Research Policy 44 (2015) 1341–1359

Contents lists available at ScienceDirect

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

Determinants of university–firm R&D collaboration and its impact on

innovation: A perspective from a low-tech industry
Ornella Wanda Maietta ∗
CSEF and DISES, Università degli Studi di Napoli Federico II, via Cinthia 45, 80126 Naples, Italy

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

Article history: The main aim of the paper is to examine the drivers of university–firm R&D collaboration while at the
Received 4 April 2014 same time assessing the determinants of innovation in a low-tech industry. This includes analysing firm
Received in revised form 17 March 2015 R&D collaborations with partners different from universities.
Accepted 18 March 2015
The paper relies on a unique data-set where firm data were sourced from the Capitalia survey, covering
the 1995–2006 years, and the university data were gathered from a number of sources.
JEL classification:
Result from a multivariate probit model reiterate that university–firm R&D collaboration affects pro-
O3
cess innovation. Evidence of a more novel kind suggests that product innovation is positively affected by
I23
D22
geographical proximity to a university but is negatively affected by the amount of its codified knowledge
R1 production. Degree programmes in fields useful for local firms favour R&D collaborations. Academic poli-
cies that aim to commercialise research output negatively impact both product and process innovations
Keywords: of local firms.
University–industry interaction © 2015 Elsevier B.V. All rights reserved.
Firm innovation
Academic research quality
Geographical distance
Academic education
Gender glass ceiling

1. Introduction However, little research has focused on a systematic investi-

gation of the impact on firm product and process innovations of
University knowledge production is important for industrial various channels of knowledge and technology transfers from uni-
innovation. Knowledge may be transferred to firms through versities, public research labs and private firms. Although there are
unintended flows that generate spillovers from university-based some contributions in the literature regarding whether academic
research or through market-mediated interactions, such as contract research quality indirectly affects firms’ innovative performance
and collaborative research (D’Este and Iammarino, 2010; D’Este and through university–firm interactions (Baba et al., 2009; Mansfield
Patel, 2007). The relevance of the specific market-mediated channel and Lee, 1996; Zucker et al., 1998), the direct impact of academic
of the knowledge transfer from university to industry varies across research quality indicators on firms’ product and process innova-
disciplines and sectors depending on the degrees of knowledge tions has not yet been quantitatively measured.
codification and interdependence in firm technology (Bekkers and The primary objective of the paper is the identification of
Bodas Freitas, 2008). Academic research quality and geographical the channels through which university–firm R&D collaboration
proximity from firm to university (Audretsch and Stephan, 1996; impacts upon firm product and process innovations and how the
Mansfield, 1991, 1995) are also universally recognised as influenc- “knowledge context” in which the firm operates (in terms of
ing market-related university–firm interactions, mainly through research, education and technology transfer-related activities at
contract and collaborative research (D’Este and Iammarino, 2010; local universities) affects this relationship. The paper’s secondary
D’Este et al., 2013; Laursen et al., 2011) and licensing (Mowery and objectives include focusing on a sector that is typically considered
Ziedonis, 2015). to be codified knowledge extensive, performing a joint analysis of
the determinants of firm R&D investment and innovation and iden-
tifying the direct impact on firm innovation of academic research
quality and degree programmes. The impact of university edu-
cation on patent production is recognised for high-tech sectors
∗ Tel.: +39 81 675032.
in studies in which analyses are performed by aggregating the
E-mail address: [email protected]

http://dx.doi.org/10.1016/j.respol.2015.03.006
0048-7333/© 2015 Elsevier B.V. All rights reserved.
1342 O.W. Maietta / Research Policy 44 (2015) 1341–1359

graduation fields that are relevant for a specific industry (Leten that can be grouped into the categories of proximity, university,
et al., 2014). The paper, by relying on disaggregated graduation firm and territory characteristics.
fields, provides novel evidence on this issue. Following D’Este et al. (2013), geographical proximity plays a
The conceptual framework of the analysis is the “National Sys- fundamental role as a determinant of university–industry collab-
tems of Innovation” (NSI) approach, which assumes that a firm’s oration that has been recognised by different bodies of literature,
innovative capabilities depend upon its ability to communicate including studies on localised knowledge spillovers, the systemic
and interact with external knowledge sources such as other firms, nature of knowledge and innovation (from innovation systems
customers and scientific institutions that can act as knowledge to the triple-helix model), and industrial clusters. The studies on
providers (Freeman, 1988; Lundvall, 1988; Nelson, 1993). A firm’s localised knowledge spillovers are based on the knowledge produc-
absorptive capacity shapes its demand for knowledge and technol- tion function framework that was proposed by Griliches (1979) and
ogy transfer because firms with low absorptive capacity depend first implemented by Griliches and Pakes (1984) and Jaffe (1986,
more on local high-quality universities (Laursen et al., 2011) for 1989). Following this conceptual framework, knowledge output,
industrial research and for the expertise and training that are which is proxied by patent applications and/or innovation cita-
offered to the local market for skilled labour. This latter acts as a tions, is produced according to a Cobb–Douglas technology using
medium for the diffusion of academic knowledge spillovers (Beise R&D efforts, namely, business and university R&D expenditures.
and Stahl, 1999) which may particularly benefit small and medium- The “Griliches (1979) – Jaffe (1989) knowledge production func-
sized firms with a lower capacity to compete in the national labour tion” has typically been estimated for nations (Acs et al., 1991;
market. In the specific case of family-run firms, owners’ children Anselin et al., 1997; Feldman and Florida, 1994; Jaffe, 1989) or terri-
often choose to attend a degree programme at a local university. torial regions (Acs et al., 2002; Fritsch, 2002; Greunz, 2002; Varga,
Furthermore, institutional changes may contribute to reinforcing 2000). Studies on the systemic nature of knowledge and innova-
the relevance of certain NSI actors as local providers of external tion typically focus on the interactions and networks among actors
firm knowledge (Robin and Schubert, 2013). aimed at the production, diffusion and use of knowledge. The inno-
The novel contribution of the paper is that it disentangles and vation systems approach privileges the firm as the core agent of
quantifies the direct and indirect impacts of research, education the network, whereas the triple-helix model places the university
and technology transfer-related activities at local universities on at the centre of a relationship among firms and the government. The
firm innovative inputs and outputs. Local university structure and empirical analysis in the original formulation for both approaches
behaviour characterise the “knowledge context” in which the firm is based on national data. A sub-national level of analysis has been
mainly operates and gathers information for R&D outsourcing. The made possible through studies on industrial clusters that consider
local university is represented, in the paper, by the closest fac- university–industry collaboration as the key factor for the compet-
ulty of agricultural studies, along with being knowledge spillovers itiveness and growth of local economic systems.
from technical universities more geographically bounded (Beise The bodies of literature discussed above assume that firms
and Stahl, 1999). that are located near universities may frequently collaborate with
The present analysis employs a representative sample of Ital- them and benefit from knowledge spillovers. Geographical prox-
ian food and drink (F&D) firms that have at least 10 workers imity (Morgan, 2004) enables the transmission of tacit knowledge,
and that are included in the 7th (1995–1997), 8th (1998–2000), which is personal and context-dependent. This knowledge cannot
9th (2001–2003) and 10th (2004–2006) waves of the Capitalia be easily bought via the market and is difficult to communicate
survey. A long period is necessary to ascertain the effects of col- other than through personal interaction in the context of shared
laboration between NSI actors and industry after accounting for experiences. In particular, geographical proximity matters when
cross-sectional and time heterogeneity. A complementary and very knowledge spillovers are informal and in the event of informa-
rich panel data-set has been built to describe the local university tion asymmetry between researchers and research users, which
characteristics, which have been found to impact university–firm arises when users cannot precisely evaluate the applicability of the
collaborations. The empirical approach is a multivariate probit sys- transferred research until they attempt to translate it into new or
tem that allows for simultaneity between intra moenia and extra improved products or processes (Audretsch and Feldman, 1996;
moenia R&D investments (Belderbos et al., 2004; Veugelers, 1997) Jaffe et al., 1993; Landry et al., 2007). In the context of asymmetry,
and the endogenous nature of R&D decisions (Crépon et al., 1998; the transfer of knowledge is unlikely if researchers and research
Robin and Schubert, 2013). users do not have frequent interactions. The number of universities
The remainder of the paper is divided into six sections. The sec- within the region in which a firm is located also affects the probabil-
ond section reviews different bodies of literature that address the ity of interacting with a nearby university because it increases the
issue discussed here. Sections 3 and 4 focus on the specificities of range of options that are available to a firm (D’Este and Iammarino,
the Italian public agri-food research system and of the Italian F&D 2010).
industry in the European context, respectively. Section 5 describes Conversely, codified knowledge, which is explicit and stan-
the sources of the data and the methodology that have been used, dardised, can be transmitted over longer distances and across
and section six presents the empirical results. Section seven pro- organisational boundaries with little cost. The capability of shared
vides concluding remarks. codification creates non-spatial proximity (Boschma, 2005): cogni-
tive proximity, which is the extent to which two organisations share
the same knowledge, and organisational proximity, the result of
2. The determinants of university–industry collaboration the accumulation of experience between the same or similar actors
and university–industry collaboration as a determinant of (D’Este et al., 2013). When knowledge is transmitted through for-
innovation mal ties between researchers and firms, geographical proximity
is not necessary because face-to-face contact does not occur by
The paper brings together two strands of literature: the former chance but is instead carefully planned (Audretsch and Feldman,
is focused on the identification of drivers of university–firm R&D 1996). Cognitive proximity is generally lower in social science
collaboration and the latter is related to the determination of the research than in natural science research because social science
impact of university–firm R&D collaboration on innovations. knowledge is less codified than natural science knowledge and is
Within the former body of literature, a number of studies have not based on a unified and established scientific methodology. Thus,
analysed the determinants of university–industry collaboration geographical proximity to universities may be more important for
 O.W. Maietta / Research Policy 44 (2015) 1341–1359 1343

accessing social science research than for accessing natural science for tacit-knowledge-intensive industries (Fitjar and Rodríguez-
research (Audretsch et al., 2005). The amount of tacit knowledge Pose, 2012).
also varies along the life cycle of a research project in that it is gen- Among the personal characteristics of scholars, age and profes-
erally higher in its earliest phase, which requires a relatively high sional status are taken into account because older scientists and full
degree of affinity between the firm and the university (Broström, professors are expected to accept multiple offers of firm involve-
2010). ment, whereas younger scientists and research assistants are more
Among university characteristics, the determinants of likely to be involved with a local firm than with a non-local firm
university–industry collaboration that have been identified in or to not be involved at all (Audretsch and Stephan, 1996; Landry
the literature are academic research quality, university size and et al., 2007). Gender is also used as a control variable (Landry et al.,
faculty/discipline composition, department size, intermediation 2007).
and the age, seniority and gender of researchers. The firm characteristics that are identified in the literature as
Academic research quality (Mansfield, 1991) is expected to act drivers of university–industry R&D collaboration are size, age, intra
as a catalyst for industrial labs that are interested in conducting and extra muros R&D, ownership structure and innovation subsidies
joint research activities by attracting firms with cutting-edge tech- (Huynh and Rotondi, 2009; Laursen et al., 2011; Medda et al., 2005;
nologies. Mansfield (1995) provides evidence that higher-quality Motohashi, 2005; Piga and Vivarelli, 2004). Among the territory
universities make greater academic contributions to industrial characteristics, location in industrial clusters (D’Este et al., 2013)
innovation. Mansfield and Lee (1996) argue that firms prefer to and regional R&D intensity (Laursen et al., 2011) may influence
work with local university researchers and with more distinguished university–firm collaboration.
university departments; however, the impacts of academic quality Within the strand of literature related to the impact of
and geographical proximity are not homogeneous across disci- university–firm interactions on innovation, the finding that prox-
plinary fields. The effect of geographical proximity on businesses’ imity to a university is positively associated with innovation is well
choices with respect to university partners is more pronounced established in studies based on a production-function approach to
for applied research than for basic research. Firms that con- investigating academic spillovers (Fritsch and Franke, 2004). Rel-
duct basic research predominantly collaborate with high-quality atively fewer papers quantify the impact of university–industry
departments. D’Este and Iammarino (2010) disentangle the effects collaboration in terms of outcome variables such as firm innovation.
of geographical proximity and university research quality on These latter studies differ in the types of samples (for example, only
the frequency and distance of university–industry research col- firms with R&D activities, only innovators or samples that include
laborations. For engineering-related departments, proximity is innovators and non-innovators), in the use of simultaneous equa-
key to explaining the frequency of collaborations with industry, tion models, in the analysis of product and/or process innovations,
whereas it is not important for basic-science related depart- in the type and number of university–firm interaction channels and
ments, for which the positive impact of research quality prevails. in the comparison with other external sources of firm knowledge.
However, the relationship between academic research quality Estimates related to innovating firms only may suffer from
and distance of collaborations is non-linear because collabora- sample selection bias if university–industry collaboration is an
tions with industry turn out to be geographically closer after a important means of innovation for firms that would otherwise not
certain threshold of research quality is reached. Laursen et al. innovate because they had no other innovation strategies, such as
(2011) find that firm choices regarding collaborating with local intra moenia R&D or some form of external knowledge acquisition.
high-quality universities depend on the firm absorptive capac- Samples of innovators and non-innovators are used by Amara and
ity: firms with low absorptive capacity choose to collaborate Landry (2005), Becker and Dietz (2004), Chen et al. (2011), Fitjar
with a high-quality local university or, as second best, with a and Rodríguez-Pose (2012), Karlsson and Olsson (1998) and Nieto
high-quality non-local university. For firms with high absorptive and Santamaria (2007).
capacity, geographical proximity to a top university has no effect To take into account the endogenous nature of R&D deci-
on collaboration choice. Muscio and Nardone (2012) find that sions, a simultaneous equation approach is suggested for modelling
academic research quality positively impacts the private fund- internal and external R&D expenditures (Veugelers, 1997), R&D
ing of university research activities, particularly with respect to collaborations with different partners (Belderbos et al., 2004) and
food sciences departments. Academic reputation is also proxied internal and external R&D expenditures and innovation (Becker
by university age, measured in years (Audretsch and Lehmann, and Dietz, 2004). Both product and process innovation are analysed
2005). by Fitjar and Rodríguez-Pose (2012), González-Pernía et al. (2014),
To account for the fact that academic institutions require a criti- Nieto and Santamaria (2007) and Robin and Schubert (2013). R&D
cal mass of researchers to improve their chances of interacting with collaborations with universities or public research labs are most
firms, scholars have introduced university and department size into frequently investigated (Becker and Dietz, 2004; Belderbos et al.,
the analysis, which are quantified as the number of researchers (or 2015; Broiwer and Kleinknecht, 1996 González-Pernía et al., 2014),
the percentage of time) devoted to research activities (D’Este and whereas citations are less frequently studied (Beise and Stahl, 1999;
Iammarino, 2010; Landry et al., 2007; Muscio and Nardone, 2012) Jiang et al., 2010; Mansfield and Lee, 1996).
or the R&D intensity of the higher education sector (Huynh and The presence of multiple university–firm interaction channels
Rotondi, 2009). (Arvanitis et al., 2008; Jang et al., 2010; Karlsson and Olsson, 1998)
The composition of the university faculty/discipline or the aca- and firm R&D partners (Belderbos et al., 2004; González-Pernía
demic scientific specialisation are introduced into the analysis of et al., 2014; Monjon and Waelbroeck, 2003; Robin and Schubert,
university spillovers to capture the higher familiarity with net- 2013) enriches the analysis of university–firm collaboration safe-
working of basic versus applied research, the different production guarding it from omitted variable bias because the impacts are
of tacit knowledge and the capability of technology transmission typically numerous and difficult to separate from other firm activ-
(Audretsch and Lehmann, 2005; Audretsch et al., 2012; Bonaccorsi ities (Arvanitis et al., 2008).
et al., 2013; D’Este and Iammarino, 2010; Landry et al., 2007). The The nature of knowledge flows between a firm and its part-
latter is also proxied by the existence of an intermediation struc- ners and the consequent innovation-related benefits may differ
ture, such as a technology transfer office, that is established to according to the interaction channel, the firm’s R&D partner and
minimise the cognitive distance between business and academics the project phase. Furthermore, firms’ R&D partners complement
(Muscio and Nardone, 2012), or by the university’s regional location or substitute each other, in the latter case mostly because academic
1344 O.W. Maietta / Research Policy 44 (2015) 1341–1359

researchers may be time constrained and less available to engage concept of “research of regional interest”.1 Following this policy,
in technology-development projects. regional administrations created their own public lab networks.2
Firms choose universities as R&D collaboration partners in cases The Italian national agri-food research system appears to be
of multi-purpose problems and learning-focused projects in areas one of the more fragmented among European countries. It is
involving new science and producing long-term benefits, such as poorly coordinated, with a strong presence of universities and a
strengthening their absorptive capacities and radical innovation; weak presence of private research labs. The German national agri-
the motivation is that these projects are more likely not to be food research system is also considered rather fragmented and
aborted prematurely (Amara and Landry, 2005; Arvanitis et al., decentralised, with the particular characteristic that universities of
2008; Bodas Freitas et al., 2011; Broström, 2010; De Fuentes and applied sciences are typically run by the federal governments with
Dutrénit, 2012; Fitjar and Rodríguez-Pose, 2012; Hall et al., 2003; an orientation towards diffusion. By contrast, the French national
Kaufmann and Tödling, 2001; Thether and Tajar, 2008). Distant uni- agri-food research system is concentrated, mission-oriented (with
versities are firms’ R&D partners (De Fuentes and Dutrénit, 2014), missions being defined at the national level) and fairly coordinated.
but generally not in the earliest phase of the project (Broström, In particular, it is characterised by the very low importance of uni-
2010); foreign universities usually partner highly innovative firms, versities and the absence of links between research and higher
at the frontier of the academic knowledge in their industry, when education in the agri-food industry instead delivered by colleges.
these latter need to acquire new forms of academic knowledge Alternatively, in the Netherlands, the national agri-food research
(Monjon and Waelbroeck, 2003). Smaller firms prefer to interact system is well coordinated and rather concentrated, and univer-
with local public research labs for applied research (De Fuentes sities are growing in importance (Chartier, 2007; Menrad, 2004;
and Dutrénit, 2014), whereas private research labs are generally Robin and Schubert, 2013).
firms’ R&D partners in cases of incremental innovation (Thether Most of the Italian academic research on agri-food topics
and Tajar, 2008). Short-term benefits, such as those associated with is performed by faculties of agricultural studies3 (INEA, 2009),
quality control, usually accrue from informal interactions to imi- which employed 2111 scientists in 2000 and numbered 24 before
tators or firms adopting incremental innovation (De Fuentes and the last university reform.4 Each university is autonomous (Law
Dutrénit, 2012; Monjon and Waelbroeck, 2003). In the specific 168/1989)5 but is under the control of the Higher Education Depart-
case of the F&D industry, firms collaborate with universities and ment. The number of scientists employed by equivalent university
public research labs to access new ideas and government funding, institutes in Germany numbered 812 in 2000 (Menrad, 2004), and
develop internal expertise and reduce time to market with new the total number of equivalent faculties is 10 (Chartier, 2007). Some
technologies, particularly for process innovation and new market of the Italian faculties of agricultural studies, which are highly dis-
penetration. Informal contacts and training courses are also impor- persed throughout the country, are very young: 5 were established
tant knowledge transfer channels (Avermaete et al., 2004; Kelly in the 1980s, and 4 were established in the 1990s. Yet Italy also
et al., 2008; Minarelli et al., 2015). hosts the oldest faculty of agricultural studies in the world, which
To disentangle the impact on firm innovation of R&D collab- was founded in 1871 at the University of Pisa.
orations with universities, other variables that are customarily Technology transfer to firms is achieved through regional devel-
used in the literature as innovation determinants must be con- opment agencies, one for each region, which generally conduct only
trolled for. These variables include firm age, size, R&D intensity, applied research, beginning with the basic research supplied by
workforce skills, territorial location, innovation subsidies (Arvanitis universities.
et al., 2008; Avermaete et al., 2004; Benfratello et al., 2008; Brouwer The huge number of labs and technology institutes has cre-
and Kleinknecht, 1996), retail involvement in innovation (Stewart- ated problems of coordination and communication among labs and
Knox and Mitchell, 2003) and cooperative status (Huiban and between labs and firms. For this reason, a reform of the public
Bouhsina, 1998). research labs under the control of the Department for Agricul-
Summing up the literature review, very few studies adopt a ture was undertaken in 1999 by merging them into one funding
simultaneous multi-equation approach which models intra muros agency (CRA). In the same year, the activities of the main Italian
and extra muros R&D investment and innovation adoption deci- non-university public research institute (CNR), which is under the
sions. No study jointly analyses the determinants of university–firm control of the Department of Higher Education, were also restruc-
R&D collaboration and of product and process innovations while tured and unified under specific themes. A third important event
controlling for (various indicators of) academic research quality. for Italian public research occurred at the beginning of 1999, when
These are the considerations upon which the paper relies in order the Italian Evaluation of Research Quality established an evalu-
to bring about novel evidence in this field. ation procedure for research activities through periodic reports.
This evaluation was the result of a general debate on the Ital-
ian public research system; the resulting Bassanini Law 59/1997
3. The Italian public agri-food research system reformed the entire system by introducing coordination, evaluation
and participative research planning (through National Research
A conceptual framework that is appropriate for analysing the
specific factors that influence firm innovative capabilities in the
F&D industry (Menrad, 2004) is the NSI approach, as proposed by 1
The average share of university budgets sourced from regional administrations
Freeman (1988), Lundvall (1988) and Nelson (1993). Within this was 2% in 2003 and 12% in 2011 (Netval, 2005, 2013).
approach, national boundaries are relevant for a firm’s innovation 2
Law 491/1993 and Legislative Decree 143/1997 further decentralised compe-
performance because of differences in government policies and in tencies in the farming and agri-food industries to regional administrations. The
amendment to Title V of the Constitution institutionalised this decentralisation of
common social and cultural values that shape institutions in an
powers in 2001.
innovation system. 3
In 1998, 51% of scholars who were active in the field of public research on agri-
As with other European countries, agri-food research in Italy food topics were employed by universities, 27% by public labs under the control of
is conducted primarily in public sector institutions, experiment the Department for Agriculture or the Department of Higher Education and 11% by
stations and laboratories run by universities and departments (agri- regional public labs (INEA, 1999).
4
Law 240/2010 abolished faculties and transferred teaching to departments.
culture, higher education, public health, environment, productive 5
University statutes and internal regulations address external fundraising and
activities and foreign affairs). Italian regional governments have technology transfer activities that represent the only sources of external income
also financed public research since DPR 606/1976 introduced the because of limitations on student fees.
 O.W. Maietta / Research Policy 44 (2015) 1341–1359 1345

Plans). Law 297/1999 completed this reform by reordering the sig- increasing8 as a consequence of university incentives given to
nificant number of subsidies to industrial research and explicitly professors in the form of profit sharing, research funds or career
introducing incentives for collaboration between public and private advancement (Fantino et al., 2012).
researchers. Concurrently, to attract new students and decrease the In 2007, the R&D expenses for private for-profit firms as a
dropout rate, Ministerial Decree 509/1999 reformed the length of fraction of the total R&D investment in Italy ranges from 70% in
the degree programmes6 by introducing a 3-year degree that can northern Italy to 31% in southern Italy, where universities and pub-
be followed by a 2-year specialisation degree. lic labs finance 66% of the total R&D (ISTAT, 2011). The amount of
A higher education evaluation system was formally introduced R&D from universities and public research labs that was devoted to
by Law 537/1993. To assess the quality of the tertiary education sys- the F&D industry was equal to 440 ml D in 2008, whereas, on aver-
tem, the Department of Higher Education began to collect a large age, F&D firms spent 127 ml D annually in R&D over the 1998–2008
amount of data for the purpose of developing qualitative and quan- period (Sorrentino and Capozzi, 2010). Italy ranks third in Europe,
titative indicators. These data, which are summarised in annual after Spain and Germany, in terms of the amount of R&D from uni-
reports by the National Committee for the Evaluation of the Univer- versities and public research labs that was devoted to the F&D
sity Sector (CNSVU), are publicly available and have been used in industry during the 1998–2008 period (Sorrentino and Capozzi,
this paper. They also represent the basis for faculty ranking that is 2010).
published annually by Italian newspapers (la Repubblica and Il Sole
ventiquattrore) to assist high-school students in choosing a degree 4. The Italian F&D industry
programme (Magistà, 2013).
Beginning in the 1980s, Italian academic research and education The Italian F&D sector makes an interesting case study of tacit-
put greater emphasis on content related to processing industries knowledge-based small firms whose absorptive capacity is poorly
through new research lines and the birth of specialised gradua- measured by their R&D expenditures. In fact, the sector is char-
tion fields (Santini, 2003). An additional development during these acterised by (a) the significant presence of small firms with no
years, that is relevant for university–industry interactions, was the R&D intensity and, therefore, a potentially more important role
introduction into degree programmes of compulsory pre-graduate for university–industry collaboration; (b) the presence of cognitive
student internships at agri-food firms. These internships directly gaps that are linked to knowledge regarding the effects of pedo-
connected universities and firms or indirectly connected these enti- climatic conditions on local production; and (c) the demand for
ties when students, who were offered internships, remained at the innovation based on the public good attributes of food products,
firms. such as food safety.
With respect to disciplines, the composition of the Italian fac- The F&D industry has traditionally been considered a low-tech
ulties of agricultural studies has not changed substantially over industry on the basis of R&D intensity, which was 0.33% in 2010 for
the past decades; for example, the percentage of researchers in Italy. The average R&D expenditure for Italian F&D firms was 32,000
the biotechnology fields is only 4% of the total research personnel, D over the 2001–2003 period and 35,000 D in the 2004–2006
whereas the percentage of researchers in the traditional agron- period. Although the level of intra-muros R&D intensity is low, it
omy and livestock production fields is 30% (Pennacchi, 2008). is increasing (ISTAT, various years).
Conversely, the budget for public agri-food research in Italy The use of R&D intensity as an indicator of the knowledge inten-
has increased throughout the 1998–2008 period (Sorrentino and siveness of the F&D industry has been criticised recently (Rama,
Capozzi, 2010). At the same time, preliminary discussion regarding 2008) because F&D uses advanced technology that is developed
the allocation of regional plan funds for rural development research by high-tech sectors, such as the chemical, pharmaceutical and
has involved representatives not only from the public research and biotechnology sectors, which strongly invest in R&D. Furthermore,
extension systems and the farming and agri-food industries but also investment expenditures in terms of industry innovation expen-
from consumer and environmentalist associations (Ascione et al., ditures are high for F&D because most innovation is embodied in
2006). equipment and capital goods (Rama, 2008). Registering a trade-
The realisation of socially and environmentally sustainable mark is one of the most important forms of innovation in this
production processes is necessarily geographically targeted and industry9 . Other non-R&D inputs to innovation are learning by
context dependent and has yielded a renewed importance of doing and learning by interacting. Interacting with suppliers and
geographical proximity in effective collaboration between firms consumers is important in this sector because of the concern for
and universities. Some examples of the new university research food quality. The recent demand for socially responsible consump-
projects that have been financed, which provide evidence of this tion, which is particularly aimed at the F&D industry, is an example
aspect of collaboration, are those that aim to identify the effect of innovation that stems from interactions with consumers and is
of organic fertilisers on sustainable local cultivation systems and generated by a non-R&D input.
those that aim to identify the local genotypes cultivated in geo- Public regulation is strong in the F&D sector because of certain
graphical areas7 (Andreakis et al., 2004; Sacchi et al., 2010). Of public good attributes of food products, such as food safety and
course, for non-geographically targeted production processes, the health concerns. Public regulation generally alters the incentives
distance between agri-food firms and universities can be substan- for innovation. In this sector, it induces the type of innovation that
tial and can overcome national borders, particularly in the case of is in line with food safety, health enhancement and food quality
multinationals. standards (Ranieri and Silvestri, 2006; Scordamaglia, 2006). This
Generally speaking, because of cuts in public financing from specific attribute of the F&D industry, together with the direct
national sources to Italian universities, the share of university public support from the European Union (EU) Common Agricul-
budgets sourced from R&D collaborations with private firms is tural Policy, can enhance the innovation capacity of small firms
compared with the same capacity of small firms in other sectors.

6
Some universities had already experienced 3-year degrees following Law
8
341/1990. On average, this share was 16% in 2002 and 25% in 2009 (Netval, 2005, 2011).
7 9
This project is intended to inform the disciplinary regulation of products that Among EU countries, Italy has the greatest number of products with a desig-
are protected by a denomination of origin trademark or by a guaranteed origin nated and/or guaranteed denomination of origin trademark (De Devitiis and Maietta,
certification authenticating their historical production areas. 2013).
1346 O.W. Maietta / Research Policy 44 (2015) 1341–1359

Table 1
Innovation expenditure in the Italian supplier-dominated sectors in 2002–2004.

% Innovation expenditure for:

Sectors R&D expenditure R&D intra-muros R&D extra-muros Physical capital Immaterial capital Development Training Mk tg
(ml D )

Per worker

Food and drink 1016 6.9 14.3 3.6 66.5 3.0 1.5 2.0 9.1
Textiles 548 5.6 31.8 1.8 53.0 5.3 3.6 1.5 3.0
Apparel 128 3.4 22.5 0.8 55.8 10.6 3.7 3.9 2.7
Leather and footwear 233 5.9 27.8 0.2 62.2 0.3 6.6 1.1 1.8
Wood 222 6.9 14.6 1.1 69.3 7.0 3.2 2.1 2.7
Paper 364 9.1 14.5 1.7 80.1 1.0 1.5 0.5 0.7
Printing 706 12.0 7.1 2.4 74.2 5.3 3.7 1.2 6.1

Source: Istat, Rilevazione sull’innovazione nelle imprese.

Table 2
F&D firms, workers, turnover, labour unit cost and firm investment by size class in Italy in 2009.

Firm size class No. Firms % No. Workers % Turnover % Unit labour cost Investment per firm

Micro (1–9) 50531 87.5 154315 36.0 16070284 14.3 23.3 38

Small (10–49) 6360 11.0 115562 27.0 28280920 25.2 30.7 174
Medium (50–249) 738 1.3 73204 17.1 31998713 28.5 40.8 1151
Large (>250) 120 0.2 85282 19.9 35967056 32.0 44.7 9656
Total 57749 100.0 428363 100.0 112316973 100.0 34.3 87

Source: ISTAT; size class defined in number of workers, value in th. D .

Farace and Mazzotta (2015), for example, find that the percentage bakeries and confectioneries, which are generally family-run firms,
of innovative firms in the F&D industry is greater than that in other represent 56% of the total F&D sector firms and employ an average
traditional sectors. of 4 workers (INEA, 2006). The presence of micro-sized firms in the
The respect for food safety and food quality standards makes F&D sector also indicates that average R&D intensity is low because
innovation in the F&D industry the result of multidisciplinary of structural factors that shape R&D intensity (Moncada-Paternò-
activity in which different skills (biological, chemical, engineer- Castello et al., 2010) within the sector.
ing, nutritional, economic and legally compliant) are necessary to According to 2011 Eurostat data, Italian F&D firms represent 12%
develop the path from the formulation of an idea to its industrial of all EU F&D firms with at least 10 workers. Looking at the distribu-
realisation.10 tion in size classes of EU F&D firms with at least 10 workers, small
Table 1 reports R&D expenditures for the Italian F&D and other firms are relatively more widespread only in Germany,12 whereas
supplier-dominated sectors (Pavitt, 1984) in 2004. R&D investment the distribution of EU F&D firms with at least 10 workers by sub-
per worker was not particularly high in the F&D industry. However, sector shows that the distribution of Italian F&D firms is similar to
one feature distinguishes the F&D industry with respect to innova- that of other Mediterranean countries, such as Spain.
tion expenditures: the value for the percentage of extra-muros R&D A bipolar structure with a plethora of small firms and a few very
expenditure is the greatest for all sectors. This figure may reflect large companies also characterises the F&D sector of other coun-
the development of a publically supported decentralised agri-food tries, such as Germany. Small firms are strongly present in other
research system (Ruttan, 2001), as in other industrialised coun- sectors as well, such as EU wood, leather and textiles (Pagano and
tries, in which collaboration among firms and universities is more Schivardi, 2003), making the results from this study interesting
widespread than it is for other small and medium-sized firms in the from a broader perspective.
traditional sectors (Istituto Guglielmo Tagliacarne, 2004). In more
recent years, despite an average R&D investment per worker that is 5. The empirical framework
lower than the average investment of all manufacturers (Monducci,
2011), interactions of Italian F&D firms with the scientific commu- 5.1. The data
nity have also been greater than those of other supplier-dominated
industries and of manufacturers as a whole.11 The percentage of The firm data used in this paper are sourced from the “Survey
Italian F&D firms (with at least 10 workers) that interact with uni- of Italian manufacturing firms”. This survey was formerly con-
versities and/or public research labs is greater than that in other ducted by Mediocredito Centrale and is currently conducted by
European countries, including Germany and France (Pasetto, 2011). Capitalia, which are both Italian credit institutions. The analysis is
Table 2, which reports the distribution of Italian F&D firms and built on four waves, which cover the 1995–1997 (7th), 1998–2000
workers by size class, emphasises the sector’s dualistic structure, (8th), 2001–2003 (9th), and 2004–2006 (10th) periods. Each wave
which is characterised by a very small number of medium-sized and includes more than 4000 firms. The survey design includes all firms
large firms and a very large number of micro-sized firms. The huge with at least 500 workers and a sample that is representative of Ital-
percentage of micro-sized firms in the F&D sector is explained by ian manufacturing firms that employ between 10 and 500 workers
and that is stratified by firm size, sector and geographical area.
In the Capitalia surveys, firms are asked whether process, prod-
uct and/or other innovations were introduced during the previous
10
Some examples of F&D innovation include health-enhancing food and drink three years. The questionnaire also collects information regarding
products (Ziggers, 2005) and the use of ozonated water to decrease the amount of
chemicals required for equipment sterilisation (Baregheh et al., 2012).
11
In the 2008-2010 period, according to ISTAT data, 3% of F&D firms had formal
12
interactions with universities or public research labs compared with an average of In 2011, small firms accounted for 66% of the national F&D firms with at least
2% for manufacturers as a whole. 10 workers in Germany and 62% in Italy.
 O.W. Maietta / Research Policy 44 (2015) 1341–1359 1347

whether R&D was intra muros or acquired from external sources and 2004–2010 periods; and the research project and the
such as universities,13 public research labs and other private firms international mobility grades (measured annually), which were
along with other firm characteristics, such as the presence of skilled kindly offered by Censis for the 1998–2006 period.
employees (that is graduates), non-standard jobs14 and subsidies.15 The academic research quality of each agricultural studies fac-
Using their ATECO classification, F&D firms have been extracted, ulty is measured by the VQR grade. The VQR grade is a composite
resulting in a pool of 1744 firms for the 1995–2006 period. After indicator of the quality of the research output produced by uni-
checking ex post representativeness, it was determined that the versities and/or public research labs under the supervision of the
derived sample is representative of Italian F&D firm by region, as Higher Education Department during the evaluation period. Groups
shown in Table A1 in Appendix A. of Experts of Evaluation, which are coordinated by the National
Size classes have been defined following the AGRA (2004) clas- Agency for the Evaluation of Universities and Research Institutes,
sification with respect to turnover thresholds, which are expressed evaluated the research output using both bibliometric analysis and
in constant 2006-based D : very small firms <5 ml; small firms informed peer review. There is evidence that these two evalua-
between 5 and 25 ml; medium-sized firms between 25 and 50 ml; tion systems give similar grades for the same set of journal articles
large firms between 50 and 100 ml; and very large firms ≥100 ml. (Bertocchi et al., 2015).
Information about the municipality in which the firm is located The VQR evaluation is unavailable for the 1995–2000 period;
or, in its absence, of the province (as in Benfratello et al., 2008) alternatively, a battery of indicators of academic research quality
have been used to identify the three closest faculties of agricul- and of society’s perception of a faculty’s reputation has been used
tural studies. The choice to focus on these faculties is supported by for the entire period. These latter have been added because less
the evidence that most university collaborations with F&D firms innovative firms may adopt reputation indicators not based on aca-
are with the regional faculty of agricultural studies. Furthermore, demic research quality which is the criterion of leader firms, at the
a firm that has university collaborations is likely to have multiple forefront in their industry, to select universities as R&D partners.
university or public research lab partners (Avermaete et al., 2004; Academic research quality is proxied by two indicators of
Bodas Freitas et al., 2011), and the probability that one of these codified knowledge production, the numbers of ISI-Scopus articles
partners is the regional agricultural studies faculty is very high. The and citations. The two indicators of codified knowledge are built
choice of three universities follows Laursen et al. (2011). For each using the medians of the ISI-Scopus indexed scientific production,
firm, three distances in kilometres from each of the three relevant which is measured by the number of articles and citations in the
faculties have been downloaded.16 As usual (D’Este et al., 2013; populations of full professors20 of the Italian agricultural studies
Laursen et al., 2011), these distances between the firm and the fac- faculties, grouped by scientific discipline over the 2002–2012
ulty main location are measured as the crow flies. The choice of the period. No other measure of scientific production was operational
faculty’s main location is based on the evidence that research labs because scholars’ names for the 1995–1999 period are unavailable
are located there, even if specific courses might sometimes be held on the website. The use of the medians that referred to the
in peripheral towns. A fourth variable for geographical proximity 2002–2012 period is based on the assumption that the differences
is a dummy that takes the value of 1 if the closest faculty of agri- among scientific disciplines in the median production of ISI-Scopus
cultural studies is more than 150 km away; this value was chosen indexed journals have not changed materially with respect to the
after testing at different thresholds.17 1995–2001 period.
With respect to the closest agricultural studies faculty, the Society’s perception of a faculty’s reputation is proxied by the
following information was also gathered: whether the faculty is research project and the international mobility grades (which are
extra-regional; whether it is public; the year in which it was used to compute the annual faculty rankings), the faculty’s age and
established; its size in terms of researchers/professors (computed the percentage of women who are full professors, which is an indi-
annually)18 ; the annual composition of researchers/professors in rect expression of meritocratic versus hierarchical institutions. The
terms of (i) gender, (ii) birth year, (iii) carrier status (researchers, research project grade (which is based on the number of research
associate and full professors), and (iv) scientific disciplines; the projects financed by national and international institutions) and the
annual number of graduates (ISTAT Statistiche sulla Ricerca international mobility grade (hereafter, international grade), which
Scientifica, various years); the presence of a food technologist 3- is based on the international mobility of scholars and students, have
year degree programme and a food technologist 5-year degree been supplied for the 1998–2006 period. This information is miss-
programme19 (Ministero dell’Università e della Ricerca Scientifica, ing for the 1995–1997 period; thus, the two grades for 1998 have
various years); the presence and the birth year of the university been used for the first period. For the remaining periods, the two
technology transfer office, which were kindly offered by Netval and grades are the average of the grades for the three corresponding
are used as proxies of academic policies that are oriented towards years.
the commercial exploitation of research results; the Italian Evalua- Related to the university regional supply are the number of
tion of Research Quality, hereafter, VQR grades for the 2001–2003 biotechnologist degree programmes (Ministero dell’Università e
della Ricerca Scientifica, various years; ISTAT, various years; INEA,
various years) and the number of faculties, which is introduced
13
to capture agglomeration externalities and the social capital com-
Only the last wave provided information about whether the universities are
regional. For the 2004–2006 period, 4 F&D firms had R&D collaborations with extra-
ponent of university–firm interactions through the creation of
regional universities. networks between industry and government.
14
These are defined as open-ended part-time, fixed-term part-time and fixed-term Territorial characteristics are sourced from ISTAT (2005) for
full-time jobs. the agricultural and food districts and by INEA (various years)
15
Any financial subsidy for applied research and technological innovation via Ital-
for regional R&D, which is measured as the amount of accredited
ian national laws, such as laws 46/1982 and 297/1999, or EU and regional laws.
16
http://distanzechilometriche.net/. funds at constant 2006-based prices normalised by the number of
17
Alternatively, focusing on the faculty of the same province/region in which the regional F&D firms. The latter is used as a proxy of regional R&D
firm is located was not operational because two regions and most provinces do not intensity.
host an agricultural studies faculty, and the geographical distance from the faculty
would be not computable in such cases.
18
http://cercauniversita.cineca.it and http://www.cnvsu.it/.
19
Since the reform that was ushered in with Ministerial Decree 509/1999, the
20
programme consists of a 3-year degree followed by a 2-year specialisation degree. http://abilitazione.miur.it/.
1348 O.W. Maietta / Research Policy 44 (2015) 1341–1359

5.2. The variables years in each period, is low at 0.28% of turnover. Two features that
are idiosyncratic to the sector examined are the relatively high
Literature recommends that the empirical framework should presence of cooperatives (17% of firms) and the distribution chain
take into account the interdependencies between innovations agreements (25.49% of sales occur through this channel). Nearly a
and external collaborations in R&D while addressing simultaneity quarter of the firms in the sample use non-standard jobs because
between innovations and (internal and external) R&D investment of the introduction in Italy of flexibility legislation that lowered the
decisions and the simultaneity between different forms of external cost for firms of this form of labour. Nearly half of the firms are
collaborations in R&D. subsidised, particularly those located in southern Italy because of
In order to allow for this, the econometric model of the paper legislation that assists disadvantaged areas. The average distance
consists of six simultaneous equations related to the following from the closest faculty of agricultural studies is 47.71 km, whereas
dependent variables: (the existence of) intra muros R&D invest- the third-closest faculty is only 144.71 km away, on average. The
ment, R&D collaboration with universities, R&D collaboration with short distance between F&D firms and agricultural studies faculties
public labs, R&D collaboration with private firms, process innova- is partly because the study chose to use linear distance but mainly
tion and product innovation. Among these, the variables of R&D reflects the presence of a highly decentralised agri-food research
collaboration with universities, R&D collaboration with public labs system based on agricultural studies faculties. Only 2% of F&D firms
and R&D collaboration with private firms are also used as regres- are more than 150 km from an agricultural studies faculty; gener-
sors. All these indicators are binary variables. ally, very small firms are located closer to these faculties.
The empirical specification of the six equations can be summed Intermediation structures, when they are present, are recent.
up as follows: The first two technology transfer offices of the universities included
Intra muros R&D investment = f1 (R&D collaboration with uni- in the data set were established only in 2001. The average agricul-
versities, R&D collaboration with public research labs, R&D collab- tural studies faculty is 50 years old, graduates 167 students per
oration with private firms, skilled employees, non-standard jobs, year and employs 110 researchers or professors. The gender glass
co-op dummy, subsidies, firm-size dummies, year dummies, terri- ceiling is present in that women account for only 11% of full profes-
tory characteristics, geographical distance, and university/faculty sors, but it is absent in young faculties, which also employ relatively
characteristics). few researchers. The choices that are related to the combination of
R&D collaboration with partnerm = fk (extra muros R&D inten- research and didactics differ from faculty to faculty, partly reflect-
sity from partnern , extra muros R&D intensity from partnero , ing the regional productive structures sine faculties are specialised
intra muros R&D intensity, skilled employees, non-standard jobs, in the same technical fields as local firms. However, 0.63% only
co-op dummy, subsidies, firm-size dummies, year dummies, terri- refers to industrial engineers because food engineers are included
tory characteristics, geographical distance, and university/faculty in the (not reported) ample residual macro-area of food scientists.
characteristics21 ),where m, n, o = universities, public research labs, Regional governments devoted, as a period average, 2180 D per
private firms and m = / n =/ o and k = 2, 3, 4. F&D firm, at constant 2006-based prices, to agri-food research. The
Product innovation = f5 (R&D collaboration with universities amount increased over the years.
and/or public research labs,22 R&D collaboration with pri- The key regressors in all equations are related to the “knowledge
vate firms, R&D intensity, skilled employees, non-standard jobs, context” represented by the closest agricultural studies faculty.
co-op dummy, sales through distribution chains, subsidies, firm- First of all, the geographical distances from the three closest fac-
size dummies, year dummies, sub-sector dummies, territory ulties have been alternatively tested, in models 1–4 of Tables 4–10,
characteristics, geographical distance, and university/faculty char- in order to choose the appropriate specification of this variable
acteristics). for each equation. The description of the “knowledge context”
Product innovation = f6 (R&D collaboration with universities begins by adding the following key independent variables related
and/or public research labs, R&D collaboration with private to the closest agricultural studies faculty only: food technologist
firms, R&D intensity, skilled employees, non-standard jobs, 3-year and 5-year degree programmes, the society’s perception of
co-op dummy, sales through distribution chains, subsidies, firm- a faculty’s reputation indicators and the intermediation structure
size dummies, year dummies, sub-sector dummies, territory indicators (technological transfer office or technological transfer
characteristics, geographical distance, and university/faculty char- office’s age) in model 5 of Tables 4–10. The description is com-
acteristics). pleted by the three academic quality indicators, which have been
Table 3 reports the descriptive statistics of the variables, which alternatively tested in models 6–8 of Tables 4–10.
indicate that 5% of the firms in the sample have R&D collaborations
with a university, the same percentage have R&D collaborations 5.3. The econometric approach
with a public research lab and 9% of firms have R&D collaborations
with private firms that are, on average, three times as expensive Following what has been said at the outset of § 5.2, the econo-
than the two previous ones. R&D collaborations decrease in the metric model consists of six simultaneous equations that are jointly
last sub-period, particularly those with public research labs and, modelled as a multivariate probit system. The model follows a six-
to a lesser extent, those with universities. Among all firms in the equation structure in which the estimation results of the second,
sample, 34% have introduced product innovation, and 49% have third and fourth equations are used as regressors in the first, fifth
introduced process innovation. R&D intensity, which is measured and sixth equations, as follows:
as the average ratio of R&D expenditures to sales over the three ⎧ ∗ ∗ +  y∗ +  y∗ + x  ˇ + 
⎪
⎪ y1i = 12 y2i 13 3i 14 4i 1i 1 1i
⎪
⎪
⎪
⎪ ∗ = x ˇ + 
⎪
⎪
y 2i 2i 2 2i
⎪
⎪
21
⎨ y∗ = x ˇ3 + 3i
Scholars’ characteristics are included in the university-firm R&D collaboration 3i 3i
, (1)
equation only since the literature suggests that these characteristics may impact
⎪
⎪ ∗ = x ˇ + 
y4i
university-firm collaboration. ⎪
⎪ 4i 4 4i
22
This specification is preferred in the two innovation equations since the two ⎪
⎪
⎪
⎪
∗ ∗ 
y5i = 54 y4i + x5i ˇ5 + 5i
variables R&D collaboration with universities and R&D collaboration with public ⎪
⎪
research labs are weakly or not significant when used separately because they are ⎩ ∗ ∗ 
strongly collinear.
y6i = 64 y4i + x6i ˇ6 + 6i
 O.W. Maietta / Research Policy 44 (2015) 1341–1359 1349

Table 3
Variables and descriptive statistics.

Variables Mean Std. Dev.

Firm characteristics
R&D collaboration with universities dummy 0.05 0.22
R&D collaboration with public labs dummy 0.05 0.22
R&D collaboration with universities and/or public research labs dummy 0.09 0.28
R&D collaboration with private firms dummy 0.09 0.29
Intra muros R&D investment dummy 0.27 0.44
Product innovation dummy 0.34 0.48
Process innovation dummy 0.49 0.50
R&D intensity (% turnover) 0.28 1.23
Intra muros R&D intensity (% turnover) 0.24 1.34
Extra muros R&D intensity from universities (% turnover) 0.01 0.07
Extra muros R&D intensity from public labs (% turnover) 0.01 0.06
Extra muros R&D intensity from private firms (% turnover) 0.03 0.18
Skilled employees (%) 5.13 7.82
Sales through distribution chain agreement (%) 25.49 34.62
Subsidies dummy 0.48 0.50
Non standard job dummy 0.24 0.49
Co-op firm dummy 0.17 0.38
Firm age (years) 30.96 24.09
Very small-sized firm dummy 0.31 0.46
Small-sized firm dummy 0.51 0.50
Medium-sized firm dummy 0.08 0.27
Large-sized firm dummy 0.05 0.21
Meat processing dummy 0.16 0.36
Fruit&vegetables processing dummy 0.12 0.33
Dairy products’manufacture dummy 0.18 0.38
Grain mil and starch products manufacture dummy 0.07 0.25
Prepared animal feeds manufacture dummy 0.05 0.22
Beverage manufacture dummy 0.19 0.39
Oils and fats manufacture dummy 0.04 0.19
Fish processing dummy 0.03 0.16

Territorial characteristics
North dummy 0.52 0.50
South dummy 0.35 0.48
Food district dummy 0.08 0.26
Agricultural district dummy 0.03 0.16

Regional R&D
Accredited funds (constant 2006-based th D /No. F&D firms) 2.18 3.12

Geographical distances
1st distance (km) 47.71 37.24
2nd distance (km) 108.94 74.40
3rd distance (km) 144.71 81.40
Distance >150 km dummy 0.02 0.15

University characteristics
No. biotechnologist degree programmes 0.61 0.49
No. regional faculties of agricultural studies 1.53 1.00
Public university dummy 0.97 0.18
Technological transfer office dummy 0.22 0.41
Technological transfer office’s age 1.80 1.92

Faculty characteristics
Extra-regional faculty of agricultural studies dummy 0.12 0.32
Food technologist 3-year degree programme dummy 0.54 0.50
Food technologist 5-year degree programme dummy 0.42 0.49
Faculty of agricultural studies’ age (years) 50.00 24.93
No. researchers/professors 109.93 51.55
No. graduates 166.77 127.02
Women on full professors (%) 10.74 9.85
Researchers on total researchers/professors (%) 34.56 10.39
Average age of researchers/professors 48.19 4.68
No. scientific macro-fields 5.56 1.81
Industrial engineers on total scholars (%) 0.63 1.53
Biologists on total scholars (%) 8.53 10.79
Chemicals on total scholars (%) 5.98 8.03
Physicians on total scholars (%) 1.03 3.71
Geologists on total scholars (%) 1.11 2.08
International grade 64.62 28.39
Research project grade 82.46 16.49
VQR grade 68.43 9.03
Codified knowledge indicator (No. journal articles) 18.57 1.76
Codified knowledge indicator (No. citations) 14.04 4.00

Temporal dummies
Dummy for 1998–2000 0.29 0.45
Dummy for 2001–2003 0.29 0.46
Dummy for 2004–2006 0.18 0.38
1350 O.W. Maietta / Research Policy 44 (2015) 1341–1359

Table 4
Significance and value of the correlation coefficients among the errors of the Eqs. (1)–(6).

Coefficients Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8

Rho21 −0.01 −0.01 −0.01 −0.01 0.02 0.03 0.03 −0.16a

Rho31 0.05 0.05 0.04 0.05 0.05 0.06 0.06 0.08a
Rho41 0.07 0.07 0.07 0.07 0.10 0.09 0.09 −0.21
Rho51 0.06 0.06 0.06 0.07 0.06 0.06 0.06 −0.06
Rho61 0.09* 0.09* 0.10* 0.10* 0.08 0.09 0.09 0.02
Rho32 0.50* 0.50*** 0.50*** 0.50*** 0.52*** 0.54*** 0.54*** 0.97a
Rho42 0.48* 0.47*** 0.47*** 0.47*** 0.48*** 0.50*** 0.50*** 0.56***
Rho52 −0.11 −0.12 −0.12 −0.11 −0.11 −0.11 −0.11 0.23
Rho62 −0.22*** −0.22*** −0.22*** −0.22*** −0.20*** −0.20*** −0.20*** 0.20
Rho43 0.29** 0.28** 0.28** 0.29** 0.32*** 0.32** 0.32** 0.53***
Rho53 −0.06 −0.06 −0.06 −0.06 −0.05 −0.05 −0.05 0.20
Rho63 −0.07 −0.07 −0.07 −0.07 −0.07 −0.07 −0.07 0.21
Rho54 −0.13* −0.12* −0.12* −0.13** −0.12* −0.11 −0.11 0.23
Rho64 −0.27*** −0.27*** −0.27*** −0.28*** −0.24*** −0.24*** −0.24*** 0.26
Rho65 0.44*** 0.44*** 0.44*** 0.43*** 0.44*** 0.44*** 0.44*** 0.50
a
In the case of model 8 only, the software did not provide significance levels for the rhos.

Table 5
Multiprobit regression. Marginal effects for the dependent variable (existence of) intra muros R&D investment.

Variables dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8

R&D collaborations with universities 0.55*** 0.55*** 0.55*** 0.55*** 0.54*** 0.54*** 0.54*** 0.53***
R&D collaborations with public labs 0.49*** 0.49*** 0.50*** 0.50*** 0.50*** 0.50*** 0.50*** 1.01***
R&D collaborations with private firms 0.41*** 0.41*** 0.41*** 0.41*** 0.40*** 0.40*** 0.41*** 0.48***
Skilled employees 0.003** 0.003** 0.003** 0.003** 0.003** 0.003** 0.003** 0.004***
Co-op firm −0.08*** −0.08*** −0.08*** −0.08*** −0.09*** −0.09*** −0.09*** −0.07**
Subsidies 0.06*** 0.06*** 0.06*** 0.06*** 0.06*** 0.06*** 0.06*** 0.10***
Non standard jobs 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02
Firm age 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Very small-sized firm −0.11*** −0.10*** −0.10*** −0.11*** −0.10*** −0.10*** −0.10*** −0.07
Small-sized firm −0.06* −0.06 −0.06 −0.06* −0.05 −0.05 −0.06 −0.02
Medium-sized firm −0.03 −0.03 −0.03 −0.03 −0.02 −0.02 −0.03 0.00
Large-sized firm 0.04 0.04 0.04 0.04 0.05 0.05 0.04 −0.02
North 0.01 0.01 0.01 0.01 0.00 0.00 0.00 0.05
South −0.02 0.00 −0.01 −0.02 −0.05 −0.03 −0.03 −0.08
Food district −0.02 −0.02 −0.02 −0.02 −0.02 −0.02 0.02 0.05
Agricultural district −0.01 0.01 0.01 0.00 0.01 0.00 0.00 −0.12
1st distance 0.000
2nd distance −0.0002** 0.00 0.00 0.00 0.00
3rd distance −0.0001*
Distance > 150 km 0.03
Regional R&D – Accredited funds 0.00 0.00 0.00 −0.001***
No. biotechnologist degree programmes 0.04 0.04 0.04 0.01
Food technologist degree 5-year programme −0.02 −0.02 −0.01 0.05
Food technologist degree 3-year programme −0.01 −0.01 −0.01 −0.10**
Extra-regional faculty of agricultural studies 0.02 0.03 0.03 −0.04
Faculty of agricultural studies’ age −0.0001 0.00 0.00 0.00
No. researchers/professors 0.0003 0.00 0.00 0.00
No. graduates 0.0001 0.00 0.00 0.00
Industrial engineers on total scholars 0.01** 0.01** 0.01** 0.03***
Biologists on total scholars 0.00 0.00 0.00 0.00
Chemicals on total scholars 0.00 0.00 0.00 0.00
Physicians on total scholars 0.00 0.004*** 0.01** 0.01
Geologists on total scholars 0.00 0.00 0.00 0.01
No. scientific macro-fields −0.02*** −0.02** −0.02** −0.01
No. regional faculties of agricultural studies 0.00 0.01 0.00 −0.04***
Public univerisity −0.03 −0.02 −0.02 −0.14*
Technological transfer office −0.08** −0.08** −0.07** −0.05
Technological transfer office’s age 0.00
International grade 0.00 0.001**
Research project grade 0.00 0.00
Codified knowledge indicator (No. articles) −0.02*
Codified knowledge indicator (No. citations) −0.01*
VQR grade 0.00
Year dummies Yes Yes Yes Yes Yes Yes Yes Yes
No. observations 1531 1531 1531 1531 1531 1531 1531 719
LogL −3176.0 −3178.6 −3177.2 −3175.6 −3115.4 −3101.8 −3101.6 −1421.4
***
Signficant at 1% level.
**
Signficant at 5% level.
*
Signficant at 10% level.
Table 6
Multiprobit regression. Marginal effects for the dependent variable R&D collaboration with universities.

Variables dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx

Model 1 Model 2 Model 3 Model 4 Model 5 Model 6 Model 7 Model 8

Extra muros R&D intensity from public labs −0.04 −0.04 −0.04 −0.04 −0.05 −0.05 −0.06 −0.22**
Extra muros R&D intensity from private firms −0.05 −0.05 −0.04 −0.04 −0.04 −0.04 −0.04 −0.05
Intra muros R&D intensity 0.01* 0.01* 0.01* 0.01* 0.01** 0.01** 0.01* 0.01***
Skilled employees 0.002*** 0.002*** 0.002*** 0.002*** 0.002*** 0.002*** 0.002*** 0.003***
Co-op firm −0.03* −0.03* −0.03* −0.03* −0.03** −0.03* −0.03* −0.02
Subsidies 0.04*** 0.04*** 0.04*** 0.04*** 0.04*** 0.04*** 0.04*** 0.05***
Non standard jobs 0.01 0.01 0.01 0.01 0.00 0.00 0.01 0.01
Firm age 0.0004** 0.0004** 0.0004** 0.0004** 0.0004** 0.0004** 0.0004** 0.0006**
Very small-sized firm −0.10*** −0.10*** −0.10*** −0.10*** −0.09*** −0.09*** −0.09*** −0.08**
Small-sized firm −0.04 −0.04 −0.04* −0.04* −0.03 −0.03 −0.03 −0.01
Medium-sized firm −0.03 −0.03 −0.03 −0.03 −0.02 −0.02 −0.02 −0.04
Large-sized firm −0.03 −0.03 −0.03 −0.03 −0.03 −0.03 −0.04 −0.01
North 0.00 0.00 0.00 0.00 −0.05*** −0.05*** −0.05*** −0.06
South 0.00 0.00 0.00 0.00 −0.02 −0.03 −0.03 0.02
−0.03 −0.03 −0.03 −0.03 −0.03 −0.02 −0.02 −0.01

O.W. Maietta / Research Policy 44 (2015) 1341–1359

Food district
Agricultural district 0.02 0.01 0.01 0.02 0.02 0.01 0.01 0.01
1st distance 0.00
2nd distance 0.00
3rd distance 0.00
Distance > 150 km 0.04* 0.04*** 0.04*** 0.04*** 0.03
Regional R&D – Accredited funds −0.004** −0.003** −0.003** −0.005*
No. biotechnologist degree programmes 0.01 0.01 0.01 −0.01
Food technologist degree 5-year programme 0.04** 0.05*** 0.05*** 0.04
Food technologist degree 3-year programme −0.03** −0.03** −0.03** −0.03
Extra-regional faculty of agricultural studies 0.00 0.01 0.01 0.05*
Faculty of agricultural studies’ age 0.00 0.00 0.00 0.00
No. researchers/professors 0.001** 0.001* 0.001* 0.00
No. graduates 0.00 0.0002* 0.0002* 0.00
Industrial engineers on total scholars 0.00 0.00 0.00 0.01
Biologists on total scholars 0.00 0.00 0.00 0.006**
Chemicals on total scholars −0.002** −0.003** −0.003** −0.02***
Physicians on total scholars 0.00 0.00 0.00 0.00
Geologists on total scholars 0.01** 0.01** 0.01** 0.01
Women on full professors 0.002** 0.002∗∗
0.002** 0.002***
Researchers on total researchers/professors −0.002*** −0.003*** −0.003*** 0.00
Average age of researchers/professors 0.00 0.00 0.00 −0.01
No. scientific macro-fields 0.00 0.00 0.00 0.00
No. regional faculties of agricultural studies −0.01** −0.01* −0.01* 0.00
Public univerisity −0.14*** −0.14*** −0.14*** −0.07
Technological transfer office 0.03 0.03 0.03 −0.02
Technological transfer office’ s age 0.00
International grade 0.00 0.00
Research project grade 0.00 0.00
Codified knowledge indicator (No. articles) 0.01
Codified knowledge indicator (No. citations) 0.01*
VQR grade 0.01*
Year dummies Yes Yes Yes Yes Yes Yes Yes Yes
***
Signficant at 1% level.
**
Signficant at 5% level.
*
Signficant at 10% level.

1351
1352 O.W. Maietta / Research Policy 44 (2015) 1341–1359

Table 7
Multiprobit regression. Marginal effects for the dependent variable R&D collaboration with public research labs.

Variables dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx

Model1 Model2 Model3 Model4 Model5 Model6 Model7 Model8

Extra muros R&D int. from universities −0.07 −0.07 −0.07 −0.07 −0.07 −0.08 −0.08 −0.12***
Extra muros R&D int. from private firms −0.07 −0.07 −0.07 −0.07 −0.07 −0.07 −0.07 −0.70
Intra muros R&D intensity 0.01* 0.01* 0.01* 0.01* 0.00 0.01 0.00 0.00
Skilled employees 0.001** 0.001** 0.001** 0.001** 0.001** 0.001** 0.001** 0.003***
Co-op firm −0.04** −0.04** −0.04** −0.04** −0.04** −0.04** −0.04** −0.29
Subsidies 0.04*** 0.04*** 0.04*** 0.04*** 0.03*** 0.03** 0.03** 0.05***
Non standard jobs 0.02** 0.02** 0.02** 0.02** 0.02** 0.02** 0.02** 0.00
Firm age 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Very small-sized firm −0.04*** −0.04** −0.04** −0.04** −0.04** −0.07** −0.04** −0.32
Small-sized firm −0.01 0.00 −0.01 0.00 −0.01 −0.01 −0.01 0.04
Medium-sized firm 0.00 0.00 0.00 0.00 −0.01 −0.01 −0.01 0.00
Large-sized firm −0.02 −0.02 −0.02 −0.02 −0.02 −0.02 −0.02 0.00
North 0.00 0.00 0.00 0.00 −0.01 0.00 0.00 0.05
South −0.03* −0.04* −0.04* −0.04* −0.05*** −0.07*** −0.07*** −0.08***
Food district 0.01 0.01 0.01 0.01 0.00 0.00 0.00 −0.05
Agricultural district −0.02 −0.02 −0.03 −0.02 −0.02 −0.02 −0.02 0.01
1st distance 0.0003***
2nd distance 0.00
3rd distance 0.0001***
Distance > 150 km 0.05*** 0.03** 0.03** 0.03** 0.05
Regional R&D – Accredited funds 0.002* 0.002** 0.002** 0.002*
No. biotechnologist degree programmes 0.00 0.00 0.00 −0.24***
Food technologist degree 5-year programme 0.03*** 0.03*** 0.03*** −0.17**
Food technologist degree 3-year programme −0.02 −0.01 −0.01 −0.04
Extra-regional faculty of agricultural studies 0.00 0.00 0.00 −0.32***
Faculty of agricultural studies’ age −0.001*** −0.001*** −0.001*** 0.003**
No. researchers/professors 0.00 0.00 0.00 0.00
No. graduates 0.00 0.00 0.00 0.00
Industrial engineers on total scholars 0.00 0.00 0.00 0.02
Biologists on total scholars 0.00 0.00 0.00 0.01
Chemicals on total scholars −0.002*** −0.004*** −0.004*** 0.01
Physicians on total scholars 0.002** 0.002* 0.002* 0.04*
Geologists on total scholars 0.00 0.00 0.00 0.06***
No. scientific macro-fields 0.00 0.00 0.00 0.00
No. regional faculties of agricultural studies −0.01** −0.01** −0.01** −0.12**
Public univerisity 0.02 0.03 0.03 −0.61***
Technological transfer office -0.01 -0.01 -0.01 −0.04*
Technological transfer office’s age −0.02
International grade −0.001** −0.001**
Research project grade 0.001*** 0.001***
Codified knowledge indicator (No. articles) 0.01
Codified knowledge indicator (No. citations) 0.00
VQR grade 0.00
Year dummies Yes Yes Yes Yes Yes Yes Yes Yes
***
Signficant at 1% level.
**
Signficant at 5% level.
*
Signficant at 10% level.

where y1i *, y2i *, y3i *, y4i *, y5i * and y6i * are latent variables defined y*k where m = universities, public research labs, private firms and
as follows: y1 * is intra muros R&D investment; y2 * are R&D collab- k = 2, 3, 4; and product and process innovations >0 for y*5 and y*6 ,
orations with universities; y3 * are R&D collaborations with public respectively.
research labs; y4 * are R&D collaborations with private firms; y5 * are The equations that refer to y1 , y2 , y3 and y4 have been included
product innovations and y6 * are process innovations; xki are vectors to identify the determinants of the presence of intra and extra muros
of exogenous variables, which influence those probabilities for firm R&D investment that aims at introducing product or process inno-
i; ˇk are parameter vectors;  kl are scalar parameters; and ki are vations and to take into account the simultaneity of firm decisions
error terms, which are assumed to be jointly normal with unknown relating to the types of intra and extra muros R&D investments. Fur-
correlation coefficients kl and correlated with something else in thermore, the common latent factor structure of the multivariate
the model. The covariate vectors xki are not restricted to containing probit framework allows to control for the potential endogeneity
the same variables of interest as long as there is at least one varying of the R&D investment decision and to correct potential sample
exogenous regressor in each equation in system (1) (Wilde, 2000). selection.
The realisation of the latent variables yki * is not observed; how- The resulting recursive multivariate probit model can be esti-
ever, the realisation of the binary variables yki is observed, and these mated using a simulated maximum likelihood method.23
are linked to the former according to the following rule:

∗ >0
yki = 1, if yki
. (2) 23
yki = 0 otherwise; k = 1,. . ., 6 No exclusion restrictions on the exogenous variables, in recursive multiple equa-
tion probit models with endogenous dummy regressors, are required for parameter
identification when there is sufficient variation in the data. This last condition is
The binary variables are equal to 1 when intra muros R&D invest- ensured by the assumption that each equation contains at least one varying exoge-
ment >0 for y*1 , extra muros R&D expenditure with partnerm > 0 for nous regressor (Wilde, 2000).
 O.W. Maietta / Research Policy 44 (2015) 1341–1359 1353

Table 8
Multiprobit regression. Marginal effects for the dependent variable R&D collaboration with private firms.

Variables dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx

Model1 Model2 Model3 Model4 Model5 Model6 Model7 Model8

Extra muros R&D int. from universities −0.25** −0.26** −0.26** −0.26** −0.24** −0.24** −0.24** −0.22
Extra muros R&D int. from public labs −0.15 −0.15 −0.15 −0.15 −0.13 −0.14 −0.14 −1.72**
Intra muros R&D intensity 0.01* 0.01* 0.01* 0.01* 0.01* 0.01* 0.01* 0.02***
Skilled employees 0.002** 0.002** 0.002** 0.002** 0.002** 0.002** 0.002** 0.004***
Co-op firm −0.05** −0.05** −0.05** −0.05** −0.05** −0.06*** −0.06*** −0.02
Subsidies 0.07*** 0.07*** 0.07*** 0.07*** 0.07*** 0.07*** 0.07*** 0.11***
Non standard jobs 0.02 0.02 0.02 0.02 0.02 0.02 0.02 0.02
Firm age 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Very small-sized firm −0.07 −0.07 −0.07 −0.07 −0.06 −0.06 −0.06 −0.03
Small-sized firm −0.06 −0.06 −0.06 −0.06 −0.06 −0.06 −0.06 −0.02
Medium-sized firm 0.00 0.00 0.00 0.00 0.00 0.01 0.01 0.02
Large-sized firm 0.06 0.06 0.06 0.06 0.07 0.07 0.07 0.07*
North 0.02 0.02 0.02 0.02 −0.02 −0.03 −0.03 0.01
South −0.04** −0.04** −0.04** −0.04** −0.06* −0.08** −0.08** −0.02
Food district −0.02 −0.02 −0.02 −0.02 −0.01 −0.01 −0.01 0.00
Agricultural district 0.05 0.05 0.05 0.05 0.05 0.04 0.04 −0.05
1st distance 0.00
2nd distance 0.00
3rd distance 0.00
Distance > 150 km 0.02 0.00 0.01 0.01 0.04
Regional R&D – Accredited funds −0.01** −0.01** −0.01** −0.01**
No. biotechnologist degree programmes 0.03 0.03 0.03 0.06**
Food technologist degree 5-year programme 0.05** 0.05** 0.05** 0.01
Food technologist degree 3-year programme −0.03 −0.03 −0.03 −0.07*
Extra-regional faculty of agricultural studies 0.00 0.00 0.00 0.01
Faculty of agricultural studies’ age 0.001* 0.00 0.00 0.002*
No. researchers/professors −0.001** 0.00 0.00 0.00
No. graduates 0.00 0.00 0.00 −0.0003**
Industrial engineers on total scholars −0.02* −0.02* −0.02* −0.02*
Biologists on total scholars −0.002*** −0.002*** 0.00 −0.006*
Chemicals on total scholars 0.00 0.00 0.00 0.01**
Physicians on total scholars 0.00 0.00 0.00 −0.07***
Geologists on total scholars 0.01*** 0.01*** 0.01*** 0.04**
No. scientific macro-fields 0.00 0.01 0.01 0.02**
No. regional faculties of agricultural studies 0.00 −0.01 −0.01 −0.04**
Public univerisity −0.02 −0.03 −0.03 −0.17**
Technological transfer office −0.04** −0.03 -0.03 −0.04*
Technological transfer office’s age 0.00
International grade 0.00 0.00
Research project grade 0.00 0.00
Codified knowledge indicator (No. articles) 0.00
Codified knowledge indicator (No. citations) 0.00
VQR grade 0.00
Year dummies Yes Yes Yes Yes Yes Yes Yes Yes
***
Signficant at 1% level.
**
Signficant at 5% level.
*
Signficant at 10% level.

6. The empirical evidence rately for Eq. (1) plus a multivariate five-equation structure run
for Eqs. (2)–(6).24
The results of the multivariate probit regressions are reported The significance and high values of ␳32 show a high correla-
for various specifications (including different subsets of regressors) tion between R&D university–firm collaborations and R&D public
in Tables 4–10. The standard errors (not reported) of the coeffi- research lab-firm collaborations. The correlation among the errors
cients have been clustered around the regions in which the firm is of the equations is also significant and strong for ␳65 and increases
located because the institutional setting is homogenous within the in the final period.
same region given that regional governments are responsible for Table 5 reports the marginal effects for Eq. (1). The results of
implementing agri-food policies. the regressions for the entire period (models 1–7) are mainly dis-
The likelihood ratio test, which was conducted on the hypoth- cussed. Model 1 and model 4 are virtually identical, although model
esis that the ␳s are jointly null, is highly significant and supports 4’s log likelihood value is higher. Hence, model 4 is chosen for fur-
the multivariate six-equation framework. The value of the statis- ther development in the paper. Regarding geographical proximity,
tics for the specification of variables relative to model 4 of the 2nd and 3rd distance variables from the agricultural studies
Tables 5–10 is equal to 178.28 with 15 degrees of freedom, faculties are negative and are significant and weakly significant,
compared with a critical value of 30.58 at the 1% significance respectively. Being closer to more agricultural studies facul-
level. As all the correlation coefficients relative to Eq. (1) are, ties induces intra muros R&D investment; however, the distance
at the best, weakly significant, a second test, suggested by
Monfardini and Radice (2006), was conducted on the multivariate
six-equation framework against the univariate probit run sepa-
24
The value of the statistics for the specification of variables relative to model 4
of Tables 5–10 is equal to 13.53 with 5 degrees of freedom, compared with a critical
value of 11.07 at the 5% significance level.
1354 O.W. Maietta / Research Policy 44 (2015) 1341–1359

Table 9
Multiprobit regression. Marginal effects for the dependent variable product innovation

Variables dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx

Model1 Model2 Model3 Model4 Model5 Model6 Model7 Model8

R&D collaborations with universities and/or public labs 0.13* 0.13* 0.13* 0.13* 0.13* 0.13* 0.13* −0.05
R&D collaborations with private firms 0.20*** 0.20*** 0.20*** 0.20*** 0.19*** 0.19*** 0.19*** −0.04
R&D intensity 0.02* 0.02* 0.02* 0.02* 0.02* 0.02* 0.02* 0.02**
Sales through distribution chain 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Skilled employees 0.002* 0.002* 0.002* 0.003** 0.003** 0.003** 0.003** 0.003**
Co-op firm −0.04** −0.04** −0.04** −0.04** −0.05** −0.05** −0.05** −0.04
Subsidies 0.16*** 0.16*** 0.16*** 0.16*** 0.16*** 0.17*** 0.17*** 0.34***
Non standard jobs 0.03 0.03 0.03 0.03 0.03 0.03 0.03 0.05
Firm age 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Very small-sized firm −0.08** −0.09** −0.09** −0.09** −0.08* −0.07* −0.08* −0.07
Small-sized firm −0.05 −0.05* −0.05* −0.05* −0.04 −0.04 −0.04 −0.02
Medium-sized firm 0.05 0.04 0.04 0.04 0.06 0.06 0.06 0.08
Large-sized firm −0.04 −0.04 −0.04 −0.04 −0.04 −0.04 −0.04 −0.10
North −0.04* −0.04* −0.04* −0.04* −0.05 −0.06* −0.06* 0.04
South −0.02 −0.02 −0.01 −0.01 0.05 0.06 0.04 0.04
Food district −0.04 −0.05 −0.05 −0.04 −0.06 −0.06 −0.06 −0.01
Agricultural district −0.04 −0.05 −0.05 −0.06 −0.06* −0.07 −0.07 −0.15***
1st distance −0.001***
2nd distance 0.00
3rd distance 0.00
Distance > 150 km −0.18*** −0.20*** −0.20*** −0.20*** −0.08
Regional R&D – Accredited funds 0.00 0.00 0.00 −0.01**
No. biotechnologist degree programmes −0.02 −0.02 −0.01 0.04
Food technologist degree 5-year programme 0.01 −0.03 −0.03 −0.01
Food technologist degree 3-year programme 0.05* 0.05* 0.05* 0.01
Extra-regional faculty of agricultural studies 0.04 0.04 0.04 0.00
Faculty of agricultural studies’ age 0.00 0.00 0.00 0.00
No. researchers/professors 0.00 −0.001* −0.001* 0.00
No. graduates 0.00 0.00 0.00 0.00
Industrial engineers on total scholars 0.01 0.01 0.01 0.00
Biologists on total scholars 0.00 0.00 0.00 −0.01*
Chemicals on total scholars 0.00 0.004*** 0.004*** 0.01
Physicians on total scholars 0.00 0.00 0.02** 0.00
Geologists on total scholars 0.00 0.00 0.00 0.00
No. scientific macro-fields 0.02** 0.02** 0.03*** 0.01
No. regional faculties of agricultural studies 0.03** 0.03** 0.03** 0.00
Public univerisity −0.04 −0.04 −0.04 −0.06
Technological transfer office −0.02 0.00 0.00 0.05
Technological transfer office’s age −0.06***
International grade 0.00 0.00
Research project grade −0.001*** −0.002***
Codified knowledge indicator (No. articles) −0.03***
Codified knowledge indicator (No. citations) −0.02***
VQR grade −0.01*
Year dummies Yes Yes Yes Yes Yes Yes Yes Yes
Sub-sector dummies Yes Yes Yes Yes Yes Yes Yes Yes
***
Signficant at 1% level.
**
Signficant at 5% level.
*
Signficant at 10% level.

variables lose significance after faculty characteristics are intro- with any universities via direct interactions or through informal
duced. More precisely, the presence of a technology transfer office contacts with academics that could act as spearheads for other col-
appears to have a displacement effect on intra muros R&D invest- laborations (Arvanitis et al., 2008) that are knowingly planned and
ment, particularly in the case of faculties with more disciplines. not necessarily local.
Table 6 reports the marginal effects for Eq. (2). Among university characteristics, the presence of an interme-
Regarding geographical proximity, the 1st, 2nd and 3rd distance diation structure has no direct effects on firm R&D collaboration
variables from the agricultural studies faculties are not significant. with universities. This result, which is apparently not in accordance
However, when the distance from the closest faculty is greater than with Muscio and Nardone (2012), can be explained by observing
150 km, this proximity is highly significant. Isolated firms that are that the introduction of the technology transfer offices was too
more than 150 km away from the closest agricultural studies fac- recent, and the diffusion of administrative practices that lever-
ulty have a 0.04 greater probability of R&D collaboration with a age new opportunities for university–industry interactions takes
university, which may or may not be the closest agricultural stud- a minimum amount of time to demonstrate a direct effect. How-
ies faculty. The former case might be explained by the so-called ever, an indirect effect is evident in that the presence of a transfer
“stray dog syndrome” (Howells et al., 2012), in which isolated firms technology office produces a displacement effect on the existence
tend to value any contact with universities more than less-isolated of intra muros R&D investment (Table 5) and R&D collaboration with
firms do because of the difficulty of identifying and maintaining private firms (the latter effect is less robust; see Table 8). The pub-
these contacts and because they are relatively unusual. Another lic status of the university is significant but negative because of
possible explanation is that for more distant firms, the absence of the many interactions between the private university research and
universities at a closer distance prevents them from collaborating industrial communities that are more deeply connected.
 O.W. Maietta / Research Policy 44 (2015) 1341–1359 1355

Table 10
Multiprobit regression. Marginal effects for the dependent variable process innovation.

Variables dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx dF/dx

Model1 Model2 Model3 Model4 Model5 Model6 Model7 Model8

R&D collaborations with universities and/or public labs 0.25*** 0.25*** 0.24*** 0.25*** 0.24*** 0.24*** 0.24*** 0.02
R&D collaborations with private firms 0.29*** 0.29*** 0.29*** 0.29*** 0.27*** 0.26*** 0.26*** −0.06
R&D intensity 0.02* 0.02* 0.02* 0.02* 0.02* 0.02* 0.02* 0.04***
Sales through distribution chain 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.001*** 0.00
Skilled employees −0.0001 −0.0001 0.00 0.00 0.00 0.00 0.00 0.00
Co-op firm 0.02 0.02 0.02 0.02 0.02 0.02 0.03 −0.04
Subsidies 0.18*** 0.18*** 0.18*** 0.18*** 0.19*** 0.19*** 0.19*** 0.25***
Non standard jobs 0.01 0.01 0.01 0.01 0.01 0.01 0.01 0.02
Firm age 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
Very small-sized firm −0.03 −0.03 −0.04 −0.03 −0.04 −0.04 −0.04 −0.06
Small-sized firm 0.01 0.01 0.01 0.01 0.01 0.01 0.00 −0.06
Medium-sized firm 0.05 0.05 0.05 0.04 0.04 0.04 0.04 −0.04
Large-sized firm 0.05 0.05 0.05 0.05 0.05 0.05 0.05 0.00
North −0.03 −0.03 −0.03 −0.03 −0.05 0.00 −0.05 0.04
South −0.07* −0.06 −0.07* −0.06 −0.01 0.03 0.00 −0.10
Food district −0.03 −0.03 −0.03 −0.03 −0.03 −0.04* −0.04* −0.01
Agricultural district 0.01 0.02 0.01 0.01 -0.01 0.01 0.01 −0.05
1st distance 0.00
2nd distance 0.00
3rd distance 0.000
Distance > 150 km −0.12 −0.11 −0.13 −0.11 −0.19
Regional R&D - Accredited funds 0.01* 0.01* 0.01* 0.00
No. biotechnologist degree programmes −0.02 −0.01 0.00 −0.13**
Food technologist degree 5-year programme −0.03 −0.05 −0.05 −0.16***
Food technologist degree 3-year programme 0.07 0.03 0.00 −0.02
Extra-regional faculty of agricultural studies 0.04 −0.03 −0.03 −0.06
Faculty of agricultural studies’ age 0.00 0.00 0.00 0.00
No. researchers/professors 0.00 0.00 0.00 0.00
No. graduates 0.00 0.00 0.00 0.00
Industrial engineers on total scholars 0.01 0.01 0.01 −0.02
Biologists on total scholars 0.00 0.00 0.00 −0.01
Chemicals on total scholars 0.00 0.00 0.00 0.02
Physicians on total scholars 0.00 0.00 0.00 0.01*
Geologists on total scholars 0.00 0.00 0.01 0.02
No. scientific macro-fields 0.02 0.02 0.01 0.04***
No. regional faculties of agricultural studies 0.02** 0.02** 0.02** 0.00
Public univerisity −0.24** −0.22** −0.22** −0.41***
Technological transfer office 0.02 0.03 0.01 0.01
Technological transfer office’s age −0.05***
International grade 0.00 0.00
Research project grade 0.002** 0.002**
Codified knowledge indicator (No. articles) 0.00
Codified knowledge indicator (No. citations) 0.00
VQR grade 0.00
Year dummies Yes Yes Yes Yes Yes Yes Yes Yes
Sub-sector dummies Yes Yes Yes Yes Yes Yes Yes Yes

Among the education variables, the food technologist 5-year Table 7 reports the marginal effects for Eq. (3).
programme is a channel for R&D university collaboration, whereas Public research labs compete with universities as firm R&D
the 3-year programme acts against these alliances. The inter- partners because collaborations with public research labs increase
pretation of this result is that university education can act as a with the shortest distance from an agricultural studies faculty and
mechanism for university–firm interactions if graduates from local decrease with its age, its internationalisation and the number of
universities find jobs in local firms. Freshmen will likely choose a regional agricultural studies faculties. Concurrently, public labs
graduation field with more local occupational chances. Then, when appear to be partners of universities with more research projects in
they are employed in local firms, they may preferentially turn to that the existence of R&D collaboration with public labs is positively
their alma maters for R&D collaboration if they experienced a good related to the research project grade of the closest agricultural stud-
relationship with their lecturers and thesis supervisors. ies faculty.
Among scholars’ personal characteristics, the presence of female Table 8 reports the marginal effects for Eq. (4). Among the
full professors induces R&D university collaboration because, on characteristics of the closest agricultural studies faculty, univer-
one hand, women have greater ability to cooperate, sensitivity to sity intermediation appears to be detrimental (in the absence of
social cues and context dependency (Croson and Gneezy, 2009; codified knowledge indicators) to R&D collaboration with private
Kuhn and Villeval, 2013). On the other hand, the cost of knowl- firms.
edge exchange with meritocratic and non-hierarchical institutions Table 9 reports the marginal effects for Eq. (5). Product inno-
is lower. vation is strongly determined by R&D collaboration with private
Codified knowledge, as measured by the number of citations in firms whereas R&D collaboration with universities and/or public
ISI-Scopus indexed journals, and the VQR are positive and weakly research labs is weakly significant. Regarding geographical proxim-
significant. Along with having agricultural studies faculties a tech- ity, the 1st distance variable from the agricultural studies faculty is
nical nature, this result is in accordance with the findings of D’Este highly significant and negative, whereas the 2nd and 3rd distance
and Iammarino (2010) for engineering-related departments. variables are not significant. Analogously, whether the distance
1356 O.W. Maietta / Research Policy 44 (2015) 1341–1359

from the closest faculty is greater than 150 km is highly signifi- intensity in the last sub-period, whereas the former appears to be
cant and negative. A firm that is within a radius of 150 km25 of an based on tacit knowledge because of the impact of direct interac-
agricultural studies faculty has a probability of product innovation tions or informal contacts with academics, as proxied by the geo-
that is 0.20 times greater (after accounting for faculty characteris- graphical distance from the closest faculty of agricultural studies.
tics) than the probability for a more distant firm. This result may
express the effect of informal contacts or direct interactions with
academics and of university–industry–government networks that 7. Concluding remarks
are beneficial for product innovation. The research project grade
and codified knowledge indicators are significant and negative The main objective of the paper is to identify how
and the VQR grade is weakly significant and negative. Consultan- university–firm R&D collaboration impacts firm product and
cies or informal collaboration may be too demanding for faculties process innovations and how the “knowledge context” in which
that are involved in projects and in codified knowledge produc- the firm operates (in terms of research, education and technol-
tion and scholars tend to concentrate on academic publications ogy transfer-related activities at local universities) affects this
because industry-oriented research may deteriorate their publi- relationship.
cation profiles (Bonaccorsi et al., 2006). The negative sign (only The results obtained show that knowledge spillovers from local
weakly significant) of the faculty size impact confirms that larger universities can be important because a firm within a radius of
faculties promote the commercial exploitation of their academic 150 km from a university has a higher likelihood of product inno-
research results and inhibit informal technology transfer (Landry vation than does a more distant firm. However, local knowledge
et al., 2007). spillovers and codified knowledge appear to be university non-joint
The marginal effects for Eq. (6) are reported in Table 10. Process outputs because the direct impact of the ISI-Scopus indexed journal
innovation is determined by R&D collaboration with private firms production on local firms’ product innovation is negative. Degree
and R&D collaboration with universities and/or public research programmes, in fields useful for local firms, act as a channel for R&D
labs, confirming what already evidenced in literature (González- collaborations with universities, public research labs and private
Pernía et al., 2014). firms.
Geographical distance from a faculty of agricultural studies is The implications for public science and technology policy of
generally not significant, whereas the public status of the university the results obtained in this study show that the NSI structure
is negative and significant with a strong impact, which means that influences the nature and the size of local knowledge spillovers.
only geographical proximity to a private agricultural studies faculty The same amount of knowledge produced by the public research
matters for process innovation. system—when the areas of expertise offered by universities are
The research project grade indicator is significant and positive: those required by the local industry—can spill over throughout
projects financed at universities have effects on the process inno- the local economy and connect the research and industrial com-
vation of local firms. The amount of codified knowledge is not munities through the education channel in the case of NSI based
significant. on universities, such as the German and Italian public agri-food
Summing up the results from all of the equations, firm age con- research systems. The positive impact of geographical proximity
tributes to explaining the choice of a university as an R&D partner, on product innovation suggests that a territorially dispersed NSI
and faculty age has a negative effect on the choice of a public structure produces local knowledge spillovers for a sector with a
research lab. Thus, long co-location contributes to explaining the plethora of small firms that use tacit-knowledge-based technolo-
linkages between F&D firms and universities together with the gies. However, the geographical distance from local universities to
education channel, which exerts a stronger impact for collabora- firms, which is relevant for knowledge externalities, is not particu-
tion with universities than for collaboration with public research larly small (150 km), whereas the marginal impact of an additional
labs. When graduates are employed in local firms, they have fewer agricultural studies faculty is limited in magnitude. Note, how-
links with public research labs; a possible explanation is that their ever, that other intrinsic characteristics of new faculties (which
turnover is higher in firms that collaborate in R&D with public probably are less hierarchical academic institutions) were already
research labs because these firms tend to use non-standard jobs. taken into account through other variables (women as full profes-
Very small firms collaborate in R&D with private firms but sors; researchers as a percentage of total researchers/professors;
encounter problems in the choice of public R&D partners, partic- discipline composition). A dispersed and polycentric NSI structure
ularly of universities. The motivations are the “problem solving” runs the risk of conflicting interests among different public players,
approach to innovation, focused on new product development, of such as universities and national and regional research labs, with a
these firms and their difficulties in identifying scientific compe- resulting increase in the information asymmetry in a firm’s choice
tences which are often fragmented and competing for fund-raising of R&D partners. Finally, the choice of both scientific disciplines and
(Ranieri and Silvestri, 2006). graduation fields influences the path of local development; some
Over the 2001–2006 period, geographical distance loses sig- economic activities might benefit, whereas others that do not use
nificance for product innovation and R&D collaborations with the knowledge produced by that specific scientific discipline or the
universities and/or public research labs for process innovation. expertise supplied by that specific graduation field might instead
Academic policies that aim at commercialising research output, be sacrificed.
which are proxied by the technology transfer office’s age, negatively The third role played by universities conflicts with research
impact both product and process innovation. and higher education in the absence of adequate resources (to be
After comparing the product and process innovations of the Ital- devoted to this specific aim) and of indicators of this type of out-
ian F&D firms, the latter appears to be based on codified knowledge put, which are taken into account to evaluate the advancement of
because of the stronger impacts of R&D collaborations and R&D scholars’ careers.
From the university perspective, particularly in the case of large
faculties, achieving high-quality teaching by monitoring scholars’
25
teaching performance should be perceived as a potential future
This threshold was selected by comparison with the alternative dummies for
50, 75, 100 and 200 km, which produce weak or non-significant results. Conversely,
source of private funding to augment university budgets. The same
if the marginal effect of the 1st distance variable in model 1 is multiplied by 150, a can be said about the gender glass ceiling, the elimination of which
comparable value of −0.15 is obtained. would increase the probability of university–firm collaboration.
 O.W. Maietta / Research Policy 44 (2015) 1341–1359 1357

From the firm perspective, very small size and an isolated loca- References
tion are related to difficulties in choosing a public R&D partner.
Reducing information asymmetry should be undertaken by trust- Acs, Z., Anselin, L., Varga, A., 2002. Patents and innovation counts as measures of
regional production of new knowledge. Res. Policy 31, 1069–1085.
worthy third parties, such as regional development agencies. Acs, Z., Audretsch, D., Feldman, M., 1991. Real effects of academic research: a
A limitation of the present work and a direction for its future comment. Am. Econ. Rev. 81, 363–367.
extension is that because of data limitations, the analysis excludes AGRA, 2004. Annuario industria alimentare in Italia. Rome.
Amara, N., Landry, R., 2005. Sources of information as determinants of novelty of
micro-sized firms (with less than 10 workers). As the paper innovation in manufacturing firms: evidence from the 1999 statistics Canada
evidenced that firm size does not directly matter for process inno- innovation survey. Technovation 25, 245–259.
vation and R&D collaboration with private partners, it could be Andreakis, N., Giordano, I., Pemtangelo, A., Fogliano, V., Graziani, G., Monti, L.M.,
Rao, R., 2004. DNA fingerprinting and quality of Corbarino cherry-like tomato
interesting to test whether this also holds in case of micro-sized landraces. J. Agric. Food Chem. 52, 3366–3371.
firms. Anselin, L., Varga, A., Acs, Z., 1997. Local geographic spillovers between university
research and high technology innovations. J. Urban Econ. 42, 422–448.
Arvanitis, A., Sydow, N., Woerter, M., 2008. Do specific forms of university-industry
knowledge transfer have different impacts on the performance of private
Acknowledgements enterprises? An empirical analysis based on Swiss firm data. J. Technol.
Transfer 33, 504–533.
The author would like to thank two anonymous referees and Ascione, E., Di Paolo, I., Vagnozzi, A., 2006. La ricerca agro-alimentare promossa
the editor for their thoughtful and very helpful comments and sug- dalle Regioni italiane nel contesto nazionale ed europeo. Quali peculiarità nei
contenuti e nella gestione. Rivista di Economia Agraria, 61, 479–518.
gestions; Attilio Luigi Pasetto and Antonio Riti (UniCredit S.p.a.) for Audretsch, D.B., Feldman, M.P., 1996. R&D spillovers and the geography of
the Mediocredito Centrale-Capitalia survey data; Roberto Ciampi- innovation and production. Am. Econ. Rev. 86, 630–640.
cacigli (Censis) for the Italian agricultural studies faculties and Audretsch, D.B., Lehmann, E.E., 2005. Does the knowledge spillover theory of
entrepreneurship hold for regions? Res. Policy 34, 1191–1202.
reputation indicators; Claudia Daniele (Netval) for information Audretsch, D.B., Lehmann, E.E., Warning, S., 2005. University spillovers and new
regarding the technology transfer offices at Italian universities; firm location. Res. Policy 34, 1113–1122.
and Massimo Aria (University of Naples Federico II), Maria Rosaria Audretsch, D.B., Stephan, P.E., 1996. Company-scientist locational links: the case of
biotechnology. Am. Econ. Rev. 86, 641–652.
Carillo (University of Naples Parthenope), Domenico Carlucci (Uni- Avermaete, T., Viane, J., Morgan, E.J., Pitts, E., Crawford, N., Mahon, D., 2004.
versity of Bari), Dimitris Christelis (CSEF), Anna Irene De Luca Determinants of product and process innovation in small food manufacturing
(Mediterranean University of Reggio Calabria), Sergio Destefanis firms. Food Sci. Technol. 15, 474–483.
Baba, Y., Shichijo, N., Sedita, S.R., 2009. How do collaborations with universities
(CSEF and University of Salerno), Luciano Gutierrez (University of affect firms’ innovative performance? The role of Pasteur scientists in the
Sassari), Tullio Jappelli (CSEF and University of Naples Federico II), advanced materials field. Res. Policy 38, 756–764.
Massimo Marrelli (University of Naples Federico II), Paolo Masi Baregheh, A., Rowley, J., Sambrook, S., Davies, D., 2012. Food sector SMEs and
innovation types. Br. Food J. 114, 1640–1653.
(University of Naples Federico II), Gianluca Nardone (University of
Becker, W., Dietz, J., 2004. R&D cooperation and innovation activities of firms:
Foggia), Yolanda Pena Boquete (University of Vigo) and Francesco evidence for the German manufacturing industry. Res. Policy 33, 209–223.
Pennacchi (University of Perugia) for the useful feedback they Beise, M., Stahl, H., 1999. Public research and industrial innovations in Germany.
provided on previous versions of this paper. Earlier versions of Res. Policy 28, 397–422.
Bekkers, R., Bodas Freitas, I.M., 2008. Analysing knowledge transfer channels
this paper were presented at the following meetings: 3rd AIEAA between universities and industry: to what degree do sectors also matter? Res.
Conference, Alghero, 25–26 June 2014, 14th EAAE Congress, Ljubl- Policy 37, 1837–1853.
jana, 26–29 August 2014, and 55th SIE Conference, Trento, 23–25 Belderbos, E., Carree, M., Diederen, B., Lokshin, B., Veugelers, R., 2004.
Heterogeneity in R&D cooperation strategies. Int. J. Ind. Org. 22, 1237–1263.
October 2014. The usual disclaimer applies. Belderbos, E., Carree, M., Lokshin, B., Fernández Sastre, J., 2015. Inter-temporal
patterns of R&D collaboration and innovative performance. J. Technol. Transfer
40, 123–137.
Benfratello, L., Schiantarelli, F., Sembenelli, A., 2008. Banks and innovation:
Appendix A
microeconometric evidence on Italian firms. J. Financial Econ. 90, 197–217.
Bertocchi, G., Gambardella, A., Jappelli, T., Nappi, C.A., Peracchi, F., 2015.
Bibliometric evalaution vs informed peer review: evidence from Italy. Res.
Policy 44, 451–466.
Table A1 Bodas Freitas, I.M., Guena, A., Rossi, F., 2011. Patterns of Collaborations between
Regional distribution of F&D firms with ≥10 workers. Regional Firms and Universities: evidence from the Piedmont region in Italy.
Working Papers, S. Cognetti de Martiis Departiment of Economics, 5.
Regions Populationa Sample Bonaccorsi, A., Colombo, M.G., Guerini, M., Rossi-Lamastra, C., 2013. University
specialization and new firm creation across industries. Small Bus. Econ. 41,
Piemonte 8 8
837–863.
Valle d’Aosta 0 0
Bonaccorsi, A., Dariao, C., Simar, L., 2006. Advanced indicators of productivity of
Lombardia 15 15 universities: an application of robust nonparametric methods to Italian data.
Trentino Alto Adige 3 2 Scientometrics 66, 389–410.
Veneto 10 11 Boschma, R., 2005. Proximity and innovation: a critical assessment. Reg. Stud. 39,
Friuli Venezia Giulia 3 2 61–74.
Liguria 2 2 Broström, A., 2010. Working with distant researchers? Distance and content in
Emilia Romagna 14 13 university–industry interaction. Res. Policy 39, 1311–1320.
Toscana 6 6 Brouwer, E., Kleinknecht, A., 1996. Firm size, small business presence and sales of
Umbria 2 2 innovative produtcs: a micro-econometric analysis. Small Bus. Econ. 8,
Marche 3 3 189–201.
Lazio 5 3 Chartier, O., 2007. Agri-Food Research in Europe: Final Report. EU Agri Mapping
Abruzzo 3 4 Project Report (downloaded from http://www.agrifoodresearch.net).
Chen, J., Chen, Y., Vanhaverbeke, W., 2011. The influence of scope, depth, and
Molise 1 1
orientation of external technology sources on the innovative performance of
Campania 8 9
Chinese firms. Technovation 31, 362–373.
Puglia 6 7 Crépon, B., Duguet, E., Mairesse, J., 1998. Research innovation and productivity: an
Basilicata 1 1 econometric analysis at the firm level. Econ. Innovation New Technol. 7,
Calabria 2 1 115–158.
Sicilia 5 6 Croson, R., Gneezy, U., 2009. Gender differences in preferences. J. Econ. Lit. 47,
Sardegna 3 5 448–474.
Total 100 100 D’Este, P., Iammarino, S., 2010. The spatial profile of university-business research
a
partnerships. Pap. Reg. Sci. 89, 335–350.
Own elaboration on Istat – Census of Industry 2001. D’Este, P., Iammarino, S., Guy, F., 2013. Shaping the formation of
university–industry research collaborations: what type of proximity does
really matter? J. Econ. Geogr. 13, 537–558.
1358 O.W. Maietta / Research Policy 44 (2015) 1341–1359

D’Este, P., Patel, P., 2007. University–industry linkages in the UK: what are the Lundvall, B.A., et al., 1988. Innovation as an interactive process—from
factors underlying the variety of interactions with industry? Res. Policy 36, user-producer interaction to the National System of Innovation. In: Dosi, G.
1295–1313. (Ed.), Technical Change and Economic Theory. Pinter, London,
De Devitiis, B., Maietta, O.W., 2013. Regional patterns of structural change in pp. 349–369.
Italian agriculture. In: Ortiz-Miranda, D., Moragues-Faus, A.M., Arnalte-Alegre, In: Magistà, A., (Ed.) 2013. Grande Guida Università, la Repubblica-Censis.
E. (Eds.), Agriculture in Mediterranean Europe Between Old and New Mansfield, E., 1991. Academic research and industrial innovation. Res. Policy 20,
Paradigms. Emerald Group Publishing Limited, Bingley, pp. 173–205. 1–12.
De Fuentes, C., Dutrénit, G., 2012. Best channels of academia-industry interaction Mansfield, E., 1995. Academic research underlying industrial innovations: sources,
for long-term benefit. Res. Policy 41, 1666–1682. characteristics, and financing. Rev. Econ. Stat. 77, 55–65.
De Fuentes, C., Dutrénit, G., 2014. Geographic proximity and university–industry Mansfield, E., Lee, J.Y., 1996. The modern university: contributor to industrial
interaction: the case of Mexico. J. Technol. Transfer, innovation and recipient of industrial R&D support. Res. Policy 25,
http://dx.doi.org/10.1007/s10961-014-9364-9. 1047–1058.
Fantino, D., Mori, A., Scalise, D., 2012. Collaboration between firms and universities Medda, G., Piga, C.A., Siegel, D.S., 2005. University R&D and firm productivity:
in italy: the role of a firm’s proximity to top-rated departments. Working evidence from Italy. J. Technol. Transfer 30, 199–205.
Papers, Bank of Italy, 884. Menrad, K., 2004. Innovations in the food industry in Germany. Res. Policy 33,
Farace, S., Mazzotta, F., 2015. The effect of human capital and networks on 845–878.
knowledge and innovation in SMEs. J. Innovation Econ. Manage. 16 (1), Minarelli, F., Raggi, M., Viaggi, D., 2015. Determinants of the type of Innovation: an
forthcoming. analysis of European food SMEs. Bio-based Appl. Econ. 4, 33–53.
Feldman, M., Florida, R., 1994. The geographic sources of innovation: technological Ministero dell’Università e della Ricerca Scientifica, various years. Guida
infrastructure and product innovation in the United States. Ann. Assoc. Am. all’Università, CIMEA, Fondazione RUI, Rome.
Geogr. 84, 210–229. Moncada-Paternò-Castello, P., Ciupagea, C., Smith, K., Tübke, A., Tubbs, M., 2010.
Fitjar, R.D., Rodríguez-Pose, A., 2012. Firm collaboration and modes of innovation Does Europe perform to little corporate R&D? A comparison of EU and non-EU
in Norway. Res. Policy 42, 128–138. corporate R&D performance. Res. Policy 39, 523–536.
Freeman, C., 1988. Japan: a new national system of innovation. In: Dosi, G., Monducci, R., 2011. Innovazione e competitività delle imprese dell’industria
Freeman, C., Nelson, R., Silverberg, G., Soete, L. (Eds.), Technical Change and alimentare, Paper presented, CIBUS TEC Workshop L’innovazione tecnologica
Economic Theory. Pinter, London, pp. 331–348. come leva di competitivitá pr l’industria alimentare, Parma,
Fritsch, M., 2002. Measuring the quality of regional innovation systems: a 18 October 2011.
knowledge production function approach. Int. Reg. Sci. Rev. 25, 86–101. Monfardini, C., Radice, R., 2006. Testing exogenity in the bivariate probit model: a
Fritsch, M., Franke, G., 2004. Innovation, regional spillovers and R&D cooperation. Monte Carlo study. Dept. of Economics, University of Bologna.
Res. Policy 33, 245–255. Monjon, S., Waelbroeck, P., 2003. Assessing spillovers from universities to firms:
González-Pernía, J., Parrilli, M.D., Peña-Legazkue, I., 2014. STI-DUI learning modes, evidence from French firm-level data. Int. J. Ind. Org. 21,
firm-university collaboration and innovation. J. Technol. Transfer, 1255–1270.
http://dx.doi.org/10.1007/s10961-014-9352-0. Morgan, K., 2004. The exaggerated death of geography: learning, proximity and
Greunz, L., 2002. The innovation process of European regions. Brussels Econ. Rev. territorial innovation systems. J. Econ. Geogr. 4, 3–21.
45, 59–94. Motohashi, K., 2005. University–industry collaborations in Japan: the role of new
Griliches, Z., 1979. Issues in assessing the contribution of research and technology-based firms in transforming the National Innovation System. Res.
development to productivity growth. Bell J. Econ. 10, 92–116. Policy 34, 585–594.
Griliches, Z., Pakes, A., 1984. R&D at the firm level: a first look. In: Griliches, Z. (Ed.), Mowery, D.C., Ziedonis, A.A., 2015. Market versus spillovers in outflows of
R&D, Patents and Productivity. Chicago University Press, Chicago. university research. Res. Policy 44, 50–66.
Hall, B.H., Link, A., Scott, J.T., 2003. Universities as research partners. Rev. Econ. Muscio, A., Nardone, G., 2012. The determinants of university–industry
Stat. 85, 485–491. collaboration in food science in Italy. Food Policy 37, 710–718.
Howells, J., Ramlogan, R., Cheng, S.-L., 2012. Innovation and university Nelson, R.R. (Ed.), 1993. Oxford University Press, New York.
collaboration: paradox and complexity within the knowledge economy. Netval, 2005. Seconda indagine sull’attività di valorizzazione della ricerca
Cambridge J. Econ. 36, 703–721. scientifica presso le università italiane, 2nd Report (downloaded from
Huiban, J.P., Bouhsina, Z., 1998. Innovation and the quality of labour factor: an www.netval.it).
empirical investigation in the French food industry. Small Bus. Econ. 10, Netval, 2013. Seminiamo ricerca per raccogliere innovazione, 10th Report
389–400. (downloaded from www.netval.it).
Huynh, K.P., Rotondi, Z., 2009. R&D Spending and Knowledge Spillovers Nieto, M.J., Santamaria, L., 2007. The importance of diverse collaborative networks
(downloaded from SSRN 968,666). for the novelty of product innovation. Technovation 27, 367–377.
INEA, 2006. L’agricoltura italiana. Sfide e prospettive di un settore vitale per Pagano, P., Schivardi, F., 2003. Firm size distribution and growth. Scandinavian J.
l’economia della nazione. ESI, Naples. Econ. 105, 255–274.
INEA, various years. Annuario dell’agricoltura italiana. Il Mulino, Bologna. Pasetto, A.L., 2011. Food & Beverage (Executive Summary), Industrial Efige Report,
INEA, various years, 2006. Annuario dell’agricoltura italiana. ESI, Naples. Unicredit, Milan.
ISTAT, several years. Rilevazione sull’Innovazione nelle Imprese, Rome. Pavitt, K., 1984. Sectoral patterns of technical change: towards a taxonomy and a
ISTAT, 2005. Distretti industriali e sistemi locali del lavoro 2001. Rome. theory. Res. Policy 13, 343–373.
ISTAT, various years. Rilevazione sull’Innovazione nelle Imprese, Rome. Pennacchi, F., 2008. Struttura e ruolo dell’Università. Agriregionieuropa 4,
ISTAT, various years. Statistiche sulla Ricerca Scientifica, Rome. 11–13.
Istituto Guglielmo Tagliacarne, 2004. L’innovazione nelle imprese del Piga, C.A., Vivarelli, M., 2004. Internal and external R&D: a sample selection
Mezzogiorno. DIRCE, Rome. approach. Oxford Bull. Econ. Stat. 66, 457–482.
Jaffe, A., 1986. Technological opportunity and spillovers of R&D: evidence from In: Rama, R., (Ed.), 2008. Handbook of innovation in the Food and Drink industry,
firms’ patents and market value. Am. Econ. Rev. 76, 984–1001. The Haworth Press, New York.
Jaffe, A., 1989. Real effects of academic research. Am. Econ. Rev. 79, 957–970. Ranieri, R., Silvestri, M., 2006. Innovazione quale fattore determinante per la
Jaffe, A.B., Trajtenberg, M., Henderson, R., 1993. Geographic localization of competitività delle imprese, In: Pennacchi, F. (Ed.), Ricerca nell’area delle
knowledge spillovers as evidenced by patent citations. Q. J. Econ. 63, scienze agrarie. Università degli Studi di Perugia–Facoltà Agraria, Perugia, pp.
577–598. 63–68.
Jiang, L., Tan, J., Thursby, M., 2010. Incumbent firm invention in emerging fields: Robin, S., Schubert, T., 2013. Cooperation with public research institutions and
evidence from the semiconductor industry. Strategic Manage. J. 32, success in innovation: evidence from France and Germany. Res. Policy 42,
55–75. 149–166.
Karlsson, C., Olsson, O., 1998. Product innovation in small and large enterprises. Ruttan, V.W., 2001. Technology, Growth and Development: An Induced Innovation
Small Bus. Econ. 10, 31–46. Perspective. Oxford University Press, New York.
Kaufmann, A., Tödling, F., 2001. Science-industry interaction in the process of Sacchi, R., Parisini, C., Paduano, A., Della Medaglia, D.A., Savarese, M., Ambrosino,
innovation: the importance of boundary-crossing between systems. Res. Policy M.L., 2010. Relationship between sensory profile and volatiles compounds:
30, 791–801. identification of sensory tipicality in PDO Italian live oil, In: Amenta, P.,
Kelly, D., Henchion, M., O’Reilly, P., 2008. Knowledge transfer in the Irish D’Ambra L., Danzart, M., (Eds.), Proceedings of the 11th European Symposium
Innovation System: Industry and Researcher Perspectives. Paper presented, on Statistical Methods for the Food Industry. Academy School, Afragola, pp.
12th EAAE Congress, Gent, 26–29 August 2008. 65–72.
Kuhn, P.J., Villeval, M.C., 2013. Are women more attracted to cooperation than Santini, A., 2003. La Ricerca Scientifica nella Facoltà di Agraria, Università degli
men? Working Papers, NBER, 19277. Studi di Napoli Federico II. La Buona stampa, Ercolano.
Landry, R., Amara, N., Ouimet, M., 2007. Determinants of knowledge transfer: Scordamaglia, L.P., 2006. Innovazione quale fattore determinante per la
evidence from Canadian university researchers in natural sciences and competitività delle imprese. In: Pennacchi, F., (Ed.), Ricerca nell’area delle
engineering. J. Technol. Transfer 32, 561–592. scienze agrarie. Università degli Studi di Perugia–Facoltà Agraria, Perugia, pp.
Laursen, K., Reichstein, T., Salter, A., 2011. Exploring the effect of geographical 69–74.
proximity and university quality on university–industry collaboration in the Sorrentino, M., Capozzi, P., 2010. Spin-off da ricerca in Campania. Giannini Editore,
United Kingdom. Reg. Stud. 45, 507–523. Naples.
Leten, B., Landoni, P., Van Looy, B., 2014. Science of graduates: how do firms benefit Stewart-Knox, B., Mitchell, P., 2003. What separates the winners from the losers in
from the proximity of universities? Res. Policy 43, 1398–1412. the new food product development? Trends Food Sci. 14, 58–64.
 O.W. Maietta / Research Policy 44 (2015) 1341–1359 1359

Thether, B.S., Tajar, A., 2008. Beyond industry-university links: Sourcing Wilde, J., 2000. Identification of multiple equation probit models with endogenous
knowledge for innovation from consultants, private research organisation and dummy regressors. Econ. Lett. 69, 309–312.
the public science-base. Res. Policy 37, 1079–1095. Ziggers, G.W., 2005. Radical Product innovation in the Dutch Food Industry: An
Varga, A., 2000. Local academic knowledge transfers and the concentration of Empirical Exploration. J. Food Prod. Marketing 11, 43–65.
economic activity. J. Reg. Sci. 40, 289–309. Zucker, L.G., Darby, M.R., Armstrong, J., 1998. Geographically localized knowledge
Veugelers, R., 1997. Internal R & D expenditures and external technology sourcing. spillovers or markets? Econ. Inq. 36, 65–86.
Res. Policy 26, 303–315.