«Der Open-Access-Publikationsserver der ZBW – Leibniz-Informationszentrum Wirtschaft The Open Access Publication Server of the ZBW – Leibniz ...»
With regard to a subsample of 412 firms reporting patent applications he investigates the influence of the reach of collaboration on the innovation output of firms (number of patents). Regional connectivity, that is the number and intensity of links to regional partners, has a highly significant positive impact on innovation output. The same result holds as to supra-regional links, i.e. national and international partners, outside the metropolitan area under scrutiny. A specification of a negative binomial regression including regional and international links and omitting the national reach does not alter the results: both regional and international collaboration has a significant and positive influence.
A comparison of the different studies turns out to be rather difficult. The main reason is the diversity of concepts and indicator variables to be found in the literature as to collaboration, innovation and spatial distance.
As to collaboration some papers assume the existence of links – the so called spillovers - without any further specification as to the precise meaning of this term (Jaffe 1989; Anselin et al. 2000). Other authors focus on formal collaboration between firms and public research institutions (Robin & Schubert 2010) or joint EU funded research projects (Autant-Bernard et al. 2007). Some use joint patents as an indicator of collaboration (Canter & Graf 2008, Broekel et al. 2011) and Cummings and Teng (2003: 49) define three interdependent types of knowledge transfer activities. Drejer
and Vinding (2007) in their survey asked for the main partner of innovation collaboration, but did not specify the type of collaboration (similar: Fritsch 2000). To the contrast Isaksen and Onsager (2010) distinguish nine knowledge transfer channels.
With respect to innovation many studies rely on patents as indicator of innovation (Jaffe 1989; Krätke 2010) or only include product innovation (Anselin et al. 2000, Drejer & Vinding 2007). Cummings and Teng (2003) ask for successful knowledge transfer and Antonelli and Fassio (2011) refer to product and process innovations. Isaksen and Onsager (2010) include product- and process-innovations and in addition patents.
As to spatial distance the indicators refer to administrative boundaries (Jaffe 1989, Fritsch 2000), functional delineations (Anselin et al. 2000, Broekel & Meder 2010), more or less precisely defined local/regional, national and international spatial levels (Drejer & Vinding 2007, Krätke 2010) or precise definitions as to miles or kilometers (Cummings & Teng 2003, Autant-Bernard et al. 2007, De Jong & Freel 2010).
With regard to the populations and control variables these studies ground on different sets of data. Some studies cover only a certain group of firms: Manufacturing and services with 20 employees and more (Robin & Schubert 2010), manufacturing (Fritsch 2000; Antonelli & Fassio 2011), micro and nanotechnologies (Autant-Bernard et al.
2007), biotechnology (Audretsch & Stephan 1996), science-based firms (Krätke 2010), high-technology corporations (Cummings & Teng 2003), certain two-digit industries (Anselin et al. 2000). As to control regressors, variables such as size of firms, R&D capacity, factors hampering innovation, management strategies, degree of competition and many more play a role.
Last not least, the analysis of regional innovation systems has to cope with problems of causality and internal relationships of innovation, cooperation and regional reach. Chart 1 depicts the idea that as to any of the three poles both directions of influence are possible (Cassiman & Veuglers 2002, Okamuro et al. 2011). Besides, the dashed arrows illustrate that substitution and complementarity of different forms of innovation, collaboration and regional reach have to be considered.
To sum up, the empirical literature covers a tremendous diversity of indicator variables, populations and in addition methods. Even so some generalizations with regard to the spatial reach of collaboration for innovation are possible.
First, as to the spatial reach the descriptive results are to some extent similar. About 30 percent of all collaborations can be found at the local and regional level. National cooperation relationships amount to about 40 to 50 percent and international linkages have a share of more or less 20 percent.
Second, the results of several empirical investigations do not indicate a negative or positive influence of a local spatial reach of cooperative links per se as to innovation (Fritsch 2000; Cummings & Teng 2003; Drejer & Vinding 2007; De Jong & Freel 2010;
Isaksen & Onsager 2010; in addition see Freel et al. 2009). There is only on study directly dealing with the question of the influence of the spatial reach of collaboration on innovation (Krätke 2010). But his results show no difference as to regional or supraregional collaborative relationships, either. Both have a positive influence on the number of patents of an establishment.
This points to a remarkable difference in comparison to the whole body of literature as to the importance of clustering of innovations at the regional level. The empirical fact of regional clustering in general is explained by reference to local and regional networking, i.e. collaboration. Therefore the outcomes at the firm level contradict to some extent the results at the regional level. The reasons may be special circumstances as to specific forms of collaboration and innovation (e.g. patents and patent related collaborations) or collaboration partners (e.g. public research institutions with a predominantly regional reach of collaboration activities).
According to the literature a range of controls exist. In our study several control variables have been integrated, which are selected based on theoretical and empirical reasons. Thereby we distinguish two kinds of factors: internal and external to the firm (chart 2).
Firm size With regard to the firm size as a factor of the innovation performance the literature shows different results. On one hand it is mentioned that the process of innovation is driven by large established companies with a high market share. This idea of Schumpeter (1946) led to an ongoing discussion. The following aspects corroborate a lower innovative activity of SME: the internal financing out of profits is difficult due to lower production capacity. Beside of this SME mostly exhibit only a small or less diverse R&D base (R&D department), so that R&D capacity is correspondingly low (Nelson 1959). In addition, the access to external financing sources for the implementation of innovation projects is very difficult, especially for smaller firms (Rottmann 1995). On the other hand, SME have shorter decision paths, they focus more on market niches and because of their flexibility and specialization, especially in terms of customer needs, they often develop new products and processes (see also De Jong & Vermeulen 2006).
For this reason the relationship between firm size and innovation activity is amply discussed in the literature, which shows different results (Hausman 2005, Freel 2005, Shefer & Frenkel 2005, Wagner et al. 2005, Avermaete et al. 2004, Bhattacharya & Bloch 2004, Rogers 2004). For example a positive relationship between firm size and the number of product innovations was found by Kang and Kang (2009), Tether (2002) and Griffith et al. (2006). But Garcia-Torres and Hollanders (2009), De Jong and Freel (2010) and Hanson (1992) revealed a negative significant coefficient for firm size (Kang et. al. 2009). In addition, Kuemmerles´ (1998) results indicate a concave relationship between laboratory size of multinational companies and research performance. Also Chang and Robin (2006) show an “inverted-U” pattern between the size of Taiwan firms and R&D intensity and/or technology import intensity.
Age of firm The theoretical background and the empirical results of this control variable are manifold. For example Audretsch (1995) deals with the relationships among entry, post-entry growth, the role of incumbents and innovation. On one hand there is the
proposal that incumbents have more experience, e.g. substantial R&D knowledge, and the firm performance will improve over time due to organizational learning (Kuemmerle 1998). On the other hand start-up firms tend to innovate more quickly than incumbents due to the fact that in the stage of entry, firms have to explore the value of new ideas in an uncertain context (Kang & Kang 2009).
The influence of firm age on innovation performance has been investigated in several studies, which showed different results. Agarwal (1998: 215) relates small firms´ survival to innovative performance. But Kang and Kang (2009) could not detect this positive significant influence of a “start-up” variable on the number of product and process innovations. Hanson (1992) discovered that both firm size and firm age tend to be inversely related to innovative output. Huergo and Jaumandreu (2004) find that the probability of innovation varies as to entry, post-entry and advanced-ages. Their results are that entrant firms tend to present the highest probability of innovation while the oldest firms tend to present lower probabilities. But there are also empirical studies, that the firm age has no significant influence on product and process innovations (Freel 2005, De Jong & Vermeulen 2006).
Industry dummies Already Malerba und Orsengio (1997) discussed the existence of differences across sectors in the patterns of innovation and similarities across countries in the patterns of innovation for a specific technology. They proposed “that the specific pattern of innovative activity of a sector can be explained as the outcome of different technological regimes that are implied by the nature of technology and knowledge. The notion of technological regime provides a synthetic representation of some of the most important economic properties of technologies and of the characteristics of the learning processes that are involved in innovative activities” (Malerba & Orsengio 1997). On closer examination of the literature it turns out, that most of the studies cover only a certain group of firms: Manufacturing and services with 20 employees and more (Robin & Schubert 2010), manufacturing (Antonelli & Fassio 2011), micro and nanotechnologies (Autant-Bernard et al. 2007), biotechnology (Audretsch & Stephan 1996), science-based firms (Krätke 2010), high-technology corporations (Cummings & Teng 2003), certain two-digit industries (Anselin et al. 2000). To detect the possible variations across sectors in the determinants of innovation performance, numerous Projekt KompNet2011 Erfolgsfaktoren regionaler Innovationsnetze studies include industry dummies as control variables. The delineation of the industry variable and also the findings of these studies are heterogeneous. For example Mohnen, Mairesse and Dagenais (2007), based on the micro-aggregated firm data from CIS1, compare manufacturing industries in seven European countries. They select a number of explanatory variables for the propensity to innovate and the intensity of innovation. They conclude that their “innovation framework already accounts for sizeable differences in country innovation intensity, more so in the hightech than in the low-tech sectors” (Mohnen et al. 2002).
Upon a database of 1250 small firms De Jong and Vermeulen (2006) analyse the determinants of product innovation across seven industries (manufacturing, construction, wholesale and transport, retail, hotel and catering, knowledge-intensive service and financial service firms). They detect that “firms from manufacturing, knowledge-intensive services and financial service industries scored better on most innovative practices and realised new product introductions more often compared to firms from construction, wholesale and transport, retail services and hotel and catering services” (De Jong & Vermeulen 2006).