these additional robustness tests as reasonable additional evidence in support
of my theoretical predictions.

66

TABLE 2.4. Robustness Test II

MODEL 1 MODEL 2
Intercept
-5.87 ***
(0.94)
-5.51 ***
(1.03)
Firm Local Experience
0.49 ***
(0.06)
0.45 ***
(0.06)
Firm Local Experience
2
-0.03 ***
(0.01)
-0.04 ***
(0.01)
Firm Distal Experience
-0.03 **
(0.01)
-0.04 **
(0.01)
Firm Distal Experience
2
0.002 ***
(0.001)
0.002 ***
(0.001)
Directors’ Local Experience
0.02
(0.02)
0.03
†

(0.02)
Directors’ Distal Experience
-0.016 *
(0.007)
-0.025 ***
(0.007)
Firm Local Experience
2
x Directors’ Local Experience
0.001 *
(0.001)
0.001
†

(0.001)
Firm Distal Experience
2
x Directors’ Distal Experience
0.0003 ***
(0.0001)
0.0002 ***
(0.0001)
Controls
Patents
-0.001
(0.002)
0.001
(0.003)
Distal Experience Overlap
-0.04
(0.06)
-0.03
(0.07)
Exploration Alliances Local
-0.07
(0.13)
-0.04
(0.13)
Exploration Alliances Distal
-0.19 *
(0.09)
-0.09
(0.08)
Exploitation Alliances Local
0.20 ***
(0.04)
0.14 ***
(0.04)
Exploitation Alliances Distal
0.02
(0.01)
0.02
(0.01)
Acquisitions
0.09
(0.09)
0.08
(0.08)
Board Meetings
-0.01
(0.01)
0.01
(0.01)
Board Independence
-0.19
(0.33)
0.12
(0.45)
Board Size
0.04
†
(0.02)
0.05 *
(0.02)
Directors’ Average Tenure
-0.05 **
(0.02)
-0.02
(0.02)
Directors’ Tenure Dispersion
-0.25
(0.27)
0.13
(0.29)
Directors’ Average Age
0.02
(0.01)
0.01
(0.01)
Directors’ Age Dispersion
0.39
(0.92)
-0.24
(1.10)
Therapeutic Area Diversity
0.85 ***
(0.25)
0.71 **
(0.25)
R&D Intensity
-0.005
(0.006)
-0.001
(0.007)
R&D Intensity Lagged
-0.001
†
(0.001)
-0.002 **
(0.001)
Firm Size
0.17 ***
(0.04)
0.11 *
(0.04)
Firm Leverage
-0.010 ***
(0.002)
-0.011 ***
(0.002)
Firm Slack
-0.001
†

(0.001)
-0.001
(0.001)
N
-2 Log Likelihood
9,095
4,779.2
5,150
2,986.2
Significance levels: ***p < 0.001, ** p < 0.01, * p < 0.05,
†
p < 0.10.
Model 1 estimates the number of new-drug developments on the whole sample, while Model 2 estimates the
number of new-drug developments on the selected sub-sample.

67

2.5 Discussion
The purpose of this study was to explore how different sources of experience
available to a firm explain the extent of new-product developments. I differentiated
among four types of experience available to a focal firm by looking at two different
dimensions that capture where the experiential learning occurred—an organizational
dimension (prior development efforts performed inside vs. outside the focal firm)—and a
technological market dimension (prior development efforts performed in the same vs. in a
different technological market). Combining these two dimensions, I examined the
following four specific types of experience: firm local experience, firm distal experience,
directors’ local experience, and directors’ distal experience.
First, I hypothesized and found empirical support for an inverted U-shaped effect
of firm local experience on new-product developments (Hypothesis 1). Basically, I
argued that a firm’s local experience in a given technological market will have a positive
effect on the number of new products developed for that focal technological market, up to
a point after which “too much” experience might actually have a negative effect on
further developments. Once a firm has already relied heavily on its prior local
development experience, its ability to identify unsatisfied needs in that market and its
capacity to come up with new original products to satisfy those needs will be hampered.
Similarly, I proposed and also found empirical support for a U-shaped effect of firm
distal experience on new-product developments (Hypothesis 2). This finding supports my
theoretical argument that firms need to acquire a sufficiently high level of distal
experience before they can successfully transfer such experience into a local (and thus

68

different) technological market. In spite of the benefits that might accrue from applying
new distal experience into a focal technological market, initial increases in distal
experience have a negative impact on the development of new products for a local
technological market. Yet, once firms have obtained a sufficiently deep understanding of
distal know-how, the benefits of applying distal experience in a local technological
market outweigh the costs.
In addition I predicted independent main effects of directors’ experience on new-
product developments. First, I found empirical support for the hypothesized positive
effect of directors’ local experience on new-product developments for a local
technological market (Hypothesis 3). I also hypothesized and found support for a
negative effect of directors’ distal experience on new-product developments for a local
technological market (Hypothesis 4). This finding confirms my a priori theoretical
expectations that because the costs of integrating directors’ distal experience into a local
domain are likely to be high, and because such experience is more likely to be valuable in
its original technological domain, directors’ distal experience will reduce the extent of
new-product developments for a local market.
Finally, I examined how directors’ experience moderates the relationships
between firm experience and new developments. First, I found empirical support for the
theoretical expectation that directors’ local experience will help a firm to further exploit
its own internal local experience for a local technological domain, thereby prolonging the
positive effects of firm local experience on new-product developments (Hypothesis 5).
Similarly, I argued that directors’ distal experience will help the firm to access the

69

benefits of its own distal experience sooner. Consistent with this expectation, I found that
the inflexion point in the U-shaped relationship between firm distal experience and the
number of new products developed occurred earlier as the level of board distal
experience increased (Hypothesis 6).
Taken together, these findings are consistent with the core theoretical premise of
my study: firm experience and director experience are two distinct sources of experience,
and as such each has a distinct impact on new-product developments. In addition, these
results, especially the two moderating effects of board experience, are consistent with the
ideas that directors’ and firm experience complement each other and that directors’
experience may indeed help firms overcome two basic constraints they face when trying
to improve their ability to develop new products for specific technological markets. First,
by delaying the point at which firms reach the exhaustion threshold, directors are able to
improve the firm’s overall ability to continue leveraging its internal expertise. Second,
directors’ experience seems to improve the firm’s ability to apply distal experience in a
new domain, thereby broadening the range of innovation opportunities provided by a
diverse range of internal experiences.

2.5.1 Implications for Theory and Practice
I believe that this study makes several theoretical and empirical contributions to
the existing literature. First, it provides compelling new insights into the role that
directors play in corporate innovation activities. Consistent with recent research (Kroll et
al., 2008; Westphal & Fredrickson, 2001), and attending to recent calls in the literature

70

(Daily et al., 2003; Hillman & Dalziel, 2003), I explore the impact of directors on
specific strategic activities beyond their monitoring role. Specifically, I provide evidence
that directors’ experience in specific technological domains has a direct effect on the
number of products that a firm will start developing for specific technological markets. In
addition, I also show that directors have an indirect effect on a firm’s innovation strategy
by helping the firm to exploit its own internal experience in the form of new-product
developments. Thus, my study suggests that directors have broader and more complex
effects on firms’ innovation activities beyond the monitoring role emphasized in prior
research (Baysinger et al., 1991; Hoskisson et al., 2002; Kor, 2006; Zahra, 1996). In
conceptualizing and testing the effects of directors’ experiences, I am able to expand our
understanding of the impact that boards have on firms’ innovation-related activities.
Also, my results are quite consistent with conventional wisdom among practitioners that
directors are often appointed for their ability to provide valuable strategic advice, based
on their experience and prior innovation activities with other firms in the industry. Hence,
I believe that my study, in addition to developing a theoretical perspective for the link
between directors and innovation, also serves to reduce the current disconnect between
practitioners’ beliefs and academic research.
Second, my study contributes to the innovation literature by showing that there is
an alternative channel (beyond strategic alliances and corporate acquisitions) through
which a company can access other firms’ experience. Prior research has shown that a
firm can acquire unique expertise and capabilities from other organizations through
allying (Deeds & Hill, 1996; Rothaermel & Deeds, 2004) or by simply taking over these

71

innovative organizations (Ahuja & Katila, 2002; Cassiman & Veugelers, 2006).
However, even after controlling for these alternative mechanisms, I find that directors
still play a significant role in explaining variations in the extent of new-product
development. This finding is particularly noteworthy if I believe that directors may
represent a rather inexpensive alternative mechanism for knowledge acquisition. It is true
that strategic alliances and corporate acquisitions might provide access to a richer (and
broader) set of capabilities that reside in other organizations, but these strategies may be
rather costly, and even irreversible in some instances, and they are also quite disruptive
for a focal firm (Paruchuri, Nerkar, & Hambrick, 2006). Our study identifies an
additional source of knowledge that has been overlooked by prior research, thereby
improving our understanding about the different range of knowledge sources that firms
can access as a way to improve their innovation activities.
Finally, my study also contributes to the innovation literature by theoretically
explicating limitations that firms face when trying to exploit different types of experience
through new-product developments, as well as by showing a way in which firms can
overcome—at least to some extent—these constraints. Consistent with other studies
(Katila & Ahuja, 2002), my research also confirms that firms that rely too much on the
same knowledge base reach, at some point, the exhaustion threshold and become unable
to continue innovating from the same stock of knowledge. Similarly, consistent with prior
research (Katila & Ahuja, 2002; King & Tucci, 2002), my study suggests that firms face
important constraints when trying to apply experience from distal domains in a local one.
Firms without the required diversity and understanding of their distal experiences are less

72

likely to apply this knowledge in a local market, perhaps because the integration costs are
significantly high, and because such knowledge is more easily applicable in the original
domain.
However, going beyond prior literature, I also demonstrate that directors of a firm
can play a crucial role in helping firms overcome these limitations. First, by accessing
directors’ local experience developed in another organization, a firm may be able to delay
the exhaustion point and continue to exploit its internal local experience for a longer
period of time. In addition, by accessing directors’ distal experience, also developed in
other organizations, a firm can improve its level of understanding of its own distal
experience and thus increase its ability to integrate knowledge across technological
domains. In sum, my study suggests that by combining their own experience with
experience generated in different organizational environments, firms may be able to
overcome the specific limitations associated with leveraging internal knowledge.
From a broader theoretical standpoint, my findings can be related to research that
has focused on the exploration vs. exploitation nature of corporate innovation (Katila &
Ahuja, 2002; March, 1991; Rosenkopf & Nerkar, 2001). When a firm has experience in a
given technological domain (local experience) and uses such experience to further
innovate within the same domain, it can be viewed as following an “exploitation”
strategy. However, when the firm uses experience developed in more distant domains
(distal experience) to innovate within a local domain, it is more likely to be engaging in
“exploration.” It appears that a board’s local experience enables the firm to prolong the
exploitation period (and continue to reap the benefits of local experience), while a

73

board’s distal experience enables the firm to reap the benefits of exploration (the ability
to leverage distal experience in a local domain). Thus, my study also suggests that a
firm’s ability and motivation to rely on either one of these two strategies may be affected
by the presence of experienced boards.

2.5.2 Limitations and Directions for Future Research
In closing, I would like to acknowledge certain limitations of this study and
suggest some directions for future research. First, because I wanted to be consistent with
prior research that has examined how interlocks explain directors’ ability to play an
advisory role, I also focused on directors’ experience obtained from their involvement in
innovation activities of other corporate boards. However, there might be alternative
sources of knowledge arising from boards and directors that might be available for the
firm, beyond the ones I decided to focus on in this study. For instance, firms might also
be able to obtain useful knowledge about specific therapeutic domains from their
scientific advisory boards. Some companies (especially in the pharmaceutical industry)
have this additional corporate source of expertise, which might indeed help the firm
improve its product-development activities. In addition, it might be the case that directors
of corporate boards have additional expertise in unique therapeutic domains obtained
from their own scientific backgrounds, and this can also affect the overall knowledge
available to the firm. While exploring these alternative mechanisms was beyond the
scope of the present study (and, such data was anyway not available from the archival
sources available to me), future research that focuses on these alternative sources of

74

expertise will contribute to a more complete understanding of the range of knowledge
sources available to firms engaging in innovation activities.
Second, my theoretical predictions were based on how firms are likely to
distribute their efforts across different technological markets in the form of new-product
developments as a function of the presence of diverse types of knowledge. However, I
was unable to directly measure the specific allocation of resources across different
therapeutic areas. Our proxy measure focused on the entry into pre-clinical tests of a drug
candidate, which is the earliest stage at which the presence of a new-product development
project can be identified through archival data. Ideally, I would like to use data on the
actual allocation of research resources across different therapeutic domains (and thus
identify the motivation to develop a new product for a given domain before the product
becomes a drug candidate and reaches the pre-clinical stage). Hence, future research that
utilizes qualitative or quantitative data from primary sources may be more able to
actually capture the firm’s initial allocation of resources across different technological
domains, thus providing a more fine-grained test of my theory.
Finally, although focusing on a single industry surely increases internal validity,
the generalizability of my findings is limited. As explained before, the pharmaceutical
industry represented an ideal setting in which to empirically test my theory to the extent
that it allowed me to clearly identify technological markets (therapeutic areas) and new-
product developments at quite an early stage (pre-clinical tests). While I found that cross-
directorships (directors sharing multiple boards) are a frequent phenomenon in this
industry, this might not be the case in other industries. Moreover, while it appears that

75

directors are quite involved in decision-making processes that deal with new drug
developments in this industry, board involvement may not be as prominent in other
industries. For all of these reasons, I believe that future research is necessary to assess
whether out theoretical story is consistent across diverse industries, and if not, what the
industry-specific characteristics may be that influence the effects of directors on firms’
innovation strategies.
In conclusion, my study of how different sources of experience affect new-
product developments provides a broader insight into the factors that influence firms’
ability and motivation to develop new products for specific technological domains. In
particular, I hope that I have taken important steps in the direction of developing a greater
understanding of how boards of directors affect innovation activities.

76

CHAPTER 3
ESSAY TWO: ARE ALL “SHARKS” DANGEROUS? NEW BIOTECHNOLOGY
VENTURES AND PARTNER SELECTION IN R&D ALLIANCES
3.1 Introduction
New technology-based ventures usually lack the necessary experience and
resources that are required to transform their own knowledge into final products
(Rothaermel & Deeds, 2004; Teece, 1992). For these organizations, alliances with
incumbent organizations represent a way of gaining access to those skills (Rothaermel
&Deeds, 2004; Ahuja, 2000). However, new ventures seeking alliances with incumbent
firms face a very critical tension (Katila et al., 2008). On the one hand, they need the
resources that these established firms provide. Yet, on the other hand, these
collaborations imply putting their technology at risk of appropriation (Alvarez &Barney,
2001). This situation has been defined as the “swimming with sharks dilemma” (Katila et
al., 2008). Understanding how new ventures deal with this tension is a fundamental
question in the field of strategy. However, in spite of its relevance, partnering decisions
as a function of appropriation risks are still under-researched (Katila et al., 2008; Lavie,
2007).
Prior studies relying on a transaction-cost-economics perspective claim that firms
can reduce appropriation risks by selecting equity-based alliances (Oxley, 1997). This
logic implies that new ventures can reduce the risks of appropriation by choosing a more
hierarchical governance mode. However, in addition to the fact that equity alliances
might still not provide sufficient protection against appropriation risks in R&D

77

environments (Dutta & Weiss, 1997; Oxley & Sampson, 2004; Teece, 1986), this
approach assumes that partner-selection decisions have already been taken, overlooking
the impact of appropriation risks in decisions prior to the selection of the type of alliance
(Katila et al., 2008). Social network scholars, on the other hand, adopt a different
perspective and claim that firms can reduce appropriation risks by allying with a
trustworthy organization (defined by prior ties) (Gulati, 1995; Li at al., 2008). Thus, new
ventures might solve the “swimming with sharks” dilemma by allying with those
organizations that a firm already knows. However, this stream of research does not
address the question of whether trustworthy organizations have the resources new
ventures need, nor does it provide alternative solutions for entrepreneurial firms that may
lack such prior ties.
In an alternative approach, recent studies show how entrepreneurial ventures in
industries in which the risk of appropriation is greater (weaker intellectual property
protection) delay, or even avoid, relationships with “sharks” (Katila et al., 2008; Katila &
Mang, 2003). Therefore, the normative implication of these studies is that new ventures
should delay or avoid allying with powerful incumbent organizations if intellectual
property rights cannot be enforced efficiently in their industry. However, these studies
look at appropriation risks at the industry level, failing to account for the existing
heterogeneity across partners within the same industry. It might be the case that not every
incumbent firm represents the same threat to every new venture. Moreover, not
establishing or delaying a relationship with an incumbent company might not be a viable

78

option for many entrepreneurial ventures because these companies indeed need to access
established firms’ resources and skills.
The goal of this study is to address some of the limitations of prior literature and
to increase our theoretical and empirical understanding of how new ventures deal with
the dilemma identified above (i.e., the need to balance benefits and appropriation risks
when making R&D partner-selection decisions). Relying on the assumption that firms are
highly heterogeneous even within the same industry, I argue that not every incumbent
firm (shark) will be perceived by each specific new venture as equally dangerous, or as
equally attractive. Hence, I propose that a new venture will ultimately be more likely to
partner with the “shark” that provides the greatest potential for value creation but also
represents sufficiently low risks of appropriation.
The rest of the essay is organized as follows. First, I describe the context in which
I explore R&D partnering decisions. Second, I develop the theory and hypotheses. Third,
I describe data, measures, methods, and results. Finally, I conclude with the contributions
for theory and practice, identify limitations, and propose directions for future research.

3.2. Context: Biopharmaceutical Industry
I examine partner-selection decisions in R&D alliances between new
biotechnology firms (NBFs) and established pharmaceutical companies for the following
reasons. First, alliances between NBFs and pharmaceutical firms are very frequent
phenomena in this industry because accessing each other’s capabilities is a key to success
(Rothaermel & Boeker, 2008; Rothaermel, 2001). In these partnerships, the NBF brings a

79

new technology with some basic properties (research stage), whereas the pharmaceutical
firm is responsible for transforming that technology into a final drug with unique
therapeutic properties (development stage). On the one hand, established pharmaceutical
firms need to access the most novel technologies, which are frequently developed by
NBFs (Rothaermel & Deeds, 2004). For instance, around 70% of the most successful
drugs marketed by pharmaceutical firms in the late 1990s were actually discovered by
NBFs (Ernst & Young Biotechnology Reports). On the other hand, NBFs strongly need
the development experience in specific therapeutic areas (disease-specific) of established
pharmaceutical firms (Macher & Boerner, 2006). It is important to note that only 0.4% of
drug candidates that enter the preclinical stage, and 20% of those entering the clinical
stage, end up reaching the market (PhRMA, 2007). Moreover, even those drugs that
succeed in getting FDA approval need to go through a rather long (around 10 years) and
expensive (around $200 million) process (Cockburn & Henderson, 2001; PhRMA, 2007).
Therefore, NBFs really need to access the development experience of pharmaceutical
firms in order to maximize the chances of success.
A second reason why I selected this industry is because R&D alliances between
NBFs and pharmaceutical firms represent the type of relationship that fits into the
“swimming with sharks” dilemma. Because an NBF needs to share its knowledge with
the pharmaceutical company in order to maximize the chances of success in the
development process, it is likely to be subject to a risk of appropriation (Pisano, 1997;
Durand, Bruyaka, & Mangematin, 2008). Prior research has shown that firms establishing
R&D collaborations with technology-intensive organizations are able to improve their

80

subsequent patenting activities, and these new patenting efforts are frequently in
technology domains that are very close to those of their prior R&D partners (Dushnitsky
& Lenox, 2005a, 2005b; Mowery, Oxley, & Silverman, 1996; Sampson, 2007). This
seems to suggest that, inside R&D alliances, pharmaceutical firms have the opportunity
to absorb and benefit from the NBFs’ technology. Consistent with this evidence, Durand
et al. (2008) found that establishing alliances with pharmaceutical firms had a significant
positive effect on NBFs innovation activities, yet such alliances had no significant effect
on the actual NBF’s economic performance. This finding suggests that while NBFs may
have been able to generate superior rents through alliances with pharmaceutical firms,
they were unable to appropriate the whole value generated by their technologies. Also,
consistent with this logic, Gulati and Higgins (2003) report no main effects for the impact
of strategic alliances between big pharmaceuticals and small biotech firms on the initial
public offering (IPO) success of the later. Overall, extant research suggests that to the
extent that partners are highly exposed to each other’s knowledge in R&D collaborations,
there is a high risk of knowledge appropriation.
Moreover, the asymmetric nature of the type of R&D partnerships that I explore
(i.e., the pharmaceutical company has the needed complementary assets –development,
manufacturing, and marketing skills– to commercially exploit any appropriated
knowledge in the market) implies that the risk of appropriation is substantially higher for
the NBF. While it is true that an NBF can rely on legal instruments (patents) as a way to
protect its knowledge (Katila & Mang, 2003), a significant proportion of an NBF’s
internal know-how is tacit and therefore difficult to codify in a patent, and it is precisely

81

this type of knowledge that can be accessed by a pharmaceutical firm in an R&D
relationship (Pisano, 1997). Moreover, even patents do not seem to represent a
sufficiently strong protection mechanism against knowledge appropriation. Even though
this industry is believed to be one in which patents represent a greater level of protection
(Matraves, 1999), there is strong evidence suggesting that this protection mechanism is
not strong enough, especially for small entrepreneurial firms. Lanjouw and Schankerman
(2001) show that, in the late 90s, an infringement suit was filed for around 2.5% of all the
patents issued in this industry (4% litigation rate for patents of small biotech companies).
This implies that a significant number of firms in this industry believed, at some
moment in time, that their intellectual property was being appropriated by a competing
firm. If patents were indeed a strong protection mechanism in this industry, we would
expect that in the majority of those cases the patentee’s property rights were successfully
enforced. Yet, in only 33% of the cases between 2000 and 2003 did the patentee actually
win the case (Biotechnology Industry Report, 2005). This number is in fact misleading in
that around 95% of all cases were settled prior to court judgment (Lanjouw &
Schankerman, 2001). To the extent that only those with strong legal and financial
resources—or enough confidence about their case—pursue court judgment, these data
suggest that the actual proportion of patentees that were actually able to enforce their
patent rights is much lower than 33% (Lanjouw & Schankerman, 2001). A potential
infringer has multiple and effective defensive strategies to the charge of patent
infringement (Rockman, 2004). For example, if the potential infringer is able to show that
the number of inventors included in the patent does not really encompass all of the true

82

inventors, or that the innovation was disclosed more than one year before the patent was
filed, the patent will be declared invalid and no infringement activity can be claimed
(Rockman, 2004).
Overall, it seems that patentees face strong difficulties when trying to enforce
their intellectual property rights. In fact, these difficulties are much greater for small
entrepreneurial ventures because they lack the legal and financial resources needed to
undergo such a costly process. The costs of litigation in this industry are around $3
million for the first patent and $1 extra million for every additional patent, in addition to
an extra $2 million per patent if the patentee files a counterclaim (American Bar
Association, 2007). In addition, prior research in the law literature has shown how small
and young firms are at a significant disadvantage because their greater litigation risk is
not offset by more rapid resolution of their suits (Lanjow & Schankerman, 2004). In sum,
all of this evidence suggests that NBFs have great difficulties in trying to enforce their
intellectual property rights against powerful pharmaceutical firms, which suggests that
R&D alliances indeed seem to fit into the “swimming with sharks” dilemma.
Finally, an additional reason why I selected this industry is because NBFs have
some discretion in partner-selection decisions. Pharmaceutical firms usually compete
among themselves to access NBFs’ breakthrough technologies because the great majority
of blockbusters in this industry are discovered by NBFs (Rothaermel & Deeds, 2004).
Thus, it seems reasonable to assume that NBFs will have some level of discretion when
making partner-selection decisions. Nevertheless, it is important to clarify that I am not
suggesting that NBFs unilaterally select pharmaceutical partners. Instead, I rely on the

83

assumption that for an alliance to take place both firms must be willing to partner with
each other. I will also carefully account for the pharmaceutical firms’ preferences in the
theory and the empirical analysis.

3.3 Theory and Hypotheses
The goal of this study is to look at NBFs’ choices in partner-selection decisions as
a function of each potential pharmaceutical firm’s attractiveness and appropriation risks.
In order to identify the attractiveness and the appropriation risks that each pharmaceutical
firm represents for a given NBF, I will look at the following partner characteristics that
have been identified by the innovation literature as inherently beneficial for innovation:
technological relatedness and the pharmaceutical firm’s development experience
(Rothaermel & Boeker, 2008; Macher & Boerner, 2006; Nerkar & Roberts, 2004). In the
first section, consistent with prior research, I argue that these characteristics will increase
partner (pharmaceutical firm) attractiveness, and I therefore claim that NBFs will be
more likely to select firms with these characteristics as partners (Rothaermel & Boeker,
2008). However, in the second section I extend prior research by identifying a set of
contextual factors under which these characteristics (technological relatedness and
development experience) might actually be perceived as highly undesirable (due to
appropriation risks), which would reduce the likelihood that NBFs will select
pharmaceutical firms with these “abilities” as R&D allies.

84

3.3.1 Partner Attractiveness: Knowledge Relatedness
First, I look at the technological relatedness between an NBF’s knowledge base
and a pharmaceutical firm’s knowledge base. Prior research looking at alliance
performance and partner-selection decisions has shown that firms relying on similar
knowledge bases (i.e., possessing technological knowledge in similar technological
domains) are more likely to establish and create better-performing R&D alliances
(Rothaermel & Boeker, 2008; Lane & Lubatkin, 1998; Mowery et al., 1996). These
studies rely on an organizational learning perspective and propose that firms whose
know-how is based on similar technological domains are better able to understand each
other’s technologies, which ultimately improves their R&D collaboration. These studies
rely on the assumption that technological relatedness implies a greater absorptive
capacity—that is, the ability to assimilate and utilize each other’s know-how (Cohen &
Levinthal, 1990).
In the present study, I rely on a similar logic to look at R&D alliances established
by NBFs as a way to access pharmaceutical firms’ development capabilities. As
described earlier, in these partnerships the NBF brings a new discovery with some basic
properties (research stage), while the pharmaceutical firm is responsible for transforming
such technology into a final drug with unique therapeutic properties (development stage).
Even though the pharmaceutical company is not usually involved in the discovery
process, it is important to acknowledge that a sufficient degree of understanding of the
technology behind the discovery of the NBF is strongly beneficial in the development
process (Henderson & Cockburn, 1994). In other words, pharmaceutical firms with

85

technological experience in the scientific and disciplinary bases from which the NBF has
drawn in discovering the new technology will have a greater ability to assimilate,
understand, and extract to a greater extent the full therapeutic potential of the technology
that the NBF brings into the R&D alliance.
Therefore, I propose that NBFs will have a stronger preference for pharmaceutical
companies that possess the necessary absorptive capacity to understand their know-how,
and thus a greater ability to transform their new discovery into a successful drug (high
technological relatedness). From the perspective of the pharmaceutical firm, it is also
reasonable to expect that these organizations will prefer to ally with NBFs with whom
they share high technological relatedness. First, such alliances will allow the
pharmaceutical firm to understand to a greater extent the technology of the NBF, which
will significantly increase the chances of developing a successful drug (Rothaermel &
Boeker, 2008). Second, through R&D collaborations with new ventures with whom they
have high technological relatedness, pharmaceutical firms have a way to access the most
novel technologies in their own technological domains, and thus protect themselves
against competence-destroying cycles (Dushnitsky & Lenox, 2005a; Hill & Rothaermel,
2003). These arguments lead to my first research hypothesis:
H1: The level of technological relatedness between an NBF and a pharmaceutical
firm will have a positive effect on the likelihood that these two firms will establish
an R&D alliance.

86

3.3.2 Partner Attractiveness: Development Experience
I also look at the development experience of the pharmaceutical firm as a
predictor of R&D alliance formation. Prior research looking at the motivations that
organizations have to establish alliances has shown that firms look for partners who
provide complementary capabilities—that is, provide access to competences they lack
(Ahuja, 2000; Rothaermel & Deeds, 2004). Specifically, these studies show how
established firms provide the necessary competences for the downstream activities of the
innovation process that new technology ventures need. The theoretical arguments behind
this evidence invoke the logic of the resource-based view and suggest that these
collaborations are motivated by asset complementarities and the ability to generate rents
through economies of specialization (Teece, 1992).
In this study, I look at R&D alliances established by NBFs and pharmaceutical
firms, and in this context, as explained by prior research, strong development
competences in specific diseases or therapeutic domains are necessary in order to
transform the technology brought by the NBF into a successful new drug (Henderson &
Cockburn, 1994; Macher & Boerner, 2006). In order to identify the firms that possess
stronger development competences, prior studies have relied on the assumption that firms
with more experience in drug-development activities possess stronger competences
(Macher & Boerner, 2006; Nerkar & Roberts, 2004). The logic behind this assumption is
that these development competences are built as the firm accumulates experiential
learning and develops specific routines (Nelson & Winter, 1982). Consistent with this
logic, prior research has provided evidence that the likelihood (and also the speed) of

87

obtaining FDA approval depends upon a firm’s prior experience in developing new drugs
for a given therapeutic area (Macher & Boerner, 2006).
Therefore, to the extent that NBFs lack strong development competences and thus
require a pharmaceutical firm to take care of the development stage, I expect that NBFs
will have a preference for pharmaceutical firms with the strongest resource
complementarities—that is, with the greatest development competences (Ahuja, 2000;
Rothaermel & Deeds, 2004). Such a pharmaceutical firm will have the required know-
how in specific therapeutic domains (diseases) to extract the full therapeutic potential
from the discovery that the NBF brings into the alliance, and thus to maximize the
chances of FDA approval (Rothaermel, 2001).
From the perspective of the pharmaceutical firm, it is reasonable to expect that
pharmaceutical firms with deep development competences will also be willing to
establish R&D alliances with NBFs. Basically, these pharmaceutical companies will be
able to exploit the strong economies of scale and specialization that their deep
development competences provide (Teece, 1992; Macher & Boerner, 2006), while
leaving the more uncertain and risky task of discovering a new molecule in the hands of
these new ventures. Moreover, as explained above, around 70% of the most successful
drugs (blockbusters) marketed by pharmaceutical firms in the late 1990s were actually
discovered by NBFs (Ernst & Young Biotechnology Reports). Thus, through these
alliances, pharmaceutical firms have a way to access the most novel technologies and
transform them into highly successful drugs (Hill & Rothaermel, 2003). Thus, I propose:

88

H2: The development experience of a pharmaceutical firm will have a positive
effect on the likelihood that a given NBF and a given pharmaceutical firm will
establish an R&D alliance.

3.3.3 The Other Side of the Coin: Appropriation Risks
So far, consistent with prior research, I have focused on the benefits that an NBF
obtains from partnering with a pharmaceutical firm with high technological relatedness
and development experience (Rothaermel & Boeker, 2008; Mowery et al., 1996;
Rothaermel & Deeds, 2004). Yet, in this study I aim to extend this approach by proposing
that these same characteristics might also represent appropriation risks for the NBF. First,
a greater technological relatedness provides the pharmaceutical firm with the necessary
absorptive capacity to assimilate and privately utilize the NBF’s knowledge in ways and
through activities that might go beyond the purpose of the alliance. Similarly, strong
development competences provide the pharmaceutical firm with the ability to understand
the therapeutic potential of the NBF’s technology for diseases or treatments that may also
fall outside the scope of the alliance. Basically, both technological relatedness and
development competences provide the pharmaceutical firm with the ability to assimilate
the NBF’s know-how, identify the therapeutic potential of such technology, and rapidly
develop that technology into a marketable drug. However, the usefulness of those skills is
not just restricted to the innovation project that is performed within the boundaries of the
R&D alliance. Thus, an NBF, when allying with a pharmaceutical firm that has deep
development competences and with whom it has a high technological relatedness, faces

89

the following dilemma: is such a pharmaceutical firm going to apply its abilities to
mainly cooperate with the NBF, or is it, on the contrary, going to behave
opportunistically and use these abilities to appropriate the NBF’s knowledge and exploit
it privately?
To answer to this question, I rely on prior research that has analyzed this tension
between partners’ cooperative and competitive behavior in R&D alliances (Khanna,
Gulati, & Nohria, 1998; Khanna, 1998). Basically, the theoretical model proposed in
these studies suggests that alliance partners are more likely to depart from cooperative to
competitive behavior the greater the ratio of private to common benefits they face within
the alliance. Private benefits are those that accrue to individual firms from activities not
governed by the alliance, whereas common benefits are those that accrue collectively to
all participants (activities governed by the alliance). In order to identify the ratio of
private to common benefits that a partner faces within a given alliance, these studies
propose the concept of “relative scope.” The relative scope in an R&D alliance measures
the extent of applications of partners’ know-how in activities that fall outside the scope of
the alliance, as a proportion of the total number of applications that such knowledge stock
has. That is, a high relative scope means that the knowledge being shared inside the R&D
collaboration has multiple applications that are unrelated to the alliance purpose.
In sum, this theoretical framework suggests that if there are many opportunities to
exploit any knowledge privately through activities that fall beyond the scope of a given
R&D alliance (high relative scope), competitive rather than cooperative behavior can be
expected (Khanna et al., 1998). Under competitive behavior, the alliance suffers from

90

what has been defined as a “learning race” (Khanna et al., 1998). Both firms will be more
focused on racing to achieve a greater level of learning from each other and to identify
alternative applications of the knowledge at stake rather than on fulfilling the alliance
objectives, and this will adversely affect alliance performance. Moreover, the
“appropriated” firm will see not only the current alliance success jeopardized but also its
future ability to exploit its own knowledge in other projects (Khanna, 1998).
Thus, in this study, I rely on this theoretical framework and claim that the tension
between appropriation risks and partner attractiveness that technological relatedness and
development experience represent will tend to be resolved in favor of competitive rather
than cooperative behavior as the relative scope of the R&D alliance increases. That is,
when the pharmaceutical firm faces multiple opportunities to apply the NBF’s knowledge
in applications that go beyond the purpose of the R&D collaboration (larger relative
scope), such a pharmaceutical firm is more likely to behave opportunistically (i.e., pursue
competitive rather than cooperative behavior) and use its abilities (technological
relatedness and development competences) to appropriate the NBF’s knowledge and
exploit it privately. Thus, the larger the “relative scope” is, the more likely it is that
technological relatedness and development experience will represent relatively greater
appropriation risks. Therefore, in the next section I propose that the relative scope of the
alliance context will have a negative moderating impact on the positive effect of
technological relatedness and development experience on the likelihood of alliance
formation. Next, I propose two indicators that I believe capture the concept of relative

91

scope for a given R&D alliance in this industry: pharma therapeutic area diversity and
NBF’s technology breadth.

3.3.4 Partner Attractiveness vs. Appropriation Risks: Therapeutic Area Diversity
Khanna et al. (1998) claim that the relative scope of an alliance can be captured
by the number of alternative markets (product or geographical) in which a partner can
privately exploit any learned knowledge. In this study, I rely on a conceptual dimension
of scope similar to the one proposed in Khanna et al. (1998). Similar to Baum, Calabrese,
& Silverman (2000), I look at therapeutic areas (rather than product or geographic
markets) in which a pharmaceutical firm is active (pharma therapeutic area diversity) as
a proxy for relative scope. I basically claim that if a pharmaceutical firm has experience
in developing new drugs for many diverse therapeutic areas, such a firm will face the
opportunity to transform and exploit any absorbed knowledge into a marketable drug for
many of those alternative therapeutic markets. Prior research has highlighted the presence
of strong economies of scope and knowledge spillovers across different therapeutic areas
(Macher & Boerner, 2006; Cockburn & Henderson, 2001; Henderson & Cockburn,
1994). Specifically, these studies show how knowledge from one therapeutic area is also
potentially useful for development projects in other therapeutic domains (Macher &
Boerner, 2006; Nerkar & Roberts, 2004). In other words, new discoveries with specific
properties can be used to develop new drugs for different therapeutic treatments
(diseases), as long as the firm has the necessary understanding of the technology and the

92

required development experience in those other therapeutic areas (Macher & Boerner,
2006; Nerkar & Roberts, 2004).
Thus, I propose that a pharmaceutical firm that is active in many different
therapeutic areas (therapeutic area diversity) will face a greater relative scope, to the
extent that this represents a greater range of opportunities to privately exploit any
knowledge appropriated from an NBF (i.e., the ratio of private to common benefits is
higher) (Baum et al., 2000). Therefore, I claim that a greater therapeutic area diversity
will signal a deviance toward competitive rather than cooperative behavior (Baum et al.,
2000).
Hence, I argue that in the presence of high therapeutic area diversity (high relative
scope), the degree of appropriation risks that technological relatedness and development
experience represent will be greater, and thus NBFs will show a lower willingness to ally
with pharmaceutical firms with those abilities. First, a greater technological relatedness in
the presence of high therapeutic area diversity provides the pharmaceutical firm with both
the necessary absorptive capacity to assimilate the NBF’s knowledge and the required
development experience in specific therapeutic areas to come up with alternative uses for
the NBF’s technology that might go beyond the purpose of the alliance. Similarly, strong
development competences in the presence of therapeutic area diversity provide the
pharmaceutical firm with the ability to understand and extract the therapeutic potential of
the NBF’s technology for diseases or treatments that may also fall outside the scope of
the alliance. As a consequence, the positive effect of technological relatedness and
development experience on the likelihood of establishing an alliance predicted in

93

Hypotheses 1 and 2 will be negatively moderated by the pharmaceutical firm’s degree of
therapeutic area diversity. Therefore, I propose the next two hypotheses:
H3a: A pharmaceutical firm’s therapeutic area diversity will negatively moderate
the positive effect of technological relatedness between a given NBF and a given
pharmaceutical firm on the likelihood that these two firms will establish an R&D
alliance.
H3b: A pharmaceutical firm’s therapeutic area diversity will negatively moderate
the positive effect of a pharmaceutical firm’s development experience on the
likelihood that a given NBF and a given pharmaceutical firm will establish an
R&D alliance.

3.3.5 Partner Attractiveness vs. Appropriation Risks: NBF’s Technology Breadth
The first proposed dimension of relative scope referred to the pharmaceutical
firm’s ability to identify and exploit alternative uses of the NBF’s technology in other
therapeutic domains. In this section, I propose an additional indicator of the presence of
alternative uses for the NBF’s technology by looking not at pharma-specific
characteristics but at idiosyncratic aspects of the NBF’s technology itself. I rely on the
assumption that there will be a significant degree of heterogeneity among NBFs’
technologies, with some having very narrow applicability and others with the potential to
be broadly applicable within the context of multiple and diverse innovation projects.
Thus, I propose a second dimension of relative scope based on the NBF’s technology
breadth.

94

Basically, I argue that if the NBF’s technology relies on (and thus has links with)
a broader set of technological domains, its potential applicability in the form of
alternative innovation projects (e.g., new drugs) will be greater. Specifically, prior
research has found that increasing a firm’s technological breadth has a positive impact on
new product introductions (Katila & Ahuja, 2002). The logic behind this argument is that
knowledge that has links with multiple technological domains increases the number of a
firm’s new products through enhancing recombinatory search (Henderson & Cockburn,
1994; Nelson & Winter, 1982). There is a limit to the number of new ideas that can be
created by using the same set of knowledge elements, so greater breadth implies the
presence of new elements in the set that improve the possibilities for finding a new useful
combination (Katila & Ahuja, 2002; Kogut & Zander, 1992). In consequence, when the
NBF’s knowledge has multiple applications (high relative scope) the pharmaceutical firm
might be able to exploit in a private fashion many of those alternative uses. Thus, I claim
that a greater NBF technology breadth will signal a deviance toward competitive rather
than cooperative behavior.
Based on the preceding arguments, I claim that for broader NBF technologies,
(i.e., the alliance context is characterized by greater relative scope), the degree of
appropriation risks posed by technological relatedness and development experience will
be greater, so NBFs will show a diminished willingness to ally themselves with
pharmaceutical firms with those abilities. First, a greater technological relatedness
provides the pharmaceutical firm with the necessary ability to assimilate and identify
some of the potential applications that such a broad technology has that might not fall

95

inside the scope of the current alliance. Similarly, strong development competences
provide the pharmaceutical firm with the experience required to exploit some of the
possible alternative therapeutic uses of such a broad technology. In consequence, the
positive effect of technological relatedness and development experience on the likelihood
of establishing an alliance predicted in hypotheses 1 and 2 will be negatively moderated
by the NBF’s technology breadth. Therefore, I propose the final two hypotheses:
H4a: An NBF’s technology breadth will negatively moderate the positive effect of
technological relatedness between a given NBF and a given pharmaceutical firm
on the likelihood that the two firms will establish an R&D alliance.
H4b: An NBF’s technology breadth will negatively moderate the positive effect of
pharmaceutical firm’s development experience on the likelihood that a given NBF
and that pharmaceutical firm will establish an R&D alliance.

All of the proposed hypotheses are presented in the theoretical model shown in
Figure 3.1.

96
Technological
Relatedness
Development
Experience
Pharma Therapeutic
Area Diversity
NBF’s Technology
Breadth
+
+
–
Likelihood of R&D
Alliance Formation
–
– –
H4b H4a
H1
H2
H3a H3b

FIGURE 3.1. Likelihood of Alliance Formation

97

3.4 Methods and Data
3.4.1 Data and Sample
I obtained data from several sources. First, data on pharmaceutical firm and NBF
characteristics, as well as data on R&D alliances, were obtained from Medtrack. In
addition to other firm-level data, this database provides information about the number of
drugs developed by each pharmaceutical firm in each therapeutic area,
12
and also
information about additional development projects and their respective development
stages. Second, in order to create a measure of firms’ technological relatedness and other
technology-based variables, I rely on patent data available up to 2002 from the National
Bureau of Economic Research (NBER) patent database (Hall, Jaffe, & Trajtenberg,
2001).
13
Finally, to ensure accurateness and completeness of the data, I double-checked
the alliance data provided in Medtrack with information from SDC Thompson and all of
the relevant companies’ Web sites.
Since the explanatory variables are available only up to 2002 because of the
limitations of the patent data sources, I look at alliance formation activities between the
NBFs and pharmaceutical firms in the 2003–2007 window as a function of explanatory
variables created from data up to 2002. As in prior studies, I look at firms in the
biopharma industry that focus on human therapeutic and diagnostic applications

12
There are 17 therapeutic areas identified by Medtrack: autoimmune and inflammation, blood and
lymphatic system, cancer, cardiovascular and circulatory, central nervous system, dental, dermatology,
digestive system, genetic diseases, infections, kidneys and genitourinary system, metabolic and
endocrinology, musculoskeletal disorders, ophthalmology and optometry, respiratory and pulmonary
system, substance abuse, and women’s health.
13
Prior studies have highlighted that in this industry patents are a reliable proxy for firms’ knowledge to the
extent that companies mainly rely on this mechanism—i.e., firms in this industry self-reported that around
80% of their innovations are patented to protect their knowledge (Arundel and Kabla, 1998; Mansfield,
1986).

98

(Rothaermel & Boeker, 2008). Also consistent with prior studies, I define a
biotechnology firm as a technology-based company in the biopharmaceutical industry
with at least one patent granted (Rothaermel & Boeker, 2008). The main reason to
include only those NBFs with at least one patent is that in order to create a measure of
technological relatedness and to capture characteristics of the NBF’s technology, I need
to rely on patent data. In addition to that, there is empirical evidence suggesting that those
NBFs that have already patented proprietary technologies are the ones at real risk of
establishing an alliance (Katila et al., 2008). Thus, in order to identify the number and
characteristics of the patents that each biotechnology firm holds, I matched data from the
NBER patent database with data from Medtrack, and I ended up with a sample of 274
companies that were identified as biotechnology firms and for which I had patent data
(out of the 567 biotechnology firms identified from Medtrack that were active by 2002).
14
Next, I defined new biotechnology ventures as those biotechnology firms that were
founded after 1997, which provided a final sample of 86 NBFs.
15

With respect to pharmaceutical firms, I rely on prior research and define
pharmaceutical companies as those organizations in the biopharmaceutical industry that
market human pharmaceuticals (Rothaermel & Boeker, 2008). Basically, pharmaceutical
firms are those such as Merck or Pfizer that possess manufacturing, marketing, and

14
Those firms for which I had patent data engaged in more alliance activity than those for which patent
data were not available (1.52 and 1.28 alliances per firm during the period of study, respectively), although
the difference was not statistically significant.
15
Prior studies looking at new entrepreneurial firms relied on similar definitions of “newness” and included
new ventures whose average age was similar to or even greater than that of the firms I include in my
sample (Rothaermel and Boeker, 2008; Park, Chen, and Gallagher, 2002). In addition, I have run sensitivity
analysis with alternative definitions of new biotechnology ventures (founded after 1996 and after 1998),
and the proposed hypotheses received similar statistical support.

99

distribution capabilities, in addition to experience in the development process (clinical
trials). Because complete data on all explanatory variables for pharmaceutical firms are
only available for public companies, all pharma firms for the study were selected from
the universe of pharma public companies (115) provided in Medtrack. The final sample
included only those public companies for which data on all explanatory variables were
available, which yielded a usable sample of 92 pharmaceutical firms.
16

In sum, my final sample included 86 new biotechnology ventures and 92
pharmaceutical firms (all of them U.S.-based companies) that focus on human therapeutic
and diagnostic applications (Rothaermel & Boeker, 2008).

3.4.2 Measures
Dependent Variable. I rely on the dyad as the unit of analysis. Consistent with
prior research that has examined partnering decisions, I considered all possible
combinations between an NBF and a pharmaceutical firm (Rothermel & Boeker, 2008;
Gulati, 1995; Stuart, 1998), which encompasses a possible 7,912 dyads (86 NBFs x 92
pharmaceutical firms = 7,912). Thus, the dependent variable ALL
ij
will take a value of 1
if NBF i and pharmaceutical firm j established an R&D alliance in the 2003–2007 time
window, and 0 otherwise. Consistent with prior research, I define R&D alliances as those
with stated objectives (as reported in Medtrack) that included research and development
(Lane & Lubatkin, 1998; Oxley & Sampson, 2004). In the final sample, 51 R&D

16
Non-selected pharmaceutical firms were mainly those firms for which a reliable matching with the patent
database was not possible. Selected and non-selected public pharmaceutical firms showed no significant
differences in terms of alliance activity, drug development activity, firm size, or firm age during the study
period.

100

alliances were formed between an NBF and a pharmaceutical company (which implies
that around 0.65% of the dyads actually entered into an alliance).
17
It is important to note
that none of the 51 R&D relationships involved an equity relationship (joint venture), so
we can be confident that all partnerships are comparable in terms of the nature of the
relationship. In addition, all 51 R&D alliances captured in my sample refer to different
dyads, so I do not face the situation in which the same dyad entered multiple alliances
during the selected time window.
Technological Relatedness. Based on prior research, I look at technological
relatedness by examining the extent to which firms patent in the same technology classes
(Ahuja & Katila, 2001; Dushnitsky & Lenox, 2005a; Rosenkopf & Almeida, 2003).
Consistent with these studies, I consider absorptive capacity to be domain-specific, so I
argue that the knowledge stocks of two firms will be more related the greater the overlap
in technological domains. I first look at all patents granted to each firm (the NBF and the
pharmaceutical firm) during the period 1999–2002, and I create a list of all of the patent
classes to which these patents belong. Choosing a 4-year window allows me to attenuate
fluctuations while providing a window short enough to capture relevant and recent
knowledge stocks (Rothaermel & Boeker, 2008). Then, I create a count of the number of
patent classes in which both firms have had patenting activities (overlap) during this time

17
Prior studies found a similar, though slightly higher, alliance formation rate (e.g., Rothaermel and Boeker
(2008) found a 1.5% alliance formation rate). However, these studies included all types of biotechnology
firms, not just new ventures as I do. So this might be the reason why my alliance formation rate is lower
than the one found in these other studies. As Katila et al. (2008) found, entrepreneurial firms are very likely
to delay relationships with incumbent firms when they are exposed to high appropriation risks.

101

window.
18
Basically, following prior research (Dushnitsky & Lenox, 2005a, 2005b), I
claim that the larger the number of common patent classes, the greater the amount of
knowledge that the pharmaceutical firm can assimilate from the NBF—i.e., the greater
the pharmaceutical firm’s absorptive capacity (Cohen & Levinthal, 1990). As prior
research has suggested, I also believe that this measure is more appropriate than the
cross-citation measure of technological overlap (see Mowery et al., 1996) for the purpose
of measuring a firm’s absorptive capacity (Dushnitsky & Lenox, 2005a, 2005b). Cross-
citation measures might underplay information about the similarities of firms that do not
cite each other (a likely event for new ventures given that they have not been around for a
long time) (Dushnitsky & Lenox, 2005a, 2005b).
Development Experience. Consistent with prior research, I look at the number of
drugs developed in the past as a proxy for development experience (Macher & Boerner,
2006; Nerkar & Roberts, 2004). However, to the extent that development knowledge is
particularly relevant for the same therapeutic area from which that experience was
obtained, I cannot just look at the overall number of prior drugs developed by a
pharmaceutical firm to create a proxy for development experience (Macher & Boerner,
2006). For instance, a company that has developed five drugs, each of them for a
different therapeutic area, has less development experience in the areas in which it is
active (i.e., these five therapeutic domains) than a firm that has developed five drugs in

18
I also examined several different normalized versions of this same measure (one in which I divide this
value by the NBF’s total number of patent classes, and a second measure in which I divide this value by the
total number of different patent classes of the NBF and the pharmaceutical firm), and the results go in the
same direction (but with lower significance levels). I retain the measure proposed initially because it
provides a more efficient estimation model (i.e., better fit). The results are available upon request.

102

the same therapeutic area. Thus, there is a need to look at two different dimensions when
exploring drug development experience across diverse therapeutic areas: depth and
diversity (Macher & Boerner, 2006).
Consistent with prior studies, I rely on a measure of depth to proxy for
development experience and a measure of scope to proxy for therapeutic area diversity
(Macher & Boerner, 2006; Cockburn & Henderson, 2001). In this manner, I am able to
capture the two orthogonal constructs proposed in the theory section. Consequently, I
create a measure of development experience that only captures a pharmaceutical firm’s
competences with respect to developing new drugs for the therapeutic areas in which the
firm is active (i.e., how well a firm knows what it knows). For this, I first create a
measure of development experience for each therapeutic area in which the firm is active
by looking at the number of new drugs developed for each particular area in the previous
4 years. As before, I look at development activities in a 4-year window as a way to
attenuate fluctuations and at the same time account for the fact that older experience
might not be that relevant (Rothaermel & Boeker, 2008; Macher & Boerner, 2006).
Finally, I average these values across all therapeutic areas in which the pharmaceutical
firm is active as a way to create a measure of development experience at the firm level.
19

Therefore, a higher value means that the firm has deep development competences in the
areas in which it is active, and this final measure is independent of the total number of
therapeutic areas in which the firm has activities.

19
I have also examined the maximum development experience instead of the average value across all
therapeutic areas, and the findings are substantially the same—identical direction of coefficients and
similar significance levels (data available upon request).

103

Pharma Therapeutic Area Diversity. As explained above, I conceptually
differentiate between depth and diversity of development experience (Macher & Boerner,
2006). In order to create a measure of therapeutic area diversity, I rely on prior studies
and look at the distribution of new drug developments across different therapeutic areas
(Macher & Boerner, 2006). For that, I look at the number of new drugs developed in the
previous 4 years in each therapeutic area. With these values, I rely on prior studies in the
diversification literature (e.g., Miller, 2004) and create a measure of diversity across
therapeutic areas by looking at the Herfindahl index (sum of squared shares). The
Herfindahl index is higher the greater the level of concentration across therapeutic
domains, so in order to create a measure that increases with the level of diversity, I
subtract this Herfindahl index value from 1. That is, the final measure is: [1 – ∑ (TA
i
×
TA
i
)], where TA
i
is the proportion of drugs developed in the previous 4 years for
therapeutic area i. Therefore, a greater value of the final measure of therapeutic area
diversity implies that the firm has development competences across many diverse
therapeutic areas.
20

NBF’s Technology Breadth. I proxy technology breadth as the degree to which
an NBF’s technology relies on diverse knowledge domains. For this measure, I look at
the diversity of patent classes of the patents cited by the NBF (Wuyts, Dutta, & Pai,
2007; Rosenkopf & Nerkar, 2001). I claim that if an NBF cites previous patents that
belong to a diverse set of technological domains, the NBF’s technology will have a

20
I also tried an alternative and simpler measure consisting of a simple count of the number of areas in
which the firm is active, and the results are very similar (same direction of the coefficients and similar
significance levels). Yet, the model that includes the original measure (diversity calculated using the
Herfindahl index) provides a stronger fit.

104

greater breadth (Hall, Jaffe, & Trajtenberg 2001). Prior research has also measured
technology breadth by simply looking at the primary technology class to which a firm’s
patents have been assigned (Miller, Fern, & Cardinal, 2007; Wadhwa & Kotha, 2006).
Yet, the class of the patent itself might not capture the full range of links that such
technology has with other areas, which might be better captured by looking at the
knowledge that such technology has relied upon (Rosenkopf & Nerkar, 2001).
21
Thus, I
rely on an alternative approach proposed by prior research (Wuyts et al., 2007;
Rosenkopf & Nerkar, 2001) and look at the technological domains of the patents cited by
the NBF in order to capture the links of the NBF’s technology with other technological
fields. First, I list all of the patents cited by an NBF in the previous 4 years. Then, I look
at the technology classes of these cited patents. Once I have this list of technology classes
cited, I rely on prior studies’ methodologies and create a measure of technological
breadth by looking at how concentrated these citations are across technology classes
(Hall, Jaffe & Trajtenberg 2001; Miller et al., 2007; Wadhwa & Kotha, 2006). Similar to
the measure of therapeutic area diversity just described, I create a measure of
concentration across patent classes by looking at the Herfindahl index (sum of squared
shares), which is higher the greater the level of concentration across technological
classes. Thus, in order to create a measure that increases with the level of diversity, I
subtract this Herfindahl index value from 1. Hence, the final measure is: [1 – ∑ (p
i
× p
i
)],
where p
i
is the proportion of times that an NBF’s citations correspond to technology class

21
I nevertheless have tested the main estimations using a measure of technology breadth based on
technology classes of the firm’s patents (Miller et al., 2007; Wadhwa and Kotha, 2006) and not on the
patents cited by the firm. The results are very similar (same sign and similar levels of significance). These
results are available upon request.

105

i. This also means that a greater value of the final measure of technology breadth implies
that the NBF draws from, and thus has links with, many technological classes (i.e.,
greater breadth).
Control Variables. I include the following set of pharmaceutical-specific, NBF-
specific, and dyad-specific controls. First, I add several control variables used in prior
research to account for pharma-level effects that could drive alliance formation
(Rothaermel & Boeker, 2008; Gulati, 1995; Ahuja, 2000). These pharma-level controls
include: the pharmaceutical sales in 2002 as a proxy for pharmaceutical firm size, the
pharmaceutical firm’s average return over equity for the 2000–2002 period as a proxy for
pharmaceutical performance (Rothaermel, 2001),
22
the age of the pharmaceutical firm,
the natural logarithm of the number of patents obtained by the pharmaceutical firm in the
previous 4 years, the distribution of the pharmaceutical firm’s patents across technology
classes over the previous 4 years (pharma technology breadth), the number of R&D
alliances in the previous 4 years as a proxy for firm centrality, and a count of the number
of acquisitions in the previous 4 years. In addition, as explained above, there is a need to
account for each specific pharmaceutical firm’s need for R&D partnerships. In order to
be confident that the reason a given NBF and a given pharmaceutical firm do not ally
under the circumstances described in hypotheses 3–6 is appropriation risks, I need to
account for alternative reasons that might be salient from the pharmaceutical firm’s point
of view. It might be the case that a pharmaceutical firm prefers not to ally with an NBF
firm under the circumstances predicted in hypotheses 3–6 because such a pharmaceutical

22
I have also tried alternative measures of firm performance such as return over sales (ROS) and return
over assets (ROA), and the empirical support for the proposed hypotheses is substantially the same (results
available upon request).

106

firm does not have the need for alternative R&D projects. In order to control for the
actual need for R&D partnerships that each single pharmaceutical firm has, I include a
measure of the number of products in the pharmaceutical firm’s pipeline (in the pre-
clinical or clinical stage) divided by the total number of products developed in the
previous 4 years. In essence, this measure captures a pharmaceutical firm’s availability of
resources and necessity for additional development projects. Thus, this variable is
expected to capture the motivation that each pharmaceutical firm will have to look for
R&D partners. I expect that firms with a fuller pipeline will have a lower need to look for
R&D alliances with NBFs.
In addition, I follow prior research and include a set of NBF-specific variables
aimed at capturing the NBF’s attractiveness to account for alternative pharmaceutical
firms’ motivations to ally with each NBF (Rothaermel & Boeker, 2008; Colombo, Grilli,
& Piva, 2006). I include NBF age, NBF centrality (number of alliances in the previous 4
years), the natural logarithm of the number of patents in the previous 4 years, and finally
a measure of the dollar amount of Venture Capital (VC) investment that each NBF has
received since its inception. Colombo et al. (2006) provide evidence that NBFs that
received VC investment had a greater likelihood of establishing alliances. Given the
uncertainty that pharmaceutical firms face about the real quality of each NBF, I claim
that the presence of VC investment represents a strong signal about the quality of an
NBF’s technology.
Finally, I add two dyad-level controls. First, I include a control for geographic
proximity. This measure takes a value of 1 when a given NBF and a given pharmaceutical

107

firm are located in the same city (Rosenkopf & Almeida, 2003).
23
Firms that are
physically closer are more likely to contact each other to explore collaboration
opportunities (Rothaermel & Boeker, 2008). Second, I include a control for prior
alliance. This measure takes a value of 1 if a given dyad has entered into an alliance in
the past, and 0 otherwise (Rothaermel & Boeker, 2008). One could argue that firms that
have already collaborated in the past might trust each other, so appropriation risks may be
less salient in such situations.

3.4.3 Analysis
The unit of analysis, as explained in the theory section, is the dyad, so the
estimation is based on the likelihood that a given dyad established an R&D alliance in the
selected time window (Rothaermel & Boeker, 2008). Because the dependent variable is
binary in nature, I estimated the likelihood of R&D alliance formation using a logit
model (Rothaermel & Boeker, 2008).
Although I have multiple years of data, cross-sectional analysis is preferred in this
context for several reasons. First, because many of the independent and control variables
vary either very little or not at all over time, panel data analysis with almost identical
explanatory variables would add very little information and would increase dependence
across residuals. In other words, I would be “artificially” increasing the sample size
without adding much statistical power to my estimation. In fact, I would just be adding a
lot of 0s into the final sample because I would be looking at multiple years of data

23
Alternative levels of proximity (state or zip code) were tried, and all hypotheses received identical
statistical support.

108

without actually increasing the number of alliances formed. Second, the event that I am
studying is so infrequent that observing every potential dyad on an annual basis would
dramatically increase the relative number of 0’s over 1’s in the final sample, reducing the
statistical power to discriminate between these two values. For these reasons, I believe
that a more conservative and adequate test is to look at a longer time period (2003–2007),
and thus, consistent with prior studies, I relied on a cross-sectional rather than a
longitudinal estimation method (Rothaermel & Boeker, 2008).
It is, however, also important to note that while I had a cross-sectional data
structure, I also had multiple observations per firm to the extent that I look at all potential
NBF-pharmaceutical combinations. Thus, firm-specific error terms could be highly
correlated, which implied that the assumption of independence across error terms was
likely to be violated (Greene, 2003). To deal with this problem, I account for unobserved
heterogeneity using a generalized estimating equations (GEE) approach (Liang & Zeger,
1986). The main advantage of GEE is that it accounts for serial dependence across
observations, which relaxes the assumption of independence across residuals
(Wooldridge, 2002). The estimations provided are based on models in which I cluster
standard errors at the pharmaceutical firm level.
24

3.5 Results
Table 3.1 displays descriptive statistics and correlations for each of the variables
described in the prior section. I find that the main independent variables (technological

24
I have run the tests accounting for dependence at the NBF firm level as well, and the findings are
basically the same (similar coefficients with similar levels of significance). These additional estimations are
available upon request.

109

relatedness, development experience, therapeutic area diversity, and NBF’s technological
breadth) have very low correlations (all lower than 0.09). More specifically, the
correlation between development experience and therapeutic area diversity is only 0.04,
which suggests that these measures are indeed capturing two orthogonal dimensions of a
firm’s development competences. In addition, it is important to note that both
technological relatedness and the NBF’s technology breadth are uncorrelated (0.09),
which also provides some confidence that these two variables are capturing different
constructs. In addition, I find high correlations between pharmaceutical size,
pharmaceutical age, and pharmaceutical centrality (between 0.75 and 0.87). In order to
avoid multicollinearity problems, I ran additional estimations (available upon request)
without some of these controls, and the main results did not change.

110
TABLE 3.1. Descriptive Statistics and Correlations (n = 7,912)
Mean s.d. 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18
1
Technological
relatedness
0.67 0.97 1.00
2
Development
experience
3.42 3.70 0.07** 1.00
3
Pharma therapeutic
area diversity
0.56 0.29 0.08** -0.04** 1.00
4
NBF’s technology
breadth
0.57 0.28 0.09** 0.01 0.01 1.00
5 Pharma size 3.81 10.1 0.14** 0.51** 0.37** 0.01 1.00
6 Pharma performance 0.17 1.33 0.01 0.09 ** 0.05** 0.01 0.16** 1.00
7 Pharma age 29.9 31.0 0.17** 0.40** 0.36** 0.01 0.84** 0.25** 1.00
8 NBF age 2.95 1.64 0.01 0.01 0.01 -0.12** 0.01 0.01 0.01 1.00
9 Pharma patents 2.86 1.75 0.35** 0.14** 0.26** 0.01 0.39** 0.06** 0.47** 0.01 1.00
10 NBF patents 1.40 0.84 0.38** 0.01 0.01 0.30** 0.01 0.01 0.01 -0.01 0.01 1.00
11 VC investment 19.4 37.0 0.08** 0.01 0.01 0.02* 0.01 0.01 0.01 0.18** 0.01 0.09 1.00
12
Pharma technology
breadth
0.63 0.27 0.23** 0.07** -0.01 0.01 0.12** 0.08** 0.15** 0.01 0.57** 0.01 0.01 1.00
13 Pharma pipeline 0.21 0.25 -0.03** -0.16** -0.16** 0.01 -0.23** -0.06** -0.28** 0.01 -0.10** 0.01 0.01 -0.02 1.00
14 Pharma centrality 13.8 23.2 0.18** 0.48** 0.39** 0.01 0.87** 0.09** 0.75** 0.01 0.47** 0.01 0.01 0.18** -0.21** 1.00
15 NBF centrality 0.25 1.29 0.08** 0.01 0.01 0.24** 0.01 0.01 0.01 0.20** 0.01 0.22** 0.29** 0.01 0.01 0.01 1.00
16 Pharma acquisitions 0.29 1.09 0.09** 0.24** 0.20** 0.01 0.44** 0.07** 0.37** 0.01 0.21** 0.01 0.01 0.17** -0.16** 0.43** 0.01 1.00
17 Prior alliance 0.01 0.08 0.02 0.07** 0.07** -0.02 0.13** 0.02 0.12** 0.02* 0.09** -0.01 0.03** 0.03** -0.03** 0.16** 0.13** 0.04** 1.00
18 Geographic proximity 0.01 0.09 0.04* -0.01 -0.03* 0.02 -0.02 -0.01 -0.02 -0.02* 0.02 0.03* 0.03* 0.02* -0.02 -0.01 0.04** -0.02 0.01 1.00
Significance levels: ** p < 0.01, * p < 0.05

111

3.5.1 Effects of Technological Relatedness and Development Experience on the
Likelihood of Alliance Formation: Hypotheses 1 and 2
I tested research Hypotheses 1 and 2 in a set of logistic regressions. These
estimations are presented in Table 3.2 and are discussed next. In Model 1, I just included
the effects of the control variables on the likelihood that a given dyad established an
R&D alliance. Consistent with prior research (Colombo et al., 2006), I found that dyads
in which the NBF received higher VC investment were more likely to establish an
alliance. Also, I found that pharmaceutical firm performance has a significant positive
effect on alliance formation, suggesting that better-performing pharmaceuticals might be
more attractive to NBFs. Also, I found that firms with a fuller pipeline were less likely to
form R&D alliances, which provides some support for the expectation that this measure
captures a pharmaceutical firm’s need for additional R&D projects. Moreover, I found
that firms that are geographically closer are more likely to establish R&D alliances.
Finally, again consistent with prior studies, I found that dyads that established alliances in
the past were indeed more likely to establish additional collaborations (Rothaermel &
Boeker, 2008).
Model 2 adds the measures of technological relatedness and development
experience. Overall, the incremental variance explained by Model 2 over Model 1 is
significant (χ
2
= 7.0, p < 0.05). More specifically, I found that technological relatedness
has a positive and significant effect on the likelihood of alliance formation (β = 0.15, p <
0.05), providing support for Hypothesis 1. In addition, I found that development

112

experience has a positive and significant effect (β = 0.11, p < 0.05) on the likelihood of
alliance formation, which provides support for Hypothesis 2.

113

TABLE 3.2. Estimation of the Likelihood of Alliance Formation

Significance levels: ***p < 0.001, ** p < 0.01, * p < 0.05,
†
p < 0.10.
Likelihood ratio (LR) values test for the increment in the overall model fit after including additional
variables. Model 2 is compared with Model 1, and Models 3, 4, and 5 are compared with Model 2.
I provide pharma-level clustered standard errors.
MODEL 1 MODEL 2 MODEL 3 MODEL 4 MODEL 5
Intercept
-8.49 ***
(1.10)
-8.36 ***
(1.14)
-8.88 ***
(1.30)
-7.99 ***
(1.45)
-8.68 ***
(1.30)
Technological relatedness -
0.15 *
(0.07)
0.66 **
(0.22)
0.33 *
(0.15)
0.78 **
(0.28)
Development experience -
0.11 *
(0.05)
0.17 ***
(0.03)
0.09
(0.06)
0.15 ***
(0.04)
Technological relatedness x pharma
therapeutic area diversity
- -
-1.64 *
(0.66)
-
-1.66 *
(0.75)
Development experience x pharma
therapeutic area diversity
- -
-0.43 ***
(0.11)
-
-0.47 ***
(0.13)
Technological relatedness x NBF’s
technology breadth
- - -
-0.55
†

(0.33)
-0.38
(0.32)
Development experience x NBF’s
technology breadth
- - -
-0.21
†

(0.12)
-0.22 *
(0.09)
Pharma therapeutic area diversity
5.24 **
(1.75)
4.44 **
(1.46)
8.88 ***
(2.33)
4.53 **
(1.46)
9.18 ***
(2.54)
NBF’s technology breadth
1.17
(0.74)
0.94
(0.74)
0.80
(0.75)
1.58
†

(0.95)
1.52
(0.98)
Pharma size
0.02
(0.02)
0.01
(0.02)
0.01
(0.02)
0.02
(0.02)
0.01
(0.02)
Pharma performance
0.28 ***
(0.07)
0.24 ***
(0.07)
0.22 **
(0.08)
0.24 ***
(0.07)
0.22 *
(0.09)
Pharma age
-0.008
(0.006)
-0.003
(0.005)
-0.006
(0.006)
-0.002
(0.005)
-0.004
(0.006)
NBF age
0.07
(0.10)
0.09
(0.11)
0.09
(0.11)
0.08
(0.11)
0.09
(0.11)
Pharma patents
0.27 *
(0.13)
0.24
†

(0.13)
0.22
(0.15)
0.19
(0.13)
0.19
(0.14)
NBF patents
0.17
(0.18)
0.02
(0.22)
-0.05
(0.24)
-0.06
(0.26)
-0.10
(0.26)
VC investment
0.007 **
(0.002)
0.007 **
(0.002)
0.007 **
(0.002)
0.006 **
(0.002)
0.006 **
(0.002)
Pharma technology breadth
0.44
(1.30)
0.30
(1.31)
0.10
(1.42)
0.24
(1.31)
0.01
(1.43)
Pharma pipeline
-1.76 *
(0.76)
-1.09
(0.82)
-1.60
†

(0.92)
-1.18
(0.83)
-1.71
†

(0.95)
Pharma centrality
0.002
(0.008)
0.001
(0.008)
0.001
(0.008)
0.001
(0.008)
0.001
(0.008)
NBF centrality
0.13 ***
(0.03)
0.13 ***
(0.03)
0.13 ***
(0.03)
0.14 ***
(0.03)
0.14 ***
(0.03)
Pharma acquisitions
-0.06
(0.08)
-0.05
(0.10)
-0.06
(0.08)
-0.06
(0.10)
-0.06
(0.09)
Prior Alliance
1.53 **
(0.51)
1.54 **
(0.50)
1.48 **
(0.51)
1.48 **
(0.50)
1.42 **
(0.51)
Geographic proximity
2.00 ***
(0.44)
2.06 ***
(0.44)
1.99 ***
(0.43)
1.98 ***
(0.43)
1.92 ***
(0.44)
N
-2 Log Likelihood
LR (χ
2
)
7,912
438.2
-
7,912
431.2
7.0 *
7,912
422.0
9.2 **
7,912
426.2
5.0
†

7,912
419.2
12.0 **

114

3.5.2 The Moderating Effect of Pharma Therapeutic Area Diversity and the NBF’s
Technology Breadth on Technological Relatedness and Development Experience:
Hypotheses 3a, 3b, 4a, and 4b
I tested the final four hypotheses in a set of logistic regressions. In Model 3, I just
included the interactions with the pharma therapeutic area diversity, and in Model 4 I just
added the interactions with the NBF’s technology breadth, but in Model 5 I included all
four interactions together. All variables were mean-centered prior to creating the
interaction term to avoid collinearity problems (Aiken & West, 1991). In Model 3 (just
adding the interactions with pharma therapeutic area diversity), the likelihood ratio test
showed that the inclusion of these interaction terms significantly improved the model (χ
2

= 9.2, p < 0.01). Specifically, as expected, I found that the interaction between pharma
therapeutic area diversity and technological relatedness is negative and significant (β = -
1.64, p < 0.05). Similarly, the interaction between pharma therapeutic area diversity and
development experience is also negative and significant, as expected (β = -0.43, p <
0.001).
In Model 4, in which, as mentioned, I just introduced the interactions with the
NBF’s technology breadth, the likelihood ratio test showed overall that the improvement
of the model achieved from the inclusion of these interaction terms was marginally
significant (χ
2
= 5.0, p < 0.10). Specifically, the interaction between the NBF’s
technology breadth and technological relatedness was negative, as predicted, and
marginally significant (β = -0.55, p < 0.10). Similarly, I also found a negative (yet also

115

marginally) significant interaction between the NBF’s technology breadth and
development experience (β = -0.21, p < 0.10).
Finally, Model 5 included all four interactions. The likelihood ratio test showed
that the model was significantly improved when I tested all four interactions
simultaneously (χ
2
= 12.0, p < 0.01). Specifically, I found that the interactions of
technological relatedness and development experience with the pharma therapeutic area
diversity were, as before, negative and significant (β = -1.66, p < 0.05 and β = -0.47, p <
0.001, respectively). In addition, I found that the interactions of technological relatedness
and development experience with the NBF’s technology breadth were negative as before,
but with slightly different levels of significance. For instance, although the interaction
between the NBF’s technology breadth and technological relatedness was negative as
predicted, it was not significant (β = -0.38, n.s.). However, I did find a negative and more
significant interaction between the NBF’s technology breadth and development
experience (β = -0.22, p < 0.05).
Overall, I interpret these findings as strong empirical support for the theoretical
arguments underlying Hypotheses 3a and 3b. Both interaction terms are negative and
significant in a consistent manner across different model specifications. Thus, these
findings suggest that pharma therapeutic area diversity increases the level of
appropriation risks associated with technological relatedness and development
experience, and hence weakens the strength of their positive effects on the likelihood of
alliance formation.

116

In addition, I interpret these results as providing support for Hypothesis 4b. The
interaction between the NBF’s technology breadth and development experience is
negative and significant in both model specifications. Thus, these findings suggest that an
NBF’s technology breadth increases the level of appropriation risks that development
experience represents, diminishing the main positive effect that this variable has on the
likelihood of alliance formation. Yet, these findings fail to provide strong support for
Hypothesis 4a. The interaction between an NBF’s technology breadth and technological
relatedness, although always negative, is only marginally significant in one of the
models.
It is important to note that once I introduced all of the interaction terms in Model
5, I found that the main effects of technological relatedness and development experience
became much more significant (β = 0.78, p < 0.01 and β = 0.15, p < 0.001, respectively).
Thus, overall, the findings in Model 5 seem to suggest that technological relatedness and
development experience do indeed increase the likelihood of alliance formation, yet these
effects are diminished when the pharmaceutical firm has development competences in
multiple therapeutic areas or when the NBF’s technology is rather broad.
Finally, by further analyzing the coefficients estimated in Model 5, I find that an
increase in the value of environmental relatedness by one standard deviation from its
mean value results in an increase of 115% in the overall likelihood of alliance formation
(for average values of all other variables). However, that same increase in technological
relatedness leads to increments of 29.8% and 98.7% (rather than the reported 115%) in
the overall likelihood of alliance formation when pharma therapeutic area diversity and

117

the NBF’s technology breadth are one standard deviation above their mean values,
respectively.
Similarly, increasing the value of development experience by one standard
deviation from its mean value results in an increase of 73.3% in the overall likelihood of
alliance formation (for average values of all other variables). However, that same
increase in development experience leads to increments of 4.7% and 38.1% (rather than
the reported 73.3%) in the overall likelihood of alliance formation when therapeutic area
diversity and NBF’s technology breadth are one standard deviation above their mean
values, respectively.

3.5.3 Robustness Test
Consistent with prior studies looking at the likelihood of alliance formation at the
dyad level, the reported estimations of the proposed model specification in Table 3.2 rely
on a fundamental assumption: all NBFs in the sample are at the risk of establishing an
R&D alliance in the period of study (Gulati, 1995; Rothaermel & Boeker, 2008; Stuart,
1998). As explained above, NBFs strongly need the development experience in specific
therapeutic areas (disease-specific) of established pharmaceutical firms (Macher &
Boerner, 2006). Thus, maybe it is not too unrealistic to assume that all NBFs were
actually looking for R&D partners in the period of study. Hence, it might be the case that
those that were unable to find a pharmaceutical firm that provided the required skills
while representing a sufficiently low appropriation risk had no other option but to avoid

118

establishing an alliance. If that is the case, the inclusion of all possible dyads in the
analysis is recommended.
However, it might be the case that this assumption is not realistic and that some of
the NBFs of the sample were not really contemplating the option of establishing an R&D
alliance at that point in time. In that case, including those dyads that are not really at risk
of forming an alliance can be a source of bias. Moreover, the conclusions obtained from
the reported findings could be incorrect to the extent that I might be interpreting the lack
of alliance formation as the NBF’s unwillingness to ally because of appropriation risks,
when in fact the alliance was never an option for that company. The solution to this
problem is not easy to identify in that I lack any objective criteria to determine a priori
which dyads are at real risk of entering an alliance and which ones are not. In order to
ensure the robustness of the results, prior studies have conducted sensitivity analysis with
only a subset of the initial sample that included only those firms that had established at
least one alliance before the period of study (Gulati, 1995). This solution is problematic
in my context to the extent that my sample focuses on new technology ventures, so many
of the firms that established an alliance in the period of study were actually establishing
their very first alliance (more than 90%).
Given the idiosyncracy of my empirical context, I relied on an alternative
methodology to test the robustness of my results under a different set of assumptions. I
relax the assumption that all NBFs are at real risk of establishing an alliance, and I rely
on the alternative assumption that only those firms that actually entered an alliance
during the period of study were at real risk of establishing an alliance. It is important to

119

note that this alternative assumption is also quite strong because it assumes that all NBFs
that did not establish an R&D alliance in the period of study were not looking for R&D
partners. To the degree that these assumptions look at two very different scenarios,
finding empirical support for the proposed hypotheses under this alternative (and also
strong) assumption will increase the robustness of my conclusions.
Under this alternative assumption, I need to test the proposed hypotheses on the
selected subset of dyads that includes only those NBFs that are at risk of establishing an
alliance (those dyads that include NBFs that actually established an alliance during the
period of study). Yet, to the extent that this proposed selection process is endogenous
(since it is based on the dependent variable of the main model specification) the
estimations from the selected subsample might provide biased estimates (Greene, 2003).
One way of looking at this proposed scenario and understanding the inherent endogeneity
is by assuming that NBFs are taking the decision of partner-selection in two steps: in the
first step they decide whether to look for an R&D partner or not (decision to establish an
R&D alliance), while in the second they decide who to partner with conditional on the
first-step decision.
Thus, one possible way to account for this endogeneity is to estimate a two-step
model analogous to a Heckman (1979) selection model. That is, in the first step I estimate
the likelihood of entering the selected sample (decision to establish R&D partnerships),
whereas in the second step I estimate the model specification proposed in the main
analysis (who to partner with) but only for the selected sample, conditional on the
estimation from the first step. Specifically, in the first step I estimate a Probit model in

120

which I look at the probability that a given NBF will establish an alliance in the period of
study as a function of NBF-specific covariates (NBF age, the natural logarithm of the
number of patents of the NBF in the previous 4 years, the dollar amount of venture
capital investment that each NBF has received since its inception, the NBF’s technology
breadth, and the number of prior alliances established by the NBF in the past). Then, in
the second step, I estimate the likelihood of alliance formation between each NBF and
each pharmaceutical firm for the selected subsample (only dyads that include NBFs that
established an alliance in the period of study), correcting for the selection bias by
including the inverse Mills ratio (IMR) generated from the first step (Heckman, 1979). I
estimated this procedure and provide the main findings in Table 3.3, which reports two
different models. The first model shows the estimations obtained with the full sample
(i.e., Model 5 from Table 3.2), whereas the second model reports the estimations with the
two-step selection model.
25

Basically, I found similar support for the proposed hypotheses using this
alternative model specification. The main effects of technological relatedness and
development experience were both positive and significant (β = 0.73, p < 0.01 and β =
0.16, p < 0.001, respectively). In addition, the moderating effects of pharma therapeutic
area diversity on technological relatedness and development experience were both
negative and significant (β = -1.73, p < 0.05, and β = -0.51, p < 0.001, respectively).
Finally, the moderating effect of the NBF’s technology breadth on technological

25
Note that when using two-step estimation, those regressors included in the second step as a function of
estimations of a first step are measured with sampling error (Greene, 2003), so the covariance matrices
estimated in the second step are biased. In order to obtain an unbiased covariance matrix, I follow Murphy
and Topel (1985), who provided a method for correcting standard errors in two-step estimations when both
steps are estimated with maximum likelihood.

121

relatedness was negative but not significant (β = -0.31, n.s.), while the moderating effect
on development experience was negative and significant (β = -0.22, p < 0.05). In sum, I
interpret these findings as strong evidence that dyads with high technological relatedness
are more likely to establish R&D alliances when the pharmaceutical firm is narrowly
focused (low therapeutic area diversity). Also, dyads in which the pharmaceutical firm
has deep development experience are more likely to establish R&D alliances when the
pharmaceutical firm is more narrowly focused (low therapeutic area diversity) or when
the NBF’s technology is not very broad. Independent of the assumption I rely on (i.e., all
NBFs are at risk of establishing an alliance or only those firms that actually allied were at
risk of forming R&D collaborations), the empirical findings remain consistent with the
proposed theoretical story.

122

TABLE 3.3. Robustness Test: Selected Sample

Significance levels: ***p < 0.001, ** p < 0.01, * p < 0.05,
†
p < 0.10.
Inverse Mills ratio calculated from the estimated probability that each NBF established an R&D alliance.
Standard errors are corrected to account for the sampling error from the first-step estimation (Murphy &
Topel, 1985).
FULL SAMPLE SELECTED SAMPLE
Intercept
-8.68 ***
(1.30)
-6.07 ***
(1.33)
Technological relatedness
0.78 **
(0.28)
0.73 **
(0.26)
Development experience
0.15 ***
(0.04)
0.16 ***
(0.04)
Technological relatedness x pharma
therapeutic area diversity
-1.66 *
(0.75)
-1.73 *
(0.70)
Development experience x pharma
therapeutic area diversity
-0.47 ***
(0.13)
-0.51 ***
(0.13)
Technological relatedness x NBF’s
technology breadth
-0.38
(0.32)
-0.31
(0.27)
Development experience x NBF’s
technology breadth
-0.22 *
(0.09)
-0.22 *
(0.09)
Pharma therapeutic area diversity
9.18 ***
(2.54)
9.80 ***
(2.45)
NBF’s technology breadth
1.52
(0.98)
1.65
†

(0.89)
Pharma size
0.01
(0.02)
0.01
(0.01)
Pharma performance
0.22 *
(0.09)
0.22 *
(0.09)
Pharma age
-0.004
(0.006)
-0.006
(0.006)
NBF age
0.09
(0.11)
0.01
(0.11)
Pharma patents
0.19
(0.14)
0.21
(0.14)
NBF patents
-0.10
(0.26)
-0.38
(0.23)
VC investment
0.006 **
(0.002)
-0.007
(0.005)
Pharma technology breadth
0.01
(1.43)
0.19
(1.39)
Pharma pipeline
-1.71
†

(0.95)
-1.78
†

(0.96)
Pharma centrality
0.001
(0.008)
0.001
(0.007)
NBF centrality
0.14 ***
(0.03)
0.08 **
(0.03)
Pharma acquisitions
-0.06
(0.09)
-0.07
(0.09)
Prior alliance
1.42 **
(0.51)
1.20 *
(0.50)
Geographic proximity
1.92 ***
(0.44)
2.20 ***
(0.50)
Inverse Mills ratio -
1.89 ***
(0.53)

N
-2 Log likelihood
7,912
419.2
4,692
394.2

123

3.6 Discussion
The purpose of this empirical essay was to explore how new technology-based
ventures make R&D partner-selection decisions as a function of both appropriation risks
and partner attractiveness. I examined R&D alliances between NBFs and pharmaceutical
firms in the biopharmaceutical industry, and I specifically focused on two main
capabilities that have been identified by the literature as beneficial and likely to increase
the likelihood of alliance formation: technological relatedness between the NBF and the
pharmaceutical firm and the pharmaceutical firm’s development competences. Consistent
with prior studies (Rothaermel & Boeker, 2008), my empirical results provided evidence
that technological relatedness (Hypothesis 1) and development experience (Hypothesis 2)
have a positive effect on the likelihood of alliance formation. Hence, I interpret these
findings as evidence that these factors increase partner attractiveness also for the type of
alliances analyzed in this study.
Yet, drawing on the same organizational learning and RBV logic used to highlight
the benefits of technological relatedness and development experience, I proposed that
these same capabilities might represent relatively stronger appropriation risks under
certain circumstances, therefore reducing the likelihood of alliance formation. Basically, I
argued that these capabilities might allow the pharmaceutical firm to appropriate the
NBF’s knowledge and exploit it in a private way in activities that fall outside the scope of
the alliance. Thus, in these collaborations, there seems to be a tension between
competitive and cooperative behavior, which I argued is resolved in one direction or
another depending upon the alliance’s relative scope. A greater relative scope implies that

124

the pharmaceutical firm faces greater opportunities to benefit from appropriating the
NBF’s knowledge and applying it privately in activities that go beyond the purpose of the
collaboration, and therefore the pharmaceutical firm will have a greater incentive to
behave opportunistically (i.e., competitively rather than cooperatively). Specifically, I
operationalized relative scope in terms of two partner characteristics: the pharmaceutical
firm therapeutic area diversity and the NBF’s technology breadth.
Consistent with this theory, I found empirical evidence suggesting that when the
pharmaceutical firm is highly diversified across different therapeutic domains the positive
effect of technological relatedness and development experience on the likelihood of
alliance formation is diminished. NBFs seem to avoid allying with pharmaceutical firms
with high technological relatedness when they also have high therapeutic area diversity
(Hypothesis 3a). The reason seems to be that such a combination of skills provides the
pharmaceutical firm with the necessary absorptive capacity to assimilate the NBF’s
knowledge, as well as the required experience in specific therapeutic domains, to
privately exploit alternative uses for the NBF’s technology that might go beyond the
purpose of the alliance. Similarly, NBFs seem to avoid pharmaceutical firms with strong
development competences when these firms also have high therapeutic area diversity
(Hypothesis 3b). Again, the reason seems to be that these competences provide the
pharmaceutical firm with the ability to understand and extract part of the therapeutic
potential that the NBF’s technology has for diseases or treatments that fall outside the
scope of the alliance.

125

Also consistent with the logic described before, I found that when the NBF’s
technology has broader applicability, the positive effect of development experience on
the likelihood of alliance formation is diminished (Hypothesis 4b). These findings
suggest that NBFs with broader technologies avoid pharmaceutical firms with strong
development competences because these skills might provide the pharmaceutical firm
with the ability to exploit in a private fashion some of the alternative uses that such a
broad technology has.
Overall, these findings suggest that new ventures actually avoid relationships with
incumbent organizations that possess the types of capabilities that they need (i.e.,
technological relatedness and development experience), when these incumbent firms also
possess complementary abilities that would allow them to appropriate and exploit the
NBF’s knowledge in activities that go beyond the purpose of the collaboration. Thus,
allying with the most capable firm is not necessarily the best option for an NBF. This
study shows that new ventures face a rather heterogeneous group of potential partners,
each of them representing a different level of attractiveness and appropriation risk. Thus,
my findings suggest that in the process of making partner-selection decisions new
entrepreneurial firms try to balance the benefits and risks that each potential partner
represents, and that appropriation risks seem to play a key role in this decision-making
process. That is, not all “sharks” are equally dangerous—or equally attractive—for a
given new venture.
Moreover, this study suggests that NBFs with broader technologies might avoid
relationships with a firm that otherwise might be their optimal partner (high technological

126

relatedness and strong development competences). Interestingly, prior research has found
strong support for the claim that broader technologies lead to more and better innovations
(Henderson & Cockburn, 1994; Katila & Ahuja, 2002). Hence, those new ventures with
broader technologies, who are therefore a priori in a better position to be successful, are
less able to access the complementary resources that they need. Thus, they seem to be in
a weaker position than other NBFs with narrower technologies who may not face similar
appropriation risks.
In addition, from the perspective of the pharmaceutical firm, these findings
suggest that some “skilled” pharmaceutical companies might find themselves seriously
constrained when looking for R&D collaborations because of the amount of
appropriation risks that they represent for small entrepreneurial firms. Ironically, being
“too good” might turn out to be problematic for pharmaceutical firms in some
circumstances. Skills that provide a significant advantage over competitors in some
dimensions of the innovation process might represent a huge disadvantage in others. For
instance, having broad experience across diverse therapeutic domains has been identified
in the literature as highly useful for the process of internal innovation because it allows
firms to exploit economies of scope and knowledge spillovers across different therapeutic
domains (Macher & Boerner, 2006). However, the ability to identify and exploit
alternative uses of a given technology might be perceived as highly undesirable by a new
venture, therefore hampering the ability to ally with that small firm. This, in turn, might
inhibit the ability of the pharmaceutical firm to protect itself and participate in the
competence-destroying cycles initiated by innovative NBFs.

127

3.6.1 Implications for Theory and Research
I believe that this study makes several theoretical contributions to the existing
literature. First, it provides a new theoretical perspective to research looking at alliances
(and more generally at inter-firm collaborations) in the context of R&D activities. Prior
research has frequently relied on an organizational learning perspective and RBV logic to
look at how learning and complementary capabilities provide an organization with the
ability to absorb and exploit knowledge from its R&D partners (Sampson, 2007;
Dushnitsky & Lenox, 2005a, 2005b). One of the main normative implications of some of
these studies is that incumbent organizations should try to partner with those
entrepreneurial firms from which they can absorb the greatest amount of knowledge. Yet,
scholars have by and large overlooked the perspective of the new venture with respect to
its preferences about a partner’s ability to learn and absorb its knowledge. A few studies
have recently started to challenge the accepted idea that entrepreneurs are weak partners
dominated by powerful established firms (Katila et al., 2008), and I believe that the
findings of the present study also challenge conventional wisdom. It seems to be the case
that entrepreneurial firms have some discretion in partner-selection decisions, and this
study suggests that they actually try to avoid those partners that represent the greatest risk
of appropriation. Thus, a novel question this study seeks to answer is: when are new
technology-based ventures really willing to ally with incumbent firms that have high
levels of absorptive capacity and complementary capabilities? I believe that this study
represents a first step in trying to provide an answer to this question.

128

Second, I believe that this study also makes important contributions to the
entrepreneurship literature. Overall, it provides a deeper understanding of how
entrepreneurial ventures choose alliance partners in the context of R&D collaborations.
As explained before, entrepreneurial firms face a very critical tension when entering into
alliances with established organizations, and this study provides new insights into how
new ventures balance appropriation risks and partner attractiveness in their choice of
alliance partners. More specifically, this study contributes to the literature exploring
appropriation risks in the context of new ventures by suggesting that, even within the
same industry, not all “sharks” are equally dangerous. For example, NBFs seem to
perceive pharmaceutical firms with high development experience and therapeutic area
diversity as riskier partners, whereas pharmaceutical firms with high development
experience but narrow scope seem to be perceived as highly attractive partners. Thus, this
study provides empirical evidence of the existing heterogeneity in terms of appropriation
risks that different incumbent firms represent for different entrepreneurial ventures.
Finally, I believe that by looking at the dual nature of a partner’s ability to absorb
and exploit other partners’ know-how, which leads to the aforementioned tension
between competitive and cooperative behavior in the context of R&D collaborations, this
study provides a natural extension (as well as an empirical test) of Khanna et al. (1998)
and Khanna’s (1998) theoretical framework. Specifically, I attend to recent criticisms of
this framework by accounting for not only partners’ incentives but also partners’ ability
to appropriate other firms’ knowledge (Inkpen, 2000; Lavie, 2007). My findings suggest
that relative scope leads to competitive behavior, particularly when combined with

129

partners’ absorptive capacity and development competences to assimilate and exploit
such knowledge in a private fashion. Moreover, I propose two different ways to
operationalize the concept of relative scope by looking at context-specific partner
characteristics. The extent of alternative uses of a given technology that fall outside the
scope of a given alliance may be captured by two indicators: therapeutic area diversity
and technology breadth. On the one hand, I argued that a greater experience of the
pharmaceutical firm in other therapeutic areas could increase the amount of private
benefits that such a firm is able to obtain. On the other hand, I proposed that the broader
the technology of the NBF, the greater the number of alternative uses that the
“appropriating” firm can exploit privately (outside the alliance). Thus, I believe the use of
context and firm-specific factors in the operationalization of the concept of relative scope
improves our ability to understand how the tension between competitive and cooperative
behavior is resolved in these R&D collaborations.

3.6.2 Limitations and Directions for Future Research
In closing, I would like to acknowledge certain limitations of this study and
suggest several directions for future research. First, this study is descriptive in nature to
the extent that I explore partners’ selection decisions, but not the actual consequences of
firms’ partnering behavior. However, the proposed theoretical story implies several
normative statements. Specifically, the study suggests that firms that fail to make the
right choice of alliance partner (i.e., the risk that the partner will appropriate the new
venture’s knowledge is very high) will experience a higher risk that their knowledge will

130

be appropriated. The negative outcomes that a new venture would experience from
allying with the “wrong” partner can be identified across several dimensions. First, we
would expect that the actual collaboration will suffer from the pharmaceutical firm’s
greater incentive to behave opportunistically. In such a case, the pharmaceutical firm
might prefer to spend more time and effort on learning from its partner than on trying to
meet the alliance goals. Hence, we would expect that choosing the “wrong” partner will
have a negative impact on alliance performance. In addition, if the pharmaceutical firm is
able and motivated to appropriate the NBF’s knowledge, we would expect that the NBF’s
future performance or ability to find future R&D partners will be hampered as well, to the
extent that some potential applications of its technology are being exploited by its former
(pharmaceutical) partner. Thus, I believe that future research might benefit from
extending this analysis by examining the consequences of partner-selection decisions that
deviate from the behaviors identified as optimal in the present study.
Second, because of the lack of available data I cannot look into alternative
defense mechanisms that NBFs might use to protect themselves against knowledge
appropriation. Katila et al. (2008), for example, found that firms in industries in which
trade secrets represent a more reliable protection mechanism were more likely to
establish links with “sharks.” Because I focus on a single industry, I cannot rely on the
same industry-level data to account for this alternative protection mechanism, yet it might
be the case that even within the same industry NBFs show a high degree of heterogeneity
in terms of their ability to protect their intellectual property. Thus, future studies looking
at this important source of heterogeneity will definitely provide a richer picture about the

131

different alternatives that new ventures have to solve the tension they face when
collaborating with powerful incumbent organizations.
Finally, although focusing on a single industry and a specific type of alliances
surely increases internal validity, the generalizability of my findings is limited.
Regulatory processes in this industry are time and capital intensive, so small ventures
show a rather strong need for incumbents’ resources, which might not be the case in other
industries. However, when it comes to patenting, the biopharmaceutical industry is
believed to be one of the industries in which patents represent a higher degree of
intellectual property protection (Matraves, 1999). Thus, the presence of strong empirical
evidence in an industry where appropriation risks seem to be weaker than the average
suggests that the effects found in this study might be even stronger in other industries.
Another aspect that may potentially limit the generalizability of my findings is that I only
focus on a sub-segment of the alliances that take place in the biopharmaceutical industry
(i.e., alliances between NBFs and established pharmaceutical firms). This decision was
basically driven by the motivation to explore the dilemma faced by new technology-
based ventures when looking for R&D allies. Yet, an unanswered question is whether the
same proposed theoretical logic—even in the same industry—applies to symmetric
partnerships (e.g., large pharma vs. large pharma) in which no firm has a clearly
advantageous position. The answer to this question is not straightforward, and future
research could benefit from a broader analysis of how different firm-level capabilities
(development competences and absorptive capacity) shape partner behavior for different
types of partnerships (i.e., with different degrees of asymmetry).

132

In conclusion, I believe that this study of partner-selection decisions as a function
of partner attractiveness and appropriation risks provides a broader insight into the
criteria that entrepreneurial ventures appear to choose in selecting R&D partners. I hope
that my study has taken important steps in the direction of developing a greater
understanding of how appropriation risks shape new ventures’ alliance activities.

133

CHAPTER 4
CONCLUSIONS
4.1 Summary of Findings
The goal of this dissertation was to explore how different biopharmaceutical
organizations (i.e. traditional pharmaceutical firms and new entrepreneurial
biotechnology ventures) face a very different set of challenges in an industry that is going
through a deep process of radical technological change. More specifically, the goal was
to understand how these challenges shape firms’ strategic choices related to innovation
activities. I attempted to achieve this purpose by focusing on a specific set of challenges
faced by these two types of organizations in two distinct empirical essays.
In the first essay, I identified a set of limitations that established pharmaceutical
firms face when trying to exploit their internal prior experience for new drug
developments. Specifically, I found that once a firm has already relied heavily on its prior
local development experience, its ability to identify unsatisfied needs in a specific
technological market and come up with new original products to satisfy those needs are
hampered. Similarly, I found that firms appear unable to apply distal experience into a
local technological market until they have obtained a sufficiently deep understanding of
their experience in distal technological markets. These findings suggest that, even though
internal experience is an important driver of new drug developments, incumbent
pharmaceutical firms face strong constraints if they simply rely on their internal
expertise. In essay 1 I identified a potential source of expertise (i.e. boards of directors)
overlooked by prior research that seems to complement firms’ internal experience, and

134

helps incumbent pharmaceuticals to exploit to a greater extent the potential of their own
internal knowledge in the form of new drug developments. Specifically, my findings
suggested that directors’ local and distal experience help pharmaceutical firms overcome,
to some degree, the limitations identified above. These findings are especially significant
in that they are quite consistent with conventional wisdom among practitioners that
directors are often appointed for their ability to provide valuable strategic advice. It
seems to be the case that many pharmaceutical firms in this industry try to buffer
themselves from the environmental uncertainties arising from the emerging
biotechnology “logic”, by appointing directors based on their experience and prior
innovation activities in other firms in the industry.
In the second essay I explored how new biotechnology-based ventures emerging
from the new biotechnology logic deal with a very fundamental challenge, i.e. the
“swimming with sharks” dilemma. On one hand, even though these new organizations
enjoy a clear technological advantage over incumbent firms in that they bring the new
biotechnology logic into the industry, they still need to access the resources that
incumbent pharmaceutical firms possess. Thus, these new ventures have a very strong
motivation to establish collaborations with powerful and resourceful incumbent
pharmaceuticals. However, on the other hand, sharing valuable knowledge with a highly
skilled organization presents high appropriation risks. Incumbent pharmaceuticals with
strong complementary resources are potentially very risky partners because such firms
are in the need for technological “ideas” and inputs, especially given their strong
necessity to adapt to the emerging biotechnology logic. In essay 2 I explored how biotech

135

new ventures deal with this challenge, and proposed that NBFs will balance benefits and
risks through partner selection decisions. Specifically, this essay’s findings suggested that
new ventures actually avoid relationships with incumbent organizations that possess the
types of capabilities that they need (i.e., technological relatedness and development
experience), when those incumbent firms also possessed complementary abilities that
would allow them to appropriate and exploit the NBF’s knowledge in activities that went
beyond the purpose of the collaboration.
Taken together, the findings from my two empirical essays suggest that under the
emergence of a new disruptive technological logic, it might be the case that no player
faces a clear dominant position, and both end up coexisting in equilibrium. It seems to be
the case (at least in this industry) that radical technological change is not enough for
emergent technology-based firms to occupy the dominant position in this industry. These
firms are in fact facing some fundamental challenges in the form of appropriation risks
that might undermine their technological advantage. On the other hand, incumbent
pharmaceuticals are facing very strong pressures from the emergence of a disruptive
technological shift, and their options to survive and adapt to this new logic seem to be
contingent on their ability to combine internal and external sources of experience.

4.2 New Directions for Future Research
In closing I would like to identify potential directions for future research that will
complement and extend the findings of the present dissertation. First, as I acknowledged
in the limitations section of both essays, these studies are descriptive in nature. Basically,

136

they both explore firms’ strategic actions driven by the presence of specific challenges.
Yet, an unanswered question is whether firms that deviate from the “optimal behavior”
identified in both theoretical stories will show a lower performance. For example, one
potential research question is whether pharmaceutical firms that combine internal and
external (from board of directors) experience are more successful in their new-drug
developments. In addition, from the findings of the second essay we would expect that
choosing the “wrong” partner will have a negative impact on alliance performance for a
small biotechnology venture. Hence, future studies testing the normative implications of
both essays of my dissertation will provide a stronger test for the proposed theoretical
frameworks, and improve our understanding of how firms’ innovation activities are
affected by the specific set of challenges identified in this dissertation.
Second, in both essays I assumed that firms facing the described challenges are
homogenous in terms of the range of available tools they have to deal with these threats.
However, it might be the case that firms have alternative solutions at their disposal
(beyond the ones identified in both essays), and therefore the risks they face due to a
radical technological change are very different. For instance, in the second essay I
proposed that firms reduce appropriation risks by avoiding “dangerous” partners. Yet, it
might the case that some NBFs can use alternative defense mechanisms to protect
themselves against knowledge appropriation, and therefore do not face the need to avoid
such relationships. For example, some new ventures might have a unique ability to
protect their intellectual property in the context of R&D relationships, so they might be
willing to ally with dangerous partners because the real appropriation risks they face are

137

substantially lower. Similarly, in the first essay I proposed that pharmaceutical firms are
able to keep up with their new-drug development activities by combining internal
experience with experience provided by directors. Yet, firms might be able to access
other organizations’ expertise through other channels (e.g. acquisitions or strategic
alliances). So an unanswered question is whether these different channels are independent
of or complements to each other.
In sum, I believe that by adopting a broader perspective and exploring a wider
range of different solutions that firms might adopt to deal with these threats, future
research might provide a more comprehensive picture about how these challenges shape
firms’ overall innovation strategies.
In conclusion, the two essays of my dissertation explored a set of challenges that
firms in the biopharmaceutical industry face as a result of radical and uncertain
technological change. The empirical evidence from these studies is consistent with the
claim that the two types of firms coexisting in this industry face very different
constraints. Specifically, my dissertation showed how the presence of these constraints
shape the strategic choices made by these firms. First, a key strategic choice made by
established pharmaceutical firms is the direction of new product development efforts.
Apparently, pharmaceutical firms appear to gear these efforts to take advantage of both
internal and directors’ local and distal experience bases. Second, a key strategic choice
made by NBFs is who to establish R&D relationships with. It appears that NBFs trade off
partner attractiveness along certain dimensions against approrpriation risks, which result
in allying decisions that may not appear optimal at first glance but meet the conflicting

138

criteria that these firms seem to optimize. In sum, I hope that this dissertation will spur
additional research that helps improve our understanding of how firms’ innovation
strategies are affected by and respond to the emergence of a new technological logic.

139

REFERENCES

American Bar Association. 2007. Biotechnology and the law: Part one and two. Section
of Science & Technology Law. Chicago: The ABA Center for Continuing Legal
Education.

Ahuja, G. 2000. The duality of collaboration: Inducements and opportunities in the
formation of inter-firm linkages. Strategic Management Journal, 21: 317–343.

Ahuja, G., & Katila, R. 2001. Technological acquisitions and the innovation performance
of acquiring firms: a longitudinal study. Strategic Management Journal, 22: 197–220.

Aiken, L.S., & West, S.G. 1991. Multiple regression: Testing and interpreting
interactions. Newbury Park, CA: SAGE Publications.

Alvarez, S.A., & Barney, J.B. 2001. How entrepreneurial firms can benefit from alliances
with large partners. Academy of Management Executive, 15: 139–148.

Argote, L. 1999. Organizational learning: Creating, retaining and transferring
knowledge. Kluwer Academic: Boston, MA.

Arora, A., & Gambardella, A. 1994. Evaluating technological information and utilizing
it: Scientific knowledge, technological capability, and external linkages in biotechnology.
Journal of Economic Behavior & Organization, 24: 91–114.

Arundel, A., & Kabla, I. 1998. What percentage of innovations are patented? Empirical
estimates for European firms. Research Policy, 27: 127–141.

Banbury, G., & Mitchell, W. 1995. The effect of introducing important incremental
innovations on market share and business survival. Strategic Management Journal, 16:
161–182.

Baum, J.A., Calabrese, T., & Silverman, B.S. 2000. Don’t go it alone: Alliance network
composition and startups’ performance in Canadian biotechnology. Strategic
Management Journal, 21: 267–294.

Baysinger, B., Kosnik, R.D., & Turk, T.A. 1991. Effects of board and ownership
structure on Corporate R&D strategy. The Academy of Management Journal, 34: 205–
214.

Belsley, D., Kuh, E., & Welsh, R. 1980. Regression diagnostics: Identifying influential
data and sources of collinearity. New York: John Wiley and Sons.

140

Biotechnology innovation report. 2005. Finnegan, Henderson, Farabow, Garrett &
Dunner, LLP. Washington D.C.

Carpenter, M.A., & Westphal, J.D. 2001. The strategic context of external network ties:
Examining the impact of director appointments on board involvement in strategic
decision making. The Academy of Management Journal, 44: 639–660.

Cassiman, B., & Veugelers, R. 2006. In search of complementarity in innovation
strategy: Internal R&D and external knowledge acquisition. Management Science, 52:
68–82.

Chaney, P., & Devinney, T. 1992. New product innovations and stock price performance.
Journal of Business Finance & Accounting, 19: 677–695.

Cockburn, I.M., & Henderson, R.M. 2001. Scale and scope in drug development:
Unpacking the advantages of size in pharmaceutical research. Journal of Health
Economics, 20: 1033–1057.

Cohen, W.M., & Levinthal, D.A. 1990. Absorptive capacity: A new perspective on
learning and innovation. Administrative Science Quarterly, 35: 128–152.

Colombo, M.G., Grilli, L., & Piva, E. 2006. In search of complementary assets: The
determinants of alliance formation of high-tech start-ups. Research Policy, 35: 1166–
1199.

Daily, C.M., Dalton D.R., & Cannella, A.A. 2003. Introduction to special topic forum.
Corporate governance: Decades of dialogue and data. Academy of Management Review,
28: 371–382.

Deeds, D.L., & Hill, C.W.L. 1996. Strategic alliances and the rate of new product
development: An empirical study of entrepreneurial biotechnology firms. Journal of
Business Venturing, 11: 41–55.

Durand, R., Bruyaka, O., & Mangematin, V. 2008. Do science and money go together?
The case of the French biotech industry. Strategic Management Journal, 29: 1281–1299.

Dushnitsky, G., & Lenox, M.J. 2005a. When do incumbents learn from entrepreneurial
ventures? Corporate venture capital and investing firm innovation rates. Research Policy,
34: 615–639.

Dushnitsky, G., & Lenox, M.J. 2005b. When do firms undertake R&D by investing in
new ventures? Strategic Management Journal, 26: 947–965.

141

Dutta, S., & Weiss, A.M. 1997. The relationship between a firm’s level of technological
innovativeness and its pattern of partnership agreements. Management Science, 43: 343–
356.

Ernst and Young biotechnology annual industry reports. Diverse years. Palo Alto, CA:
Ernst and Young.

Fleming, L. 2001. Recombinant uncertainty in technological search. Management
Science, 47: 117–132.

Fleming L. 2002. Finding the organizational sources of technological breakthroughs: The
story of Hewlett-Packard’s thermal ink-jet. Industrial and Corporate Change 11: 1059–
1084.

Fleming, L., & Sorenson O. 2004. Science as a map in technological search. Strategic
Management Journal, 25: 909–928.

Greene, W.H. 2003. Econometric analysis, Third Edition. Englewood Cliffs, NJ: Prentice
Hall.

Griffin, A. 1997. Drivers of NPD success: The 1997 PDMA report. Chicago: Product
Development Management Association.

Gulati, R. 1995. Social structure and alliance formation patterns: A longitudinal analysis.
Administrative Science Quarterly, 40: 619–652.

Gulati, R., & Higgins M.C. 2003. Which ties matter when? The contingent effects of
interorganizational partnerships on IPO success. Strategic Management Journal, 24:
127–144.

Hall, B.H., Jaffe A.B., & Trajtenberg, M. 2001. The NBER patent citations data file:
lessons, insights and methodological tools. NBER Working Paper 8498. Cambridge MA.

Hambrick, D.C., Cho, T.S., & Chen, M. 1996. The influence of top management team
heterogeneity on firms' competitive moves. Administrative Science Quarterly, 41: 659–
684.

Heckman, J. 1979. Sample selection bias as a specification error. Econometrica, 47: 153–
161.

Henderson, R.M., & Cockburn, I.M. 1994. Measuring competence? Exploring firm
effects in pharmaceutical research. Strategic Management Journal, 15: 63–84.

142

Hill, C.W.L., & Rothaermel, F.T. 2003. The performance of incumbent firms in the face
of radical technological innovation. Academy of Management Review, 28: 257–274.

Hillman, A.J., & Dalziel, T. 2003. Boards of directors and firm performance: Integrating
agency and resource dependence perspectives. Academy of Management Review, 39:
383–396.

Hoskisson, R.E., Hitt, M.A., Johnson, R.A., & Grossman, W. 2002. Conflicting voices:
The effects of institutional ownership heterogeneity and internal governance on corporate
innovation strategies. The Academy of Management Journal, 45: 697–716.

Inkpen, A.C. 2000. A note on the dynamics of learning alliances: Competition,
cooperation, and relative scope. Strategic Management Journal, 21: 775–779.

Katila, R., & Ahuja, G. 2002. Something old, something new: A longitudinal study of
search behavior and new product introduction. The Academy of Management Journal,
45(6): 1183–1194.

Katila, R., & Mang, P.Y. 2003. Exploiting technological opportunities: The timing of
collaborations. Research Policy, 32: 317–332.

Katila, R., Rosenberg, J.D., & Eisenhardt, K.M. 2008. Swimming with sharks:
Technology ventures and corporate relationships. Administrative Science Quarterly, 53:
295–332.

Khanna, T. 1998. The scope of alliances. Organization Science, 9: 340–355.

Khanna, T., Gulati, R., & Nohria, N. 1998. The dynamics of learning alliances:
Competition, coopertation, and relative scope. Strategic Management Journal, 19: 193–
210.

King, A.A., & Tucci, C.L. 2002. Incumbent entry into new market niches: The role of
experience and managerial choice in the creation of dynamic capabilities. Management
Science, 48: 171–186.

Kogut, B., & Zander, U. 1992. Knowledge of the firm, combinative capabilities, and the
replication of technology. Organization Science, 3: 383–397.

Kor, Y.Y. 2006. Direct and interaction effects of top management team and board
compositions on R&D investment strategy. Strategic Management Journal, 27: 1081–
1099.

Kroll, M., Walters, B.A., & Wright, P. 2008. Board vigilance, director experience, and
corporate outcomes. Strategic Management Journal, 29: 363–382.

143

Lane, P.J., & Lubatkin, M. 1998. Relative absorptive capacity and interorganizational
learning. Strategic Management Journal, 19: 461–477.

Lanjow, J.O., & Schankerman, M. 2001. Enforcing intellectual property rights. NBER
Working Paper No. 8656. Cambridge, MA: National Bureau of Economic Research.

Lanjow, J.O., & Schankerman, M. 2004. Protecting intellectual property rights: Are small
firms handicapped? The Journal of Law and Economics, 47: 45–74

Lavie, D. 2007. Alliance portfolios and firm performance: A study of value creation and
appropriation in the U.S. software industry. Strategic Management Journal, 28: 1187–
1212.

Levitt, B., & March, J.G. 1988. Organizational learning. Annual Review of Sociology, 14:
319–340.

Li, D., Eden, L., Hitt, M.A., & Ireland, R.D. 2008. Friends, acquaintances or strangers?
Partner selection in R&D alliances. The Academy of Management Journal, 51: 315–334.

Liang, K.Y., & Zeger, S.L. 1986. Longitudinal data analysis using generalized linear
models. Biometrika, 73: 13–22.

Macher, J.T., & Boerner, C.S. 2006. Experience and scale and scope economies: Trade-
offs and performance in development. Strategic Management Journal, 27: 845–865.

Mansfield, E. 1986. Patents and innovation: An empirical survey. Management Science,
32: 173–181.

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.

Matraves, C. 1999. Market structure, R&D and advertising in the pharmaceutical
industry. Journal of Industrial Economics, 42: 169–194.

Miller, D.J. 2004. Firms’ technological resources and the performance effects of
diversification: A longitudinal study. Strategic Management Journal, 25: 1097–1119.

Miller, D.J., Fern, M.J., & Cardinal, L.B. 2007. The use of knowledge for technological
innovation within diversified firms. The Academy of Management Journal, 50: 308–326.

144

Mowery, D.C., Oxley, J.E., & Silverman, B.S. 1996. Technological overlap and interfirm
cooperation: Implications for the resource-based view of the firm. Research Policy, 27:
507-523.

Murphy, K.M., & Topel, R.H. 1985. Estimation inference in two-step econometric
models. Journal of Business and Economic Statistics, 3: 88–97.

Nelson, R., & Winter, S. 1982. An evolutionary theory of economic change. Cambridge,
MA: Belknap.

Nerkar, A., & Roberts, P.W. 2004. Technological and technological market experience
and the success of the new product introductions in the pharmaceutical industry. Strategic
Management Journal, 25: 779–799.

Oxley, J.E. 1997. Appropriability hazards and governance in strategic alliances: A
transaction cost approach. Journal of Law, Economics and Organization, 13: 387–409.

Oxley, J.E., & Sampson, R.C. 2004. The scope and governance of international R&D
alliances. Strategic Management Journal, 25: 723–749.

Palmer, D.A., Jennings, P.D., & Zhou, X. 1993. Late adoption of the multidivisional form
by large U.S. corporations: Institutional, political and economical accounts.
Administrative Science Quarterly, 38: 100–131.

Park, S.H., Chen, R., & Gallagher, S. 2002. Firm resources as moderators of the
relationship between market growth and strategic alliances in semiconductor start-ups.
The Academy of Management Journal, 45: 527–545.

Paruchuri, S., Nerkar, A., & Hambrick, D. 2006. Acquisition integration and productivity
losses in the technical core: Disruption of inventors in acquired companies. Organization
Science, 17: 545–562.

PhRMA annual report. 2007. Washington, D.C.: PhRMA.

Pisano, G.P. 1997. R&D performance, collaborative arrangements, and the market-for-
know-how: A test of lemons hypothesis in biotechnology. Working Paper.

Rockman, H.B. 2004. Intellectual property law for engineers and scientists. New Jersey:
IEEE Press/Wiley Inter-Science Press.

Rosenkopf, L., & Almeida, P. 2003. Overcoming local search through alliances and
mobility. Management Science, 49: 751–766.

145

Rosenkopf, L., & Nerkar, A. 2001. Beyond local search: Boundary-spanning,
exploration, and impact in the optical disk industry. Strategic Management Journal, 22:
287–306.

Rothaermel, F.T. 2001. Incumbent’s advantage through exploiting complementary assets
via interfirm cooperation. Strategic Management Journal, 22: 687–699.

Rothaermel, F.T., & Boeker, W. 2008. Old technology meets new technology:
Complementarities, similarities, and alliance formation. Strategic Management Journal,
29: 47–77.

Rothaermel, F.T., & Deeds, D.L. 2004. Exploration and exploitation alliances in
biotechnology: A system of new product development. Strategic Management Journal,
25: 201–221.

Sampson, R. 2007. R&D alliances and firm performance: The impact of technological
diversity and alliance organization on innovation. The Academy of Management Journal,
50: 364–386

Stevens G.A., & Burley J. 1997. 3,000 raw ideas equals one commercial success.
Research Technology Management, 40: 16–27.

Stuart, T.E. 1998. Network positions and propensities to collaborate: An investigation of
strategic alliance formation in a high-technology industry. Administrative Science
Quarterly, 43: 668–698.

Stuart, T. E., Hoang, H., & Hybels, R.C. 1999. Interorganizational endorsements and the
performance of entrepreneurial ventures. Administrative Science Quarterly, 44: 315–349.

Teece, D.J. 1986. Profiting from technological innovation: Implications for integration,
collaboration, licensing and public policy. Research Policy, 15: 285–305.

Teece, D.J. 1992. Competition, cooperation, and innovation: organizational arrangements
for regimes of rapid technological progress. Journal of Economic Behavior and
Organization, 18: 1–25.

Utterback J.M. 1994. Mastering the dynamics of innovation. Boston: Harvard Business
School Press.

Wadhwa, A., & Kotha, S. 2006. Knowledge creation through external venturing:
Evidence from the telecommunications equipment manufacturing industry. The Academy
of Management Journal, 49: 819–835.

146

Westphal, J.D., & Fredrickson, J.W. 2001. Who directs strategic change? Directors’
experience, the selection of new CEOs, and change in corporate strategy. Strategic
Management Journal, 22: 1113–1137.

Wooldridge, J.M. 2002. Econometric analysis of cross section and panel data, First
Edition, Cambridge, MA: MIT Press.

Wuyts, S., Dutta, S., and Pai, S. 2007. Superior new product development: The interplay
between internal knowledge creation and alliance activity. Working Paper.

Zahra, S.A. 1996. Governance, ownership, and corporate entrepreneurship: The
moderating impact of industry technological opportunities. The Academy of Management
Journal, 39: 1713–1735.

Abstract (if available)

Abstract In this dissertation I discuss two empirical studies that examine how different types of firms in the biopharmaceutical industry approach different types of challenges posed by radical and uncertain technological change. In the first empirical essay I explore the role of different sources of experience in understanding incumbent pharmaceutical firms' decisions to develop new drugs. In the second empirical essay I explore how emerging new biotechnology ventures make alliance partner selection decisions as a function of both partner attractiveness and the risks of appropriation that arise from establishing alliances with incumbent pharmaceuticals.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Empirical essays on relationships between alliance experience and firm capability development

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

Essays in non-market strategy and entrepreneurship

PDF

CEO selection performance: does board experience matter?

PDF

How do acquirers govern the deal-making process? Three essays on U.S. mergers and acquisitions 1994 – 2017

PDF

Essays in retail management

PDF

Three essays on distance: examing the role of institutional distance on foreign firm entry, local isomorphism strategy and subsidiary performance

PDF

Short-term project organizations for corporate entrepreneurship: evidence from the Japanese animation industry (2000–2008)

PDF

Essays on consumer product evaluation and online shopping intermediaries

PDF

Essays on strategic categorization

PDF

Essays on supply chain management

PDF

The effects of accounting performance and professional relationships on promotion, dismissal, and transfer decisions in a conglomerate

PDF

Competing across and within platforms: antecedents and consequences of market entries by mobile app developers

PDF

Automated contracts and the lawyers who don't review them: adoption and use of machine learning technology

PDF

Quality investment and advertising: an empirical analysis of the auto industry

PDF

The structure of strategic communication: theory, measurement, and effects

PDF

Essays on the role of entry strategy and quality strategy in market and consumer response

PDF

Essays on the turning points of the product life cycle

PDF

Exploring the role of learning in contract design: empirical evidence using buyer-supplier IT services contracts

Asset Metadata

Creator Diestre, Luis (author)

Core Title Empirical essays on alliances and innovation in the biopharmaceutical industry

School Marshall School of Business

Degree Doctor of Philosophy

Degree Program Business Administration

Publication Date 08/05/2009

Defense Date 05/05/2009

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag biotechnology,corporate governance,entrepreneurship,innovation and technology,OAI-PMH Harvest,strategic alliances,strategic management

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Rajagopalan, Nandini (committee chair), Dutta, Shantanu (committee member), Mayer, Kyle (committee member), Moon, Hyungsik Roger (committee member)

Creator Email diestre@usc.edu,luisdiestre@hotmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m2483

Unique identifier UC1255188

Identifier etd-DIESTRE-2967 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-182514 (legacy record id),usctheses-m2483 (legacy record id)

Legacy Identifier etd-DIESTRE-2967.pdf

Dmrecord 182514

Document Type Dissertation

Rights Diestre, Luis

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