An Analysis of Regional Policies Promoting Networks for Innovation

Abstract

The diffusion of system-based innovation policies calls for the development of an appropriate evaluation framework. Such a frame requires a careful definition of the unit of analysis and evaluation, since evaluation based on the sum of individual additional results cannot assess appropriately the emergence of system effects. Moreover, it requires experimenting with new tools for measuring the relational effects of the new policies. The paper aims to contribute to this challenge, proposing an analysis of the inner structure and the organization of regional innovation systems in terms of network relations, and considering its results for their potential contribution to the evaluation of innovation policies in a systemic perspective. The empirical application focuses on a set of policies implemented by the Tuscany Region in Italy. It highlights that the same policy intervention may lead to the emergence of different relational architectures connecting the world of research and that of production, depending on the different relational context (technological/sectoral and territorial) in which they are grounded. These two aspects—the structure of the relations and the context in which they develop—must be carefully combined in order to identify the effects of policies aimed at promoting innovation.

Acknowledgements

Sections 4–7 are to be attributed to Annalisa Caloffi. The authors wish to thank Margherita Russo from the University of Modena and the participants to the XI EUNIP Conference for their valuable comments. Part of the research work has been carried out within the framework of the DISTRICT-RFO project INTERREGIIIC EU programme (Bellandi & Caloffi, 2008).

Notes

See the cases of Sweden (Klofsten et al., 1999), Finland (Gheorghiou et al., 2003) and Germany (Eickelpasch & Fritsch, 2005).

The database includes a set of seven interventions implemented by the Tuscan regional government within the DOCUP-SPD 2000–2006 and the RPIA-ITT programme (Regional Programme of Innovative Actions “Innovazione Tecnologica in Toscana” 2001–2004, funded within the ERDF Innovative Actions framework) during the programming period 2000–2006. Though such period ends in 2006, some of the funded projects have been implemented during the following year. The empirical research was carried out over an extended time span, from 2004 to 2006, since the authors participated in the monitoring and analysis of two specific regional programmes implemented during this period, namely the RPIA-ITT programme 2001–2004 and the DOCUP-SPD, line 1.7.1, 2005–2006. The research is based on desk analysis of the cited projects and programmes, and on questionnaires and in-depth interviews with the project participants (Lazzeretti et al., 2007).

The first set of network projects was funded by the regional government in 2003 as a response to the EU RPIA framework, which asked the region to implement innovative policy actions. The Tuscany region has extended further this experimental intervention, including the promotion of innovative networks within some structural lines of intervention. The projects have benefited from a small amount of public funds (180,000 Euro on average for each innovative project).

The database includes all the projects funded in the period under consideration. The projects submitted by networks of actors that were not accepted for funding have been excluded. The project data refer to definitive projects, drafted in the format scheduled in the funding specifications. This format did not expressly require indication of the names of possible subcontractors, who have thus not been taken into consideration in our analysis.

The different departments of a university and the different Institutes of the CNR (National Research Council) have been considered as different actors.

Participation in the different calls for tenders were not restricted to SMEs, although only SMEs could directly benefit from the public funds. Our database includes only five large enterprises, which have been indicated as members of the project partnerships.

The classification is made on the basis of the location of the participant's registered office. Since almost all the observed actors have unique location, the latter coincide with the location of the productive plants. Although participation in the programmes was not restricted to the actors localized in the Tuscany Region, almost all of the actors included in our database (99%) are localized within the 10 provinces of the region. Only this kind of participant could benefit from the public funds.

The analysis has been performed using Ucinet Software (Ucinet 6 for Windows, provided by Analytic Technologies). For some specific operations, we have used also Pajek software, available at: http://vlado.fmf.uni-lj.si/pub/networks/pajek/ (De Nooy et al., 2005).

Besides the relational proximity generated by the co-participation to the innovative projects, we will analyse the nature, the sector of activity and the location of the different actors involved. While location is taken as a proxy of territorial proximity (or distance), the other two elements may be considered as proxi of cognitive and productive proximity.

The degree centrality measures the number of direct links possessed by an actor: the more links the latter has with the other nodes the more central it is within the network. The closeness centrality measures the shortest path (that is in terms of the minimum number of nodes) necessary for an actor to reach the other actors of the web. The most central actor is the one with the shortest path. In our network, a high centrality index is associated not only with those actors who participate in a large number of innovation projects or have broader partnerships, but also with those who have the largest number of indirect links, that is, to those collaborating with central actors. The third index is the betweenness centrality, which takes into account the indirect links an actor has, that is the number of actors to which an actor is indirectly connected through their direct links. The betweenness centrality index focuses on the location of an actor within a network, considering the control it has on the communication flows which develop among the other actors of the web. Considering all the pairs of nodes of the network, the more central an actor is, the more it lies upon pathways between pairs. This structurally advantaged position of being between other actors may give the actor the capacity to mediate contacts between other actors and/or to isolate actors or prevent contacts. In our network, an actor is central when it lies at the crossroads of an (important) set of relations.

In order to do that, the timescale observed has been divided into four different sub-periods: 2000–2003 (two programmes launched in 2003); 2003–2005 (three programmes); 2005–2006 (one programme); 2006–2007 (one programme). The core-network has been constructed by keeping only those relations which remain active for at least two different sub-periods.

The core-network is partitioned into four components: the largest one includes 152 agents (i.e. 94% of the total actors of the core-network); the remaining 10 actors are separated within three small groups (two dyads and a small group of six connected actors).

In our case, a local containment within the province is quite significant, since most of the clusters and industrial districts localized in the Tuscany region have a territorial extension which is smaller or approximately equivalent to that of the province. Therefore, infra-provincial linkages focussing on the specific technological/sectoral axes which is at the core of the cluster and district localized within the province are considered here as a proxy of the infra-cluster relations.

As defined by De Nooy et al. (2005), an “island” identifies a significant cluster of central agents connected with each other in a robust manner, since the total weights of the linkages connecting the agents inside the cluster must be larger than that of the linkages outside the cluster. The five islands displayed in Figure 1 are the result of the “Island” procedure performed with Pajek and subsequently imported in Ucinet. After some trials, we have extracted those line islands having a minimum dimension of three nodes and a maximum dimension of 20 nodes. By increasing the lower threshold, we lose the smaller islands of innovative actors, while the largest island of the Made in Italy remains visible.

Furthermore, most of the SMEs have participated only for 1 year-projects, which can be considered as a short time for an innovative activity.

In order to identify the four sub-networks, we have considered all the relations connecting each of the four groups of actors included in the islands with the other agents of the total network. Therefore, in this stage, we have reintroduced the analysis of the temporary relations.

As regards the “made in Italy” sectors, we refer to the definition adopted in the various funding specifications: textiles/clothing, footwear, marble, furniture and light (instrumental) mechanics. The public funding was aimed at promoting IT technology within these sectors.