Firm-university alliances: does distance matter?
Massimo G. Colombo[1], Massimiliano Guerini [2], Cristina Rossi Lamastra [3]
Abstract. This paper explores whether and how geographical, social and cognitive proximity and affect the likelihood of collaboration between high-tech entrepreneurial ventures and universities. In addition, the impact of university quality on firm-universitiy collaboration is considered. We use a sample composed by 79 Italian young high-tech entrepreneurial ventures that have collaborated with at least an Italian university in the period 2004-2008, resulting in 96 university-industry collaborations. We run a logistic regression where the dependent variable is the probability that the entrepreneurial venture collaborates with a university. We also interact social, cognitive proximity and university quality with geographical proximity. Results show that geography matters in shaping university-industry collaborations. Furthermore, our findingssuggest that social and cognitive proximity are among the main determinants of collaboration, while university quality do not play a significant role. Finally, we find that social proximity actually reduces the negative effect of distance on the likelihood that the entrepreneurial venture collaborates with the university.
Keywords: university-industry collaborations, geographical proximity, cognitive proximity, social proximity, high-tech entrepreneurial venture.
Acknowledgements: the financial support of Regione Toscana Project LILIT: I Living Labs per l’Industria Toscana (PAR FAS REGIONE TOSCANA Linea di Azione 1.1.a.3) is kindly acknowledged.
Preliminary draft
1 Introduction
The study of university-industry collaborations is nowadays a highly relevantissue that has key policy implications. Indeed,in the last years, universitieshave significantly re-orientated their academic research towards industrial applications and have shown a stronger propensity to collaborate with firms so as to obtain research funds from them (Bonaccorsi et al 2012a). Likewise, collaborationswith universities help firmsin accessing scientific and technological knowledge developed by universities (hereafter: university knowledge), thus playing a crucial role in allowing firms to keep the rapid pace of the science and technology This is particularly important for entrepreneurial ventures operating in high-tech industry (henceforth: high-tech entrepreneurial ventures). These firms requireleading-edge knowledge to conduct their production processesand successfully implementing their business model. However, due to their young age and limited internal resources, they cannot produce all this knowledge within their boundaries.
Moving from these premises, in this paper we speculate on the differential role of geographical, social and cognitive proximity in enabling collaborations between high-tech entrepreneurial ventures and universities. In addition, the effect ofuniversity quality shaping these collaborations is considered.Available empirical evidence highlights the importance of geographical proximity to access university knowledge (see e.g. the recent contributions of Bonaccorsi et al 2012b; Belenzon and Schankerman 2012; Laursen et al 2011).We expand on these results andanalyzehow the different forms of proximity interact in determining the likelihood that high-techentrepreneurial ventures collaborate with the university. Specifically, we intend to answer the following research questions. Is geographical proximity the main determinant of university-industry collaborations in high-tech industries? Do other forms of proximity (social and cognitive) and university quality substitute for geographical proximity? In other words, do high-tech entrepreneurial ventures collaborate with distant universities if social, cognitive proximity and university quality are high?
We explore the role of proximity in shaping university-industry collaborations by using a sample composed by 79Italian young high-tech entrepreneurial ventures that have collaborated with at least an Italian university in the period 2004-2008, resulting in 96 university-industry collaborations.We combine data from a number of rich information sources.The sample of young high-tech entrepreneurial ventureshas been extracted from theRITA (Research on Entrepreneurship in Advanced Technologies) database.This database constitutes the most complete source of information availableon Italian high-tech firms and has been developed by Politecnico di Milano. Sample firms are young (less than 25 years old), independent and operate in high-tech sectors in manufacturing and services. Data on Italian universities have been extracted from the EUMIDA database. This database has been developed under a European Commission tender and it is based on official statistics produced by National Statistical Authorities in all 27 EU countries plus Norway and Switzerland (for details see European Commission, 2010).Our final dataset consists in 6,241 entrepreneurial venture-university dyads, which correspond to all possible combinations between the collaborating entrepreneurial ventures and the Italian universities (79 entrepreneurial ventures * 79 universities).
The dependent variable is the probability that the high-tech entrepreneurial venture collaborates with a university.We use a logistic regression to estimate the effect of geographic, cognitive and social proximity and of university quality on the likelihood of collaboration between the high-tech entrepreneurial venture and the university. As a (inverse) proxy for geographical proximity, we measure the distance between the high-tech entrepreneurial ventureand the university.In order to measure social proximity, we consider ifat least one of the founders of the entrepreneurial venture had a previous experience in the university. As far as concerns cognitive proximity, we evaluate whether the university is specialized in engineering sciences. Finally, order to measure university quality, we refer to the Shanghai Ranking.
Results show that geography matters in shaping university-industry collaborations, since the likelihood that a high-tech entrepreneurial ventures collaborates with a university decreases with the distance from that university. In other words, geographical proximity is extremely relevant to access university knowledge. Moreover, results suggest that social and cognitive proximity between the entrepreneurial venture and the university are among the main determinantsof collaboration, while university quality does not play a significant role. Finally, we find that social proximity actually reduces the negative effect of distance on the likelihood that the entrepreneurial venture collaborates with the university.
The structure of the paper is as follows. The next section reviews the extant literature and provides the theoretical background for the empirical analysis. Section 3 describes data and section 4 the methodology. Section 5 reports the results of the econometric estimates. The final section provides conclusions and discussion.
2 Theoretical background
Conventional wisdom indicates that research conducted within universities is a source of significant innovation-generating knowledge for high-tech entrepreneurial ventures CITE E AGGIUNGERE UN PARAGRAFO COSI E’ TROPPO STRINGATO. However, there is quite unanimous agreement that the successful transfer of knowledge from universities to industry is shaped by geography.Various studies conducted in the US have found that knowledge transfers from universities to industry crucially depends on geographical distance, being to a large extent confined to the area in which the university is located (e.g. Jaffe 1989; Anselin et al 1997; Belenzon and Schankerman, 2012; for a similar evidence in European countries see, among the others, Bottazzi and Peri 2003; Bonaccorsi et al 2012b).In a similar vein, recent contributions have found that distance from the university decreases the likelihood that a firm collaborates with the university (e.g Laursen et al 2011; Hong and Su 2012).
Such an evidence is explained by the importance of direct interactions and face-to-face contacts for the university knowledge exchanges. In most cases, knowledge developed within universities is not directly usable by firms for commercial purposes. It is still fluid and only partially formed (Storper and Venables 2004) and has a significant component of tacit knowledge, Geographical proximity to university enablesintense and frequent direct interactions with the academic personnelthat developed this knowledge (Morgan 2004; Bruneel et al 2010), thus being an important requisite of successful knowledge transfer.Along this line of reasoning, it has been noted that collaborationsbetween firms and neighboring universities aremore likely than distant collaborations to generate impulses to innovation and create significant learning effects for the collaborating firms (e.g., Broström 2010).Furthermore, geographical proximity favors the establishment of common interests and aligns incentives between collaborating entrepreneurial ventures and their academic partners. Put differently, geographical proximity can compensate the lack of institutional proximity(Cooke et al 1997).Institutional proximity helps bring organizations together through sharing similar values and norms at the macro-level (Boschma 2005). If institutional proximity is missing, geographical proximity helps to create a common background and a shared set of expectations and understandings about the nature of the collaboration (Gertler 1995). Institutional proximity is generally low in university-industry relations (e.g. Ponds et al 2007) so that geographic proximity plays an important rolein helpingheterogeneous agentsto overcome the differences in institutional contexts. Accordingly, we posit the following research hypothesis:
H1: The likelihood that high-techentrepreneurial ventures collaborates with a university decreases with the geographical distance from the university.
However, the importance of geographical proximity cannot be assessed in isolation, but it should be analyzed in relation to other dimensions of proximity, that might facilitate knowledge exchanges.. Rooting on mainstream literature, we consider here the role of cognitive and social proximity. Cognitive proximity indicates the extent to which two organizations share the sameknowledge base while social proximity the extent to which the members of the two organization have friendly relationships (Boschma, 2005).
Cognitive proximity is an essential ingredient for successfulcollaborations as it provides a shared knowledge base for partners (Knoben and Oerlemans 2006; Nooteboom et al 2007). As far as university-industry collaborationsare concern, the successful transfer of university knowledge depends on the absorptive capacity of the entrepreneurial venture (Cohen and Levinthal 1990). Accordingly, the cognitive base of the entrepreneurial venture should be close enough to the knowledge developed within the university. Specifically, increasing the cognitive proximity enables the partners tosuccessfully exchange and use knowledge in their joint work building on sharedtechnological experiences and knowledge bases. We argue here that cognitive proximity is higher if the university has a strong specialization in applied sciences. Using data from the Carnegie Mellon Survey of Industrial R&D,Cohen et al (2002) find that the impact of public research, at least in most industries, is exercised through engineering and applied science fieldsrather than through basic sciences. Hence, we may expect that if cognitive proximity is high (i.e. the collaboration involves universities specialized in applied sciences), the likelihood of collaborations increases. Moreover, the effect of cognitive proximity might also overcome the negative effect of distance in shaping university-industry collaborations. Specifically, we expect that high-tech entrepreneurial ventures will tend to collaborate with universities specialized in applied and engineering sciences, independently from their distance. Following this line of reasoning, we posit the following research hypothesis:
H2a: The likelihood that ahigh-techentrepreneurial venture collaborates with a university increases with the cognitive proximityto university.
H2b: Cognitive proximityreduces the negative effect of geographical distance on the probability that an entrepreneurial venture collaborates with the university.
Let us now focus on the role of social proximity. Boschma (2005) defined social proximity in terms of socially embedded relations between agents at the micro level. Specifically, the presence of previous relationships between a high-tech entrepreneurial venture and a university increase trust between agents, reduceuncertainty, promoteseffective learning (Lundvall, 1993) and facilitatesthe transfer of tacit knowledge (Maskell and Malmberg, 1999). Hence, we expect that high-tech entrepreneurial ventures in which at least one of the founders has a previous experience in the university have an higher likelihood to collaborate with that university. Moreover, we expect that social proximity not only increases the likelihood of collaboration, but also has a moderating effect on the negative impact of distance. In other words, the presence of social proximity increases university-industry collaborations independently from distance. Accordingly, we posit the following research hypotheses:
H3a: The likelihood that the entrepreneurial venture collaborates with a university increases with the social proximityto university.
H3b: Social proximityreduces the negative effect of geographical distance on the probability that an entrepreneurial venture collaborates with the university.
Finally, it is important to consider that not all universities are endowed equallywith resources and networks. The most valuableresources are likely to be concentrated in high-quality universities.Therefore, given a choice of partners,firms will prefer to collaborate with these universities. In a recent contribution, Laursen et al (2011) have shown that firms’ decisions to collaborate with universities are influenced by both geographical proximity and university quality. Being located close (within 100 miles) to a high quality university increases the propensity for firms to collaborate locally. Conversely, co-location with a low quality university discourages local collaboration. In the absence of a high quality university nearby, the second-best choice is to collaborate with a distant (more than 100 miles) high quality university. In other words, firms appear to prefer university quality to geographical proximity, in line with the view that the benefits of leveraging high quality knowledge overcome the cost of long distance collaborations.Hence, we posit the following research hypotheses:
H4a: The likelihood that the entrepreneurial venture collaborates with a university increases with university quality.
H4b: University qualityreduces the negative effect of geographical distance on the probability that an entrepreneurial venture collaborates with the university.
3 Data
To test our conjectures, we take advantage of the RITA (Research on Entrepreneurship in Advanced Technologies) database, which was developed by the RITA Observatoryresearch team at Politecnico di Milano and it is the most authoritative source of information presently available on Italian high-tech entrepreneurial ventures.The Observatory aims at extending the knowledge of Italian high-tech entrepreneurial ventures, by monitoring new start-ups, their economic, financial and managerial characteristics and post-entry performances in terms of innovation and growth.The RITA database was created at the beginning of year 2000 at Politecnico di Milano and was updated and extended in the years 2002, 2004, 2007, and 2009.
The database contains information on 1,949 high-tech entrepreneurial ventures, established in 1983 or later, independent at start-up time (i.e. not controlled by another business organization even through other organizations may have held minority shareholdings in the new firms) and operating in high-tech industries, both in manufacturing and services. In order to obtain a measure of university-industry collaboration, we drew on the responses to 6 questions on the RITA 2009 survey that asked whether the entrepreneurial venture i) obtained a license from the university; ii) has utilized technical (non patented) knowledge developed within the university; iii) has financed joint R&D projects with the university; iv) has take advantage of university laboratories and equipment provided by the university; v) has receivedconsulting services (technical, managerial, commercial) from the university and vi) has received support from the university to obtain finance.We thus considered as “collaborating” each entrepreneurial venture that declared that used at least 1 of the 6 forms of collaborations defined above. The RITA 2009 survey also asked the name of the university with which the entrepreneurial venture has collaborated. After cleaning data because of missing information, we were able to identify 79 entrepreneurial ventures that collaborated with at least a Italian university between 2004 and 2008. Since some entrepreneurial ventures collaborated with more than a university, the total number of collaborations in our sample is 96. Table 1 reports the distribution of the 79 collaborating entrepreneurial ventures by industry, foundation year and geographical localization.
[Table 1]
Data on collaborating entrepreneurial ventures from the RITA database are complemented by data on Italian universities which have been extracted from the EUMIDA database. This database has been developed under a European Commission tender and it is based on official statistics produced by National Statistical Authorities in all 27 EU countries plus Norway and Switzerland (for details see European Commission, 2010).For the purpose of this paper, we extracted the data relating to the size of the university (measured by the academic staff) and their scientific specialization.Moreover, in order to assess the quality of the university, we use the Shanghai ranking.[4]
We then considered all the possible combinations between the 79high-tech entrepreneurial ventures in our sample and every Italian university (79 universities) with which a high-tech entrepreneurial venture could potentially have established a collaboration. Hence, our final dataset consists in 6,241 entrepreneurial venture-university dyads, which correspond to all possible combinations between the collaborating entrepreneurial ventures and the Italian universities (79*79).
In order to calculate distances between entrepreneurial ventures anduniversities, we used the database of the Italian National Institute of Statistics (ISTAT) to obtain data on the latitude and longitude of each municipality in which entrepreneurial ventures and universities are located. Hence, we calculated the linear distance between the centroids of the municipalities in which the high-tech entrepreneurial venture and the university are located. Considering the 96 collaborations in our sample, the average distance of the collaboration is 91.26 km, ranging from 1km (the entrepreneurial venture and the university are in the same municipality) to 854.21 km.
4 Method
We use a logistic regression to estimate the effect of geographic, cognitive, and social proximity and university quality on the likelihood of collaboration between the high-tech entrepreneurial venture and the university. We estimate models of the type:
. (1)
The dependent variablecolli,j is a dummy variable that equals 1 if the entrepreneurial venture i collaborates with the university j. Among covariates, as a (inverse) proxy for geographical proximity, we consider the distance between the high-tech entrepreneurial venture i and the university j.Following Sorenson and Stuart (2001), we use the logarithm of distance (log(dist)i,j) to deal with the non-linearity between a distance and the time and money consumed for that distance.In order to measure social proximity, founderi,j is a dummy variable assuming value 1 if at least one of the founders of the entrepreneurial venturehas a previous experience in the university j. The variable engj is a dummy variable that equals 1 if the university j is specialized in engineering sciences. As previously discussed, we could expect that cognitive proximity is higher if the university is specialized in engineering sciences. It has been constructed by first calculating for each Italian university a Balassa Index (Balassa 1965), considering the share of academic staff specialized in engineering sciences. Specifically, the Balassa Index has been calculated as follows: