University of Southern California Dissertations and Theses

Essays on technology innovation and new product development strategy

(USC Thesis Other)

Essays on technology innovation and new product development strategy

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content ESSAYS ON TECHNOLOGY INNOVATION AND NEW PRODUCT
DEVELOPMENT STRATEGY
Copyright 2003
by
Shaoming Qu
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 2003
Shaoming Qu
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UMI Number: 3116771
INFORMATION TO USERS
The quality of this reproduction is dependent upon the quality of the copy
submitted. Broken or indistinct print, colored or poor quality illustrations and
photographs, print bleed-through, substandard margins, and improper
alignment can adversely affect reproduction.
In the unlikely event that the author did not send a complete manuscript
and there are missing pages, these will be noted. Also, if unauthorized
copyright material had to be removed, a note will indicate the deletion.
®
UMI
UMI Microform 3116771
Copyright 2004 by ProQuest Information and Learning Company.
All rights reserved. This microform edition is protected against
unauthorized copying under Title 17, United States Code.
ProQuest Information and Learning Company
300 North Zeeb Road
P.O. Box 1346
Ann Arbor, Ml 48106-1346
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
UNIVERSITY OF SOUTHERN CALIFORNIA
THE GRADUATE SCHOOL
UNIVERSITY PARK
LOS ANGELES, CALIFORNIA 90089-1695
This dissertation, written by
5>M AOM t1sl£i
under the direction o f h \% dissertation committee, and
approved by all its members, has been presented to and
accepted by the Director o f Graduate and Professional
Programs, in partial fulfillment o f the requirements fo r the
degree o f
DOCTOR OF PHILOSOPHY
/ f a y *
/ Q / " Dire Director
Date A u gu st 1 2 , 2003
Dissertation Committee
n
Chair
L —
f''(^\y ..■
\
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Dedication
This dissertation is dedicated to my parents, Zheng Qu and Kunfan Yang, and my
wife Siyan Wang.
1 1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Acknowledgments
This dissertation was written under the supervision of my chief advisor,
Professor Shantanu Dutta. Professor Dutta's deep and broad knowledge in Marketing
practices and theories, Economics and Strategic Management inspired this research.
Without his helpful advisement, continuous encouragement and stimulating
discussion, this work may never be completed. His working style, down-to-earth
personality, and self-imposed unusually high standard on his work have made the
most important influence on me.
I am very grateful to Professor Om Narasimhan and Professor Rakesh Niraj for their
collaboration and guidance on this dissertation. I have benefited a lot from discussion
with them on my research from various perspectives.
I would also like to thank Professor Cheng Hsiao for serving on my dissertation
committee. The completion of my dissertation owns a lot to the outstanding
experience in econometric theory and empirical application of Professor Hsiao. His
insightful comments and suggestions are greatly appreciated.
During my five years' study at USC, I have received numerous help and support
from the faculty, staff and fellow students at the Department of Marketing,
iii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Department of Accounting and Department of Economics. I am also very indebted to
all of them.
Finally, I would like to thank my family for being so supportive during my study
at USC. My parents' love and sacrifice have helped me go through tough times, and
will remain so throughout my life. My wife, who also finished her graduate study
and received her doctoral degree at USC, shares with me so much laughter and tears.
She is the source of energy encouraging me to face any challenges in life.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Contents
Dedication ii
Acknowledgments iii
List of Tables vii
List of Figures viii
Abstract ix
1. Introduction 1
2. Intellectual Property Protection: Strategic Marketing Considerations and
Financial Market Reaction 9
2.1. Introduction...........................................................................................................9
2.2. Literature Review...............................................................................................15
2.3. Conceptual Framework..................................................................................... 19
2.3.1. Factors Affecting The Probability of Litigation..............................22
2.3.2. Factors Affecting The Market Reaction to Litigation..................... 30
2.4. Empirical Analysis.............................................................................................33
2.4.1. Model Specification.............................................................................33
2.4.2. Description of The Data and Sample Construction..........................39
2.4.3. Variable Operationalization................................................................ 45
2.5. Empirical Results and Discussion....................................................................51
2.5.1. Summary of Results............................................................................. 51
2.5.2. Substantive Insights and Managerial Implication............................55
2.6. Conclusions and Future Research.....................................................................59
3. Product Diversification, Technology Diversification and Market
Performance in Pharmaceutical Industry 62
3.1. Introduction.........................................................................................................62
3.2. Literature Review...............................................................................................65
v
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.3. The Context: Pharmaceutical Industry and Drug Development.................. 67
3.4. Conceptual Framework......................................................................................69
3.5. Data Sources, Data Structure and Descriptive Statistics................................85
3.6. Variable Operationalization.............................................................................. 89
3.7. Empirical Analysis............................................................................................101
3.7.1. Selection of Empirical M odel............................................................101
3.7.2. Discussion of Empirical Results........................................................104
3.8. Conclusions........................................................................................................108
References 110
Appendix A. Estimating Marketing and R&D Capabilities: Stochastic
Frontier Estimation (SFE) 120
Appendix B. Standard Random Sampling and Choice-Based Sampling 126
Appendix C. Drug Classification System and Patent Classification System 130
VI
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Tables
2.1. Variables, Measures and Data Sources.......................................................... 40
2.2. Descriptive Statistics - Mean Value (Standard Deviation)..........................51
2.3. Analysis Results of Probability of Patent Litigation.....................................52
2.4. Analysis Results of Financial Market Reaction.............................................54
3.1. Summary of Major Hypotheses.......................................................................84
3.2. Descriptive Statistics of Major V ariables.................................................... 100
3.3. Estimation Results for Product Diversity Equation....................................104
3.4. Estimation Results for Technology Diversity Equation..............................105
3.5. Estimation Results for Market Performance Equation................................105
vii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
List of Figures
2.1. Conceptual Framework for Probability of Litigation....................................32
2.2. Conceptual Framework for Market Reaction.................................................32
3.1. Relationships Between Diversification and Their Impact on Financial
Performance........................................................................................................ 83
viii
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Abstract
This dissertation consists of three chapters. Chapter One outlines the motivation
and the structure of this dissertation. It also summarizes the major empirical
findings and managerial applications. Chapter Two constructs a conceptual
framework based upon Resource-Based View of the firm and applies Type II Tobit
Sample Selection model to investigate the importance of intellectual property
enforcement to firm’s marketing strategy. The empirical results indicate that the
level of firm’s marketing and R&D capability, the centrality of individual
technology to firm’s future product development/technology innovation trajectory
and the quality of intellectual property base generate significant impact on both its
decision to enforce intellectual property and financial market reaction. Chapter
Three focuses on the relationship between two different, but inherently correlated
diversification processes of firm, i.e. product diversification and technology
diversification. The financial consequence of these processes is also explored in
this chapter. The study is deployed on the basis of theories of evolutionary
economics and Resource-Based View of the firm. Dynamic panel data model is
applied to obtain empirical evidence. The results show that path-dependency exits
ix
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
in these processes, product diversification and technology diversification have
reciprocal impact on each other in the evolution process, and the level of diversity
increases with firm’s know-how about product and technology markets. These
factors are also found to have sound financial meanings to firm.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 1
Introduction
This dissertation consists of two essays on firm’s technology innovation and new
product development strategy.
The first essay, "Intellectual Property Protection: Strategic Marketing
Considerations and Financial Market Reaction", is an empirical analysis of
marketing implication in firm’s intellectual property enforcement.
In many technology-intensive markets, such as semiconductors, software,
pharmaceuticals and biotechnology, intellectual property (IP), which refers to
assets such as patents, trademarks, etc., accounts for as much as 70% of the firms’
value. The importance of intellectual property is enhanced by the facts that, in such
markets, successive generations of technologies and products build upon their
predecessors. Firms in these markets generate revenue through marketing their
intellectual property as well as converting it into products and services. Prior
literature has made the implicit assumption that if firm possesses intellectual
property, it will automatically be able to appropriate the value from its ownership.
1
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This is not bome out by industry trends. Patent infringement litigation, for example,
has increased by more than tenfold since the early 80's with much of the growth
occurring during the 90's.
The above trends seem to suggest that the gains from IP ownership are not
guaranteed unless firms have an enforcement strategy in place. Such a strategy is
not, however, without costs. Apart from direct costs ($1MM-$3MM on average),
litigation activities get a mixed reaction from the financial market (average change
of shareholder value of $67.9 million). Despite the important role that IP
enforcement activities, such as patent litigation, can play in aiding firms’ product
and competitive strategies, and the varied market reaction to these decisions, patent
litigation decisions have been left largely to lawyers. There has been recent
recognition in industry that patent litigation can be employed strategically as an
important mechanism to appropriate the value of a firm’s innovation and should be
influenced by strategic marketing considerations. The academic literature has,
however, not responded to this shift of emphasis on the part of firms. In particular,
there is no framework that helps understand differences in litigation strategy across
firms, or the differential market reaction to such strategies. This paper attempts to
fill this gap.
First, I attempt to demonstrate the importance of IP enforcement to marketing
strategy. To do this, I build a conceptual framework to bring in marketing explicitly
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
through incorporation of the firm’s marketing capabilities and through a
consideration of the role of IP in the firm’s future technology/product trajectories.
Second, I attempt to explain why firms vary in their propensity to litigate and why
market reaction varies to firm's patent infringement litigation. Resource-Based
View (RBV) of the firm has been applied as the theoretical base for conjecturing
about firm-specific differences. Finally, Type II Tobit Sample Selection Model is
employed to find empirical evidence on the framework by examining variation in
willingness to litigate, as well as variation of financial market reactions to such
activities, on a dataset of firms in high technology markets.
A number of interesting insights emerge from this analysis. First, increase in a
firm’s marketing capability reduces the likelihood of litigation. Further, the market
reacts negatively to litigation decisions of firms that have a high marketing
capability. This suggests that firms that have higher ability to take their products
and technologies to the market are better off maintaining that focus instead of
allocating resources to litigation activities. Similar results hold for the firm’s R&D
capability. These two results taken together highlight that senior marketing and
R&D managers have to play a more proactive role in their firms to prevent lawyers
from excessive allocation of resources towards litigation activities that may end up
reducing shareholder value. Similarly, the centrality of the IP in question to the
firm’s future product development/technological innovation trajectories is found to
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
be one of the most important factors influencing a firm’s propensity to litigate.
Finally, the market rewards litigation by firms which have a high quality IP base -
a reputation for toughness is especially useful for such firms. Analogous to any
brand-building exercise, this reputation has benefits that extend well beyond the
current period.
The second essay, “Product Diversification, Technology Diversification and
Market Performance in Pharmaceutical Industry”, attempts to find empirical
evidence on the patterns of product diversification and technology diversification in
pharmaceutical industry and the consequential financial impact.
This study is mainly motivated by a set of widely reported stylized facts. First,
pharmaceutical firms seem to differ greatly in the diversity of their product
offerings. For example, Aventis produces drugs targeting a relatively few number
of diseases, such as diabetes (Amaryl) and allergy (Allegra). On the other hand,
Merck, the flagship of the industry, has a much broader drug portfolio. It markets
drugs targeting cardiovascular (Cozzar), asthma (Singulair), cholesterol-control
(Zocor) and pain-control (Vioxx), etc. Second, pharmaceutical firms seem to differ
significantly in the diversity of their technology portfolios. For example, Pfizer has
comparatively greater technology profile. It has developed technology to make
diazine as an effective compound preventing infection, thyromimetic antiobesity as
the method useful in the treatment of obesity, Tetrahydroquinazoline-2-4-diones for
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
cardiovascular therapeutic uses, etc. Abbott Lab, on the other hand, is endowed
with relatively concentrated technology to produce polypropylene for inhalation
anesthetic treatment and alkynyl amides for psychiatric disorders control, etc.
The first fact, i.e., the distinct differences in firm's product portfolio, has been
the focus of a massive body of literature, encompassing Economics, Strategy and
Marketing. The second fact, i.e. the significant discrepancy in technological know
how has attracted considerably less attention by comparison. What's more, there
has been almost no work looking at the reciprocal impact of these two diversities
on each other (Stephan, 2002; Grantrand, 1998). Cursory evidence, especially in
high-technology markets, would suggest an intimate relation between a firm’s
technology diversity and its product diversity (Osterloff, 2003). The main task of
the paper is to explore this relationship and its impact on firm’s financial
performance.
There are many reasons why a more thorough examination of technology
diversity and product diversity is in order. First, there seems to be no obvious
pattern to technological and product explorations, i.e., there seem to be firms that
have technologies in a number of areas, but products in much fewer areas (e.g.
AstraZeneca). The reverse holds for some other firms (e.g. Pharmacia). Arguably,
each of these groups of firms suggests different strategic choices by firms. Second,
the nature of the diversities is important. Whether entering diverse product markets
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
generates the kind of market pull influences that translate to a diverse technological
portfolio, or the other way around, i.e. technology push effect (Granstrand et al.,
1990) is of great managerial implication.
In all of this, an important point needs to be kept in sight, and that is the need
for a thorough exploration of the product and technology diversity issues within
one comprehensive framework. This research does this in two ways. First, the
conceptual framework recognizes the dependencies between technology and
product diversity, as well as the points of departure between them, by appropriately
specifying the firm-specific resources and capabilities that would affect each
strategy. What's more, "path dependency" of the two decisions is taken into account
explicitly. Second, econometric specification estimates the impact of various
factors on product and technological diversity. Beyond that, how the diversities and
the interaction between them, together with firm-specific tangible and intangible
resources, and capabilities, affect financial market valuation is further explored. All
the arguments are presented in the context of the pharmaceutical industry, which
provides the location for this empirical investigation.
The empirical investigation supplies strong evidence on the existence of
reciprocal effect of these diversification processes on each other. It also finds
significant path dependency in the trajectory of product and technology
diversification. Firm's knowledge of product and technology market cumulated
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
over time and the endowment of high capability in marketing and R&D affect the
diversification process significantly. Financial valuation reacts positively to these
factors.
The dissertation proceeds as follows. Chapter 2, "Intellectual Property
Protection: Strategic Marketing Considerations and Financial Market Reaction",
has six sections. Section 2.1 gives the introduction of this essay. Section 2.2
discusses the literature related to this research. Section 2.3 develops conceptual
framework and research hypotheses. Section 2.4 describes the data sources, the
operationalization of major variables and the selection and specification of the
empirical model. Section 2.5 discusses the empirical results. The managerial
implications and suggestions for future research are discussed in Section 2.6.
Chapter 3, “Product Diversification, Technology Diversification and Market
Performance in Pharmaceutical Industry”, consists of eight sections. Section 3.1
gives an overall introduction of this essay. Section 3.2 reviews the relevant prior
literature. Section 3.3 provides some basic information on pharmaceutical industry
and drug development. Section 3.4 lays out a conceptual framework and develops
research hypotheses. Section 3.5 introduces information on data sources, data
structure and gives general descriptive statistics. Variable operationalization is
given in Section 3.6, and the choice of empirical model and empirical results are
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
discussed in Section 3.7. Section 3.8 elaborates the managerial implications and
future extensions to this study and concludes this essay.
8
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 2
Intellectual Protection: Strategic Marketing Considerations and
Financial Market Reaction
2.1. Introduction
High technology markets are marked by a high rate of innovation, with a high
priority placed on R&D. In such markets, both the academic literature and the
popular press (e.g. Fortune, 2002) have pointed to the pre-eminent role played by
intellectual property (IP)1 , rather than physical assets, as the principal source of
competitive advantage. Not surprisingly, the acquisition and management of
intellectual property rights in general, and patents in particular, has become an
important element of a firm’s innovation management strategy. A recent study
(Koen, 1991) found that 98 percent of large technology firms and 87 percent of
1 Intellectual property includes not only the output of R&D, but also general business know-how
that a firm has developed in the course of running its business. It refers to the propriety knowledge
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
small firms felt that intellectual property protection was an important consideration
in their corporate strategy.
The importance of IP is enhanced by the facts that in most high-technology
markets, successive generations of technologies and products build upon their
predecessors (John et al., 1999). Thus, a firm’s current technological know-how
(IP) can play an important role in the firm’s future product and technology
development strategy. Further, the role of intellectual property like patents or
trademarks in enabling firms to accrue first mover advantage or generate higher
profits through pricing or licensing their technology has been recognized in the
marketing strategy literature (Robertson et al., 1995; Alpert, 1993; Kerin et al.,
1992). Interestingly, there seems to be an implicit assumption that if firms possess
IP, they will automatically be able to appropriate the value from their ownership. In
other words, firms that own the IP will continue to have unfettered flexibility in
determining how best to use it for future product development, with no attempts by
other firms to infringe on their IP.
The assumption that ownership of patents or trademarks confers automatic
protection is not borne out by industry evidence in recent years. The inter
dependence among firms in developing new technologies and the role of IP in
enhancing firms’ competitive advantage through innovations have led to significant
expressed as a recipe, formula, trade secret, invention program or process. Patents, copyrights and
trademarks are the typical formats.
10
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
increases in IP infringement (Rivette et al., 2000). For instance, in recent decades,
there has been significant increase in patent infringement and it has been
accompanied by a concomitant increase in the prosecution of such violations
through litigation (Lanjouw et al., 2000; Lemer, 1995). Patent infringement
litigation has increased by more than tenfold since the early 80’s, with much of this
growth occurring during the 90's. Patent litigation now constitutes over 60% of IP
enforcement activities in technology intensive markets like electronics and
semiconductors (Rivette et al., 2000).
The above trends seem to suggest that the gains from IP ownership are not
guaranteed unless firms have an enforcement strategy in place. However, IP
protection can be costly. For instance, patent litigation incurs high direct cost.
Some case estimates have reported that direct legal costs for each side can range
from $1.00MM-$3.00MM in 1997 dollars (AIPLA, 1997). It also imposes high
indirect costs on the firm, since the process absorbs the time and energy of key
managers and technical personnel (Somaya, 2001). Interestingly, financial markets
seem to pay close attention to these patent litigation activities. As highlighted in the
literature (Lemer, 1995), within two days of a report of involvement in a patent
litigation, a company's market value could vary by as much as 3.1%, with an
average change of shareholder's value of $67.9 million. The market thus recognizes
the perils of a ‘shoot first and ask questions later’ approach to IP protection through
11
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
patent litigation. Clearly, as in the case of any other strategic activity, there are
times to hold and times to fold.2
Despite the important role IP enforcement activities, patent litigation in
particular, can play in aiding firms’ product and competitive strategies, and the
varied market reaction to these decisions, patent litigation decisions have been left
largely to lawyers (Rivette et al., 2000). However, there has been increasing
recognition that patent litigation can be employed strategically as an important
mechanism to appropriate the value of a firm’s innovation and requires active
participation by senior managers in different functional areas within the firm
(Berman, 2002; Brandt, 2002). Exploiting and protecting IP assets has become less
focused on technology, but more influenced by strategic marketing considerations.
This is brought out saliently by a recent comment from Broadcom, on patent
litigation initiated against it by Intel, “Intel’s claims against our product were
driven by marketing concerns rather than legal ones” (CNET News, 2000). The
academic literature has however not responded to this shift of emphasis on the part
of firms.
Our paper attempts to fill this gap in marketing literature. Our aim is three-fold.
First, we attempt to demonstrate the importance of IP enforcement to marketing
strategy. To do this, we build a conceptual framework that brings in marketing
2 Our focus in this paper is on owners of IP choosing to initiate litigation relating to the infringement
of their IP. Unless specified otherwise, litigation strategies refer to the strategies of such firms, not
12
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
explicitly, through incorporation of the firm’s marketing capabilities and through a
consideration of the role of IP in the firm’s future technology/product trajectories.
Second, we attempt to explain why firms vary in their propensity to litigate and
why market reacts differently to firm's patent litigation activities. Our use of the
Resource-Based View (RBV) of the firm as a theory base is explicitly tied to this
motivation, since the main focus of RBV is to explain how firm strategy is
influenced by heterogeneity in their endowment of resources and capabilities.
Finally, we offer empirical evidence on our framework by examining variation in
willingness to litigate, as well as variation of financial market reactions to such
activities, on a dataset of firms in high technology markets.
A number of interesting insights emerge from our empirical analysis. First, we
find that increases in a firm’s marketing capability reduce the likelihood of
litigation. Further, the market reacts negatively to litigation decisions of firms that
have a high marketing capability. This suggests that firms that have higher ability
to take their products and technologies to the market are better off maintaining that
focus instead of allocating resources to litigation activities. Similar results hold for
the firm’s R&D capability. These two results taken together highlight that senior
marketing and R&D managers have to play a more proactive role in their firms to
of the firms being sued for infringement.
13
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
prevent lawyers from excessive allocation of resources towards litigation activities
that may end up reducing shareholder value.
A second insight is that markets react positively to litigation by firms that have
IP of high quality. We conjecture that enforcement action by such firms helps build
a reputation that has positive spillover effects for the firm beyond the specific
technology that is being litigated. This result highlights the interesting difference
between resources and capabilities that is at the heart of the RBV. Resources (even
technology resources) are more easily imitable - firms with a superior quality
resource base therefore would try their best to protect this base. Conditional on the
quality of this resource base, however, firms with superior capabilities in coming
up with innovative technologies have higher opportunity costs in diverting
resources to litigation rather than R&D. The opposing findings on technology
resources versus technology capabilities verify this fundamental intuition of the
RBV.
Finally, the likelihood of litigation of IP goes up as its role in the firm’s future
product and technology development strategy becomes more critical. All the results
taken together highlight the key role of firm specific resources and capabilities in
enabling firms to create and appropriate shareholder value through strategic
protection of their intellectual property. Importantly, our results go to the heart of
14
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
RBV theory in pointing out the differential role of resources and capabilities in
influencing firm’s strategy.
2.2. Literature Review
Our study draws on, and contributes to, a number of related research streams in
Marketing, Economics and Strategic Management.
In marketing, our work is most closely tied to the literature on technological
know-how and product innovation in high-technology markets. John et al. (1999)
suggest that in most high-technology markets, successive generations of
technologies and products build upon their predecessors. This makes IP crucial to
the firm’s future product and technology development strategy. The importance of
such path dependence is clearly recognized in the marketing literature (Alpert
1993). We explicitly draw on this literature by developing constructs that take into
account the role of IP in future technology/product strategies. However, this
literature makes the implicit assumption that firms, which own the IP, will continue
to have unfettered flexibility in determining how best to use it and there will be no
attempts by other firms to infringe on their IP. This assumption is clearly not borne
out by actual trends in high-technology industries, as the number of patent
infringement suits has risen almost 10 times since 1978 (Lanjouw et al., 2001). Our
15
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
paper contributes to the literature by questioning this assumption, and examining
differences in the enforcement strategies of firms.
Second, we draw on literature that looks at the link between a firm’s market
orientation and the process of innovation. There has been some work (Dutta et al.,
1999) that has suggested that a firm’s marketing capability is critical to enhancing
the firm’s ability to generate valuable intellectual property in technology intensive
markets. Further, this stream of research also highlights the important role played
by innovation in the market orientation/profitability link (Han et al., 1998; Hurley
et al., 1998). Finally, there is some literature in marketing that has looked at the
role played by IP in enabling firms to accrue first mover advantage or generate
higher profits through pricing or licensing of technology (Robertson et al., 1995;
Alpert, 1993; Kerin et al., 1992).
However, in the above literature there is no conceptualization and evidence of
the role that marketing can play in protecting a firm’s intellectual property and how
it influences shareholder wealth creation. Our work extends these streams of
research in marketing by suggesting that IP ownership by itself may not enable
firms to appropriate its full value unless firms take steps to minimize infringement
of their IP. In particular, we articulate the role that a firm’s marketing capability
can play in the firm’s decision to protect its intellectual property through patent
litigation, and its role in influencing market reaction when firm does decide to
16
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
litigate infringements. Our focus on financial market evaluation of the firm’s patent
litigation decision adds to the relatively sparse literature in marketing (Srivastava et
al., 1998) which links the role of marketing activities and resources to the creation
of shareholder's value (Sorescu et al., 2001; Agrawal et al., 1995). Finally, our
framework and evidence builds on prior literature in marketing that has studied
how appropriability conditions vary across industries (Boulding et al., 1995), by
suggesting how appropriability can be enhanced through strategic enforcement of
patent infringements.
There is a significant number of literature in the field of law and economics that
has addressed the issue of litigation generally and the issue of patent litigation
specifically. Cooter and Rubenfeld (1989) discuss the determinants of litigation
generally, while Priest and Klein (1984) and Bebchuk (1984) build stylized models
to explain the determinants of litigation outcomes. Theoretical models that have
specifically looked at patent litigation have pointed out that patent litigation could
serve as a mechanism to transmit information (Jain, 2001; Choi, 1998). We
incorporate the theoretical arguments suggested by this literature in our conceptual
framework. Empirically, there have been a few papers examining the determinants
and outcomes of litigation in general (Waldfogel, 1995), and patent litigation in
particular (Somaya, 2001; Lanjouw et al., 2000; Lemer, 1995). The focus of these
17
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
studies has been on patent characteristics that might influence the propensity of
patent litigation.
Strikingly, almost none of the law and economics literature cited above has
focused on why firms would differ in their litigation strategies, except for broad
reasons such as firm size. Our focus, different from that of prior studies, is
precisely on how firm-specific capabilities and resources, marketing and R&D
capabilities and the value of a firm’s technological resources in particular, influence
its propensity of litigation. We thus explicitly tie litigation strategy to the
constraints imposed by hitherto unexamined capabilities that are of central
importance to marketing. We further extend the literature by examining the impact
of litigation on market reactions and the creation or destruction of shareholder
wealth, rather than on court outcomes. This has the benefit of focusing on outcomes
of more direct relevance to marketers. It also helps us avoid a problem endemic to
past literature in that litigation outcomes are (a) often private and hence
unobservable to third parties, and (b) often occur considerably after the filing of
litigation.
The discussion above provides a natural segue into the third major stream of
literature, which serves as our theoretical base, namely the Resource-Based View
(RBV) of the firm (Wemerfelt, 1984; Teece, 1980). Since the central tenets of the
RBV are about the heterogeneity of firms in resources and capabilities, and the
18
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
impact these differences have on their strategies, the RBV is a natural theory base
for us. We contribute to this literature by being the first study to bring RBV into the
domain of enforcement of intellectual property. A number of the resources that we
suggest in our framework are therefore new to the RBV literature, since they are
tailored to our institutional context. We also distinguish between resources and
capabilities in a manner consistent with the RBV framework and which addresses
criticisms of researchers like Williamson (1999). Finally, the careful distinction of
resources and capabilities, both in the theorizing and in operationalization, leads to
interesting payoffs in our results. For instance, we find opposite effects for the
impact of resources and capabilities on a firm’s propensity to litigate, a result
puzzling at first glance, but entirely consistent with the RBV.
2.3. Conceptual Framework
Before we delve into the conceptual framework, it is useful to lay out the broad
question we are trying to address. At the outset, it should be recognized that the
purpose of IP rights is to confer a legal monopoly to the recipient firm. Every firm
would like to extract as much value as possible from this monopoly. And every
firm, would, ostensibly, fight against anything that leads to the dilution of this
monopoly. Asking why firms differ in their IP litigation strategies therefore
19
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
amounts to asking why firms would be more or less motivated to protect this legal
monopoly. The obvious answer is that firms would weigh the net benefits of
protecting their monopoly rights when they decide on a course of action - it is not
obvious, however, why these net benefits would differ across firms. In other words,
are the monopoly rights conferred upon firms differentially attractive, and if yes,
what factors cause them to be so?
Since our basic premise is that there are firm-specific differences that cause
firms to adopt different litigation strategies and markets to respond differently to
such strategies, a natural theory to turn to for building our conceptual framework is
the RBV. According to the RBV, firms are defined by their bundles of resources
and capabilities (Day, 1994; Rumelt, 1984; Wemerfelt, 1984) that are distributed
heterogeneously across firms and constitute the determinants of competitive
position and advantages in the industry. A number of tenets of the RBV are key to
our conceptual framework. The first is the notion of resources and capabilities. A
firm’s resources are productive factors that a firm uses to achieve its business
objectives (Wemerfelt, 1984). They can be viewed as assets conferring competitive
advantages to the firm - examples would include marketing research expenditure
and patents. By contrast, firm capabilities represent the ability of the firm to
combine efficiently a number of resources to engage in productive activity and
attain a certain objective (Amit et al., 1993). Thus one can think of capabilities as
20
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the efficiency with which a firm uses the inputs available to it (i.e., its resources,
such as R&D expenditure) and converts them into whatever output(s) it desires
(i.e., its objectives, such as developing innovative technologies). In the RBV
literature, firm capabilities are clearly an “intermediate transformation ability”
between resources (i.e., inputs) and objectives. Capabilities, being this
transformation ability, rely on a host of tacit firm-specific routines and resources,
which makes them hard to imitate or to buy. In turn, this lack of transferability and
inimitability create ex-ante and ex-post barriers to competition, and enable
capabilities to act as sources of competitive advantage. The distinction between
resources and capabilities is key to both the RBV and to our conceptual framework,
as we shall discuss shortly.
The second key tenet is that firms differ in their endowments of resources and
capabilities. These differences are what enable resources and capabilities to serve
as sources of competitive advantage. And, these differences in turn guide and
constrain firm strategies, specifically the IP litigation strategies in our case.
The third is firms whose strategies are a better fit with their resources and
capabilities gain competitive advantage (Teece, 1980). This third point, in fact,
offers one way of testing the RBV. What is needed first is an appropriate definition
of resources and capabilities suitable to the institutional context being studied.
Then one hypothesizes on what strategies are appropriate for a given
21
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
resource/capability configuration. And finally, one tests if firms follow strategies
(as hypothesized above) appropriate to their resources and capabilities do better
(e.g., are rewarded by the market). We follow this logic exactly, which leads us to
two sets of hypotheses and two equations to test them. The first set consists of
explaining the specific resources and capabilities that affect a firm’s litigation
strategy, and the second set consists of hypotheses on the market reactions to such
strategies. We discuss each set in turn. Within each set, in discussing the firm-
specific factors that affect the outcomes, we start with a discussion of firm-specific
capabilities, followed by a discussion of firm-specific resources.
2.3.1. Factors Affecting The Probability of Litigation
Firm-Specific Capabilities:
Marketing Capability: Marketing capability in the RBV framework refers to the
efficiency with which a firm identifies market opportunities, gathers competitive
intelligence, and utilizes its resources such as technological base and marketing
resources (e.g., advertising and marketing research expenditure) to achieve its
marketing goals (Dutta et al., 1999; Day, 1994). The role of marketing in high-
technology markets has been discussed extensively in the literature (Griffin et al.,
1993; Gupta et al., 1986). Right from shaping technological trajectories through
22
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
customer input, to creating demand-side advantages for the finished product of
such technology, marketing capability is integral to the success of high-technology
firms. Specifically for our case, IP is the building block of both technological and
product strategies, which are central to marketing. Clearly, marketing and IP are
closely tied together; the question is how a firm’s marketing capabilities would
affect its decision to litigate IP.
Why does marketing capability have an influence in IP enforcement? The
opportunity costs for a firm with a high marketing capability that engages in patent
litigation are high. Such a firm could profitably engage in marketing related
activities at which it is good, rather than wasting resources on litigation. Two
pieces of evidence support this conjecture. Berman (2002), Brandt (2002), and
AIPLA (1997) report that litigation imposes not only a direct cost on the firm but
also imposes high indirect costs on the firm. It requires senior managers from
different functional areas like R&D, marketing and corporate strategy to allocate
their time to the litigation decision when they could be allocating their efforts to
other strategic endeavors. Somaya (2001) further discusses how disruptive
litigation is to the normal activities of a firm. Also, two surveys across industries
(Cohen et al., 2000; Levin et al., 1987) asking managers about how they can protect
and exploit their intellectual property highlight the relative superiority of lead-time
and speed to market over patent litigation as a source of competitive advantage.
23
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
This evidence seems to suggest that if firms are strong at taking their products to
the market faster and are more efficient than their competitors they are better off
focusing on that dimension to maintain their competitive advantage. It is important
to realize that exploiting these market driven strategies is disproportionately
beneficial to firms with a higher marketing capability and this capability relies on
factors that are firm specific and less imitable.
The arguments above suggest that a firm with higher marketing capability will
be less vulnerable to infringement threats, since it knows that it has the ability to
deliver the product more efficiently and ensure a better match with the benefits
sought by the customer. This suggests:
HI a: The higher a firm’s marketing capability, the lower the likelihood that the
firm will litigate patent infringements
R&D Capability: R&D capability refers to the efficiency of the firm in utilizing
its resources, such as R&D investment, prior technology base, etc., to develop
innovative technologies (Dutta et al., 1999). Firms with higher R&D capability are
better able to come up with innovations that are of higher quality; in other words,
the marginal productivity of their R&D investment is higher. As in the case of
marketing, R&D capability is hard to imitate or buy in the market.
The arguments for R&D capability are similar to those for marketing capability.
The opportunity costs for a firm with a high R&D capability that engages in patent
24
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
litigation are high - the firm could profitably engage in developing new generations
of innovative products and technologies instead of litigating. In particular, in
technology intensive environments it is important for firms to be able to generate
new generations of innovations and sometimes even leapfrog existing technologies
(Purohit, 1994). In such cases, a firm that is good at coming up with repeated
innovations is less reliant on current technology than a firm that finds such repeated
innovation hard. Thus, a firm with higher R&D capability should be less
susceptible to infringements of its technologies and is better of focusing efforts to
continue its development of newer generations of innovations and products. It is
important to realize that exploiting these innovative paths is disproportionately
beneficial to firms with a higher R&D capability and this capability relies on
factors that are firm specific and less imitable. We thus hypothesize:
H2a: The higher a firm’s R&D capability, the lower the likelihood that the firm
will litigate patent infringements
Firm-Specific Resources:
While firm-specific capabilities, are central to what causes heterogeneity across
firms in litigation strategy, differences in such strategy could also be caused by the
resources that the firm possesses. A resource, as defined earlier, is an asset owned
by the firm that would be hard to imitate or acquire in the short run (Amit et al.,
25
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
1993). In our context, the most important resource a firm possesses is its base of
technology. 3
Technology Resource Base: We suggest that firms with a superior technology
resource base are more likely to litigate IP . 4 The reason for this goes back to the
need to signal a reputation for tough enforcement of IP rights. In the markets we
are interested in, firms operate in conditions of rapid technological change and high
uncertainty. Given the interrelated nature of technologies in such industries and the
complexity of products embodying those technologies, a firm frequently finds itself
needing to make decisions on whether a certain path would lead to infringement of
some other firm’s IP (Hall et al., 1999). At this point, it has to take into account the
probability that the firm whose rights it infringes upon will litigate - the merits of
the case aside, what matters here is clearly the reputation of the infringed firm. We
are speaking here of a firm’s reputation for vigorous enforcement of its IP rights.
To the extent that such a reputation acts as an additional constraint on the decision
making of rival firms, it can delay their products or force them to make sub-optimal
technological decisions (Shankar et al., 1998). In this sense, this reputation is very
3 To reiterate a point made earlier, the firm’s technological resource base is different from its
marginal productivity in coming up with more innovations (referred to earlier as its R&D
capability). These two are conceptually very different, corresponding to the distinct RBV concepts
of resources and capabilities respectively. The veriable operationalization explicitly brings out this
distinction.
4 It is important to note that this effect is over and above the effect caused by the quality of the
individual patent that might be the subject of dispute. Given an identical patent, a firm with a high
quality technological base is more likely to litigate it than a firm with a low quality technological
base.
26
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
similar to any brand-building exercise - its effects last well beyond the current
period and enable the firm to capture value more efficiently.
Such a signal is most useful for firms that have a valuable IP portfolio, because
they are (a) most likely to see infringement of their rights (i.e., the probability of
infringement goes up), and (b) have the most value to gain from litigation due to
spill-over effects of that reputation. In other words, the expected loss from
infringement is much greater for firms with a superior technological base if they do
not litigate infringements of their patents, since it could encourage other firms to
infringe other valuable technologies in their portfolio. By aggressively litigating
infringements, the firm deters future infringement and thus gains from both
enhanced value capture and a simple saving of resources expended on future
litigation. We thus hypothesize that:
H3a: The more valuable the technological base of a firm, the higher its
propensity to engage in IP litigation
While the above has talked of the value of a firm’s overall technological base as
a resource, quite clearly the value of the individual technology itself should
influence a firm’s desire to litigate any infringements of it. In what follows, we
discuss different dimensions of the value of an individual technology - the
centrality of the technology to its future product development trajectory, and the
breadth of applicability of the technology.
27
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Centrality o f Individual Technology: Often times, certain specific technologies
are crucial to the future technological trajectory the firm sees itself taking. This
notion of path dependence, which is central to the RBV, imparts a firm-specificity
to the value. Further specificity to the value is provided by the fact that the firm has
a host of complementary resources and capabilities surrounding this particular IP.
Thus, the firm might use a complementary capability, such as marketing, to acquire
competitive advantage in a product stream - Intel’s initial advantage in
microprocessor architecture, in combination with its successful branding efforts,
have led to advantages that are yet to erode. As Teece (1988) suggests, protecting
the firm’s IP also protects investments in complementary resources, and helps the
firm develop capabilities that are hard to imitate or buy. Somaya (2001) points to
the example of Eli Lilly, whose vigorous exploitation of its insulin related IP
helped it reap the benefits even when a biotech version of the product came out.
The firm-specificity implies that even if the IP were to be transferred to another
firm, it might not have the same marginal productivity. This high firm-specific
valuation of individual technology, of course, increases the likelihood of litigation
over tit once in question. We therefore hypothesize:
28
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
H4: The higher the centrality of the technology to the firm’s future product
development /technology innovation trajectory, the higher the probability of it
being litigated
Breadth o f Application o f Individual Technology: The broader the applicability
of the technology, the more the revenue the firm could expect to get from both
licensing and embodying in its products (Lanjouw et al., 2000; Lemer, 1994).
While this is a resource, it is not clear how firm specific the technology is.
Regardless of the firm-specificity, our hypothesizing remains unchanged - we
suggest that the broader the scope of an individual technology, the higher the
likelihood of it being litigated. Specifically,
H5a: The broader the applicability of a technology, the higher the probability of
it being litigated
Litigation History: Prior literature suggests that there is a substantial "learning
curve" in patent litigation (Lemer, 1995). The existence of this learning curve
suggests that firms with more litigation experience are likely to be more
knowledgeable about the patent litigation process, understanding the nature of
coordination needed across different groups in the firm, and better at acquiring
information needed about potential violators in order to proceed with litigation.
Thus they are likely to be more efficient at the task. This suggests litigation costs for
such firms would be lower, making them more likely to litigate patent. We hypothesize:
29
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
H 6 a: The higher the experience of the firm in patent litigation, the higher the
likelihood that the firm will litigate patent infringements
2.3.2. Factors Affecting the Market Reaction to Litigation
We had suggested earlier that firms, which follow a litigation strategy
consistent with their resources and capabilities, should gain competitive advantage.
Our second set of hypotheses accordingly deals with the market reaction to the
litigation decisions of firms. If the litigation strategies suggested by our first set of
hypotheses are indeed correct, the market should reward firms following these
strategies. Not surprisingly, therefore, our second set of hypotheses is almost
identical to the first one (i.e., same directionality). The only important distinction
between the two sets relates to the important idea of the centrality of an individual
technology to the firm. We argue that to the extent the centrality of this IP to the
firm’s future product/technological strategy is specific to the firm itself, the market
at large is not likely to possess this information. This is in keeping with literature
that suggests the existence of private information possessed by firms (Eckbo et al.,
1990). The implication is that the market cannot take this information into account
when deciding a reaction to the firm’s litigation decision. 5
5 Of course there are caveats. The market obviously has some conjectures about the possible
importance of the E P to the firm’s future strategies. In the absence of compelling arguments as to
what these conjectures are based on, the most parsimonious way to account for the fact that this is
private information is to not include it in the market reaction process.
30
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The second set of hypotheses then lay out as follows (the numbering is to
maintain consistency with the first set of hypotheses):
Hlb: The higher a firm’s marketing capability, the lower the increment in the
market value to the firm’s litigation decision
H2b: The higher a firm’s R&D capability, the lower the increment in the
market value to the firm’s litigation decision
H3b: The more valuable the technological base of a firm, the higher the
increment in the market value from the firm’s litigation decision
H5b: The broader the applicability of a technology, the higher the increment in
the firm’s market value from its litigation decision
H6 b: The higher the experience of the firm in patent litigation, the higher the
increment in the market value from the firm’s litigation decision
Figure 2.1-2.2 show direction of impact of determinant variables on probability of
litigation and financial market reaction. The signs indicate the hypothesized
directionality of relationship of interest.
31
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Figure 2.1 Conceptual Framework for Probability of Litigation
Probability of Patent
Litigation
Characteristics of Individual Patent:
1. Breadth of Applicability (+)
2. Centrality to Product
Development trajectory (+)
Control Variables:
1. Market Concentration (+)
2. Firm Size (+)
3. Litigation History (+)
Firm Specific Resource and
Capabilities:
1. Marketing Capability (-)
2. R&D Capability (-)
3. Tech Resource Base (+)
Figure 2.2 Conceptual Framework for Market Reaction
Market Reaction to
Patent Litigation
Characteristic of Individual Patent:
1. Breadth of Applicability (+)
Control Variables:
1. Relative Firm Size (+)
2. Market Overlap (+)
3. Case Size (+)
Firm Specific Resource and
Capabilities:
1. Marketing Capability (-)
2. R&D Capability (-)
3. Tech Resource Base (+)
32
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.4. Empirical Analysis
We start this section with a discussion of our model specification and how it deals
with the sample selection issue inherent in our data. Since the construction of the
sample for our analysis is a non-obvious task, we deal with this in some detail.
Following this we discuss our data set and detail the operationalization of the
variables.
2.4.1. Model Specification
One obvious way to test our hypotheses is as follows. To test the impact of
factors that affect the probability of litigation, one can regress the occurrence of
litigation from a sample of litigated and non-litigated patents on these factors. And
to test for market reaction, one can regress market reaction on the relevant factors,
for the sample of litigated patents only. As far as the second equation goes, this is
indeed how most market reaction studies have been done in the literature (Houston
et al., 2000; Horsky et al., 1987). However, considering only the sample of litigated
patents is equivalent to treating the litigation decision as exogenous to the firm.
Researchers (Eckbo et al., 1990; Acharya, 1988) have argued that if firms are
rational, voluntary corporate actions like litigation are an outcome of endogenous
firm decisions. Ignoring this endogeneity will lead to sample selection problem
33
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
resulting in biased estimates (Kyriazidou, 1997). Formally, in the process of
litigation decision-making, firms expect to obtain private benefits or utility (y*)
from patent enforcement. No litigation will be filed and hence no change in market
value is observed unless y* is above a certain threshold, which is unobservable to
the public (hence the econometrician). The standard solution to this type of
selectivity problem relies upon an auxiliary model of the process generating y *
(Greene, 1999).
As it turns out, in our case the process that generates y* is precisely what the
first part of our conceptual framework is about. There we speculated on the firm-
specific resources and capabilities that would affect the utility a firm would derive
from litigating IP, and hence determine the probability that IP would be litigated.
Therefore, the sample selection methodology is a perfect fit for testing our
conceptual framework.
To account for the endogeneity of the litigation decision, we resort to Type II
Tobit Sample Selection Model (Amemiya, 1985). This method considers market
reaction to litigation, conditional on a firm’s choice of litigation decision, and
explicitly models this latter choice as a function of various factors (effectively, our
first set of hypotheses).
34
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Firm’s decision to litigate is models by the following equation. Formally, if the
decision to litigate is denoted by L/7) ( 1 denoting litigation and 0 denoting no
litigation) we have:
patent, Xj is a vector of covariates that explains the litigation decision, and s is the
error term satisfying N(0, a \) . The following equation captures our hypotheses
relating to the propensity of litigation:
prob(LITj = 1) = prob{yj > 0) = prob(X f j3 + e . > 0)
Denoting the market reaction to litigation by CARj, we have the second stage of the
sample selection model as:
CAR, = CAR* if LIT, = 1
J 3 J J
where C A R *-W jY + j U j
W j is a vector of covariates that explain the market reaction, y is a vector of
associated coefficients, and // is the error term satisfying N(0, < j 2 m ). CAR* is the
*
where yj - X j ' p + £j
y f is a latent variable measuring the net private valuation of litigation of the / h
35
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
change of market value of the plaintiff firm due to filing of patent litigation, which
is observable to the public. This equation captures our hypotheses relating to the
market reaction to litigation (effectively, our second set of hypotheses).
We further assume that error terms e and // are distributed bivariate normally,
< p and are respectively the density function and distribution function of the
standard normal distribution evaluated at {X'j/3)l<7e • The term A above is known as
the inverse Mill's Ratio. If it is significantly different from zero, sample selection
becomes an issue, and a standard linear regression of CARj on the independent
variables Wj alone will generate inconsistent estimates of parameters. This relates to
Putting this together, we get:
CARj = CAR* | (l ITj = l) = e (cAR* \ LITj =1 ) + Vj =W j ’r + p a mX{aie) + v j
E(CAR* | LITf = 1) = E(CAR* \ y s * > 0)
= E (C A R 'j\X 'j0 + £j > 0)
= W 'j y + E(/zj \£] >-X*j3)
= W'jY + p a jtlA (a j£)
where, v is error term satisfying normal distribution, and
36
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the committing of missing variable specification error. The significance of X is thus an
indicator of the appropriateness of the current model specification.
Estimation of the parameters in the market reaction regression is done using
Heckman's two-stage approach. In stage one, we use a Probit model to get the
estimates of filoe , and hence the value of inverse Mill's ratio X (a e) . In stage two,
we apply linear regression of CARj on W , and X(a£)io estimate y and the
coefficient of inverse Mill's Ratio, which is p a E.
To sum, our estimation equations are as follow:
LITj = j3o + /?/*Marketing Capability + /%*R&D Capability + /%*Tech Resource
Base + f y *Centrality + /?5 *Breadth + /?6 *Litigation History + /?z*Firm Size +
/?s*Market Concentration +£
CAR* = yo + ^M ark etin g Capability + ^*R&D Capability + j^T ech
Resource Base + Breadth + %*Case Size +^7*Relative Firm Size + f§*Market
Overlap + v
We now provide details on the estimation of market reaction, which is stage
two of the sample selection model. We analyze market reaction using the event
study method that has been commonly used to determine the financial consequence
of a firm's marketing decisions, corporate takeovers, change of firm name, etc.
37
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(Houston et al., 2000; Horsky et al., 1987). This is in line with current marketing
research (Srivastava et al., 1998), which argues for examining the impact of
marketing actions on shareholder value. The event study method builds upon the
efficient markets theory (Campbell et al., 1997; Fama, 1991), which posits that all
information publicly available is incorporated into the stock price of a firm.
Similarly, the expected cash flows, positive or negative from the litigation, are
incorporated into the stock price of litigating firm. Any abnormal increases
(decreases) in stock prices therefore reflect the value of that particular litigation. In
essence, this amounts to using stock prices to calculate the present value of
expected future cash flows generated by litigation.
The CAR is measured by the product of three terms: (a) the cumulative
abnormal return over the 3-day event window6, where abnormal return is defined as
the difference between return on the firm's stock and return on the overall stock
market, (b) the stock price prior to the event window, and (c) shares outstanding
during the event window. (Sorescu et al., 2001; Campbell et al., 1997). Formally,
CAR = f d(Rl ~ R m)Pt_x Nsharest_x
r= 0
P + d
Rt is the rate of return on the firm’s stock, calculated as follows: R. = — -----' ■ — 1
P
1 t - 1
6 We use a window of three days around the announcement of the litigation decision, to assess the
response of the market. This approach has been commonly used in current literature (e.g., Fama,
1991).
38
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Pt is the stock price at time t
dt is the dividend per share paid by the firm on day t
Rm t is the rate of return of all publicly traded equities on the market
Nsharest-i is the number of outstanding shares that the firm has the day
before the announcement
2.4.2. Description of the Data and Sample Construction
Our conceptual framework posits hypotheses whose testing requires diverse
pieces of data. Thus, we need information on a firm’s resources (e.g., technological
base), its capabilities (e.g., marketing and R&D), technological factors (e.g., the
breadth of applicability of the patent), market factors (e.g., market concentration),
litigation details (e.g., was the patent litigated or not) and financial outcomes (e.g.,
stock market reaction). To this end, we have created a unique archival database
integrating information from different sources. Table 2.1 supplies a summary of the
variables and data sources. Our sample consists of 82 publicly traded US firms, in
the electrical and electronic sectors (with SIC 36) that have received at least 10
patents by 1999. Our analysis covers the time period 1991-1999. Two concerns
relating to the choice of our sample have to be addressed here. First, the focus is on
the firms with the same two-digit SIC code (i.e. same industry). The major reason
for choosing firms with the same SIC code is that prior studies (Cohen et al., 2000;
39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Levin et al., 1987) have pointed out that there exists significant variation in the
value and the role of patent protection across industries.
Table 2.1 Variables, Measures and Data Sources
Conceptual
Variable
Measured Variable Data Source
Market Reaction
Quality of Tech
Base
Marketing
Capability
R&D Capability
Centrality
Breadth
Litigation History
Firm Size
Market
Concentration
Case Size
Market Overlap
Abnormal Stock-Market Return Due
to Litigation
CRSP
Citation-Weighted Patent Stock USPTO, COS
Efficiency of Marketing Estimated
Using SEE
Efficiency of R&D Estimated Using
SFE
Forward Self-Citations Received by
Patent
Total Number of Claims Made by
Patent
Litigation Frequency Relative to
Industry Average
Total Assets of Litigating Firm
Herfindahl Index
Number of Patents Involved in
Litigation Case
Comparison of 4-Digit SIC Code
Between Involving Firms
COMPUSTAT,
USPTO, COS
COMPUSTAT,
USPTO, COS
COS
COS
Derwent Co.
COMPUSTAT
COMPUSTAT
Derwent Co.
COMPUSTAT
40
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Narrowing the scope of sample firms helps minimize these variations. This is
also consistent with our focus on examining the impact of firm-specific differences
in resources and capabilities on IP litigation strategies. Observe, however, that to
the extent that firms within an SIC are likely to be more homogenous than firms
across SICs, our sample provides a conservative test of the hypotheses.
The second concern relates to the choice of the particular SIC. We focused on
this particular SIC code for two reasons. First, we needed an industry where IP
accounted for a significant fraction of a firm’s value, which narrows our search to
‘high-technology’ industries. Second, the sector we choose should be featured that
IP is embodied largely in patents, which makes the measurement of IP and the
tracking of litigation activities associated with it very convenient (as opposed to the
software industry, where most IP value resides in trademarks and copyright).
We now describe the components of our data, dividing them into four broad
categories for convenience.
Litigation Information
Since this is a key piece of our empirical analysis, and since much of this data
is new to the marketing literature, we describe it in greater detail. We obtained
patent litigation information of the 82 sample firms, for the years 1991-1999 from a
proprietary database owned by Derwent Co. This database, which is constructed
41
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
from information collected by the U.S. Patent and Trademark Office (USPTO), has
been used by other researchers who have also explored patent litigation issues
(Somaya, 2001; Lanjouw et al., 2000). The database records patent litigation date,
the identities of the parties involved in litigation, and the patent numbers of
litigated patents. In all, we obtained litigation information on 122 patents.
As discussed in our model specification, it is erroneous to consider only the
sample of patents that have been litigated and check for the impact of this litigation
on market outcomes. In fact, as our conceptual framework highlights, various firm
and technological factors impact the probability of a patent being litigated - the
litigation event is therefore not exogenous to the firm. As suggested earlier, the
solution to correct for possible sample selection bias is to introduce an auxiliary
equation that explicitly models the probability of a patent being litigated.
Conditional on being litigated, we then examine market outcomes for litigated
patents. The issue then becomes the appropriate choose of sample patents for the
litigation probability equation, since clearly the market reaction equation can only
be run on litigated patents. A first cut is to run the litigation probability equation on
the universe of patents relevant to us. This is however an infeasible task, given that
there are over 100,000 patents for sample firms. As a possible solution, one could
pick randomly from the universe of patents that have not been litigate, and then run
the litigation probability equation. Consistent with past literature, we follow a
42
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
variation of this random-pick approach. This approach creates a control group that
n
consists of a ‘matched’ sample of patents that are not litigated .
Following Lanjouw and Schankerman (2000), we create a matched sample by
using three filters - the SIC code8, the issue year and the USPTO technology class.9
Thus, for each litigated patent that we have, we have a matched non-litigated patent that
comes from the same SIC code, belongs to the same technology class and has the same
issuance year. The restrictions we impose in the construction of our matched sample help
us control for any variation due to technology effects (some technology classes see waves
of litigation activity), cohort effects (there might be systematic age related effects on
probability of litigation), and industry effects (litigation behavior in semiconductors
being different from that in pharmaceuticals). Of the 122 litigated patents in the
sample, we were able to obtain matching non-litigated patents for 90. Thus our
patent sample comprises of 180 patents with half of them being litigated.
We should point out that constructing the sample in this fashion necessarily
over-weights the litigated patents (the probability of litigation in our sample is
50%, which is significantly higher than the population average of about 0.06).
Since the formation of our sample is choice-based rather than random, we correct it
appropriately in our estimation (Manski et al., 1981).
7 Refer to Appendix B for retail on the Choice-Based Sampling approach we apply here.
8 By this we mean that the non-litigated patent belonged to a firm within the same SIC code as the
firm holding the litigated patent - in our case, SIC code 36.
43
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Firm-Specific Financial and Accounting Information
Financial and accounting information for the sample, such as sales, R&D
investment, advertising expenditures, administrative expenditures, etc., is obtained
from the COMPUSTAT and First-Source databases.
Technology Information
With respect to IP, we collect information on a firm’s patent portfolio from the
USPTO and COS databases. This gives us information on the patent’s issuance
date, the patentee firm and the technology classes the patent is classified into. This
does not give us information, however, on the patent’s impact on the firm’s future
product trajectory - for this we have to do content analysis of the patent and engage
in a recursive search of other patents on the basis of the content analysis.
Financial Market Response
We collected information relating to financial market response to patent
litigation, such as abnormal market returns over an event window, stock prices and
outstanding share stock from the CRSP database.
9 The USPTO assigns one or more 9-digit technology classes to each patent. For example, Patent
5592111 (Intel Co.) has been assigned into four different technology classes. They are 327-045-000,
44
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
2.4.3. Variable Operationalization
Dependent Variables
Occurrence of Patent Litigation (LIT): This is represented as a dummy variable,
coded as 1 for a litigated patent and 0 for a matched non-litigated patent.
Market Reaction (CAR): Briefly, market reaction is measured by the changes of
shareholder value due to patent litigation over an event window of three days. In
order to control for the effects of unexpected leakage of litigation information to
the stock market, we define the day prior to the litigation filing date as the starting
day of the three-day event window.
Independent Variables
Marketing Capability (MKT_CAP): Following Amit and Shoemaker (1993),
we conceptualize a firm’s capability as an “intermediate transformation ability”
between resources (i.e., inputs) and business objectives. Since capabilities are an
intermediate step between resources and outputs, one can hope to see the inputs
that a firm uses and the outputs it achieves, but one can only infer its abilities in
converting one to the other. As such, we employ an econometric technique called
327-042-000, 327-291-000, 327-048-000.
45
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Stochastic Frontier Estimation (SFE) (Dutta et al., 1999) to infer capabilities based
on an observation of a firm’s inputs and outputs10.
Specifically, marketing capability is defined as the ability of the firm to
efficiently deploy resources, such as advertising expenditure, marketing research
expenditure, prior technology base, along with its installed customer base, to
maximize its sales, which is the objective we assume the firm wishes to achieve. A
firm, which is more efficient at transforming the resources above into the desired
objective (e.g., sales), is said to have a higher marketing capability. This can be
obtained by estimating the maximum sales the firm could have achieved. Then,
since we observe the actual sales it did achieve, we can estimate the shortfall of the
actual from the maximum. The smaller this shortfall, the higher the efficiency, and
hence the higher the marketing capability of the firm.
R&D Capability (RD_CAP): Our estimation of R&D capability follows
identical logic to that of marketing capability above. We suggest that the objective
of R&D activities is to maximize the production of innovative technologies
(measured using citation-weighted patent counts). The major resources that the firm
has at its disposal to fulfill this objective include current and past R&D
expenditure, and the extant technological knowledge base of the firm. A firm,
which transforms these resources more efficiently into the production of innovative
10 For further details on the methodology, please refer to the Appendix A.
46
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
technologies, is said to have a higher R&D capability. This can be obtained by
estimating the maximum innovative output the firm could have achieved. Since we
observe the actual innovative output the firm did achieve, we can estimate the
shortfall of the actual from the maximum. The smaller this shortfall, the higher the
efficiency and hence the higher the R&D capability of the firm.
Technological Resource Base (TECH_RESOURCE): This measure is obtained
by summing up the citation-weighted value of each of the firm’s patents. The need
for this weighting arises because there exists considerable variation in the value or
quality of patents (Trajtenberg, 1990). As such, it is inappropriate to treat each and
every patent equally. When a patent is highly cited, it is likely to contain important
technological advances that many subsequent developments have built upon. 11
Consistent with prior literature (Dutta et al., 1999; Trajtenberg, 1990), we divide
the number of citations each patent receives by the sample average number of
citations, and then use this as the appropriate weight. Call the sum of all such
weighted patents for each year an intermediate quantity, TECHt. To form the final
TECH_RESOURCEt measure, we resort to a Koyck lag function of TECHt, with a
weighting factor of 8 = 0.4 (Griliches, 1984), i.e., we have a declining weight on
past technological resources. Formally, we have:
1 1 This is similar to the notion of viewing journal articles with higher citations as being more
influential.
47
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
k -t
TECH_RESOURCEt = £ 8'k x TECHk.
*=i
Centrality of Technology (CENTRALITY): This variable attempts to capture
the centrality of a piece of IP to the firm’s technology and product trajectories. We
suggest that the more central the D P , the more other innovations of the firm would
draw on (or refer to) it. To do this in a precise way, we count all the forward
12
citations that the particular patent has received from the firm itself. The more the
number of these ‘self-citations’, the more critical the patent is in firm’s future
technology strategy (Lanjouw et al., 2001). A high number here suggests that the
firm is engaged in subsequent inventions that build on this earlier patent.
Breadth of Applicability of Patent (SCOPE): A patent is comprised of a set of
claims that define the boundaries of property rights provided by the patent
(Lanjouw et al., 2001; Lemer, 1994). It is associated with a technology or product
space to which it can be applied. Hence, consistent with previous research, the
number of claims that the patent makes is used as a measure of the patent's
applicability across product markets.
Litigation History (LITIG_HIST): A firm’s experience in litigation, or litigation
history is measured by the relative frequency of litigation that a particular firm has
experienced. It is a dummy variable, which takes the value 0 if the number of
1 2 Forward citations for a patent are similar to citations for a journal article. This is as different from
backward citations for a patent that are similar to the references in a journal article.
48
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
litigation cases a firm has been involved in is less than the sample average, and 1
otherwise (i.e., it represents a median split of the data on litigation history for the
entire sample)13.
Control Variables
Firm Size (FIRMSIZE): Past literature has suggested that larger firms would
have better organized and larger legal departments, and hence enjoy lower costs of
engaging in litigation activity. To control for this, we include firm size as a variable
measured by the total value of the firm’s assets (Lemer 1995).
Relative Firm Size (REL_FIRMSIZE): This is measured as the ratio of the total
assets of the plaintiff firm to the total assets of the defendant firm.
Size of Litigation Case (CASESIZE): It is measured by the total number of
patents involved in a litigation case.
Market Concentration (HERFINDAHL): We expect the extent of market
concentration to affect a firm’s probability of litigating a patent. The effects run
two ways. First, the more concentrated a market, the greater the monopoly power
each individual firm enjoys, which translates to greater foregone profits if its
intellectual property is infringed by other firms. Such firm, therefore, would be
1 3 Ideally we would have measured this by the proportion of successful litigation events a firm had
engaged in. However, litigation outcomes are notoriously hard to find - settlements are frequently
private. To the extent that taking part in litigation, regardless of outcome, leads to the enhanced
experience in the process, ours is a good proxy.
49
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
more inclined to prosecute infringements of its IP. On the hand, the more
concentrated a market, the more the possibility of ‘interaction’ between the firms.
The theory of repeated games and that of multi-market oligopoly (Bemheim et al.,
1998) suggests that firm behavior can become more cooperative if there is repeated
interaction over time. This suggests that there might be less recourse taken to
litigation as a means of enforcement. The relative importance of these two effects is
unclear.
We measure concentration by the commonly used Herfindahl Index., which is
n
the sum of squares of firm's market share H = ^ f t2/n , where n is the total
;= i
number of firms within the same industry and fi is the marker share of firm i. We
use the 4-digit SIC code as our industry classification.
Overlap of Product Market (PROD_OVERLAP): This is a dummy variable,
coded as 1 if the plaintiff and defendant firms share the same product market (i.e.,
the same 4-digit SIC code), and 0 otherwise. This variable is used in the market
reaction equation and is a counterpart to the market concentration variable, which
we use in the choice model of litigation decisions1 4 .
1 4 The two variables differ for the following reason. When studying market reactions to litigation
activity, we actually know the identities of the firms involved in the litigation, and can hence
construct a precise measure of market overlap. When we look at propensity to litigate, however,
there is only the single firm involved, and hence market concentration is a better measure.
50
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 2.2 presents the general descriptive statistics. In general, the average level
of marketing capability and R&D capability for firms with litigation are lower than
those with no litigation. However, the quality of technology base and the level of
centrality of individual litigation technology are much high for firms involved in
litigation than otherwise.
Table 2.2 Descriptive Statistics - Mean Value (Standard Deviation)
Under Litigation Not Under Litigation Overall
MKT_CAP 1.51 (.56) 1.81 (0.97) 1.66 (0.78)
RD_CAP 0.39 (1.35) 0.46 (1.60) 0.43 (1.46)
TECHJRESOURCE 4.42 (4.29) 2.30 (3.16) 3.42 (4.30)
CENTRALITY 27.35 (32.05) 2.98 (5.32) 15.84 (26.49)
SCOPE 33.68 (23.51) 19.76(15.58) 27.11 (21.27)
2.5. Empirical Results and Discussion
2.5.1. Summary of Results
The first point to note is the appropriateness of our sample selection model. The
coefficient of A . (inverse of M ill’s ratio) is highly significant, which suggests that
correcting for sample selection is appropriate.
51
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 2.3 reports the estimates of parameters that are hypothesized to influence
the patent litigation decision. As hypothesized, firm-specific factors play a
significant role in the firm's litigation decision. Both marketing (MKT_CAP, pi = -
0.8313) and R&D capabilities (RD_CAP, P2 = -0.541) have a significant negative
effect on the probability of a specific patent being litigated. If a firm’s relative
marketing capability increases by one unit, the probability of litigation of the firm’s
patent decreases by 4 percentage points, while one unit increase in the firm’s
relative R&D capability decreases the probability of litigation by 2.6 percentage
points.
Table 2.3 Analysis Results of Probability of Patent Litigation
Variable Name Coefficient (Standard Deviation)
MKT_CAP -0.8313 (0.4341)**
RD_CAP -0.5410 (0.3071)**
TECH_RESOURCE 0.3358 (0.1838)**
CENTRALITY 0.0515 (0.0177)*
SCOPE 0.0156 (0.0127)
LITIGJHIST 1.1498 (0.7281)
FIRMSIZE -0.2386 (0.1843)
HERFINDAHL -1.2453 (3.0714)
* Significant at 5%
** Significant at 10%
52
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
We had suggested that, in addition to its capabilities, a firm’s resources,
specifically its technological resource base (TECH_RESOURCE) would have a
significant positive impact on litigation probabilities. This conjecture is supported
by empirical result (P3 = 0.3358). Indeed, as the value of a firm’s technological
base goes up by one unit, the probability of litigating a patent that belongs to it goes
up by 1.62 percentage points.
We had proposed that the likelihood of a patent being litigated would also
depend on the role this patent played in the firm’s future technological and product
strategy. The centrality of this role would impart a firm-specificity to the valuation
of the patent - the more central the role, the greater a firm’s incentives to litigate
any infringements of it. This is strongly supported (CENTRALITY, P4 = 0.0515),
in that a one-unit increase in the firm-specific value of a patent increases the
probability of litigation by 0.25 percentage points.
Some of the variables that were hypothesized to be significant are not so. Of
these, the most surprising is the breadth of applicability of a patent (SCOPE, P5 =
0.0156). We think this might due to the fact that the more important aspect of the
value of a patent is being captured by our measure of centrality, leaving little
variation in value to be explained by the breadth measure. Also, litigation history is
not significant (LITIG_HIST, fa = 1.149). This is perhaps largely because of the
53
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
crudeness of our measure - recall that we measured this by a dummy capturing the
number of litigation events a firm took part in, not by the number it had won.
Table 2.4 reports the estimates of parameters that are hypothesized to influence
the financial market reaction to patent litigation. Not surprisingly, the significance
and signs of the coefficients are very similar to the litigation probability equation.
To some extent, this does support our use of the efficient markets approach, as well
as the endogeneity of the litigation decision, in that markets and firms seem to be
using similar factors to assess the worth of a litigation.
Table 2.4 Analysis Results of Financial Market Reaction
Variable Name Coefficient (Standard Deviation)
MKT_CAP -1276.28 (307.97)*
RD_CAP -262.62 (135.72)**
TECH.RESOURCE 386.72 (41.86)*
SCOPE 4.09 (6.84)
REL_FIRMSIZE 1.61 (1.09)
CASESIZE -25.43 (68.43)
PROD_OVERLAP 146.09 (317.31)
X
528.42 (233.29)*
* Significant at 5%
** Significant at 10%
A firm’s marketing and R&D capabilities seem to affect market reaction
adversely (MKT_CAP, Y i = -1276.28 and RD_CAP, 7 2 = -262.62). One unit
54
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
increase in a firm’s marketing capability causes a drop of about $934.47 million in
shareholder wealth, while one unit increase in a firm’s R&D capability causes a
drop of about $40.17 million. Markets look favorably upon the litigation of patents
that belong to firms with a valuable technological resource base overall
(TECH_RESOURCE, 7 3 = 386.72). Indeed, a unit increase in the technological
base of a firm is linked to a $248.66 million rise in its market valuation on the
pursuit of litigation. As before, the breadth of applicability of the patent does not
seem to matter to market valuation.
2.5.2. Substantive Insights and Managerial Implications
The central premise of our paper is that a firm’s endowment of resources and
capabilities matters to what kind of IP litigation strategy it pursues and to how the
market looks upon these strategies. Our results and the insights that flow from them
suggest that the characteristics of who owns the IP are crucial - the owner’s
capabilities and resources turn out to have significant impacts on litigation
strategies and market outcomes.
Our first set of insights deals with the role of marketing and R&D capabilities
and suggests why some firms might be better off with a low propensity to litigate.
Our results indicates that increasing levels of marketing and R&D capabilities
reduce the likelihood of litigation by a firm, and litigation undertaken by such firms
55
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
is viewed negatively by market. On the marketing capability side, this suggests that
firms that have higher ability to take their products and technologies to the market
are better off maintaining that focus instead of allocating resources to litigation
activities. This result on marketing capability is in keeping with surveys on the
effectiveness of various appropriability mechanisms (Cohen et al., 2000; Levin et
al, 1987), which suggest that firms value complementary capabilities such as
marketing (and associated first-mover advantages) well above patent protection.
The R&D capability result similarly suggests value creation activities through
coming up with successive innovations have much more marginal impact for high
R&D capability firms than engaging in litigation to protect existing IP.
Our premise when we start out is that patent litigation is too important a
business to be left to lawyers alone. The results above support that premise - when
it comes to a trade-off, managers in firms with high marketing and R&D
capabilities should actively articulate the superiority of their capabilities in value
creation and capture, versus engaging in IP litigation. The latter imposes huge
direct costs, as well as large opportunity costs, which are greater for high capability
firms.
A second insight comes from a consideration of firm resources. Firms with a
technological resource base of high quality are both more prone to litigate, and are
rewarded by the market for such litigation. Similarly, technology that is more
56
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
central to a firm’s future product and technological trajectories is more likely to be
litigated. Both these results bring out a common sense intuition - the more valuable
a resource, the more the incentive to protect it. The results also speak to the
importance of building an effective reputation for toughness in IP enforcement for
firms with high-quality IP. Marketing has a rich literature discussing the
importance of branding. Our result is, in some sense, the counterpart of a brand
reputation in the IP domain. Exactly as in the case of a brand, the positive
spillovers from litigation extend well beyond the current period. The canonical
example in this case is Texas Instruments, which initiated the trend of aggressive IP
enforcement. In successive years, it has seen its licensing revenues increase, with
more and more firms negotiating agreements with them rather than risk litigation.
This reputation has thus yielded direct benefits (in terms of increased licensing
revenue), as well as indirect benefits (ease of negotiating cross-licensing
agreements, reducing the design freedom of competitors), extending into the future.
Combining the above result on resources, together with the one on capabilities,
brings out an important insight on the relative roles of resources and capabilities in
this context and in the RBV generally. Firms with IP resources of higher value are
more prone to litigation; but superior capabilities can overturn the effects one
would expect from a simple consideration of the value of the IP alone. This is
because firms with higher capabilities can tolerate larger reductions in the value of
57
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
their resource base through infringement. Once again, this highlights our central
theme - a consideration of litigation strategy in isolation from the other capabilities
of a firm presents a misleading picture. Thus, while technology resources are the
building blocks of future innovation, marketing and R&D capabilities are what
really make those innovations, and the rewards thereof, happen. The enduring
sources of competitive advantage in the RBV are a firm’s capabilities. Our results
back this point up by suggesting how superior capabilities have the power to shape
IP litigation strategy in directions that might run counter to a simple logic of entry
deterrence or competition prevention.
We have suggested earlier that our hypotheses, taken together, constitute a test
of the RBV. To reiterate, does the market reward firms that are pursuing strategies
in consonance with their resource/capability profile? To make this non-tautological,
we had first suggested what kind of strategies would make sense with what kind of
resource and capability endowments. Our results do seem to indicate that the
market follows the logic of the RBV - the market prefers firms with high
capabilities not litigating, and rewards firms for litigating IP of high quality. While
this test comes with a host of caveats (e.g., we don’t have any formal measure of
‘fit’ between strategies and profiles), it nevertheless is evidence of the
appropriateness of the use of RBV as the theoretical base for our study.
58
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
To sum, the results taken together highlight the key role of firm specific
resources and capabilities in enabling firms to create and appropriate shareholder
value through strategic protection of their intellectual property. They highlight that
senior marketing and R&D managers have to play a more proactive role in their
firms to prevent lawyers from excessive allocation of resources towards litigation
activities that may end up reducing shareholder value. While enforcement of IP has
its place, our results suggest that the decision to litigate and the value capture that
results from it should be balanced with alternative mechanisms such as exploiting
one’s R&D and marketing capabilities.
2.6. Conclusions and Future Research
This paper has dealt with the issue of intellectual property and its enforcement in
technology intensive markets. Even though marketing literature has dealt with
innovation in technology-intensive markets in great detail, there has been a
complete absence of research on the important issue of value capture through the
enforcement of IP rights. Our study is a first step at filling this gap.
Our framework and results provide a number of insights to managers, in terms
of the firm-specific factors that influence the decision to litigate and the subsequent
market reaction to such litigation. We emphasize the marketing implications of
59
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
protecting IP and develop a set of variables that explicitly articulate the centrality
of the IP to the firm’s technological/product trajectories. We also contribute to the
RBV literature by extending the RBV to the study of strategic IP enforcement and
by linking firm's resources and capabilities to financial metrics in the IP domain.
Our two sets of hypotheses and the empirics provide a test of the RBV formulated
in a non-tautological manner, and hence addressing the criticism against much of
RBV research.
Our empirical results suggest that firms with a highly valuable technological
base, which would presumably benefit most from the reputational spillovers of
litigation, do show a greater propensity to litigate, and markets view this behavior
positively. Interestingly, firms with a superior marketing capability seem to be less
inclined to litigate. When they do litigate, the market won’t look them upon
favorably. Similar results also hold for firms with high R&D capabilities. The
differences in these results also highlight the differential role of capabilities and
resources in the firm’s quest for developing and sustaining competitive advantage.
Since our study is the first in marketing to deal empirically with the issue of
intellectual property enforcement, there are a number of avenues for future research
that remain. The principal line of research would be to explicitly examine the
dynamic implications of litigation strategy. For instance, one could construct a time
series of litigation behavior and examine whether and by how much litigation
60
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
succeeds in delaying entry. Similarly, it would be interesting to examine whether
firm litigation behavior over time varies depending on the nature of competition
facing it. We have taken a partial cut at some of these questions, but better data are needed
to address them satisfactorily.
61
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Chapter 3
Product Diversification, Technology Diversification and Market
Performance in Pharmaceutical Industry
3.1. Introduction
In this work, we investigate the patterns of product diversification and technology
diversification in pharmaceutical industry and the consequential financial impact. It
is mainly motivated by a set of widely reported stylized facts. First, pharmaceutical
firms seem to differ greatly in the diversity of their product offerings. For example,
Aventis produces drugs targeting a relatively few number of diseases, such as
diabetes (Amaryl) and allergy (Allegra). On the other hand, Merck, the flagship of
the industry, has a much broader drug portfolio. It markets drugs targeting
cardiovascular (Cozzar), asthma (Singulair), cholesterol-control (Zocor) and pain-
control (Vioxx), etc. Second, they seem to differ significantly in the diversity of
their technology portfolios. For example, Pfizer has comparatively greater
62
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
£9
•§■3) sb o jb J3M3J qonui ui sjonpojd jnq ‘sb s jb j o jaquinu b u i saiSopuqoaj 3ABq
jBqj suiqj aq oj mass 3J3qj ‘-3'i ‘suoijBJOfdxa janpoid puB [BOiSopuqoaj oj majjBd
snoiAqo ou sq oj suiaas sisqj ‘jsiiq -japjo ui si AjisisAip jonpojd puB AjisJ3Aip
Agojouqosj jo uotjBuraiBxa qSnojoqj s j o u i b AqA\ s u o s b o j A u b u i s j b 3J3 q x
•3 3 UBUUOJJ3 d
pqouBUij s (uxqj uo jOBduii sji puB diqsuoijBpj siqj sjojdxa 0 1 si aouaq jsdsd aqj
jo qsBj uibui aqx '(£002 ‘ JJ°IJ9J S0 ) AjisiaAip jonpojd sji puB AjisjaAip A§opuqaaj
s tuxiij b uaa/Wjaq uoijBpj ajBuqjui ub jsaSSns pjnoM ‘sjaqiBUi A3oiouqoaj-q3iq
ui AqBiaadsa ‘aauapiAa Ajosjii3 (8661 ‘puBiqiBJQ ‘Z 0 0 Z ‘UBqdsjs) jaqjo qaBa uo
S3 iqSJ3A ip OMJ 3 S3 q j JO P B d u il [B3 0 j d l 3 3 J a q j JB §U D [0 0 { qJOAV OU JSOUIJB U3 3 q SBq
a ja q j ‘a jo r a s j B q /^ -u o s u B d u io o A q u o tju a jjB s s a j A jq B J a p isu o o p ajaB ijjB SBq Avoq
-M o u q jB a iS o jo u q o a j u i A o u B d s jo sip ju b o ijiu S is a q ; a i ‘jo b j p u o o a s a q x 'S u ija q jB jy
pUB A §3 JBJJS ‘S3 IUI0 U0 3 3 §UISSBdlU0 3 U3 ‘aJtlJBJajIJ JO A p o q 3 AISSBUI B JO SU30J 3 q i
u s a q SBq ‘o q o j q o d j a n p o id s ,iu j ij u i s a a u a j a jjip j a u ijs ip a q j ‘ a t ‘p b j j s j ij a q x
•op Joijuoa siapjosip oupiqoAsd j o j sapiure [AuAqp puB juauijBajj
aijaqjsauB uoqBjBqui j o j aua[AdojdAjod aonpojd oj ASopuqaaj pajBJjuaauoa
ApAippj qq/A poMopua s i ‘puBq jaqjo aqj uo ‘q c q qoqqy p p ‘sasn oijnadBjaqj
JB[ti3 SBAOipjB3 j o j sa u o ip -p -Z -a u q o z B U ir tb o jp A q B J ja x ‘Ajisaqo jo juauijBOJj aqj
ui [njasn poqpui aqj sb AjisaqoijUB aijauquiojAqj ‘uoijaajui §ui;u3A3id punoduioo
3Aip3jj3 ub sb auizBip sqBUi oj ASojouqaaj padopAap SuiABq ‘aqjojd ASojOUqaaj
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
AstraZeneca). The reverse holds for some other firms (e.g. Pharmacia). Arguably,
each of these groups of firms suggests different strategic choices by firms. Second,
the nature of the diversities is important. Whether entering diverse product markets
generates the kind of market pull influences that translate to a diverse technological
portfolio, or the other way around, i.e. technology push effect (Granstrand et al.,
1990) is of great managerial implication.
In all of this, an important point needs to be kept in sight, and that is the need
for a thorough exploration of the product and technology diversity issues. Our
paper does this in two ways. First, our conceptual framework recognizes the
dependencies between technology and product diversity, as well as the points of
departure between them, by appropriately specifying the firm-specific resources
and capabilities that would affect each strategy. What's more, we take explicitly the
nature of "path dependency" of the two decisions into account. Second, our
econometric specification estimates the impact of various factors on product and
technological diversity. Beyond that, we explore further how the diversities and the
interaction between them, together with firm-specific tangible and intangible
resources, and capabilities, affect financial market valuation. We present arguments
in the context of the pharmaceutical industry, which provides the location for our
empirical investigation.
64
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Our empirical analysis supplies strong evidence on the existence of reciprocal
effect of these diversification processes on each other. It also finds significant path
dependency in the trajectory of product and technology diversification. Firm's
knowledge of product and technology market cumulated over time and the
endowment of high capability in marketing and R&D affect the diversification
process significantly. Financial valuation reacts positively to these factors.
3.2. Literature Review
It is important at the outset to delineate the domain of our enquiry. Diversification,
in its various stripes, e.g., across industries, across different geographic regions or
countries, is not what we consider here. Our attention, basically, is on two types of
highly linked diversification: product diversification and technology
diversification.
Product diversification, in our research setting, refers to the development of
both the new products with new, distinct utility and the products with incremental
improvement. It would encompass both the creation of new product lines, as well
as the creation of variants within existent lines (the latter being the question
marketing has examined). Technological diversification, similar to product
diversification, refers to firm's coming up with innovations in technological areas
65
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
hitherto new to it. The innovation either belongs to firm's existent technologic areas
or expands beyond what it has.
Research in economics and management has extensively investigated product
diversification issues from a number of perspectives. Research questions include,
for example, what the underlying driving forces are, what the economic
consequences will be, etc. It covers both business and geographic expansion
(Montgomery, 1994; Hoskisson et al., 1990; Ramanujam et al., 1989; Wemerfelt,
1984). However, marketing literature so far has mainly focused more on within-
product-line decisions (Putsis et al., 2001; Bayus et al., 1999; Reddy et al., 1994;
Kekre et al., 1990). Factors, such as market share, price of focal and competing
firms, market concentration, etc. are suggested to likely impact product-line
extension decision.
Technology diversification, in contrast, has attracted much less attention, and
literature on it is much sparser (Breschi et al., 2003; Osterloff, 2003; Granstrand,
1998, 1997). Among the few related works, Breschi et al. (2003) examine the key
role that relatedness of knowledge might play in technology diversity; Granstrand
et al. (1997) investigate the development and consequences of technology
competencies in technology diversified corporation.
Even though the important role technology diversity and product diversity may
play to a firm has long been recognized, none of the past literature in economics,
66
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
management or marketing supplies the ample empirical evidence on the interplay
between technology and product diversity and the financial consequences due to
this interplay. Our work is among the first attempts to fill this research gap.
Specifically, we look at the determinants of both technology and product diversity,
and explore the impact of these diversities on firm's financial performance.
3.3. The Context: Pharmaceutical Industry and Drug Development
Most diversification studies have so far used multi-industry samples and
uncritically pooled cross-sectional data (Verweire, 2003). There are some typical
problems associated with the use of this methodology. At least, firms pooled over
across a number of different industries are very heterogeneous. This heterogeneity
is one of the major causes leading to estimation problems (Hill et al., 1988). More
and more researchers are convinced that restriction should be deliberately made to
narrow the sample of firms to those operating in one or similar industries in order
to get better interpretations of the relationship between diversification strategy and
financial performance. Our choice of sample firms within one particular industry,
i.e. pharmaceutical industry in US, is in line with this trend.
Pharmaceutical industry is highly research-intensive. It has a greater R&D
intensity than any other industry. R&D accounted for 15.6% of sales in 2000 for
67
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
US research-based pharmaceutical industry, compared to 10.5% for the next
highest industry (computer software), 8.4% for electrical and electronic firms, and
3.9% for US companies overall (Pharmaceutical Research Manufactures
Association, 2001). The average R&D cost per new chemical entity is estimated at
over $800 million. In recent decades, drug development has observed a significant
shift to the scientific basis of innovation, from the more random process used
earlier towards the rational model building upon genetic engineering, monoclonal
antibodies, immunology and cellular biology, etc. (Nature, 1998; Henderson et al.,
1996). Accompanying with this change is firm's continuous diversification of drug
portfolio and technology portfolio.
Technology discovery and drug development in this industry is a highly time-
consuming process, usually taking 7-12 years to advance a drug from technology
discovery through regulatory approval to market. It's a high-risk and high-stakes
venture. For each new chemical entity approved, roughly five enter human clinical
trials and 250 enter pre-clinical testing (Danzon et al., 2003). This industry is also
notable for its unusually close link between new drug development and technology
innovation. This close link thus provides us an ideal setting for us to explore our
research issues.
68
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.4. Conceptual Framework
In this section, we build on the prior literature on the resource-based view of the
firm (RBV) and theories of evolutionary economics to develop a conceptual
framework that addresses the causes of diversity in firm's technology portfolio and
product portfolio, and the consequential impact on market performance.
Literature on RBV points out that any firm is a bundle of unique, firm-specific
resources and capabilities. These endowments vary significantly across different
firms. The heterogeneity of firm's performance and choice of strategy can be
appropriately explained by the differences in firm's resources and capabilities
(Teece et al., 1997; Wemerfelt, 1984).
Further more, research in evolutionary economics suggests that the strategic
posture of a firm is determined not only by its current status of endowments, but
also by its learning processes and by the coherence of this processes and incentives
(Teece et al., 1997). Firms' current position is partly shaped by the path it has
traveled (Coombs et al., 1998; Teece et al., 1997). The notion of "path
dependence” recognizes that "history matters". The underlying reason is that firm's
learning process tends to be "local", which is defined by the specific technologies
and markets experienced. Hence the opportunities in the neighborhood of a firm's
prior research areas and markets are likely to impact its option with respect to both
69
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
the new technology it might explore and the new market it would enter. Thus,
firm's knowledge base and its routine of operation reinforce path dependence and
influence the integration of external knowledge or production of radical new
knowledge (Coombs et al., 1998). "Bygones are rarely bygones, a firm's previous
investments and its repertoire of routines (i.e. history) constrains its future
behavior" (Teece et al., 1997). Thus, "the variety generation with in the firm is
constrained variety generation" (Coombs et al., 1998).
In this research, we incorporate these theories into the construction of our
conceptual framework through two perspectives. (1) Product diversity and
technology diversity will be explained by a set of variables that are cumulative and
related to firm-specific resources and capabilities. It is in line with the studies in the
RBV. (2) Product diversity and technology diversity will be explicitly modeled as a
function of the level of diversification in previous periods, which is consistent with
the path-dependence theory.
Clearly, the commitment of resources by profit-motivated firms must involve
both the perceptions of some sorts of opportunities and an effective set of
incentives (Dosi, 1988). Opportunities stem mainly from the knowledge
endogenously accumulated by firms. However, they are also affected by the
changes of exogenous market situation. Given any level of opportunities for
technology innovation or new product development, the incentive to commit
70
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
resources to technology discovery and new product development will depend on
resource availability (or the flip side of it, resource constraints) and the amount of
effort committed to innovation and new product (i.e. appropriability conditions).
Thus, firm's strategic choice of trajectory of product diversification and technology
diversification will be a function of its decision made in history, availability of
resources, appropriability conditions and opportunities in markets under which it
operates.
In the following, we first describe a set of variables that are hypothesized to
affect the process of technology diversification and product diversification; and
then explore how these processes, together with other firm-specific endowment
variables, influence financial market valuation on firm.
Relationship Between Product Diversification and Technology Diversification
As suggested by prior literature, any firm may engage in two fundamental types
of diversification— product (or business) diversification, technology diversification
(Granstrand, 1998). Technology diversification and product diversification have
often been found to follow each other’s development (Osterloff, 2003). The
interaction between these two diversification processes is one important source of
dynamics in the evolution of firm (Granstrand, 1998).
71
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
On the one hand, high level of product portfolio diversity will expose the firm
to a larger number of distinct product markets, which will enable it to get better
sense of changes of customer needs and market trend. More diversified customer
requirements generate new market opportunities, which, in turn, increase the
demand for making transitions in firm's technology base accordingly. Cumulative
knowledge about product market is important in shaping future pattern of
technology innovation of firm (Galambos et al., 1995). Product diversification is
among one of the prerequisites causing technology diversification. It is often
labeled as "market pull" type of technology diversification (Granstrand et al.,
1990).
On the other hand, technology diversification lays solid basis for product
diversification. This is particular true in tech-intensive market, such as the
Pharmaceuticals. New drugs development ultimately relies on technologies making
various chemical entities or different combination of them that supply certain
therapeutic functions to patients. Diversified technology portfolio enables the firm
to have higher freedom of choosing promising product markets. Knowledge
cumulated over years across different therapeutic areas facilitates the flow of know
how between these areas and improves the development of more diversified drugs
portfolio efficiently. Technological exploration in a wide range of technologies is a
prerequisite for expansion of product portfolio. It anticipates the product
72
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
diversification (Pavitt, 1998). It is often called "technology push" type of product
diversification (Granstrand et al. 1990).
In sum, we propose that each of these diversification constitutes a prerequisite
for the evolution process of the other.
Additional Factors Explaining Product Diversification
Consistent with the prior discussion based on RBV and evolutionary economics
theories, we explicitly incorporate the property of "path dependence" in product
diversification into our conceptual model. Specifically, we argue that current level
of product diversity will be a function of its level in prior period.
In addition, we propose a set of factors hypothesized to influence firm's choice
of product diversity and classify them into three categories. These factors, to a
variable extent, also capture the logic of path-dependence implicitly. These
categories are: (1) the Richness of Resources, (2) the Capability of Marketing, and
(3) the Condition of Product Markets
We make the case for each of these below:
The Richness o f Resources - Clearly, for a firm to foray into new markets by
way of new product development, it needs resources that it can devote to the task
(Peteraf, 1993; Chatterjee et al., 1991). Firms, in the face of intense competition in
market and under increasing pressure to sustain competitive advantage, have strong
73
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
incentive to identify new market opportunities, diversify and expand its product
portfolio into new territories. The strategic decision on diversification always has to
take into account the resource constraints (Teece et. al., 1993). If a firm has a
higher amount of resources at disposal and a higher level of flexibility of allocating
them, it will be able to enjoy more synergy benefits and have much more freedom
in diversification decision than otherwise (Campbell, 1994; Porter, 1987).
Building upon prior literature, we propose there are two types of resources that
are relevant to product diversification process.
Marketing Investment - It refers to the amount of resources a firm commits
directly to marketing related activities, such as marketing research, advertising and
sales force. High level of marketing investment makes the exploration of more
product-markets possible, which in turn enhances the chances of identifying new
market opportunities. Clearly, large amount of resources a firm can allocate to these
marketing practices is one of the prerequisites for firm to diversify its product
portfolio. Diminishing return of investment of resources into marketing makes it
rational for the firm to allocate these marketing resources across product
development projects for more diversified markets (Klette et al., 2002). So we
propose large amount of marketing investment would lead to greater product
diversity.
74
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Product Market Presence - Prior literature indicates that product-market
presence matters in firms' choice of strategy (Teece et al., 1993). The greater the
number of markets the firm presents in, and the more the market specific know
how will be accumulated. These know-how and experience constitute valuable
resources for the firm in the diversification of its product portfolio. They enable the
firm to have synergy benefits obtained by the cumulative knowledge about the
different markets, expose the firm to more diversified voice of customers, and
facilitate the firm to know what other applications can be got from firm's
technology, i.e., what diverse product applications are possible. Thus presence in a
large number of markets offers firm substantial opportunities to expand its product
diversity.
The Capability o f Marketing - While a firm’s resource base is definitely a key
factor influencing its strategic decisions, such as those on the choice of product
diversity, the RBV also suggests that marketing capability is also of great
importance. High capability in marketing would imply a superior ability of
efficiently analyzing market opportunities, selecting target markets and leveraging
resources to quickly accomplish reconfiguration and transformation ahead of
competition (Teece et al., 1993; Kohli et al., 1990). The capability to calibrate the
requirements for change and to effectuate the necessary adjustments are obtained
over long-time accumulation. It is hard to imitate and are very specific to the firm.
75
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Thus, we propose that marketing capability facilitates the product diversification in
reaction to the changes of market condition.
The Condition o f Product Markets - Though firm-specific resources and
capabilities are vital to firm's product diversification decision, external market
situations may also play significant roles (Teece et al., 1997). For example, if a
firm's businesses are mainly focused in highly concentrated drug markets, its
advantageous appropriatability condition will make it less aggressive to diversify
its product portfolio into other markets in which it might have less dominant power.
One the other hand, if the firm operates in the markets of high volatility and risk,
the quest for stable and manageable business will push it to diversify its product
portfolio across more product categories (Rogers, 2001; Berger et al., 1995). We
hence suggest that product market variables, such as market concentration and
volatility, should also be included in the analysis of product diversification
decision.
So far, we have outlined the major factors that are hypothesized to influence
product diversification. In sum, the main hypotheses include:
- the level of product diversity in prior periods will generate positive impact on
the product diversification in the future
76
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
- the level of technology diversity in prior periods will generate positive impact
on the product diversification in the future
- the richness of resources, such as marketing investment and product market
presence (expertise), and marketing capability will positively influence product
diversification
Additional Factors Explaining Technology Diversification
Following the same logic we have applied in the prior discussion about factors
influencing product diversification, we propose that, in addition to the technology
diversification and product diversification in prior period, there are a number of
firm-specific resource and capability related factors that influence firm's choice
technology diversity. Accordingly, we classify them into three categories. They are:
(1) the Richness of Resources, (2) the Capability of R&D, and (3) the Condition of
Technology Markets.
We make the case for each of these below:
The Richness o f Resources - Just as the practice of product diversification,
technology innovation and diversification also requires commitment of variable
resources. Based on the relevant theories and current practices, we propose there
are two kinds of resources that contribute to technology diversification process:
77
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
R&D Investment - Pharmaceutical companies target more diverse populations
and explore therapeutic areas at an unprecedented level. Clearly, the greater the
amount of money available to spread across different R&D areas, the more likely
the pursuit of diverse technologies. However, there exists diminishing returns to
R&D investment (Klette et al., 2002). Given the amount of R&D resources at hand,
it would be rational for firm to allocate them across more R&D projects instead of
pouring them in only a few number of research programs. Thus the level of R&D
investment is expected to have positive impact on the diversity level of technology.
The Cumulative Technology Base - As highlighted in prior literature, a firm’s
previous cumulative technological output, which bears the impression of previous
choices and chance events (Coombs et al., 1998), would shape future possibilities
for the development of new technological know-how. Innovative output cumulated
through prior R&D activities plays a crucial role in opening up opportunities of
new technological advances and easing the flow of knowledge across different
areas. The diversification of technology portfolio inevitably builds on technological
base cumulated through prior periods.
The Capability o f R&D - Similar to marketing capability, R&D capability refers
to the efficiency of firm to utilizing firm’s R&D resources to come up with new
innovations. Firms with superior R&D capabilities will be more efficient in
discovering research opportunities, developing new technologies (i.e. new chemical
78
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
entities of potential therapeutic effectiveness), as they are more efficient in pooling
distinctive knowledge from different areas that lays out the foundation for the
future entry into new technology classes (or new therapeutic areas) (Katila et al.,
2000). So, we propose that high R&D capability will lead to a highly diversified
technology portfolio.
The Condition o f Technology Markets - Just as competition condition and
stability of product market may generate impact on decision of product
diversification, some technology market related factors may also influence firm's
decision on the diversification in technology innovation. Accordingly, we suggest
that technology market concentration and volatility should be included in the
analysis of diversification decision. More specifically, we propose that high
volatility of market increases the incentive of diversification, while high dominance
status in market reduces the motivation for expanding technology portfolio.
So far, we have outlined the major factors that are hypothesized to influence
technology diversification. In sum, the main hypotheses include:
- the level of technology diversity in prior periods will generate positive impact
on the technology diversification in the future
- the level of product diversity in prior periods will generate positive impact on
the technology diversification decision in the future
79
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
- the richness of resources, such as R&D investment and cumulative technology
base, and R&D capability will positively influence technology diversification
Factors Explaining Financial Performance
Though identifying factors that influence the process of firm's product
diversification and technology diversification is of great managerial implication,
how firm's financial performance will be affected by these two distinct processes is
also of critical importance. Firm's choice of diversification will eventually
determined by whether these diversification will help sustain and strengthen its
advantageous position in markets. Thus this empirical study will also examine the
financial consequences of these diversification.
We propose that firm's financial performance will be affected by its product
diversity, technology diversity, together with some resource, capability and market
condition related variables. In particular, we have:
Product Diversity - A rich literature has indicted that broadening product scope
can be a value-generating strategy (Matsuska, 1994). For example, widening the
product portfolio will enable the firm to exploit economies of scale and scope
(Henderson et al., 1996; Baumol et al., 1982), to enjoy large customer base and
high customer loyalty (Moothy et al., 1998; Quelch et al., 1994) and to preempt
rival firms competing in its niche markets (Gilbert et al., 1993; Bhatt, 1987), etc.
80
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
However, there are also strong competing arguments emphasizing that product
diversification reduces firm's value. These arguments stress the fact that increasing
product diversity requires numerous complementary resources, such as
manufacturing, marketing, sales, and distribution assets, as well as managerial time
and attention to guide and coordinate the process (Teece et al., 1997). To the extent
that the diversity if often unrelated, the applicability of resources is even more of a
problem. Thus, the performance implications of increasing product diversity are
likely to be negative. Given the contradictory findings from past literature (Moothy
et al., 1998; Berger et al., 1995; Lang et al., 1994; Lloyd et al., 1994), we do not
hypothesize in either direction for the impact of product diversity on financial
performance, and leave this as an empirical issue.
Technology Diversity - The arguments here are similar to those for product
diversity. Highly diversified technology portfolio will preempt rival firms in the
competition. It strengthens firm's advantageous position in the market. It can also
be applied as powerful bargaining tools to obtain strategic position in markets.
However, this diversification can only be sustained by commitment of numerous
complementary resources, high replacement rate of technology will incur
substantial cost to firm if the diversified technology could not be put into
application efficiently and quickly. Technology diversification alone may not
always a value-enhancing process. We do not hypothesize in either direction for the
81
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
impact of technology diversity on financial performance, and leave this as an
empirical issue.
Interaction Between Product Diversity and Technology Diversity - As a proxy
for the level of match between these two diversities, we are interested in how
financial performance will react to it. Based on prior arguments, we propose that
high level of interaction, i.e. high level of match, will generate positive effect on
financial reaction.
Marketing Capability - As one of the important firm-specific capability
valuables, strong marketing capability will enable firm to be more efficient in
identifying customer's needs, predicting the changes of market trend, better
targeting and positioning its products than competing firms. Consistent with prior
literature (Dutta et al., 1999), we postulate that marketing capability will have a
positive influence on firm's market performance.
Cumulative Technology Base - Recent literature has pointed out the link
between a firm's technology base and it's market performance (Deng et al., 2000).
More and more firms have begun realizing that they can leverage substantial
benefits from their intangible resources (patented technologies in this case) not only
from incorporating them into the development and marketing of their products, but
also through direct selling or licensing. We thus propose that a technology base of
high quality has a positive effect on a firm's financial performance.
82
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Product Market Presence - Prior literature has outlined the strategic importance
of rich product portfolio. Rich product portfolio, hence high level of product
market presence, represents valuable knowledge of markets. High level of product
market presence has a positive impact on a firm's financial performance.
In addition, we include firm size as a ‘control’ variable in the diversification
analyses and market performance study. Again, since past literature has been
completely ambiguous on the impact of firm size on performance, we leave this as
an empirical issue and do not hypothesize in either direction. Market condition
factor, i.e. market concentration, is also included in the financial performance study
as a control variable, with which financial performance is expected to increase due
to its impact on the condition of value appropriation.
The major relationships we attempt to explore in this paper are represented in
Figure 3.1.
Figure 3.1 Relationships Between Diversification And Their Impact on Financial
Performance
(+/-) (+/-)
Firm’s Financial
Performance
Product
Diversification
Technology
Diversification
83
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3.1 lists the variables along with the suggested direction of impact
product diversification, technology diversification and financial performance.
Table 3.1 Summary of Major Hypotheses
Equation Hypothesis
Product Diversification
Product Diversity At t-1 (+)
Technology Diversity At t-1 (+)
Marketing Capability (+)
Marketing Investment (+)
Product Market Presence (+)
Technology Diversification
Technology Diversity At t-1 (+)
Product Diversity At t-1 (+)
R&D Capability (+)
R&D Investment (+)
Cumulative Technological Base (+)
Market Performance
Product Diversity (+/-)
Technology Diversity (+)
Interaction Between Diversities (+)
Marketing Capability (+)
Product Market Presence (+)
Cumulative Technological Base (+)
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.5. Data Sources, Data Structure and Descriptive Statistics
As the central objective of this paper is to explore the relationship between firm's
innovation strategy (i.e. technology diversification) and product development
strategy (i.e. drug product diversification), and the impact of these strategies on
firm’s financial performance, we collect data on a sample of pharmaceutical firms
from a number of sources. They contain rich information on firm’s technology,
new product development and financial status.
Technology Information
As pointed out in prior literature (Levin et al., 1987), one of the prominent
features of pharmaceutical industry is that firm’s technology innovation is highly
presented by patent application. W hat’s more, firm’s patent information is always
accessible to the general public. The approach using patent information as proxy of
firm's technology innovation information has been widely applied in recent
empirical analysis on firm’s technology innovation (Trajtenberg, 1990). We thus
use firm’s patent information as proxy to measure its technology diversification
process. In particular, we have:
Patent Portfolio Information - Patent portfolio information for these original
sample firms is obtained from US Patent and Trademark Office (USPTO) and
85
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Community of Science (COS) database for original sample firms since their
creation. From these two patent sources, we not only get the number of patents
issued by USPTO to each sample firm for each year, but also get the information
on the citations each patent receives.
Technology Diversification Information - Technology diversification
information is obtained based on the information of patent technology
classification. As each patent is assigned with 9-digit US Classification code by
USPTO, which corresponds to a position within the hierarchical technology
classification system. Each 3-digit code indicates different technology classes1 .
Product Information
Data on sample firm’s product portfolio contains two subsets of information:
Drug Portfolio Information - For each sample firm, we collect information on
total marketed drugs from the database of US Food and Drug Administration
(FDA). Since all the new drugs sold in the U.S. must first get approval from FDA,
its dataset includes the universe of all new drugs marketed in the U.S. during our
research period. Different from some prior literature on innovative drug
introduction, we don’t restrict collection of drug data only to the new molecular
1 For more detailed information, refer to Appendix C.
86
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
entities (NMEs)2. Because only very few NMEs happen to immediately fulfill
therapeutic efficacy and required safety standard, many NMC prototypes, on the
contrary, show various clinical shortcomings which might be overcome by
alternations to the original structures, or the introduction of compound
combinations, or different ways of administering them. NMC only captures a
fracture of firm’s technology innovation. As highlighted in prior section, we adopt
a broad definition of technology innovation, and include in the sample all the
approved new drugs (NDAs), no matter whether they stand for path-breaking
innovation build upon NME, or just represent incremental improvement though
new combination of original entities.
Drug Classification Information - Intercontinental Medical Statistics America
(IMS) has developed the well-known Uniform System of Classification (USC),
which uses a hierarchical structure with 5-digit code to categorize drug into
different therapeutic areas3. The major (first-two-digit) therapeutic classes include:
Analgesics, Antacids and Antiflatulents, Antiarthritics, Anti-Infectives,
Antispasmodic/Antisecretory, Biologicals, Respiratory Therapy, Cancer/Transplant
Therapy, Cardiovascular Therapy, Contraceptives, Cough/Cold Preparations,
Dermatologicals, Diabetes Therapy, Diagnostic Aids, Diuretics, Hormones,
2 FDA’s definition of NME is that it is an active moiety that has not previously been approved either
as the parent compound or as a salt, ester, or derivative of the parent compound in the United States
for use in a drug product either as a single ingredient or as part of a combination.
3 For more details, refer to Appendix C.
87
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hospital Solutions, Nutrients and Supplements, Ophthalmic Preparations,
Psychotherapeutic Drugs, and Vitamins. Each of them is divided into smaller
therapeutic markets down to the five-digit level. Although different drugs sold
within the same therapeutic class are often imperfect substitutes, the degree of
substitutability of them is much greater within, as opposed to across, therapeutic
class.
USC has been recognized as a well-established classification system and
become frequently applied in recent pharmaceutical research (e.g. Roberts and
McEvily, 2000). We apply the same system to classify our sample drugs into
different therapeutic classes.
Financial Information
The construction of our data set includes the collection of financial information
for all publicly traded firms with SIC-2834 (Pharmaceutical sector) from
COMPUSTAT since 1982 -2001. In particular, we obtain firm's annual information
on sales, value of assets, profits, investment in R&D and Marketing, etc.
Though there are over three hundred pharmaceutical firms publicly traded in US,
our sample consists of thirty of them. It is because we set some criteria in the
sample selection process. They are: (1) sample firms should have become public
88
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
for at least 10 consecutive years and keep public status until 2001, (2) sample firms
should have been issued at least 10 patents since creation, and (3) sample firms
should have at least 5 drugs in market by year 2001. We impose these restrictions
on the selection of sample firm with the intention that the technology portfolio and
product portfolio of sample firms should show certain level of variations over
years. It will inevitably exclude some small firms. However, further overview of
the sample firms and overall industry indicates that the number of drugs introduced
by these sample firms constitutes over 80% of drug listed in FDA database, and the
number of patents issued to these firms constitutes similar proportion of overall
patents. Hence, this sample of firms does not suffer the lack of representativeness.
3.6. Variable Operationalization
Based on the conceptual framework outlined in prior section, we lay out the
measurement of variables of interest in the following order. We start with the
operationalization of product diversity and technology diversify, and then move on
to describe the underlying logic of choosing Tobin’s Q as the measure of financial
performance. The rest of this section devotes to the discussion of the
operationalization of the exogenous variables.
89
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Diversity Measures
Product Diversity (PROD_DIVit) - Product Diversity refers to the distribution
of drug variations across different therapeutic classes. One approach prior literature
often adopts to measure product diversity is the sample count of different models
(or SIC) of product the firm has in its product portfolio (Putsis et al., 2001).
However, this approach basically assigns different model the same weight in the
estimation of portfolio diversity. It is inappropriate in the sense that, if some
models have large number of variations within, the firm’s product portfolio will be
regarded to be more concentrated on these few models than that of other firm with
number of variations more equally distributed across models, even these two firms
might carry the same number of models of product.
Palepu (1985) developed an entropy measure combining the objectivity of the
earlier product count measures based on the number of models (or SIC code) and
the relatedness concept of the categorical measures. In particular, this measure
enables us to take into consideration two elements of drug diversification: (a) the
number of therapeutic fields in which a firm operates, (b) the relative importance of
each of these fields in the overall drug portfolio.
We thus apply entropy measure in our study. Specifically, at year t, firm i's
product diversity (at 5-digit level, for example) is:
90
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
td „ = Z p , 1 "
j = 1
Pj
Pj — the proportion of drugs developed by period t, which can be classified into
the jth therapeutic class at 5-digit level
N — the total number of 5-digit level therapeutic fields in which firm operates
To measure drug diversity, one has to first define the nature of the product. The
measure of product diversity depends crucially on the level of data aggregation.
The greater the level of aggregation, the smaller the observed variety (Gronau et al.,
2001). As a first approximation, we measure drug diversity at 2-digit level. It is
appropriate in the sense that, to a large extent, drugs within the same 2-digit level
can be used as substitutes targeting similar diseases, however, drugs cross different
2-digit levels won't be of similar functions and hence represent different therapeutic
areas. Finally, we would like to stress that our measure of product diversity is
cumulative. It measures product diversity for a firm by the time of interest.
Technology Diversity (TECH_DIVit) - Technology diversity refers to the extent
of the coverage of a firm's technology portfolio that can be applied directly or
indirectly to facilitate the development of new drugs or new chemical entities. The
approach to grasp technological diversity is to view technology as consisting of a
number of distinct “technological areas” (Jaffe, 1989). Patent information is of
91
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
particular relevance to assess the spectrum of technological activities of companies
(Gavetti, 1994; Pavitt, 1988).
We adopt the technology-oriented classification system based on the US
Classification developed by USPTO to categorize firm's patented technology.
Analogous to the measure of product diversity, we apply an entropy measure for
technology diversity. Specifically, at year t, firm i's technology diversity (at 9-digit
level, for example) is:
TD, = £ P/ ] n
1
P
Pj — the proportion of patents issued by period t, which can be classified into jth
technology class at 9-digit level
N - the total number of 9-digit level tech classes in which firm participates
Applying the logic we use in the choice of digit level for drug diversity
measure, we measure product diversity at 6-digit level. Based on the approach that
USPTO has applied to classify patented technology, we can tentatively derive that
technologies in the same 6-digit level share high level of commonality in
technological features and thus can be applied to develop products offering similar
utilities. We measure firm’s technology diversity at 6-digit level. Similar to product
diversity measure, the technological diversity measure is cumulative, i.e., it
measures technological diversity by the time of interest.
92
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Financial Performance Measure
Financial Market Valuation (TOBINjt) - The use of accounting based
performance measures, including profits, market share, return to assets, etc.
attributes to the mixed results in diversity research (Verweire, 2003). These
measures are subject to severe criticism, as they can be easily distorted by
systematic risks, tax laws and arbitrary conventions" (Montgomery et al., 1988).
At the same time, there is emerging evidence showing that the range of firm's R&D
and productive activities follows some "purposiveness", firm's new product
development is far from being driven be short-term quest for profits (Scott. 1993).
These accounting measures don't fit our research scenario in particular, because the
impact of firm's technology diversification on its financial performance can not be
represented directly by changes of its financial performance in product markets
over a short period.
The financial market measure, Tobin’s Q, which is the ratio of market value of
a firm over the replacement cost of its physical assets, incorporates both
accounting-based and market-based elements (Lloyd et al., 1994) and is widely
applied in analyses (Lang et al., 1994). It is thought to effectively minimize the
distortions caused by the use of the accounting measure of performance
(Montgomery et al., 1988). Any deviation of market value of the firm from its
93
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
replacement value is interpreted as signifying an unmeasured source of value.
Furthermore, no risk adjustment or normalization is required to make comparisons
between firms. Hence, we adopt this measure in this study. Specifically, at year t,
firm i's financial performance is measured by Tobin's Q, which is
MVt — market value of the firm at year t
PVt — replacement value of the physical assets at year t
Although various methods have been proposed to calculate the Q, Chung and
Pruitt (1994) show that these different approaches tend to yield similar values.
Following Kedia and Mozumdar (2002), we calculate the market value of assets as
the sum of book value of assets (Compustat data item 6) and market value of
equity, less the book value of common equity (item 60) + balance sheet deferred
taxes (item 74).
The list of exogenous variables includes resource richness, marketing capability
and R&D capability, market conditions etc. We describe them below in order.
94
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Resource Richness for Product Diversification
Product Market Presence (PRODBASEu) - Each different drug targets patients
in different therapeutic class (drug market), so we measure Product Market
Presence by the sum of different drugs that a firm has in its product portfolio4.
Marketing Investment (MKTJNVjt) - Marketing Investment is measured by the
log sum of the prior service, general management and administration expenditure
(SGA) by year t with 15% annual depreciation rate (Crepon et al., 1998). Selection
of SGA is necessitated by the lack of firm-level data on marketing research
expenditure. In the absence of direct measure of it, SGA is particularly appropriate
and superior to the more common practice of using advertising expenditure as a
proxy for marketing investment (Vinod et al., 2000), as sales activities and
promotion costs, marketing research expenditure is accounted for by SGA item.
We take "log" measure in consistent with the assumption in prior literature
about the diminishing return on R&D investment. We take "stock" (sum) measure
to account for the cumulative feature of this measure and the "lag" issue5. We don't
make explicit assumptions about distributed lags between R&D investment and the
successful development of new chemical compounds and the issuance of patents.
4 Different drugs have different effect. In contrast to employing citations o f patents to value relative
quality, we just use simple count as proxy. As we don't have objective measures that help weight the
relative importance of different drugs.
5 Explicitly accounting for the lag issue will lead throwing out much of our data in order to have 3-6
lags. Considering the limited observation over time span, substitute measures would be applied
instead.
95
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
We simply include "stock" of R&D investment, as if the annual flows are
reasonably smooth, using stock measure is equivalent in many senses to imposing a
lag structure (Henderson et al., 1996). And consistent with prior literature, we
apply a 15% depreciation rate in our study. Specifically, for firm i at year t, its
marketing investment is calculated by:
3
MKT _ INVit = £ MINVT^j) x (l - y)J
,=o
MINVTt — annual marketing investment at year t
r — depreciation rate
Resource Richness for Technology Diversity
Cumulative Technology Base (TECHBASEit) - This measure is obtained by
summing up the value of each of the firm’s patents. To do this, we weight each
patent by the citations it has received6. The need for this weighting arises because
there exists considerable variation in the value or quality of patents (Trajtenberg,
1990). As such, it is inappropriate to treat each and every patent equally. When a
patent is highly cited, it is likely to contain important technological advances that
many subsequent developments have built upon7. Consistent with prior literature
6 We take into account the truncation issue faced by the patent issued recently.
7 This is similar to the notion of viewing journal articles with higher citations as being more
‘impactful’.
96
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
(Dutta et al., 1999; Trajtenberg, 1990), we divide the number of citations received
by each patent by the sample average number of citations, and use this as the
appropriate weight. Call the sum of all such weighted patents for each year an
intermediate quantity, TECHt. To form the final TECHBASEt measure, we form a
Koyck lag function of TECHt, with a weighting factor of S = 0.4 (Griliches, 1990),
i.e., we have a declining weight on past technological resources. Specifically, we
have:
k=t
TECHBASEit = £ x TECH k.
* = i
R&D Investment (RD_INVjt) - R&D Investment is measured by the log of the
sum of the past R&D investment by year t with 15% annual depreciation rate
(Crepon et al., 1998). The logic of taking "log" and "stock" measure is similar to
what we apply to construct marketing investment measure. We use similar formula
to calculate R&D investment as we do in the calculation of Marketing Investment,
and replace annual market investment with annual R&D investment.
Capability Measures
Marketing Capability (MKT_CAP;t) - Following Amit and Shoemaker (1993),
we conceptualize a firm’s capability as an “intermediate transformation ability”
between resources (i.e., inputs) and business objectives. Since capabilities are an
intermediate step between resources and outputs, one can hope to see the inputs
97
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
that a firm uses and the outputs it achieves, but one can only infer its abilities in
converting one to the other. As such, we employ an econometric technique called
Stochastic Frontier Estimation (SFE) (Dutta et al., 1999; Stevenson 1980) to infer
capabilities based on observations of a firm’s inputs and outputs8.
Specifically, marketing capability represents the ability of the firm to efficiently
sense market opportunities, deploy marketing and technologic resources, along
with its installed customer base, to maximize its sales, which is the objective we
assume the firm wishes to maximize. Briefly, the marketing capability is measured
by how close the realized sales are to the maximum sales the firm could have
achieved. Clearly, the closer the realized to the maximum, the higher the marketing
capability of the firm. We detail the estimation of annual marketing capability in
the Appendix A.
As the current level of product diversity is also affected by marketing capability
in prior periods, we apply a cumulative measure of this variable in diversity
analysis. This is calculated as the weighted sum of marketing capability over the
prior 4 years with a discount rate of 15%9. For the performance equation, however,
we use the current level of marketing capability as the explanatory variable, which
is based on the assumption that performance reacts contemporaneously to the
current level of capability.
8 Further details on the methodology, please refer to the Appendix A.
9 We apply the same discount rate as used by Crepon et al. (1998) and Henderson et al. (1996).
98
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
R&D Capability (RD_CAPjt) - Our estimation of R&D capability follows the
logic identical to that of marketing capability above. We suggest that the objective
of R&D activities is to maximize the production of innovative technologies. The
major resources that the firm has at its disposal to fulfill this objective include
current and past R&D expenditure, and the extant technological knowledge base of
the firm. We employ quality-adjusted patent counts as our measure of technological
output. The superiority of this measure to a raw patent count measure has been
discussed at length in the literature (Trajtenberg, 1990)1 0 . We apply the same
technique to estimate R&D capability1 1 as we do towards marketing capability.
As the current level of technology diversity is also affected by R&D capability
in prior periods, we apply a cumulative measure of this variable in diversity
analysis. This is calculated as the weighted sum of R&D capability over the prior 4
years with a discount rate of 15%1 2 .
Market Condition Variables and Firm Size
Consistent with prior literature, we propose some market condition variables in
the diversification analysis. Specifically, they consist of market concentration and
1 0 The main criticism is that raw counts, by weighting patents equally, do not take into account the
great deal of heterogeneity in the quality of patents.
1 1 For further details on the methodology, please refer to Appendix A.
1 2 We apply the same discount rate as used by Crepon et al. (1998) and Henderson et al. (1996).
99
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
market volatility. We measure market concentration by the using the traditional
Herfindahl concentration index. There exist many market volatility measures. Here
we use as a proxy the square of the difference between market growth rate in two
consecutive years exactly prior to year t of interest. We measure firm size by the
log of firm's asset at year t.
We summarize general descriptive statistics of the major variables in Table 3.2.
Table 3.2 Descriptive Statistics of Major Variables
Variable Name Mean Std. Error
TOBIN 1.57 0.19
PROD_DIV 3.18 0.05
TECH.DIV 7.65 0.47
MKT_CAP 2.62/0.99* 1.19/0.49*
R&D CAP 2.32/0.84* 1.36/1.02*
MKTJNV 6.76 2.42
R&DJNV 5.48 2.30
PRODBASE 115.23 160.21
TECHBASE 124.72 170.62
* Cumulative/Current
100
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.7. Empirical Analysis
3.7.1. Selection of Empirical Model
As pointed out by Hsiao (2002), standard regression analysis always assumes
that factors that may affect the value of the dependent variables, but can not been
explicitly listed in the set of independent variables, can be appropriately
summarized by a random disturbance. But "when numerous individual units are
observed over time, it is sometimes assumed that some of the omitted variables will
represent factors peculiar to individual units and time periods for which
observations are obtained, ..." (Hsiao, 2002). Traditional cross-sectional regression
analysis won't help exclude the effect from these individual-specific unobservable
factors.
The data we have collected are longitudinal, time series feature will enable us to
obtain sufficient variability in decision on technology diversification and product
diversification for the sample firm, while the cross-section property will allow us to
exploit inter-firm variability in these decisions (Bayus et al., 1999).
We apply dynamic panel-data model to investigate the diversification process.
This model explicitly captures the possible impact of "path dependence" that is
hypothesized to play important role in these processes. More specifically, we have
y« = « * y u -1 + x u * B + v,. + s it i = 1,A A; t = 1,A Tt
101
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
where:
a is the coefficient for the "path dependence"
B is a vector of parameters for the exogenous explanatory variables
V j is the random effect that is independent and identically distributed (iid)
over the individual i with variance a 2 (v)
B it are iid over the whole sample with variance a (e)
Vi, B it are assumed to be independent for each i over all t.
Further more, as we are also interested in examining the impact of product and
technological diversity on firm's performance, we resort to random effect single
equation model using the same data set to conduct the investigation.
According to Hsiao (2002), we have the following model:
T« = r + A 'X it + u jt
where: i=l,2, ..N; t=l,2,..Ti
“ ii =4- + n u
uu consists of two components: with the first part A, being firm-specific random effect
vector1 3 , and the second part mt being standard random error term vector. General
1 3 We apply Hausman test to justify our selection of model (random effect vs. fixed effect) and find
that selection of random effect model is supported. For detail, refer to Hsiao (2002) and Greene
(1997).
102
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
assumptions relevant to these two error terms will be applied here, i.e., these two error
terms are independently distributed and have zero mean value. They have no
correlation with independent variables.
In addition, they satisfy:
i f i ~ j
' J lO i f i * j
En n = JA r = ( < / ) if i = J a n d t = s
“ p I 0 otherwise
The following three equations summarize the relationships of interest:
— Equation One (Product Diversity)
PROD_DlVit = /
PROD_ DIVit_ v TECH_ DIVit_ x, M KT _ CAPit, M KT _ INVit,
PRODBAS%, CON_ PRODit, VOL_ PRODit SIZE,
— Equation Two (Technology Diversity)
TECH _D IVit = /
TECH_ D1V1 , PRO D _D IV, , R D _ CAP,, R D _ INV, ,
TECHBASEit, CON _ TECH„, VOL_ TECH, , SIZE,
— Equation Three (Market Performance):
f rr r ? r ' T i t \ t i t D D /in n n / rA7TPD h/i v t r i D V
TOBIN, = /
TECH _ DIV, , PROD _ DIV, , INTER _ DIV, , MKT _ CAP, ,
kTECHBASE, , PROBASE, , CON _ MKT, , SIZE, ,
103
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
3.7.2. Discussion of Empirical Results
The estimation results for diversities are reported in Table 3.3 and Table 3.4,
and the estimation results for market performance are presented in Table 3.5.
Table 3.3 Estimation Results for Product Diversity Equation
Variable Coefficient Std. Error P-Value
PROD_DIV (prior period) 0.7884 0.0326 0.000
TECH_DIV (prior period) 0.0899 0.0287 0.002
MKT_CAP 0.0061 0.0112 0.058
MKTJNV -0.0414 0.0217 0.560
PRODBASE 0.0226 0.0049 0.000
SIZE 0.0019 0.0008 0.023
VOLATILITY 0.0251 0.0672 0.709
CONCENTRATION -0.0299 0.0753 0.691
CHI-SQUARE: 1052.07
104
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Table 3.4 Estimation Results for Technology Diversity Equation
Variable Coefficient Std. Error P-Value
TECH_DIV(prior period) 0.7812 0.0377 0.000
PROD_DIV(prior period) 0.0802 0.0316 0.011
RD_CAP 0.0152 0.0054 0.005
RDJNV 0.0076 0.0149 0.611
TECHBASE 0.0001 0.0001 0.085
SIZE 0.0034 0.0012 0.004
VOLATILITY 0.0712 0.1667 0.669
CONCENTRATION -0.1308 0.0595 0.028
CHI-SQUARE: 669.51
Table 3.5 Estimation Results for Market Performance Equation
Variable Coefficient Std. Error P-Value
PROD_DIV 0.0102 0.0425 0.081
TECH.DIV -0.0353 0.0169 0.370
INTER_DIVERSITY 0.0009 0.0084 0.910
MKT_CAP 0.0003 0.0001 0.011
PRODBASE 0.0062 0.0027 0.021
TECHBASE 0.0002 0.0001 0.010
SIZE -0.0028 0.0009 0.004
CONCENTRATION 0.0178 0.0067 0.008
R-SQUARE: 0.39
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The results from product diversification analysis show that major effects are as
hypothesized. The coefficient of product diversity in prior period is 0.79 with high
significance. It strongly supports the existence of effect of "path dependency" in the
trajectory of product diversification. Technology diversity in prior periods is found
to generate positive and significant impact on the product diversification. It is in
line with the argument of the existence of "technology push" in product
diversification, and consistent with prior theory outlining the predictability of
technology diversity on product diversity. As expected, a firm’s high efficiency in
marketing activities, represented by its marketing capability, also leads to increase
in product diversity. As argued earlier, superior marketing capability implies higher
marginal benefits for the firm, so expanding into more different product markets
makes sense. The coefficient of product market presence, which is one of the
variables capturing the resource richness of the firm, is also significant and
correctly signed. However, marketing investment is not found to positively link to
product diversification in a significant way. Though the coefficients of market
condition variables, i.e. market concentration and volatility, are correctly signed,
they are not found to have significant impact. What's more, firm size matters in this
diversification process, it is found to have positive and significant influence on
diversity.
106
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In the equation of technology diversification, most of the variables are
significant and of the expected sign. Again, we found strong evidence to support
the "path dependency" argument in technology diversification process. Level of
technology diversification in prior periods has positive and significant impact on
technology diversity at current period. The coefficient of product diversity is 0.08
and of high significance, which in turn indicates the technology diversification is
partly "pulled" by product diversity. R&D capability, which represents efficiency in
coming up with innovations, has a significant positive impact. The cumulative
technological base, as a measure of R&D related resources, has positive and
significant effect on technology diversification, while the R&D investment is not
found to have voice in the technology diversification process. High market
volatility will lead the firm to diversify its technology portfolio, however, the effect
is not significant. The second market condition variable, i.e. market concentration,
is found to have enough impact on the decision of diversification process. Firms
operating in highly concentrated market will be less motivated to diversify its
technology portfolio, monopoly power over the market warrants advantageous
position the firm wants. What’s more, the size of firm is also found to play
important role in technology diversification.
Product diversity has significant influence on firm’s market performance.
However, technology diversity has negative but insignificant impact on
107
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
performance. This suggests in some ways that the disadvantages of ‘spreading out
too thin’ dominates the advantage of staying abreast of multiple technologies, and
hence staving off possible obsolescence. The coefficient of interaction of the two
diversities is positive, but no significance is found. That the coefficient of expertise
or knowledge in both product market and technology market is significantly
positive indicates that financial market values highly of the firms rich in these
resources. Market valuation is also found to increase with marketing capability.
What's more, consistent with competition theory, the firm operating in highly
concentrated market enjoys substantial monopolistic benefits, and hence financial
market values its advantageous market position in a positive way. However, large
firm size itself will not necessarily be valued positively by market.
3.8. Conclusions
In response to the lack of research that examines the interplay between a firm's
product diversity and technology diversity, and the corresponding financial
consequences, we construct a unified conceptual framework using the resource-
based view of firm and evolutionary economics theory to investigate the
relationship between them. Further more, we apply a dynamic regression model to
test how these diversities affect each other and a random-effect model to explore
108
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
how financial market values these diversities by using panel data on a sample of
firms in the pharmaceutical industry over the period 1982-2001.
Our empirical study shows that a firm’s richness of resources and capabilities
positively impact its product and technology diversification. We also find evidence
for both market pull and technology push effects, in contrast to earlier studies that
have tended to emphasize one of these effects to the exclusion of the other. Finally,
the findings are consistent with "path dependency" arguments derived from
evolutionary economics theories.
Our research, which is the first to examine the interplay between technology
diversity and product diversity decisions, suffers from some limitations. For
example, we impost strict criteria when selecting sample firms. Inclusion of more
sample firms with low product profile or technology profile will increase the
applicability of our findings. As we have highlighted in prior section, this study
focuses on just one particular industry, i.e. US pharmaceuticals. While it is with an
advantage of reducing cross-industry heterogeneity, some of our findings may be
industry-specific, impeding their generalization. Further expansion of industry
scope will help increase the generalization of our conclusion, however, special
caution should be paid to control cross-industry heterogeneity.
109
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
References:
Acharya, S. (1988): "A Generalized Econometric Model and Tests of a Signaling
Hypothesis with Two Discrete Signals", Journal o f Finance, 43(2), 413-429.
Agrawal, J. and W. Kamakura (1995): "The Economic Worth of Celebrity
Endorsers: An Event Study Analysis", Journal o f Marketing, 59, 56-62.
Alpert, F. (1993): "Breadth of Coverage For Intellectual Property Law:
Encouraging Product Innovation by Broadening Protection", Journal o f Product
and Brand Management, 2, 5-17.
Amemiya, T. (1985): Advanced Econometrics, Harvard University Press.
Amit, R.and P. J. Shoemaker (1993): ’’Strategic Assets and Organizational Rent”,
Strategic Management Journal, 13, 33-46.
Baumol, W., J. Pnazer and R. Willig (1982): Contestable Markets and The Theory
o f Industry Structure, New York: Wiely.
Bayus B. and W. Putsis Jr. (1999): "Product Proliferation: An Empirical Analysis
of Product Line Determinants and Market Outcomes," Marketing Science, 18 (2),
137-153.
Bebchuk, L. (1984): "Litigation and Settlement under Imperfect Information", The
Rand Journal o f Economics, 14, 404-415.
Berger, P. and E. Ofek (1995): "Diversification’s Effect on Firm Value", Journal o f
Financial Economics, 37, 39-65.
Berman, B. (ed.) (2002): From Ideas to Assets: Investing Wisely in Intellectual
Property, John Wiley & Sons, Inc. New York.
110
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Bemheim D. and M. Whinston (1998): “Incomplete Contracts and Strategic
Ambiguity”, American Economic Review, 88 (4), 902-933.
Bharadwaj, A., S. Bharadwaj and B. Konsynski (1999): "Information Technology
Effects on Firm Performance: A Measure by Tobin’s Q", Management Science, 45,
1008-1024.
Bhatt, S. (1987): "Strategic Product Choice in Differentiated Markets", The Journal
o f Industrial Economics, XXXVI, 207-216.
Bottazzi, G., G. Dosi, M. Lippi and M. Richaboni (2000): "Process of Corporate
Growth in the Evolution of An Innovation-Driven Industry: The Case of
Pharmaceuticals", Working Paper.
Boulding, W. and R. Staelin (1995): “Identifying Generalizable Effects of Strategic
Actions On Firm Performance: The Case Of Demand-Side Returns To R&D
Spending”, Marketing Science, 14 (3), 222-237.
Breschi, S., F. Lissoni and F. Malerba (2003): "Knowledge-Relatedness in Firm
Technological Diversification", Research Policy, 32, 69-87.
Campbell. J., A. Lo. and C. MacKinlay (1997): The Econometrics o f Financial
Markets. Princeton University Press, Princeton, NJ.
Campbell A. (1994): "Managing Diversification and Business Development", Long
Range Planning, 27 (2), 128-130.
Chatterjee, S. and B. Wemerfelt (1991): "The Link Between Resources and Types
of Diversification: Theory and Evidence", Strategic Management Journal, 12, 33-
48.
Choi, J. (1998): "Patent Litigation as An Information-Transmission Mechanism",
American Economic Review, 88, 1249-1263.
Chung, K. and S. Pruitt (1994): "A Simple Approximation of Tobin’s Q", Financial
Management, 23, 70-74.
Clark, B. and D. Montgomery (1998): “Competitive Reputations, Multi-market
Competition and Entry Deterrence”, Journal o f Strategic Marketing, 6 (2), 81-97.
Ill
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Cohen, W., R. Nelson and J. Walsh (2000): "Protecting Their Intellectual Assets:
Appropriability Conditions and Why U.S. Manufacturing Firms Patent (Or Not)",
NBER Working Paper.
Cohen, W. and D. Levinthal (1990): "Absorptive Capacity: A New Perspective on
Learning and Innovation", Administrative Science Quarterly, 35, 128-152.
Coombs, R. and R. Hull (1997): "Knowledge Management Practices and Path-
Dependence in Innovation", Research Policy, 27, 237-253.
Cooter, R.and D. Rubinfeld (1989): "Economic Analysis of Legal Disputes and
their Resolution", Journal o f Economic Literature, 27, 1067-1097.
Crepon B., E. Duguetand and J. Mairesse (1998): "Research, Innovation and
Productivity: An Economic Analysis at The Firm Level", NBER Working Paper.
Danzon, P., S. Nicholson and N. Pereira (2003): "Productivity I Pharmaceutical-
Biotechnology R&D: The Role of Experience and Alliances", NBER Working
Paper.
Davis, R. and L. Thomas (1993): "Direct Estimation of Synergy: A New Approach
to the Diversity-Performance Debate", Management Science, 39, 1334-1346.
Day, G. (1994): "The Capabilities of Market-Driven Organization", Journal o f
Marketing, 58 (October), 37-52.
Deng, Z., B. Lev and F. Narin (1999): "Science and Technology as Predictors and
Stock Performance", Financial Analysis Journal, 55 (May/June), 20-32.
Dosi, G. (1988): "Sources, Procedures, and Micro-economic Effects of Innovation",
Journal o f Economic Literature, 26 (3), 1120-1171.
Dutta, S., O. Narasimhan and S. Rajiv (1999): "Success in High-Technology
Markets: Is Marketing Capability Critical?", Marketing Science, 18 (4), 547-568.
Eckbo, B., V. Maksimovic and J. Williams (1990): "Consistent Estimation of
Cross-Sectional Models in Event Studies", Review o f Financial Studies, 3(3), 343-
365.
Fama, E. (1991): "Efficient Capital Market", Journal o f Finance, 46, 1575-1617.
112
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Galambos, D. and J. Sewell (1995): Networks o f Innovation, Cambridge University
Press.
Gatignon, H., B. Weitz and P. Bansal (1990): "Brand Introduction Strategies and
Competitive Environments", Journal o f Marketing Research, 27, 390-401.
Gavetti, G. (1994): "Strategies of Multinational Firms in The Patent Domain in
Europe", Working Paper.
Gilbert R. and C. Matutes (1993): "Product Line Rivalry with Brand
Differentiation", Journal o f Industrial Economics, 23 (Sep), 223-240.
Granstrand, O. (1998): "Towards A Theory of The Technology-Based Firm",
Research Policy, 27(5), 465-490.
Gradstrand, O., P. Patel and K. Pavitt (1997): "Multi-Technology Corporations:
Why They Have Distributed' Rather than 'Distinctive' Core Competencies",
California Management Review, 39 (4), Summer.
Granstrand, O. and S. Sjolander (1990): "Managing Innovation in Multi-
Technology Corporations", Research Policy, 19 (1), 35-60.
Greene, W. (1999): Econometric Analysis, Prentice Hall, Upper Saddle River, NJ.
Griffin, A. and J. Hauser (1993): “The Voice of Customer”, Marketing Science,
12(1), 1-27.
Griliches, Z. (1990): "Patent Statistics as Economic Indicators", Journal o f
Economic Literature, 28, 1661-1707.
Griliches, Z (ed.) (1984): R&D, Patents and Productivity, University of Chicago
Press, Chicago, IL.
Gronau R. and D. Hamermesh (2001): "The Demand for Variety: A Household
Production Perspective", NBER Working Paper.
Gupta, A., S. P Raj and D. Wilemon (1986): “A Model for Studying R&D-
Marketing Interface in the New Product Development Process”, Journal o f
Marketing, 50 (April), 7-17.
113
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Hall B. and R. Ham (1999): "The Patent Paradox Revisited: Determinants of
Patenting in the US Semiconductor Industry 1980-1994", NBER Working Paper.
Han, J., K. Namwoon and H. Kim (1998): “Market Orientation and Organizational
Performance: Is Innovation a Missing Link?”, Journal o f Marketing, 62 (4), 30-45.
Henderson, R. and I. Cockbum (1996): "Scale, Scope and Spillover: The
Determination of Research Productivity in Drug Discovery", RAND Journal o f
Economics, 27 (1), 32-59.
Horsky, D. and P. Swyngedouwn (1987): "Does It pay to Change Your Company's
Name? A Stock market Perspective", Marketing Science, 13, 320-335.
Houston, M. and J. Shane (2000): “Buyer-Supplier Contracts Versus Joint
Ventures: Determinants and Consequences of Transaction Structure”, Journal o f
Marketing Research, 37(1), 1-15.
Hsiao, C. (2002): Analysis o f Panel Data, Cambridge Press.
Hurley, R.F. and G.T.M. Hult (1998): "Innovation, Market Orientation, and
Organizational Learning: An Integration and Empirical Examination", Journal o f
Marketing, 62 (July), 42-54.
Jaffe, A. (1989): "Characterizing the Technological Position of Firms with
Application to Quantifying Technological Opportunity and Research Spillovers",
Research Policy, 18, 87- 97.
Jain, S. (2001): "An Analysis of the Impact of Patent Litigation on Product Pricing
and Licensing", Working paper.
John, G., A. Weiss and S. Dutta. (1999): "Marketing in Technology Intensive
Markets: Towards and Conceptual Framework", Journal o f Marketing, 63, 78-91.
Katila, R. and G. Ahuja (2000): "Something Old, Something New: A Longitudinal
Study of Search Behavior and New Product Introduction", Working Paper.
Kedia S, and A. Mozumdar (2002): "Performance Impact of Employee Stock
Options", Working Paper.
114
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Kekre, S. and K. Srinivasan (1990): "Broader Product Line: A Necessity to
Achieve Success?", Management Science, 36, 1216-1231.
Kerin, O., R. Varadarajan and R. Peterson (1992): " First-Mover Advantage: A
Synthesis, Conceptual Framework and Research Propositions", Journal of
Marketing, 56, 44-59.
Kholi, A. and B. Jaworski (1990): "Market-Orientation: The Construct, Research
Propositions and Managerial Implications", Journal o f Marketing. 4 (April), 1-18.
Klette, T. and S. Kortum (2002): "Innovating Firms and Aggregate Innovation",
NBER Working Paper.
Koen, M. (1991): Survey o f Small Business Use o f Intellectual Property Protection,
Washington, D.C. US Small Business Administration.
Kyriaziou, E. (1997): "Estimation of A Panel Data Sample Selection Model",
Econometrica, 65, 1335-1364.
Lang, L. and R. Stulz (1994): "Tobin’s Q, Corporate Diversification and Firm
Performance", Journal o f Political Economy, 102 (6), 1248-1280.
Lanjouw, J. and M. Schankerman (2001): "Enforcing Intellectual Property Rights",
Working Paper
Lanjouw, J. and M. Schankerman (2000): "Characteristics of Patent Litigation: A
Window on Competition", The Rand Journal o f Economics, 32 (1), 129-151.
Lemer, J. (1995): "Patenting in the Shadow of Competitors", Journal o f Law and
Economics, 38, 463-496.
Lemer, J. (1994): "The Importance of Patents Scope: An Empirical Analysis", The
Rand Journal o f Economics, 25, 319-333.
Levin, R., R. Klevorick, R. Nelson, and S. Winter (1987): "Appropriating Returns
From Industrial Research and Development", Brookings Paper on Economic
Activities, 783-820.
Lloyd, W. and J, Jahera (1994): "Firm-Diversification Effects on Performance as
Measured by Tobin’s Q", Managerial and Decision Economics, 15, 255-269.
115
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Loken, B. and D. John (1993): "Diluting Brand Beliefs: When Do Brand Extension
Have a Negative Impact", Journal o f Marketing, 23 (Julu), 71-84.
Manski, C.F. and D. McFadden (1981): Structural Analysis o f Discrete Data with
Econometric Applications, The MIT Press.
Matsusaka, J. and V. Nanda (1994): "A Theory of The Diversified Firm,
Refocusing, and Divestiture", Working Paper.
Montgomery, C. (ed.) (1995): Resources-Based and Evolutionary Theories o f the
Firm, Kluwer, Boston.
Montgomery C. (1994): "Corporate Diversification", Journal o f Economic
Perspective, 8(3), 163-178.
Montgomery C. and B. Wemerfelt (1988): "Diversification, Ricardian rents and
Tobin’s Q", The Rand Journal o f Economics, 19 (4), 623-632.
Moothy, S. and P. Papatla (1998): "SKUs as Traffic Builders", Working Paper.
Nayyar, P. (1992): "On the Measurement of Corporate Diversification Strategy:
Evidence From Large US Service Firms", Strategic Management Journal, 13, 219-
235.
Osterloff, M. (2003): "Technology-Based Product Market Entries: Managerial
Resources and Decision-Making Process", Working Paper.
Palepu, K. (1985): "Diversification Strategy, Profit Performance and The Entropy
Measure", Strategic Management Journal, 6, 239-255.
Pavitt, K. (1998): "Technologies, Products and Organization in The Innovating
Firm: What Adam Smith Tells Us and Joseph Schumpeter Does Not", Industrial
and Corporate Change, 7 (3), 433-452.
Pavitt, K. (1988): "Uses and Abuses of Patent Statistics", Handbook o f Quantitative
Studies o f Science and Technology, Amsterdam.
Peteraf, M. (1993): "The Cornerstones of Competitive Advantage: A Resource-
Based View", Strategic Management Journal, 14, 179-191.
116
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Porter, M. (1987): "From Competitive Advantage to Corporate Strategy", Harvard
Business Review, May-June, 43-59.
Priest, G. and B. Klein (1984): "The Selection of Disputes for Litigation", Journal
o f Legal Study, 8, 1-56.
Purohit, D (1994): "What Should You Do When Your Competitors Send in The
Clones?” Marketing Science, 13(4), 392-412.
Putsis, W. and B. Bayus (2001): "An Empirical Analysis of Firm’s Product Line
Decision", Journal o f Marketing Research, XXXVIII, 110-118.
Quelch, J. and D. Kenny (1994): "Extend Profits, Not Product Lines", Harvard
Business Review, Sep-Oct. 153-160.
Reddy, S. and S. Bhat (1994): "To Extend or Not To Extend: Success Determinants
of Line Extensions", Journal o f marketing Research, 31(May), 243-262.
Rivette K. and D. Kline (2000): Rembrandts in the Attic: Unlocking the Hidden
Value of Patents, Harvard Business School Press, Boston, MA.
Roberts, P. and S. McEvily (2000): "Product-Line Expansion and Resource
Cannibalization in Pharmaceutical Firms", Working Paper.
Robertson, T., J. Eliasburg and T. Rumon (1995): "New Product Announcement
and Incumbent Reactions", Journal o f Marketing, 59(3), 1-15.
Rogers, M. (2001): "The Effect of Diversification on Firm Performance", Working
Paper.
Rumelt, R. P. (1984): "Toward a Strategic Theory of The Firm", Competitive
Strategic Management, R. Lamb (ed.), Prentice Hall, Englewood Cliffs, NJ.
Scott, J. (1993): Purposive Diversification and Economic Performance,
Cambridge, NY.
Shankar, V., G. Carpenter, and L. Krishnamurthi (1999): “The Advantages of
Entering Early in the Product Life Cycle”, Journal o f Marketing Research, 36(2),
269-277.
117
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Shankar, V. and G. Carpenter (1998): “Late Mover Advantage: How Innovative
Late Entrants Outsell Pioneers", Journal o f Marketing Research, 34(1), 54-71.
Somaya, D. (2001): "My Strategy Says ' See You in Court!' Determinants of
Decisions Not to Settle Patent Litigation in Computer and Research Medicines",
Working Paper.
Sorescu, A., R Chandy and J. Prabhu (2001): “Sources and Financial Consequence
of Radical Innovation”, Working Paper.
Srivastava, R., T. Shervani and L. Fahey (1998): "Market-Based Assets and
Shareholder Value: A Framework for Analysis", Journal o f Marketing, 62, 2-18.
Stephan, M. (2002): "Diversification Profiles of Multinational Corporations: An
Empirical Investigation of Geographical Diversification, Product Diversification
and Technological Diversification", Working Paper.
Stevenson, R. (1980): "Likelihood Functions for Generalized Stochastic Frontier
Estimation", Journal o f Econometrics, 13, 57-66.
Sutton, J. (1998): Technology and Market Structure: Theory and History, MIT
Press, Cambridge, MA.
Teece, D., J. Pisano and A. Shuen (1997): "Dynamic Capabilities and Strategic
Management", Strategic Management Journal, 18, 121-135.
Teece, D. (1988): " Capturing Values from Technological Innovation: Integration,
Strategic Partnering and Licensing Decisions", Interfaces, 18(3), 45-61.
Teece, D. (1980): "Economics of Scope and The Scope of Enterprises", Journal of
Economic Behavior and Organization, 1, 223-247.
Trajtenberg, M. (1990): "A Penny for Your Quotes: Patent Citations and The Value
of Innovations", RAND Journal o f Economics, 21 (1), 172-187.
Verweire, K. (2003): "An Overview of The Literature on Corporate
Diversification", Working Paper.
118
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Vinod, H. and P. Rao (2000): "R&D and Promotion in Pharmaceuticals:
Conceptual Framework and Empirical Exploration", Journal o f Marketing, Theory
and Practice, Fall, 10-20.
Von Flippel, E. (1986): "Lead Users: A Source of Novel Product Concepts",
Management Science, 32 (7), 791- 805.
Waldfoget, J. (1995): "The Selection of Hypothesis and The Relationship Between
Trial and Plaintiff Victory", Journal o f Political Economy, 103, 229-260.
Wemerfelt, B. (1984): "A Resource Based View of The Firm", Strategic
Management Journal, 5, 171-180.
Williamson, O. (1999): "Strategy Research: Governance and Competence
Perspective", Strategic Management Journal, 20 (12), 1087-1108.
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix A
Estimating Marketing and R&D Capabilities: Stochastic Frontier
Estimation (SFE)
This appendix details the estimation of marketing and R&D capabilities. We start
with an overview of our estimation procedure, called Stochastic Frontier Estimation
(SFE), and then give details of its application to the estimation of both marketing
and R&D capabilities. Formally, we model a firm’s activities as an efficient frontier
or transformation function (akin to the notion of a “production frontier/function” in
economics) relating the resources used by a firm to the optimal attainment of its
objective(s). Conceptually, this involves two steps:
Step 1: We measure the maximum possible objective (output) the firm could
have achieved, given its set of resources (inputs). This tells us the best the firm
could have done, if it had used the resource level at its disposal efficiently, to
achieve its objective.
Step 2: We observe a firm’s actual performance, i.e., the level of objective it
actually attained. Given the estimate of the best, from step a) above, it is possible to
120
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
measure how far away its actual performance was from this best. The greater the
gap between its maximum achievable objective and its actual performance, the
lower its efficiency, and hence, the lower its capability.
The general specification can be written as:
Yu = f ( X it, a ) + eit = f ( x u, a ) + eu - Jjit
where Yit denotes the objective or output for the z th sample firm, i= 1,2,... N, in the
? th time period, t =1,2,...T. Xlt represents a set of explanatory variables describing
the resources (input) required to achieve the objective, a is a vector of coefficients
associated with the independent variables. Thus, f(X,,; a) gives the deterministic
component of the efficient frontier and represents the maximum expected output if
the firm could deploy its resources Xit at the most efficient level.
eit represents the composite error term. It is divided into two bits. The first, elt
represents the random error term, reflecting stochasticity in the environment. It has
a normal distribution, with zero mean and variance < y s . Formally:
= J -— exP
a £
r \2
£it_
\(J£J
121
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
The second part, rju represents the inefficiency error term, which is the key to
our measurement of capabilities. It has a truncated normal distribution, with mode
// and variance a n 2. Formally:
' M =
1 -o (-M /a r n)
exp
v &r] y
for T jjt £ [0,+°°]
or = 0 if else
where OQ is the standard normal distribution function.
While our interest is in estimating the inefficiency error term rjlU what we
actually observe is the composite error term, eit, i.e., eit - r]it = elt. Now, we know
the individual distribution of £ * , and r/lt. From this, we can compute the distribution
of eit. This distribution is given by (see Stevenson, 1980):
g { e n ) =
1
■ y j Orj + <7§
1 - 0
eit O 'r }
o eV
0 £TjOr]+ C?e Ojj-\jor]+Oey
X 0
f \
eit + V
^cr^ + Oey
1 - 0
r \
-JL
V Os)
From the distribution of eit above, we can form the required likelihood function
as follows:
122
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
N T 1
^ = n n . x
i-lt-l y (jjj +
X < f )
1 - 0
0’ £tJo*t]+ &e 0’ rj-^(Tri+ cr|y
/ \
Yit-f{Xit,a) + M
X
i
i
* — 1
t
y ^(Tri + Os
V < J e)
-1
But this likelihood function still does not give us estimates of rju which is what
we are interested in. To get that, we need to derive the conditional distribution of
rjit, given e ,-f. Formally:
V i = E[v\ei = e ] = f i + a l
f ~ \
-
(
- -1
_ M l .
2
1 - 0
M i
2
^ O * * ^
( 7 * *
V y
where:
M i =
2 — , M i<5e
~ane i + ------
2+ d :
n
-1
2 2
J 2 & V&£
and o** = -----------
u l + T icrl
A consistent estimate of firm i's inefficiency is given rj/y. Thus the consistent
estimate of capacity of firm i is:
CAP =
(C r / \
xlOO
which is the inverse of inefficiency term of the firm over that of industry average.
123
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
We now briefly mention the particular specifications for R&D and marketing
capabilities respectively. This involves a specification of the resources (inputs) and
objectives (outputs) for each of these capabilities. The estimation technique is
identical to that outlined above.
R&D Capability
We suggest that the objective of R&D activities is to maximize the production
of innovative technologies (Dutta et al., 1999). The major resources that the firm
has at its disposal to fulfill this objective include current and past R&D
expenditure, and the current technological knowledge base of the firm. Formally,
using a Cobb-Douglas production formulation, the R&D frontier can be specified
as:
\n(TECHINNVit ) = a 0 + a l x In (CUM _ RDEXPit ) + a 2x In (TECHBASEit) + eit - f]u
Marketing Capability
We suggest that the objective of marketing activities is to efficiently deploy
resources, such as advertising expenditure, along with its installed customer base,
to maximize sales. Formally, using a Cobb-Douglas production formulation, the
sales frontier can be specified as:
124
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
In(SALESit) = f i + f i x (ADSTOCKit) + f i x (CUSTOMERBASEit)
+ f i x (TECHBASEit) + £it~ B it
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix B
Standard Random Sampling and Choice-Based Sampling
Discrete choice model specifics probability for each of the possible choice an
individual can choose. P(i\Z,0) stands for the probability of choosing choice i by
individual, Z is a set of exogenous variables describing observed attributes of the
individual and factors that are supposed to affect the choice, 0 is a set of parameters
to be estimated based on the information about the choice of the individuals. The
estimation method depends on (1) the functional form of P{i\Z,ff) , (2) on sampling
procedure, and (3) on the extent of prior knowledge of the distribution of
exogenous variables Z (Manski and McFadden, 1981).
Different from random sampling, which is typified by the fact the individuals in
the sample are randomly selected from the population, choice-based sampling first
classifies the population into subsets based on the choice each individual makes
(which is endogenous)-, and then for each subset, a random sample is created. The
sample of population is the aggregate of all of these samples from subsets.
126
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Overlooking the difference among sampling procedures will lead to biased
estimates.
Likelihood function for standard random sample
In a random sample, the likelihood of observing choice i with information on
individual’s characteristics Z is:
f{i,Z | d) = P(i | Z,d)- /u(z)
Continuing the notation in the first paragraph, where p(Z) is the density
function for the exogenous variables Z. The log likelihood for a sample of size N is
therefore:
= ! > / > ( ;„ | z „ 0 ) + f ; M z . )
n = 1 n =1
Maximization of this likelihood function with respect to 0 (hence maximum
likelihood estimator) doesn’t require any knowledge of p(Z).
Likelihood function for choice-based sample
The log likelihood for choice-based sample is:
*■» ( M = Z i " p ( i, l z , . e) + £ '■> M z. ) ■ - Z s.l n a
n = 1 n~ 1 5=1
127
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
where (i and 0 are subject to the constraints:
\dZJ u{z)p{i\Z,9) = Qi
N - the total number of cases
Ni — the observed number of chases choosing alternative i, for i =1,..., M
S — the total number of sub-samples
N s— the number of cases in sub-sample s, for s=l,2,3,---, S
Qi - the proportion of the population choosing alternative I
< 2 , = ^ jeg(s)G, " the aggregate choice proportions Qi are known in advance
Hi = N i/N and Hs = N s/N
Replacement of p(Z) by an empirical distribution with weight factor 0)n leads to the
likelihood:
n=l n=1
which is to be maximized over d e © and a) e W , subject to the constraints:
128
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
J^(OnP{i\Zn,e) = Qi, i = 1,K ,M.
n=1
where:
W - jty | mn > 0 and 'Y jz n = l |
Prior knowledge of aggregate shares (proportions of the whole population that
select each alternative) will constitute a constraint in the estimation procedure and
enable the purely choice-based sample identifiable. This constrained maximum
likelihood estimator is then consistent and asymptotically efficient (Manski et ah,
1981).
129
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
Appendix C
Drug Classification System & Patent Classification System
IMS has developed a 5-digit hierarchical structure to classify any FDA approved
drug according to its therapeutic functions. It contains eighty-two major therapeutic
classes with each being represented by a different first 2-digit code. Each 2-digit
code (hence therapeutic class) has variant number of 3-digit codes representing
sub-therapeutic classes. Each 3-digit code (hence sub-therapeutic class) has variant
number of 4-digit codes representing sub-sub-therapeutic classes, so on and so
forth. Each digit code (except the first 2-digit code) represents different level of
class and the whole classification system has 4 hierarchical levels in total.
The following illustrates the drug classification system developed by IMS:
30000: Cancer/Transplant Therapy
31000: Cardiovascular Therapy
31100: Anti-hypertension Drugs
31110: ....
31120: ....
130
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
31130: ....
31140: Ace Inhibitors
31141: Ace Inhibitors, Alone
(Submarket)
31142: Ace Inhibitors with Diuretic
(Submarket)
USPTO has created a 9-digit hierarchical structure to classify patented technology
or technologic know-how. The first 3-digit code indicates major technology field.
Each of the first 3-digit code has variant number of sub-classes, represented by
different first 6-digit codes. Each 3-digit code represents different level of class and
the whole classification system has 3 hierarchical levels in total.
The following illustrates the patent classification system created by USPTO:
260,xxx,xxx: Chemistry of carbon compounds
260,568,xxx: Azodioxy
260,568,557: Sulfur attached directly to a ring by nonionic bonding
260,568,558: Diazo or dizonium
261,xxx,xxx: Gas and liquid contact apparatus
285,xxx,xxx: Railway mail delivery
131
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

An investigation of repeat visit behavior on the Internet: Models and applications

PDF

Essays on new product preannouncements

PDF

Customer satisfaction, customer bargaining power, and financial performance

PDF

Essays on technological evolution and financial returns to innovation

PDF

Information diffusion and the boundary of market efficiency

PDF

Inventory and pricing models for perishable products

PDF

Licensing and absorptive capacity in high -technology markets

PDF

Effects of redundancy in media coverage on nonprofessional investors' earnings forecasts

PDF

Information systems capabilities and indirect relational value: Sustaining networked organizations

PDF

Asymmetry and risk aversion in first -price sealed bid auctions: Identification, estimation, and applications

PDF

Does adding nonfinancial value drivers to a summary financial measure improve the learning and performance of managers?

PDF

Individual heterogeneity and program evaluation

PDF

Income inequality and economic growth: A theoretical and empirical analysis

PDF

"How will I feel about it?": The asymmetric impact of affective misforecasting on consumer satisfaction

PDF

Essays on commercial real estate and commercial mortgages

PDF

Influencing individual innovation through technology features that support cross -departmental understanding

PDF

Beyond skill development: The effects of training and development on the attitudes and retention of employees

PDF

Biopharmaceutical strategic alliances: Interorganizational dynamics and factors influencing FDA regulatory outcomes

PDF

Developing business communication skills: Leveraging stage versus global processes of change in skills improvement approaches

PDF

Identification and characterization of transcriptional regulatory elements in the Msx2 promoter

Asset Metadata

Creator Qu, Shaoming (author)

Core Title Essays on technology innovation and new product development strategy

School Graduate School

Degree Doctor of Philosophy

Degree Program Business Administration

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

Tag business administration, marketing,OAI-PMH Harvest

Language English

Contributor Digitized by ProQuest (provenance)

Advisor Dutta, Shantanu (committee chair), Hsiao, Cheng (committee member), Narasimhan, Om (committee member), Niraj, Rakesh (committee member)

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c16-647650

Unique identifier UC11335091

Identifier 3116771.pdf (filename),usctheses-c16-647650 (legacy record id)

Legacy Identifier 3116771.pdf

Dmrecord 647650

Document Type Dissertation

Rights Qu, Shaoming

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the au...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA