Seminal Paper

Research Policy 41 (2012) 16071619
Contents lists available at SciVerse ScienceDirect
Research Policy
journal homepage: www.elsevier.com/locate/respol
Intangible resources and technology adoption in manufacturing rms

Jaime Gmez 1 , Pilar Vargas
Universidad de La Rioja, Departamento de Economa y Empresa, Edicio Quintiliano, Cige
na 60, 26004 Logro
no (La Rioja), Spain
a r t i c l e
i n f o
Article history:
Received 9 November 2009
Received in revised form 5 March 2012
Accepted 28 April 2012
Available online 21 May 2012
Keywords:
Intangibles
Technology adoption
Diffusion
Complementarities
a b s t r a c t
Our objective in this paper is to analyse the determinants of the use of advanced manufacturing technologies in manufacturing rms. We go beyond more traditional approaches and consider the role of
complementarities in technology adoption at two levels. First, we adapt Teeces (1986) framework to
study the incentives to use new technology that stem from investments in R&D, human capital and advertising. Second, we analyse whether technology use is conditioned by a system effect that arises from the
use of related technologies. We test our hypotheses on a representative sample of manufacturing rms
in Spain. Our results fully support the idea that R&D investments increase the likelihood of technology
use, but only offer partial support for human capital and advertising investments. Export intensity, being
part of a business group and epidemic effects are also important determinants of adoption.
2012 Elsevier B.V. All rights reserved.
1. Introduction
Our objective in this paper is to understand the role of intangible resources in technology adoption. The idea that intangible
resources contribute to the explanation of technology diffusion
among rms has its roots in some analyses of the impact of information technology (IT) on performance. A part of this literature has
argued that IT are strategic necessities in the sense that they have to
be adopted in order not to suffer from competitive disadvantages,
but that they cannot produce sustainable competitive advantages
(Clemons and Kimbrough, 1986; Clemons and Row, 1991). Recent
research on complementarities has shown that the performance
of IT frequently depends on their interaction with organizational
elements (Powell and Den-Micallef, 1997). In particular, the application of the resource-based view of the rm suggests that, in order
to achieve inimitability, rms have to combine IT with resources
protected by isolating mechanisms such as tacitness, causal ambiguity and time compression diseconomies2 (Barney, 1991; Dierickx
Corresponding author. Tel.: +34 941 299 572; fax: +34 941 299 393.
E-mail addresses: jaime.gomez@unirioja.es (J. Gmez), pilar.vargas@unirioja.es
(P. Vargas).
1
Tel.: +34 941 299 373; fax: +34 941 299 393.
2
Time compression diseconomies make reference to the idea that the returns
diminish when one input, in this case time, remains constant (Dierickx and Cool,
1989). Intangible assets, such as the ones created through investments in R&D or
advertising are likely to be subject to them. Dierickx and Cool (1989, p. 1507) provide
an example for R&D investments: 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 over half the time interval.
0048-7333/$ see front matter 2012 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.respol.2012.04.016
and Cool, 1989). In other words, research on the impact of IT on

performance suggests that rms possessing resources protected by
isolating mechanisms should have higher incentives to adopt IT.
Despite this, research on technology diffusion tends to focus on
the role of some classical variables that characterise rms, such
as size and investments in absorptive capacity (see, for example,
Karshenas and Stoneman, 1993). In this paper, we shift the focus
of our attention to the idea that the complementarities between
resources and new technologies should increase the incentives
of rms to adopt. Taking into account the arguments provided
by the resource-based view (Barney, 1986; Wernerfelt, 1984), we
examine the role of intangible resources through the framework
provided by Teece (1986, 2006). Given the weak appropriability of
new technologies, we focus on difcult-to-obtain assets to explain
the different stimuli that rms may have to use new technologies. Accordingly, we develop hypotheses that relate investments
in research and development, in personnel qualications, and in
advertising with the likelihood of a rm using a technology. These
resources not only interact with technologies in order to create rm
value, but they can also help the rm to appropriate the returns of
its investment in them.
We test our hypotheses on a representative sample of Spanish
manufacturing rms. Our main focus is on the use of four process
technologies (numerically controlled machines, robotics, computer
aided design and exible manufacturing) that have previously been
considered as being part of what is termed Advanced Manufacturing Technologies (AMT). Using information on several technologies
is in line with the literature, particularly the theory of Milgrom and
Roberts (1990) that complementary technologies form a production system, which implies that adoption of the technologies should
be positively correlated. This allows us to explore the hypothesis
1608
J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619
that the use of any technology could also be explained by the

use of other technologies with which it forms a system (Colombo
and Mosconi, 1995; Sinha and Noble, 2008) and reinforces the
idea that complementarities have a key role to play in the understanding of performance differences and adoption behaviour. To
further check this system effect, we use a second sample in which
we test whether the four process technologies are related to certain e-business technologies. More precisely, we use information
on whether rms have adopted supplier-to-business, business-tobusiness and business-to-consumer technologies to estimate the
complementarities arising within and among different groups of
technologies. The use of different technologies will also help us to
understand the degree to which our predictions can be generalized.
From a methodological point of view, a problem introduced
by the consideration of the role of complementarities is that the
decisions to adopt several technologies could be related. In such a
setting, the independent estimation of the models explaining the
use of each technology would be inadequate, given that the decisions to adopt are potentially correlated. A solution to this problem
is the estimation of a multivariate model, which takes the potential correlation into account. Testing for differences in correlations
among different types of technologies makes it possible to evaluate
the degree to which the system effect is present in our data.
The rest of the paper is organized as follows. In the next section, we integrate Teeces (1986) framework, the resource-based
view and the literature on technology diffusion in order to develop
four hypotheses on the role of complementary assets. A section
that explains the methodology, describes the sample and denes
the variables follows. After this, we present the results and perform some robustness tests in order to conrm them. The nal
section concludes and discusses the implications for technology
management.
2. Intangible resources and technology adoption
Our objective in this paper is to use the idea of complementarity to explain technology adoption by manufacturing rms. The
concept of complementarity refers to the idea that if the investments in one activity are increased, the returns on investment
of others would also increase (Milgrom and Roberts, 1995). The
analysis of complementarities between organizational elements is
receiving increasing attention in the economics and management
literatures (Ennen and Richter, 2010). Different papers have studied
the interactions arising between rm resources, practices, policies,
strategies and environmental factors.3
The idea of complementarities may help us to understand the
factors that underlie a rms technology adoption. In fact, it is
surprising that, given the close relationship between innovation
and technology adoption (Battisti et al., 2009; Koellinger, 2008;
Santamara et al., 2009), research on the two has progressed with
relative independence. Research has shown the importance of new
technologies on product and process innovation. For example,
some AMT have been related to quality and exibility strategies.
These strategies are characterised by high innovation frequency
and product R&D, among other features (Parthasarthy and Sethi,
1992, 1993). Similarly, Koellinger (2008) shows that process and
product innovations are enabled by the use of IT.
In this paper, we use Teeces (1986) framework to integrate the
two lines of research and to explain the different incentives rms
3
For example, Cassiman and Veugelers (1999) study the complementarities
between external and internal R&D investments, whereas Belderbos et al. (2006)
assess the complementarities arising in R&D cooperation strategies. A review of the
literature on complementarities and on the types of factors considered may be found
in Ennen and Richter (2010).
may have to use technologies. Following Teece (1986), the capacity of a rm to appropriate the returns from innovation mainly
depends on two types of factors (Teece, 1986): the appropriability
regime and the possession of complementary assets. The appropriability regime is dened by (1) legal and (2) technological factors.
Additionally, the rents accruing to the innovator also depend on
the possession of complementary assets. By complementary assets,
Teece (1986) understands all the resources that need to be jointly
used with the innovation in order to exploit it. They include, for
example, the manufacturing, distribution and service resources
that are needed to operate the different stages of the value chain.
Teeces framework could be easily adapted to understand the
incentives of rms to adopt new technologies. The fact that new
technologies are freely available in the market (there are no legal
barriers protecting the adopter) focuses attention on complementarity assets.4 Despite the weak appropriability regime of new
technologies, the returns from their use could be captured if
the rm possesses complementary resources that are difcult to
obtain, such as the ones created by investments in research and
development and advertising. In fact, research has shown that
the link between IT and protability could be explained through
the consideration of complementary resources under the control
of the rm (Powell and Den-Micallef, 1997). However, their use
as determinants of adoption has been scarce. The existence of
difcult-to-imitate complementary resources could be an indicator of the stimuli that a rm has to adopt a new technology, given
that they could provide the basis for competitive advantage.
Selecting the types of resources more likely to be combined with
new technologies to provide competitive advantages is a difcult
task. Although both tangible and intangible assets play an important role in creating value, intangible assets frequently present time
compression diseconomies and they are typically tacit and hard
to codify (Conner and Prahalad, 1996; Kogut and Zander, 1992).
Additionally, they are likely to be traded in imperfect factor markets (Barney, 1986) and are not consumed by their use (Collins and
Montgomery, 1998). Teece (2000) suggests that a rms superior
performance depends on its ability to defend and use the intangible
assets it creates. Accordingly, we argue that technology adoption is
conditioned by the possession of certain complementary resources
that are difcult to acquire or copy, namely, technological, human
and marketing resources.5 Similarly, we also analyse the presence
of complementary relationships between technologies. As mentioned in the Introduction, we focus on production technologies
and, in particular, the group of technologies that has been termed
advanced manufacturing technologies.6
4
In fact, this has led some researchers to conclude that, in the long run, technology adoption might only provide competitive parity (Barney, 1991). Some authors
have enunciated this as the strategic necessity hypothesis (Clemons and Kimbrough,
1986; Clemons and Row, 1991). Following this hypothesis, a focal rm would have
incentives to adopt a valuable new technology. However, this would create incentives in rival rms to invest in the same technology in order not to obtain below
normal prots. As long as new technologies are freely available in the market, the
long-run result would be that rival rms would erode the rents accruing to rst
adopters, leading to competitive parity.
5
We are not claiming that difcult-to-imitate complementary resources are the
only ones likely to interact with new technologies to provide competitive advantage.
In fact, Rivkin (2000) points out that complex systems could be inimitable, even if
their elements are not.
6
As suggested by an anonymous referee, we expect differences between the
effects of each of the variables depending on the type of technology considered.
For example, Rogers and Shoemaker (1971) argue that the diffusion of innovations
depends on their attributes, namely, relative advantage, compatibility, complexity,
trialability and observability. For example, we would expect the effect of technological resources to be more important in those cases in which the technology is more
complex. Therefore, the magnitude of the associated effect should be higher for
exible manufacturing than for the other technologies. Similarly, we would expect
a higher effect of the qualications of human resources in those cases in which the
2.1. Technological resources and technology adoption

Investments in research and development contribute to the
knowledge base of the rm, increasing its capabilities, and may
also result in process and product innovations. Like other intangible
resources, they tend to be tacit, idiosyncratic and deeply embedded
in the rm (Winter, 1987), which makes it difcult for competitors to copy them. Imitation may be also complicated by the fact
that this knowledge may be path dependent and subject to time
compression diseconomies (Dierickx and Cool, 1989).
There are different ways in which the knowledge base of the rm
interacts with advanced technologies increasing the possibility to
create complex combinations of resources (Denrell et al., 2003;
Rivkin, 2000). On the one hand, the knowledge base of the rm
has frequently been related to technology adoption through the
absorptive capacity concept. A rms absorptive capacity is dened
as the ability of a rm to recognize the value of new, external
information, assimilate it and apply it to commercial ends (Cohen
and Levinthal, 1990, p. 128). Although absorptive capacity tends
to be related to adoption through the greater ability of rms to
understand valuable information, we should emphasize its role in
the exploitation of new technologies. In other words, once a new
technology is adopted, its use would require relating the knowledge embodied in the innovation to the internal processes taking
place inside the organization. Therefore, a better knowledge base
could help rms to exploit new technologies more effectively and
to capture the results. For example, this knowledge could be used
to better understand and to rene a rms productive process.
On the other hand, new technologies may also help the rm
to take advantage of the results of the innovation process. In fact,
research has shown that new technologies frequently enable process and product innovations (Koellinger, 2008). In this sense,
several authors have considered that the use and appropriation of
AMT also accelerate the resolution of more radical problems, opening the pace for a greater degree of novelty in the innovation (Amara
et al., 2008; Wuyts et al., 2004). Therefore, new technologies may
provide an adequate channel for converting the knowledge base of
the rm into protable products and services. Hence, they are more
likely to be adopted when the rm possesses a larger knowledge
base.
Hypothesis 1. Firms with more technological resources are more
likely to use a new technology.
2.2. Human resources qualications and technology adoption
The use of AMT could be also related to workers qualications.7
The most common theory is the skill-bias technical change hypothesis, which states that there is a relationship of complementarity
between new technologies and skilled workers because the latter
are the only ones able to fully implement these technologies (Piva
et al., 2005, p. 143). Doms et al. (1997) maintain that the positive
correlation between technologies and the attributes of the workforce may be explained using three arguments. First, at a general
level, AMT increase the level of automation in a factory. Workers using these machines must, at least, have reasonable language,
technology is more complex. However, the fact that we do not have information on
all these ve dimensions (or other that could be considered as important) precludes
as from being more precise in our arguments.
7
Some papers in the agricultural economics literature have used education to
explain adoption. However, its use seems to be more related to the concept of
absorptive capacity. For example, Huffman (1974, p. 85) refers to an allocative
effect, understood as the human agents ability to acquire, decode, and sort market and technical information efciently. We attempt to emphasise the idea that
these investments are important for exploiting and appropriating the returns to the
technology.
1609
reading and basic math skills and they should also able to deal
with higher levels of abstraction and to act quickly on complex
information (Parthasarthy and Sethi, 1992). Thus, more automated
plants will employ relatively more educated and skilled workers.
Second, the introduction of a more technology-specic level may
affect the organization of the workforce and will require skilled
operators and technicians who replace skilled craftspeople but also
less skilled workers. Finally, many of these technologies require
qualied support staff to install and maintain them.
It could be argued that these abilities have commodity-like
character, given employee mobility, not constituting a difcult-toimitate resource. However, researchers have maintained that more
qualied workers are more likely to invest in rm-specic investments. The reason is that their mobility opportunities are less than
for employees with low qualications and their activities tend to
imply the use of their intellectual capabilities in idiosyncratic tasks
(Vicente-Lorente, 2000). This makes it highly likely that more qualied workers invest more in rm-specic abilities than less qualied
ones.
More qualied workers also provide other abilities related to
innovation and management that are important in the use of new
technologies. First, as mentioned above, new technologies frequently enable process and product innovations (Koellinger, 2008).
Education contributes to the innovation process by increasing a
persons capacity to think systematically and creatively about
techniques (Wozniak, 1984, p. 71). Second, education may also be
related to management skills, in other words, the increasing ability of educated workers to effectively integrate new technologies
into the activities of the rm. In fact, given that technology-related
knowledge may be contracted and is not rm- (but technology-)
specic, it has been argued that management skills are the only
likely source of competitive advantage: they are path dependent,
they tend to be tacit and rm-specic and they may be socially
complex (Mata et al., 1995).
Hypothesis 2. Firms with more highly qualied personnel are
more likely to use a new technology.
2.3. Marketing resources and technology adoption
Complementarities are also present in marketing activities
(Teece, 1986). The literature on management understands marketing investments as rm-specic assets (see, for example,
Balakrishnan and Fox, 1993; Vicente-Lorente, 2001). Like technological resources, investments in marketing create intangible
resources, such as reputation, brand image and closer relationships
with customers. These resources are frequently difcult to imitate:
they may be subject to time compression diseconomies (Dierickx
and Cool, 1989) or be socially complex (Barney, 1991). Investments
in marketing activities help the rm to develop a good reputation or
a brand image. These assets are quasi-public, in the sense that they
can continue to deliver services after being used. In other words,
once they have been developed, they can be used to support product development, market development and diversication (Ansoff,
1965).
To understand the complementarities between AMT and
marketing investments it is rst necessary to emphasize two characteristics of the former. First, AMT can be used to improve the
information processing capabilities of rms (Kotha and Swamidass,
2000). Firms can use new technologies in order to cope with the
uncertainty surrounding the activities that they perform, given
that they help to collect and manage information. Those rms
that follow a strategy that is intensive in information processing
requirements will benet most from the use of new technology.
Second, AMT also improve the exibility of rms to adapt to changing demands (Parthasarthy and Sethi, 1992). In particular, they
1610
allow more frequent changes in the production line, which could

be used to satisfy the diverse needs of consumers and provide the
rm with a way to obtain scope economies.
These two characteristics are key to understand the interdependence that arises between advanced manufacturing technologies
and the marketing function (Blois, 1985; Kotha and Swamidass,
2000). Although the benets of higher information processing capabilities and more exibility may be important for a variety of
rms, they are likely to be critical for rms investing more intensively in marketing assets. Both Kotha and Swamidass (2000) and
Parthasarthy and Sethi (1992) argue that investments in AMT are
more important in rms following a product differentiation strategy. These rms focus on offering a more varied set of products and
present more product innovation than rms pursuing a cost leadership strategy (Porter, 1980). On the one hand, this means that
the exibility provided by AMT is of great importance. On the other
hand, rms following a differentiation strategy are likely to have a
higher level of complexity and they present more discontinuities
in the production process (Kotha and Swamidass, 2000). Again, this
will make the information processing capabilities of AMT critical.
An important observation in this argumentation is that investments in marketing are a relevant part of a differentiation strategy
(Hill, 1988). Advertising outlays may be used to build the reputation
and brand image necessary to support differentiation. Therefore,
although the complementarity between marketing investments
and AMT may arise in any rm, they are more likely to appear
in those cases in which marketing investments are more intense,
providing incentives for adoption.8
Hypothesis 3. Firms with more marketing resources are more
likely to use a new technology.
2.4. Complementarities between technologies
As well as the complementarities between the use of new technologies and rm assets, complementarities between technologies
may also arise (Stoneman and Kwon, 1994). According to Milgrom
and Roberts (1990), the adoption of a cluster of technological innovations, which share some basic technological properties, is subject
to signicant complementarities. Under such circumstances, interdependences must be taken into account as they are likely to affect
diffusion.
Adopting powerful combinations of advanced technologies can
leverage the technology gap a rm possesses over its competitors
(Castellaci, 2002). The key advantage that arises from the use of
multiple process technologies is that they can build complex manufacturing systems, which make manufacturing programmable and
lead to timely information transfer across departments, employees, customers and suppliers. These systems are important since
most manufacturing technologies are relatively available on the
open market and any particular competitive advantage is difcult
to defend. One way to achieve competitive advantage consists of
building systems in which the number of elements and the degree
of interaction among them is what makes them inimitable (Rivkin,
2000). In other words, while the adoption of individual technologies
may no have effect on competitive advantage, certain technology combinations may help in building and sustaining it (Sinha
and Noble, 2008) due to complementarities with past adoptions
(Colombo and Mosconi, 1995).
Previous work on complementary technologies has provided
theoretical arguments and empirical evidence on the existence
of complementarities between different advanced technologies

(Arvanitis and Hollenstein, 2001; Astebro et al., 2005; Colombo and
Mosconi, 1995; Milgrom and Roberts, 1990; Stoneman and Kwon,
1994). From an empirical point of view, Arvanitis and Hollenstein
(2001), Colombo and Mosconi (1995), Gmez and Vargas (2009)
and Stoneman and Kwon (1994), demonstrate the importance of
complementarities across adopted technologies. Beede and Young
(1996) found enormous diversity in adoption patterns within the
same industry, as well as important differences in the effect of
various technology combinations on performance, suggesting the
need to consider technology bundles in assessing adoption scenarios. More recently, Sinha and Noble (2008) found that rms
that adopted several technologies were best positioned to survive,
illustrating the importance of cumulative technology adoption.
Hypothesis 4. The use of a technology is positively related to the
use of the other technologies with which it forms a system.
3. Methodology, sample and variables
3.1. Sample description
The dataset used for this study is drawn from the Survey of Business Strategies (ESEE). This is an annual survey on the activity of
Spanish manufacturing rms and their business strategies nanced
by the Ministry of Industry and carried out by the SEPI foundation. Although it is not specically designed to analyse technology
adoption, it includes information on a number of technologies that
are used by rms. The survey covers rms which have 10 or more
employees and whose principal economic activity is listed in one
of the two digit manufacturing industries of the NACE-Rev.1. In the
base year, surveyed rms employing between 10 and 200 people
were selected by means of a random sampling scheme, while rms
with more than 200 were surveyed on a census basis. Although
the survey has been administered annually to rms since 1990,
questions about adoption behaviour were only included in 1994,
1998, 2002 and 2006. This does not permit us to establish causality relationships because the date of adoption of the technology
is unknown. Fortunately, we do have lagged information on other
rm and market characteristics that will be used as explanatory
variables, as is explained below. Therefore, our objective is to nd
the combination of resources that is more likely to be used with
new technologies. After selecting all the observations for which
data on the independent variables is available, we are left with 4418
observations that will be used in the empirical analysis.9
Our main analysis is carried out on four process technologies. We focus on computer numerical controlled (CNC) machines,
computer aided design (CAD), robotics and exible manufacturing
systems (FMS), given that these are the technologies for which a
longer observation window is available.10 These technologies have
been analysed in previous diffusion studies. In fact, and as we have
mentioned before, they are part of what is termed Advanced Manufacturing Technologies. AMT is used as an umbrella term to describe
an automated production system of people, machines and tools
for the planning and control of the production process, including the procurement of raw materials, parts and components and
the shipment and service of nished products (Pennings, 1987,
p. 198).
A characteristic of these technologies is that they are easy
to integrate electronically. The key advantage of AMT is that
As mentioned by the two anonymous referees, the link between marketing

resources and AMT is expected to be weaker than in the case of the other two
elements (technological and human resources), given that their interaction is less
direct.
The usual tests show that the sample is representative of the total population.
In the survey, rms are asked whether their production process uses any of
the following systems: (1) computer numerically controlled (CNC) machines, (2)
robotics, (3) computer assisted design (CAD), (4) combination of some of the above
systems through a central computer (FMS).
10
1611
Table 1
Percentage of adopters and non adopters by technology and year.
CNC
1994
1998
2002
2006
ROBOTICS
CAD
FMS
B2B
B2C
S2B
Yes
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
No
Yes
No
33.1
43.9
48.3
51.5
66.9
56.1
51.7
48.5
18.1
26.3
26.4
31.5
81.9
73.7
73.6
68.5
24.2
35.3
36.9
40.0
75.8
64.7
63.1
60.0
21.4
26.7
22.3
27.1
78.6
73.3
77.7
72.9
5.8
9.9
94.2
90.1
4.1
5.1
95.9
94.9
13.8
25.9
86.2
74.1
The expressions Yes and No indicate whether the rms have adopted or not
complex manufacturing systems may be built that make manufacturing programmable and lead to timely information transfer
across departments, employees, customers and suppliers. This adds
a new element to the study of complementarities: these technologies not only interact with the endowment of resources of the rm,
but they are frequently used as a system. This reinforces the idea
that complementarities have a key role to play in the understanding
of performance differences and adoption behaviour.
Although these four technologies will be the main focus of
our analysis, the Survey also includes information on the use of
e-business in 2002 and 2006, the last two periods for which information has been published. In particular, the available data refer to
the adoption of business-to-business (B2B), business-to-consumer
(B2C) and supplier-to-business (S2B) technologies. Although the
main objective of these technologies is to manage information
electronically, they are also used for enhancing communication
between rms. In other words, similarly to Powell and Den-Micallef
(1997), we distinguish between technologies that are used within
rms (in-rm technologies, namely, CNC, CAD, robotics and FMS)
and those used beyond their limits (beyond-rm technologies,
namely, B2B, B2C and S2B). Following Hypothesis 4, our estimations
should show a stronger relationship between in-rm technologies
than between in-rm and beyond-rm technologies.
Table 1 offers a rst approximation to the data, showing the
distribution of adopters and non-adopters by technology, and year.
As can be seen, the number of adopters is different depending on the
technology. CNC is the most-used technology (51.5% of the rms use
it in 2006), followed by CAD (40.0% in 2006). Robotics and FMS have
the lowest gures (31.5% and 27.1%) among in-rm technologies.

However, the adoption of e-business technologies presents much
lower percentages. Looking at the gures for 2006, we can observe
that B2B is adopted by 9.9% of the rms, B2C by 5.1% and S2B by
25.9%.
Similarly, Table 2 presents the distribution of adopters by industry. It shows clear differences between the sectors included in the
sample. Firms in the Motors and autos industry are the most
active in the adoption of CNC machines, followed by the ones operating in Other transport material and Furniture. The adoption
of robotics shows its highest values in the Motors and autos sector, with the Beverages industry and Other transport material
following. Firms belonging to Other transport material, Motors
and Autos and Machinery for agriculture and industry are the
most frequent users of CAD. Finally, in the case of FMS, the most
frequent users are Motors and Autos, Other transport material
and Paper. In contrast, Meat products is the least active industry in CNC, CAD and FMS and Other manufacturing in the case of
Robotics.
The gures on the use of e-business technologies are, as mentioned above, generally lower. The most active sectors are Paper
(in the case of B2B) and Edition and graphical arts (B2C and S2B).
Contrarily, rms belonging to Leather and footwear (B2B and B2C)
and Wood (S2B) are the least active in adoption.
Finally, Table 3 shows the conditional aspects of technology
adoption, revealing the complementarities that arise in the use of
AMT. For any of the four main technologies studied, it shows the
percentage of adopters that are also using none, one, two and three
Table 2
Percentage of adopters in 2006 by technology and industry.
CNC
ROBOTICS
Meat products
Food and tobacco
Beverages
Textiles and clothing
Leather and footwear
Wood industry
Paper
Edition and graphical arts
Chemical products
Plastic and rubber
Non-metallic minerals
Metallurgy
Metallic products
Machinery for agriculture and industry
Machinery for ofces, data processing, etc.
Electrical material and accessories
Motors and autos
Other transport material
Furniture
Other manufacturing
28.6
41.5
37.5
36.9
34.8
47.5
56.4
37.9
34.8
52.5
46.1
62.0
62.1
64.5
46.2
57.4
75.4
75.0
67.2
42.1
21.4
27.7
43.8
16.7
17.4
25.0
41.0
17.2
18.8
42.6
35.5
42.0
34.7
32.9
30.8
38.9
62.3
43.8
21.3
15.8
Total manufacturing
2
51.5
72.2***
31.5
68.5***
*
***
Coefcient statistically signicant at the 90% level.

CAD
3.6
10.6
25.0
42.9
17.4
17.5
46.2
31.0
23.2
42.6
31.6
56.0
45.2
69.7
53.8
57.4
70.5
87.5
39.3
26.3
40.0
160.2***
FMS
B2B
B2C
S2B
7.1
20.2
37.5
21.4
8.7
12.5
38.5
24.1
31.9
29.5
22.4
38.0
29.8
31.6
23.1
31.5
45.9
43.8
19.7
15.8
10.7
5.3
6.3
8.3
0
2.5
23.1
15.5
15.9
11.5
9.2
14.0
7.3
7.9
15.4
13.0
4.9
6.3
13.1
10.5
3.6
2.1
12.5
6.0
0
0
5.1
22.4
4.3
4.9
2.6
4.0
2.4
5.3
0
5.6
6.6
0
6.6
5.3
21.4
17.0
37.5
25.0
13.0
10.0
23.1
39.7
24.6
23.0
17.1
36.0
29.0
30.3
30.8
25.9
32.8
37.5
29.5
21.1
27.1
44.0***
9.9
26.3
5.1
48.4***
25.9
29.3*
1612
Table 3
Probability of adoption of AMT depending on the number of other technologies adopted and unconditional probability of adoption (2006).
CNN
ROBOTICS
CAD
FMS
Unconditional probability
28.4%
16.9%
9.2%
10.0%
53.4%
30.0%
31.9%
16.8%
78.7%
49.1%
61.6%
39.4%
89.3%
64.3%
81.2%
60.7%
51.5%
31.5%
40.0%
27.1%
of the other technologies. The data clearly shows an increasing

probability of using CNC, Robotics, CAD and FMS as rms increase
the number of other AMT technologies in which they have invested.
Therefore, this preliminary evidence seems to support the idea that
the adoption of a given technology is favoured by the use of any of
the other technologies with which it forms a system.
3.2. Main variables

3.2.1. Stock of technological resources
We approximate the stock of technological resources by using
the ratio of research and development capital to sales. For any given
year (t) for which data on technology adoption is available, we
compute research and development capital (K) through a partial
inventory of past (3 previous years) and present annual internal
R&D11 expenditures (R) with a constant depreciation rate, :
Kt =
3
(1 ) Rtk
k=0
The annual depreciation rate () was assumed to be 15% in

accordance with Adams (1999), Griliches (1981), Griliches et al.
(1981) and Villalonga (2004). The use of a depreciation rate can be
explained by the decay of knowledge over time (Argote et al., 1990)
and by the loss of economic value due to the development of new
knowledge and technologies (Oriani and Sobrero, 2003).
3.2.2. Human resources

The available information also allows us to calculate a measure
of employee skills using the number of employees who possess a
university degree (see, for example, Vicente-Lorente, 2000).12 With
this information, we calculate the ratio of the number of employees
with a university degree to the total number of employees of the
organization. While this is not a perfect measure of employee skills,
it has been used in previous papers (see, for example, Arvanitis and
Louikis, 2009; Doms et al., 1997), and it is considered a reasonable
proxy.
3.2.3. Stock of marketing resources

Similarly to technological resources, the stock of marketing
resources is measured as the ratio of advertising capital to sales.
As with technological capital, for any given year (t) for which data
on technology adoption is available, the value of marketing capital
(P) is constructed using a partial inventory of past (3 previous years)
11
The survey collects data on (1) external expenses, (2) internal expenses and
(3) total expenses ((1) + (2)) in research and development. To calculate the stock of
technological resources, we have used internal R&D expenses.
12
Firms are asked to classify all the workers according to their qualications.
The options available are: (1) engineers and graduates, (2) middle level engineers,
experts and qualied assistants and (3) other personnel. From this information we
calculated the percentage of workers with a university degree.
and present annual advertising expenditures13 (A) with a constant

depreciation rate, :
Pt =
3
(1 ) Atk
k=0
Even though there is no consensus in the literature on the

rate of depreciation, we follow Hirchey and Weygandt (1985) and
Villalonga (2004) and use a depreciation rate of 45% going back
three years.
3.3. Control variables
3.3.1. Firm size
Firm size has traditionally played a prominent role in rank
models of diffusion, usually presenting a positive effect on the probability of adoption. There are several explanations provided for this
impact.14 First, rm size has been positively related to the existence
of complementary assets within the rm. Hence, larger rms would
obtain more protability from the technology and would have more
incentives to adopt it early (Colombo and Mosconi, 1995; Teece,
1986). A second explanation focuses on the concept of economies
of scale. Larger rms would be able to spread the costs of the adoption of new technologies among a larger number of units. Firm size
is measured as the total number of employees working for the rm.
3.3.2. Financial constraints
The literature on diffusion attributes a role to nancial constraints in the determination of a rms adoption behaviour. The
arguments in favour of a negative relationship between nancial constraints and adoption take three factors into account:
uncertainty on the cash ows to be perceived from the adoption, information asymmetries between borrowers (the adopting
rm) and lenders and, lastly, the frequent association between the
diffusion process and investment in intangible and technologyspecic assets whose economic value could be difcult to recover
(Canepa and Stoneman, 2005; Stoneman, 2001). As a consequence,
we expect the availability of nancial resources to have a positive
impact on technology adoption. This variable is measured as the
ratio of debts to total assets.
3.3.3. Corporate status
The corporate status of a rm may also have an impact on the
decision to adopt. By corporate status we mean whether the rm
is a part of a larger business group or not. As noted by Baptista
(2000), Bartoloni and Baussola (2001), Karshenas and Stoneman
(1993) and Rose and Joskow (1990), the effect of this variable on
the adoption decision is likely to be ambiguous. On the one hand,
independent rms may be better positioned with regard to implementation, once the decision to adopt has been made. On the other,
rms that are part of a larger institution may be better informed
and bear less risk in adopting new technologies. Corporate status is
13
Firms provide some information on their accounts. For example, they are asked
to report the expenses in publicity, advertisement and public relations.
14
Recent efforts have tried to explain this relationship (see, for example Astebro,
2002), although some confusion about the underlying mechanism still remains.
measured through a dummy that takes a value of one when the

rm is part of a larger corporate unit.
3.3.4. Ownership
The fourth variable considered in the analysis is the type of
ownership. The argument in favour of a positive effect is that the
subsidiaries of a foreign rm may have access to the resources
of the parent rm. Foreign investment may be a vehicle for the
introduction of superior technology and scientic knowledge. Previous evidence on adoption models revealed a positive impact of
foreign ownership (Baldwin and Diverty, 1995; Bosworth, 1996;
Faria et al., 2003), although this relationship was only signicant in
the case of certain technologies. The presence of foreign capital in
the focal rm is measured through a dummy that takes a value of
one when the presence of foreign investors in the capital is higher
than 30%.
3.3.5. Propensity to export

The literature on international technology diffusion has argued
that international trade is a channel through which rms can obtain
information on new technologies (Keller, 2004). Exporting rms
can also be expected to face more competitive international markets and be more likely to adopt new technologies in order to
confront the higher levels of competition in the international arena.
Previous empirical works have found either a positive relationship
between exports and the adoption of new technologies (Cohen,
1975; Riedel, 1975) or no relationship at all (Lal, 1999, 2002). The
ratio of total exports to sales is used to capture the effect of this
variable.
3.3.6. Industry-specic rank effects

We also consider two industry-specic inuences on technology adoption. First, market structure has frequently been linked
to the incentives of the rm to adopt a new technology, although
with ambiguous inuences (see, for example, Reinganum, 1981).
We measure market concentration through the market share of
the four largest rms (CR4 ). Second, one of the conclusions from
the literature on technology diffusion is the importance of industry
effects. Technology diffusion is largely determined by the technological characteristics of a given production process and this,
in turn, is intimately linked to the sector of activity in which the
rm is operating. Accordingly, we introduce 19 dummy variables
to account for the 20 different sectors identied in the survey.
Finally, and due to the fact that traditional models of diffusion have also taken other determinants of technology use into
consideration, we introduce epidemic and stock effects into our
model (Karshenas and Stoneman, 1993). Both these effects are
taken into account through the inclusion of time variables. The
passage of time has two opposite inuences on technology adoption. On the one hand, the stock effect should mean that, as the
number of adopters grows, the probability of adoption should be
lower, given the lower returns. On the other, the epidemic effect
acts through a learning-by-contact process. Hence, the number of
previous adopters would also produce a positive effect on adoption as information ows increase in the industry. Such an effect
could counteract the expected stock effects, making the net effect
ambiguous (Baptista, 2000; Luque, 2002). We measure epidemic
and stock effects through three dummy variables that take a value
of one for the years 1998, 2002 and 2006, leaving 1994 as the base
year.15
15
Appendix A contains some descriptive statistics of the covariates.
1613
3.4. Model and estimation strategy

The model that we estimate is similar to other adoption models that have been used in the literature (see, for example, Barbosa
and Faria, 2008; Battisti et al., 2009). Let A(t)ijk be the net expected
protability from the adoption (at time t) of technology k by rm i,
that operates in industry j. The net expected returns from the technology depend on a set of rm-specic variables (xi ) that capture
rank effects, a set of industry characteristics (zj ), epidemic, order
and stock effects (st ) that are expected to vary over time, and an
error term (ij ) that measures unobservable effects, in the following
way:
A(t)ijk = xi + zj + st + ij
The data set we use imposes two restrictions on this model. The
rst is that we do not have information on the date at which a rm
adopts the technology, but only on whether it has adopted it at predetermined points in time. In other words, as in other papers in the
literature (see, for example, Battisti et al., 2009), we cannot estimate
a truly dynamic model. Second, the data describe adoption as a discrete choice, which means that we do not observe the returns to
innovation, but only an indicator variable (Iijk ) that takes a value of
one if rm i adopted technology k at time t (A(t)ijk 0), and zero
otherwise. This suggests using qualitative dependent variable models (either logit or probit). In other words, we model the probability
of adoption of the technology as a function of multiple explanatory
variables capturing the rank, stock and epidemic effects described
above. The probability of adoption may be expressed as a function
of a vector of variables reecting all three effects. Given the documented similarities between the logit and probit specications and
the considerations that we make below, we chose the probit link
to perform our estimations.
As mentioned when developing our empirical model, technologies may be complementary. In the case of the technologies
analysed here, it has been argued that they form a cluster in which
we could include numerically controlled machines, computer aided
design, robots and exible manufacturing. Recent research has
extended the number of factors affecting adoption in order to take
into account that technologies are complementary.
The idea is that some technologies are difcult to use in isolation
and, therefore, need to be adopted as systems that are jointly used
in certain activities. From an empirical perspective, the assumption that technologies do not operate in isolation has an important
methodological implication. If the protability of adopting a given
technology is related to the adoption of some other technology,
this means that the two decisions are interdependent. Therefore,
the estimation of any model of adoption should consider the fact
that the rm may be using other complementary technologies.
Given that our data on adoption provides us with information on the use of several technologies, we estimate the decision
to use the technology through a multivariate probit model. This
model can be seen as a generalization of the bivariate probit model,
allowing more than two equations to be simultaneously estimated.
The model captures the complementarities in the adoption of
different technologies by allowing the disturbances of the different equations to be correlated. These correlations may be due to
complementarities (positive correlation) or substitutabilities (negative correlation) between different technologies. Nevertheless,
correlation could also be the result of unobservable rm-specic
characteristics that affect adoption decisions but that are not easily
captured by measurable proxies (Belderbos et al., 2004; Santamaria
and Rialp, 2007). In fact, the arguments that we have used in this
paper about the existence of complementarities between resources
and technology could be extended to other assets that are unobservable. The multivariate probit model takes into account the
1614
interrelationships arising in adoption, although it is not able to

distinguish between the two sources of correlation described. However, if correlation exists, the estimates of the separate equations
would provide us with inefcient estimates.
The estimation is carried out using Statas mvprobit command, which applies the method of simulated maximum likelihood
(SML) that uses the GewekeHajivassiliourKeane (GHK) smooth
recursive-conditioning simulator to evaluate the multivariate normal distribution. Following Cappellari and Jenkins (2003), the
simulated probabilities are unbiased and bound within the (0, 1)
interval. The variancecovariance matrix V of the cross-equation
error terms has values of 1 on the leading diagonal, and the offdiagonal elements, correlations jk = kj, are to be estimated. The
parameter jk is the co-variance between the error terms of equations j and k.
4. Results
Table 4 shows the results of estimating a multivariate probit on
the 4418 observations available.16 Columns 14 estimate four models in which only control variables are included. They are used as
a benchmark. A second set of estimations is presented in columns
58, which introduce the variables that help us to test our hypotheses. As we can see, all the models are globally signicant, given the
high values of the Wald statistic. Furthermore, their comparison
favours the estimations presented in columns 58. As shown at the
bottom of Table 4, the LR test is highly signicant, pointing to the
importance of complementary resources for explaining adoption.
We should highlight that not only the coefcients but also the tratios accompanying them remain fairly stable across the two sets
of estimations.17
Hypothesis 1 argued that rms with more technological
resources would be more likely to use advanced technologies. This
is, in fact, the case for the four technologies: the coefcients accompanying the technological resources variable are positive and
signicant. Therefore, the evidence conrms our hypothesis that
the use of new technologies is favoured by the presence of higher
levels of R&D-derived knowledge inside the rm.
Hypothesis 2 stated that technology adoption should be also
be conditioned by the qualications of human resources. In this
case, we observe clear differences between the four advanced
technologies analysed. The variable human resources is highly
signicant in the case of computer aided design and exible manufacturing, whereas the other two technologies do not show any
impact. This result seems to resemble that of Dunne and Troskes
(2005) paper. These authors found that the effect of skilled labour
on technology use varied across technologies, nding a stronger
correlation between technologies associated with design and engineering functions and human resources (CAD) than those more
closely associated with production activity (CNC and robotics).18
16
Following the comments of one of the anonymous referees, we also estimated an
ordered probit model that considers whether the rm adopts 0, 1, 2, 3 or 4 manufacturing technologies. The results are highly consistent with the ones presented here:
rm size, the availability of nancial resources, technological and human resources,
export propensity and integration in a business group are signicant predictors of
adoption. The estimates also show the relevance of epidemic effects.
17
We calculated the variance ination factors (VIF) in order to detect multicollinearity problems. The maximum VIF was 5.02 and the average VIF was 2.47.
Therefore, we do not have reasons to suspect that they could be present.
18
The variable used to test this hypothesis is calculated as the percentage of workers with a university degree. As one anonymous reviewer points out, this has the
problem that some of these employees will be working on R&D activities and this
may lead to double counting. Although we do have information on the number of
employees working on R&D activities, the classication of the workers into different
groups does not match the one for the whole rm. In any case, we rerun the estimations with a measure of human resources from which we subtracted the number of
Similarly, Hypothesis 3 stated that rms with more marketing

resources would be more likely to use new process technologies. Again, we nd clear differences between the technologies
analysed. Whereas the variable marketing stock is signicant
for robotics and, to a lesser extent, for computer aided design
marketing resources are not signicant in the case of computer
numerically controlled machines and exible manufacturing. These
ndings are consistent with the ones of Kotha and Swamidass
(2000). Using information on several AMT, they conclude that those
rms following a differentiation strategy (an strategy associated
with high marketing investments) are more likely to use product
design technologies (including CAD) and high volume automation
technologies (including robotics).
Our analysis also included a set of control variables that are frequently used in the literature. Firm size is an important predictor
of the likelihood of using a technology, as it shows a positive and
signicant effect in all the estimations. Having controlled for size,
the importance of nancial constraints is also signicant for both
robotics and exible manufacturing, but not for CNC and CAD.
Three other rm-specic variables present a signicant impact
on the adoption of new technologies. First, our results show that the
propensity to export (ratio exports to sales), is positively associated
with adoption in all cases. This result is interesting from the perspective of the international diffusion of technology (Keller, 2004).
Several authors have argued that international trade is an important diffusion channel (see, for example, Eaton and Kortum, 2002;
Grossman and Helpman, 1991). In the case of exports, the anecdotal evidence shows that rms may benet from interacting with
foreign customers, given that they tend to require higher standards
and also provide information on how to meet them (Keller, 2004:
767). Furthermore, this result also seems to be consistent with the
idea that exporters are different from non-exporters in terms of
productivity (Bernard and Jensen, 1999) and innovation. In fact, the
evidence from Spain shows that exporters are more productive and
more likely to introduce process innovations than non-exporters.
et al. (2009) nd that not

Using the same database, Mnez-Castillejo
only the most productive rms have a higher probability of exporting, but also that process innovations increase the probability of
exporting.
Second, the data conrm that the integration of the rm into a
business group has an impact on adoption. This result is in line with
the idea that establishments that are part of a larger corporation
will experience less uncertainty and less nancial constraints when
adopting new technologies. Third, we nd that foreign ownership is
only signicant in the case of robotics.19 This result is partially consistent with previous evidence that points out that foreign-owned
companies invest more in advanced manufacturing technologies
and obtain higher operational performance (Beaumont et al., 2002).
Third, our results are also generally consistent with the existence of an epidemic effect. The time dummies show that the use
of all the four technologies is more frequent in 2006.20
Apart from assessing the impact of certain complementary
resources on adoption, the use of the multivariate probit model
employees working for the R&D function that were (1) graduates and (2) middle
level technicians. The results were similar to the ones presented here.
19
An anonymous reviewer suggested investigating on the sources of this capital. Unfortunately, our data do not allow us to identify the country in which these
investments originate. We were, however, able to identify the country in which a
rm invests for the two last periods of our sample. Our results are in line with the
work of Grifth et al. (2006), suggesting that investments in OECD countries are
signicant for explaining Robotics and CAD adoption.
20
We also estimated a model that included both the percentage of adopters per
industry and in the economy as a whole. The only variable that remains signicant
in this model is the rst, offering support for the existence of an industry specic
epidemic effect.
1615
Table 4
The effect of technological, human and marketing resources on AMT adoption.
CNC (1)
Robotics (2)
CAD (3)
FMS (4)
CNC (5)
Robotics (6)
Stock of technological resources
0.17***
(3.76)
0.08
(0.87)
0.38***
(7.77)
0.29***
(2.92)
0.19***
(4.47)
0.10
(1.11)
0.48***
(9.35)
0.33***
(3.41)
Human resources
Stock of marketing resources
Export propensity
0.44***
(5.17)
0.38
(1.32)
0.30***
(5.69)
0.02
(0.01)
0.24***
(3.98)
0.34***
(5.69)
0.40***
(6.48)
Yes
0.87***
(4.32)
0.63***
0.51***
(7.13)
(5.74)
0.26
0.20
(0.84)
(0.64)
***
0.47
0.33***
(8.39)
(6.01)
0.16**
0.04
(2.49)
(0.64)
0.23***
0.35***
(3.36)
(5.37)
0.27***
0.38***
(3.95)
(5.84)
0.39***
0.45**
(5.72)
(6.84)
Yes
Yes
1.44***
2.47***
(6.30)
(9.28)
0.409*** (17.46)
***
0.457 (21.18)
0.292*** (11.80)
0.405*** (16.77)
0.297*** (11.69)
0.413*** (17.13)
4418
1907.70***
0.16***
(3.50)
0.07
(0.83)
1.61***
(2.86)
0.12
(0.71)
0.22
(0.91)
0.41***
(4.81)
0.35
(1.22)
0.29***
(5.37)
0.00
(0.08)
0.24***
(4.02)
0.34***
(5.66)
0.40***
(6.44)
Yes
0.89***
(4.38)
0.36***
(7.28)
0.28***
(2.80)
2.42***
(4.12)
0.14
(0.73)
0.42***
(2.70)
0.60***
(6.65)
0.23
(0.74)
0.45***
(7.95)
0.18***
(2.82)
0.25***
(3.64)
0.29***
(4.17)
0.42***
(5.98)
Yes
1.49***
(6.52)
Firm size
Firm debt ratio
Market concentration
Integrated in a business group
Foreign capital
Year 1998
Year 2002
Year 2006
Industry dummies
Constant
Rho2,1
Rho3,1
Rho4,1
Rho3,2
Rho4,2
Rho4,3
No. observations
Wald test
Comparison test
LR test of Rho2,1 = Rho3,1 = Rho4,
1 = Rho3,2 = Rho4,2 = 0 Rho4,3 = 0
*
**
***
0.33***
(3.75)
0.31
(0.99)
0.39***
(7.10)
0.01
(0.21)
0.14**
(2.16)
0.02
(0.31)
0.13*
(1.93)
Yes
1.74***
(7.32)
1047.75***
CAD (7)
FMS (8)
0.17***
(3.89)
0.12
(1.29)
3.66***
(6.32)
0.66***
(3.64)
0.32*
(1.74)
0.44***
(4.92)
0.23
(0.75)
0.28***
(4.93)
0.01
(0.22)
0.37***
(5.60)
0.39***
(5.87)
0.47***
(6.88)
Yes
2.54***
(9.52)
0.409*** (17.40)
***
0.459 (21.16)
0.290*** (11.67)
0.403*** (16.53)
0.294*** (11.48)
0.404*** (16.54)
4418
1992.44***
111.98***
1025.56***
0.46***
(8.82)
0.33***
(3.36)
2.81***
(4.86)
0.71***
(3.88)
0.33
(0.95)
0.28***
(3.13
0.32
(1.02)
0.35**
(6.21)
0.00
(0.02)
0.14**
(2.17)
0.01
(0.09)
0.11*
(1.68)
Yes
1.76***
(7.36)

also allows us to estimate the correlations between the technologies once the variables included in Table 4 have been controlled
for. These correlations are presented at the bottom of Table 4
and provide us with a way of testing Hypothesis 4. They are all
positive and signicant, suggesting that the arguments leading to
Hypothesis 4 are correct. However, these correlations could be
interpreted in two ways. First, a positive correlation could be due
to the inuence of rm-specic factors that are not included in our
estimations and that determine the propensity of some rms to
adopt all the technologies. Therefore, they provide us with a way
of assessing the inuence of unobservable rm effects. Second, a
positive correlation could mean that the technologies form part of
a system. Hence, the adoption of one technology would increase
the probability of the adoption of the other technologies of the
system.
One possibility to investigate Hypothesis 4 further is to reestimate our full model adding other technologies that we suspect
are not closely related to the four process technologies analysed.
In other words, the correlations between non-related technologies
should capture the inuence of non-observable rm-specic effects
related to the adoption of all the technologies. Fortunately, our data
set also contains information on the adoption of three e-business
technologies. From 2000 on, the survey includes a question on
whether the rm buys goods or services (providers) through the
Internet (S2B), whether it sells to nal consumers through the
Internet (B2C) and whether it sells to rms through the Internet (B2B). Given that the three technologies supporting these
systems are not obviously related to process technologies, a dramatic reduction in the value of the correlations should be observed.
Unfortunately, this signicantly restricts our observation window
Table 5
Correlation matrix between process and information technologies.
CNC
Robotics
CAD
FMS
B2B
B2C
S2B
***
CNC
Robotics
CAD
FMS
B2B
B2C
1
0.414***
0.470***
0.337***
0.105***
0.138***
0.080***
1
0.396***
0.357***
0.090***
0.130***
0.067***
1
0.436***
0.075***
0.009***
0.131***
1
0.122***
0.199***
0.178***
1
0.797***
0.373***
1
0.355***
1616
(to years 2002 and 2006) and the number of available observations
(to 2367).
The correlations resulting from estimating a full multivariate
probit model with the seven technologies are presented in Table 5.
The conclusion is that the correlations between process and ebusiness technologies are clearly lower than those within the
groups. The highest correlation between the two groups is 0.199
(B2B and FMS), the value is non-signicant in two of the cases
(CAD-B2B and CAD-B2C) and three correlations are below 0.10.
In order to make sure that these differences were signicant, we
compared all the correlations between in-rm and beyond-rm
technologies.21 For a given correlation between two technologies belonging to the in-rm technologies (for example, CNC and
Robotics), we tested whether the coefcient was different from that
of the in-rm technology and each of the beyond-rm technologies (i.e., we compared it with CNC-B2B, CNC-B2C and CNC-S2B).
The results (not shown) of all the comparisons are signicantly
different, showing that within-group correlations are higher than
inter-group correlations. In other words, our results conrm that
complementarities between related technologies exist and that
they are very different from the ones corresponding to unrelated
technologies, supporting Hypothesis 4.
5. Conclusions and implications
In this paper, we have focused on the different incentives
that rms have to adopt advanced technologies depending on the
possibilities of appropriating their returns. We have argued that
complementary resources are one of the mechanisms that rms
use to prot from the use of new technologies. One implication of
this hypothesis is that the use of AMT should be more likely in rms
that possess more complementary assets. The evidence presented
in this paper seems to conrm our conjecture: complementary
resources are associated with a rms technology use.
Clear differences in adoption behaviour are identied depending on the technologies analysed. More precisely, only technological resources are unequivocally related to the use of new
technologies in manufacturing rms. The fact that we have studied the use of AMT could help us to understand this result. These
technologies are related to product manufacturing and design.
Therefore, investments in the knowledge base of the rm could
have served to understand and rene both processes, detecting the
need and the role of new technologies. Our results seem to suggest that more complex technologies benet to a larger extent from
these investments, as proposed by Rogers and Shoemaker (1971).
Human and Marketing resources are also important, although they
present a weaker relation and different effects depending on the
technologies analysed. The quality of the services provided by
human resources is positively related to computer aided design
and exible manufacturing, whereas marketing investments are
only related to robotics and computer aided design. These results
are consistent with previous literature (Dunne and Troske, 2005;
Kotha and Swamidass, 2000). The interpretation of these results
has, however, to be taken with some caution, given that we lack
data on other characteristics of the technologies (see, for instance,
Rogers and Shoemaker, 1971) and that, at the time of the study, the
diffusion process had already started.
We also explain the use of advanced technologies through the
complementarities that arise within systems of technologies. In
fact, the interrelations between some of the technologies are the
main reason that we observe positive correlations between them,
once we control for the determinants of adoption. This supports the
21
We followed Buis (2011) in order to get access to the correlations. We thank
Maarten Buis and Stephen Jenkins for their advice.
view of those that contend that the diffusion of new technologies

should not be analysed in isolation.
Our results have implications for the study of technology diffusion and competitive advantage. With respect to the former, they
support the move from epidemic to rank models undertaken by the
literature on diffusion in recent years. Epidemic models contend
that rms adopt new technologies once information on their characteristics reduces the uncertainty surrounding the innovation.
However, although important, research has shown that information diffusion is not the only mechanism that explains why some
rms are early users and others delay the adoption of the technology until the latest stages of the diffusion process or do not invest
in it at all. Although the range of factors involved adds complexity
to any explanation provided, the literature and our results show
that complementarities play an important role.
Complementarities are likely at two levels. First, they should
be important to understand the rm-technology dyad. Both the
resource-based view of the rm (Barney, 1991; Dierickx and Cool,
1989; Wernerfelt, 1984) and Teeces (1986) framework provide
support for the idea that the interconnectedness of resources
should be understood in order to assess the quality of the services provided. In other words, rm resources cannot be analysed
or understood in isolation. Accepting this argument and understanding new technologies as a resource, this implies that the
endowment of resources of a given rm should determine its needs
for or the adequacy of a given technology. Although research in
the information systems and technology adoption literatures has
argued that complementary resources are needed to appropriate
value creation, it has mainly focused on the analysis of performance. This paper, thus, contributes by showing that the incentives
of rms to adopt new technologies depend on the amount of complementary resources that they possess. This not only adds support
to the hypothesis that complementary resources could help us to
understand the link between technology adoption and competitive advantage, but also introduces a new, rm-specic element
into the explanation of diffusion patterns. We have focused on the
analysis of certain intangible resources, based on the argument that
they tend to be more difcult to imitate. However, the range of rm
assets likely to interact with and, therefore, to determine the use
of a given technology is difcult to specify. As our results show,
not all the technologies are expected to be equally related to rm
resources.
Second, complementarities should also be important in
the technologytechnology dyad. Several papers (Colombo and
Mosconi, 1995; Milgrom and Roberts, 1990; Stoneman and Kwon,
1994) have argued that technologies cannot be analysed in isolation. On the contrary, they need to be adopted as systems that are
jointly used in certain activities. Several papers have shown that
multiple adoption behaviour can help the rm to build complex
systems in which complementary relationships between the elements will arise. Moreover, the multiple adoption consideration
opens new questions for research into the factors that affect the
acquisition of new technologies because the technological trajectory and the resource endowments of a rm greatly affect adoption
choices.
These results also have implications for understanding rm heterogeneity and competitive advantage in the context of technology
diffusion. The main argument for maintaining that, despite their
positive effects on rm activities, new technologies are not related
to sustainable competitive advantage is based on the idea that
they can be acquired in the market. In other words, laggard rms
could easily replicate the technology strategy of pioneers by buying units of the new technology. This would lead to a situation
of competitive parity (Barney, 1991), providing no advantages to
any rm. The literature on information systems has coined the
term strategic necessity hypothesis to refer to this idea (Clemons
1617
Table A1
Descriptive statistics of dependent and independent variables.
1. CNC
2. Robotics
3. CAD
4. FMS
5. B2B
6. B2C
7. S2B
8. Firm size
9. Firm debt ratio
10. Stock of technological resources
11. Human resources
12. Stock of marketing resources
13. Export propensity
14. Market concentration
15. Integrated in business group
16. Foreign capital
Mean
Observations
St. Dev.
Min.
Max.
0.45
0.26
0.35
0.24
0.08
0.05
0.19
0.23
0.56
0.01
0.11
0.03
0.19
0.41
0.33
0.18
4418
4418
4418
4418
2367
2367
2367
4418
4418
4418
4418
4418
4418
4418
4418
4418
0.50
0.44
0.48
0.43
0.27
0.21
0.39
0.62
0.23
0.04
0.13
0.13
0.26
0.13
0.47
0.39
0
0
0
0
0
0
0
0.001
0
0
0
0
0
0.20
0
0
1
1
1
1
1
1
1
14.20
1
0.65
1
4.32
1
0.95
1
1
and Kimbrough, 1986; Clemons and Row, 1991). However, recent

research in strategic management has concluded that the process
of technology diffusion among rms may not be as regular and
widespread as expected. Greve (2009) has shown that social networks and cluster theory affect the diffusion of technologies. His
ndings indicate that there may be other sources of competitive
advantage because resources that, in principle, are available to all
will, in fact, be obtained by few rms, allowing them to get a longlasting competitive advantage. Our results contribute to this line
of research by showing that the initial levels of heterogeneity of
rm resources could produce very different results. In fact, they
suggest that complementarities between rm resources and new
technologies create varying incentives for rms to adopt, providing an additional dimension (the technology) through which to
differentiate themselves.
In other words, far from leading to resource similarity and competitive parity, the process of technology diffusion could create
further heterogeneity, at least in the short run. The questions that
remain are whether this increased heterogeneity is permanent
and whether it can create competitive advantages. The answer,
again, seems to depend on the balance between differentiation
and homogenization. Although research on the interrm dimension of diffusion tends to show that the process of adoption affects
the majority of the rms operating in an industry (but see Greve,
2009), research on the intrarm dimension shows that differences in the intensity of technology adoption increase over time
(Fuentelsaz et al., 2009). However, the role of competitive imitation seems to be important when determining the effect of IT
on performance. Although research on the diffusion of new technologies has recognized that a rms returns cannot be assessed
in isolation when arguing in favour of stock and order effects (see,
for example, Karshenas and Stoneman, 1993) and has also studied
the inuence of rivals adoptions on the diffusion of new technologies (Hannan and McDowell, 1987), the predominant view
conceives competitors as information diffusers. Therefore, the role
of competition, heterogeneity generation and competitive advantage creation should be more explicitly considered in technology
diffusion studies. In the end, imitation would depend on the characteristics of the systems created and their inimitability and on
the actions of competitors, which could progressively erode the
rents of early adopters (Koellinger, 2008). In any case, the conrmation of the idea that complementary resources explain technology
adoption and increase rm heterogeneity should not necessarily
lead us to conclude that the net effect of technology on protability is positive in the long run. Even if the relative positions of the
rms are maintained or even if some rms improve their advantage
over their rivals, the high investments necessary to acquire new
technologies could imply that all the rms reduce their initial levels
of protability if they are not able to compensate for them.
Acknowledgements
We acknowledge nancial support from the Spanish Ministry
of Science and Technology and FEDER (projects ECO2008-04129
and ECO2011-22947), the Regional Government of Aragn (project
S09/PI138-08) and Universidad de La Rioja (projects API 19/08, API
11/25 and EGI 11/29). We are grateful for the comments and sugges
tions provided by Dolores Ann,
Sergio Palomas, two anonymous
referees, the Editor of this Journal, participants at the Conferences
of ACEDE, IAMB, EURAM and at seminars held at the Universities
of Zaragoza, La Rioja and Rey Juan Carlos. A previous version of
this paper was included in the Working Paper Series of the Spanish Fundacin de las Cajas de Ahorros (FUNCAS) (document number
505/2010).
Appendix A. Descriptive statistics
(See Table A1)
References
Adams, J.D., 1999. The structure of rm R&D, the factor intensity of production, and
skill bias. The Review of Economics and Statistics 81 (3), 499510.
Amara, N., Landry, R., Becheikh, N., Ouimet, M., 2008. Learning and novelty of innovation in established manufacturing SMEs. Technovation 28, 450463.
Ansoff, I., 1965. Corporate Strategy: An Analytic Approach to Business Policy for
Growth and Expansion. McGraw-Hill, New York.
Argote, L., Beckman, S.L., Epple, D., 1990. The persistence and transfer of learning in
industrial settings. Management Science 36, 140154.
Arvanitis, S., Hollenstein, H., 2001. The determinants of the adoption of advanced
manufacturing technology. An empirical analysis based on rm-level data
for Swiss manufacturing. Economics of Innovation and New Technology 10,
377414.
Arvanitis, S., Louikis, E., 2009. A comparative study of the effect of the ICT, organizational and human capital on labour productivity in Greece and Switzerland.
Information Economics and Policy 21, 4361.
Astebro, T., 2002. Noncapital investment costs and the adoption of CAD and CNC in
U.S. metalworking industries. Rand Journal of Economics 33 (4), 672688.
Astebro, T., Colombo, M.G., Seri, R., 2005. The Diffusion of Complementary
Technologies: An Empirical Test (downloaded on 8 November 2009 from
http://ssrn.com/abstract=690981).
Balakrishnan, S., Fox, I., 1993. Asset specicity, rm heterogeneity and capital structure. Strategic Management Journal 14, 316.
Baldwin, J., Diverty, B., 1995. Advanced Technology Use in Canadian Manufacturing Establishments. Research Paper, 85. Analytical Studies Branch.
Statistics Canada, Ottawa (downloaded on 8 November 2009 from
http://publications.gc.ca/collections/Collection/CS11-0019-85E.pdf).
Baptista, R., 2000. Do innovations diffuse faster within geographical clusters? International Journal of Industrial Organisation 18, 515535.
Barbosa, N., Faria, A.P., 2008. Technology adoption: does labour skill matter? Evidence from Portuguese rm-level data. Empirica 35, 179194.
1618
Barney, J., 1991. Firm resources and sustained competitive advantage. Journal of
Management 17 (1), 99120.
Barney, J.B., 1986. Strategic factors markets: expectations, luck and business strategy. Management Science 32 (10), 12311241.
Bartoloni, E., Baussola, M., 2001. The determinants of technology adoption in Italian
manufacturing industries. Review of Industrial Organization 19, 305328.
Battisti, G., Canepa, A., Stoneman, P., 2009. E-Business usage across and within rms
in the UK: protability, externalities and policy. Research Policy 38 (1), 133
143.
Beaumont, N., Schroder, R., Sohal, A., 2002. Do foreign-owned rms manage
advanced manufacturing technology better? International Journal of Operations
and Production Management 22 (7), 759771.
Beede, D.N., Young, K.H., 1996. Patterns of advanced technology adoption and manufacturing performance: employment growth, labor productivity and employee
earnings. Research Paper: Economics and Statistic Administration, Ofce of Policy Development, Washington, DC.
Belderbos, R., Carree, M., Diederen, B., Lokshin, B., Veugelers, R., 2004. Heterogeneity
in R&D cooperation strategies. International Journal of Industrial Organization
22, 12371263.
Belderbos, R., Carree, M., Lokshin, B., 2006. Complementarity in R&D cooperation
strategies. Review of Industrial Organization 28, 401426.
Bernard, A.B., Jensen, J.B., 1999. Exceptional exporter performance: cause, effect, or
both. Journal of International Economics 47, 125.
Blois, K.J., 1985. Marketing strategies and the new manufacturing technologies. International, Journal of Operations and Production Management 6 (1),
3441.
Bosworth, D., 1996. Determinants of the use of advanced technologies. International
Journal of the Economics of Business 3 (3), 269293.
Buis, M., 2011. Stata tip 97: getting at s and s. The Stata Journal 11 (2), 13.
Canepa, A., Stoneman, P., 2005. Financing constraints in the inter rm diffusion of new process technologies. The Journal of Technology Transfer 30,
159169.
Cappellari, L., Jenkins, S.P., 2003. Multivariate probit regression using simulated
maximum likelihood. The Stata Journal 3 (3), 278294.
Cassiman, B., Veugelers, R., 1999. Make and buy in innovation strategies: evidence
from Belgian manufacturing rms. Research Policy 28 (1), 6380.
Castellaci, F., 2002. Technology gap and cumulative growth: models and outcomes.
International Review of Applied Economics 16 (3), 333346.
Clemons, E.K., Kimbrough, S., 1986. Information systems, telecommunications, and
their effect on industrial organization. In: Proceedings of the 7th International
Conference on Information Systems, San Diego, CA.
Clemons, E.K., Row, M.C., 1991. Sustaining IT advantage. The role of structural differences. MIS Quarterly 15 (3), 275292.
Cohen, B., 1975. Multinational Firms and Asian Exports. Yale Univ. Press, New Haven,
CT.
Cohen, W.M., Levinthal, D.A., 1990. Absorptive capacity: a new perspective on learning and innovation. Administrative Science Quarterly 35, 128152.
Collins, D.J., Montgomery, C.A., 1998. Creating Corporate Advantage Harvard Business Review 76 (3), 7083.
Colombo, M.G., Mosconi, R., 1995. Complementary and cumulative learning effects
in the early diffusion of multiple technologies. Journal of Industrial Economics
43, 1348.
Conner, K.R., Prahalad, C.K., 1996. A resource-based theory of the rm: knowledge
versus opportunism. Organizational Science 7 (5), 477501.
Denrell, J., Fang, C., Winter, S.G., 2003. The economics of strategic opportunity.
Strategic Management Journal 24, 977990.
Dierickx, I., Cool, K., 1989. Asset stock accumulation and sustainability of competitive
advantage. Management Science 35 (12), 15041511.
Doms, M., Dunne, T., Troske, K., 1997. Workers, wages and technology. The Quarterly
Journal of Economics 112 (1), 253290.
Dunne, T., Troske, K., 2005. Technology adoption and the skill mix of US manufacturing plants. Scott Journal of Political Economy 52 (3), 387405.
Eaton, J., Kortum, S., 2002. Technology, geography, and trade. Econometrica 70,
17411779.
Ennen, E., Richter, A., 2010. The whole is more than the sum of its parts- or is
it? A review of the empirical literature on complementarities in organizations.
Journal of Management 36 (1), 207233.
Faria, A., Fenn, P., Bruce, A., 2003. A count data model of technology adoption. Journal
of Technology Transfer 28, 6379.
Fuentelsaz, L., Gmez, J., Palomas, S., 2009. The effects of new technologies on
productivity: an intrarm diffusion-based assessment. Research Policy 38,
11721180.
Gmez, J., Vargas, P., 2009. The effect of nancial constraints, absorptive capacity and
complementarities on the adoption of multiple process technologies. Research
Policy 38 (1), 106119.
Greve, H., 2009. Bigger and safer: the diffusion of competitive advantage. Strategic
Management Journal 30 (1), 123.
Grifth, R., Harrison, R., Van Reenen, J., 2006. How special is the special relationship?
Using the impact of U.S. Spillovers on U.K. rms as a test of technology sourcing.
The American Economic Review 96 (5), 18591875.
Griliches, Z., 1981. Market value, R&D and patents. Economic Letters 7, 183187.
Griliches, Z., Hall, B.H., Pakes, A., 1981. R&D, patents and market value revisited: is
there a second technological opportunity factor? Economics of Innovation and
New Technology 1, 183202.
Grossman, G., Helpman, E., 1991. Innovation and Growth in the World Economy.
MIT Press, Cambridge, MA.
Hannan, T.H., McDowell, J.M., 1987. Rival precedence and the dynamics of technology adoption: an empirical analysis. Economica 54, 155171.
Hill, C.W.L., 1988. Differentiation versus low cost or differentiation and low cost: a
contingency framework. Academy of Management Review 13, 401412.
Hirchey, M., Weygandt, J.J., 1985. Amortization policy for advertising and
research and development expenditures. Journal of Accounting Research 23,
326335.
Huffman, W.E., 1974. Decision making: the role of education. American Journal of
Agricultural Economics 56 (1), 8597.
Karshenas, M., Stoneman, P., 1993. Rank, stock, order, and epidemic effects in the
diffusion of new process technologies: an empirical model. Rand Journal of
Economics 24 (4), 503528.
Keller, W., 2004. International technology diffusion. Journal of Economic Literature
42, 752782.
Koellinger, P., 2008. The relationship between technology, innovation, and rm performance empirical evidence form e-business in Europe. Research Policy 37,
13171328.
Kogut, B., Zander, U., 1992. Knowledge of the rm, combinative capabilities, and
replication of technology. Organization Science 3 (3), 383397.
Kotha, S., Swamidass, P.M., 2000. Strategy, advanced manufacturing technology
and performance: empirical evidence from U.S. manufacturing rms. Journal
of Operations Management 18 (2), 257277.
Lal, K., 1999. Determinants of the adoption of Information Technology: a case study
of electrical and electronic goods manufacturing rms in India. Research Policy
28, 667680.
Lal, K., 2002. E-business and manufacturing sector: a study of small and mediumsized enterprises in India. Research Policy 31, 11991211.
Luque, A., 2002. An option value approach to technology adoption in U.S. manufacturing: evidence from microdata. Economics of Innovation and New Technology
11 (6), 543568.
J.A., Rochina-Barrachina, M.E., Sanchis-Llopis, J.A., 2009. SelfMnez-Castillejo,

selection into exports: productivity and/or innovation? Applied Economics
Quarterly 55 (3), 219241.
Mata, F.J., Fuerst, W.L., Barney, J., 1995. Information technology and sustained competitive advantage: a resource based analysis. Management Information System
19 (4), 487505.
Milgrom, P., Roberts, J., 1990. The economics of modern manufacturing: technology,
strategy and organization. The American Economic Review 80 (3), 511528.
Milgrom, P., Roberts, J., 1995. Complementarities and t: strategy, structure, and
organizational change in manufacturing. Journal of Accounting and Economics
79 (2-3), 179208.
Oriani, R., Sobrero, M., 2003. A meta-analytic study of the relationship between
R&D investments and corporate value. In: Caldirini, M., Garrone, P., Sobrero, M.
(Eds.), Corporate Governance, Market Structure and Innovation. Edward Elgar,
Cheltenham, pp. 177199.
Parthasarthy, R., Sethi, S.P., 1992. The impact of exible automation on business
strategy and organizational structure. Academy of Management Review 17 (1),
86111.
Parthasarthy, R., Sethi, S.P., 1993. Relating strategy and structure to exible automation: a test of t and performance implications. Strategic Management Journal
14 (7), 529549.
Pennings, J.M., 1987. Technological innovations in manufacturing. In: Pennings,
J.M., Buitendam, A. (Eds.), New Technology as Organizational Innovation: The
Development and Diffusion of Microelectronics. Ballinger, Cambridge, MA, pp.
197216.
Piva, M., Santarelli, E., Vivarelli, M., 2005. The skill bias effect of technological and
organisational change: evidence and policy implications. Research Policy 34,
141157.
Porter, M.E., 1980. Competitive Strategy. The Free Press, New York.
Powell, T.C., Den-Micallef, A., 1997. Information technology as competitive advantage: the role of human, business, and technology resources. Strategic
Management Journal 18 (5), 375405.
Reinganum, J.F., 1981. Market structure and the diffusion of new technology. The
Bell Journal of Economics 12 (2), 618624.
Riedel, J., 1975. The nature and discriminants of export-oriented direct foreign investment in a developing country: a case study of Taiwan.
Weltwirtschaftliches Archiv 111, 505528.
Rivkin, J.W., 2000. Imitation complex strategies. Management Science 46, 824844.
Rogers, E.M., Shoemaker, F.F., 1971. Communication of Innovations. The Free Press,
New York.
Rose, N.L., Joskow, P.L., 1990. The diffusion of new technologies: evidence from the
electric utility industry. Rand Journal of Economics 21 (3), 354373.
Santamara, L., Nieto, M.J., Barge-Gil, A., 2009. Beyond formal R&D: taking advantage of other sources of innovation in low- and medium-technology industries.
Research Policy 38, 507517.
Santamaria, L., Rialp, J., 2007. La eleccin del socio en las cooperaciones tecnolgicas: una anlisis emprico. Cuadernos de Economia y Direccin de Empresas 31,
6796.
Sinha, R.V., Noble, S.H., 2008. The adoption of radical manufacturing technologies
and rm survival. Strategic Management Journal 29 (9), 943962.
Stoneman, P., 2001. Technological diffusion and the nancial environment, Working paper, no-01-3. The United Nations University,
Institute for New Technologies (downloaded on 1 of March 2012 from
http://www.intech.unu.edu/publications/eifc-tf-papers/eifc01-3.pdf).
Stoneman, P., Kwon, M.J., 1994. The diffusion of multiple process technologies. The
Economic Journal 104, 420431.

Teece, D., 1986. Proting from technological innovation: implications for integration, collaboration, licensing and public policy. Research Policy 15,
285305.
Teece, D., 2006. Reections on proting from innovation. Research Policy 35 (8),
11311146.
Teece, D.J., 2000. Managing Intellectual Capital: Organizational, Strategic, and Policy
Dimensions. Oxford University Press, Oxford.
Vicente-Lorente, J.D., 2000. Inversin en intangibles y creacin de valor
Economa Industrial 332, 109

en la industria manufacturera espanola.
123.
Vicente-Lorente, J.D., 2001. Specicity and opacity as resource-based determinants
of capital structure: evidence for Spanish manufacturing rms. Strategic Management Journal 22, 157177.
1619
Villalonga, B., 2004. Intangibles resources, Tobins q, and the sustainability of performance differences. Journal of Economic Behaviour and Organization 54 (2),
205230.
Wernerfelt, B., 1984. A resource based view of the rm. Strategic Management Journal 5 (2), 171180.
Winter, S., 1987. Knowledge and competence as strategic assets. In: Teece, D. (Ed.),
The Competitive Challenge. Center for Research in Management, Berkeley, CA,
pp. 159184.
Wozniak, G.D., 1984. The adoption of interrelated innovations: a human capital
approach. The Review of Economics and Statistics 66 (1), 7079.
Wuyts, S., Dutta, S., Stremersch, S., 2004. Portfolios of interrm agreements in
technology-intensive markets: consequences for innovation and protability.
Journal of Marketing 68, 88100.

Seminal Paper

Загружено:

Сведения о документе

Авторское право

Доступные форматы

Поделиться этим документом

Поделиться или встроить документ

Параметры публикации

Этот документ был вам полезен?

Это неприемлемый материал?

Авторское право:

Доступные форматы

Seminal Paper

Загружено:

Авторское право:

Доступные форматы

Research Policy 41 (2012) 16071619

Contents lists available at SciVerse ScienceDirect

Intangible resources and technology adoption in manufacturing rms

and Cool, 1989). In other words, research on the impact of IT on

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

that the use of any technology could also be explained by the

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

2.1. Technological resources and technology adoption

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

allow more frequent changes in the production line, which could

of complementarities between different advanced technologies

As mentioned by the two anonymous referees, the link between marketing

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

the lowest gures (31.5% and 27.1%) among in-rm technologies.

Coefcient statistically signicant at the 90% level.

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

of the other technologies. The data clearly shows an increasing

3.2. Main variables

The annual depreciation rate () was assumed to be 15% in

3.2.2. Human resources

3.2.3. Stock of marketing resources

and present annual advertising expenditures13 (A) with a constant

Even though there is no consensus in the literature on the

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

measured through a dummy that takes a value of one when the

3.3.5. Propensity to export

3.3.6. Industry-specic rank effects

Appendix A contains some descriptive statistics of the covariates.

3.4. Model and estimation strategy

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

interrelationships arising in adoption, although it is not able to

Similarly, Hypothesis 3 stated that rms with more marketing

et al. (2009) nd that not

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

Stock of technological resources

Stock of marketing resources

Coefcient statistically signicant at the 90% level.

Coefcient statistically signicant at the 99% level.

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

view of those that contend that the diffusion of new technologies

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

and Kimbrough, 1986; Clemons and Row, 1991). However, recent

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

J.A., Rochina-Barrachina, M.E., Sanchis-Llopis, J.A., 2009. SelfMnez-Castillejo,

J. Gmez, P. Vargas / Research Policy 41 (2012) 16071619

Economa Industrial 332, 109

Вам также может понравиться