Post on 21-Mar-2018
Inter-firm R&D networks and firms’ technological knowledge base: a co-evolutionary
perspective
Abstract
This paper investigates how a firm’s technological knowledge base co-evolves with firm’s
position within the network of R&D alliances. In particular, we argue that the more a firm
moves towards the core of the R&D network the more it will develop a generalist
technological knowledge base; furthermore, the more a firm develops a generalist
technological knowledge base the more it will move towards the core of the network. To test
the existence of this self-reinforcing dynamics, we analyze a large panel data set describing
patent and financial variables for all the firms involved in all the R&D alliances registered in
the US between 1984 and 2002. Based on a simultaneous equation model, the data strongly
support our argument and hypotheses.
2
Introduction
For several decades now, the complex relationship linking an organization to its
environment has been a chief topic of investigation in organization theory (Emery and Trist
1965; Andrews 1980; Blau & Schoenherr 1971; Burns & Stalker 1961; Grinyer & Yasai-
Ardekani 1981; Hofer & Schendel 1978; Lawrence & Lorsch 1967; Prescott 1986; Pugh et al
1969; Thompson 1967). Over the years, multiple lines of theoretical and empirical research
have emphasized different aspects of the organization-environment interface. While early
works were primarily concerned with the role of environmental uncertainty (Duncan 1972a,
1972b; Lawrence & Lorsch, 1967), later research expanded this notion by suggesting that the
external environment encompasses a wider range of relevant dimensions (Aldrich 1979, Dess
and Beard 1984). Accordingly, a voluminous body of literature has studied how organizations
are affected by environmental characteristics such as munificence (Castrogiovanni 1991),
dynamism (Li and Simerly 1998), complexity (Child 1972, Pennings 1975, Tung 1979),
consensus (Aldrich 1979), connectedness (Pfeffer and Salancick 1978), density and
legitimation (Hannan and Freeman 1989). Similarly, many have investigated how
organizations purposefully seek to position themselves where environmental conditions are
more conducive to their success (e.g., Child 1972, Porter 1980, Hrebiniak and Joyce 1985).
Whether their analytical focus was on the role of the environment in affecting
organizations or, conversely, on the role of firms in navigating the environment, thus far the
predominant approach among studies of the organization-environment interface has been to
treat the environment as exogenous to the organization. However, "environments affect
organizations through the process of making available or withholding resources” (Aldrich
1979, p. 61), which for the most part are controlled by other organizations (Pfeffer and
Salancick 1978). As a consequence, one chief way in which firms seek to secure access to the
resources residing in their environment is by developing a network of exchange relationships
3
with other organizations. Contrary to macro characteristics of the environment, such as
density or dynamism, some of the characteristics of these networks can hardly be regarded as
exogenous to the organization. In particular, to the extent that an organization pursues its best
interests in choosing who to include in its network, many of these choices will reflect
characteristics of the organization itself.
While the importance of investigating the role of endogenous dynamics between an
organization and its environment has been recently emphasized (McKelvey 1997, Calori et al.
1997, Koza and Lewin 1998), research on the subject is still in its infancy. The goal of the
present paper is to cover part of this gap in the literature by examining how the technological
knowledge base of an organization co-evolves with the network of research and development
alliances it maintains with other organizations. More specifically, we will argue and show
that there exists a mutually reinforcing dynamics between the degree of generality of a firm’s
technological knowledge base and the “coreness” of its position in the economy-wide inter-
organizational R&D alliance network. That is, the more a firm develops technological
knowledge that has applications spanning across technological sectors the more it will move
towards the core of the network; and, the more a firm moves towards the core of the network
the more it will develop technological knowledge that is applicable across multiple
technological sectors.
There are three principal reasons to focus on this particular aspect of the organization-
environment relationship. First, the linkage between a firm’s technological knowledge base
and its R&D network promises to offer a fertile empirical ground for the exploration of
endogenous dynamics in the organization-environment interface. On the one hand, the
influence of inter-organizational R&D networks on the development of firms’ technological
knowledge bases has been widely established (e.g., Powell et al. 1996; Hagedoorn and
Schankenraad 1994; Shan et al. 1994, Walker et al. 1997; Stuart 2000; Ahuja 2000). On the
4
other hand, evidence also suggests that in selecting their R&D alliance partners, firms will
take into serious consideration the potential complementarities existing between their own
and other firms’ technological knowledge bases (Grant and Baden-Fuller 2004, Lavie and
Rosenkopf 2006, Sampson 2007). Second, technological knowledge is a crucial source of
competitive advantage in the knowledge-based economy (Romer 1990). Consistent with this
notion, an interesting stream of research has recently emerged that characterizes
organizations based on their technological knowledge base (e.g., Brusoni et al. 2001, Katila
and Ahuja 2002, Nerkar and Paruchuri 2005). Therefore, by focusing on the technological
knowledge base of organizations, we aim to shed new light on an organization-level construct
that is substantively relevant for contemporary students of organization. Similarly, and third,
understanding under which conditions firms generate general purpose technologies or,
conversely, specialized ones, is essential to explain how productivity shifts and economic
growth occur, an issue with huge implications for the material and welfare conditions of our
societies (Breshnan and Trajtenberg 1995, Helpman 1998). Quite unexplainably, thus far
organizational scholars have failed to contribute to this important debate; hopefully, this
study will provide a useful starting point for students of organization to engage more actively
in the productivity debate.
The paper proceeds as follows. We start by clarifying the notions of network coreness
and technological knowledge base, after which we develop our theory and hypotheses.
Subsequently, we describe the data, its operalization, and the statistical methods. We
conclude the paper by discussing the results and implications of our analysis.
Core-periphery structures in R&D collaboration networks
Inter-organizational R&D collaboration represents a major mechanism by which firms
absorb technological knowledge from their environment (e.g., Powell et al. 1996). Compared
5
to hierarchy or market transactions, R&D alliances often provide a superior means to learn
externally generated technological knowledge since much of that knowledge is tacit and
organizationally embedded (Kogut 1988). In line with this view, Ahuja (2000, p. 448) argued
that alliances “serve as sources of resources and information” and demonstrated a positive
link between the extent of a firm’s alliance activity and firm patenting or innovation.
Similarly, Baum et al. (2000) showed that biotech start-ups were more innovative when they
had many alliances, suggesting that alliances contribute to a firm’s knowledge base. While
most extant research has analyzed R&D alliances at the dyadic level, recently increasing
attention has been paid to the whole network of R&D relationships within which a firm is
embedded, both within and across industries (Powell et al. 2005, Goerzen and Beamish
2005). In the present study, we build on this line of research in order to explore how a firm’s
position within the core-periphery structure of the R&D network affects the firm’s trajectory
of technological development.
Core/periphery structures are characterized by a cohesive clique of densely
interconnected core actors surrounded by a fringe of weakly connected peripheral actors
(Borgatti and Everett 1999: 375). Figure 1 visually illustrates an ideal-typical core/periphery
structure. The cloud of densely connected dark nodes at the center of the network represents
the network core. Peripheral members, conversely, are tied to the core and to each other
mostly through indirect connections. Sociologists have long argued that such core/periphery
macro structures exert a deep influence on social actors’ behavior by shaping both their
access to information and resources and their inducements (e.g., Mintz and Schwartz 1981,
Barsky 1999, Cummings and Cross 2003). However, the possibility that core-periphery
structures may surface and play a role in the R&D alliance system has never been
investigated. To explore this possibility, we analyze in this paper the entire network of R&D
relationships formed in the United States between 1985 and 2002. Hence, our R&D network
6
includes a wide variety of inter-firm alliances spanning across all existing economic and
technological sectors. In particular, our analysis focuses on the notion that the actors that
occupy core positions in a network are proximate not only to each other but to all actors in
the network; by contrast, the actors on the outskirts of the network are no more than
moderately close to the core and, furthermore, they are far away from the vast majority of
other peripheral actors. For that reason, the actors occupying the core of the network “play
the key coordinating roles…, whereas the periphery is occupied by actors with less
integrative importance.” In the context of our study, this notion is reflected in the fact that the
closer is a firm to the core of the R&D network, the more closely it is connected to firms
across technological and economic sectors. Accordingly, ampler opportunities to exploit
complementarities among these sectors are likely to accrue to core firms because, for ideas to
travel between mutually unconnected peripheral firms, these ideas must necessarily pass
through the core of the network.
As we will argue in the next sections, the extent to which a firm is in the core of the R&D
network has important implications for the technological knowledge base the firm will be
able and willing to develop. Furthermore, the technological knowledge base a firm develops
influences whether the firm will be able and willing to move towards to core of the network.
Before we explicate in more detail our theory and hypotheses, we now turn to describing the
construct of technological knowledge base.
Generality of firms’ technological knowledge base
Firms’ success increasingly depends on continuously improving their productivity, either
through internally developed innovations or through the integration of innovations developed
by other organizations. Accordingly, "knowledge-creating companies" and "learning
organizations" are widely celebrated for their ability to generate and integrate both internal
7
and external technological knowledge (Nonaka and Takeuchi 1995; Simonin 1997; Leonard-
Barton 1995). By the same token, firm-level differences in developing and managing
technological knowledge are argued to influence strategic outputs such as market
diversification (Kim and Kogut 1996), technological innovation (Katila and Ahuja 2002;
Ahuja 2000; Nerkar and Parachuri 2005), new product introduction (Nerkar and Roberts
2004), and firm boundaries (Brusoni et al. 2001). Furthermore, prior research has shown that
firm’s technological knowledge is important to organizational adaptation in a technologically
dynamic environment (Ahuja and Katila 2001, Rosenkopf and Nerkar 2001).
Organizations adapt to their environments through innovations, which for the most part
stem from local search and recombination of familiar knowledge (Lewin et al. 1999). Hence
firms generate new technological knowledge in a path dependent fashion, and their current
technological knowledge base demarcates the space of technological search and opportunities
salient to the firm (Stuart and Podolny 1996; Ahuja 2000b). Various aspects of a firm’s
technological knowledge have been analyzed based on the stock of patents the firm produced
over time (e.g., Katila and Ahuja 2002, Rosenkopf and Nerkar 2001), its R&D expenditures
(e.g., Helfat 1994) or its human resources (e.g., Chang 1996). Scholars have characterized the
technological knowledge base of a firm in terms of its internal heterogeneity (Pavitt 1997),
breadth and depth (Prencipe 2000), and complexity (Singh 1997). Furthermore, Yayavaram
and Ahuja (2004) conceptualized a firm’s technological knowledge base as a network of
knowledge elements, where the technological knowledge of a firm is embodied both in the
elements themselves and in the combinative relationships between these elements. Also,
firms’ technological knowledge base has been specified by its distance from competitors
(Stuart and Podolny 1996; Ahuja 2000b) or a technology cluster (Jaffe 1989). In sum, the
significantly increased importance of technological knowledge as an economic and
8
organizational asset for the firm has led many scholars to characterizing firms in terms of
their technological knowledge base.
In this study, we characterize the technological knowledge base of a firm along the
generality-specialism dimension. In particular, we distinguish firms based on the extent to
which the technological knowledge they develop finds applications across few or, conversely,
many technological sectors. Accordingly, in our conceptualization an extreme form of
technological generalist is a firm developing technological knowledge applicable across
virtually all sectors of the economy. Historically, this happened in the case of firms
developing ideal-typical general purpose technologies such as the electric dynamo, the steam
engine, or the computer (Breshnan and Trajtenberg 1995, David 1990, Rosenberg and
Trajtenberg 2004). By contrast, an extreme technological specialist in our conceptualization
is a firm that develops technological knowledge with applications in one technological sector
only. In actuality, of course, most firms fall somewhere in between these two extreme cases.
Our hypothesis is that the generality of a firm’s technological knowledge base co-evolves
with firm’s degree of “coreness” within the broader inter-organizational network of research
and development relationships. Namely, we submit that the closer is a firm to the core of the
R&D network the more it will have the capabilities and incentives to build a generalist
technological knowledge base; furthermore, the more a firm has built a generalist
technological knowledge base the more it will have the capabilities and incentives to move
towards the core of the R&D network.
The co-evolution of technological generality and network coreness
While the use of R&D alliances is evident and increasing (Morris and Hergert 1987;
Mowery 1988), their performance varies widely and failures abound (Bleeke and Ernst 1993;
Kogut 1989). Given the importance of inter-organizational learning and technological
knowledge sharing to outcomes from R&D alliances, characteristics of the technological
9
knowledge bases of alliance partners are a crucial factor in determining the value generated
through an alliance. Extant research, for example, has shown that the degree of similarity
between the technological knowledge bases of alliance partners has a strong influence on
whether and how well firms learn from each other (Mowery et al.1996; Lane and Lubatkin
1998; Ahuja 2000). Similarly, Baum et al. (2000) found that biotech firms that partnered with
organizations having different kinds of relevant knowledge, such as pharmaceutical firms,
universities, and government labs, were more successful after their initial public offerings
than firms engaging in alliances with only single types of partners.
More generally, the added value of an R&D alliance depends to a considerable extent on
the complementarities existing among partners’ technological knowledge bases (Sampson
2007). As said, a generalist technological knowledge base is one featuring complementarities
across a wide spectrum of technological sectors. Therefore, we expect a firm characterized by
a generalist technological base to be perceived as a potentially valuable R&D partner by
firms from multiple and possibly diverse technological sectors. By the same token, a firm
with a general technological knowledge base should find it particularly attractive to establish
R&D partnerships across technological sectors so as to be able to exploit more fully the wide
applicability of its technological knowledge. Because, as we argued, being in the core implies
being directly exposed to a wider array of technological knowledge bases than being in the
periphery, we submit that the more a firm is characterized by a general technological
knowledge base the more it is attractive for, and the more it is attracted to, firms that occupy
a core position in the R&D network.
By contrast, a specialist technological knowledge base is one that generates applications
only in one or very few technological domains. Therefore, a firm characterized by a specialist
technological knowledge base ought to be regarded as a potentially attractive R&D partner by
firms from a more limited set of technological sectors than a technological generalist.
10
Similarly, a firm with a specialist technological knowledge base has no incentive to develop
R&D partnerships outside its field of application because no added value can be anticipated
in return of the costs of those alliances. Therefore, the more a firm has developed a specialist
technological knowledge base the more we expect it to be attractive for, and attracted to,
firms occupying a peripheral position in the network. Takes together, these arguments lead us
to our first hypothesis:
Hypothesis 1: The more generalist is the technological knowledge base of a firm the more
the firm will move towards the core of the network; conversely, the more specialized is the
technological knowledge base a firm the more the firm will move towards the periphery of the
network.
We believe that the causal relationship between the generality of a firm’s
technological knowledge base and its degree of coreness in the network runs in the opposite
direction too. There are two reasons for our conjecture. First, firms strive to shape their
supply of technology so as to meet demand and maximize their return on investment (Scherer
1965). When firms make investment decisions concerning the development of their
technological knowledge base, the structure of incentives they face is likely to vary
depending on their degree of coreness in the network. In particular, we expect firms in the
core to have a strong incentive to pursue general purpose technologies because they are
directly exposed to a wide array of potential complementarities, and hence potential R&D
partners, across multiple technological sectors. Conversely, the choices of technology
investment made by firms in the periphery are driven by the fact that their complementarities
with potential R&D partners, and hence their opportunities for R&D collaborations, are
confined within a limited domain of technological application. Therefore, we expect firms in
11
the periphery to more inclined towards specializing, rather than generalizing, the scope of
applicability of their technological knowledge base in order to more fully rip the
complementarities between them and their potential R&D partners.
Second, as said, firms absorb technological knowledge from their environment and
R&D inter-organizational collaboration networks are a prime source of learning and
innovation (Powell et al. 1996). Because firms in the core are exposed to more numerous and
more distant technological sectors than firms in the periphery, we can expect the former to
absorb from the environment a wider base of technological knowledge than the latter.
Therefore, firms in the core should develop greater capabilities and environmental resources
to build a generalist technological knowledge base than do firms located in the periphery of
the network. By the same token, exploiting the complementarities inherent in a specialist
technological knowledge base requires mastery of the domain and a deep knowledge of the
specificities of the technological and economic sector wherein potential R&D partners
operate. Because firms in the periphery are surrounded by a more focused network of R&D
collaborators, such deep knowledge is more easily generated in peripheral than in core
network positions. Taken together, these arguments lead us to our second and last hypothesis:
Hypothesis 2: The closer is a firm to the core of the network the more it will develop
a generalist technological knowledge base; conversely, the closer is a firm to the periphery of
the network the more it will generate a specialist technological knowledge base.
To test these hypotheses, we use a large data set on US-based R&D alliances. In the next
section, we turn to describing the data.
Data
12
The population for this study was all R&D alliances registered with the US Department of
Justice under the NCRA between 1984 and 2005. Thus our data is similar to that reported in
the CORE database in that both are drawn from filings reported in the Federal Register.
However, the unlike the CORE database, we have obtained and tracked membership in the
various research joint ventures (RJV) at the firm and arrangement level. However, each filing
is made by a distinct organizational entity and does not conform to a common standard for
naming its members. This leads to an inconsistency in naming of organizational members
across Federal Register filings. Therefore, we limit our analysis to publicly traded firms in the
US, for which we have identified a unique standard identifier: the CUSIP number, used in the
Standard and Poors, Compustat database. Limiting the sample to only publicly traded firms
which joined or formed an RJV in the period 1984-2005 gives a final sample of 762 firms for
a total of 4941 firm-year observations, with the average firm being in 6.5 consortia over the
observation period.
Measures
We hypothesized a self-reinforcing dynamic between the extent to which a firm occupies a
core position within the inter-organizational R&D network and the generality of the firm’s
technological knowledge base. This hypothesis therefore entails two dependent variables,
which we measure as follows.
Coreness: The degree of coreness of a firm within the R&D network is measured following
Borgatti and Everett (1999). The authors propose a continuous model in which each node is
assigned a measure of coreness depending on how far they are to the core of the network. In a
Euclidean representation, their model corresponds to distance from the centroid of a single
point cloud. Our network data consist of continuous values representing the strength
13
relationships, as reflected by the number of R&D alliances between pairs of firms at any
point in time. In this situation, Borgatti and Everett (1999) suggest using the following model
to define the coreness of each firm in the network:
Xij=cicj
where X is a vector of nonnegative values indicating the degree of coreness of each node.
Thus, the pattern matrix has (i) large values for pairs of nodes that are both high in coreness,
(ii) middling values for pairs of nodes in which one is high in coreness and the other is not,
and (iii) low values for pairs of nodes that are both peripheral. Thus, the model is consistent
with the interpretation that the strength of a relationship between two firms is a function of
the closeness of each to the center. It may be worthwhile noting that this is the same situation
found in factor analysis, where the correlations among a set of variables are postulated to be a
function of the correlation of each to the latent factor (Nunnally 1978), and in consensus
analysis (Romney et al., 1986), where agreements among pairs of takers of a knowledge test
is seen as a function of the knowledge possessed by each one. Thus, when the continuous
model fits a given dataset, it provides an extremely parsimonious model of all pairwise
interactions (Borgatti and Everett 1999). We therefore use this formulation of the core-
periphery model to estimate coreness empirically on our R&D data.
Technological generality. To measure how generalist is the technological knowledge base of
a firm we looked collected patent and patent citation data for each firm in our population.
Following Trajtenberg, Jaffe and Henderson (1997), and leaving time subscripts aside for the
sake of simplicity, we first measured the generality of each patent granted to any firm as:
14
Pi = 1 - ∑=
n
jijs
1
2
where sij indicates the percentage of citations received by patent i that belongs to patent class
j, out of ni patent classes. Therefore, if patent i is cited by subsequent patents that belong to a
wide range of technological fields, the measure will be high, whereas if most citations are
concentrated in a few fields, the measure of generality will be close to zero. A high generality
score suggest that a patent had a widespread impact influencing subsequent innovations in a
variety of fields. Then, the generality of firm k’s technological knowledge base is calculated
as the average generality across all patents granted to k:
Gk = ∑=
z
i
i
ZP
1
In our analysis, we control for the following variables, which we obtained from Standard &
Poor’s Research Insight database:
Total assets. The sum total of all assets of each firm
Employees. The total number of employees in each firm
Intangibles. This variable represents the intangible assets included in a company’s balance
sheet. Assets covered in this valuation include, but are not limited to copyrights, patents,
licenses, trademarks, trade names, and goodwill.
Advertising costs. We assess the extent to which firms incur advertising costs as one
indicator of complementary assets.
15
Firm performance. As there is no single best indicator of firm performance, we included
separate measures of firm’s:
- Sales
- Return on investment
- Return on equity
- Return on sales.
Econometric specification
As said, our goal is to test the hypothesis of a two way causal relationship between the degree
of coreness of a firm within the R&D network and the generality of the firm’s technological
knowledge base. From an econometrics standpoint, therefore, we need to model and
parametrize an endogenous process between these two variables. Our hypotheses are
corroborated to the extent that (i) there is an endogenous component linking coreness and
generality; furthermore, net of this endogenous component, (ii) coreness positively influences
technological generality and (iii) technological generality positively influences coreness. The
most straightforward econometric approach to test our hypotheses is a simultaneous
equations model of the following kind:
y it = a1 + b1x it + i1r 1it + c1z it + d1ki + f1ht + e1it
x it = a2 + b2y it + i2r 2it + c2z it + d2ki + f2ht + e2it
where e1it and e2it are assumed to be independently and identically distributed disturbances
within each equation, although possibly correlated across equations; ht represents unit-
invariant time-varying factors; ki models firm-level fixed effects (i.e., dummies); y it and x it
16
measure firm’s coreness and firm’s technological generality, respectively; z it is a set of
control variables common to both equations; r 1it is an instrumental variable for x it; and, r 2it is
an instrumental variable for y it.
All variables on the right-hand side are assumed to be exogenous with the exception
of x it in the first equation and of y it in the second. Therefore, this model assumes y it and x it to
be endogenously related. To the extent that the used instruments are valid, this model
provides an efficient and unbiased estimator in the presence of endogeneity (Wooldridge
2006). As a robustness check, we used both 2-stage and 3-stage least square estimation for
our test, obtaining identical results. Further robustness checks were performed by lagging the
independent variables; again, results were perfectly consistent with those reported here.
Results
To test the presence of endogeneity between y it and x it, we performed a Hausman
test (1978) and compared the model estimates with and without the endogeneity assumption
(Wooldridge 2006). According to the test, the null hypothesis that the two variables are
exogenously related must be rejected. Hence, it can be concluded that the degree of coreness
of a firm and the generality of its technological knowledge base are endogenously related.
Table 1 reports a correlation matrix and descriptive statistics of our variables of interest.
Table 2 reports the results of the simultaneous equations model. Equation 1 specifies
Coreness as a dependent variable while, Generality as an endogenous variable, and a dummy
variable representing service firms according to the NAICS classification as an instrument for
Generality. Equation 2, by contrast, specifies Generality as a dependent variable, Coreness as
an endogenous variable, and the average number of claims per patent made by a firm as an
instrument for Coreness. Sales, R&D costs, Advertising costs, and Total assets are treated as
exogenous independent variables in both Equation 1 and Equation 2. Furthermore, in both
17
equations we modeled a non-linear and possibly non-monotonic time trend by introducing the
variables Time and Time squared; firm-level fixed effects are modeled away by means of
firm dummies. All continuous variables are logged in both equations.
------------------------------TABLE 1 AROUND HERE----------------------------
Starting with Equation 1, there turns out to be a humped time trend of Coreness; that
is, the average degree of coreness of the firms in our population declines in the initial phase
of our observation period, but then starts to increase again. Not surprisingly, the more a firm
spends on R&D the more it is close to the core of the R&D network. However, firms that
spend more on advertising and that have larger sales tend to be closer to the periphery. As
predicted by our hypothesis, the more a firm has a technological knowledge base with wide
applications across technological sectors the more the firm is close to the core of the network.
It is important to stress again that the positive effect of Generality on Coreness is net of the
endogenous component inherent in the relationship between these two variables.
Turning to Equation 2, there appears to be no time trend in the extent to which firms
generate generalist or specialist technological knowledge. The larger is the amount of assets
owned by a firm the more the firm tends to develop a generalist technological knowledge
base. By contrast, firms that spend more on advertising and firms that achieve larger sales
tend to develop less general knowledge. Interestingly, R&D investments appear to have no
effect on the generality of a firms’ technological knowledge base. Lastly, and most
importantly, the more a firm is in the core the more it develops technological knowledge with
broad applications across technological sectors; again, we would like to stress that this effect
is net of endogenous dynamics. Hence, the estimates in Equation 2 provides support for our
hypothesis 2.
--------------------TABLE 2 AROUND HERE-----------------------
18
Conclusions
The paper investigated the relationship between the coreness of a firm within the R&D
alliance network and the generality of the firm’s technological knowledge base. It was argued
and showed that the more a firm moves towards the core of the network the more it tends to
build a technological knowledge base with wide applicability across technological and
economic sectors. Furthermore, the more a firm develops a generalist technological
knowledge base the more it tends to move towards the core of the network. Therefore, there
is a self-reinforcing endogenous dynamics between a firm positioning in the core of the R&D
network and the generality of its technological knowledge base.
This finding has notable theoretical implications. The complex relationship linking an
organization to its environment has always been a chief topic of investigation in organization
theory (Emery and Trist, 1965; Andrews 1980; Blau & Schoenherr 1971; Burns & Stalker
1961; Grinyer & Yasai-Ardekani 1981; Hofer & Schendel,1978; Lawrence & Lorsch 1969;
Prescott 1986; Pugh et al. 1969; Thompson 1967). While so far the predominant approach has
been to treat the environment as exogenous to the organization, the importance of
investigating the role of endogenous dynamics between an organization and its environment
has been recently acknowledged (McKelvey 1997, Calori et al. 1997, Koza and Lewin 1998).
By examining how the technological knowledge base of an organization co-evolves with the
network of research and development alliances it maintains with other organizations, the
present paper represents one of the very first attempts to systematically study the role and
relevance of one such dynamics.
The present study offers interesting results also from a substantive standpoint. The
question of how inter-organizational networks evolve over time is an important one, and
numerous scholars have begun to analyze it in recent years (e.g., Koza and Lewin 1998,
Powell et al. 2005). We identified and investigated the role of an organizational variable – the
19
generality of a firm’s technological knowledge base, and we showed that it plays a crucial
role in shaping the evolution of one of the most widely studied forms of inter-organizational
networks – the one formed by R&D alliances among firms. Our study also suggests that the
co-evolution of the R&D network and of firms’ technological knowledge base tends to yield
a well-defined core-periphery network structure, and that firms choose their technology
strategy based o their position within that structure. Indeed, the very existence of such core-
periphery structure appears to be a non-trivial finding in its own right, which we are
investigating in a companion paper.
Lastly, this study makes a substantive contribution to the important debate on the
determinants of general purpose technologies. Understanding under which conditions firms
generate general purpose technologies or, conversely, specialized ones, is essential to explain
how economy-wide productivity shifts occur, an issue with huge implications for the material
and welfare conditions of our societies (Breshnan and Trajtenberg 1995; Helpman 1998).
Quite unexplainably, thus far organizational scholars have failed to contribute to this
important debate. This study showed that the extent to which firms are willing and able to
develop general or specialist technological knowledge depends on their position within the
economy-wide network of R&D alliances. We hope this study will provide a useful starting
point for students of organization to engage more actively in the productivity debate.
Ahuja, G., 2000a, Collaboration Networks, Structural Holes, and Innovation: A Longitudinal
Study. Administrative Science Quarterly 45 425-455.
Ahuja G. 2000b. The duality of collaboration: Inducements and opportunities in the
formation of interfirm linkages. Strategic Management Journal 21: 317-344.
20
Ahuja G, Katila R. 2001. Technological acquisitions and the innovation performance of
acquiring firms: A longitudinal study. Strategic Management Journal 22: 197-220.
Aldrich, H. (1979). Organizations and environments. Englewood Cliffs, NJ: Prentice-Hall.
Andrews, K. (1980). The concept of corporate strategy. Homewood, IL: Richard D. Irwin.
Baum, J. A. C., Calabrese, T., & Silverman, B. S. 2000. Don't Go it Alone: Alliance Network
Composition and Startups' Performance in Canadian Biotechnology. Strategic Management
Journal, 21: 267-294.
Blau, P., & Schoenherr, R. (1971). The structure of organizations. New York: Basic Books.
J. Bleeke and D. Ernst. (1993). Collaborating to Compete: Using Strategic Alliances and
Acquisitions in the Global Marketplace. Wiley, New York (1993).
Borgatti, S. and M. Everett. (1999). Models of core/periphery structures. Social Networks,
21: 375-395.
Bresnahan, T. F. and Trajtenberg, M., 1995. General purpose technologies 'Engines of
growth'?,"Journal of Econometrics, Elsevier, vol. 65(1), pages 83-108, January.
Brusoni, S., A. Prencipe A., K. Pavitt. 2001. Knowledge specialization, organizational
coupling, and the
boundaries of the firm: why do firms know more than they make? Administrative Science
Quarterly.
46 (4) 597-621.
21
Burns, T., & Stalker, G. (1961). The management of innovation. London: Tavistock.
Calori, R., M. Lubatkin, P. Very and J. F. Veiga, 1997, Modelling the Origins of Nationally-
Bound Administrative Heritages: A Historical Institutional Analysis of French and British
Firms, Organization Science, 8, 6, 681-696.
Castrogiovanni, G., 1991. Environmental munificence a theoretical assessment. Academy of
Management Review Vol. 16, pp. 542-565.
Chang, S.J. 1996. An evolutionary perspective on diversification and corporate restructuring:
Entry, exit, and economic performance during 1981-89. Strategic Management Journal 17:
587-611.
Child, J. (1972). Organizational structure, environment and performance: The role of
strategic choice. Sociology, 6, 1-22.
David, P. A. 1990. “The Dynamo and the Computer: An Historical Perspective on the
Modern Productivity Paradox.” American Economic Review 80: 355-361.
Dess, G., & Beard, D. 1984. Dimensions of organizational task environments. Administrative
Science Quarterly, 29, 52-73.
Duncan, R. (1972a). Characteristics of organizational environments and perceived
environmental uncertainty. Administrative Science Quarterly, 17, 313-327.
22
Duncan, R. (1972b). Multiple decision-making structures in adapting to environmental
uncertainty: The impact on organizational effectiveness. Human Relations, 26 ,273-291.
Emery, F.E., and E. Trist 1965 "The causal texture of organizational environments." Human
Relations, 18: 21-31.
Goerzen A. and P.W. Beamish 2005, The effect of alliance network diversity on
multinational enterprise performance, Strategic Management Journal 26, pp. 333–354
Grant, R. and C. Baden-Fuller 2004, A knowledge access theory of strategic alliances,
Journal of Management Studies 41: 61–84
Grinyer, P., Yasai-Ardekanii, M., & Al-Bazzaz, S., 1980. Strategy, structure, the environment
and financial performance in 48 United Kingdom companies. Academy of Management
Journal, 23, 193-220.
Hagedoorn, J. and J. Schankenraad. 1994. The Effect of Strategic Technology Alliances on
Company Performance., Strategic Management Journal, 15: 291-309.
Hannan M. e Freeman J., 1989, Organisational Ecology, Cambridge Mass, Harvard
University Press
Helfat C. 1994. Evolutionary trajectories in petroleum firm R&D. Management Science 40:
1720-1747.
23
Helpman, E. (ed.) General Purpose Technologies and Economic Growth , Cambridge: MIT
Press, 1998
Hofer, C., & Schendel, D. (1978). Strategy formulation: Analytical concepts. St. Paul, MN:
Westview.
Hrebiniak, L., & Joyce, W. (1985). Organizational adaptation: Strategic choice and
environmental determinism. Administrative Science Quarterly, 30, 336-349.
Jaffe, A. 1989, Real Effects of Academic Research, American Economic Review, 79: 957–70.
Lane, P.J., & Lubatkin, M. 1998. Relative absorptive capacity and interorganizational
learning.
Strategic Management Journal, 19: 461-477.
Lavie, D., L. Rosenkopf. 2006. Balancing exploration and exploitation in alliance formation.
Academy of Management Journal 49 797–818.
Lawrence, P., & Lorsch, J. (1967). Organization and environment. Boston, MA: Harvard
Business School.
Leonard-Barton, D. (1995), Wellsprings of Knowledge-Building and Sustaining the Sources
of Innovation, Harvard Business School Press, Boston, MA, .
Li, M. and R. L. Simerly, 1998. The moderating effects of environmental dynamism on the
ownership
24
and performance relationship, Strategic Management Journal, 19(2), pp. 169–179.
Arie Y. Lewin, Chris P. Long, Timothy N. Carroll. The Coevolution of New Organizational
Forms. Organization Science, Vol. 10, No. 5, pp. 535-550
Katila R, Ahuja G. 2002. Something old, something new: A longitudinal study of search
behavior and new product introductions. Academy of Management Journal 45: 1183-1194.
Koza, M. P., A. Y. Lewin. 1998. The coevolution of strategic alliances. Organization Science
9 (3) 255-264.
Kim D-J, Kogut B. 1996. Technological platforms and diversification. Organization Science
7: 283-301.
Kogut, B. 1988. Joint Ventures: Theoretical and Empirical Perspectives, Strategic
Management Journal, 9: 319-332
B. Kogut, 1989. The stability of joint ventures: reciprocity and competitive rivalry. Journal of
Industrial Economics.38: 183–198
McKelvey, B. 1997, Quasi-natural Organization Science, Organization Science, 8, 351–380.
Morris, D., & Hergert, M. 1987. Trends in International Collaborative Agreements. Columbia
Journal of World Business, 22: 15-21.
Mowery, D. C. (ed.) (1988). International Collaborative Ventures in U.S. Manufacturing.
Ballinger, Cambridge, MA.
25
Mowery, D. C., Oxley, J. E., & Silverman, B. S.,1996. Strategic Alliances and Interfirm
Knowledge Transfer. Strategic Management Journal, 17(Winter Special Issue), 77-91.
Nerkar, A. and Paruchuri, S., 2005. Evolution of R&D Capabilities: The Role of Knowledge
Networks Within a Firm. Management Science, Vol. 51, No. 5, May, 771-785.
Nerkar, A., P. Roberts. 2004. Technological and product-market experience and the success
of new product introductions in the pharmaceutical industry. Strategic Management Journal
25(8) 779-799.
Nonaka, I., Takeuchi, H. (1995), The Knowledge Creating Company, Oxford University
Press, New York, NY, .
Nunnally, J.C., (1978). Psychometric Theory. McGraw-Hill, New York.
Pavitt, K., Academic research, technical change and government policy. In: J. Krige and D.
Pestre, Editors, Science in the Twentieth Century, Harwood Academic, Amsterdam (1997),
pp. 143–158.
Pennings, J., 1975. The relevance of the structural-contingency model for organizational
effectiveness. Administrative Science Quarterly, 20, 393-410.
Pfeffer, J., & Salanick, G. (1978). The external control of organizations: A resource
dependence perspective. New York: Harper & Row.
26
Porter, M. (1980). Competitive strategy: Techniques for analyzing industries and
competitors. New York: Free.
Powell W.W. and K. W. Koput and L. Smith-Doerr, 1996. "Interorganizational Collaboration
and the Locus of Innovation: Networks of Learning in Biotechnology," Administrative
Science Quarterly, 41, 116- 145.
Powell, W., White, D., Koput, K., Owen-Smith, J., 2005. Network dynamics and field
evolution: the growth of inter-organizational collaboration in the life sciences. American
Journal of Sociology, 110 (4).
Prencipe A., 2000, Breadth and depth of technological capabilities in CoPS: the case of the
aircraft engine control system Research Policy, Volume 29, Number 7, pp. 895-911(17)
Prescott, J. 1986. Environments as moderators of the relationship between strategy and
performance. Academy of Management Journal, 29, 329-346.
Pugh, D., Hickson, D., Hinings, C., & Turner, C. (1969). The context of organizational
structures. Administrative Science Quarterly, 14, 91-114.
Romer, P.M., 1990, Endogenous technical change, Journal of Political Economy, 98,
October.
27
Rosenberg, N. and M. Trajtenberg, 2004, A general purpose technology at work: the Corliss
steam engine in the late-nineteenth-century United States, Journal of Economic History 64
(1), pp. 61–99.
Rosenkopf L, Nerkar A. 2001. Beyond local search: Boundary-spanning, exploration, and
impact in the optical disc industry. Strategic Management Journal 22: 287-306.
Sampson, R.C. (2007), "R&D alliances and firm performance: the impact of technological
diversity and alliance organization on innovation", Academy of Management Journal, Vol. 50
pp.364-86.
Shan, W., G. Walker and B. Kogut. 1994. Interfirm Cooperation and Startup Innovation in
the Biotechnology Industry. Strategic Management Journal. 15 387-394.
Scherer, F.M., 1965, Firm Size, Market Structure, Opportunity, and the Output of Patented
Inventions The American Economic Review, Vol. 55, No. 5, Part 1, pp. 1097-1125
Singh, K. 1997, The Impact of Technological Complexity and Interfirm Cooperation on
Business Survival The Academy of Management Journal, Vol. 40, No. 2, Special Research
Forum on Alliances and Networks, pp. 339-367
Simonin, B., 1997, The importance of collaborative know-how: an empirical test of the
learning organization, Academy of Management Journal, Vol. 40 No.5, pp.509-33.
Stuart, T. 2000. Interorganizational Alliances and the Performance of Firms:A study of
Growth and Innovation Rates in a High-Technology Industry. Strategic Management Journal
21 791-811.
28
Stuart, T., J. Podolny. 1996. Local search and the evolution of technological capabilities.
Strategic Management Journal 17(Special Issue: evolutionary perspectives on strategy
(Summer) 21-38.
Thompson, J. (1967). Organizations in action. New York: McGraw-Hill.
Trajtenberg, M., Bresnahan T. 1995, General Purpose Technologies: Engines of Growth?
Journal of Econometrics, 65(1), pp. 83-108.
Trajtenberg, M., Henderson H. and Jaffe, A., 1997, University versus Corporate Patents: A
Window on the Basicness of Invention. Economics of Innovation and New Technology, 5 (1),
pp. 19-50.
Tung, R., 1979. Dimensions of organizational environments: An exploratory study of their
impact on organization structure. Academy of Management Journal, 22, 672-693.
Walker G. B. Kogut, and W.J. Shan, 1997. Social Capital, Structural Holes, and the
Formation of an Industry Network. Organization Science 8 109-125.
Yayavaram, S. and Ahuja, G. 2004. Structure of a firm’s knowledge base and the
effectiveness of technological search. Paper presented at the Academy of Management 2004
Conference, August 2004, New Orleans.
29
FIGURES
Figure 1. A core-periphery structure. (Source: Cattani and Ferriani 2008)
30
TABLES
Table 1. Correlation matrix Mean Std. Dev. Coreness 0.028 0.043 1.0000 Generality 0.35 0.269 0.1130 1.0000 R&D costs 459.64 899.92 0.3977 -0.1161 1.0000 Sales 9660.33 18389.29 0.2905 -0.0962 0.8125 1.0000 Adv Costs 281.93 496.17 0.0846 -0.1755 0.6406 0.7485 1.0000 Total Assets
13127.23 35452.59 0.1732 -0.0839 0.6038 0.7685 0.4767 1.0000
Time 9.60 4.35 -0.1404
-0.6136 0.1020 0.0475 0.1113 0.0774 1.0000
Table 2. 3-Stage Least Squares Regression
Equation 1 Dependent Variable: Coreness Observations: 4941 Endogenous Variable: Generality R-Sq: 0.168 Chi-Sq: 1297.6 Variable Coefficient Z value P value Constant -4.80*** -20.26 0.000 Generality 0.52*** 6.78 0.000 Sales -0.13*** 0.03 0.000 R&D costs 0.61*** 19.97 0.000 Adv. Costs -0.18*** -8.02 0.000 Total assets -0.03 -1.04 0.298 IV (dummy) -1.82*** -16.69 0.000 Time -0.17*** -3.23 0.001 Time-squared 0.01*** 3.64 0.000 Generality 4941 0.126 794.6
Equation 2
Dependent Variable: Generality Observations: 4941 Endogenous Variable: Coreness R-Sq: 0.126 Chi-Sq: 794.6 Constant 0.15 0.64 0.522 Coreness 0.17*** 4.65 0.000 Sales -0.06** -2.51 0.012 R&D costs -0.01 -0.30 0.761 Adv. Costs -0.12*** -7.31 0.000 Total assets 0.07*** 3.71 0.000 IV (dummy) 0.11*** 16.85 0.000 Time -3e-3 -0.10 0.922 Time-squared -9e-4 -0.42 0.675