:2. ,. l:
i 3.. 3... if. .x J, Ll.

Jr
. I: n. .m : .Akfi 1'
$33 a»? L a. “weak.“
arm”, 1... Lmu

43..

, 3 . ,
. {:22}. . 2..
5591.. A93. .r.§~
Rafi-fin A yuan»

..

L

5..
5. .5 v.
.7: .3 2

. .
a
- Ema»...
MIC-“trifikmubvofith
.. 23...; 2%:
t
sinn.

.9.

r
, .zl}, iii}.
.uu‘t‘Airinff

\ arm; ‘

.‘
{lg-i -
‘3: .u‘v‘.
‘ ..}.....“....slz.:ns!. at” ..
Ltmmqt, . . .H...‘ 5......
3.1L; , 3.2.45.3 .... rawuu
I i. .1...

I), 2.551).!

a... 2 ...., s I.

.5..- 13; irissofiiun
3:;

Cit-iii! !
.2 illtll .5

E69

1.81.4. , x3 3
2.1.. 1&1)
:3.

9.)

 

THESIS

(7

~.—-

,‘ L If (7

EXAMINING THE RELATIONSHIP BETWEEN
RESEARCH AND DEVELOPMENT RESOURCE FLOWS AND KNOWLEDGE-
BASED CAPABILITIES: INTEGRATING RESOURCE-BASED AND
ORGANIZATIONAL LEARNING THEORY
By

Michael Eric Wasserman

A DISSERTATION
Submitted tO
Michigan State University

In partial fulfillment Of the requirements
for the degree Of

DOCTOR OF PHILOSOPHY
Department Of Management

2000

ABSTRACT
EXAMINING THE RELATIONSHIP BETWEEN
RESEARCH AND DEVELOPMENT RESOURCE FLOWS AND KNOWLEDGE-
BASED CAPABILITIES: INTEGRATING RESOURCE-BASED AND
ORGANIZATIONAL LEARNING THEORY
By

Michael Eric Wasserman

This dissertation addresses the development of knowledge-based
research and development capabilities. The resource-based view of the firm
suggests that knowledge-based capabilities can be an important source of
sustainable competitive advantage. Thus, the question Of how such capabilities
are developed is of great interest to both managers and researchers. A process
model is presented linking the concepts of research and development resource
flows, research and development capabilities, and firm performance. The model
is predicated on theories of learning and knowledge development that consider
the structure of knowledge as consisting of data and nonlinear relationships
among those data. The model is tested using fifteen years Of data from the
pharmaceutical industry. Although the data did not support any of the
hypotheses, the theoretical development Of nonlinear flow and stock

relationships provides a new lens to develop future research.

ACKNOWLEDGMENTS

This dissertation was challenging on many fronts. I greatly appreciate the
support and perseverance of my committee over this long process. Each one
contributed in unique ways above and beyond what can normally be expected
from a dissertation committee.

I also thank my family for their unconditional support and love - those two
intangible but tremendously valuable resources were dearly appreciated
throughout this process. The value placed on education by my parents and
grandparents has been markedly impressed upon me, and has most certainly
deeply influenced how I’ve spent the last decade.

Over the years, I have been extraordinarily fortunate to develop many
Close friends, each of whom has helped me in a variety of important ways. Most
of these folks I still talk with regularly - they know how grateful I am! Some of
them I’ve lost touch with, but all have been important and are appreciated.

I also want to thank someone whom I never met, and who I am quite
certain knows nothing about me. When I was a struggling adolescent, fighting
just to make it through junior high school, I read an article on the back page of
the Detroit Free Press about an old woodcutter from ldlewild, Michigan named
Andy Horujko. At age 66, he was retiring from woodcutting and returning to
college after a 45-year absence. The article included a quote that really stuck
with me and served as the core of the personal philosophy that I still hold strictly

to: “When a man stops learning he starts dying. I’ve got to learn something

every day or at least get a good laugh.” I’m sure if this old woodcutter is still
around, and knew how much that article 20 years ago meant to me, he’d get a
goodlaugh.

Most importantly, I thank my wife Sandy for her love, support, partnership,

and help. Without her, this undoubtedly would never have happened.

TABLE OF CONTENTS

INTRODUCTION ....................................................................................... 1
LITERATURE REVIEW ............................................................................. 4
The Evolution of the Resource-Based View of the Firm

— Three Eras .......................................................................... ‘ ............ 4
Resource Based Theory, 1991 -1 995 - Economic Justification &

Tacit Resources .................................................................................. 7
1996-1997: Integrating Resource-Based Theory ................................ 14
The Future of Resource-Based Theory — Focus on Knowledge-

Based Resources ............................................................................ 17
Back to Basics - Reexamining the Idea of Resource Flows and

Stocks ............................................................................................... 21
The Nature of Learning ....................................................................... 29
The Geometric Nature of the Learning Process ................................. 35
Learning and Innovation: An Organizational Perspective ................... 36
Potential Limits on the Development of Knowledge-Based

Capabilities ....................................................................................... 45
Summary ............................................................................................. 47

PROCESS MODEL AND HYPOTHESES ................................................ 49
A Process Model Of Knowledge Development in Organizations ......... 50
Step One - Flows and Stocks ............................................................. 44
Step Two - Consistency ...................................................................... 57
Step Three - Linkages Between Stocks and Performance ................. 61
Step Four - The Full Model ................................................................. 65

RESEARCH METHODOLOGY ................................................................ 69
Population, Sample, Timeframe .......................................................... 69
Dependent Measures ......................................................................... 73
Independent Measures .......... L ............................................................ 78
Data Sources ...................................................................................... 86

RESULTS ................................................................................................ 87
Data Analysis Strategy ....................................................................... 87
Data Structure and Results of the Diagnostic Analyses ..................... 88
Results of the Pooled Analyses .......................................................... 91

DISCUSSION ......................................................................................... 100
Summary of Findings ........................................................................ 100
Limitations ......................................................................................... 105
Future Research Directions .............................................................. 108

Conclusions ...................................................................................... 109

REFERENCES ...................................................................................... 111

vi

LIST OF TABLES

Table 1 - Foundation Resource-Based Citations (1984-1991) ................ 10
Table 2 - Selected Studies 1991-1995: Examining Tacit Resources ....... 13

Table 3 - Important Resource-Based Studies 1996-1997: Integrating and

Extending RBV ...................................................................................... 18
Table 4 - Linkages Between Data and Relationships ............................. 38
Table 5 - Companies Included in the Study ............................................. 70
Table 6 - Selected Studies Examining the Pharmaceutical Industry ....... 74
Table 7 - Hypotheses, Methods, and Variables ....................................... 80
Table 8 - Pharmaceutical Journals Used for Citation Counts .................. 81

Table 9 - Durbin-Watson d Statistics for Repeated Measures
MANOVA .............................................................................................. 90
Table 10 - Results of Repeated Measures MANOVA with Time Varying
Covariate .............................................................................................. 93

Table 11 - Repeated Measures Regression with an Auto-Regressive Error

Structure: Descriptives and Correlation Matrix ..................................... 94
Table 12 - Pooled Analysis, z-test Parameters - Hypothesis 1 ............... 96
Table 13 - Pooled Analysis, z-test Parameters — Hypothesis 2 .............. 98
Table 14 - Pooled Analysis - Hypothesis 3 ............................................. 99

vii

LIST OF FIGURES

Figure 1 - Influence of Additional Data Places on Knowledge Structure. 39
Figure 2 — Linkages Among Data — Building Knowledge ......................... 40
Figure 3 - A Process Model of Knowledge Development in

Organizations - Step One ..................................................................... 53

Figure 4 - Hypothesized Relationships Between Resource Flows and

Capability Stocks .................................................................................. 56
Figure 5 - The Importance of Consistency in Learning ............................ 62
Figure 6 - The Moderating Impact of the Consistency of Flows .............. 63
Figure 7 - Hypothesized Moderated Relationship ................................... 64
Figure 8 - The Complete Process Model ................................................. 67

viii

INTRODUCTION

Why have some firms consistently outperformed rivals? In search Of
answers to this question, the strategic management literature has extended
beyond its roots in industrial organization economics and its emphasis on
industry structure as the prime driver of firm performance. Indeed, the strategy
field has blossomed as researchers have begun to examine firm-level sources of
variance, originally thought to include diversification (Rumelt, 1974) and generic
business-level strategies (Porter, 1980). Students of organizations have
followed the development of the strategy field through theoretical growth in
areas such as strategic groups (Cool & Schendel, 1988) and upper echelons
theories (Hambrick & Mason, 1984), into the present. Currently, the resource-
based view of the firm is one theory rapidly becoming popular in the literature. It
is apparent that the shift to resource-based theory has made a significant impact
on the way managers manage, students learn, and academicians think about
competitive advantage (Barney, 1996; Montgomery, 1995; Wernerfelt, 1995).

As knowledge becomes an increasingly important aspect of the economy
(Grant, 1996; Spender, 1996), researchers must find a way to adequately
develop and test theory that reflects this phenomenon (Conner & Prahalad,
1996). Resource-based theory shows great promise to improve understanding
Of competition in knowledge-intensive industries, that is, those firms that rely on
knowledge as an important input, including the semiconductor, pharmaceutical,

and telecommunication industries. As the world shifts from a labor-based to an

information-based economy, the development and management of knowledge
becomes increasingly important. In knowledge-intensive industries, differences
in both the amount of knowledge and in the rate of learning (or the speed at
which the existing knowledge base changes) become key issues in
understanding interfirm performance differences, or why some firms consistently
outperform rivals.

Yet, there is considerable debate about the usefulness of resource-based
theory. The underlying logic of the resource-based view is intuitively appealing,
yet the combination of undefined internal processes and difficult to measure key
variables creates skepticism (Montgomery, 1995). This dissertation attempts to
Clarify these processes and measure some of the key variables by adopting
concepts from a range of disciplines and applying them to the notion of resource
flows and stocks, ideas previously mentioned by Dierickx and Cool (1989), but
given short shrift by many resource-based theorists (notable exceptions being
Conner, 1991 and Black and Boal, 1994). In an effort to develop and extend
resource-based theory, the impact of knowledge on the transformation Of
research and development resource flows into capability stocks becomes a
prime theoretical driver of interfirm differences in knowledge-intensive industries.
Concepts from cognitive psychology, sociology, and organizational theory are
reviewed and integrated in an effort to complement recent ideas addressed in
the strategic management literature about the development of sustained

competitive advantage in knowledge-intensive industries.

The body Of this dissertation is presented as follows: The second chapter
reviews resource-based theory and the literature on learning and innovation.
The third chapter integrates the issues raised in the literature review, and a
process model with related hypotheses is developed. The fourth chapter
addresses data collection and analytical methodology. The fifth Chapter details
the results Of the data analysis. Finally, the sixth Chapter discusses the findings,

limitations, implications, and directions for future research.

LITERATURE REVIEW

In an effort to better understand the relationship between firm resources
and firm performance in knowledge-intensive industries, this review explores and
integrates two research streams: the development of resource-based theory and
learning and innovation from an organizational perspective. This review is
particularly focused on the relationship between research and development
resource flows and knowledge-based capabilities, and the impact of both on firm

performance.

The Evolution of the Resource-Based View of the Firm - Three Eras

Over the past twenty-five years, the central premise of the strategic
management field has expanded from its roots in industrial organizational
economics to acknowledge that indeed, some firms can, and do, enjoy persistent
performance advantages over rivals (Aharoni, 1993; Barney, 1991; lstvan, 1992;
Peteraf, 1993; Wernerfelt, 1995). Therefore, the essential question for strategy
researchers has shifted from m; certain firms enjoy sustained extranormal
performance, to Mcertain firms enjoy sustained extranormal performance.

In an attempt to answer this question, strategic management researchers
have focused on studying the determinants of firm performance. In the first half
of the 1980s, the scope of strategic management research focused on product
and market characteristics, including studies based on industry environment

(Harrigan, 1980; Porter, 1980), form and level of diversification (Bettis, 1981;

Montgomery, 1982; Rumelt, 1974), and generic strategic archetypes based on
what and where the firm sold its output (Dess & Davis, 1984; Hambrick, 1983;
Miles & Snow, 1978). However, in response to the equivocal results from
research using these traditional product and market dependent variables,
Wemerfelt (1984) made a significant shift away from the popular research
streams of that time. Wemerfelt drew on the work of Edith Penrose who
proposed that the firm may be viewed as "a collection of productive resources"
(1959:24). Wernerfelt integrated this concept of viewing a firm in terms of its
inputs with the notion that certain resources can act as barriers to competition
(Porter, 1980), which gave rise to what is known today as the resource-based
view of the firm. This period of 1984-1991 marked the introduction of the
resource-based view of the firm into the strategic management literature.

The citations generally considered to be the foundation of the resource-
based view, published in this period, are summarized in Table 1. Barney (1986)
extended the resource-based view by discussing the linkage between strategies
and the resources necessary for implementing those strategies. Barney (1986)
used the concept of the strategic factor market to explain the importance of
managerial expectations about the future value of resources. The important
point here was that managers could gain superior expectations of the future
value of resources not by the traditional method of analyzing the environment
(because all managers would come to similar conclusions thus nullifying any
advantage), but in identifying the skills and resources residing within the

organization. This way, managers can acquire complementary resources that

have more value to their firms than to rivals, and thus gain an advantage in
assessing the future value of a given resource. Barney (1986) brought focus
and relevance to the Wemerfelt (1984) paper by showing that viewing the firm
as a collection of resources can fit in with traditional strategic management
techniques and offered initial insights on the link between resources and
competitive advantage.

Dierickx and Cool (1989) offered an alternative to Barney’s notion of
resource acquisition. Dierickx & Cool (1989) suggested that firms accumulate
resources over time and that the key to competitive advantage was in the
development of difficult-to-imitate bundles of interconnected resources that are
built over time. The process by which resources are accumulated was
discussed in terms of resource flows and resource stocks. This notion will be
discussed in more detail later in this chapter.

Prahalad and Hamel (1990) brought the resource-based view to
practitioners through the lens of core competencies in a widely cited Harvard
Business Review article. Prahalad and Hamel prescribed that managers should
identify the key bundles of resources that underlie their sources of competitive
advantage and both protect and leverage those bundles. In another important
article, Barney (1991) refined the underpinnings of the resource-based view,
arguing that firms can obtain extranormal performance by acquiring resources
that are rare, valuable, lack substitutes, and are difficult for rivals to imitate.
Barney cemented the linkage between resources and competitive advantage,

integrating the prescriptive logic of Prahalad and Hamel (1990) with the

conceptual clarity of Wemerfelt (1984) and Barney (1986). Barney (1991)
provided needed detailed definitions of resources and discussed how resource
based analysis fit into formal planning processes. Importantly, Barney (1991)
provided a framework (rare, valuable, lack substitutes, difficult to imitate) useful
to both researchers and practitioners.

By 1991, it was apparent to some that the resource-based view had great
potential for impacting how managers and researchers think about strategic
management of the firm (Conner, 1991). The seminal work by Wemerfelt (1984)
and Barney (1986, 1991) laid the theoretical groundwork, while Prahalad and
Hamel (1990) piqued the interest of managers and provided a real life context

for further study.

Resource-Based Theory, 1991-1995iconomic .jus_tif_ig_ation & Tmit Resogrces
The foundation laid by Wemerfelt and Barney in the 1980’s developed
into a fertile research stream in the 19905. Most recent studies follow Wemerfelt
(1984) and Barney (1986) by applying the resource-based lens to examine
competitive advantage. During the first half of the 1990’s, resource-based
theory development followed two relatively distinct paths. One stream
developed the economic justification of the resource-based view (Conner, 1991;
Peteraf, 1993). The second attempted to further define firm resources and tilted
toward the consideration of tacit resources (Castanias & Helfat, 1991; Hall,

1992, 1993).

The Economic Justification of Resource-Based Theog. Conner (1991)

and Peteraf (1993) underscored important linkages between industrial-
organization economics and the resource-based view. Their efforts provided
important economic justification for the proposition that firm resources are an
important source of interfirm performance differences and have helped the
resource-based view gain acceptance among management researchers.

Conner (1991) accelerated the justification of resource-based theory in
economic terms, providing a detailed comparison among traditional industrial
organization economic theories and resource-based theory. Her central
conclusions were that the resource-based view exists as both a complement and
an alternative to traditional industrial organization (l0) economic theory. Conner
asserted that resource-based theory shares key characteristics with the
neoclassical and Chicago economic views about the role of the firm, but is at
odds with several central assumptions, including perfect information and
resource mobility. Resource-based theory is considered similar to the
Schumpeterian process of creative destruction and Bain-type l0 theories in the
important assumption that persistent extranormal profits are possible in the long
run.

In Conner’s (1991) final analysis, two points stand out. First, an important
economic justification of resource-based theory is predicated on the idea that
interfirm performance differences can be attributed to the possession of unique
resources. Second, the recognition of resource stocks and flows, representing

the level at which resources are defined, is important in “preventing resource-

based theory from becoming tautological” (1991: p. 145). Both points are
important in later research using resource-based theory.

Peteraf (1993) presented the first resource-based model of firm
performance. The model was based in part on the generation of economic rents
through resource mobility barriers and other limits to competition. Thus, Peteraf
built on the justifications made by Conner (1991), and integrated work by
Wemerfelt (1984) and Barney (1986) into a concise statement of the
“cornerstones of competitive advantage.” Peteraf (1993) provided descriptive
examples of how her model could be applied across both business and
corporate levels of study within strategic management research.

Despite the increase in theoretical specification that resulted from the
work of Conner (1991) and Peteraf (1993), relatively few pure empirical tests of
resource-based theory had been conducted. Much of this can be traced to
difficulty in identifying and measuring resources (Conner, 1991). In an effort to
address the paucity of quantitative research, researchers started to tackle
specification and measurement issues, focusing at first on tacit resources.

Specifling Tacit Resources. Conner (1991) explicitly specified that 1)
unique resources are the key to competitive advantage and 2) that the
identification and measurement of these unique resources is a major obstacle to
empirical research using resource-based theory. However, many researchers
attempted to develop explicit resource definitions, primarily of tacit resources, in
an effort to bridge the gap that was preventing broad empirical testing of

resource-based theory.

Table 1

Foundation Resource-Based Citations (1984-1991)

 

 

 

 

 

Citation Contribution

Penrose (1959) Introduced the notion of the firm as a collection of
resources.

Wemerfelt (1984) Reintroduced the notion of the firm as a collection of
resources. Defined a resource as “anything which
can be thought of as a strength or weakness of a
given firm” (p. 172). Concluded that some resources
could be considered as ‘position’ barriers that can
inhibit competition.

Barney (1986) Resources are traded on factor markets, and play an
important role in firm performance

Dierickx & Cool Differentiated the ideas of asset stocks vs. asset

(1989) flows; specified the notion that time compression

diseconomies, interconnectedness, asset mass
efficiencies, asset erosion, and causal ambiguity are
drivers of the asset accumulation process, by which
inimitable resource stocks that support competitive
advantage are accumulated over time

 

Prahalad & Hamel
(1990)

Defined the term ‘core competence’ and discussed
the applicability of the resource-based view to
practitioners

 

Barney (1 991 )

 

 

Defined the nature of resources that support
competitive advantage: Rare, valuable, difficult to
imitate, lack substitutes

 

1O

 

Castanias and Helfat (1991) extended the ideas of Barney (1991),
specifying managerial expertise as an essential firm resource that generated
rent. Hall (1992, 1993) followed this stream by creating a framework of the
intangible resources and capabilities that support competitive advantage. These
resources include patents, trademarks, reputation, expertise, data bases,
culture, and organizational networks. Each of these resources was broken down
along three dimensions; the people dependence or independence of the
resource, skill versus asset, and the nature of the capability: functional, cultural,
positional, or regulatory.

Using case study data across industries and survey data, Hall (1993)
concluded that expertise and reputation were the two most durable types of
intangible resources and thus the most likely to support sustained competitive
advantage. Hall’s empirical work provided initial support for the arguments of
both Barney (1991) and Castanias and Helfat (1991) about the relationship
between resources and firm performance. Knez and Camerer (1994) defined
tacit resources such as reputation as nontradeable, and underscored the
importance of building these resources over time. Based on a lab study, Knez
and Camerer concluded that these tacit resources do indeed exist and are highly
complex, providing some additional empirical support for the fundamental
assumptions of Castanias and Helfat (1991) and Hall (1992, 1993).

Amit and Schoemaker (1993) made an effort to better specify the links
among resources, strategy, and performance. They defined important resources

and capabilities at both the industry and firm levels, and then identified decision

11

making patterns by which managers attempt to align firm-level resources and
industry-level resources in such a way as to produce sustainable rents. This
paper was an exemplar of work in 1992-93 that specified tacit resources in more
defined terms and set out logical boundaries of what should and should not be
considered a resource.

An important step in defining tacit resources came when researchers
started to explore and explain the complexity of these resources. In this vein,
Black and Boal (1994) theorized that resources must be utilized in
configurations, or network relationships over time to allow resource flows to
become non-tradeable asset stocks, the most likely resource configuration to
support sustained competitive advantage. Specifically, Black and Boal defined
resources in terms of their inherent nature (tradeable/non-tradeable and
flow/stock) and in the nature of the relationships among resources. Black and
Boal concluded that the longitudinal processes by which bundles of resources
are developed over time are important for both managers and practitioners in
determining the sources of sustained competitive advantage.

Barney and Hansen (1994) started to push the outer boundaries of the
types of tacit resources that can be considered sources of competitive
advantage, looking specifically at the nature of trust in economic exchanges.
Barney and Hansen (1994) identified several distinct forms of trust, relating each
level to the likelihood of that trust supporting competitive advantage. Other
studies in 1994 and 1995 continued to push on these boundaries as well. In an

interesting twist on the traditional approach in the diversification literature,

12

Robins and Wiersema (1995) adopted the commonly used construct of
relatedness as a proxy for important strategic assets. Their resource-based
index of relatedness appeared to provide an alternative to traditional entropy-
based measures and may, along with Stimpert and Duhaime (1997) who also
provided a new interpretation of relatedness, serve as a useful tool for a new
wave of more useful diversification studies.

Collis (1994) addressed the issue of capability development, delineating
three types of capabilities: the ability to perform basic functional activities of the
firm, the ability to make dynamic improvements in the firm, and the ability to
develop the “metaphysical strategic insights” (1994: p. 145) that allow managers
to develop and implement strategies that both use existing firm resources
develop important new resources. Collis acknowledged the resource-based
assumption that capabilities are asset stocks accumulated over time and thus
can form the basis of inimitability so important for sustained competitive
advantage. However, the key contribution of Collis was his identification of the
problems inherent in the use capabilities as a tool for prediction and explanation
of phenomena.

Collis (1994) concluded that capabilities are just another level in the
search for the sources of sustained competitive advantage, not the primary
source, and researchers would be well advised to remember this in future
research. Specifically, Collis advised that any given capabilities can always be
subsumed by higher-order capabilities, thus rendering any competitive

advantage gained by the original capability useless. Thus, in a few short years,

13

work on tacit resources appeared to come full circle from a theoretical standpoint
(see Table 2 for a summary of these studies), and as the preceding review
clearly shows, very few empirical resource-based studies were undertaken to
provide any support for the existence of these tacit resources.

Despite the prolonged delays in publishing solid empirical work,
increased theoretical specification of the economic bases and the nature of tacit
resources helped enhance the resource-based theory in its emergent stages
during the early 1990s. In the middle of the decade, both Barney (1996) and
Wemerfelt (1995) noted that resource-based theory was on the brink of
becoming a critical part of the mainstream strategic management literature. As
explained below, research published in late 1996 and early 1997 suggests that

these predictions may be coming to fruition.

1996-97: Integrating Resource-Based Theory

 

Recently, the resource-based view has been woven into existing research
streams, such as the strategic group, diversification, and strategic alliance
literatures, with interesting and meaningful results. Table 3 summarizes current
work using resource-based theory. For example, Mehra (1996) extended
strategic group research by introducing a resource-based explanation of
commonalities among strategic group members. Hypotheses about the sources
of performance variation, drawn from the resource-based approach, were

empirically supported.

14

Markides and Williamson (1996) extended the diversification literature by
presenting a new conceptualization and measurement of the relatedness
between businesses. They demonstrated empirically that their resource-based
measurement system improved on two traditional measures of relatedness, the
entropy measure and the Rumelt (1974) classification. Along similar lines,
Eisenhardt and Schoonhoven (1996) integrated the resource-based view with
the literature on strategic alliances, concluding that firms form strategic alliances
for more reasons than usually discussed; that is, justification built on transaction-
cost economics. These alternative reasons can be considered resource-based,
in that firms are attempting to propel themselves from threatening competitive
situations by rapidly building access to key strategic resources.

Although Eisenhardt and Schoonhoven did not specifically address
knowledge or expertise as a specific type of key strategic resource, it certainly
should be considered as a potential independent variable in future research.
Nevertheless, their data supported their hypotheses that firms undertake
strategic alliances in order to build essential resources, and marked another
successful effort at blending resource-based theory with traditional strategic
management viewpoints.

Other authors branched out farther, looking at areas often considered
outside traditional strategic management research streams, and used resource-
based theory to relate important issues back into a strategic management
framework. For example, Powell and Dent-Micallef (1997) applied the resource-

based view in a study that explored the linkages between information technology

15

Table 2

Selected Studies 1991-1995: Ex_a_mining Tacit Resources

 

 

Citation Contributions
Amit & Schoemaker Defined resources as “stocks of available factors
(1993) that are owned or controlled by the firm”

Resources “consist, inter alia, of know-how that
can be traded (e.g. patents and licenses),
financial or physical assets (e.g. property, plant,
and equipment), human capital, etc.” (p. 35)

 

Barney & Hansen
(1994)

Defined trust as a valuable resource, difficult to
imitate, and accumulated over time, and thus can
help support sustained competitive advantage.

 

Black & Boal
(1994)

Categorized resources in terms of traits,
configurations and paths using network theory
and the notion of flows and stocks.

 

Castanias & Helfat
(1991)

Defined the role of managerial expertise as a
resource essential for the development of
sustained competitive advantage. Argued that
resource-based logic better explains manager-
shareholder coalignment that aggmy theory.

 

Collis (1994)

Defined three categories of capabilities: ability to
perform basic functional activities, dynamic
improvement, metaphysical strategic insights

 

Hall (1992, 1993)

Categorized intangible resources by their
relationships to people, a firm’s asset base, and
where it lies in the organization. Concluded that
expertise and reputation were the two most
durable and important intangibles

 

Knez & Camerer
(1 994)

Defined expectational assets, or reputation, as
non-tradeable, developed over time, and
generating firm-specific rent. Concluded that
expectational assets (individual beliefs) exist and
are socially complex and difficult to develop
accurately

 

Robins &
Wiersema (1995)

 

 

Defined resources as knowledge or capability that
creates competitive advantage. Concluded that
the performance of large manufacturing firms is
impacted by relatedness, an indirect measure of
knowledge resources

 

16

 

and performance. Their data suggested that information technology alone is not
a source of sustained competitive advantage, but can support competitive
advantage when used in combination with complementary human and business
resources. Russo and Fouts (1997) looked at corporate environmental
performance as a firm resource and found an empirical link with firm
performance. Maijoor and van Witteloostuijn (1996) explored the relationships
among regulation, firm resources, and performance. This was done through a
reexamination of the levels at which resources exist. Maijoor and van
Witteloostuijn concluded that resources exist at the firm, group, and industry
level. Empirical analysis of the Dutch audit industry suggested that regulation

can be considered a resource at the industry level.

The Fugture of Resogrce-BgsLeg Theory - Focus on Knowledge-_Ba_sgt Rews
The preceding review supports one of the fundamental premises of
resource-based theory, that industry structure does not fully explain performance

patterns and that barriers to imitation do exist. As a result, firms are often
unable to imitate the specific factors that make some rivals more successful than
others. It is also apparent that capability development, not imitation, is the key
to competitive advantage (Lippman & Rumelt, 1982; Reed & DeFillippi, 1990).
This important point becomes even more relevant when considering a recent

and useful distinction of knowledge-based resources (Conner & Prahalad, 1996;

17

Table 3

Important Resogrce-Based Studies 1996-1997: Integrating and Extending RBV

 

 

Citation Contributions
Eisenhardt & Examined building firm resources through alliances.
Schoonhoven (1996) Strategic alliances are more likely when firms are in

a vulnerable strategic position, and led by large
experienced top mmt teams

 

Maijoor & van

Explored the link among regulation, firm resources,

 

Williamson (1996)

Witteloostuijn (1996) and performance. Concluded that regulation can
be considered a resource at the industry level.
Markides & Presented an alternative, resource-based

measurement of the relatedness, and demonstrated
empirically that it improved on traditional measures
of relatedness

 

Mehra (1996)

Conducted an exploratory study that showed a
significant relationship between firm resources and
performance. Certain configurations work better
than others.

 

Powell & Dent-
Micallef (1997)

Concluded that information technology supports
competitive advantage when used in combination
with other, tacit, human and business resources

 

Russo & Fonts (1997)

 

 

Examined corporate environmental performance as
a firm resource and concluded empirically that a
positive relationship with performance exists.

 

18

 

Kogut & Zander, 1996) and other types of resources. Knowledge-based
resources are important because they may impact the economic reasons that
firms organize.

Organization, as opposed to market contracting, makes available
expertise that can be jointly utilized (Conner & Prahalad, 1996). In a market
contract, expertise not related to the expressed purpose of the contract is not
shared, and cannot then be applied in future work. This is a very important point
because, by extension, it may justify the existence of the resource-based view of
the firm. That is, organization is important because it supports the accumulation
of knowledge and the development of valuable stocks of expertise that would
probably not be developed under a market contract relationship (Conner &
Prahalad, 1996).

It is important to remember that the resource-based view of the firm was
built on the idea that important interfirm performance differences are based on
valuable, rare, and difficult to imitate resources rather than products and markets
(Barney, 1986). One way to better address this notion of interfirm performance
differences in knowledge-intensive industries might be to turn back to the notion
of resource flows and stocks, a set of ideas that were presented relatively early
in the history of the resource-based view by Dierickx and Cool (1989), and is
widely cited by current resource-based researchers.

The flow and stock distinction came about as part of a discussion in the
literature between Dierickx and Cool and Barney. An early central theme was

the distinction between resource acguisition and resource accumulation. Barney

19

(1986) addressed resource acguisition, arguing that all resources were
essentially tradable on what he termed “strategic factor markets.” Barney
defined a strategic factor market as a market where resources required for the
implementation of a strategy are exchanged.

Dierickx and Cool (1989) countered this argument with a discussion of
resource accumulation, asserting that sustainable competitive advantage is built
on the deployment of nontradable assets not readily available on open markets.
Nontradable resources include reputation for quality, dealer loyalty, firm-specific
human capital, and research and development capability. Dierickx and Cool
argued that these nontradeable resource stocks are not available on any type of
strategic factor market, rather, that these can only be “built or accumulated
through a consistent time pattern of expenditures or flows” (1989: p. 1509).
This has two implications. First, since resource stocks cannot be purchased in a
market setting, these stocks must be developed by a firm over time. Second,
since these stocks cannot be sold in a market setting, these stocks must be
deployed in a product market in order to realize their value.

One way to address some of the differences between the concepts of
resource accumulation and resource acquisition is by examining the
conceptualization of how “flows” of resources develop into the configurations or
“stocks” of assets that are rare, valuable, and difficult for rival managers to
imitate. If, as Dierickx and Cool (1989) asserted, asset configurations or stocks

that develop over time are more difficult to imitate than the individual resources

20

purchased or acquired, then understanding interfirm differences in both strategy
and performance may rely on examining the underlying resource accumulation

patterns.

Back to Basics - Reexamining the Idea of Resou_rce Flows and Stocks

Dierickx and Cool (1989) presented a framework that described the
processes by which asset stocks accumulate over time to support the
development of sustained competitive advantage. The resource accumulation
process is based on assumptions that, over time, an organization develops a
stockpile of resources, and that resources “floW’ into and out of this "stock" as
resources are acquired and used up as the firm operates.

The flow and stock idea is not new, and more importantly, continues to be
refined. For example, McCarthy (1979, 1982) used the flow-stock
conceptualization as a tool for understanding the relationship between economic
events and accounting procedures. The idea was that balance sheet items,
capital assets that exist at a given point in time, are stocks. Stocks might include
plant, land, and inventory. Income statement items, cash that moves into and
out from the firm during a given time period are flows. Flows might include sales
revenues, rent, advertising expenses, and payroll.

Very recently, Samuelson (1996) called for enhanced clarification of the
conceptual framework of the FASB, and suggested that solidifying the definition
of stocks and flows would be beneficial within such a conversation. Samuelson

defined flows, in the accounting sense, as the movement of resources to or from

21

an organization during a defined set of time periods; whereas stocks are defined
as a factor that maintains the potential of creating future value for the firm at a
given time point in time.

In the econometrics literature, the flow and stock distinctions are used in
modeling continuous processes over time where variables need to be
understood 1) in terms of a certain state at a given point in time, and 2) as that
variable changes over time. Thus, the variable can be examined at a single
point in time, as a state, defined as a stock. Alternatively, the variable can also
be decomposed and studied in terms of the changes to the stock over a defined
time period. These changes in the state of the variable are defined as flows
(Lancaster 8. lmbens, 1995; Robinson, 1993).

Since Dierickx and Cool (1989), there have been two instances where the
flow and stock concepts have been discussed in detail within the strategic
management literature. Black and Boal (1994) built on the flow-stock
conceptualization as a key to understanding the nature of resources and
capabilities in their model of firm performance. Almeida (1996) used the flow
and stock idea to highlight the nature of innovation, arguing that information
flows and knowledge stocks lead to innovation.

In the few applications that applied flows and stocks in the strategy
literature, resource flows have been consistently defined as investments
intended to be transformed (or used to support the transformation) into the end
product or service; that is, flows are the building blocks of “capabilities” which

are then applied in product markets. Resource flows can be acquired

22

repetitively, as in regular purchases of raw materials or regular upgrades of
computer software. However, resource flows can also include one-time
purchases, such as raw materials, machinery, or computers. In a discussion of
flows and stocks in the accounting literature, McCarthy (1979, 1982) defined
flows as transactions whereby goods, services, or claims are brought into the
firm over a defined period of time.

For example, a firm can purchase 100 sets of pre-developed training
materials from a consulting company. This is a resource flow that is purchased
by the firm. Alternatively, the firm can hire and pay a training professional,
(where hiring costs would be a one-time resource flow and each pay period
would be considered a recurring resource flow), buy paper, materials, doughnuts
and coffee, and other essentials.

Resource stocks are defined as bundles of resources that are used to
create or add value to end products. Resource stocks can include expertise,
reputation, relationships with suppliers, and relationships with customers. These
bundles of associated resources are often associated with the term capability in
the literature (Black & Boal, 1994). In the accounting literature, McCarthy (1979,
1982) defines stocks as balance sheet assets consisting of goods, services, or
claims that grow or shrink in monetary value as resources flows move into and
out of that asset.

Examples include research and development capabilities (collections of
resources that provide a firm with the ability to invent and bring new compounds

to market), marketing capabilities (collections of resources that result in

23

developing valuable advertising, market research, and after-sale service),
reputation (for quality, for financial strength, for quality of work life, etc.),
manufacturing capabilities (resulting in flexibility, meeting tolerances, and
speed), and human resource capabilities (collections of resources that facilitate
recruiting, selection, compensation, training, even ‘downsizing’). In the earlier
training example, the expertise developed by employees that participated in a
successful training program would be the resource stock. Dierickx and Cool
concluded that “(t)he common element in all of these cases is that the strategic
asset is the cumulative result of adhering to a consistent set of policies over a
period of time. Put differently, strategic asset stocks are accumulated by
choosing appropriate time paths of flows over a period of time” (1989:1506).

Dierickx and Cool (1989) argued that sustained competitive advantage is
developed through the accumulation of interdependent stocks of resources.
Several processes that underlie this stock and flow conceptualization were
identified. These include time compression diseconomies, asset mass
efficiencies, interconnectedness of asset stocks, asset erosion, and causal
ambiguity.

Time compression diseconomies develop as assets are accumulated over
time. According to Dierickx and Cool:

Conceptually, time compression diseconomies and the notion of

“strictly convex adjustment costs” in the theory of capital

investment to which they are related express the same

fundamental mechanism: the “law of diminishing returns" when

one input, viz. time, is held constant. For example, MBA students

may not accumulate the same stock of knowledge in a one-year
program as in a two-year program, even if all inputs other than time

24

are doubled. In the case of R&D, the presence of time
compression diseconomies implies that maintaining a given rate of
R&D spending over a particular time interval produces a larger
increment to the stock of R&D know-how than maintaining twice
this rate of R&D spending over half the time interval (1989: p.
1507)

Asset mass efficiencies relate to the idea that the asset accumulation
process is facilitated as firms accumulate large asset stocks (Dierickx & Cool,
1989). According to Dierickx and Cool:

Sustainability will be enhanced to the extent that adding
increments to an existing asset stock is facilitated by possessing
high levels of that stock. The underlying notion is that ‘success
breeds success’: historical success translates into favorable initial
asset stock positions which in turn facilitate further asset
accumulation. For example, firms who already have an important
stock of R&D know-how are often in a better position to make
further breakthroughs and to add to their existing stock of
knowledge than firms who have low initial levels of know-how. . .
The competitive implication is clear: when asset mass efficiencies
are important, building asset stocks starting at low initial levels may
be difficult (1989: p. 1507-08).

lnterconnectedness of asset stocks refers to the notion that certain
resources are complimentary, and as such, can result in superior returns when
accumulated and implemented together. As Dierickx and Cool explained,
“(a)ccumulating increments in an existing stock may depend not just on the level
of that stock, but also on the level of other stocks” (1989: 1508). For example, a
firm that possesses outstanding process technology resources, resulting in low
production costs, will earn greater returns from these assets when the firm also
maintains high levels of marketing expertise, resulting in the ability to generate
high levels of brand recognition and the perception of value. According to
Dierickx and Cool, “(h)ere, the difficulty of building one stock is related, not to
the initial level of that stock, but to the low initial level of another stock which is

its complement” (1989: 1508) Thus, the interconnectedness of asset stocks

25

contributes to causal ambiguity and uncertain imitability, as the nature of these
complementary relationships may not be precisely visible to rival managers.
Asset erosion occurs when resource stocks, which deteriorate over time,
are not replenished (Dierickx & Cool, 1989). Thus, asset stocks must be
continuously maintained if competitive advantage is to be sustained. For
example, relationships with suppliers or customers must be maintained in order

for those assets to remain productive. Dierickx and Cool stated the following:

R&D know-how depreciates over time because of technological
obsolescence, brand awareness erodes because the consumer
population is not stationary (existing consumers leave the market,
while new consumers enter), consumers forget, etc. The
characteristics of this decay process have several managerial
implications. There is an important relation between an asset’s
‘half—life’ and strategic entry deterrence. To credibly deter entry,
firms must be committed to punitive post entry behavior. Thus,
output and advertising policies are not, in general, credible
vehicles for entry deterrence, whereas capacity and brand loyalty
are. The reason is that the former, pertaining to flow variables,
could be adjusted at will should entry occur, whereas the latter,
being stock variables, cannot. (1989: p. 1508).

Dierickx and Cool argued that “managers often fail to recognize that a
bundle of assets, rather than the particular product market combination chosen
for its deployment, lies at the heart of their firm’s competitive position” (1989: p.
1504), and concluded that “a firm’s current strategy involves choosing optimal
time paths of flows, whereas its competitive position and hence its potential
profitability is determined by the level of its stocks ” (1989: p. 1510). This
argument is consistent with the overall role of resource accumulation in the

strategy process.

26

Dierickx and Cool argued that resource accumulation is both the direct
result of past patterns of strategic decisions _a_rfi the driver of future strategic
decisions. This is consistent with Mintzberg’s (1973) definition of strategy as a
pattern in a stream of decisions. Thus, resource accumulation is embedded in,
and is a product of, the long term strategy of the firm. This argument makes the
ability to identify and study resource accumulation patterns even more essential
to a unified understanding of the sources of firm performance. These strategic
implications are just one result of using the Dierickx and Cool framework to study
the determinants of firm performance.

Other constructs in the strategic management literature are illuminated
when examined through an asset accumulation lens. For example, Thompson
(1967) used the term “technical core” to describe the internal processes that
firms use to transform inputs into outputs. Thompson held that the technical
core was, for many firms, a strategic asset that must be buffered from external
forces. While Thompson drew attention to the significance of the internal
processes of the firm, he did little to specify the nature of the technical core.

Since Thompson’s book, many researchers have amplified this pattern;
that is, highlighted the importance of internal processes while ignoring the
specification of those processes. The simple input-’black box’-output approach
has consistently failed to consider these internal processes, and it should be
little surprise that these researchers failed to discover consistent relationships
between input and output variables. Specifically, researchers including Cohen

and Levinthal (1989) and Narin, Noma, and Perry (1987) have looked before at

27

the relationship between R&D expenditures and patents. The problem, or what
has escaped notice or rigorous study, has been how R&D expenditures feed or
develop the innovation process, or how marketing expenditures build brand
awareness and loyalty.

There is a link between tangible resources, such as R&D expenditures,
and outputs, such as patents, that most researchers skirt over. The value of the
Dierickx and Cool framework is that it directs attention to not only resource flows,
but on the resource stocks that make up Thompson’s “technical core,” providing
clearer understanding of how R&D dollars are transformed into R&D
performance. Put another way, the key resource is not in the easily imitated
R&D expenditures, but the strategic component of how the R&D dollars are used
to build the capability to innovative.

Dierickx and Cool argued that their framework details the process by
which flows of inputs are developed into intangible resource configurations that
are then deployed in product markets and can be used to drive competitive
advantage. Consistent with the Dierickx and Cool framework, Prahalad and
Hamel asserted that “(c)ulitvating core competence does not mean outspending
rivals on research and development” (1990:83).

The key contribution of Dierickx and Cool is that the stock of resources
must not be ignored. Understanding the processes that drive the creation of
these capabilities is an essential component of identifying the internal processes
that occur within the “technical core” of the firm. In summary, the concepts of

resource flows and stocks are important conceptualizations of how rare,

28

valuable, and difficult to imitate capabilities develop. Thus, Dierickx and Cool
( 1989) added an additional and complementary dimension to the seminal
Wernerfelt (1984) and Barney (1986) cites. Specifically, the identification of
resource accumulation patterns should provide a clearer understanding of how
firm resource stocks develop over time. Better understanding of the resource
stock development should allow greater understanding of interfirm performance
differences. The next step in building a theoretical basis for understanding the
flow and stock relationship and the resulting impact on performance lies in the

literature on individual and organizational learning.

The Nye of Leaflng

The study of individual learning has been evolving for well over one
hundred years. Theorizing about learning and knowledge development spans
several disciplines, including sociology, psychology, philosophy, economics, and
more recently, organizational theory and strategic management. In the past
thirty years, a substantial literature on organizational learning has emerged as a
sub-discipline to specifically address how knowledge develops within firms
(Levitt & March, 1988; March, 1991). Within the last decade, strategic
management researchers have started to examine to impact of cognition and
associated learning processes on firm performance (Barr, Stimpert, & Huff,
1992; Prahalad & Bettis, 1986; Spender, 1989; Walsh, 1995; Zajac & Bazerman;
1991). In the strategy domain, much has been gained from this work, yet

researchers have yet to ask important questions about the nature of knowledge

29

development and how the process by which knowledge accumulates can lead to
sustained performance differences between firms (Aharoni, 1992; Walsh, 1995).

Specifically, broad questions remain about why certain firms get more
“bang for the buck” from research and development resource investments, why
firms making similar levels of research and development investments develop
widely dissimilar innovation capabilities, and why firms that cease making
investments in certain research and development resources continue to enjoy
fruits from knowledge-based capabilities for many years. The strategic
management literature suggests one key may lie in improving understanding of
the relationship between research and development resource flows and
knowledge-related resource stocks (Dierickx & Cool, 1989).

Further, as Levitt and March (1988), Senge (1990) and others have
suggested, building and maintaining “learning organizations” may be a key to
understanding the development of those key knowledge based resources that
lead to competitive advantage. If this logic is to be developed, we must first
clearly define knowledge, learning, and other related terms. As Boisot stated,
“(t)he terms ‘knowledge’, ‘information’, and ‘data’. . .tend to be used
interchangeably. . . Failure to clarify the terms used sometimes leads to some
pretty loose thinking. . .” (1995; p. 450).

Knowledge. Knowledge, within any given domain, consists of both 9%
and the relationships abstracted from data and used for decision making. A
piece of data, at the most fundamental level, is defined as “a discernible

difference in the energy states of phenomena as they occur and propagate in

30

space-time, whether as matter or electromagnetically" (Boisot, 1995, p. 22).
Data exist, then, when variance in the environment stimulate our sensory
apparati. The signals received by the brain from the sensory organs, called
perceptions, are attended to, recognized, encoded, and stored in memory.

This conceptualization of knowledge is consistent with Newell and Simon
(1972), who found that individuals store bits of information (analogous to data
herein) and subroutines (similar to linked pieces of data herein) in memory and
then recall and organize these in response to stimuli in the environment for use
in decision making. This is also consistent with F iegenbaum’s (1970) theory of
perception and memory at the individual level, and also with the action theories
of Argyris (1976) and Argyris and Schon (1978) at the organizational level.

According to theory that addresses knowledge development, there are
three types of knowledge: declarative, structural, or procedural (Jonassen, et
al., 1993). Declarative knowledge is the recollection of stored facts, or data
labels (Anderson, 1982). However, declarative knowledge may be tacit (Polanyi,
1966). For example, declarative knowledge would be recalling the atomic weight
of the element molybdenum, or recalling the average salary of tenured
organizational behavior faculty at Big 10 universities. Procedural knowledge is
knowing how to do something (Anderson, 1982; Anderson & Fincham, 1994).
For example, procedural knowledge would include analyzing a sample of the
element molybdenum in a mass spectrograph for purity, or using the World Wide
Web to locate the average salary of tenured Organizational Behavior faculty at

Big 10 universities is located using the World Wide Web.

31

Structural Knowledge. Structural knowledge is the type of knowledge that
recollects monships amongiaLa (Jonassen, et al., 1993). Structural
knowledge can be either causal or categorical. Causal knowledge is reflected in
simply understanding the impact of g on p. Categorical knowledge entails
placing a in group of similar facts or concepts, and linking it to being dissimilar to
other groups of related facts (Jonassen, et al; 1992). Regardless of the
causal/categorical distinction, this type of knowledge consists of related data
that are interconnected, or linked, and organized in such a way as to provide
explanatory power. For example, structural knowledge is the recollection of the
expected reaction when the element molybdenum is combined with hydrogen
gas under pressure. This knowledge is driven by recalling specific data about 1)
the molecular structure of molybdenum, 2) the molecular structure of hydrogen,
3) how the two elements would combine, and 4) the impact of pressure on this
combination. The data must be organized so that causal relationships are clear
and can be applied to decision making and action. In this case, the expert
chemist would know that the resulting combination would result in explosion and
would not carry out the experiment.

In the other example, structural knowledge about average tenured faculty
salaries would include the relationship, or comparison among, several pieces of
data: 1) the average Big 10 salary, 2) the Michigan State University (MSU)
average salary, and 3) knowledge about faculty research productivity at other
Big 10 schools. The comparison among 1) and 2) would lead to the addition of

data 3) and the “why” component of the knowledge. In this simple example,

32

these pieces of data would be organized and linked to determine that MSU’s
average salary is higher than most Big 10 schools because our faculty is more
productive. It should be evident that all three types of knowledge: declarative,
procedural, and structural are important; however, structural knowledge binds
declarative and procedural knowledge together and is important for the

development of knowledge related capabilities in organizations.

Cognitive Structures and Mental ModLels. The information processing
approach used by cognitive psychologists has been extended and developed
into a theory of cognitive structures. This theory that has rapidly entered into the
organization sciences literature (Klimoski & Mohammed, 1994; Walsh, 1995).
The term cognitive structure is defined as “the organization of the relationships
of concepts in long-term memory (Shavelson, 1972, pp. 226-227). It is similar to
the terms “knowledge structure” (Neisser, 1976) and “mental model” (Johnson-
Laird, 1983; Klimoski and Mohammed, 1994; Rouse & Morris, 1986) that have
appeared in the managerial literature (Walsh, 1995). The cognitive structures
represent data and the relationships among those data; thus providing an
important explanation for how and why information or data becomes knowledge
or expertise (Klimoski & Mohammed, 1994). Cognitive structures provide a
measurable construct that identifies expertise and learning, and is capable of
addressing the rate and level of change in knowledge over time.

It is relatively straightforward to apply the concept of cognitive structures

to the concepts of learning and knowledge development. Importantly, the notion

33

of cognitive structure provides an ex1planation for organizational learning that is
relevant at both the individual and organizational levels (Klimoski & Mohammed,
1994; Walsh, 1995) and provides justification for the development of expertise
that is, on one hand, difficult to imitate, yet on the other hand, transferable

across time and space.

_De_fining Legming Usinnggnitive Structures. One way to conceptualize
learning is through this idea of the cognitive structure. Learning is defined as
change in the cognitive structure, or in other words, changes in both the amount
of data, and/or the relationships among those data (Jonassen, et al., 1993).
Learning a new piece of data adds to the network and also enhances the
knowledge structure in a much more powerful way. That is, each new piece of
data builds increased relationships among several other discrete pieces of data
by providing the individual or organization with increased understanding about
those facts in relation to the newly learned bit of data.

At the most fundamental level, simple declarative knowledge is applied
recalling the stored labels that represent data. But in organizations, more
complex associative and causal forms of knowledge are invoked to make
inferences and decisions. From this, reasoning and problem solving processes
can occur. The data (or more precisely, the stored labels that represent data)
are not necessarily important in and of themselves to an organization, but they
are critical in that these data are building blocks for more complex and

commercially important relationships and associations. The amount of data

34

stored or known is important because it determines the amount of commercially

relevant and complex knowledge that can be used in the competitive arena.

The Geometric N§t_u_re of the Lea_ming Process

When looked at as a whole, the literature on cognitive structures
suggests that the relationship between the number of discrete pieces of data
stored and the number of relationships among those data (structural knowledge)
is curvilinear. The more an organization knows, the more an organization
benefits from each additional piece of data learned. For each piece of data
added to the knowledge structure, multiple relationships among data become
known or further elaborated and enriching existing relationships.

Knowledge grows through learning in a curvilinear manner -- the more
data known, the greater the knowledge gained from new data. Put another way,
learning is characterized by knowledge that is increased at the most basic level
in a linear manner by simply adding one piece of data to an existing cognitive
structure. However, a more realistic understanding consistent with recent
advances in cognitive psychology, recognizes that knowledge grows in a
nonlinear fashion. When a new piece of data is retained, the learner stores it in
the brain, and during this process, relates this new piece of data to other
previously known data. As a result, the learner now knows about the individual
piece of data, but also knows more about other related pieces of data and the

relationships among them. Thus one new piece of data adds to the knowledge

35

based, but also increases the knowledge about existing data, as those
relationships become known more fully.

The table and graphs below look at three situations. The far right column
in Table 4, represented on Figure 1 by the top (dark solid) line, represents the
relationship between the addition of individual data points and the resulting
number of direct relationships by which the knowledge structure is increased.
This assumes that one additional piece of data is linked to all other pieces of
data in that particular knowledge domain.

However, this assumption may not always hold. For a variety of reasons,
there may not be relationships among all data in a given domain. The middle
and left columns, and the middle (diamond-marked) and lowest (dotted) lines in
Figure 1 show the relationship between an additional piece of data and the size
of the knowledge structure under two additional conditions. Each line shows
knowledge structure growth under conditions where 75% and 50% of data are
‘linked’ or associated in some way with the existing fact base in the domain area.
It is important to note that these relationships only consider direct links among
data, and do not include indirect links. Thus, the number of relationships

generated from additional data is probably quite conservative.

Learning and lnnovéation: An Organizationpl Pegective
The nature of, and relationship between, the constructs of learning and
innovation have been widely addressed (Cohen & Levinthal, 1990; Cyert 8

March, 1963; David, 1985; Nelson & Winter, 1982; Simon, 1976) in the

36

management and economics Iiteratures. Learning has been described as
cumulative (Hedberg, 1981 ), and includes processes such as perception,
attention, selection, memory storage, retrieval, reasoning, and problem solving
(Ashcraft, 1989). The relationship between individual and organizational
learning is posited to be dynamic. “Organizations do not have brains, but they
have cognitive systems and memories" (Hedberg, 1981; p. 6), including
customs, values, standard operating procedures, policies, and myths.
Organizations can leverage individual learning by facilitating the development of
accurate mental models or cognitive structures, by decreasing the cognitive
effort involved in ramping up to master new tasks, and by providing “shortcuts”
that foster continued success. This can be facilitated through training or
mentoring. Hedberg (1981) emphasized that organizational learning is not
merely the accumulation of individual learning. Thus, despite the fact that
organizations generally “knoW’ less than the sum of the individuals that make up
the organization (Hedberg, 1981), two important points can be deduced. First,
individual learning can act as a “multiplier”, as one person’s individual learning
or expertise is rapidly transferred through organizational routines and other
structures to a large number of other employees. Second, new structures that
enhance organizational learning (such as new forms of “organizational memory”
or new pathways to disseminate stored organizational knowledge) can rapidly
enhance or “multiply” knowledge at both the individual and organizational levels.

These two points underscore the importance of learning in the organization, and

37

Table 4

Linkages Between Data and RelationshiLs

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Number of Number of Number of Number of
discrete data relationships relationships relationships (100%
bits (50% relatedness) (75% relatedness) relatedness)
1 0 0 0
2 1 1 1
3 1 2 3
4 3 4 6
5 5 7 1 0
6 7 10 1 5
7 1 0 14 21
8 14 21 28
9 1 8 24 36
1 0 22 33 45

 

 

 

 

200
1 80
1 60

 

 

 

3%

 

+ 1 00%
-fi— 75%
— 50%

 

120
100

 

 

 

 

G
O

 

 

 

 

a:
O

 

 

Number of Relationships
.5
O

 

 

N
O

 

 

"'"I'MIIIIIIFTVITTIIIII

0 5 10 15 20

Number of Pieces of Data

 

O
1

Figure 1. Influence of Additional Data Pieces on Knowledge Structure

39

 

 

 

 

 

Figure 2. Linkages Among Data - Building Knowledge

40

highlight the potentially powerful relationship that may exist underneath the
visible resources that the organization controls.

However, questions remain: How does learning relate to resource flows
and stocks? How does learning relate to firm performance? One way to address
this issue was broadly discussed by March (1991), who examined the link
between organizational learning and competitive advantage. March framed the
issue as a dilemma between “exploiting the known” and “exploring the new”.
Firms generally have some ability to “explore the new”, but they have continually
improve the ability to innovate in order to build competitive advantage. This
process was further delineated by the idea of absorptive capacity introduced by
Cohen and Levinthal (1990).

Absorptive Capacity. Absorptive capacity is the ability of organization to
recognize, assimilate, and apply outside knowledge to internal innovative
processes (Cohen & Levinthal, 1990). Important factors in the development of
absorptive capacity include internalization, routinization, and cumulativeness.
Thus, “(t)o the extent that an organization develops a broad and active network
of internal and external relationships, individuals’ awareness of others'
capabilities and knowledge will be strengthened. As a result, individual
absorptive capacities are leveraged all the more, and the organization’s
absorptive capacity is strengthened” (Cohen & Levinthal, 1990; 134).

Absorptive capacity is, in essence, the organizational capacity to learn.
Given that data exist in the environment, then organizations must develop the

capacity to notice, attend, encode, and store these data. Usually these data are

41

‘sensed’ (noticed, attended, and encoded) by its members, although
increasingly, computer power is driving automation of these tasks.
Organizational structures serve as memory, including databases, policies, and
myth (Hedberg, 1981). The organization of data within these structures, and
the ability to infer categories and causation from these data by the individuals in
the organization complete the development of knowledge structures within an
organization. Thus, the greater an organization’s absorptive capacity, the
greater the likelihood it can build structures that result in the development of
knowledge.

Absorptive capacity creates learning, or knowledge growth, by building
mechanisms for individuals to sense, notice, and encode data or changes in the
environment. Traditionally, knowledge is thought to be built in firms by storing
data within the organizational structure in computer databases, manuals, or
within individuals (March & Simon, 1993). Relationships among those data are
made as access to data within the organization is shared and organized such
that categorical and causal links are elucidated. Over time, these structures (the
data and the relationship among the data) change, and this is the point when
learning occurs. Structural change implies that knowledge is created and is
ready to apply in the competitive environment. The context-specific knowledge
developed can, as Prahalad and Hamel (1990) argued, create competitive
advantage.

The development of absorptive capacity is the result of resources

invested in the organization, by the organization. The firm invests in R&D

42

related infrastructure to facilitate the collection and storage of data. Employees
create relationships with others both within and outside the organization and
share information, both adding data and building relationships among data.
Examples include purchasing computer hardware and database software,
creating and supporting cross functional teams, building a culture where
employees are encouraged to interact with peers through publication and joint
research, or by developing joint ventures and technology sharing alliances. All
of these increase the amount of data available for learning and improve the
likelihood that relationships among those data can be understood.

As Cohen and Levinthal (1990) stated, absorptive capacity is largely a
function of an organization’s prior level of knowledge. This is because
awareness of new knowledge is dependent to a great extent on having some
level of existing knowledge, so that important information can be recognized as
important. Additionally, new knowledge is often built onto existing knowledge.
In this manner, a new piece of information can be compared and contrasted to a
wider base of initial knowledge, providing a greater probability that the new
information can be used in a productive manner. As such, a reinforcing
relationship exists between existing capacity in a knowledge—based activity and
the future potential for knowledge-based activity, the impact of which was
described by Cohen and Levinthal as, “dynamically self-reinforcing behavior”
(1990: p. 138). As the capacity to absorb data (learning) increases, the amount
of data absorbed increases, and vice-versa. As these data are stored,

organized and used, the organization increases its ability to further absorb or

43

learn from the environment, up to limits based on size and structure of the
organization.

An example of the learning process is evident in the time required to fully
utilize resources such as computers or buildings. Employees must learn how to
use and integrate the new resource into work routines. This is consistent with
arguments advanced by Arrow (1962), who found empirical support for his
hypothesis that technical change is derived from learning over time.

Current research is starting to reflect greater understanding of the impact
of learning on innovation. The essential nature of the accumulation of
knowledge was discussed by Helfat (1994). She posited, from an evolutionary
perspective, that tacit knowledge and the accumulation of learning would
combine to cause persistent interfirm differences in patterns, or trajectories, of
research and development expenditures. Helfat used evolutionary theory to
support her argument that research and development expenditures were path
dependent, and that persistent differences exist among firms that compete in a
single industry. Helfat’s data supported these two broad hypotheses, implying
that there are persistent differences in research and development expenditures.

While Helfat (1994) made an interesting and important set of arguments,
a gap in the literature exists. Helfat implied, but did not address the issues of
organizational learning and its impact on relationships between R&D flows or
expenditures, which she studied. In fact, in her conclusions, Helfat stated:

It would be helpful to analyze the joint evolution of bundles of

activities within the firm. For example, this study tested for the
importance to R&D activity of ties to related firm-level assets and

44

investment. While the investment and asset variables were jointly

significant, in most of the regressions the variables had relatively

low coefficient estimates (even when standardized). Path

dependence in R&D does not preclude ties of R&D to related firm

resources, or the joint evolution of R&D and other firm resources.

As a first step, the analysis performed here could incorporate

longer term lags of both R&D expenditures and of related

investment and assets to examine the persistence for all three

variables over time. It also might help to separate the analysis of

investment and assets from that of lagged R&D.

In summary, the persistent differences in petroleum firms’

R&D activities documented here suggest that the associated

knowledge bases change only slowly (Helfat, 1994, p. 1745).

Thus, from a research and development standpoint, absorptive capacity
appears to be an important property, driven by organizational learning, that firms
use to facilitate innovation. Firms use absorptive capacity to maintain existing
knowledge resources and to build new knowledge (learning). Through these two
processes, firms in knowledge-intensive industries can develop the building
blocks of competitive advantage. Helfat’s (1994) findings reinforced the
assertion that the process by which capabilities create innovations is an
important aspect in examining how firm capabilities or resource stocks develop.
However, another important issue, similar to the issue raised by Collis (1994), is

the issue of potential limitations on the development of knowledge-based

capabilities.

Potential Limits on the Development of Knowledge-Based Capabilities

Research in the strategic management literature suggests that there are

potential limits on the development of knowledge-based capabilities associated

45

with increasing organizational size. Graves and Langowitz (1993) examined a
sample of sixteen U.S. pharmaceutical firms during the period 1969 through
1987 and concluded that there are decreasing returns to scale from large
amounts of investment in research and development. They argued that
“bigness” creates inefficiencies that create limits to innovation.

These limits are likely driven by organizational constraints on individual
learning. If an individual cannot attend to or process new information, that
information has no way of integrating with existing knowledge structures and
thus Ieaming, or knowledge growth, is stunted. Put simply, the benefits of
leaming occur to an organization because individuals are able to access and
use information developed by others as well as their own individual
contributions. Organizations become shared repositories of mutual experiences.
Structures such as policies, training, and myths provide the capacity for
individual learning to become stored and allow others access to it across time
and space. As these structures become filled with information, individuals
become less able to access and use it. The amount of information stored by the
organization continues to grow, but often the individuals that work within the
organization become unable to use all of it, either because of a breadth or depth
of data that is overwhelming to individuals. This limiting function is evident in
that, as explained earlier, knowledge grows exponentially, while the number of
new employees usually grows at a much slower rate. Thus, as the knowledge
base becomes increasingly broad and deep, people are unable to take

advantage of organizational learning because of interdependencies and

46

complexities in both the information itself and the structures within the
organization that contain the information.

Another approach to understanding the limits to growth can be found in
the work of Paul Romer. Romer‘s idea that rival and excludable goods create
limits on the sharing and use of knowledge and ideas. A rival good can only be
used by one person at a time. Excludability means that the owner can prevent
others from using her or his good. The catch here is that ideas that are non-rival
(Le. a piece of software or a procedure to duplicate biological material) and
many that are non-excludable (no enforceable patent rights or no physical
ownership) can lead to exponential growth. However, as goods or ideas become
rival or excludable, limits to growth can occur (Nelson & Romer, 1996; Romer,
1990). The likely existence of limitations to innovative capabilities suggests that
the relationship between research and development resource flows and

knowledge-based capabilities maps out as a cubic, or S-shaped function.

Marx

Three relevant and important theoretical issues have emerged from
integrating the resource-based view of the firm and the organizational learning
literature. First, resource-based theory has evolved to recognize that stocks of
knowledge-based resources are becoming increasingly important drivers of firm
performance in many industries. Second, the organizational learning literature
suggests that these resource stocks may take the form of capabilities that allow

organizations the capacity to store, develop, and utilize knowledge, despite the

47

lack of existence of a “brain” or living cognitive organism. Third, and most
importantly, the building blocks for these organizational-level capabilities include
both individual and organizational knowledge structures, which are dynamic and
interdependent. Furthermore, these structures are related geometrically, and
this idea plays an important role in the dynamic nature of organizational
capabilities.

Thus, the literature review suggests that organizational learning may play
an important, and as of yet unexplored, role as a keystone in improving
specification and measurement of resource based theory. In essence,
organizational learning may be the linking pin in the relationship between
resource flows and resource stocks and thus is very important in determining the
development of capabilities that firms use to develop sustained competitive
advantage. Many questions exist, including what are the timing issues around
these processes? What are the accelerating or decelerating factors in these
processes? How do these resource flows turn into stocks or capabilities and
what is the precise impact of this process on performance? In an effort to
address these questions, the next section builds a theoretical model and

develops the hypotheses.

48

PROCESS MODEL AND HYPOTHESES

The preceding literature review suggests three important points. First,
sustainable interfirm performance differences stem from organizational
capabilities, or m, which are organized collections of associated resources
owned or controlled by the firm (Aharoni, 1993; Dierickx & Cool, 1989; Prahalad
& Hamel, 1990). Next, a developing theoretical stream suggests that interfirm
differences in the development of organizational capabilities are often
knowlegfi-reficfl (Conner & Prahalad, 1996; Grant, 1996; Spender, 1996),
especially in the growing number of industries involved with technology and/or
information processing. Finally, differences in knowledge-related capabilities
can and should be examined by studying the relationship between the flows of
research and development resources into an organization and the resulting
knowledge-based capability stocks (Dierickx & Cool, 1989). Specifically,
capabilities based on the geometric nature of the development of mental models
or cognitive structures may be a fruitful way to examine capability development
in knowledge-intensive industries such as the pharmaceutical industry.

These three points suggest the following research question: given that it
is increasingly important that firms in many industries develop capabilities to
develop knowledge and apply that knowledge in the generation of new products
and services, how do knowledge-based capabilities develop? In an effort to
answer this question, a process model and related hypotheses are developed in

the next sections.

49

The following discussion outlines a process model that, based on an
integration of the cognitive psychology, sociology, organizational theory, and
strategic management literatures, suggests an innovative explanation for the
knowledge accumulation process in organizations. For the sake of clarity, the
model is examined in steps, one relationship at a time. Step one addresses the
core of the model, the relationship between research and development resource
flows and knowledge-related resource stocks. In step two, a moderator variable,
the consistency of resource flows, is introduced. In the third and final step,

critical firm performance outcomes are integrated into the model.

Step One - Flows and Stocks

The core relationship in this process model is that between research and
development resource flows and knowledge-related resource stocks. Resource
flows, including money, equipment, and information, are accumulated by the firm
with the expectation that these resources will be transformed into capabilities,
which are in turn used to convert raw materials into outputs. In the case of the
pharmaceutical industry, one form of resource flows is expenditures into ongoing
research; that is, money needed to run laboratories and pay researchers. The
managers and owners of pharmaceutical firms want these investment flows to
develop into the capability to continuously bring patented compounds into the

marketplace, providing large, long-lasting profit streams. Non-financial resource

50

flows, especially flows of information, are also posited to directly affect resource
stocks or capabilities. This core aspect of the model is displayed in Figure 3,
and drives the first hypothesis.

The first hypothesis examines the relationship between the size of the
research and development resource flows, and the stock of research and
development capabilities. This relationship is hypothesized to be curvilinear, as
depicted in Figure 3. That is, increases in research and development resource
expenditures impact the development of knowledge-related capabilities in at
least two distinct phases. First, increasing returns to resource investments are
expected to provide modest increases in research and development capabilities.
This modest effect is a result of firms slowly building knowledge about a given
domain area. In this learning stage, people in the organization build expertise
about the product or service itself, as well as find colleagues, exchange
information, take advantage of organizational repositories of knowledge, and
build or specialize the infrastructure required to support innovation. This is the
point at which knowledge flows, or data, are absorbed into the system relatively
slowly, and organizational knowledge structures become enriched with data.

The second phase of the relationship is marked by large increases in the
knowledge base derived from any additional research and development flows.
Individual and organizational learning increase rapidly at this stage, as the
increased abilities of firms to absorb and develop innovation from within and
outside of the firm’s boundaries greatly improves the transformation of research

and development expenditures into research and development capabilities. This

51

phase reflects organizational learning and is driven by processes similar to
those discussed in the literature on learning curves. Learning curves are based
on the notion that organizational learning drives a curvilinear decrease in unit
costs relative to the number of units produced (Gruber, 1994; Spence, 1981;
Zimmerman, 1982). Levitt & March describe this same effect and call it a
“competency multiplier” (1988, p. 332).

High rates of resource flow to stock transformation may be further
increased by the development of a direction or as Sahal (1985) labeled it, the
“avenue” of innovation, where greater focus, along with organizational and
individual commitment to the innovation, greatly speed the transformation of
resource flows to resource stocks. Consequently, the relationships among data
are used by members of the organization, resulting in knowledge generation, or
innovation. This is where the non-linearity discussed earlier comes into play.
Specifically, each piece of information has the potential to impact multiple other
pieces of information. In effect, a web of relationships can be created from one
additional piece of information, driving the aforementioned nonlinear relationship
between flows (of information for example) and knowledge-related capabilities.

Finally, a third phase of the relationship between flows and stocks may
include a decrease in the rate of flow to stock transformation. This decrease
may result from the inability of organizations to adequately manage large flows

of resource inputs, which would cause a ceiling effect on the transformative

52

 

 

 

 

 

 

Flows of
Research and
Development
Resources StOCKS Of
Knowledge-
. _ Related
Frnancral Capabilities
Non-
financial

 

 

 

 

 

 

 

 

 

Figure 3. A Process Model of Knowledge Development in Organizations - Step
One

53

process (Garud & Nayyar, 1995). This could be caused by problems
organizations develop in learning during conditions of rapid change or
uncertainty.

For example, a new compound is developed in the R&D labs of a major
pharmaceutical firm. At first, the compound shows promise but little is known
about the compound’s long term effects and its potential interactions with other
compounds. Little knowledge is generated relative to the inputs of money and
efforts. Over time, more is learned about the new compound, including
alternative uses, potentially beneficial and harmful side effects, and long term
impacts that were previously unknown. This information is shared with other
development groups within the firm, who contribute to the speed of development
and find new areas to apply new knowledge generated by the initial innovation.
Here, the knowledge base is increasing rapidly through this flurry of knowledge
development. At some point, the speed of the transformation process may be
limited by constraints such as information overload or information technology
shortcomings. In this example, the curvilinear relationship between flows and
stocks would best describe the long-term pattern of capability development.

Other support for a curvilinear relationship comes from revisiting and
integrating pharmaceutical industry research and economics research.
Schwartzman (1975) reported that, in pharmaceutical research, the expected
rate of return from investment in 1960 ranged from 11.4 to 18.4 percent. While

this is interesting, it falls short of explaining why some firms outperform others.

54

Most firms see positive, double digit rates of return on R&D investment, but there
is significant firm-level variation that is left unexplained by the simple linear
models created by Clarkson (1977) and Schwartzman (1975).

Capon, Farley, and Hoenig (1990) reported results of their meta-analysis
of the literature linking a variety of organizational and industry factors. Across
studies, research and development explained 10.7% of the variance in financial
performance, which was statistically significant at p<.01. Pakes and Griliches
(1984) reported that the form of the relationship between research and
development expenditures and patents was “the logarithm of patents as a
function of time, current and five consecutive lagged values of the logarithm of R
& D expenditures, and (correlated) firm-specific constant terms” (1984: p, 70).
This may be because of a nonlinear relationship, and suggests that by positing
curvilinearity, the estimates presented in the Capon, et al., ( 1990) meta-analysis
may be significantly attenuated.

The first hypothesis examines the form of the relationship between
research and development resource flows and research and development
capabilities. The relationship is posited to be curvilinear, specifically either of a
quadratic (or J-curve) or cubic (or S—curve) form.

Hypothesis 1: The relationship between the size of the research and

development resource flows and research and development capabilities

will be curvilinear in knowledge-intensive firms, of either the form Y = a+x

+x2 orY=a+x+x2+x3

55

Quadratic
(l-CUWG) Cubic
(s-curve)

Resource Stocks (Capabilities)

 

 

 

Resource Flows

Figure 4. Hypothesized Relationships Between Resource Flows and Capability
Stocks

56

Next, an important moderator variable, consistency, is introduced in an
effort to refine the theoretical development of the model, and to improve its
explanatory and predictive capability. Consistency is suggested to be a
moderator that can enhance or constrain knowledge development. Consistency
of resource flows is related to the important role of timing and feedback in

learning.

Step Two - Cpnsistencv

The relationship between research and development resource flows and
research and development capabilities is hypothesized to be moderated by the
consistency with which the organization has directed research and development
resource flows into the firm. Consistency means low deviation over time, or a
lack of “time gaps” or temporal holes in investments. In general, consistency
should moderate the relationship between inputs and outputs, as higher
consistency results in a more effective conversion of inputs to outputs, or flows
to stocks. Specifically, the logic underlying consistency is similar to that
developed by Argyris and Schon (1978), Cyert and March (1963), Hedberg
(1981), Senge (1992), and others who described simple and higher-order
learning.

This idea is displayed in figure 5, which illustrates a primary relationship
of actions leading to outcomes, a process that repeats regularly. The results of
the outcome are used to guide future decisions, as learning takes place (Cyert &

March, 1963). Some firms build structures, and over time, gather and use the

57

feedback to impact action in the next cycle (Hedberg, 1981). These firms
minimize errors based on past performance while firms that do not utilize
feedback make avoidable mistakes or worse, change future actions without
understanding cause-effect relationships from previous iterations, confounding
the ability of employees to make sense of data from previous actions (Hedberg,
1981). Pressures for speed can thus cause arbitrary or ill-informed decisions
(Cyert & March, 1963).

Consistency in the process of building knowledge is important because of
the role of feedback and adjustment. That is, after every action in an
organization, there is an outcome, often after a delay. This outcome generates
feedback that must be returned to the originator of the action, processed and
then integrated into knowledge, where it can be reapplied. This is a basic
learning process and helps explain how firms build expertise (Hedberg, 1981;
Simon, 1991).

Consistency over time is also important in building research and
development capabilities because of the role of practice. Both individuals and
organizations need time to practice in order to refine procedures and capabilities
for implementing knowledge - or in other words, learning by doing (Levitt &
March, 1988). This can also be viewed as learning from failure. Individuals and
their organizations learn through trial-and-error. Innovative effort then results
from practice, or “doing”, but under the condition that feedback can be obtained
(Levitt & March, 1988). Practice allows the construction of a feedback loop, or

cybernetic system, to improve knowledge about phenomena. As discussed

58

above, if the feedback is not incorporated into future iterations, learning
becomes more difficult. As Levitt & March concluded, “learning procedures will
become common when they lead to favorable outcomes and that organizations
will become effective at learning when they use learning routines frequently”
(1988,p.332)

Organizations streamline knowledge structures by creating “avenues” or
“pathways” for knowledge collection from internal and external sources,
structures or storehouses for this knowledge, and pathways for knowledge to be
accessed by organizational members (Sahal, 1985). These pathways can
include information technology, interpersonal networks, or policies
(Hedberg,1981). All three of the structural elements (collection, storage, and
access) can be lost if not maintained or used. Moreover, these processes
require consistent activity to collect, organize, prioritize, and absorb data
effectively. “Forgetting” or decay can occur readily at the organizational level if
key resources leave, databases become outdated, or communication networks
are severed (Hedberg, 1981).

The second hypothesis suggests that research and development
capabilities cannot be built in a short amount of time by making huge, one-time
investments, or by intermittent investment in the development of knowledge.
This assertion was supported by an empirical study by Harrison, Hall, and
Nargundkar (1993), who found a weak but significant positive relationship
between consistency in research and development investments and return on

assets using a sample of 96 firms across a variety of industries. Their results

59

suggest that consistency may indeed be an important factor in determining firm
performance. However, from the logic developed above, consistency is
hypothesized to impact performance not as a main effect, but as a moderating
variable in the relationship between flows and stocks.

Hypothesis 2 posits the moderating influence of the consistency of
resource flow investments on knowledge-related capabilities (see Figures 6 and
7). A lower intra-organizational standard deviation of research and development
flows over time can be interpreted as meaning that firms do not greatly alter the
level of R&D flow investments. As noted in the previous section, consistency is
important for Ieaming or knowledge—development because under situations of
consistent investment, researchers have the opportunity to use feedback to
improve the accuracy of the relationships among data. Swings in investment
inhibit the construction of relationships among data as projects become
abandoned in periods of low funding and flurries of new projects start in periods
of high funding levels. The moderating impact of consistency is relevant to the
relationship among research and development resource flows and research and
development capabilities in that consistency increases the impact of flows on the
development of capabilities and inconsistency decreases the impact of flows on
the development of capabilities.

Hypothesis 2: The interaction between the consistency of R&D resource

flows and the size of R&D resource flows on R&D capabilities in

knowledge-intensive firms will be significant.

60

Step3 - Linkages Between Stocks and Performance

The third and final component of the process model links capabilities, or
stocks, and performance. The strategic management literature suggests that
knowledge-related capabilities drive performance (Conner & Prahalad, 1996).
The notion of core competencies (Prahalad & Hamel, 1990) suggests that areas
of expertise within the firm can provide sustained performance advantages when
applied in various product markets, and further, that these capabilities are
difficult for rivals to imitate. Figure 6 depicts the complete process model.

The outcomes of knowledge-related capabilities generate economic rent
(Peteraf, 1994) that can be used by the firm to reinvest in research and
development resource flows, or can be banked in times of prosperity to ensure
the firm’s ability to make consistent investments in resource flows in leaner
times. The implication here is that once an organization moves down the
learning curve, knowledge builds rapidly and creates an asset that is difficult to
imitate, thus creating economic rent and building a cornerstone of competitive
advantage (Barney, 1986; Peteraf, 1994).

In general, many researchers have noted empirical results that suggest
that research and development capabilities impact firm performance (see for
example, Grabowski & Vernon, 1990; Helfat, 1997; Henderson & Cockburn,
1994). However, these data are generally input related, focusing only on
research and development expenditures, and the researchers merely imply that

research and development capabilities are the unmeasured intervening

61

 

 

 

 

Action

 

 

 

 

 

Outcome

 

 

 

 

 

 

 

Learning
(cognitive
change)

 

 

 

A

Consistent feedback needed here to drive

the system to correct or improve
organizational actions or else firms will repeat

mistakes, or at the very least, stagnate.

Figure 5. The Importance of Consistency in Learning

62

 

 

Flows of
Research and
Development

Resources

 

Financial

 

 

 

 

Non-
financial

 

 

 

 

 

 

 

 

 

Stocks of
Knowledge-
Related
Capabilities

 

 

 

Consistency of
Research and
Development

Resource Flows

 

Financial

 

 

 

 

Non-
financial

 

 

 

 

 

Figure 6. The Moderating Impact of the Consistency of Flows

63

 

High consistency

Low consistency

R&D Resource Capabilities

 

 

R&D Resource Flows

Figure 7. Hypothesized Moderated Relationship

64

variable. Hypothesis 3 broadly tests specific and often unmeasured linkage
between research and development capabilities and performance in knowledge-
intensive firms.

Hypothesis 3: Firm performance is positively related to research and

development capabilities in knowledge-intensive firms.

Step Four — The Full Model

Hypothesis 4 addresses the full model, including research and
development resource flows, the moderating impact of consistency, research
and development capabilities, and firm financial performance. This model is
depicted in Figure 8. The implication of Hypothesis 4 is that the curvilinear
model relating flows and stocks is a better predictor of firm performance than a
linear model. The logic driving this hypothesis is predicated by the fact that
strategic management researchers have previously emphasized the relationship
between research and development and performance in the pharmaceutical
industry (Bogner, Thomas, & McGee, 1996; Cool & Schendel, 1988; Grabowski
& Vernon, 1990; Henderson & Cockburn, 1994). Further, these studies have
primarily examined linear relationships between inputs (research and
development dollars) and outputs (measured by firm performance).

However, if the premise of the resource-based view is correct, the
standard input-output model is not sufficient to fully capture either the interesting
distinctions between firms on the input side of the equation or the sustained

performance differences on the output side of the equation. This would be

65

consistent with an important issue raised by Henderson and Cockburn (1994),
who concluded that despite controlling for quantitative measures of firm size,
scope, research program size, and intra- and inter-firm spillovers in their
econometric model, they “still found surprisingly large and persistent
heterogeneities among firms in their research performance” (p. 69). Henderson
and Cockburn’s response to this issue was to “expand the set of explanatory
variables” (p. 69).

Henderson and Cockburn concluded that “(o)ur results provide
considerable support for the importance of ‘competence’ as a source of
competitive advantage in research productivity. Idiosyncratic firm effects
account for a very substantial fraction of the variance in research productivity
across the firms in our sample” (1994, p. 77). However, Henderson and
Cockburn used linear econometric models. This research extends the literature
by using learning theory as a reason to hypothesize non-linear relationships

between resource flows and resource stocks.

Hypothesis 4: The variance explained in performance by the indirect
effect of R&D flows on R&D capabilities (and through R&D capabilities, on
performance) will be significantly greater when the curvilinear effect of

lows on capabilities is included in the model.

66

 

Flows of
Research and
Development

Resources

 

Financial

 

 

Non-
financial

 

 

 

 

 

 

 

Figure 8. The Complete Process Model

 

Consistency of
Research and
Development

Resource Flows

 

Financial

 

 

Non-
financial

 

 

 

 

 

67

 

 

Stocks of
Knowledge-
Related
Capabilities

 

 

 

Outcomes
(Firm
Performance)

 

 

 

 

Hypotheses 1—3 test each part of the model individually, in an effort to
answer the question how do knowledge-based capabilities develg? The final
hypothesis posits that, as a whole, the curvilinear model linking research and
development expenditures and firm performance offers a) a clearer picture of the
drivers of firm performance in knowledge-intensive industries, and b) addresses
the relevance of a resource-based approach focusing on knowledge-based
capabilities vis-a-vis the traditional linear model. The next section addresses

research methodology.

68

RESEARCH METHODOLOGY

Population, Sample, and Time Frame

This study examined all 22 publicly traded firms that competed in the
pharmaceutical industry during the period between 1978 and 1996 (see Table 5
below) and were either U.S.-based or reported in US. dollars using US.
accounting procedures as a result of their financial structure. The 1978 start
date was selected to balance data availability limitations with sample size
concerns. This particular date, 1978, is a natural breakpoint in the industry,
where changes in pricing, advertising, and substitution regulations created, to
some degree, a level playing field (Bogner, Thomas, & McGee, 1996). The last
year of the time frame, 1996, is the most recent year for which complete data are
available. The fifteen year time period also helps address fluctuations in
financial measures, which are frequently affected by accounting changes, one-
time charges against earnings, and other financial activity which may not be
representative of the underlying construct of performance.

The pharmaceutical industry was selected for several reasons. First, this
population has been extensively studied in the strategic management literature
(Bogner, Thomas, & McGee 1996; Cool & Schendel, 1988; Graves 8. Langowicz,
1993). There appears to be a consensus as to the nature of competition in the
pharmaceutical industry, providing some signposts to guide the present research

in sample, time frame, and variable selection (Bogner, Thomas, & McGee, 1996;

69

Table 5

Companies Included in the Study

 

Abbott Laboratories

 

American Home Products

 

Bausch and Lomb

 

Becton Dickenson

 

Bristol Myers Co.

 

Carter Wallace

 

Glaxo HoldingPLC

 

lCN Pharmaceuticals

 

Johnson & Johnson

 

Lilly (Eli) & Co.

 

Mallinkrodt

 

Merck & Co.

 

Monsanto

 

Pfizer

 

Rhone Poulenc Rorer

 

Schering—Plough

 

Searle, G. D.

 

SmithKline Beckman

 

Squibb Corp.

 

Syntex Corp.

 

Upjohn Co.

 

Warner-Lambert Co.

 

 

 

Cool & Schendel, 1989). Table 6 outlines a range of recent research that has
examined the pharmaceutical industry.

Second, there is evidence that the pharmaceutical industry environment
has been fairly stable and that particular points in time, including the 1978
datewithin this sample, can be clearly identified as break points in industry
continuity (Bogner, Thomas, & McGee, 1996; Cool & Schendel, 1988).
Environmental consistency is an important aspect of selecting the population in
any resource-based study, especially in light of findings presented by Miller and
Shamsie (1996) that suggested environment does play a significant role in the
development of resources. Thus, for the purposes of this study, the examination
of firms under a single set of industry conditions limits one potentially large
source of error variance.

Finally, the characteristics of this industry make several specific
resources especially important, especially given the fact that large components
of the value of pharmaceutical firms consists of information and expertise.
Research and development expenditures are widely recognized as important
drivers of innovation and new product development (Bierly & Chakrabarti, 1996;
Halliday, Drasdo, Lumley, & Walker, 1997). Innovation and new product
development, central to pharmaceutical firm profits, are themselves important
capabilities that are derived from information and expertise held by employees
and organizations (Bierly & Chakrabarti, 1996). Thus, this population is

appropriate for assessing issues surrounding knowledge-based resources, an

71

area identified in the literature review as an important stream driving the
development of resource-based theory.

Merged Firms. Four firms in the sample merged during the time frame
under study. COMPUSTAT combines data for these firms into one for the years
preceding the mergers. Bristol Myers and Squibb, as well as Monsanto and
Searle, are restated into a combined firm for the years prior to their merger. This
process is carried out consistent with generally accepted accounting practices.
By following this practice, only two of the firms were dropped from the sample,
for a total sample size of 20, but no data were lost, as the restated data for the
single firm contains all the data from both firms.

_Ppwi The present sample is broader and covers more years than almost
all other studies that have studied the pharmaceutical industry (e.g. Bogner, et
al, 1993; Gambaradella, 1992; Koenig, 1983; Pisano, 1994). However,
statistical power is an area of concern. The sample size was limited by the
number of firms in the industry and data availability. The second component of
power to examine is estimated effect size. However, traditional estimates of
effect size are questionable in repeated measure studies. Commonly, this is
handled by assuming a small effect size, which creates low power.

However, Dunlap, Cortina, Vaslow, and Burke (1996) argued that in a
within-subjects design, there is 'built-in' control for extraneous error variance, so
moderate effect sizes can be used for power calculations. Given this, power
calculations for a sample size of 20 for a .05 level of significance are in the .30-

.34 range and go up to 39-42 range using an alpha of .10. This research,

72

where sample size is limited by the level of analysis (due to the limited number
of firms) seems suited for such an adjustment. While this is still low, it is an
improvement on other single industry, firm-level research that has appeared in
the strategy literature. This is true both in the sense that the power issue is
even addressed (Mone, Mueller, & Mauland, 1996) and that power is being
improved over existing studies. Power is being improved by using a slightly
larger sample size and by using a higher alpha, recognizing the issue of
practical vs. statistical significance where in this case, the sample size is
constrained by the number of firms that participate in this segment of knowledge-
intensive firms. This limitation is described in light of the fact that in the strategic
management literature, a high degree of statistical power is often difficult to
achieve, since the focal unit of study is the firm, whereas in the behavioral
sciences greater numbers of the unit of study, the individual or team, are

available (Mone, Mueller, & Mauland, 1996).

Dependent Measures

The dependent variable in the first two hypotheses is the stock of
research and development capabilities (a summary table of the methods and
variables for all five hypotheses is presented in Table 7). This variable was
operationalized by using the stock of patent-years possessed by each firm; more
specifically, the total number of years of patent protection possessed by each

firm. Patent data has been used as a proxy variable for research and

73

Table 6

Selected Studies Examining the Pharmaceutical lndustm

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Citation Time- Sample Indep. Vars. Dep. Vars. Findings
frame Size

Bogner, 1969- 25 firms R&D, Market Long term competitive

Thomas, & 1988 market entry, exit posture is impacted by

McGee (1996) position, nature of firm resources

Cool and 1963- Not profitability industry Declines in ROS are

Dierickx 1982 reported (ROS) structure, linked to rivalry, not

(1993) rivalry structure

Gambardella 1973- 14 firms publications patents patents are positively

(1 992) 1988 R&D related to scientific

expendits. publications even after
controlling for R&D
expend.

Graves 8. 1969- 187 obs. # of new Avg. R&D Returns to scale in R&D

Langowitz 1987 16 firms chemical exp (6 8. 3 increase as R&D

(1993) entities yr avgs.), expenditures increase

firm size,
regulation

Henderson 1975- 3120 knowledge Patents productivity drove firm

and Cockburn 1988 abs, 10 flow across effects, research portfolio

(1994) firms firm and variations were large,

national persistent, and impacted
boundaries, productivity

emphasis on

publications

Koenig (1983) 1970- 24 firms publications Important publications and citations

1974 & citations, thera- are positively correlated
R&D peutic with successful
expendits., gains pharmaceutical research
expen
judgment

Nann, Noma, 1975- 17 firms publications patents strong positive

& Peny (1987) 1983 8. citations, relationship between

expert patent data and
judgment, technological strength,
financial using expert opinions and
performance citation freq.

Pisano (1994) Cross- 11 firms Ieaming lead time lab Ieaming drives
ectn'l 23 proj- strategy, org increasing lead times in
analy. ects structure, traditional, chemical

technical based research, but not
content in biotechnology

 

74

 

development capability in pharmaceutical industry research by many, including
Helfat (1994, 1997), Henderson and Cockburn (1994), and Pakes and Griliches
(1 984).

Some have used simple patent counts as an output, but there are
compelling reasons to support the use of the stock of patent years as a proxy for
research and development capabilities. A recent stream of research has
thoroughly examined the relationship between patent data, patent citations,
scientific publication, and performance in the pharmaceutical industry in similar
studies using different data and a variety of sources (Archibugi & Pianta, 1992;
Gambardella, 1992; Narin, Noma, & Perry, 1987). These researchers
consistently found that correlations among patents and a variety of performance
measures were relatively stable across firms and over time. Nevertheless,
patent citations and R&D expenditures are both imperfect indicators of firm
performance (Capon, et al., 1990; Gambardella, 1992). However, patent data
have been shown to correlate strongly with innovation capabilities in the
pharmaceutical industry (Narin, et al, 1987) when the innovation capabilities are
assessed using survey data collected from multiple sources.

The more recent work discussed above can be contrasted with a stream
of economic work from the 19703 and 19805 led by Griliches and his colleagues,
who studied the relationship between research and development and patents.
Pakes and Griliches (1984) concluded that the relationship between research
and development expenditures and patents was indeed curvilinear. Pakes and

Griliches (1984) could not explain the curvilinearity clearly using existing

75

economic theory. These conclusions suggest that there are significant
opportunities to explain the curvilinear relationships using theory drawn from
organizational Ieaming and the resource based view. The resource-based view
suggests that patents may represent a stock of knowledge resources that are
applicable in the marketplace, and as such are most likely an antecedent of firm
performance. The organizational Ieaming literature suggests that there likely
exists intermediate learning processes that impact the relationship between the
flows of research and development resources and patents. Integrating the two
theoretical approaches, it is apparent that patents represent an important
variable that is both impacted by research and development resource flows, and
impacts firm performance.

This view of patents as a linkage between resources flows and
performance can be supported by carefully reviewing the large body of research
addressing patents. Narin, et al., (1987) suggested that patent data is a reliable
and valid source of data on pharmaceutical knowledge and Ieaming. While
patents are related to performance and are considered by some to be an output,
there is evidence to suggest that patents, specifically the sum total of patent-
years accrued to the firm, are more accurately a stock of potential performance.
Viewed in this state, this stock of potential performance can be viewed in two
ways. First, the stock is representative of the capability of the firm to develop,
internalize, and formally obtain intellectual property rights to knowledge

(Kerssens-Van Drongelen, de Weerd-Nederhof, & Fisscher, 1996). Second, the

76

patent stock is a stable of accumulated knowledge that potentially can be
applied in the marketplace with some associated performance outcome.

The key aspect is that patents have limited value if the knowledge
represented by the patent is not applicable in the market. Patent counts do not
and cannot accurately assess the market potential of an innovation. Often times
“can’t miss” blockbuster patents fail to succeed on the market (i.e. interferon) or
are determined to be unsafe after reaching the market (i.e. Seldane), while little-
heralded advances often are later realized to be breakthroughs (i.e. some of the
drugs now being used as part of AIDS fighting “cocktails”). As such, a patent
really has no value until it is applied in the marketplace, either by bringing the
associated product to market, or by licensing or selling the patent itself to
another company. In short, patents merely reflect a non-subjective assessment
of new or unique, and thus patentable, information generated by the firm. It is
neither relevant nor possible to assess the revenue generating potential of the
patent. Patents are useful measures if the variable under study is simply an
assessment of a firm’s ability to generate new knowledge, which is the case in
the present study.

The patent stock data was collected from the US. Patent Office CASSIS
database at the US. Patent Office in Alexandria, VA. Each patent granted to a
firm or an individual associated with that firm was coded and a measure of total
patent life accrued to each firm was computed for each firm for each year in the

sample. This was computed by subtracting the expiration year from the year

77

under study. This process was repeated for each patent held and summed to
create a stock of patent-years for each firm for each year of the study.

The dependent variable in hypotheses 3 and 4 is financial performance,
measured two ways. The first measure, Return on Assets, was used following
Geroski (1990), Mueller (1986), Russo and Fouts (1997), and Helfat (1994,
1997). These data were collected from the COMPUSTAT databases,
augmented by audited company financial reports. The second measure, Tobin's
q was also collected using data from the COMPUSTAT database and calculated

following Chung and Pruitt (1994).

Independent Measures

Research and Development Flows. Two different types of research and
development flows were examined; financial and non-financial. Financial
resource flows were operationalized using research and development
expenditures. This includes money spent on equipment, supplies, facilities,
salaries and expenses of R&D staff. Research and development expenditures
are reasonable proxy of research and development intensity and have been
often used, especially in pharmaceutical industry research (Clarkson, 1977;
Schwartzman, 1975). More recently, in the strategic management literature,
Helfat (1994, 1997), and Bogner, Thomas, and McGee (1996) used R&D
expenditures to measure R&D resources acquired by the firm.

Non-financial resource flows were examined in terms of

interorganizational links. lnterorganizational links create resource flows driven

78

from structural links developed between specific domains of knowledge outside
the firm. lnterorganizational links are defined as the structural links that exist
between an organization with human and informational resources outside the
firm (Goes & Park, 1997). Firms that can successfully integrate multiple types of
data from strategic partners can take advantage of the cross-fertilization of ideas
as information from different sources is exchanged and processed.
lnterorganizational links provide both new information and can serve as a
catalyst for existing data to be integrated in new ways. This improves
understanding of causal and categorical relationships among data by members
of the organization (Goes & Park, 1997).

These non-financial interorganizational linkages were operationalized in
two forms. The first operationalization was corporate-level strategic partnering
with other firms, following Goes and Park (1997), Colombo and Garrone (1996),
Hagedoorn and Schakenraad (1994), and Joly and Mangematin (1996). This
partnering was accomplished through strategic alliances, joint ventures,
mergers, acquisitions, or licensing agreements. Following Goes and Park
(1997), the number of external links made by an organization included the
number of mergers, acquisitions, joint ventures, and strategic alliances made for
each firm, each year in the sample. Goes and Park found that there was
variation in patterns of resource and development flows, and that this variation
was correlated with performance differences. These findings are consistent with
the idea of building complementary knowledge (Teece, 1981) by developing

opportunities for sanctioned exchange of information with other organizations.

79

Table 7

Hypotheses, Methods, and Variables

 

 

 

 

 

 

 

 

 

 

 

H D.V. D.V. I.V. I.V. Measures Test
Measures
1 R&D Patent life Size of R&D 1, R&D expenditures Polynomial
capability in years resource flows 2. Number of intra- regression
51°C“ org. information
links (citations).
3. Number of intra-
ogstmduml links
2 R&D Patent life Consistency of Variance of hyps 1, 2, Polynomial
capability in years R&D resource and 3, above regression
stocks flows
3 Firm Return on R&D capability patent life in years Multiple
performance assets, stocks regression
Tobin's Q
4 Firm Return on 1. Size of R&D 1. R&D expenditures Polynomial
performance assets, resource flows 2, Number of external regression
Tobin's Q 2. Consistency of links
R&D resource 3. Number of cites
"0W5 .. with external orgs
3- R&D capability 4. patent life in yrs
stocks

 

80

 

Table 8

Pharmaceutical Journals Used for Citation Counts

 

Journal of Pharmaceutical Sciences

 

General Pharmacology

 

Pharmacology & Therapeutics

 

Journal of Clinical Pharmacology

 

Pharmacology

 

Progress in Biochemical Pharmacology

 

Journal of Pharmacology and Experimental Therapeutics

 

Molecular Pharmacology

 

Clinical Pharmacology and Therapeutics

 

 

Biochemical Pharmacology

 

 

81

A second operationalization of non-financial resource flows was the
number of citations of published work done jointly between a member of the
focal firm and researchers from another institution (see Table 8 for list of journals
used for citation counts), following logic set out by Koenig (1983) and
Gambaradella (1992). Both studies examined the relationship between patents
and citations in scientific journals. Gambaradella concluded that these external
links, developed through publication in scientific journals, were an important part
of the innovation process because they help drive the rate of knowledge
development within the firm. Especially in the pharmaceutical industry, where
knowledge is complex and rapidly changing, organizations must be able to
absorb and utilize information from the external environment. In fact, Koenig
(1983) found that pharmaceutical company research was as highly cited in
medical journals as research from top medical schools. This suggests that these
citations may be interesting drivers of capability development, as they represent
an active sharing of information among the scientific community. Gambaradella
(1992) found that patents are positively correlated with scientific publications by
firms even after partialling out the size of the firm’s research and development
expenditures. Gambaradella argued that firms that were best at developing in-
house innovations were more effective at exploiting outside research.

These operationalizations are representative of the nature of research
work in the pharmaceutical industry. Drug discovery teams consist of
researchers focusing on similar task areas, but often with different sets of

expertise. Thus, a common body of knowledge or team mental model (Klimoski

82

& Mohammed, 1994) becomes the basis for speeding discovery. This contrasts
to a cross-functional team, where a common set of understandings often does
not exist, creating tension and slowing the decision making process. It is the
common understandings of the specialty research teams that facilitate new
knowledge development as new relationships are understood when new data
are added to the milieu.

This interorganizational linkage variable is important because the
research and development process is complex and involves much more than
pure scientific research in the laboratory. Effective research and development
requires building knowledge from inside and outside the firm, integrating that
knowledge, and using that knowledge to spur discovery. Without an adequate
breadth of knowledge, research and development capabilities will be stunted by
a failure to take advantage of knowledge that lies outside the walls of the
research and development department. Breadth of knowledge is important
because almost always, discovery is built on the knowledge of others, either
within the same domain of knowledge, or by taking information outside the
domain of knowledge and applying it in a different context. The more
interorganizational links at all levels (between individuals, departments,
organizations, and industries), the higher the likelihood this cross-fertilization
process can occur and result in innovation.

For both interorganizational linkage variables, three data collection
guidelines were developed to clarify exactly what constitutes a link. For the

interorganizational linkage variable, a linkage was only recorded if it is clear that

83

1) a merger or acquisition includes considerable integration of at least one
functional unit, 2) a strategic alliance includes a target purpose that involves
coordination of business units from both partners. 3) Strictly marketing linkages
were not counted as there is likely little learning transferred across firms. The
citations were culled from the pharmaceutical journals listed in Table 8 above. A
random subset of the primary data was cross-checked by a specifically trained
George Mason University graduate student for accuracy and inter-rater
reliability. No differences in these ratings were reported between the raters in
the random sample.

Consistency. Consistency was measured using the variance in research
and development intensity for the three years prior to the given year sampled.
This straightfonlvard measure follows Harrison, et al., (1993). Harrison et al.
used the variance in research and development intensity to study consistency of
resource allocation decisions. In a footnote, the authors noted that “(t)here was
no theoretical reason to believe that variance was superior to other potential
methods of dispersion. Consequently, in pretests we substituted standard
deviation and a measure that compared the highest to the lowest intensity and
found comparable results” (1993: p. 1035).

Consistency was operationalized using a three-year time period in an
effort to balance the development of a stable measure of variance with the
benefits of having as many discrete time periods as possible so as to maximize

the ability to detect fine-grained changes in resource development. Interaction

84

terms were created by multiplying each resource flow term by its associated
consistency.

Firm Performance. Firm performance was measured using both financial
and market based data. There is much discussion about the use of financial
based performance measures (Lubatkin & Shrieves, 1986). Of the various
accounting based measures it is apparent that Return on Assets (ROA) is the
measure of choice in the strategic management literature. Many researchers
concluded that ROA is better because it is broadly used by managers as a
measure of performance (Keats & Hitt, 1988; Bettis & Mahajan, 1985). Others
(Ball & Brown, 1968; Gonedes, 1973) argued that ROA is a better measure of
performance than other market related variables because assets are much more
under the control of managers, and ROA has been shown to be closely related
to market value. ROA is considered better than other accounting measures
because it is not as vulnerable to variability as a result of changes in leverage
(Meeks & Meeks, 1981; Ramaswamy, 1997). The widespread use of ROA
creates another incentive to use this measure. Robins and Wiersema strongly
supported the use of ROA in their 1995 SMJ article that applied resource-based
theory to diversified firms. They argued that "(t)he use of ROA as a performance
measure allows the results of the analysis to be directly compared with a
substantial body of work on related topics in strategy, and it helps to make the
research replicable and cumulative" (1995: p . 287).

However, given the general advisability of using multiple performance

measures (Venkatramen & Ramanujam, 1986), firm performance was also

85

operationalized using Tobin‘s q, a market-based measure of firm performance,
following Wemerfelt and Montgomery (1988) and Chung and Pruitt (1994).
Tobin's q is widely used in the research literature (Chung & Pruitt, 1994) and is
held by Industrial Organization Economists to be a strong alternative to financial

based measures.

Data Sources

Archival data from the years 1978-1996 were extracted from a variety of
sources. Research and development expenditures and firm performance data
were gathered primarily from the COMPUSTAT database, and checked against
audited company financial reports and Moody’s Industrial Manual.
lnterorganizational linkage data were collected from the Wall Street Journal, and
Moddv’s Industrial Manual. Citation data were collected from a selection of
major pharmaceutical journals. Patent data were collected from the the US.

Patent Office CASSIS database at the US. Patent Office in Alexandria, Virginia.

86

RESULTS

melvsis Stlgtpgy

The overall intent of this research has been to examine the nature of
relationships among resource flows, capability stocks, and firm performance. To
what extent are capability stocks related to performance? Is the relationship
between flows and stocks curvilinear? Does consistency of resource flow
investments impact the development of knowledge-based resource stocks?
Does a curvilinear model of stocks and flows explain firm performance better
than a linear model? The data analysis strategy outlined in the previous chapter
on research methodology provides a blueprint to address these research
quesfions.

However, the threats of autocorrelation and time dependencies inherent
in analyzing time-series data must be considered. In an effort to address these
threats, two diagnostic analyses were initially conducted on the data. First, the
Durbin-Watson test for autocorrelation was conducted. Then, a repeated
measures MANOVA with a time-varying covariate was conducted to assess time
dependencies within the data. Once these tests detailed the data structure,
pooled analyses were run on the data using repeated measures regression with

an autoregressive error structure to test the four hypotheses.

87

Data Structure and Resdlts of the Diagnostic Analyses

Autocorrelation occurs when error terms in regression equations are
positively correlated over time because key variables are unspecified. This is a
common problem with time series data in strategic management research, and it
is an important problem because when error terms are autocorrelated the
variance of the error term regression estimates may be seriously
underestimated. This may reduce the efficiency of the regression estimators,
and invalidate standard statistical tests (Neter, Kutner, Nachtsheim, &
Wasserman, 1996).

The Durbin-Watson test is commonly used to identify first-order
autocorrelation. The D statistic yielded by the Durbin-Watson was compared to
the lower and upper bounds of 0.77 and 1.25 (calculated using alpha=.01 and
n=17 and p = 2 (number of X variables) using tables from Durbin & Watson,
1951 ). If the D statistic falls above the upper bound, the test indicates no first-
order autocorrelation. If the D statistic falls below the lower bound, the test
indicates first-order autocorrelation. If the test statistic falls between the bounds,
the test is inconclusive (Neter, et al., 1996). Durbin-Watson tests for
autocorrelation were conducted on the data, the results of which are displayed in
Table 9 below. The results of the Durbin-Watson tests suggest that the
research and development and patent years variables appear to be impacted by
autocorrelation. The presence of autocorrelation in this test suggests that
additional diagnostic analyses should be run to assess the degree and form of

the autocorrelation so that the error matrices used in the hypothesis testing

88

phase of the analysis can be structured so as to properly reflect the correlated
terms.

The impact of time on the variance explained in each relationship,
between flows and stocks and stocks and performance, is another issue with
important implications for the data analysis. To assess time dependencies in the
dataset, a repeated measures Multiple Analysis of Variance (MANOVA) with a
time varying covariate was performed to assess the possibility that the
relationships in the data were
spurious, that is due solely to growth trends in either the dependent or
independent variables, rather than because of the hypothesized causal factors.
Each independent variable was analyzed using patent years as the dependent
variable, and then patent years was used as the independent variable with ROA
and Tobin’s q used as the dependent variables in separate analyses. These
tests were limited to variable pairs due to constraints on degrees of freedom that
were inherent in the study design. Thus, no interaction effects or nonlinear
models could be tested with repeated measures MANOVA.

The results of the MANOVA, summarized in Table 10 below, suggest that
when time is used as a covariate, the relationship between the various
independent variables and patent year dependent variable is significant at the
.05 level for 6 of the 11 relationships and in 8 of 11 relationships at the .10 level.
Additionally, the F-ratios were not significant for the relationships between
patent years and both Return on Assets and Tobin’s q. These findings suggest

that the correlations may indeed be spurious, that is driven by trends in the

89

Table 9

Durbin-Watson d Statistics for Repeated Measures MANOVA

 

‘ all values of this were 0, thus D-W statistic was not computable (/0 error)

 

Below lower bound

 

 

 

 

 

 

 

O
(autocorreclgtion present) 20 1 1 19 E 3 “3.
Between upper& lower of g "- 3
bounds (inconclusive) O 0 1 1 {'1 S. E E
Above upper bound 3 I'. §
(autocorrelation not present) 0 19 14 O '3 ‘5’ .8 g.

 

 

 

90

 

 

independent and dependent variables, rather than by causal relationships
between the variables. Thus, an autoregressive error structure is necessary to
analyze the pooled data.

The information about the data structure yielded by these initial diagnostic
analyses drove the final step in addressing the autocorrelation and time
dependence issues, which was to perform a repeated measures regression with
an auto-regressive error structure on the pooled dataset. By using an auto-
regressive error structure, the autocorrelation suggested by the initial Durbin-
Watson tests was more clearly assessed in both degree and form and this
information was used to test the hypotheses controlling for the tendency of the
errors in the data to be correlated over time.

The autoregressive structure is the standard tool used in time-ordered
data and reflects the decreasing correlation in errors over time. The repeated
measures regression with an auto-regressive error structure was conducted in
SAS using the AR( 1) first-order structure using the PROC MIXED procedure on
data that were pooled over time. The descriptive analysis and correlation matrix

are presented in Table 11.

Resdlt_s of the Pooled ADM

All four hypotheses were tested by analyzing the data using repeated measures
regression with an autoregressive error structure. The data were pooled across
all companies for all years. Hypotheses 1, 2 and 4 addressed the linear versus

curvilinear question. Each of these hypotheses was tested using a linear model

91

and two different polynomial models, quadratic (j-curve) and cubic (s-curve).
Hypothesis 3 addressed the relationship between capability stocks and firm
performance. This relationship was hypothesized to be linear; thus only a linear
model was used in this analysis.

Hymtpesis 1: Flows and stocks. Hypothesis 1 suggested that the
relationship between the size of the research and development M and
research and development capability Mg is positive and curvilinear.
Separate linear, quadratic, and cubic regression analyses were conducted using
each of the three independent variables, research and development
expenditures, interorganizational linkages and interorganizational citations,
presented in Table 12.

The analysis of the first hypothesis confirmed that the patent years variable is
severely autocorrelated, as suggested by the large z-test for parameter
estimates in Table 12. The results suggest that none of the independent
variables explained a significant amount of variance in patent years using the
linear, the quadratic, or the cubic model. Therefore, there is no evidence for a
curvilinear relationship between flows and stocks; Hypothesis 1 is not supported.

Hypothesis 2: Consistency of flows. The second hypothesis examined
the impact of the consistency of resource flows on stock or capability
development as measured by total patent years held by the firm. Specifically,
the relationship between flows and stocks was hypothesized to be moderated by

consistency, as measured by the average variance in the flow variable across

92

Table 10

Resiilis of Repeated Measures MLNOVA with Time Va_rying Covariate

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

p
.o 2
.59 n
E 2 3
8 g c
c u 8
o c o r: A
8 a “a 'E ‘-
0 o .. .9 J'-
is Q u. U) E
R&D Exmnditures Pgtent Ye_a_rs 18.66 .000
, Organizational Unks Pg_tent Yea_r_s 23.63 000
Citation Links Pa_tent Years 11.86 003
Variance in R&D Exp. Pam“ Yegrs 3.62 .086—
Variance in Org. Links Paltent Yem 4.78_ .042
Variance in Cit. Links Patent Years §_.5_2 .009
Variance by R&D Interaction Patent Ye_fl 4.44 .061
Variance by Org. Links Interaction Pa_tent Yfls 4.67 .044
Variance by R&D Cit. Links Interaction Pgtent Years 2.79 .112
Eatent years Retumm 9.60 .451
_P_atent years Tobin's q 0.33 .572

 

 

 

93

Table 11

Regated Measures Regression with an Auto-Regressivejrror Structure:
Descriptives and Correlation Matr_ix_

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Variable Abbr Mean Std. Dev ]
R&D Expenditures RDEXP 31923.762 30806.44 336|
Van'ance in R&D Expenditures VRDEXP 43056238 1.17E+08 320|
Eganizational Linkages ORGLNX 3.0882 2.4273 340I
Van'ance in Organizational Linkages VORGLNXI 3.8206 4.7045 340I
lntraorganizational Citation Linkages CITLNX 0.7265 1.2894 340I
Variation in Citation Linkages VCITLNX 0.7784 2.1965 340I
Patent Years (Capability Stock) PATYRS 10536.58 8182.08 340]
Return on Assets ROA 0.107049 0.06709 337|
Tobin's q TOBQ 2.9245 0.9808 311]
RDEXP VRDEXP ORGLNX VORGLX CITLNX VCITLNX PATYRS ROA

RDEXP

VRDEXP .510“
ORGLNX .342” .002
VORGLX .228" .000 .480“

CITLNX .466” .161” .290” .080
VCITLNX .339“ .084 .214” .048 .605“

PATYRS .293” -.052 .366“ .231“ .259“ .117"

ROA 215” -.005 .125“ .056 .071 -.018 .125“

T080 .494” .187” .182” .105 .250” .060 .059 .569“

um; ‘p<.05 ”p<.01

94

 

three prior years. This was tested by regressing stock of patent years on
research development expenditures, the consistency of those expenditures over
time, the product of the two preceding terms, organizational linkages, the
variance in those linkages, and the product of the two preceding terms, and
citation linkages, the variance in those linkages, and the product of the two
preceding terms as the independent variables.

The results of the analysis for Hypothesis 2 suggested that the only
significant interaction term was between R&D expenditures and variance of R&D
expenditures using the quadratic model (see Table 13). However, plotting the
interaction indicated that the direction of the interaction was the opposite of the
hypothesized direction. Specifically, the data suggested that high variance in
R&D expenditures, coupled with high levels of R&D expenditures, was
associated with higher ROA. Given the overall lack of evidence of significant
interactions in the hypothesized direction, these results indicate that Hypothesis
2 was not supported.

fiypoihesis 3: Stocks and performance. The third hypothesis examined
the relationship between capability stocks, measured by accrued patent years,
and firm performance, as measured by 1) a financial measure of performance,
return on assets and 2) a market-based measure of performance, Tobin's q.
This hypothesis was tested by regressing firm performance on patent years. As
shown in Table 14, capability stocks generally did not explain a significant
amount of variance in firm performance as measured by either return on assets

or Tobin's q. Thus, Hypothesis 3 was not supported.

95

Table 12

PoolerLAnalvsis. g-test Par—ameters - vaothesis 1

 

 

 

 

 

 

Patent Years

Linear [Quadraticl Cubic
Covariance Parameter Z 923.10” 930.91 " 904.74”
Independent Variables
YEARID -3.23** -2.66** -2.92**
RDEXP 1 .03 -0.70 0.86
ORGLNX 1 .28 0.45 -0.27
CITLNX -0.17 -0.15 -0.09
RDSQ 1 .23 -1 .13
ORGSQ 0.03 0.58
CITSQ 0.06 0.04
RDCUB 1 .41
ORGCUB -0.59
CITCUB -0.03

 

 

 

 

Note: *p<.05 “p<.01

Hypothesis 4: The full process mocfl. The fourth hypothesis examined
the entire process model as shown in Figure 8. Given that the first three
hypotheses were not supported, there is no possible way hypothesis four can be

supported, so there was no reason to test the fourth hypothesis.

97

Table 13

Pooled Analysis, z-test Parameters - Hypothesis 2

Without Interaction Term

 

 

Degndent Variable = PATYR

Degndent Variable = ROA

 

 

Linear Quadratic Cubic Linear Quadratic Cubic
Covariance Parameter 2 839.44” 856.28“ 833.35” 11.96“ 11.21“ 11.51“
lndegndent Variables
YEARID -2.87** -2.10* -2.29" -2.89*" -2.83** -2.55*
RDEXP 1.18 -0.48 0.96 2.94“ 1.29 2.71“
RDSQ 1.11 -1.26 0.03 -2.38*
RDCUB 1.52 2.42
VRDEXP (Var. in R&D) -0.49 -0.08 0.20 0.24 0.22 1.00
ORGLNX 0.71 0.47 -0.10 1.04 0.14 0.65
ORGSQ -0.24 0.34 0.30 -0.63
ORGCUB -0.41 0.76
VORGLNX (Var. in Org Lnx) 0.53 0.57 0.58 -1.04 1.07 -1.11
CITLNX 1.13 -0.05 0.01 0.40 0.35 0.89
CITSQ 1.20 0.21 -0. 1 1 -0. 89
CITCUB 0.03 0.87
VCITLX (Var. in Cit Lnx) -1.91 -2.30* -2.20 -1.27 1.10 -0.79

 

Note: *p<.05 ”p<.01

With Interaction Term

 

 

 

Dependent Variable = PATYR

 

Dependent Variable = ROA

Linear Quadratic Cubic Linear Quadratic Cubic
Covariance Parameter Z 829. 58" 855.54“ 835.29“ 12.51” 13.02“ 12.99“
lndegndent Variables
YEARID -2.82** -2.06" 223* -2.23* -2.67** 2.47“
RDEXP 1.22 -1.03 0.51 1.87 2.98“ 3.03“
VRDEXP (Var. in R&D) 0.05 1.12 1.34 -3.35** -4.09** -3.42**
RDSQ 1.65 -1.02 -2. 38* -2. 21 *
RDCUB 1.54 1.73
RDXVRD (Interaction) -0.35 -1.26 -1.37 3.91" 4.52 4.10”
ORGLNX 0.72 0.45 0.31 0.41 -0.04 0.64
ORGSQ -0.12 0.55 0.16 -0.70
ORGCUB -0.62 0.81
VORGLNX (Var. in Org Lnx) 0.56 0.44 0.00 -0.88 -0.90 -0.30
OLXVOL ( Interaction) -0.30 -0.09 0.39 0.37 0.26 -0.38
CITLNX 0.71 -0.40 0.00 0.47 0.05 0.71
CITSQ 1.29 0.27 0.37 -0.67
CITCUB 0.30 0.76
VCITLX (Var. in Cit Lnx) 1.77 -1.68 -1.52 -0.88 -0.81 -0.67
CLXVCL (Interaction) 0.51 -0.86 -0.88 -0. 05 -0.27 -0. 18

 

Note: *p<.05 **p<.01

98

 

 

Table 14

Pooled Analysis - Hypothesis 3

 

Dependent Variable
ROA Tobin‘s q
Covariance Parameter Z 14.25” 19.49“

Independent Variable

 

 

YEARID -0.60 2.97”
PATYR 1 .57 0.86
Independent Van'able

YEARID -0.67 3.08"
NUMBER OF PATENTS 1.39 1.52

 

 

Note: *p<.05 **p<.01

99

DISCUSSION

The purpose of this study was to examine the relationships among
resource flows, stocks, and performance in an effort to better determine why
some firms outperform rivals consistently over time. This research project was
designed to address three specific gaps in the existing strategy literature. First,
the study was designed to explore an important hole in the resource-based
literature by empirically testing the underlying notion of resource flow and stock
concepts (Dierickx & Cool, 1989), and extend this framework by positing and
testing for a curvilinear relationship between resource flows and capability
stocks. Second, this study was designed to look at a potentially major source of
unmeasured variance in firm performance: the impact of consistency in the
development of research and development capabilities. Third, and most
broadly, this study was designed to investigate the important issue of knowledge
development within and between organizations using an integrated model drawn
from both resource-based theory and the organizational learning literatures.
Sixteen years of archival data from twenty firms in the pharmaceutical industry
were analyzed using repeated measures regression with an autoregressive error

structure.

S_umma_ry of Findings
Flows. stocks. and curvilinearity. The first hypothesis addressed the

concept of resource flows and capability stocks. In their seminal paper on asset

100

stock accumulation, Dierickx and Cool (1989) described, but did not test, the
importance of asset mass efficiencies in the development of valuable resource
stocks. This study examined the relationship between flows and stocks, but
results did not suggest evidence of a significant curvilinear relationship. This
was a disappointing finding, but some interesting issues were nevertheless
raised by the data analysis. The analysis found a significant correlation between
each of the three flow variables (research and development expenditures,
organizational linkages, and citation linkages) and the stock variable, patent
years (see Table 11). This suggests that relationships do exist among these
variables, but these relationships disappear when an autoregressive error
structure is used. There are at least three possible reasons for this result.

First, there simply may be no empirical relationship between flows and
stocks, after controlling for size and time. If these findings could be replicated
across time, across samples, and with a variety of operationalizations, the logic
presented by Dierickx and Cool (1989) and almost every resource-based study
that uses the flow-stock relationship and would be weakened. Second, and
more likely, is that the operationalizations of flows and stocks are poor. This
seems especially probable for the stock variable, which had severe
autocorrelation problems, as mentioned in the previous section. Third, there may
be major sources of unmeasured variance that impact the ability of the statistical
tests used in this study to detect a relationship, including financial structure,
managerial cognition, diversification strategies, etc. Each of these three

possibilities might yield interesting findings for future work.

101

Consistency. The data do not seem to support the impact of consistency
on capability development. In fact, a strong and significant correlation in the
opposite direction of that hypothesized was found. However, the pooled nature
of the analysis makes the interpretation that increased variance in expenditures
positively impacts firm asset stock development questionable. Since the
analysis is pooled, the results are capturing the huge growth in all
pharmaceutical firms during the time period under study. The successful firms
grew the most, driving the variability, and thus driving the unexpected result.

In fact, there were many instances of significant negative parameter
estimates when performance was regressed on variance, suggesting that as
expected, lower variance (more consistency in investments) may be related to
performance, but the interaction terms in many of these cases were not
significant or significant in the opposite direction. This may be a function of
autocorrelation or the above-mentioned issue of increasing size of research and
development investments driving firm performance and masking the true impact
ofvanance.

Another possible interpretation is that perhaps consistency does not
matter. Perhaps the drivers and limits of organizational Ieaming are driven more
by the total accrued amount of research and development flow investment and
less by the consistency of resource flows. Given the general acceptance for
spaced over massed learning in the individual Ieaming literature, this may either
be a major contribution to the literature, or at the very least, fodder for a more

detailed study comparing individual to organizational Ieaming processes. The

102

focus of such a study might be to determine if scale economies have more
influence on organizational Ieaming than the timing or pace of new knowledge
acquisition.

Linkages between resodpce stocks a_nd firm performan_ce. The data did
not support the existence of a causal relationship between resource stocks, as
measured by accrued patent years and performance, as measured by either
return on assets or Tobin’s q. Thus, the mediated relationship between resource
flows and performance through resource stocks as measured by patent years
possessed by the firm was also not supported. The lack of support for a
relationship between the capability measures in this study and performance
prevents successful support for the Dierickx and Cool (1989) framework.

The absence of a relationship between stocks and performance has three
possible explanations. First, the operationalization of capability stocks, the
accrued patent years measure, is likely a deficient proxy for capability stocks
given the aforementioned problems with autocorrelation. Second, as in
Hypotheses 1 and 2, there may be other important but unmeasured sources of
variance that would mask any existing relationship. Third, and less likely, the
link between capabilities and performance may be weaker than traditionally
viewed in the literature.

A final possible interpretation is that, by taking into consideration
alternate performance measures, the data may yet yield valuable conclusions.
The ability of resource stocks to predict sales revenue and net income may

support the notion that pharmaceutical companies need many years of patent

103

protection to develop high sales, but this particular capability may not result in
the efficient use of assets. Those capabilities may be more operational in terms
of manufacturing and distribution. Further study of firm capabilities should
involve multiple capability measures, including the previously mentioned
operational types of variables.

Full model. The data failed to provide support for the full model as
described in Hypothesis 4, in that the regression of all the independent variables
against the dependent performance variables yielded no significant parameters.
This is not surprising, given the failure of each of the first three hypotheses.

The failure to find strong results in this study is especially disappointing in
light of recent trends in the pharmaceutical industry. The pharmaceutical
industry has undergone rapid change since the end of the dataset timeframe and
is continuing to experience upheaval to date. The emergence of small
biotechnology firms with genetic-based therapies has rocked the stability of the
large pharmaceutical firms. As a result, large pharmaceutical firms have
followed a two-pronged strategy. The first response has been for large
pharmaceutical firms to link up with small biotech firms. The small firms provide
a hedge against genetic-based therapies by providing access to a potentially
lucrative source of revenues, while the large firms provide distribution and
marketing expertise, along with access to low-cost financial capital.

A second response of large pharmaceutical companies has been an
acceleration of large transnational mergers. Given this pattern, any findings in

this study would have made important contributions. Any findings would have

104

built the groundwork for such important questions as how do knowledge stocks
develop between firms using different technologies, or between firms with

different national cultures? At the least, increased transnational merger activity
drives the need to better understand the lack of findings in this dissertation and
lends urgency to the need to develop better understanding of the processes at

work in the pharmaceutical and other knowledge-intensive industries.

Limitations

There are several key limitations that afflicted this research project; some
were anticipated at the outset and some were unanticipated (and very
interesting). The three central anticipated limitations were autocorrelated
predictors, sample size limitations, and weaknesses in both the flow and stock
variables. The three unanticipated limitations were the use of variance as a
measure of consistency, the impact of growth and ambiguity within the industry,
and the use of time as a control variable.

Anticipated limitations. It was expected that, as in most time series
analyses using financial data, autocorrelated predictors would be limit the
interpretability of the data analysis results. An attempt to address this limitation
was made by analyzing the data using repeated measures MANOVA with a time
varying covariate and repeated measures regression with an autoregressive
error structure. This strategy was an attempt to assess the existence and impact

of autocorrelation on the results and control accordingly. Moderate to severe

105

autocorrelation was indeed detected, reducing confidence in the results, despite
using an autoregressive error structure in the data analysis.

Limitations on sample size, and thus degrees of freedom, influenced the
ability to perform the repeated measures MANOVA in that only first order
individual variable pairs could be tested. However, almost the entire population
of US. pharmaceutical firms was used. To increase sample size, one would
have to examine firms in multiple industries, potentially blurring results, an
obvious trade-off.

The study was most clearly limited because of weaknesses in the variable
operationalizations, especially in the stock or capability variable, as measured
by patent years. This was especially evident given the general lack of
correlation between patent years and either financial or non-financial measures
of firm performance. This suggests that the patent year measurement was
simply not an effective proxy for capabilities. More specific capability
operationalizations need to be developed and assessed, probably down to the
level of studying specific product development teams for shared, accumulated,
knowledge-driven skills and knowledge.

Finally, the citation linkage variable did not have adequate variance to be
a useful predictor, given that many of the firms in the sample had no citation
linkages with other organizations in many of the time periods studied. There
were most likely many sources of unmeasured variance that further limited the

ability to make causal inferences.

106

Unanticipated limitations. A major unanticipated limitation revolved
around the use of variance as a measure of consistency in a fast growing
industry such as pharmaceuticals. Also unanticipated were questions about the
appropriateness of firm performance measures. One financial (return on assets)
and one non-financial (Tobin’s q), both commonly used, were used, however,
both adjust for size. Bigger definitely seems to mean better in the
pharmaceutical industry and perhaps by controlling for size, we lose some
important information about performance. When either of two variables
collected but not introduced in the hypotheses, sales revenue and net income,
were used in the model in lieu of ROA or Tobin’s q, strong evidence of
curvilinear and flow-to-stock relationships was found. Reasons for this may
include the existence of relationships between organizational learning and firm
size that are masked when the performance measures control for size. More
analysis using alternative performance measures is warranted to determine the
nature of these relationships.

Also, there were unexpected limitations on the study stemming from the
nature of the pharmaceutical industry itself. In addition to the limitations that we
anticipated, mentioned above, the industry’s fast growth and ambiguity about the
location and nature of asset stock development may have limited the
effectiveness of the data to answer the research questions. Again, additional
studies with different samples are warranted.

A final and intriguing limitation is that time was controlled for in the data

analysis, consistent with much of the research using time series financial-based

107

data. However, after evaluating the results, it is apparent that controlling for
time may not have been appropriate. As mentioned earlier, in the
pharmaceutical industry, flows, stocks, and performance all tend to grow
positively over time. By controlling for time, covariance among the three types of
variables is eliminated, which eliminates any possibility of detecting

relationships, linear or curvilinear.

Futdre Research Directions

The results of this study suggest several follow-up studies, each of which
in some way addresses the key limitations of the present approach. First,
cross-sectional and case study designs using methods such as growth curve
analysis might be used to identify individual patterns of resource accumulation
and capability development within organizations. These may yield insight into
specific resource and capability drivers, and help focus attention on factors that
both support and hinder organizational learning.

A second study might be to use a sample in a different industry, both as a
tool for validation and to help sample size constraints. The information
technology and biotechnology industries would be ideal candidates. Also, this
research is likely related to the emerging construct of knowledge management.
Such a study might be an effective lens through which to base better measures
of knowledge-based capabilities. Future research linking the development of
knowledge management systems to firm performance would be a valuable

addition to the research literature.

108

Refinements of the resource flow and stock variables are needed. As
mentioned earlier, reanalysis should likely use additional performance variables,
not adjusted for size, such as sales revenue and net income. Also, the need for
finer-grained research approaches is clearly evident. Examples include tracking
resource flows on specific products, and as discussed earlier, measuring
specific capabilities at the individual and team level.

Other future research questions driven by this research include, is there a
limit to the synergies possible from combination in the pharmaceutical industry?
Is a partnering strategy effective given the growth in the biotechnology sector?

Can structural linkages between firms improve performance in some situations?

Conclpsions

Results failed to support the existence of a curvilinear relationship
between research and development flows and firm performance. Many possible
explanations for the failure to find results were discussed, relating both to
possibilities that either the relationships posited do not exist or, given the
patterns of results in the data analysis, it is more likely that variables were
misspecified and models were underspecified. These results do however,
provide specific directions for future research and perhaps provide a new lens to
reassess past research. These future directions will be especially relevant given

the emergence of the biotechnology sector and the strategic shifts developed by

109

F“

large traditional pharmaceutical firms in attempts to maintain historically torrid
levels of revenue and profit growth while addressing the rapid changes in the

health sciences.

110

’14

REFERENCES

Aharoni, Y. (1993). In search for the unique: Can firm-specific
advantages be evaluated? Journal of Management Studies. 30 31 -49.

Almeida, P. (1996) Knowledge sourcing by foreign multinationals: patent
citation analysis in the US semiconductor industry. Strategic Management
Journal. 17 (Winter 1996 Sjecial Issue), 155-166.

Amit, R., & Schoemaker, P. J. H. (1993). Strategic assets and
organizational rent. Strategic Management Journal, 14, 33-46.

Anderson, J. R. (1982). Acquisition of cognitive skill. Psychological
Review. 89 369-406.

Anderson, J. R., & F incham, J. M. (1994). Acquisition of procedural skill

from examples. Journal of Experimental: Psychology: Learning, Memog, and
Cognition, 20, 1322-1340.

Archibugi, D., 8 Pianta, M. (1992). Specialization and size of
technological activities in industrial countries: The analysis of patent data.

Research Policy, 21, 79-93.

Argyris, C. (1976). Leadership, learning, and changing the status quo.
Organizational Dynamics. 4. 29-43.

Argyris, C., & Schon, D. (1978). Organizational Learning. Reading, MA:
Addison-Wesley.

Arrow, K. (1962). The economic implications of learning by doing. Review
gf__Economic Studies 29 155-173.

 

Ashcraft, M. H. (1989). Human Memogr and Cognition. HarperCollins.

Ball, R. & Brown, P. (1968). An empirical evaluation of accounting
income numbers. Journal of Accounting Research 6 159-178.

 

Barney, J. B. (1986). Strategic factor markets: Expectation, luck, and
business strategy. Management Science. 32, 1231 -1 241.

Barney, J. B. (1991). Firm resources and sustained competitive
advantage. Journal of Management, 17, 99-120.

111

 

 

Barney, J. B. (1996). The resource-based theory of the firm. Organization
Science 7 469.

 

Barney, J. B., 8 Hansen, M.H. (1994).Trustworthiness as a source of

competitive advantage. Strategic Management Journal, Winter Special Issue,
_1_5_, 1 75-190.

Barr, P. S., Stimpert, J. L., 8 Huff, A. S. (1992). Cognitive change,
strategic action, and organizational renewal. Strategic Management Journal, 12,
15-36.

Bettis, R. A. (1981). Performance differences in related and unrelated
diversified firms. Strategic Management Journal, 2, 379-393.

Bettis, R. A., 8 Mahajan, (1985) Risk-retum performance of diversified
firms. Management Science, 31, 785-800.

Bielry, P. 8 Chakrabarti, A. (1996). Determinants of technology cycle time
in the US. pharmaceutical industry. R&D Management, 26, 115-126.

Black, J. A., 8 Boal, K B. (1994). Strategic resources: Traits,
configurations and paths to sustainable competitive advantage. Strategic

Management Journal, Summer Special Issue, 15, 131 -148.

Bogner, W. C., Thomas, H., 8 McGee, J. (1996). A longitudinal study of
the competitive positions and entry paths of European firms in the US.
pharmaceutical market. Strategic Management Journal, 17, 85-107.

Bpisot, M. H. (1995). Information Space: A framework for learning in
organizations, institutions, and culture. New York: Routledge.

Capon, N., Farley, J. U., 8 Hoenig, S. (1990). Determinants of financial
performance: A meta-analysis. Management Science. 36. 1143-1159.

Castanias, R. P., 8 Helfat, C. E. (1991). Managerial resources and rents.
Journal of Management, 17, 155-171.

Chung, K. H., 8 Pruitt, S. W. (1994). A simple approximation of Tobin's q.
Financial Management, 23, 70-74.

Clarkson, K. W. (1977). Intangible Capital and Rates of Return.
Washington, DC: American Enterprise Institute for Public Policy Research.

112

Cohen, J. 8 Cohen, P. (1983). Applied Multiple Regression/Correlation
Analysis for tlhe Behayioral Sciences. (Second Edition). Hillsdale, NJ: Lawrence
Erlbaum Associates.

Cohen, W. M. 8 Levinthal, D. A. (1989). Innovation and Ieaming: The two
faces of R&D. The Economic Journal 99 569-596.

 

Cohen, W. M. 8 Levinthal, D. A. (1990). Absorptive capacity: A new
perspective on Ieaming and innovation. Adminisgative Science Quarterly, 35,
128-152.

Collis, D. J. (1994). How valuable are organizational capabilities? j
Strategic Management Journal, 15 (Winter Special Issue), 143-152. I

Colombo, MG, 8 Garrone, P. (1996). Technological cooperative
agreements and firm’s R&D intensity. A note on causality relations. Research
Policy, 25, 923-932.

Conner, K. R. (1991). A historical comparison of resource-based theory
and five schools of thought within industrial organization economics: Do we
have a new theory of the firm? Journal of Management, 17, 121 -154.

Conner, K. R. 8 Prahalad, C. K. (1996). A resource-based theory of the
firm: Knowledge versus opportunism. Organization Science. 7. 477-501.

Cool, K. 8 Dierickx, I. (1993). Rivalry, strategic groups, and firm
profitability. Strategic Management Journal, 14, 47-59.

Cool, K. 8 Schendel, D. (1988). Performance differences among strategic
group members. Strategic Management Journal, 9, 207-223.

Cyert, R. M. 8 March, J. G. (1963). A Behavioral Theory of the Firm.
Englewood Cliffs, NJ: Prentice-Hall, Inc.

David, P. A. (1985). Clio and the economics of QWERTY. American
Economic Review Proceedings, 75, 332-337

 

Dess, G. G. 8 Davis, P. S. (1984). Porter’s (1980) generic strategies as
determinants of strategic group membership and organizational performance.

Academy of Management Journal, 27, 467-488.

Dierickx, I. 8 Cool, K. (1989). Asset stock accumulation and sustainability
of competitive advantage. Management Science. 35. 1504-1511.

113

Dunlap, W. P., Cortina, J. M., Vaslow, J. B., 8 Burke, M. J. (1997). Meta-
analysis of experiments with matched groups or repeated measures designs.

Psychological Methods, 1, 170-177.

Durbin, J. 8 Watson, G. S. (1951). Testing for serial correlation in least
squares regression. Biometrica 38 159-178.

 

Eisenhardt, K. M. 8 Schoonhoven, C. B. (1996) Resource-based view of
strategic alliance formation: Strategic and social effects in entrepreneurial firms.
Organization Science- 7. 1 36-150.

Fiegenbaum, E. A. (1970). Information processing and memory. In
Donald A. Norman (ed.), Models of Human Memom: 451 -468. New York:
Academic Press.

Gambaradella, A. (1992). Competitive advantages from in-house scientific
research: The US pharmaceutical industry in the 19803. Research Policy, 21,
391-407.

Garud R. 8 Nayar, P. R. (1995). Transforrnative capacity: Continual

structuring by intertemporal technology transfer. Strategic Management Journal,
fi, 365-385.

Geroski, P. A. (1990). Modeling persistent profitability, in D. C. Mueller

(ed), The Dynamics of Company Profits. Cambridge, UK Cambridge University
Press.

Goes, J. B. 8 Park, S. H. (1997). lnterorganizational links and innovation:
The case of hospital services. Academy of Management Journal, 40, 673-696.

Gonedes, N. J. (1974). Capital market equilibrium and annual accounting
numbers - empirical evidence. Journal of Accounting Research, 12, 26.

Grabowski, H. 8 Vernon, J. (1990). A new look at the returns and risks to
pharmaceutical R8D. Management Science. 36. 804-821.

Grant, R. M. (1996) Toward a knowledge-based theory of the firm.
Strategic Management Journal. 17 (Winter 1996 Special lssfl 109-123.

Graves, S. B. 8 Langowicz, N. S. (1993). Innovative productivity and
returns to scale in the pharmaceutical industry. Strategic Management Journal,
1_4, 593-605.

Gruber, H. (1994). The yield factor and the learning curve in
semiconductor production. Applied Economics, 26, 837-844.

114

Hagedoorn, J., 8 Schakenraad, J. (1994). The effect of strategic

technology alliances on company performance. Strategic Management Journal,
_1_5_, 291 -31 1.

Hall, R. (1992). The strategic analysis of intangible resources. Strategic
Management Journal, 13, 135-144.

Hall, R. (1993). A framework linking intangible resources and capabilities
to sustainable competitive advantage. Strategic Management Journal, 14, 607-
618.

 

Halliday, R. G., Drasdo, A. L., Lumley, C. E., 8 Walker, S. R. (1997). The
allocation of resources for R&D in the world's leading pharmaceutical
companies. R8D Management, 27, 63-77.

 

Hambrick, D. C. (1983). Some tests of the effectiveness and functional

attributes of Miles and Snow’s strategic types. Academy of Management Journal,
E, 5-26.

Hambrick, D. C. 8 Mason, P. A. (1984). Upper echelons: The
organization as a reflection of its top managers. Academy of Management
Review 9: 193-206.

 

 

Harrigan, K. R. (1980). The effect of exit barriers upon strategic flexibility.
Strategic Management Journal, 1, 165-176.

Harrison, J. S., Hall, Jr., E. H., 8 Nargundkar, R. (1993). Resource
allocation as an outcropping of strategic consistency: Performance implications.
Academy of Management Journal, 36, 1026-1051.

Hedberg, B. (1981). How organizations learn and unleam. In P.C.

Nystrom 8 W. H. Starbuck (Eds) Handbook of Organizational Design (pp. 8-27).
London: Oxford University Press.

Helfat, C. E. (1994). Evolutionary trajectory in petroleum firm R&D.
Management Science, 40, 1720-1747.

Helfat, C. E. (1997). Know-how and asset complementarity and dynamic

capability accumulation: The case of R&D. Strategic Management Journal, 18,
339-360.

Henderson, R. 8 Cockburn, l. (1994). Measuring competence? Exploring
firm effects in pharmaceutical research. Strategic Management Jorgnal. Winter
Special lssde. 15. 63-84.

115

lstvan, R. L. (1992). A new productivity paradigm for competitive
advantage. Strategic Management Journal, 13, 525-537.

Johnson-Laird, P. N. (1983). Mental models: Toward a cognitive science
of language, inference, and consciousness. Cambridge, MA: Harvard University
Press

Joly, P. B., 8 Mangematin, V. (1996). Profile of public laboratories,
industrial partnerships and organisation of R&D: the dynamics of industrial
relationships in a large research organization. Research Policy. 25301-922.

Jonassen, D. H., Beissner, K. L., 8 Yacci, M. (1993). Structural
Knowledge: Technigues for RepresentinggConveying, and Acgairing Structural
Knowledge. Hillsdale, NJ: Lawrence Erlbaum.

Keats, B. W., 8 Hitt, M. A. (1988). A causal model of Llinkages among
environmental dimensions. Academy of Management Journal, 31, 570-599.

Kerssens-Van Drongelen, l. C., de Weerd-Nederhof, P. C., 8 Fisscher, O.
A. M. (1996). Describing the issues of knowledge management in R&D:
Towards a communication and analysis tool. R8D Management, 26, 213-230.

Klimoski, R. 8 Mohammed, S. (1994). Team mental model: Construct or
metaphor? Journal of Management, 20, 403-437.

Knez, M. 8 Camerer, C. F. (1994). Creating expectational assets in the
laboratory: Coordination in weakest-link games. Strategic Management Journal,
15 (Winter Special Issue), 101-120.

Koenig, M. E. D. (1983). A bibliometric analysis of pharmaceutical
research. Reseagh Policy. 12,15-36.

Kogut, B. 8 Zander, U. (1996). What firms do? Coordination, identity, and
learning. Organization Science. 7. 502-518.

Lancaster, T. 8 lmbens, G. (1995). Optimal stock/flow panel. Journal of
EconomeLrics. fl. 825-348.

Levitt, B., 8 March, J. G. (1988). Organizational learning. Annual Review
of Sociology, 14, 319-340.

Lippman, S. A. 8 Rumelt, R. P. (1982). Uncertain imitability: An analysis

of interfirm differences in efficiency under competition. Bell Journal of
Economics. 13. 418-438.

116

al.—#—

 

if“

Win—41:3...
A

Lubatkin , M., 8 Shrieves, R. E. (1986). Towards reconciliation of market
performance measures to strategic management research. Academy of
Management Review. 11. 497-513.

Maijoor, S. 8 van Witteloostuijn, A. (1996). An empirical test of the
resource-based theory: Strategic regulation in the Dutch audit industry.

Strategic Management Journal, 17, 549-569.

March, J. G. (1991). Exploration and exploitation in organizational
learning. Organization Science. a, 71-87.

March, J. G., 8 Simon, H. A. (1993). Organizatione (Second Edition).
Cambridge, MA: Blackwell.

Markides, C. C. 8 Williamson, P.J. (1996). Corporate diversification and
organizational structure: A resource-based view. Academy of Management
Journal 39 340-367.

 

McCarthy, M. E. (1979). An entity-relationship view of accounting models.
The Accopnting Review. 54. 667-686.

McCarthy, M. E. (1982). The REA accounting model: A generalized
framework for accounting systems in a shared data environment. T_he

Accounting Review, 57, 554-578.

Meeks, G., 8 Meeks, J. G. (1981). Profitability measures as indicators of
post-merger efficiency. Journal of Industrial Economics 29, 335-344.

 

Mehra, A. (1996). Resource and market based determinants of

performance in the US. banking industry. Strategic Management Journal, 17,
307-322.

Miles, G., 8 Snow, C. C. (1978). Organizational strategy, structure, and
process. New York: McGraw-Hill.

Miller, D. 8 Shamsie, J. (1996). The resource-based view of the firm in
two environments: The Hollywood film studios from 1936 to 1965. Academy of

Management Journal, 39, 519-543.

Mintzberg, H. (1973). Strategy making in three modes, California
Management Review. 16. 44-53.

Mone, M. A., Mueller, G. C. 8 Mauland, W. (1996) The perceptions and

usage of statistical power in applied psychology and management research.
Personnel Psychology. 49. 103-120.

117

Montgomery, C. A. (1982). The measurement of firm diversification: Some
new empirical evidence. Academy of Management Journal, 25, 299-307.

Montgomery, C. A. (1995). Of diamonds and rust: A new look at
resources, in C. A. Montgomery (Ed.), Resource-Based and Evolutionapy
Theories of t_he Firm. Boston: Kluwer Academic Publishers, 251 -268.

Mueller, D. C. (1986). Profits in the Long Run. Cambridge, England:
Cambridge University Press.

Narin, F., Noma, E., 8 Perry, R. (1987). Patents as indicators of corporate
technological strength. Research Policy. 16. 143-155.

Neisser, U. (1976). Cognition and reality: Principles and implications of
anitive psychology; San Francisco: W.H. Freeman.

Nelson, R. R., 8 Romer, P. M. (1996). Science, economic growth, and
public policy. Challenge, 39 (2), 9-22.

Nelson, R. R. 8 Winter, S. G. (1982). An Evoldtionm Theory of
Economic Change. Cambridge, MA: Harvard University Press.

Neter, J., Kutner, M. H., Nachtsheim, C. J., 8 Wasserman, W. (1996).
Applied Linear Regression Models Ehird Edition). Chicago: Irwin.

Newell, A., 8 Simon, H. A. (1972). Human Proplem Solving Englewood
Cliffs, NJ: Prentice-Hall.

Pakes, 8 Griliches, Z. (1984) in Z. Griliches (Ed.) R8D Patents and
Prflctivity. Chicago: University of Chicago Press, .

 

Penrose, E. T. (1959). The t_heory of the growt_h of th,_e firm. New York:
Wiley.

Peteraf, M. A. (1993). The cornerstones of competitive advantage: A
resource-based view. Strategic Management Journal, 14, 179-191.

Pisano, G. P. (1994). Knowledge, integration, and the locus of learning:
An empirical analysis of process development. Strategic Management Journal,
15 (Winter Special ISSlfiL 85-100.

Polyani, M. (1966). The Tacit Dimension. New York: Doubleday.

Porter, M. E. (1980). Competitive Strategy. Free Press: New York.

118

Powell, T. C. 8 Dent-Micallef, A. (1997). Information technology as
competitive advantage: The role of human, business, and technology resources.

Strategic Management Journal, 18, 375-405.

Prahalad, C. K. 8 Bettis R. A. (1986). The dominant logic: A new linkage
between diversity and performance. Strategic Management Journal, 7, 485-502.

Prahalad, C. K. 8 Hamel, G. (1990). The core competence of the
corporation. Harvard Business Review. 68. May-June, 79-91.

Ramaswamy, K. (1997). The performance impact of strategic similarity in
horizontal mergers: evidence from the US. banking industry. Academy of

Management Journal, 40, 697-716.

Reed, R. 8 DeFillippi, R. J. (1990). Causal ambiguity, barriers to imitation,
and sustainable competitive advantage. Academy of Management Review, 15,
88-102.

Robins, J. 8 Wiersema, M. F. (1995). A resource-based approach to the
multi-business firm: Empirical analysis of portfolio interrelationships and
corporate financial performance. Strategic Management Journal, 16, 277-299.

Robinson, P. M. (1993). Continuous-time models in econometrics:
Closed and open systems, stocks and flows. In P.C.B. Phillips (Ed.) Models

Methods, and Applications of EconometfiCS. 71-90, Basil Blackwell: Cambridge,
MA.

 

Romer, P. M. (1990). Are nonconvexities important for understanding
growth? AEA Papers and Proceedings, 80, 97-103.

Rouse, W. B., 8 Morris, N. M. (1986). On looking into the black box:

Prospects and limits in the search for mental models. Psychological Bulletin,
100, 349-363.

Rumelt, R. P. (1974). Strategy, Structure, and Economic Performance.
Cambridge, MA: Harvard University Press.

Russo, M. V., 8 Fouts, P. A. (1997). A resource-based perspective on

corporate environmental performance and profitability. Academy of Management
Journal 40 534-559.

 

Sahal, D. (1985). Technology guide-posts and innovation avenues.
Research Policy. 14. 61 -82.

119

Samuelson, R. A. (1996). The concept of assets in accounting theory.
Accopnting Horjzrons. 10. 147-157.

Schwartzman, D. (1975). Pharmaceutical R&D expenditures and rates of
return. In H. B. Helms (Ed.) Drug Develgpment and Marketing. Washington,
DC: American Enterprise Institute for Public Policy Research.

Senge, P. M. (1990). The Fifth Discipline: The Art and Practice of the
Learning Organization. New York: Doubleday.

Shavelson, R. J. (1972). Some aspects of the correspondence between
content structure and cognitive structure in physics instruction. Journal of
Educational Psychology, 63, 225-234.

Simon, H. A. (1976). A_dministrative Behavior (3rd Edition). New York:
Macmillan.

Simon, H. A. (1991). Bounded rationality and organizational learning.
Organizational Science. a, 125-134.

Spence, M. (1981). The learning curve and competition. Bell Journal of
Economics. 12549-70.

Spender, J. -C. (1989). Industry Recipes: The nature and sources of
managerial judgment. Oxford: Basil Blackwell.

 

Spender, J. -C. (1996). Making knowledge the basis of a dynamic theory
of the firm. Strategic Management Journal. 17 (Winter Special ISSIfi, 45-62.

Stimpert, J. L., 8 Duhaime, l. M. (1997). In the eyes of the beholder:

Conceptualizations of relatedness held by the managers of large diversified
firms. Strategic Management Jogrnal. 18. 111-125.

Teece, D. J. (1981). Toward an economic theory of the multiproduct firm.
Journal ficonomic Behavior and Organization, 3, 3963.

Thompson, J. D. (1967). Organizations in Act_ion. New York: McGraw—Hill.

Ventrakamen, N., 8 Ramanujam, V. (1986). Measurement of business
performance in strategy research: A comparison of approaches. Acadmey of
Management Review. 1 1. 801 -814.

Walsh, J. P. (1995). Managerial and organizational cognition: Notes
from a trip down memory lane. Organizational Science, 6, 280-321.

120

Wernerfelt, B. (1984). A resource-based view of the firm. Strategic
Management Journal 5 171 -1 80.

 

Wernerfelt, B. (1995). The resource-based view of the firm: Ten years
after. Strategic Management Journal, 16, 171-174.

Wernerfelt, B., 8 Montgomery, C. A. (1988). Tobin’s q and the importance
of focus in firm performance. The American Economic Review. 78. 246-250.

Zajac, E. J. 8 Bazerman, M. H. (1991). Blind spots in industry and
competitor analysis: Implications of inter-firm (mis)perceptions for strategic
decisions. Academy of Management Review. 16. 37-56.

Zimmerman, M. B. (1982). Learning effects and the commercialization of
new energy technologies: the case of nuclear power. Bell Journal of Economics
fl, 297-310.

 

121