You are on page 1of 15

Regional Studies

ISSN: 0034-3404 (Print) 1360-0591 (Online) Journal homepage: http://www.tandfonline.com/loi/cres20

Technological diversity in Europe: empirical


evidence from agri-food research projects

Alessandro Muscio & Andrea Ciffolilli

To cite this article: Alessandro Muscio & Andrea Ciffolilli (2017): Technological diversity
in Europe: empirical evidence from agri-food research projects, Regional Studies, DOI:
10.1080/00343404.2017.1301662

To link to this article: http://dx.doi.org/10.1080/00343404.2017.1301662

View supplementary material

Published online: 05 May 2017.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=cres20

Download by: [University of Florida] Date: 06 May 2017, At: 06:06


REGIONAL STUDIES, 2017
https://doi.org/10.1080/00343404.2017.1301662

Technological diversity in Europe: empirical evidence from


agri-food research projects
Alessandro Muscioa and Andrea Ciffolillib

ABSTRACT
Technological diversity in Europe: empirical evidence from agri-food research projects. Regional Studies. Regional
technological diversity is the capacity of regions to carry out research in multiple technology fields. Based on the
analysis of 730 competitive research projects funded by the 7th European Framework Programme in agri-food, we
show that the capacity to obtain funding and the regional degree of centrality in collaborative networks promote
technological diversity. European programmes for research tend to encourage specialization in peripheral territories and
technological diversity in central regions. This has challenging implications for a policy that wants to promote research
excellence and, simultaneously, to reduce the wide regional variation in research and innovation performance across the
European Union.
KEYWORDS
Framework Programme; technological diversity; smart specialization; European regions; agri-food industry

摘要
欧洲的技术多样性:农业食品研究项目的实证证据。Regional Studies. 区域技术多样性,是区域在多重技术场域进行
研究的能力。我们根据第七欧盟架构计画所资助的七百三十个竞争的农粮研究计画之分析显示,取得补助的能力与
在合作网络的中心性之区域层级,可促进技术多样性。欧盟的研究计画倾向鼓励边陲地域的专殊化,以及核心区域
的技术多样化。而这对企图促进卓越研究并同时减少欧盟内部研究与创新表现的大幅区域差异的政策而言,带来了
挑战性的意涵。
关键词
架构计画; 技术多样性; 智能专殊化; 欧洲区域; 农粮产业

RÉSUMÉ
Diversité technologique en Europe: données empiriques découlant de projets agroalimentaires. Regional Studies. La
diversité technologique régionale est la capacité des régions d’effectuer une recherche dans de multiples secteurs
technologiques. Sur la base de l’analyse de 730 projets de recherche compétitifs financés par le 7ème programme cadre
dans le secteur agroalimentaire, nous démontrons que la capacité d’obtenir un financement et le degré régional de
centralité dans les réseaux de collaboration encouragent la diversité technologique. Des programmes de recherche
européens ont tendance à encourager la spécialisation dans des territoires périphériques ainsi que la diversité dans des
régions centrales. Les implications sont difficiles pour une politique visant la promotion de l’excellence dans la recherche
conjointement avec le renforcement de l’innovation dans l’Union européenne.
MOTS-CLÉS
Programme cadre; diversité technologique; spécialisation intelligente; régions européennes; secteur agroalimentaire

CONTACT
a
(Corresponding author) alessandro.muscio@unifg.it
Dipartimento di Scienze Agrarie, degli Alimenti e dell’Ambiente, Università degli Studi di Foggia, Foggia, Italy.
b
ciffolilli@ismerieuropa.com
Ismeri Europa S.r.l., Rome, Italy.

© 2017 Regional Studies Association


2 Alessandro Muscio and Andrea Ciffolilli

ZUSAMMENFASSUNG
Technische Vielfalt in Europa: empirische Belege von Forschungsprojekten für Nahrungsmittelerzeugung. Regional Studies.
Bei der regionalen technischen Vielfalt handelt es sich um die Kapazität von Regionen zur Forschung in verschiedenen
technischen Bereichen. Anhand einer Analyse von 730 miteinander konkurrierenden Forschungsprojekten, die vom
7. Europäischen Rahmenprogramm für Nahrungsmittelerzeugung finanziert wurden, zeigen wir, dass die Kapazität für
den Erhalt einer Finanzierung und das regionale Ausmaß der Zentralität in kooperativen Netzwerken die technische
Vielfalt fördern. Europäische Forschungsprogramme fördern in der Regel eine Spezialisierung in peripheren Gebieten
und technische Vielfalt in zentralen Regionen. Dies ist mit problematischen Auswirkungen für eine Politik verbunden, die
erstklassige Forschung fördern und zugleich die starken regionalen Unterschiede bei der Leistung im Bereich der
Forschung und Innovation innerhalb der Europäischen Union verringern will.
SCHLÜSSELWÖRTER
Rahmenprogramm; technische Vielfalt; intelligente Spezialisierung; europäische Regionen; Nahrungsmittelbranche

RESUMEN
Diversidad tecnológica en Europa: evidencia empírica de proyectos de investigación agroalimentaria. Regional Studies. La
diversidad tecnológica regional es la capacidad de las regiones para llevar a cabo estudios en los campos tecnológicos
múltiples. A partir del análisis de 730 proyectos de investigación competitivos financiados por el VII programa marco
europeo para la industria agroalimentaria, mostramos que la capacidad para obtener financiación y el grado regional de
centralidad en redes de colaboración fomentan la diversidad tecnológica. Los programas europeos para la investigación
suelen estimular la especialización en territorios periféricos y la diversidad tecnológica en regiones centrales. Esto tiene
efectos adversos para una política que quiere fomentar la excelencia en investigación y a la vez reducir las fuertes
diferencias regionales en el campo de la investigación y la innovación en la Unión Europea.
PALABRAS CLAVES
programa marco; diversidad tecnológica; especialización inteligente; regiones europeas; industria agroalimentaria

JEL O18, O31, O38, R58


HISTORY Received 12 January 2016; in revised form 21 February 2017

INTRODUCTION across networks and member states (Breschi & Cusmano,


2004; Breschi, Cassi, Malerba, & Vonotars, 2009; Paier
In the programming period 2007–13, the European Com- & Scherngell, 2011; Pandza, Wilkins, & Alfoldi, 2011;
mission (EC) invested a significant amount of resources in Scherngell & Barber, 2011).
research and innovation. Considering only the main instru- Despite the growing empirical evidence on the scientific
ments such as the European Regional Development Fund and economic results of the FPs and on the extent and
(ERDF) and the 7th Framework Programme for Research characteristics of the promoted partnerships, there is still
and Technological Development (FP7), over €100 billion little evidence on the technological specialization and the
were devoted to supporting policies in these fields. While kind of technologies that are developed with the support
the ERDF and the Structural Funds in general finance a of this important policy instrument. In the attempt to fill
wide range of initiatives in support of innovation, from this gap, we focus on the concept of technological diversity,
the creation of an innovation-friendly environment to col- which refers to the regional capacity to carry out research in
laboration and technology transfer, the FP7 was entirely several technological fields. We investigate the determi-
devoted to subsidizing research, both basic as well as nants of technological diversity, shedding light on the fac-
applied and market oriented. This is one of the largest tors that are associated with greater regional capacity to
competitive research and development (R&D) pro- ‘diversify’ research activities or ‘specialize’ them. In particu-
grammes in the world, funding a significant volume of lar, we focus on two factors: the total amount of research
research and innovation and accounting for over 5% of funding granted to each region, and the degree of centrality
European member states’ expenditure on R&D. in interregional collaboration promoted by the FP7.
There is a growing literature that analyses the partici- The mass of co-financed research projects provides
pation in FPs (European Commission, 2016a, 2016b, unique information on the European, and often inter-
2016c), the outcomes and impacts of this important Euro- national, research trends as well as on regional assets and
pean policy instrument (Arnold, 2012; Arnold, Clark, & research specializations which have not been investigated
Muscio, 2005; Muscio, Rivera Leon, & Reid, 2015) and thoroughly. This paper presents the results of the analysis
the geographical patterns of research collaborations (Cecere of a large sample of FP7 projects pertaining to one of the
& Corrocher, 2015). Particular attention has been paid to most important European industries, the agri-food sector.
an analysis of the characteristics of regional networks and The FP7 projects considered in this paper are research
the effectiveness of the FPs for knowledge dissemination cooperation endeavours involving enterprises, universities

REGIONAL STUDIES
Technological diversity in Europe: empirical evidence from agri-food research projects 3

and other research-performing organizations. FP7 projects technology specialization or diversification, also consider-
are classified by call in CORDIS (the Community ing the concerns over an efficient implementation of an
Research and Development Information Service) and RIS3 strategy.
there is no official classification of projects by key research The debate on specialization and diversification in
area (KRA) or technology. Each call does not identify a regional economies (Boschma & Iammarino, 2009; Van
research area but only a general theme (e.g., NMP – nano- der Panne, 2004) that has taken place over the past 20
technologies, materials and new production technologies; years suggests that knowledge spillovers within a region
KBBE – knowledge-based bio-economy). The database or smaller country occur primarily among related sectors,
we use in this paper has been assembled on the basis of a and only to a limited extent among unrelated sectors
project-by-project analysis, focusing on the primary objec- (Boschma, 2005). The concept of ‘related variety’ (Fren-
tive of each project and the type of technology developed. ken, Van Oort, & Verburg, 2007) suggests that strength
These technologies have been classified on the basis of an in a particular sector, or part of a value chain, can have posi-
original taxonomy of KRAs developed by ISMERI tive spillovers in other related sectors (Muscio et al., 2015).
Europa, with the contribution of expert peer reviews for In this context, the core principle of the RIS3 literature is
each technological domain.1 The added value of our classi- that public and private resources should be concentrated
fication is threefold: (1) it takes into account the actual pro- on selected technological or market priorities that have
ject objective, allowing the identification of agri-food the potential to foster not only the emergence of new activi-
technologies even in calls that were not specifically ties but also the adoption, dissemination and adaption of
addressed to this field (e.g., those financed as part of an ‘general-purpose technologies’ across a wide range of sec-
NMP, KBBE, health or energy calls); (2) it makes it poss- tors (Foray, David, & Hall, 2011; Foray & Goenaga,
ible to classify research and technological development 2013).
(RTD) projects by research area and technology going in In this paper we do not deal with sectoral specialization
much more detail than the topics of the call and offering versus diversification, but narrow down the focus on tech-
a sectoral perspective; and (3) it allows one to break nological specialization by concentrating on KRAs and
down project costs by partner and NUTS-2 regions, con- technologies developed within a specific domain, that of
sidering the actual location of an organization or of the agri-food. We argue that investigating technological
specific branch/department participating in a certain pro- specialization and diversity in a specific domain at the
ject.2 Finally, unlike in other classification systems (e.g., regional level can provide key indications on the drivers
Frascati, World Intellectual Property Organisation – of specialization and on the coherence of research, techno-
WIPO), this paper focuses on a much narrower definition logical development and innovation (RTDI) policy in sup-
of technology areas which addresses the input side: the pro- porting it or not.
jects themselves and the KRAs under which the projects Following Wang and Von Tunzelmann (2000),
financed by FP7 fall. regional technological diversity can be defined as a region’s
The paper is organized as follows. The next section capacity to carry out research activity across different
summarizes the theoretical and policy background with knowledge domains. The expansion of technological com-
special reference to the European Union (EU) research petence into a broader range of technical areas (Granstrand
and innovation policy, the role of FP and the recent con- & Oskarsson, 1994) is reflected in the breadth of R&D
cept of smart specialization strategies (RIS3). The third output (Cecere & Ozman, 2014) and is considered as a
section presents the data and the methodology underlying vector of technological change and innovation (Van Rijn-
the taxonomy for grouping agri-food FP projects in soever, Van den Berg, Koch, & Hekkert, 2015). Some
KRAs and related key technologies. The fourth section works highlight the powerful role of firms’ technological
presents the econometric analysis. The fifth section draws diversity in driving their innovation performance (Breschi,
concluding remarks and policy implications. Lissoni, & Malerba, 2003; Gemba & Kodama, 2001;
Granstrand, 1998; Quintana-Garcia & Benavides-Velasco,
THEORETICAL AND POLICY BACKGROUND 2008). However, there is little evidence about what research
policy can do to support heterogeneity in research activity.
With the FPs, the EC has invested a significant amount of We still have little knowledge about how different policy
resources in research and innovation in both the private and tools influence diversity and insights into the underlying
public sectors. While there is a wide consensus on the posi- driving mechanisms are currently lacking (Van Rijnsoever
tive impact of FPs on the creation of networks, evidence on et al., 2015). Government funding is viewed as having a
the extent to which these measures have boosted regional strong positive effect on knowledge creation and technol-
innovation performance, fostered structural change towards ogy development to the extent that some authors (Defazio,
higher value-added activities and ultimately created a basis Lockett, & Wright, 2009; Lee & Bozeman, 2005) point
for sustained competitiveness across the EU is not conclus- out that the beneficial effects of research funding derive
ive (Clark, Muscio, Simmonds, & Von Tunzelmann, from the access to research resources rather than from col-
2004; Muscio et al., 2015). Moreover, little is known laboration and networking. However, the intensity of these
about what technologies have been developed with the sup- effects varies depending on the amount of funding (Banal-
port of FPs, while it would be vitally important to deter- Estañol, Jofre-Bonet, & Lawson, 2015; Bolli & Somogyi,
mine what EU policy has done so far in promoting 2011). For these reasons, we argue that the benefits of

REGIONAL STUDIES
4 Alessandro Muscio and Andrea Ciffolilli

research funding and collaboration could be extended to conditions for knowledge production and successful inno-
the breadth of technologies developed with the support vation (Powell & Grodal, 2005). From a regional policy
of EU collaborative schemes such as FPs. Therefore: perspective, it is therefore crucial to create the conditions
that stimulate the participation intensity of organizations
Hypothesis 1. There is a positive relationship between FP fund- in research projects. In this respect, we argue that the net-
ing and regional technological diversity. works promoted by the FP7 allow a critical mass of compe-
tencies to be reached that are positively related to the
Collaborative research programmes such as FPs are technological capabilities of regions and hence their tech-
supposed to ‘add value’ which cannot be obtained through nological diversity.
national programmes (Fahrenkrog, Polt, Rojo, Tubke, &
Zinocker, 2002; Georghiou, 2001). The most obvious Hypothesis 2. There is a positive relationship between the degree
addition is the ability to tackle technological problems of centrality in interregional networks promoted by the FP7 and
drawing together a much larger pool of resources, both technological diversity.
financial and human (Muscio, 2006). In fact, although
international collaborations undertaken under FPs may To access new knowledge that is dispersed and not
be expected to offer benefit over national programmes for available within regional boundaries, regional R&D organ-
several reasons, the opportunity to achieve a ‘critical mass’ izations increasingly rely on long-distance, cross-regional
of research performers – and users – is unlikely to be research collaborations (Muscio, 2013; Scherngell & Bar-
achieved nationally, or where a national focus would ber, 2009, 2011), which are regarded as efficient means
exclude ‘key players’ from the joint venture. Therefore, to access spatially distant knowledge in a relatively rapid
the FP provides an opportunity to tackle research questions and targeted way (Wanzenböck et al., 2015). Therefore,
at a scale that few EU member states would contemplate or the knowledge to which agents have access through estab-
could afford (Clark et al., 2004). lished network links determines its embeddedness within
Larèdo (1998) investigated the characteristics of the networks. Given the fundamental role of interaction in
networks promoted by previous iterations of the FP. The knowledge creation, networking is expected to ease and
history of FPs is characterized by a series of changes in facilitate the exploitation of external knowledge also for
institutional arrangements that determine the nature and the sake of technological diversity.
structure of research collaboration networks funded from A high level of network centrality allows regions that
EU sources (Pandza et al., 2011). Scientists get involved rely on ‘internal’ R&D activities to recognize and select
in government-funded collaborative research projects valuable linkages and to capture the know-how of the pro-
such as the FP because of the increasingly interdisciplinary, ject partners more efficiently. Thus, network centrality is
complex and costly characteristics of modern science (Lee expected to influence the relationship between FP7 fund-
& Bozeman, 2005). The question of whether collaboration ing and technological diversity, negatively moderating the
promotes research productivity has concerned the econ- impact of FP funding. In other words, in this context, net-
omic literature for quite some time. The interaction work centrality diminishes the relative importance of access
between researchers employed in academic institutions or to funding for the purpose of technological diversity.
research centres and in the business sector is expected to
bring together different sets of knowledge and cognitive Hypothesis 3: The higher the network centrality, the less effective
approaches that is likely to lead to the creation of new FP funding will be in influencing regional technological
knowledge. However, as noted by Lee and Bozeman diversity.
(2005), both collaboration and productivity may be influ-
enced by access to grants which can dictate team-based
goals.
EU-funded research networks require researchers to DATA AND METHODOLOGY
collaborate as a condition for securing research funding,
and there is evidence that collaborative networks positively So far, the technology and market orientation of FPs have
affect researchers’ productivity in the longer-term (Defazio been explored mainly by directly interviewing the stake-
et al., 2009; European Commission, 2016a). Moreover, holders (e.g., participating firms or academic institutions),
there is early evidence that collaborative network position as in Luukkonen (2002) who suggests that both technology
and the composition of collaborative innovation projects and market orientation are important reasons for compa-
influence diversity in emerging technologies (Van Rijnso- nies in their participation in the EU FP. Aggregate FP pro-
ever et al., 2015). A higher network embeddedness of a ject databases undoubtedly offer precious evidence about
region in a network is supposed to increase information relevant issues such as the networks promoted by the FPs
and knowledge access (Wanzenböck, Scherngell, & Lata, (Cecere & Corrocher, 2015), their spatial dimension
2015), contributing to creating a competitive advantage (Scherngell & Barber, 2011), the persistence of given actors
in the creation of new collaborations and alliances (Gilsing, in FP participation (Breschi et al., 2009; Breschi & Cus-
Nooteboom, Vanhaverbekec, Duystersd, & Van den mano, 2004), and the dynamics of knowledge diffusion
Oorda, 2008; Maggioni & Uberti, 2007) and the (Cassi, Corrocher, Malerba, & Vonortas, 2008).

REGIONAL STUDIES
Technological diversity in Europe: empirical evidence from agri-food research projects 5

The core of our analysis is based on an original database functional claims and KRA 4 – food quality, safety and tra-
of projects in the agri-food sector funded by the FP7 within ceability are emerging areas of research which still account
the ‘Cooperation’ and ‘Capacities’ programmes. Food and for a limited number of projects.
beverages was the fastest growing manufacturing sector In order to provide some insights into regional partner-
during the years that followed the recent world financial ships in each of the knowledge domains identified with the
crisis. This industry is mostly based on low R&D invest- peer-review process, we analysed the structure of the net-
ment levels and mature, pervasive technologies, where – works promoted by the FP. Table 3 present the results
in general, but not always – static capabilities dominate obtained from some standard indexes of social network
over dynamic capabilities (Martinez & Briz, 2000; Zahra, analysis (SNA). The results show that KRAs differ signifi-
Sapienza, & Davidsson, 2006). Despite having great inno- cantly in relation to the way regional partners network to
vative potential, the food industry is generally based on perform research tasks.
‘redundant technologies’. Science and technology offer The average centrality degree index can be used to allow
wide opportunities to change and improve the taste of pro- unbiased comparisons among several networks since its
ducts, preparation and nutritional characteristics, but the magnitude is not affected by the dimension of the network.
industrial structure is generally mainly composed of small This index measures the average number of ties between
and medium-sized enterprises (SMEs) with low R&D each region and its partners in FP7 agri-food projects.
capacity (Muscio, Nardone, & Dottore, 2010). Conse- The average number of links at the aggregate level is
quently, product innovations in the industry are hardly rad- 96,407, which is very high as our network includes just
ical and much more often of an incremental nature. This 295 regions, but, as expected, the estimated averages are
feature makes innovation dynamics in the food industry much lower at the disaggregated level, ranging from
similar to those in other mature industries such as textiles, 60,183 links in KRA 4 – food quality, safety and traceabil-
clothing and footwear, and despite its relevant innovation ity, to just 18,739 links in KRA 1 – primary production in
potential, differentiates the industry from science-based agriculture, forestry, aquaculture and fisheries.
sectors where the contribution of scientific institutions to Regions can be embedded in networks of different
innovation is much more important (Pavitt, 1984). complexity. A measure of complexity is provided by the
FP7 project data were regionalized at NUTS-2 level. average path length, a measure of relational distance
The database contains information on project budgets, between regions in FP7 networks. The diffusion of knowl-
type and name of partner organizations involved, geo- edge in Europe is geographically localized (Bottazzi & Peri,
graphical location of partners and, of course, KRAs and 2003; Cappelli & Montobbio, 2016; Maurseth & Verspa-
technologies developed through the projects. The identifi- gen, 2002) and regional openness is linked to a number of
cation of the KRAs and technologies promoted by the FPs structural factors (Gambardella, Mariani, & Torrisi, 2009).
required a peer-review process led by agri-food experts. Large distances as opposed to proximity may represent an
The team of experts identified five main KRAs. The obstacle to knowledge diffusion. In our case, we find that
KRAs represent the main scientific and technical fields of the average path length (column 6) is between 1.7 and
FP7 projects, each comprising a ‘family’ of key technol- 2.0, depending on the area, where two directly connected
ogies, as shown in Table 1. The regional-level data con- regions in a FP7 project are at a distance of 1.0. This
tained in the FP7 database were matched to shows that in agri-food FP7 is promoting tightly con-
EUROSTAT and Organisation for Economic Co-oper- nected, direct networks where each region in connected
ation and Development (OECD) data on regional econ- in less than two steps. Closely connected networks such
omic and research performance. Our dataset covers 295 as these can be considered conducive to relatively fast and
regions, including a non-EU category, which includes effective information diffusion between regions.
regions outside the EU, not considered in the econometric We estimated the clustering coefficient which represents
analysis because of missing data. The continent with the the average of the densities of the neighbourhoods of all the
largest number of organizations participating in FP7 pro- regions in the network. A high level of clustering increases
jects is Asia and the Middle East (263), followed by Africa the chances that complex knowledge is spread more effec-
(183), Central and South America (101), North America tively. The results of the estimated clustering coefficients sup-
(60) and Australasia (37). Only 12 out of 295 European port the results of the average path length discussed above,
regions did not participate in any FP7 project. Information ranging between 0.755 and 0.662, and confirm that regions
on these regions is available in Table A2 in Appendix A in are embedded in relatively highly clustered neighbourhoods.
the supplemental data online. These results were also confirmed in terms of compact-
Table 2 presents the total number of projects in each of ness of networks: with larger values indicating the greater
the five different KRAs discussed above. The large majority cohesiveness of a network, 0.62 out of the full scale of
of projects and participations is concentrated in KRA 1 – 1.0 suggested that the overall is relatively close. However,
primary production in agriculture, forestry, aquaculture the results are weaker at the level of individual KRAs,
and fisheries, which accounts for 43% of the total. This where the degree of compactness ranges between 0.20 in
area is followed by KRA 2 – innovative industrial processes the case of KRA 1 and 0.48 in KRA 4, the most compact
for food production and conservation, accounting for network. Finally, the density indicator measures the pro-
22.6% of projects; and KRA 5 – policy studies and knowl- portion of all possible ties that are actually present in the
edge transfer, accounting for 15.89%. KRA 3 – health and network. It is calculated as the sum of the existing ties

REGIONAL STUDIES
6 Alessandro Muscio and Andrea Ciffolilli

Table 1. Identified key research areas and related key technologies.


Key technologies (techniques, methods,
Key research areas knowledge and procedures)
Primary production in agriculture, forestry, aquaculture and fisheries . Highly efficient plants and trees
Research in this area addresses the challenges of sustainability and security of . Plant and tree protection
European production in agriculture, forestry, aquaculture and fisheries. It is . Improved livestock
mainly grounded in agricultural, forestry and fishery sciences, which are . Competitive and sustainable aquaculture
broad multidisciplinary fields that apply basic knowledge from exact, natural . Animal health and welfare
(in particular, biology), economic and social sciences to optimize production . Environmentally friendly farming
and processing in agriculture, forestry and fishery

Innovative industrial processes for food production and conservation . Processing technologies for high food quality
Greater efficiency in food processing, packaging and distribution is and safety
warranted in order for the agri-food industry to stay competitive, while . Innovative food packaging
maintaining or improving food quality and safety. At the same time, negative . Smart food-processing machines and
externalities of food processing have to be contained. This often requires equipment
systemic analysis of production processes. Research addressing these . Environmentally friendly processes
challenges is mainly grounded in food sciences, which is usually classified as
a subcategory under agricultural sciences

Health and functional claims . Tailor-made foods, functional foods and


In order to address health concerns, there is a need for research on diet– ingredients
health relations and on developing targeted communication strategies to . Improving diet–health relations
promote the adoption of healthy diets and lifestyles. In addition, there is a
need for research on functional foods (i.e., foods with specific health claims)
and on tailor-made foods for specific vulnerable groups. Research
addressing these challenges is usually based on a mix of food and medical
sciences

Food quality, safety and traceability . Traceability, feed/food supply chain security
Industry has primary responsibility for food safety. However, governments . Tools for food quality/safety assessment and
have an important controlling role to play. Both industry and governments control
require research to provide a robust understanding of safety issues along the
food/feed supply chains and adequate tools to assess quality and trace
contamination. Research addressing these challenges is mainly grounded in
food sciences and chemistry

Policy studies and knowledge transfer . Agriculture, fisheries and food policy
This category comprises: (1) projects that, strictly speaking, do not conduct . Knowledge transfer, RTDI coordination etc.
research, but which address issues that help to advance agri-food research
by disseminating research results or facilitating networking and cooperation;
and (2) projects that conduct research in support of policy decision-making
rather than production (i.e., socio-economic impact analysis and studies)
Source: Adapted from 2015 Ismeri Europa taxonomy of agri-food key research areas and key technologies.

divided by the number of all possible ties. The results indi- R&D are crucial factors for high regional embeddedness
cate that, again, there are large differences across KRAs in in terms of both betweenness and eigenvector centrality.
the degree of connectedness between regions. While in Table 4 presents information on the top 10 performing
KRA 1 there is only 6% of possible ties, in KRA 4 there European regions according to a number of FP7 indicators.
are 21% of possible connections. By far, the region that was awarded the largest number of
Wanzenböck et al. (2015) provide evidence that organ- projects and amount of funding was Gelderland in the
izations’ resources dedicated to R&D and regional per- Netherlands. This region hosts the ‘Food Valley’, a region
formance in high-technology patents and expenditure on where international food companies and research institutes

REGIONAL STUDIES
Technological diversity in Europe: empirical evidence from agri-food research projects 7

Table 2. Agri-food projects sponsored by the 7th Framework Programme for Research and Technological Development (FP7):
basic project statistics.
European
Average number Commission
% of participations contribution
Key research areas Projects projects Participations per region (€ millions)
1 Primary production in agriculture, forestry, 314 43.01 4275 2.13 423.61
aquaculture and fisheries
2 Innovative industrial processes for food 165 22.60 1844 2.90 151.97
production and conservation
3 Health and functional claims 73 10.00 1082 4.71 132.49
4 Food quality, safety and traceability 62 8.49 784 11.34 71.55
5 Policy studies and knowledge transfer 116 15.89 1606 4.44 105.41
Total 730 100.00 9591 5.06 885.01
Source: Authors’ elaboration on 2015 Ismeri Europa data covering FP7 agri-food projects.

Table 3. Regional network features.


Average Average graph
centrality Average clustering coefficient
Key research areas degree path length (weighted) Compactness Density
1 Primary production in 18,739 1.93 0.76 (0.53) 0.20 0.06
agriculture, forestry,
aquaculture and fisheries
2 Innovative industrial processes 23,702 1.86 0.74 (0.50) 0.24 0.08
for food production and
conservation
3 Health and functional claims 29,424 1.94 0.66 (0.43) 0.37 0.10
4 Food quality, safety and 60,183 1.74 0.67 (0.52) 0.48 0.21
traceability
5 Policy studies and knowledge 28,278 1.80 0.73 (0.51) 0.26 0.10
transfer
Total 96,407 1.65 0.72 (0.60) 0.62 0.33
Source: Authors’ elaboration on 2015 Ismeri Europa data covering FP7 agri-food projects.

such as Wageningen University are concentrated. Gelder- ECONOMETRIC ANALYSIS


land is also the region with the highest degrees of centrality.
Other top performers include Cataluña (Spain) and Île-de- Econometric specification
France, two regions with a flourishing food industry and In order to investigate regional technological diversity in
top research institutions in the food sector. At the country FP7 projects, the following model is estimated:
level, Italy is one of the best performers, with three regions
in the top 10 positions (i.e., Lazio, Lombardy and Emilia- TECHDIV i = a + b1 REGIONAL CONTROLSi
Romagna). Data disaggregated by KRAs to some extent + b2 FP7 NETWORKi + b3 FP7 FUNDINGi
confirm the regional performance at the aggregate level. + b4 FP7 NETWORKi ∗ FP7 FUNDINGi + 1i
Gelderland is still the top performer in four areas out of (1)
five, while Cataluña and Île-de-France are still in the 10
positions in each of the five areas considered. Finally, it is where TECHDIVi represents a measure of technological diver-
worth noting that regions in new member states secured sity measured with a scalar indicator ranging from one to five
10.38% of FP7 projects awarded to European countries, KRAs, taking into account the number of areas in which region
corresponding to 5.38% of funding. i has participated in one or more FP7 projects.3 REGIONAL
In conclusion, the analysis of descriptive statistics and CONTROLS is a set of explanatory variables controlling for
SNA indicators shows that while the FP7 has promoted regional socio-economic factors. FP7 FUNDING measures
relatively dense networks, with high degrees of centrality the natural logarithm of the total EC contribution, expressed
and relatively low relational distance between regions, in euros, granted to institutions and organizations located in
there is great diversity in how networks are shaped in region i. FP7 NETWORK represents a set of indicators
different research areas. accounting for the characteristics of regional networks promoted

REGIONAL STUDIES
REGIONAL STUDIES

8
Alessandro Muscio and Andrea Ciffolilli
Table 4. Top 10 regions involved in agri-food projects.
Total European
Commission
contribution Coordinated Centrality
NUTS-2 Region Projects NUTS-2 (€ thousands) NUTS-2 projects NUTS-2 degree Eigenvector
1 NL22 Gelderland 270 NL22 108,204 ES51 52 NL22 252 0.108
2 FR10 Île-de-France 199 FR10 77,334 NL22 49 FR10 238 0.105
3 ES51 Cataluña 177 ES51 48,358 FR10 34 ES51 238 0.105
4 ITI4 Lazio 143 IE02 44,223 IE02 20 IE02 230 0.103
5 DK01 Hovedstaden 129 DK01 38,753 ES52 19 DK01 229 0.104
6 IE02 Southern and Eastern 126 ITC4 33,719 DE11 17 ITI4 228 0.102
Ireland
7 ES30 Comunidad de Madrid 119 UKM2 31,408 DE50 17 ITH5 227 0.102
8 ITC4 Lombardia 118 ITI4 30,408 ITI4 16 EL30 227 0.102
9 BE10 Région de Bruxelles-Cap. 110 ES30 30,089 ITH5 16 ES30 225 0.103
10 ITH5 Emilia-Romagna 108 ITH5 28,960 DK01 15 ITC4 223 0.102

NUTS-2 KRA 1 NUTS-2 KRA 2 NUTS-2 KRA 3 NUTS-2 KRA 4 NUTS-2 KRA 5
1 NL22 22 NL22 29 ES51 51 NL22 131 NL22 61
2 ES51 22 NL31 21 ES52 33 FR10 99 FR10 47
3 IE02 16 IE02 20 NL22 27 ES51 79 ITI4 39
4 ITI4 16 DK01 19 DE50 25 ES30 64 BE10 26
5 DK01 14 BE10 18 IE02 24 DK01 59 AT13 26
6 UKE2 14 DE21 17 FR10 24 ITI4 58 DEA2 26
7 FR10 13 FR10 16 ITH5 22 UKM2 58 DK01 24
8 ITC4 13 UKH1 16 ITC4 22 ITC4 57 ITH5 24
9 DE30 12 ITH5 15 ES30 20 IE02 50 HU10 24
10 NL31 11 ES51 14 BE10 19 ES61 47 EL30 24
Note: KRA, key research area.
Source: Authors’ elaboration on 2015 Ismeri Europa data covering FP7 agri-food projects.
Technological diversity in Europe: empirical evidence from agri-food research projects 9

Table 5. Description of variables.


Variable Definition Primary source of data
Dependent variables
TECHDIV Number of key research areas (1–5) in which the region participates Ismeri Europa (based on DG
Research data)
HHID Herfindahl–Hirschman index Ismeri Europa (based on DG
Research data)
SDI Shannon diversity index Ismeri Europa (based on DG
Research data)

Regional indicators
EU-15 European Union of 15 member states –
English proficiency EF; English proficiency index (1–5) EF Education First
Distance from Brussels (ln) Distance from Brussels (km) EUROSTAT
Share students in Pupils and students in all levels of education (ISCED 0–6), share of EUROSTAT
education total population (2006)
GDP per capita (ln) Gross domestic product per capita (current market prices, 2006) EUROSTAT
GERD per capita (ln) Gross expenditure on research and development per capita (2006) EUROSTAT
Share manufacture of Manufacture of food and beverages, share of employment in EUROSTAT
food manufacturing (2006)

FP7 indicators
Distance from partners (ln) Average distance from partners (km2) DG Research
Share of intra-regional Share of FP7 links established between organizations located in the DG Research
links same region
Eigenvector Network centrality degree: eigenvector DG Research
Total EC contribution Total European Commission contribution for the analysed FP7 DG Research
projects (euros, instrumented)

Instruments
HRST (ln) Population with tertiary education or employment in science and EUROSTAT
technology (HRST) (thousands, 2006)
Share of projects Share of regional participations in FP7 projects coordinated by local DG Research
coordination organizations
Population (ln) Population (headcount) EUROSTAT
Note: ISCED, International Standard Classification of Education.

by the FP7. This set of indicators includes an indicator that con- each region a centrality that depends both on the number and the
trols for the average geographical distance of region i from other quality of its connections by examining all regions in parallel and
regions involved in the same projects,4 an indicator of the share assigning centrality weights that correspond to the average cen-
of intra-regional linkages, which measures the intensity of lin- trality of all linked regions (Bonacich, 1987).
kages between institutions and organizations within the same As a robustness check, we tested our model with
region, and an index of network centrality. This index accounts alternative measures of technological diversity such as the
for eigenvector centrality, a more sophisticated version of the Herfindahl–Hirschman index of diversification (HHID)
centrality degree discussed above. As suggested by Scherngell and the Shannon diversity index (SDI). The HHI is con-
and Barber (2011), the eigenvector is particularly useful in the ventionally used in industrial organization to measure
analysis of regional networks and stems from the idea that industry concentration, but it is becoming popular to
while having many connections ensures that a region can exert measure technological diversification (Quintana-Garcia &
influence and is able to access knowledge, it is also arguable Benavides-Velasco, 2008). The HHID can be expressed
that not all connections are the same. Typically, connections to as follows:
regions which are themselves well connected will provide regions
with more influence and knowledge resources than connections 
to poorly connected regions. Eigenvector centrality thus assigns HHIDi = 1 − HHIi = 1 − Pi2 (2)
i

REGIONAL STUDIES
10 Alessandro Muscio and Andrea Ciffolilli

Table 6. Descriptive statistics.


Variable Observations Mean SD Minimum Maximum
Dependent variables
TECHDIV 259 3.745 1.325 1.000 5.000
HHID 259 0.562 0.208 0.000 0.786
SDI 259 1.036 0.423 0.000 1.574

Regional indicators
EU-15 259 0.722 0.449 0.000 1.000
English proficiency 259 3.220 1.050 2.000 5.000
Distance from Brussels 259 887.038 588.755 0.000 3012.064
Share students in education 259 0.218 0.032 0.147 0.356
GDP per capita 259 24,177.610 12,074.370 2600.000 91,800.000
GERD per capita 259 417.282 453.871 4.000 2631.000
Share manufacture of food 245 16.469 8.620 1.975 55.215

FP7 indicators
Distance from partners 259 1157.137 391.123 574.056 3139.752
Share of intra-regional links 259 0.084 0.023 0.026 0.174
SNA eigenvector 259 0.053 0.028 0.003 0.108
Total European Commission contribution 259 6,636,419 11,000,000 0 108,000,000

Instruments
HRST 259 367.687 354.342 18.000 3492.000
Share projects coordination 259 0.067 0.092 0.000 0.600
Population 259 1,860,597 1,503,827 202,604 1.15e+07
Source: Authors’ elaboration.

where Pi denotes the proportion in a region of FP projects multicollinearity. The VIF test was passed for all variables
in KRA i. The index equals 0 when a region carries out with ample margins. The set of control variables includes
research activity only in a single area, and it is close to 1 an indicator of English proficiency proxying the ability of
when the region spreads its research activity over a broad applicants to write research proposals for the FP7 in
technological knowledge base. English language.
The SDI belongs to the group of entropy indices that
has been applied in the context of industrial diversification
(Aiginger & Davies, 2004; Attaran & Zwick, 1987). This Factors related to technological diversity
is defined as the negative sum of project shares, multiplied We test our hypotheses regarding technological diversity
by the natural logarithm of project shares of each single developed under the FP7 using the econometric strategy
KRA i. Because of the log-form, the relative weights of proposed above. Since the dependent variable TECHDIV
areas with a large number of projects are reduced compared is based on scalar values ordered progressively from 1 to 5,
with the HHID. The SDI is a measure of diversity, we chose an ordered logit model for the analysis. The
increasing with decreasing specialization so that the lower ordered logit was chosen in the attempt to evaluate the
bound (zero) gives absolute specialization and the upper impact of the selected covariates on the cumulative prob-
bound complete diversification. The SDI can be expressed ability of regions to engage in different research fields.
as follows: Table 7 reports in column 1 the ordered logit estimates
and the corresponding marginal/impact effects only for

N the top value (five) of the ordered scale. Column 2 presents
SDIi = − Pi lnPi (3) the marginal effects. The marginal effects of changes in the
i=1 independent variables in an ordered logit model are not
easy to interpret. As Greene (2002) indicates, in an ordered
Table 5 presents the definition of the variables used in the logit model, the sign of any parameter can only clearly
regressions, and Table 6 presents some descriptive determine the marginal effect of each variable on the
statistics. A correlation matrix is presented in Table A1 extreme probabilities (the probability of technological
in Appendix A in the supplemental data online. We diversity being broadest). We also run an ordinary least
used variance inflation factors (VIF) to help detect squares (OLS) model as a robustness check (column 3).

REGIONAL STUDIES
Technological diversity in Europe: empirical evidence from agri-food research projects 11

Table 7. Determinants of technological diversity.


(1) (2) (3) (4) (5) (7)
o.logit M.E. OLS OLS OLS
Variables tech.div. pr.out. ¼ 5 tech.div. HHID OLS SDI tech.div.
EU-15 0.169 0.029 0.074 –0.016 –0.016 0.084
[0.523] [0.134] [0.022] [0.044] [0.133]
English proficiency 0.132 0.023 0.077+ 0.013 0.025 0.085+
[0.158] [0.045] [0.011] [0.019] [0.045]
Distance from Brussels (ln) 0.185 0.032 0.029 –0.005 –0.005 0.141*
[0.153] [0.027] [0.007] [0.012] [0.064]
Share students in education 6.655 1.139 1.354 0.202 0.477 1.706
[4.767] [1.462] [0.360] [0.638] [1.439]
GDP per capita (ln) –1.221+ –0.209+ –0.492** –0.090* –0.177* –0.492**
[0.641] [0.176] [0.040] [0.070] [0.176]
GERD per capita (ln) 0.738* 0.126* 0.272** 0.056** 0.108** 0.277**
[0.289] [0.084] [0.020] [0.035] [0.083]
Distance from partners (ln) –0.760 –0.130 –0.122 –0.053 –0.080 –0.392
[0.843] [0.225] [0.053] [0.092] [0.272]
Share intra-regional links 27.471** 4.700** 5.989** 0.685 1.504+ 6.344**
[6.842] [2.008] [0.509] [0.866] [1.986]
Eigenvector 128.292** 21.950** 38.213** 4.166** 10.027** 37.856**
[11.305] [2.415] [0.600] [1.051] [2.409]
(IV) Total EC contribution 0.578* 0.099* 0.102* 0.033** 0.053** 0.198**
[0.288] [0.043] [0.010] [0.017] [0.053]
(IV) Total EC contribution*eigenvector –16.941** –2.898** –4.082** –1.096** –1.955** –5.382**
[5.899] [1.200] [0.269] [0.509] [1.198]
Manufacture of food and beverages (% –0.007
employment) [0.005]
Constant 4.806* 1.229** 2.016** 5.843**
[2.102] [0.407] [0.746] [2.198]

Observations 259 259 259 259 259 259


R2 0.748 0.482 0.585 0.752
Pseudo-R 2 0.471
Notes: Robust standard errors are given in brackets.
**p < 0.01; *p < 0.05; +p < 0.1.

Columns 4 and 5 present, respectively, the results obtained suggested in the economic literature, access to research
adopting the HHID and the SDI. funding is typically driven by the amount of available
We acknowledge the fact that, given the cross-sectional human resources. A large endowment in qualified human
nature of the data, our model could suffer from reverse capital facilitates the participation in research projects,
causality problems. In particular, if it is true that FP7 fund- granting access to larger FP7 funding. Similarly, as con-
ing can drive technology diversity, it is also true that regions firmed in DG Research statistics, projects coordinators in
that are technologically diversified could have better FP7 projects acquire, on average, larger resources than
chances of securing better funding. In order to address ordinary participants. Therefore, a region with organiz-
endogenous FP7 funding we introduced instrumental vari- ations coordinating research projects is likely to acquire lar-
ables (IV), instrumenting the natural logarithm of regional ger funding from the EC. Finally, regional population is
FP7 funding with three variables: the natural logarithm of included in the regression to control for regions’ size.
regional endowment of population with tertiary education The Durbin–Wu–Hausman test confirmed the necessity
or employment in science and technology (HRST) in the to instrument the variable measuring total EC contri-
year before the beginning of the programme (2006); the bution.5 The results of the estimation are presented in
share of regional participations in FP7 projects coordinated Appendix A in the supplemental data online.
by local organizations; the natural logarithm of the regional First of all, the results of the regressions show that in all
population (2006). As for the first explanatory variable, as model specifications the only control factors that matter for

REGIONAL STUDIES
12 Alessandro Muscio and Andrea Ciffolilli

the purpose of technological diversity are regional gross development of physical and human resources and via
domestic product (GDP) and gross expenditure on knowledge sharing across a multitude of regional actors.
research and development (GERD). Regional GDP is According to our evidence, the amount of FP7 funding
negatively correlated with technological diversity, with and network centrality is significantly associated with
scale effects allowing ‘richer’ regions, being more likely to regional technological diversity in agri-food science in
specialize in specific KRAs in food sciences. On the con- Europe. However, excessive reliance of participants on
trary, as expected, we find that regions with larger efforts external partners diminishes the beneficial effects of
in R&D as expressed by the GERD are more likely to research funding on technological diversity.
carry out research in more KRAs.
As far as FP7 indicators are concerned, it is especially CONCLUSIONS
important to note that participation in FP7, as expressed
by the amount of funding that regions are granted from European Framework programmes are primarily aimed at
the EC (TOT EC Contribution), always has a positive promoting research excellence and, to do so, they support
effect on the probability of having high levels of technologi- high-risk basic and applied research, transnational net-
cal diversity. Moreover, the coefficient size remains largely working between organizations and regions, mobility of
unaffected in the estimated models. people etc. Over time growing attention has been paid to
The coefficient for network centrality (SNA Eigenvec- reducing the wide regional variation in research and inno-
tor) is positive and highly significant in all model specifica- vation performance across the EU (European Commission,
tions, confirming that more centrally networked regions 2013).
have better chances of being involved in R&D activities The literature pays little attention to the regional tech-
in different areas of technology. While distance from part- nological effects of FPs and other policies. The analysis
ners is not a significant aspect for the purpose of techno- presented in this paper partly aims to fill this gap and
logical diversity, the higher the share of participants from shows that in an important domain such as agri-food
the same region involved in FP7 projects (Share intra- FP7 funding and centrality in the promoted collaborative
regional links), the higher the probability of being involved networks are strongly associated with greater technological
in the development of diverse technologies. Therefore, diversity. This implies that the most engaged and (relation-
regions with greater intra-regional networks have better ally) central regions are more likely to follow a path of tech-
chances of being technologically diversified than regions nological diversity rather than specializing in one or few
relying more extensively on external partnerships. This KRAs and in a limited number of related technologies.
reflects, in a way, regional size. In other words, while tech- On the contrary, ‘peripheral’ and less engaged regions are
nological specialization seem to need geographically close more likely to specialize in a limited set of KRAs and
partnerships, technological diversity is characterized by related technological fields.
research activities that are carried out to a higher extent Even though these results are limited to one domain,
between partners of the same region. agri-food, they suggest that the FPs implicitly tend to fos-
These results are supported by the findings on the inter- ter a core–periphery pattern where technological diversity is
action term (TOT EC Contribution*Eigenvector). All promoted in central territories that attract lots of funding
specifications include an interaction term between regional while, as can be expected, technological diversity does not
networking as expressed with the eigenvector and FP7 prosper in more isolated territories. The extent to which
funding. This variable accounts for the moderating role this kind of narrowed down diversity of periphery or, to
of regional networking on FP7 funding. The interaction put it differently, greater specialization, means higher
term is negative and highly significant, showing that capacity to compete and is in line with the philosophy
being highly central negatively moderates the positive effect underlying other European policies (e.g., smart specializ-
of FP7 funding on technological diversity. According to ation strategies) remains to be studied further. The current
our results, while FP7 funding is positively related to the exercise is a first attempt to incorporate the technological
level of technological diversity, its effect is dependent dimension in an exploration of the effect of RTD policy,
upon the level of regional centrality. There is a funding a point often disregarded by the literature, in the authors’
threshold above which the propensity to diversify technol- opinion.
ogy diminishes. This result shows that centrality does not This study has relevant policy implications, yet it faces
only influence technological diversity directly, but also some important limitations. First, the information con-
through its relationship with FP7 funding. Since Eigenvec- tained in FP projects is useful only at the aggregate level
tor centrality provides a measure of centrality that depends at the end of the programming period, in our case 2014.
both on the number and the quality of its connections, we Data from single calls are affected by cyclical trends and
conclude that the relation of FP7 funding and technology is factors that limit the information obtained from any econo-
stronger when regions are characterized by moderate inter- metric exercise. The need to use stock data requires the use
regional linkages. However, regions embedded in dense of cross-sectional analyses, which in turn expose to risks of
collaboration networks see a decline in the probability of reverse causality, with regions with higher technological
being involved in the development of diverse technologies. diversity applying more intensely for FP support. Secondly,
In conclusion, FP7 can support research and techno- given the focus of the paper on one specific scientific field,
logical development by promoting the acquisition or the agri-food sector, we can explore European engagement

REGIONAL STUDIES
Technological diversity in Europe: empirical evidence from agri-food research projects 13

in this area thoroughly, but we cannot tell whether greater REFERENCES


technological diversity or specialization is obtained at the
expense of greater diversity or specialization in other Aiginger, K., & Davies, S. (2004). Industrial specialization and geo-
science and technology fields. This brings us to future graphic concentration: Two sides of the same coin? Not for the
developments of this study which could be extended to European Union. Journal of Applied Economics, 12, 231–248.
Arnold, E. (2012). Understanding long-term impacts of R&D fund-
all scientific sectors supported by the FPs and hence pro-
ing: The EU Framework Programme. Research Evaluation, 21(5),
vide a more complete picture of technological diversity in 332–343. doi:10.1093/reseval/rvs025
Europe. Arnold, E., Clark, J., & Muscio, A. (2005). What the evaluation
record tells us about European Union Framework Programme
performance. Science and Public Policy, 32(5), 385–397. doi:10.
3152/147154305781779335
ACKNOWLEDGEMENTS Attaran, M., & Zwick, M. (1987). Entropy and other measures of
industrial diversification. Quarterly Journal of Business and
This work benefited from valuable input given by Elisa Economics, 26, 17–35.
Giuliani, Andrea Morrison, Nicoletta Corrocher, Luca Banal-Estañol, A., Jofre-Bonet, M., & Lawson, C. (2015). The
Rossi, Marco Pompili and three anonymous referees. The double-edged sword of industry collaboration: Evidence from
authors are particularly grateful to Antonio Lopolito for engineering academics in the UK. Research Policy, 44, 1160–
1175. doi:10.1016/j.respol.2015.02.006
his help with the social network analysis.
Bolli, T., & Somogyi, F. (2011). Do competitively acquired funds
induce universities to increase productivity? Research Policy,
DISCLOSURE STATEMENT 40(1), 136–147. doi:10.1016/j.respol.2010.10.001
Bonacich, P. (1987). Power and centrality: A family of measures.
American Journal of Sociology, 92, 1170–1182. doi:10.1086/228631
No potential conflict of interest was reported by the Boschma, R., & Iammarino, S. (2009). Related variety, trade lin-
authors. kages, and regional growth in Italy. Economic Geography, 85,
289–311. doi:10.1111/j.1944-8287.2009.01034.x
Boschma, R. A. (2005). Proximity and innovation: A critical assess-
SUPPLEMENTAL DATA
ment. Regional Studies, 39(1), 61–74. doi:10.1080/
0034340052000320887
Supplemental data for this article can be accessed at https:// Bottazzi, L., & Peri, G. (2003). Innovation and spillovers in regions:
doi.org/10.1080/00343404.2017.1301662. Evidence from European patent data. European Economic Review,
47, 687–710. doi:10.1016/S0014-2921(02)00307-0
Breschi, S., Cassi, L., Malerba, F., & Vonotars, N. (2009).
NOTES Networked research: European policy intervention in ICTs.
Technology Analysis and Strategic Management, 21(7), 833–857.
1. The database was compiled by extracting project infor- doi:10.1080/09537320903182314
mation from the CORDIS online repository. All FP7 pro- Breschi, S., & Cusmano, L. (2004). Unveiling the texture of a
jects financed under Cooperation and Ideas programmes European research area: Emergence of oligarchic networks
have been categorized and included in the database. Vari- under EU Framework Programmes. International Journal of
Technology Management, 27(8), 747–772. doi:10.1504/IJTM.
ables such as the real geographical location of participants
2004.004992
(NUTS-2) and the distribution of resources among part- Breschi, S., Lissoni, F., & Malerba, F. (2003). Knowledge-related-
ners have been added. ness in firm technological diversification. Research Policy, 32,
2. This is relevant for large research-performing organiz- 69–87. doi:10.1016/S0048-7333(02)00004-5
ations such as Fraunhofer, Centre National de la Recherche Cappelli, R., & Montobbio, F. (2016). European integration and
Scientifique (CNRS) etc. knowledge flows across European regions. Regional Studies,
3. The index aims at determining whether region i is 50(4), 709–727. doi:10.1080/00343404.2014.931572
Cassi, L., Corrocher, N., Malerba, F., & Vonortas, N. (2008). The
involved, at least to some extent, in the highest possible
impact of EU-funded research networks on knowledge diffusion
number of research areas. The degree of specialization in at the regional level. Research Evaluation, 17(4), 283–293. doi:10.
each of the five KRAs is instead measured by the HHID 3152/095820208X364535
and SDI indexes. Therefore, if Île-de-France has one pro- Cecere, G., & Corrocher, N. (2015). The intensity of interregional
ject in KRA 1 and two in KRA 5, it will have a value of 2 cooperation in information and communication technology pro-
(out of five). jects: An empirical analysis of the Framework Programme.
4. The distance was calculated using data on latitude and Regional Studies, 49(2), 204–218. doi:10.1080/00343404.2012.
759651
longitude of the central city of each region involved in the
Cecere, G., & Ozman, M. (2014). Technological diversity and inven-
FP7 projects. tor networks. Economics of Innovation and New Technology, 23,
5. Durbin–Wu–Hausman chi-squared test: 0.198, Chi- 161–178. doi:10.1080/10438599.2013.815473
sq(1), p-value ¼ 0.656, where H0: Regressor is exogenous. Clark, J., Muscio, A., Simmonds, P., & Von Tunzelmann, N.
(2004, May). Targeted review of added value provided by inter-
national R&D programmes. London: Office of Science and
ORCiD Technology (OST) International, Department of Trade and
Industry (DTI).
Alessandro Muscio http://orcid.org/0000-0001-5186- Defazio, D., Lockett, A., & Wright, M. (2009). Funding incentives,
2522 collaborative dynamics and scientific productivity: Evidence from

REGIONAL STUDIES
14 Alessandro Muscio and Andrea Ciffolilli

the EU framework program. Research Policy, 38, 293–305. doi:10. Applied evolutionary economics and economic geography (pp. 230–
1016/j.respol.2008.11.008 255). Cheltenham: Edward Elgar.
European Commission. (2013, August 7). Sixth FP7 monitoring report Martinez, M. G., & Briz, J. (2000). Innovation in the Spanish food &
(Monitoring Report 2012). Brussels: European Commission. drink industry. International Food and Agribusiness Management
European Commission. (2016a). An analysis of the role and engagement Review, 3, 155–176. doi:10.1016/S1096-7508(00)00033-1
of universities with regard to participation in the Framework Maurseth, P. B., & Verspagen, B. (2002). Knowledge spillovers in
Programmes, DG Research and Innovation (Study carried out by Europe: A patent citations analysis. Scandinavian Journal of
Ismeri Europa, Politecnico di Torino and IRCRES-CNR). Economics, 104, 531–545. doi:10.1111/1467-9442.00300
Brussels: European Commission. Muscio, A. (2006). The European added value of Framework
European Commission. (2016b). An analysis of the role and impact of Programmes: Evidence from the UK. Economia, Società e
research performing organisations’ participation in the Framework Istituzioni, 3, 381–412.
Programmes, DG Research and Innovation. Brussels: European Muscio, A. (2013). University–industry linkages: What are the deter-
Commission. minants of distance in collaborations? Papers in Regional Science,
European Commission. (2016c). An analysis of the role and impact of 92(4), 715–739. doi:10.1111/j.1435-5957.2012.00442.x
industry participation in the Framework Programmes, DG Muscio, A., Nardone, G., & Dottore, A. (2010). Understanding
Research and Innovation. Brussels: European Commission. demand for innovation in the food industry. Measuring Business
Fahrenkrog, G., Polt, W., Rojo, J., Tubke, A., & Zinocker, K. Excellence, 14(4), 35–48. doi:10.1108/13683041011093749
(2002). RTD evaluation toolbox: Assessing the socio-economic impact Muscio, A., Rivera Leon, L., & Reid, A. (2015). An empirical test of
of RTD policies (STRATA Project HPV 1 CT 1999-00005, the regional innovation paradox: Can smart specialisation over-
IPTS Technical Report Series No. EUR 20382 EN). Seville: come the paradox in Central and Eastern Europe? Journal of
Institute for Prospective Technological Studies (IPTS). Economic Policy Reform, 18(2), 153–171. doi:10.1080/
Foray, D., David, P. A., & Hall, B. H. (2011). Smart specialization: 17487870.2015.1013545
From academic idea to political instrument, the surprising career of Paier, M., & Scherngell, T. (2011). Determinants of collaboration in
a concept and the difficulties involved in its implementation European R&D networks: Empirical evidence from a discrete
(MTEI Working Paper 2011-001). Lausanne: École polytechni- choice model. Industry and Innovation, 18(1), 89–104. doi:10.
que fédérale de Lausanne. 1080/13662716.2010.528935
Foray, D., & Goenaga, X. (2013). The goals of smart specialisation (S3 Pandza, K., Wilkins, T. A., & Alfoldi, E. A. (2011). Collaborative
Policy Brief Series No. 01/2013 – May, S3 Platform). Seville: diversity in a nanotechnology innovation system: Evidence from
Institute for Prospective Technological Studies-Joint Research the EU Framework Programme. Technovation, 31, 476–489.
Centre (IPTS-JRC). doi:10.1016/j.technovation.2011.05.003
Frenken, K., Van Oort, F., & Verburg, T. (2007). Related variety, Pavitt, K. (1984). Sectoral patterns of technical change: Towards a
unrelated variety and regional economic growth. Regional taxonomy and a theory. Research Policy, 13, 343–373. doi:10.
Studies, 41, 685–697. doi:10.1080/00343400601120296 1016/0048-7333(84)90018-0
Gambardella, A., Mariani, M., & Torrisi, S. (2009). How ‘provincial’ Powell, W. W., & Grodal, S. (2005). Networks of innovators. In J.
is your region? Openness and regional performance in Europe. Fagerberg, D. C. Mowery, & R. R. Nelson (Eds.), The Oxford hand-
Regional Studies, 43(7), 935–947. doi:10.1080/ book of innovation (pp. 56–85). Oxford: Oxford University Press.
00343400801932268 Quintana-Garcia, C., & Benavides-Velasco, C. A. (2008). Innovative
Gemba, K., & Kodama, F. (2001). Diversification dynamics of the competence, exploration and exploitation: The influence of tech-
Japanese industry. Research Policy, 30, 1165–1184. doi:10.1016/ nological diversification. Research Policy, 37, 492–507. doi:10.
S0048-7333(00)00140-2 1016/j.respol.2007.12.002
Georghiou, L. (2001). Evolving frameworks for European collabor- Scherngell, T., & Barber, M. (2009). Spatial interaction modelling of
ation in research and technology. Research Policy, 30, 891–903. cross-region R&D collaborations: Empirical evidence from the
doi:10.1016/S0048-7333(00)00163-3 5th EU Framework Programme. Papers in Regional Science, 88,
Gilsing, V., Nooteboom, B., Vanhaverbekec, W., Duystersd, G., & 531–546. doi:10.1111/j.1435-5957.2008.00215.x
Van den Oorda, A. (2008). Network embeddedness and the Scherngell, T., & Barber, M. (2011). Distinct spatial characteristics of
exploration of novel technologies: Technological distance, industrial and public research collaborations: Evidence from the
betweenness centrality and density. Research Policy, 37, 1717– fifth EU Framework Programme. Annals of Regional Science,
1731. doi:10.1016/j.respol.2008.08.010 46(2), 247–266. doi:10.1007/s00168-009-0334-3
Granstrand, O. (1998). Towards a theory of the technology-based Van der Panne, G. (2004). Agglomeration externalities: Marshall ver-
firm. Research Policy, 27, 465–489. doi:10.1016/S0048-7333 sus Jacobs. Journal of Evolutionary Economics, 14, 593–604.
(98)00067-5 doi:10.1007/s00191-004-0232-x
Granstrand, O., & Oskarsson, C. (1994). Technology diversification Van Rijnsoever, F. J., Van den Berg, J., Koch, J., & Hekkert, M. P.
in ‘MUL-TECH’ corporations. IEEE Transactions on (2015). Smart innovation policy: How network position and pro-
Engineering Management, 41, 355–364. doi:10.1109/17.364559 ject composition affect the diversity of an emerging technology.
Greene, W. H. (2002). Econometric analysis (5th ed.). Upper Saddle Research Policy, 44, 1094–1107. doi:10.1016/j.respol.2014.12.004
River: Prentice-Hall. Wang, Q., & Von Tunzelmann, N. (2000). Complexity and the
Larèdo, P. (1998). The networks promoted by the Framework functions of the firm: Breadth and depth. Research Policy, 29,
Programme and the questions they raise about its formulation 805–818. doi:10.1016/S0048-7333(00)00106-2
and implementation. Research Policy, 27, 589–598. Wanzenböck, I., Scherngell, T., & Lata, R. (2015). Embeddedness of
Lee, S., & Bozeman, B. (2005). The Impact of research collaboration European regions in European Union-funded research and devel-
on scientific productivity. Social Studies of Science, 35, 673–702. opment (R&D) networks: A spatial econometric perspective.
doi:10.1177/0306312705052359 Regional Studies, 49(10), 1685–1705. doi:10.1080/00343404.
Luukkonen, T. (2002). Technology and market orientation in com- 2013.873119
pany participation in the EU Framework Programme. Research Zahra, S. A., Sapienza, H. J., & Davidsson, P. (2006).
Policy, 31(3), 437–455. doi:10.1016/S0048-7333(01)00118-4 Entrepreneurship and dynamic capabilities: A review, model
Maggioni, M. A. & Uberti, T. E. (2007). Inter-regional knowledge and research agenda. Journal of Management Studies, 43(4),
flows in Europe: An econometric analysis. In Frenken K. (Ed.), 917–55. doi:10.1111/j.1467-6486.2006.00616.x

REGIONAL STUDIES

You might also like