Research Policy 41 (2012) 602–613

Contents lists available at SciVerse ScienceDirect

Research Policy
journal homepage: www.elsevier.com/locate/respol

Scope and patterns of innovation cooperation in Spanish service enterprises

Alexandre Trigo ∗ , Xavier Vence 1
ICEDE Research Group and Department of Applied Economics, Facultade de Ciencias Económicas e Empresariais, University of Santiago de Compostela, Av. Burgo das Nacións, s/n,
15782 Santiago de Compostela, Spain2

a r t i c l e i n f o a b s t r a c t

Article history: Examining 2148 innovating service firms from the Spanish Technological Innovation Panel 2004, this
Received 22 September 2010 paper utilizes Latent Class Analysis to appraise the scope of innovation cooperation in services in the
Received in revised form 9 June 2011 Spanish economy, in accordance with the growing weight of external information flows throughout
Accepted 20 October 2011
innovation processes. The empirical evidence indicates that the nature of the service activity affects both
Available online 21 November 2011
the partner chosen and the cooperation intensity. The results lead to the creation of a typology of coop-
eration composed of three broad profiles: service firms intensive in techno-scientific cooperation, intensive
JEL classification:
in interactions with clients and a profile with low intensity in cooperation, called lonely innovators. The
O39
L80
probability that a firm belongs to the latter profile is 59%, which makes it reasonable to affirm that inno-
vation cooperation is not a common practice in Spanish innovating service enterprises. Innovation output
Keywords: variables have been included in order to examine the relationship between patterns of cooperation and
Cooperation innovation performance. The findings also underline the co-existence of different cooperation patterns
Innovation within the same industry.
Patterns of innovation
© 2011 Elsevier B.V. All rights reserved.
Service sector
Source of information

1. Introduction and Faulkner, 1998; Tidd et al., 2001; Gomes-Casseres, 2003; Tether
and Tajar, 2008; Hipp, 2010). These agreements are strategically
Literature on innovation management and economics has high- important in open innovation processes since they enable organi-
lighted the increasing importance of external mechanisms of sations to access knowledge, technology and know-how dissipated
knowledge creation such as external contacts and collaborations among other economic actors (see Chesbrough, 2003; Chesbrough
with other companies or entities (Rothwell, 1992, 1994; Lundvall, et al., 2006). This evidence leads to the assumption that the connec-
1992; Gibbons et al., 1994; Oerlemans et al., 1998; Chesbrough, tion and interaction between individuals with heterogeneous skills
2003; Chesbrough et al., 2006; Vega-Jurado et al., 2009). Indeed, and different but complementary experiences represent a collec-
empirical studies have shown that enterprises rarely innovate in tive and distributed process of knowledge creation and innovation
isolation of the economic system (Christensen and Lundvall, 2004). (Hayek, 1945; Andersen et al., 2000; Coombs et al., 2003).
Lately, many authors have emphasized the weight of networks As far as industrial analyses are concerned, the literature on
and partnerships as a way to incorporate external knowledge for innovation economics has historically ignored the service sec-
innovation, leading to the development of new products or pro- tor until recent years, so that the existing vision of innovation
cesses in companies (Lundvall, 1985, 1988, 2007; Gerlach, 1992; has been built based on the study of R&D and innovation in
Freeman, 1994; Gulati, 1998; Tether, 2002; Johnson and Lundvall, manufacturing branches. However, nowadays, economic literature
2003; Gomes-Casseres, 2003; Powell and Grodal, 2005). Among all acknowledges the existence of innovative and cooperative perfor-
these types of collaboration, formal cooperation is considered a key mance in services (see, e.g., Miles, 2001; Gallouj, 2002a,b; European
mechanism to fortify the innovative capacity of many firms through Commission, 2004; Tether, 2002, 2005; Hipp, 2010). After two
a synergistic atmosphere of production and value creation (Child decades, studies on innovation have proved cooperation practices
to be a cross-cutting feature in this industry; so much so that the
well-known formal representation of service innovation designed
by Gallouj and Weinstein (1997) has been revised and expanded in
∗ Corresponding author. Tel.: +34 881811653/981 56 31 00x11653.
accordance with contemporary “networked society” (Vries, 2006).
E-mail addresses: [email protected] (A. Trigo), [email protected]
In addition to this acknowledgment, many authors have recog-
(X. Vence).
1
Tel.: +34 881811567/981 56 31 00x11567. nized the existence of a plurality of innovation patterns within the
2
www.usc.es/icede. service industry (Soete and Miozzo, 1989; Gallouj and Weinstein,

0048-7333/$ – see front matter © 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.respol.2011.10.006
 A. Trigo, X. Vence / Research Policy 41 (2012) 602–613 603

1997; Den Hertog and Bilderbeek, 1999; Evangelista, 2000; Tether Hertog and Bilderbeek, 1999; Evangelista, 2000; Tether and Hipp,
and Hipp, 2000; Miozzo and Soete, 2001; Sundbo and Gallouj, 2000; 2000; Sundbo and Gallouj, 2000; Hollenstein, 2003; Hipp and
Hollenstein, 2003; Hipp and Grupp, 2005; De Jong and Marsili, Grupp, 2005; Hipp and Herstatt, 2006; De Jong and Marsili, 2006;
2006; Miles, 2008; Tether and Tajar, 2008; Vence and Trigo, 2009; Hortelano and González-Moreno, 2007; Miles, 2008).
Trigo, 2009a,b; Hipp, 2010). In effect, recent studies have laid Soete and Miozzo (1989) could be considered as pioneers in
emphasis on the diversity of innovation patterns within certain describing networking patterns of innovation in service compa-
service sub-industry such as KIBS (Knowledge Intensive Business nies. The “network-based innovation” pattern underlined by these
Services) or the tourism sector for instance (see, e.g., Hjalager, 1997; authors was not included in Pavitt’s popular typology (1984). As
Baark, 2005; Leiponen, 2005; Freel, 2006; Sundbo et al., 2007). Hipp and Grupp (2005) emphasize, this gap could be due to the
To illustrate such diversity, most of the studies on innovation in absence of a distributive nature in most manufacturing activities,
services have pointed to KIBS as the leading sub-sector with regards the source of reference for the earliest typologies. Things change
to innovation and cooperation. This prominent performance is not when we are dealing with service activities. The “network-based
only a result of the high proportion of innovating firms, but also, by innovation” pattern is composed of, on the one hand, scale-
and large, a consequence of the high number of innovating firms intensive and physical network intensive sectors (transport and
engaged in the majority of innovation activities and innovation wholesale trade), and on the other hand, information-intensive
cooperation. In contrast, other services such as distributive services networks sectors (communication, finance and insurance services).
(transport, wholesale, retail, etc.) and HORECA (hotels, restaurants Gallouj and Gallouj (2000, 30), however, note that the interactive
and catering) present a very low innovative and cooperative per- nature via networks should be considered “not so much one of
formance. the types within the taxonomy but rather a characteristic that is
Therefore, due to the increasing relevance of external informa- transversal, a trait of several if not all, types”. They further note that
tion flows to the firm’s innovative competence, in addition to the even those services classified as “dominated by suppliers” could
growing role of services in all modern economies, this paper has hold this feature. Another criticism of the contribution of Soete
two principal purposes: to appraise the real scope of innovation and Miozzo concerns the analytical perspective of technological
cooperation in Spanish service enterprises, and to examine the rela- trajectories used by them, and the absence of non-technological
tionship between cooperation and innovation performance. The innovations (Gallouj and Gallouj, 2000).
range of indicators investigated includes three sets of variables: A similar taxonomy laid out by Evangelista (2000) based on
(1) formal cooperation, (2) the significance of external information 19,000 service companies with more than twenty employees,
sources (3) as well as the types of innovation outputs. The first which were included in the 1993–1995 Italian Innovation Survey.
two sets aim to measure the level of openness of innovation pro- Factor analysis was applied to an array of aspects related to the
cesses and the importance of external knowledge. While formal innovation process, including the importance of science and tech-
cooperation represents a specific mode of collaboration, sources of nology based interactions and other information sources, in order
information permit us to estimate the value of extramural flows of to identify innovation profiles in services. However, two critical
knowledge including informal collaboration. Variables of innova- weaknesses in the study can be pointed out. First, a specific variable
tion outputs have been included in order to study the correlation to measure formal innovation cooperation is missing; and, second,
between the cooperative behaviour and innovation performance. the industrial classification is too broad, making use of highly aggre-
The empirical analysis is derived from the Spanish Technolog- gate data and preventing the identification of specific innovation
ical Innovation Panel (PITEC) carried out in 2004. PITEC, which is patterns within each sub-sector.
part of the Spanish CIS data, is a statistical panel whose sample is Hollenstein (2003) presents a five-profile innovation typol-
composed of data from the Central Business Directory (DIRCE) and ogy through cluster analysis using 2731 service firms from the
the Research Business Directory (DIRID). This paper is structured as Swiss Innovation Survey 1999. With regard to the identification of
follows: A literature review on the scope of cooperation-oriented cooperation-oriented profiles, Hollenstein includes a large range
patterns in services is tackled in Section 2. A descriptive analysis of of indicators, among them knowledge sources and R&D network
cooperation in services is the core of Section 3. In Section 4, the mul- variables. Unlike the aforementioned authors, this set of indicators
tivariate technique known as Latent Class Analysis (LCA) is used in embraces also non-technological aspects of innovation. It allows
order to create patterns of innovation cooperation in services. The for the creation of an effective and useful innovation taxonomy for
main conclusion and implication are discussed in Section 5. services as far as networks and others types of external links are
concerned. In contrast to the preceding typologies, the coopera-
tion aspect has been explored as a transversal feature of innovation
2. Cooperation-oriented patterns in existing typologies of instead of a specific attribute found in just one or few service
innovation in services branches, as suggested by Gallouj and Gallouj (2000). Compared to
Evangelista’s taxonomy, Hollenstein (2003) adds a new dimension
The literature on innovation in services has increased signif- to innovation labeled “market-oriented incremental innovators
icantly since the late 1980s. Within this emerging field, the study with weak external links”, which is composed of service activities
on patterns at the intra-sectoral level has received special attention such as business services and wholesale trade.
in recent years. Soete and Miozzo (1989) created a typology, based Hipp and Grupp (2005) carry out an interesting empirical analy-
on Pavitt’s taxonomy (Pavitt, 1984), to improve the understanding sis of innovation patterns in services. The results of this remarkable
of the diversity of innovation in services. Their typology permits study suggest that the patterns of innovation in services depend to
the identification of different patterns of innovation, beyond the a lesser degree on the sectoral classification, since each pattern can
supplier-dominated character stressed by Pavitt in his original be found in each service branch studied. This evidence strengthens
categorization (Pavitt, 1984, see also Bell and Pavitt, 1993). The the conclusions stated by the aforementioned Sundbo and Gallouj
classification created by these scholars has undoubtedly become (2000) and Hollenstein (2003). Concerning cooperation-related
an imperative contribution toward an enhanced perception of patterns, Hipp and Grupp (2005) identify a profile of network-based
innovation, incorporating more deeply the service sector in the services (information networks) composed mainly of banks and
mainstream economic research of the discipline. This categoriza- insurance companies. Technical services and R&D as well as soft-
tion of innovation in services has inspired many other studies ware services are classified as knowledge intensive. The authors
and has become the earliest reference in this field (see, e.g., Den also claim that most of the innovation typologies are suitable to
604 A. Trigo, X. Vence / Research Policy 41 (2012) 602–613

services showing a classical innovation structure. Innovation pat- order to better understand the relationship between patterns of
terns displayed in earlier studies do not embrace the whole range of cooperation and performance in innovation.
innovative services, thus new patterns must be identified by using
alternative concepts and measuring instruments. The authors sug- 3. Empirical evidence: innovation cooperation in services
gest that further theoretical and empirical studies on innovation in Spain
need to analyse manufacturing and services together, taking into
consideration their “products” instead of the sector they belong to. 3.1. Data and sample
Following Hipp and Grupp’s proposal, Tether and Tajar (2008)
elaborate a typology of innovation for all manufacturing and ser- The data used in the empirical analysis is derived from the Tech-
vices sectors, using data from 2500 European firms included in the nological Innovation Panel (PITEC) carried out in Spain in 2004,
Innobarometer 2002. This analysis enforces the synthesis approach which is a database developed by the INE (National Institute for
built by several scholars combining different innovation trajecto- Statistics), FECYT (Spanish Foundation for Science and Technology)
ries and a large set of economic activities (Gallouj and Weinstein, and COTEC (Foundation for Technological Innovation). This panel,
1997; Coombs and Miles, 2000; Hollenstein, 2003; Drejer, 2004; which is available since 2003, is an important statistical tool to
Hipp and Grupp, 2005; Tiri et al., 2006; Leiponen and Drejer, 2007; analyse the innovative performance of Spanish enterprises. The
Castellacci, 2008; Peneder, 2010). data provided by PITEC derives from the Central Business Direc-
In order to classify enterprises in terms of their innovative fea- tory (DIRCE) and the Research Business Directory (DIRID). The panel
tures, Tether and Tajar (2008) select three specific issues from survey follows the Oslo Manual methodology applied in the Com-
the survey: the firm’s orientation to innovation; the main sources munity Innovation Survey as reference to the selection of variables
of advanced technologies; and the firm’s perceived strengths in and indicators (see OECD, 2005).
innovation. Applying different statistical methods, these authors The sample of the 2003 panel was composed of two different
suggest three profiles of innovation: a product-research (PR) groups of enterprises: a sample of firms with 200 or more employ-
mode; a process technologies (PT) mode; and an organisational- ees and a sample of firms with intramural R&D expenditures. The
cooperation (OC) mode. While the first profile is composed mainly second edition of the panel, referring to 2004, sought to cover a
of medium–high-technology manufacturing and high-technology major limitation found in the first edition, incorporating firms with
activities, the second one is dominated by low and medium-low fewer than 200 employees without innovation activities, as well as
tech manufacturing. Services, especially distribution and trade firms with fewer than 200 employees whose research activities and
activities, are confined to the latter one. The PR mode tends to development are outsourced.
cooperate with universities or R&D specialists, while the OC mode The number of service enterprises presented in PITEC 2004 is
is more likely to engage in cooperative practices with suppli- 3546, among them 2148 are innovating ones. We consider an inno-
ers, customers and trade associations (supply-chain-cooperation vating firm as a firm which has implemented any type of innovation
rather than research-based cooperative practices). Furthermore, during the last two years, as recommended in the Oslo Manual
the innovation activities carried out by these firms are basi- (OECD, 2005). This concept includes, apart from technological inno-
cally oriented at organisational changes. However, although Tether vation, organisational and marketing changes. Table 1 shows the
and Tajar’s taxonomy covers an extensive range of economic
branches and European-wide countries, the aggregated classifica-
Table 1
tion at the industrial level used to display the results, particularly Distribution of service firms and innovating service firms, Spain.
for services, may conceal internal innovation and cooperation
diversity. Total firms Innovating firms %

All these contributions have provided substantial progress Repair of motor vehicles, 80 33 41%
toward a better understanding of innovation in services. However, motorcycles
Wholesale 391 224 57%
we observe two critical constraints: First, some of these studies
Retail trade 216 92 43%
have not given sufficient emphasis on the interactive dimension Hotels and restaurants 207 74 36%
through cooperation activities. Modern innovation practices bring Transport, storage and 130 65 50%
to light the importance of interaction and communication among communication
economic agents, accentuating the increasing significance of exter- Supporting transport activities; 114 57 50%
travel agencies
nal information flows in the enhancement of internal innovation Post and courier activities 17 10 59%
capacity, especially in some service activities. For this reason, we Telecommunications 67 55 82%
consider cooperation not only an important aspect but also a key Financial intermediation 217 177 82%
determinant to characterize innovation patterns at the industrial Real estate activities 61 27 44%
Renting of machinery and 32 15 47%
level, and specifically in certain service branches. The second criti-
equipment
cal constraint refers to the industrial classification. The high degree Software consultancy and 382 324 85%
of aggregation of service activities predominant in most of these supply
studies might hamper an in-depth typology for innovation. The Computer and related activities 124 105 85%
micro-data used in our empirical analysis, however, permits to Research and development 166 147 89%
Architectural and engineering 260 210 81%
classify services in 20 sub-sectors, allowing us to delimit more accu-
activities
rately the real scope of diversity of innovation cooperation patterns Technical testing and analysis 90 65 72%
in services. Other business activities 602 256 43%
Thus, the analysis of innovation and the typology proposed in Motion picture and video 28 10 36%
activities
this paper will focus essentially on the cooperation performance. As
Radio and television activities 36 27 75%
mentioned in the previous section, we have also deemed it relevant Health and social work, 326 175 54%
to include two sets of variables apart from cooperation indicators. sanitation and similar
Since the cooperation indicators refer only to the type of part- activities
ner, we have added a set of Likert scale variables on sources of Total 3.546 2.148 61%

information in order to measure the significance of each partner- Source: Trigo (2009a).
ship. Furthermore, innovation output variables are incorporated in Data from the Technological Innovation Panel (PITEC), 2004.
 A. Trigo, X. Vence / Research Policy 41 (2012) 602–613 605

Fig. 1. Proportion of innovating firms and proportion of innovating firms with cooperation, services, Spain (%).

distribution of service firms in PITEC 2004 database, where the dif- to the total of innovating firms. The number of cooperative firms
ferent innovation performances in services are displayed. While among the sample of innovating ones is only 36%. This chart also
KIBS, telecommunication and financial intermediation are in the tackles the innovation performance – the vertical axis quantifies the
forefront of innovation in services, HORECA (hotels, restaurants and proportion of innovating firms to the total of firms. The compari-
catering) and distributive services are not leading innovators. This son between these two proportions brings to light the existence
heterogeneous context reinforces the relevance of intra-industrial of two different groups of services with different propensities to
analysis of innovation patterns, given the existence of different cooperate and innovate. The diagonal display of these two groups
innovation drivers in each service branch. supports the notion that the correlation between innovation and
cooperation is directly linked.
3.2. Descriptive analysis of innovation cooperation in services Almost all highly innovative industries show a high propensity
from PITEC to cooperate. On the other hand, most branches with a low pro-
portion of innovating firms exhibit a low propensity to cooperate.
Among the different types of collaborations,3 this analysis This evidence enforces the hypothesis that the higher the capacity
focuses on the specific form of partnership commonly known as for- of interaction with other economic actors, the higher the innova-
mal cooperation in joint innovation projects between the firm and tion capacity of an enterprise will be. This result leads us to classify
other actors such as customers, suppliers, competitors, universities, service activities into two groups: one composed of those branches
technology institutes, public research institutions or government. with a high proportion of innovating firms with a high propen-
This mode of linkage must involve active cooperation on the part- sity to cooperate (Group 1). This group is constituted essentially
ners’ part including purchases of knowledge and technology as by knowledge-intensive business services, software and telecom-
described by the Oslo Manual (OECD, 2005). Therefore, the notion munications. Another group is composed of activities with a low
of cooperation adopted here implies formal relationships such as tendency to innovate and cooperate (Group 2). The branches of
strategic marketing alliances and joint development of new tech- retail trade, hotels and restaurants, motion picture and video activ-
nologies, products or processes. ities are some of the branches that comprise this class. The number
of innovating firms in these service branches is relatively low, and
few of them cooperate. This evidence is also supported by OECD
3.2.1. The scope of cooperation and innovation in services
reports, where the low propensity to cooperate of distributive ser-
As the first attempt to analyse the cooperative behaviour in
vices is highlighted (OECD, 2001).
services, the frequencies of cooperative firms by service branch
are calculated and the result is shown in Fig. 1 – the horizontal
axis measures the proportion of innovating firms with cooperation 3.2.2. Patterns of partnerships
The cooperation diversity in services is also manifested in the
choice of partners. Taking into consideration the service sector as
3
E.g., subcontract/supplier relations, licensing, consortia, strategic alliance, joint a whole, clients and universities are the most frequent partners to
venture, network (Tidd et al., 2001). services. In some branches, suppliers and consulting services are
606 A. Trigo, X. Vence / Research Policy 41 (2012) 602–613

Table 2
Proportion of firms with cooperation, by partners, services, Spain (%).

COOP COOP1 COOP2 COOP3 COOP4 COOP5 COOP6 COOP7 COOP8

Repair of motor vehicles, motorcycles 21 15 9 3 3 12 3 0 3

Wholesale 27 7 10 3 2 4 8 4 6
Retail trade 13 7 2 0 1 5 3 0 1
Hotels and restaurants 4 0 3 0 0 3 0 1 1
Transport, storage and communication 32 5 23 0 3 2 8 0 5
Supporting transport activities; travel agencies 39 11 26 9 9 2 5 4 0
Post and courier activities 10 0 10 0 0 10 0 10 0
Telecommunications 51 13 27 22 5 5 33 4 9
Financial intermediation 41 14 24 5 15 12 4 2 3
Real estate activities 15 4 4 4 4 4 4 0 7
Renting of machinery and equipment 7 0 7 0 0 7 0 0 0
Software consultancy and supply 40 6 9 16 10 6 17 7 13
Computer and related activities 30 7 8 10 6 4 12 5 8
Research and development 72 13 26 44 23 34 59 37 49
Architectural and engineering activities 52 11 14 21 12 10 27 12 22
Technical testing and analysis 42 3 14 22 8 14 28 11 15
Other business activities 25 6 13 7 4 7 6 3 5
Motion picture and video activities 10 0 0 0 10 0 0 0 0
Radio and television activities 37 19 33 4 0 7 11 0 4
Health and social work, sanitation and similar activities 33 7 13 9 6 9 14 7 6

Total 36 8 14 12 8 9 15 7 11

Source: Trigo (2009a).

Data from the Technological Innovation Panel (PITEC), 2004.
COOP = any kind of cooperation; COOP1 = other enterprise within the group; COOP2 = clients; COOP3 = suppliers; COOP4 = competitors or other firms; COOP5 = consultants,
commercial laboratories/R&D; COOP6 = universities; COOP7 = government or public research institutes; COOP8 = technology institutes.

clearly significant too. More specifically, firms whose productive tendency to cooperate with clients. However, the results suggest
and innovative activities are related to technological progress tend that firms that cooperate with actors closer to the productive
to cooperate with suppliers, universities and technology institutes system such as customers, consultants, etc., tend to develop
to a greater extent than those in any other enterprises (see also new-to-the-market innovation to a lesser extent (Kaufmann and
Tether and Hipp, 2000; Miles, 2002; Hollenstein, 2003; Hipp and Tödtling, 2001; Tödtling et al., 2009). The type of services, which
Grupp, 2005; Salter and Tether, 2006; Tether and Tajar, 2008; Vence present high propensity to cooperate with consultants, commer-
and Trigo, 2009, 2010; Trigo, 2009a). Moreover, their innovations cial labs, or private R&D institutes (labeled as “coop 5” in Table 2)
(product innovations) tend to be breakthroughs, not only new to compared to the propensity to cooperate with other agents, belong
the firm but also to the market. In this sense, the types of inter- to a set of activities with low innovative capacity, and at the same
actions along the innovation process would affect the scope of the time with reduced inclination to cooperate.
novelty of the innovations in some way. In fact, universities play a
special role for highly innovative industries with a high propensity 4. Patterns of innovation cooperation in services
to engage in R&D such as KIBS and telecommunications. Cooper-
ation with suppliers seems to have great weight for the KIBS and 4.1. Methodology and data analysis
telecommunications sectors, meanwhile suppliers seems to be less
significant to the others. Although the study of the interaction dimension in service inno-
It is worth pointing out that the sectors with significant vation is not an original trend, the analysis proposed here is unique
relationships with science and technology-related actors are fun- in two aspects: the variables chosen and the method of analysis
damentally activities with a high propensity to engage in internal applied.
R&D and have high skilled human capital (see Vence and Trigo, As previously explained, the indicators selected are classified
2009; Trigo, 2009a). This finding highlights the importance of a into three sets. The first one refers to the partners for cooperation
strong absorptive capacity of external information and knowledge while the second one includes Likert scale variables on sources of
flows, suggested by Cohen and Levinthal (1990), and its relationship information. Whereas the former permits the identification of the
with the firm’s innovative potential. This underlines the importance nature of cooperative linkages, the latter allows for the measure-
of highly skilled human capital as the main connection between ment of their significance. These variables are essential to achieve
the company and these institutions of science and technology. The our objective of identifying different patterns of innovation coop-
notion of the abovementioned absorption capacity induces us to eration in services. Furthermore, a set of output-related variables
assume that innovation and cooperation embrace a positive feed- formed by the types of innovation developed by the firm was taken
back loop. In this sense, it is not only cooperation that fosters the into account. These indicators are keys to describe the innova-
firm’s innovative capability, but also the internal innovative effort tive performance of cooperative firms as well as to verify whether
spent by enterprises confers the necessary and fundamental knowl- different types of innovation rely on specific sorts of cooperation
edge and capacity to interact with others. This second statement is arrangements. The variable selection is therefore consistent with
even more apparent in medium and high-tech sub-sectors. this article’s aims of analysing the real scope of cooperation on
Cooperation with customers is an important part of the total innovation in services in the Spanish economy, as well as that of
weight of the partnerships, mainly to activities such as support- examining the relationship between patterns of cooperation and
ing transport activities, travel agencies, financial intermediation, innovation performance.
other business activities as well as radio and television activ- As far as methodology is concerned, we use Latent Class Analysis
ities. In reality, these companies, plus telecommunications and (LCA) as an alternative to traditional Cluster Analysis since the for-
research and development firms, are the ones with the highest mer provides more accurate results for binary-type variables. LCA
 A. Trigo, X. Vence / Research Policy 41 (2012) 602–613 607

Table 3
Variables used in the Latent Class Analysis.

Type

Indicators
innobien New or significantly improved goods Binary
innoserv New or significantly improved services Binary
innfabri Implementation of new methods of production Binary
innlogis Implementation of new logistic system Binary
innapoyo Implementation of new supporting activities Binary
inorg1 Implementation of advanced management techniques within your enterprise Binary
inorg2 Implementation of major changes to your organisational structure Binary
inorg3 Changes in the relationship with other enterprises or public institutions Binary
incom1 Changes in the good/service design Binary
incom2 Implementation of new sales methods or delivery Binary
coopera Cooperation: all types Binary
coop11 Cooperation: other enterprise within the group Binary
copp21 Cooperation: clients Binary
coop31 Cooperation: suppliers Binary
coop41 Cooperation: competitors Binary
coop51 Cooperation: consultants, commercial laboratories/R&D Binary
coop61 Cooperation: universities Binary
coop71 Cooperation: government or public research institutes Binary
coop81 Cooperation: technology institutes Binary
infonew1 Significance of the information source: within the group Likert 0–3
infonew2 Significance of the information source: suppliers Likert 0–3
infonew3 Significance of the information source: clients Likert 0–3
infonew4 Significance of the information source: competitors Likert 0–3
infonew5 Significance of the information source: consultants, commercial laboratories/R&D Likert 0–3
infonew6 Significance of the information source: universities Likert 0–3
infonew7 Significance of the information source: public research institutes Likert 0–3
infonew8 Significance of the information source: technology institutes Likert 0–3
infonew9 Significance of the information source: conferences and fairs Likert 0–3
infonew10 Significance of the information source: scientific journals and other publication Likert 0–3
infonew11 Significance of the information source: professional associations Likert 0–3
Covariant
Service branches

Source: Trigo (2009a).

Table 4
Statistic results of Latent Class Analysis using variables related to cooperation, sources of information and innovation outputs, services, Spain.

LL BIC (LL) Npar L2 df p-Value Class.Err.

4-Cluster −36202.97 74405.4 268 59625.728 1470 1.5e−11450 0.0501

5-Cluster −35811.38 74159.3 340 58842.5487 1398 8.8e−11343 0.0525
6-Cluster −35450.65 73975.0 412 58121.0887 1326 1.5e−11248 0.0605
7-Cluster −35236.52 74083.9 484 57692.8251 1254 4.0e−11217 0.0816
8-Cluster −35009.88 74167.8 556 57239.5433 1182 4.8e−11181 0.0671

Source: Trigo (2009a).

Data from the Technological Innovation Panel (PITEC), 2004.

is a multivariate technique based on conditional probabilistic anal- The results of Latent Class Analysis display different solutions,
ysis. The objective of this statistical method is to verify whether the each one with different numbers of classes (clusters). The crite-
association between a set of observed categorical variables could rion for selecting the most accurate model that fit with the data set
be explained through a latent typology that is composed of differ- was the Bayesian Information Criterion (BIC) due to its consistency
ent classes.4 The variables used in each analysis are summarized in in comparison with other criteria, such as the Akaike Information
Table 3. Criterion (AIC). Most of the empirical analysis carried out through
This statistical technique has many advantages compared with Latent Class Analysis has chosen such statistical criterion for model
other tools. We can aver that one important advantage is the prob- selection. According to this criterion, the accurate model is the one
ability distribution of the clusters identified. The service branches with the lowest value for BIC (see Kashyap, 1977; Schwartz, 1978).
present different probabilities of belonging to each cluster identi- Therefore, following this principle, the precise model of coopera-
fied. One of the outputs brought about by this statistical technique, tion arrangements, sources of information and innovation output
therefore, is a classification by groups based on probabilities. consist of six clusters – each latent class (cluster) representing a
Another significant advantage is in determining the number of clus- different pattern of responses (see Table 4).
ters. This is due to the existence of rigorous statistical tests that
support the choice of the dimension of the model (choice of the
dimension regarding the best solution to data which means the 4.2. Service typology of cooperation behaviour based on
number of clusters in the model). Another important attribute is partnerships, sources of information and innovation outputs
the possibility of using categorical variables.
Table 5 summarizes the results provided by the Latent Class
Analysis, which can be grouped into three set of variables: inno-
4
Further technical information of Latent Class Analysis is provided in Appendix vation outputs, cooperation and source of information. While the
A. See also Heinen (1996) and Hagenaars and McCutcheon (2003). two former are expressed in percentage terms, the latter is the
608 A. Trigo, X. Vence / Research Policy 41 (2012) 602–613

Table 5
Cooperative and innovative performances by cluster, services, Spain.

Cluster 1 Cluster 2 Cluster 3 Cluster 4 Cluster 5 Cluster 6

Cooperation (%)a
Cooperation: all types 1% 19% 0% 100% 100% 100%
Other enterprise within the group 0% 6% 0% 21% 11% 35%
Clients 0% 6% 0% 43% 17% 58%
Suppliers 0% 2% 0% 54% 30% 18%
Competitors 0% 1% 0% 34% 15% 20%
Consultants, commercial laboratories/R&D 0% 3% 0% 40% 12% 22%
Universities 0% 0% 0% 68% 55% 14%
Government or public research institutes 0% 0% 0% 39% 21% 2%
Technology institutes 0% 0% 0% 58% 34% 5%

Source of information (Likert-scale 0–3)b

Within the group 2.43 2.00 2.41 2.72 2.41 2.44
Suppliers 1.47 1.11 1.79 1.91 1.17 1.99
Clients 1.50 0.64 1.79 2.31 1.59 1.43
Competitors 1.06 0.30 1.51 1.79 1.05 1.27
Consultants, commercial laboratories/R&D 0.51 0.18 1.39 1.88 1.02 1.08
Universities 0.12 0.06 1.32 2.06 1.43 0.45
Public research institutes 0.00 0.04 1.23 1.71 0.98 0.22
Technology institutes 0.02 0.03 1.23 1.96 1.10 0.23
Conferences and fairs 1.18 0.03 1.72 2.16 1.08 1.31
Scientific journals and other publication 1.24 0.03 1.61 2.18 1.16 1.27
Professional associations 0.64 0.00 1.37 1.87 0.62 0.91

Innovation outputs (%)a

Goods 46% 27% 43% 60% 53% 48%
Services 48% 33% 48% 79% 52% 67%
New methods of production 35% 17% 33% 60% 38% 45%
New logistic system 23% 10% 15% 21% 10% 37%
New supporting activities 54% 47% 48% 62% 32% 81%
Advanced management techniques 61% 44% 60% 84% 39% 76%
Major changes to the organisational structure 52% 37% 52% 78% 33% 66%
Relationship with other enterprises or public institutions 21% 15% 28% 59% 28% 41%
Changes in the good/service design 25% 7% 23% 26% 8% 36%
Implementation of new sales methods or delivery 29% 10% 17% 26% 13% 39%

Source: Adapted from Trigo (2009a).

Data from the Technological Innovation Panel (PITEC), 2004.
a
Each percentage shown can be understood as the probability of engagement on the selected item.
b
This is a Likert type scale from 0 to 3, where 0 = not at all relevant, 3 = highly relevant.

average of answers (Likert scale 0–3). Each percentage showed, in (similarity in the nature), but with some discrepancy in inten-
the case of the first two set of variables, can be understood as the sity (difference in the scope). The probability of cooperating with
probability of answering “yes” to the given item by a firm from universities, technology centres or with their suppliers is higher
cluster 1, cluster 2, cluster 3 and so forth. Although the LCA esti- than any other partner. This disparity in relation to the inten-
mation suggested the 6-clusters solution as the most accurate one sity is also reflected in the significance of information sources
in accordance to the BIC criterion of selection, we consider that and in the innovative intensity. It is interesting to note that the
a typology of innovation cooperation in Spanish services should cluster with the lower cooperative and innovative performance
be represented by 3 broad profiles, some of them composed of among them (with a lower probability of positive responses),
more than one clusters. Such aggregation has as the main decisive cluster 5, presents a higher propensity to product innovations
factor the nature of the partnerships (i.e., those with whom partner- than any other type (53% for innovative goods and services 52%).
ships the cooperation are developed) because of the relevance and Hence, the service firms belonging to cluster 5 are essentially
the objective proposed in this article. It is also worth mentioning techno-scientific product innovators. Indeed, the probability of
that the partnership composition presents a higher level of hetero- organisational innovations is the lowest among the others groups
geneity of answers than any other set of indicators applied in the (innovations related to knowledge management systems: 39%, and
sample. related to the work organisation: 33%). On the other hand, the more
The outcomes from this empirical analysis shows that the pro- cooperative and innovative cluster (with a higher probability of
files identified are – to a greater or lesser extent – present in all positive responses), cluster 4, presents a very high probability of
service sub-industry analysed. This evidence, also ensured by other organisational innovations, especially those related to knowledge
authors in recent publications (e.g., Tiri et al., 2006; Hortelano management systems (84%), although they have a high propen-
and González-Moreno, 2007; Leiponen and Drejer, 2007) denotes sity also for service innovations (79%). Thus, slightly different from
a criticism on the original premise supported by the techno- the cluster 5, service firms belonging to cluster 4 might be catego-
logical paradigms and technological trajectories, where different rized as techno-scientific service/organisational innovators. With
firms with the same technological regime holds similar innovation regard to the significance of information sources, the principal dif-
path. ference is in the magnitude of significances, not so much in nature
(differences in the choice of partners to cooperate). However, we
4.2.1. Intensive in techno-scientific flows of information should mention that only the most cooperative and innovative
According to the results, two of the six clusters (4 and 5) cluster (cluster 4) presents positive rates (above 1.5 – Likert scale
have an analogous profile, based on techno-scientific cooperation of 0–3).
 A. Trigo, X. Vence / Research Policy 41 (2012) 602–613 609

Fig. 2. The triplot representation of probability distribution regarding the three-profile typology of cooperation in service innovation, services, Spain (%).

The probability that an innovating service firm from the Span- approximately 48% of service firms belong to a profile similar to
ish sample belongs to this broad profile of firms intensive in that one (Hollenstein, 2003).5 This proportion is very high if we
techno-scientific flows of information (composed of cluster 4 and take into account the probability of just 13% found in our analysis.
5) is 28%. Although almost all branches have a certain probabil-
ity of belonging to these two clusters, the service activities with a
4.2.3. Lonely innovators
higher tendency to be intensive with regards to techno-scientific
The third identified profile is composed of three of the six LCA
flows of information are, “architectural and engineering activities”,
clusters (clusters 1, 2 and 3). The innovating firms that fit in this pro-
“telecommunications”, “research and development” and “technical
file show a very low probability of cooperating. This result leads us
testing and analysis”. Many other taxonomies on services iden-
to assume that those innovating firms are “lonely innovators” due
tify this profile, for instance in Hollenstein (2003) who names it
to the minor involvement with other economics actors, through
as Science-based high-tech firms with full network integration, and
formal cooperation, along their innovation processes. However,
product-research mode of innovation in Tether and Tajar (2008).
although those three clusters display very few tendencies to coop-
erate in innovation, the significance of information sources used
4.2.2. Intensive in interactions with clients is quite dissimilar. Two of the three “lonely innovator” clusters
Another profile of interactions identified from LCA makes (cluster 1 and 3), consider relatively important the information
known the role of the client as the main partner (cluster 6). There- that comes from customers and suppliers. These firms have low
fore, this cooperation behaviour is highly linked to the needs of propensity to cooperate formally but simultaneously they place
consumers and users of the service. The probability of an innovating high value on their relationships with actors along the value chain.
firm with this profile cooperating with clients is approximately 60% However, that interaction seems to be a result of business relation-
(highest of the entire set of clusters identified). In this sense, it is log- ships instead of common efforts to achieve innovation. Indeed, the
ical that the customer receives the highest rating among all possible significance of information sources is higher in cluster 3 than any
external sources (see Table 5). Client-led firms are very innovative, other cluster, except cluster 4 and 5.
especially for process and organisational innovation (e.g., probabil- With reference to innovation outputs, all lonely innovators
ity of 81% to be support processes innovators and 76% knowledge present similar tendencies: in general, these clusters are more
systems innovators). It is also worth noting the high probabil- likely to innovate in organisational aspects. However, there is a
ity of developing service innovations (67%). Among all analysed considerable difference in the propensity to innovate. Taking into
branches, “financial intermediation” is the service activities that account the three lonely innovators clusters, cluster 2 is the least
best fits to this profile. According to knowledge generation, inno- cooperative and innovative. Its probability of innovating in prod-
vating firms with client cooperation are intensive both in internal uct is the lowest in the whole model (27% for product innovation
R&D activities and training, as well as in acquisition of machin- and 33% for service innovation). On the other hand, clusters 1 and
ery and software (Trigo, 2009a,b). The probability of an innovating 3 are as inclined to be product innovators as cluster 5 (intensive
Spanish firm belonging to this profile is only 13%. Other taxonomies in techno-scientific flow of information), and even more prone in
based on innovation and networks highlight the high propensity terms of organisational innovation.
of cooperating with customers in some service activities. Those This is unquestionably the least advantageous of the three
typologies stress the high tendency of those firms to be also R&D- profiles identified given its weak cooperative and innovative per-
intensive. Client-led innovation (Den Hertog and Bilderbeek, 1999), formance. However, the probability that a Spanish firm belongs to
information network (Soete and Miozzo, 1989; Miozzo and Soete,
2001), network pattern (Sundbo and Gallouj, 2000), network basis
(Hipp and Grupp, 2005) and cost-oriented process innovators with
5
strong external links along the value chair (Hollenstein, 2003) are An important difference from the statistical method applied by Hollenstein
(2003) should be stressed. While LCA is a multivariate technique based on con-
some of the labels used by other authors to describe such a pro- ditional probabilistic analysis, the percentages of firms displayed in the Cluster
file. Some of the typologies found in the recent literature reveal the Analysis estimated by Hollenstein (2003) refer to the numbers of enterprises with
proportion of firms in each profiled identified. In the Swiss case, each profile.
610 A. Trigo, X. Vence / Research Policy 41 (2012) 602–613

Table 6
Service typology of cooperation patterns based on partnerships, sources of information and innovation outputs.

Highly interactive Lowly interactive

Intensive in techno-scientific flows of information Intensive in Lonely innovators

interactions with
clients (cluster 6)

With high cooperative With moderate With relatively high With low significance
and innovative cooperative and significance of of information sources
performance (cluster 4) innovative information sources (cluster 2)
performance (cluster 5) (cluster 1 and 3)

Principal partner in Universities, Universities, Clients, other – –

formal cooperation technology centres and technology centres and enterprise within the
suppliers suppliers group
Innovative intensitya Highly intensive Relatively intensive Highly intensive Relatively intensive Poorly intensive

Most common Service and Product (goods and Process and Organisational and Organisational
innovation organisational services) organisational product (goods and
services)
Similar typologies Science-based high-tech Product-research mode Client-led innovation Market-oriented Low profile innovators
firms with full network of innovation (Tether (Den Hertog and incremental innovators with hardly any
integration and Tajar, 2008) Bilderbeek, 1999) with weak external links external link
(Hollenstein, 2003) Information network (Hollenstein, 2003) (Hollenstein, 2003)
Product-Research mode (Soete and Miozzo,
of innovation (Tether 1989; Miozzo and
and Tajar, 2008) Soete, 2001)
Network patten
(Sundbo and Gallouj,
2000)
Cost-oriented process
innovators with strong
external links along the
value chair
(Hollenstein, 2003)
Network basis (Hipp
and Grupp, 2005)
Typical core sectors Research and Telecommunications; Financial Wholesale; Repair of motor
development; Architectural and intermediation; Retail trade; vehicles, motorcycles;
Technical testing and engineering activities. Supporting transport Post and courier Transport, storage and
analysis. activities; travel activities; communication;
agencies; Real estate activities; Health and social work,
Radio and television Software consultancy sanitation and similar.
activities. and supply;
Computer and related
activities;
Motion picture and
video activities;
Other business
activities;
Renting of machinery
and equipment;
Hotels and restaurants.

Source: Trigo (2009a).

a
Innovative intensity refers to the probability that a firm innovates (by type of innovation).

this profile is approximately 59%. This evidence is even worse when hardly any external link. In the Swiss case, the proportion of innovat-
we take into account the employment of economic activities in ing service firms with both profiles is fairly lower than the Spanish
Spain. Almost 65% of the employment in Spain is assigned to activi- sample (approximately 43%). This contrast makes evident the insuf-
ties with high probability of having this profile. As noted by Molero ficiency of a cooperative mind-set in Spanish firms compared to the
(2006), the growth of technology-intensive sectors has been sig- international context.
nificantly lower in Spain compared to other European economies The Fig. 2 shows a Triplot6 representation of this typology.7 The
(see also Arancegui, 2002). In other words, the specialization of the closer an element is to an apex of the triangle, the greater the prob-
Spanish economy has been concentrated mainly in areas with low ability, of this element having a specific profile. The Spanish service
or medium technological intensity. The branches that have a higher sector presents a general trend toward the “lonely innovator” pro-
probability of this profile are “retail trade”, “post and courier activ-
ities”, “repair of motor vehicles”, “hotels and restaurants” among
others. Indeed, most of those activities have a considerable partic- 6
We used the software for Microsoft Excel Triplot developed by David Graham
ipation in the Spanish economic structure because, among other
(Loughborough University) and Nicholas Midgley (Liverpool John Moores Univer-
reasons, “the existence of important weather and natural advan- sity), and distributed free of charge. See Graham and Midgley (2000) and additional
tages” (Gordo et al., 2006). Other authors have also highlighted the documentation in http://www.lboro.ac.uk/research/phys-geog/tri-plot/index.html.
7
existence of this profile in services. For example, in Hollenstein’ tax- The probability distribution has been recalculated taking into consideration the
three broad profiles of the typology proposed here, based on the 6-cluster LCA
onomy for Swiss services (2003), two of five groups have no leaning
model: it has been estimated the average of clusters 1, 2 and 3 for the profile “lonely
toward establishing external linkages: market-oriented incremental innovator” and clusters 4 and 5 for the profile “Intensive techno-scientific flows of
innovators with weak external links and low profile innovators with information” and recalculated of the weight of each profile in base 100.
 A. Trigo, X. Vence / Research Policy 41 (2012) 602–613 611

file for the Spanish service sector. Table 6 summarizes the main The results lead to the conclusion that the nature of the activities
aspects and characteristics of the taxonomy created here. The three affects both the nature and the intensity of the cooperation part-
profiles acknowledged in the analysis of innovation cooperation in nership in innovation. The nature of the activities referred to here
Spanish services are described regarding innovation intensity, the can be expressed by the technological capacity, in other words, the
most common innovation developed as well as the nuances about intensity of the use of technologies, the technological opportunities,
the intensity of linkages. Moreover, similar profiles found in the the growth in demand, the life cycle of the services, which describe
related literature and typical core sectors are provided. the evolution of the subsector, as well as the degree of standard-
ization or customization of the service activity. While knowledge
and technologically intensive business services demonstrate a very
5. Conclusions active behaviour as far as the linkages analysed are concerned,
innovating distributive services (transport, wholesale, retail, etc.)
This article aimed to appraise the scope of innovation cooper- and innovating HORECA (hotels, restaurants and catering) present
ation in Spanish services, in accordance with the growing weight a very low innovative performance. However, the results under-
of external knowledge exchange throughout innovation processes. line the co-existence of different cooperation pattern within the
We believe that cooperation is crucial for the differentiation of same industry. Although the existence of an increasing debate on
innovation patterns, especially to certain service branches. Inno- which one between the sectoral-determinism and strategic-choice
vation output variables have been included in order to examine is the most significant factor to shape cooperation and innovation
the relationship between cooperation behaviour and innovation patterns, the results lead to the conclusion that there exist clear
performance. association between the information flows used throughout the
Firm-level analysis carried out in this article has permitted the innovation processes and the nature of the activity. However, it is
in-depth scrutiny of cooperation patterns, beyond the boundary of important to stress that, in the same way that the decision that leads
the traditional standard industry classification largely applied so one firm to innovate is a strategic-choice, the inclination to use cer-
far. The multivariate statistical method applied, Latent Class Anal- tain information channel is also a choice in terms of managerial and
ysis, can be considered an original and innovative technique in innovation strategy.
identifying patterns of innovation in services. The analysis devel-
oped in this article lead to the conclusion that cooperation on
innovation is still not a very common practice in Spanish services.
Acknowledgements
Furthermore, the assumption of plurality has been confirmed, as
far as the nature and the dynamics of cooperation are concerned,
We are grateful to Prof. Stan Metcalfe, Prof. Ian Miles, Prof.
as demonstrated by other authors. This evidence shows the grow-
Roonie Ramlogan, Dr. Davide Consoli, Dr. Shu-Li Cheng, Prof.
ingly untenable hypothesis of a unique pattern of cooperation on
Marcela Miozzo and Dr. Yanuar Nugroho from the Manchester
innovation in services.
Institute of Innovation Research (University of Manchester, UK),
The findings of the empirical analysis led to the creation of a
for motivating and helpful discussions on the original version of
cooperation-oriented typology for service innovation composed of
this work, and also for providing the specific computer statistical
three broad profiles: intensive in techno-scientific flows of informa-
software for Latent Class Analysis. We acknowledge Prof. Ánxela
tion, intensive in interactions with clients and, finally, a profile of low
Troitiño and other members of the ICEDE Research Group from the
intensity in interactions, called lonely innovators. The innovating
University of Santiago de Compostela for interesting suggestions
firms intensive in techno-scientific flows of information are char-
on previous drafts of this work. We also appreciate the useful com-
acterized by the high probability of cooperating with agents such
ments made by Dr. Michele Mastroeni from INNOGEN (School of
as technology institutes, universities and suppliers. Lonely inno-
Social and Political Science, University of Edinburgh, UK). A pre-
vators, on the other hand, show low probabilities of carrying out
liminary version of this article has strongly benefited from the
any type of innovation project with other partners. The probabil-
valuable remarks provided by the participants in the DRUID Sum-
ity that a service firm innovates in isolation is 59%. Therefore, it
mer Conference 2009 and the XIXth International RESER 2009
seems to be reasonable to affirm that innovation cooperation is
Conference. The authors acknowledge the financial support from
scarcely performed by Spanish innovating service enterprises. The
the European Regional Development Fund (ERDF) and from Xunta
result is more understandable if one takes into account the com-
de Galicia (Competitive Reference Group 2008/041 and Project
position of the Spanish business sector and its passive profile as
08SEC008201PR). Finally, we would also like to express thanks to
far as innovation initiatives are concerned. This evidence supports
the editor and two anonymous referees for the constructive feed-
the assumption that formal cooperation in Spain is not as present in
back. The usual disclaimer applies.
the innovative dynamics of the companies compared to other coun-
tries, as indicated by the European Commission on many occasions.
The empirical evidence also confirms that the relationship
between cooperation behaviour and innovation performance is Appendix A. Latent Class Analysis
directly linked. In this sense, the higher the innovative level, the
higher cooperative level and vice versa. Indeed, innovation and Latent Class Analysis (LCA) is a statistical method that can be
cooperation embrace a positive feedback loop, which means that it applied to cluster, factor or regression analysis. Given the purpose
is not only cooperation that fosters the firm’s innovative capability. of this article, we will focus on the specific module of clustering. In
Empirical findings in literature have also proven that the internal this sense, LCA is a multivariate technique used to identify clusters
innovative effort spent by enterprises confers the necessary and of related cases (latent classes or clusters) based on the analysis
fundamental knowledge and capacity to interact with others. With of the probability distribution of observed categorical variables.
regard to the innovation performance of the three-profile typol- Its aim is to examine whether the association between categorical
ogy, firms intensive in techno-scientific flows of information tend variables observed can be explained from a structure or unobserved
to innovate mainly in product and, to a certain extent in organi- latent variable . The LCA cluster model for categorical variables
sational aspects. Client-led innovators, on the other hand, seem to is constructed from probabilistic modeling of the observed vari-
be more process innovators than any other profile. Finally, lonely ables {V1 , V2 , . . ., Vj , . . ., VJ } which is conditioned by a structure
innovators are basically organisational innovators.  of latent classes. Under the assumption of local independence
612 A. Trigo, X. Vence / Research Policy 41 (2012) 602–613

(independence between the variables Vj within the latent classes), Gordo, E., Jareño, J., Urtasun, A., 2006. Radiografía del sector de servicios en España.
the conditional probability function turns out to be: Banco de España, Documentos Ocasionales No 0607.
Graham, D.J., Midgley, N.G., 2000. Graphical representation of particle shape using
T J triangular diagrams: an excel spreadsheet method. Earth Surface Processes and
Landforms 25 (13), 1473–1477.
v v 
j
 
P = P(t ) P Gulati, R., 1998. Alliances and networks. Strategic Management Journal 19, 293–317.
 t Gallouj, F., Weinstein, O., 1997. Innovation in Services. Research Policy 26 (4–5),
t=1 j=1
537–556.
Hagenaars, J.A., McCutcheon, A.L. (Eds.), 2003. Applied Latent Class Analysis. Cam-
where
bridge University Press, Cambridge.
Hayek, F.A., 1945. The use of knowledge in society. American Economic Review XXXV
• v identifies any values {v1 , v2 , . . . , vj , . . . , vJ } of the vector of (4), 519–530.
Heinen, T., 1996. Latent Class and Discrete Latent Trait Models: Similarities and
observed variables V = {V1 , V2 , . . ., Vj , . . ., VJ }. It is the result of Differences. Sage Publications, Thousand Oaks, CA.
observing all the variables Vj , j = 1, 2, . . ., J, in any individual in the Hipp, C., Grupp, H., 2005. Innovation in the service sector: the demand for service-
population. It indicates a pattern of behaviour of each individual specific innovation measurement concepts and typologies. Research Policy 34,
517–535.
according to the vector of observed variables V. Hipp, C., Herstatt, C., 2006. Patterns of innovation and protection activities
• P( t ) is the a priori probability that the latent variable  takes a within service companies. Results from a German study on service-intensive
specific pattern  t . That is, it identifies the a priori probability that companies, Working Paper No. 45, Technologie- und Innovationsmanage-
ment/Technische Universität Hamburg-Harburg.
any individual belongs to latent class  t .
Hipp, C., 2010. Collaborative innovation in services. In: Gallouj, F., Djellal, F. (Eds.),
• P(vj /t ) is the probability that the categorical variable Vj take any The Handbook of Innovation and Services: A Multi-Disciplinary Perspective.
value vj when the latent variable takes the modality  t . Edward Elgar, United Kingdom, pp. 318–341.
Hjalager, A.M., 1997. Innovation patterns in sustainable tourism: an analytical typol-
ogy. Tourism Management 18 (1), 35–41.
References Hollenstein, H., 2003. Innovation modes in the Swiss service sector: a cluster analysis
based on firms-level data. Research Policy 32, 845–863.
Andersen, B., Metcalfe, J.S., Tether, B., 2000. Distributed innovation systems and Hortelano, M.D.E., González-Moreno, A., 2007. Innovation in service firms:
instituted economic process. In: Metcalfe, J.S., Miles, I. (Eds.), Innovation Systems exploratory analysis of innovation patterns. Management Research 5 (2),
in the Service Economy. Measurements and Case Study Analysis. Kluwer, Boston, 113–126.
pp. 15–42. Johnson, B., Lundvall, B.-Å., 2003. Promoting innovation systems as a response to
Arancegui, M.N., 2002. La cooperación para la innovación de la empresa española the globalizing learning economy. In: Cassiolato, J.E., Lastres, H.M.M., Maciel,
desde una perspectiva internacional comparada. Economía industrial 346, M.L. (Eds.), Systems of Innovation and Development. Edward Elgar Publishing,
47–66. Cheltenham, pp. 141–184.
Baark, E., 2005. New modes of learning in services: a study of Hong Kong’s consulting Kashyap, R., 1977. A Bayesian comparison of different classes of dynamic mod-
engineers. Industry and Innovation 2 (2), 283–301. els using empirical data. IEEE Transactions of Auto Control AC-22 (5),
Bell, M., Pavitt, K., 1993. Technological accumulation and industrial growth: con- 715–727.
trasts between developed and developing countries. Industrial and Corporate Kaufmann, A., Tödtling, F., 2001. Science–industry interaction in the process of inno-
Change 2 (5), 157–211. vation: the importance of boundary-crossing between systems. Research Policy
Castellacci, F., 2008. Technological paradigms, regimes and trajectories: Manu- 30, 791–804.
facturing and service industries in a new taxonomy of sectoral patterns of Leiponen, A., 2005. Organization of knowledge and innovation: the case of Finnish
innovation. Research Policy 37 (6–7), 978–994. business services. Industry and Innovation 12 (2), 185–203.
Chesbrough, H., 2003. Open Innovation: The New Imperative for Creating and Prof- Leiponen, A., Drejer, I., 2007. What exactly are technological regimes? Intra-industry
iting from Technology. Harvard Business School Press, Boston, MA. heterogeneity in the organization of innovation activities. Research Policy 36 (8),
Chesbrough, H., et al. (Eds.), 2006. Open Innovation: Researching a New Paradigm. 1221–1238.
Oxford University Press, Oxford. Lundvall, B.-Å., 1985. Product Innovation and User–Producer Interaction. Aalborg
Child, J., Faulkner, D., 1998. Strategies of Cooperation: Managing Alliances Networks University Press, Aalborg.
and Joint Ventures. Oxford University Press, Oxford, UK. Lundvall, B.-Å., 1988. Innovation as an interactive process: from user–producer
Christensen, J.L., Lundvall, B.-Å. (Eds.), 2004. Product Innovation, Interactive Learn- interaction to the national system of innovation. In: Dosi, G., Freeman, C., Nel-
ing and Economic Performance. Elsevier, Amsterdam. son, R., Silverberg, G., Soete, L. (Eds.), Technical Change and Economic Theory.
Cohen, W.M., Levinthal, D.A., 1990. Absorptive capacity: a new perspective on learn- Pinter Publishers, London, pp. 349–369.
ing and innovation. Administrative Science Quarterly 35 (1), 129–152. Lundvall, B.-Å., 2007. National innovation system: analytical focusing device and
Coombs, R., Miles, I., 2000. Innovation measurement and services: the new prob- policy learning tool, Working Paper R2007:004, ITPS – Swedish Institute for
lematique. In: Metcalfe, J.S., Miles, I. (Eds.), Innovation Systems in the Service Growth Policy Studies.
Economy. Measurements and Case Study Analysis. Kluwer, Boston, pp. 85–103. Lundvall, B.-Å. (Ed.), 1992. National System of Innovation: Towards a Theory of
Coombs, R., Harvey, M., Tether, B., 2003. Analysing distributed processes of provision Innovation and Interactive. Learning. Pinter Publishers, London.
and innovation. Industrial and Corporate Change 12 (6), 1125–1155. Miles, I., 2001. Services innovation: a reconfiguration of innovation studies, PREST
Den Hertog, P., Bilderbeek, R., 1999. Conceptualising Service Innovation and Service Discussion Paper No 01-05, Manchester United Kingdom.
Innovation Patterns, Research Programme on Innovation in Services (SIID) for Miles, I., 2002. Services innovation: towards a tertiarization of innovation
the Ministry of Economic Affairs. Dialogic, Utrecht. Studies. In: Gadrey, J., Gallouj, F. (Eds.), Productivity, Innovation and
De Jong, J.P.J., Marsili, O., 2006. The fruit flies of innovation: a taxonomy of innovative Knowledge in Services. Edward Elgar Publishing, Cheltenham, Northampton,
small firms. Research Policy 35 (2), 213–229. pp. 164–196.
Drejer, I., 2004. Identifying innovation in surveys of services: a Schumpeterian per- Miles, I., 2008. Patterns of innovation in service industries. IBM Systems Journal 47
spective. Research Policy 33 (3), 551–562. (1), 115–128.
European Commission, 2004. Innovation in services: inviting Cinderella to the inno- Miozzo, M., Soete, L., 2001. Internationalization of services: a technological perspec-
vation ball Innovation & Technology Transfer 4/04, Luxembourg. tive. Technological Forecasting and Social Change 67, 159–185.
Evangelista, R., 2000. Sectoral patterns of technological change in services. Eco- Molero, J., 2006. Informe crítico sobre la innovación en la economía española. Más
nomics of Innovation and New Technology 9 (3), 183–221. allá de la ‘caja negra’, Ponencia en el I Congreso Ibérico de la Innovación, Mérida
Freel, M., 2006. Patterns of technological innovation in knowledge-intensive busi- 21–23 noviembre 2006.
ness services. Industry and Innovation 13 (3), 335–358. OECD, 2001. Employment Outlook, Paris.
Freeman, C., 1994. The economics of technical change. Cambridge Journal of Eco- OECD, 2005. Oslo Manual—Proposed Guidelines for Collecting and Interpreting
nomics 18 (5), 463–514. Technological Innovation Data, Paris.
Gallouj, C., Gallouj, F., 2000. Neo-Schumpeterian perspective on innovation in ser- Oerlemans, L., Meeus, M., Boekema, F., 1998. Do networks matter for innovation?
vices. In: Boden, M., Miles, I. (Eds.), Services and the Knowledge-Based Economy. The usefulness of the economic network approach in analysing innovation. Tijd-
Continuum Books, London, pp. 21–37. schrift voor Economische en Sociale Geografie 89, 298–309.
Gallouj, F., 2002a. Innovation in the Service Economy: The New Wealth of Nations. Pavitt, K., 1984. Sectoral patterns of technical change: towards a taxonomy and a
Edward Elgar Publishing, Cheltenham. theory. Research Policy 13 (6), 343–373.
Gallouj, F., 2002b. Innovation in services and the attendant old and new myths. Peneder, M., 2010. Technological regimes and the variety of innovation behaviour:
Journal of Socio-Economics 31, 137–154. creating integrated taxonomies of firms and sectors. Research Policy 39,
Gerlach, M.L., 1992. Alliance Capitalism: The Social Organization of Japanese Busi- 323–334.
ness. University of California Press, Berkeley, CA/London. Powell, W.W., Grodal, S., 2005. Networks of innovators. In: Fagerberg, J., Mowery,
Gibbons, M., et al., 1994. The New Production of Knowledge: The Dynamics of Science D.C., Nelson, R.R. (Eds.), The Oxford Handbook of Innovation. Oxford University
and Research in Contemporary Societies. Sage, London. Press, Oxford, pp. 56–85.
Gomes-Casseres, B., 2003. Competitive advantage in alliance constellations. Strate- Rothwell, R., 1992. Successful industrial innovation—critical factors for the 1990s.
gic Organization 1 (3), 327–335. R&D Management 22 (3), 221–239.
 A. Trigo, X. Vence / Research Policy 41 (2012) 602–613 613

Rothwell, R., 1994. Towards the fifth-generation innovation process. International Tidd, J., Bessant, J., Pavitt, K., 2001. Managing Innovation: Integrating Technological,
Marketing Review 11 (1), 7–31. Market and Organizational Change, 2nd ed. John Wiley & Sons, Chichester.
Salter, A., Tether, B., 2006. Innovation in services: through the looking glass of inno- Tiri, M., Peeters, L., Swinnen, G., 2006. Are innovation patterns in manufacturing
vation studies. Background paper for Advanced Institute of Management (AIM) and services converging? A multivariate analysis of CIS-3 data for the Flem-
Research’s Grand Challenge on Service Science. ish business sector. Paper Presented at the DRUID Summer Conference 2006,
Schwartz, G., 1978. Estimating the dimension of a model. The Annals of Statistics 5 Copenhagen, Denmark, 18–20 June 2006.
(2), 461–464. Tödtling, F., et al., 2009. Do different types of innovation rely on specific kinds of
Soete, L., Miozzo, M., 1989. Trade and Development in Services: A Technological knowledge interactions? Technovation 29 (1), 59–71.
Perspective, Working Paper No 89-031, MERIT, Maastricht. Trigo, A., 2009a. La naturaleza y la dinámica de los flujos de información en la inno-
Sundbo, J., Gallouj, F., 2000. Innovation as a loosely coupled system in services. vación. Un análisis del sector servicios en España. Ph.D. Thesis. Universidad de
In: Metcalfe, J.S., Miles, I. (Eds.), Innovation Systems in the Service Economy. Santiago de Compostela.
Measurements and Case Study Analysis. Kluwer, Boston, pp. 43–68. Trigo, A., 2009b. Innovation patterns under the magnifying glass. Firm-level latent
Sundbo, J., Orfila-Sintes, F., Sørensen, F., 2007. The innovative behaviour of tourism class analysis of innovation activities in services, Manchester Business School
firms—comparative studies of Denmark and Spain. Research Policy 36 (1), Working Paper, 578.
88–106. Vega-Jurado, J., Gutiérrez-Gracia, A., Fernández-de-Lucio, I., 2009. Does external
Tether, B., Hipp, C., 2000. Competition and innovation amongst knowledge-intensive knowledge sourcing matter for innovation? Evidence from the Spanish man-
and other service firms: evidence from Germany. In: Andersen, B., Howells, J., ufacturing industry. Industrial and Corporate Change 18 (4), 637–670.
Hull, R., Miles, I., Roberts, J. (Eds.), Knowledge and innovation in the New Service Vence, X., Trigo, A., 2009. Diversity of innovation patterns in services. The Service
Economy. Edward Elgar Publishing, Cheltenham, pp. 49–67. Industries Journal 29 (12), 1635–1657.
Tether, B., Tajar, A., 2008. The organisational-cooperation mode of innovation and Vence, X., Trigo, A., 2010. Global and national cooperation in service inno-
its prominence amongst European service firms. Research Policy 37, 720–739. vation. In: Gallouj, F., Djellal, F. (Eds.), The Handbook of Innovation and
Tether, B., 2002. Who co-operates for innovation, and why. An empirical analysis. Services: A Multi-Disciplinary Perspective. Edward Elgar, United Kingdom,
Research Policy 31, 947–967. pp. 545–572.
Tether, B., 2005. Do services innovate (differently)? Insights from the European Vries, E.J., 2006. Innovation in services in networks of organizations and in the
innobarometer survey. Industry and Innovation 12 (2), 153–184. distribution of services. Research Policy 35 (7), 1037–1051.