A tale of persistent network additionality, with evidence from a regional policy

CIMR Research Working Paper Series, WP No. 38, 10/05/2017, ISSN 2052-062X

Annalisa Caloffi, Department of Economics and Management, University of Padua, Italy, annalisa.caloffi@unipd.it
Federica Rossi, Department of Management, Birkbeck College, University of London, United Kingdom, f.rossi@bbk.ac.uk
Margherita Russo, Department of Economics Marco Biagi, University of Modena and Reggio Emilia, Italy, margherita.russo@unimore.it

Adopting a counterfactual approach to the evaluation of a regional R&D collaboration policy, carried out in Tuscany (Italy), we investigate different types of persistent network additionality, namely persistence effect, breadth effect, composition effect, and depth effect. Our findings reveal that this R&D collaboration policy has been able to generate some persistent changes in the networking behaviour of participating firms, particularly fostering their collaboration with universities. Network additionality has been greater for firms that were previously accustomed to collaborating with other firms, than for less collaborative firms. With respect to the former firms, we also find a composition effect, which implies a change in their type of partners in innovation-related activities. We find, instead, no evidence of network
breadth and network depth effects.

Key words: Network additionality, network persistence, policy evaluation, R&D
collaboration policy, Italy.
JEL code: O38, O32, D04

Acknowledgements: We are grateful to Tuscany’s Regional Government for granting us free access to the policy data used in this work and for supporting the collection of data used in the counterfactual analysis. We also wish to thank Maurizio Tessieri, Fulvia Bracciali and the staff of Sviluppo Toscana for their support with data collection and with the management of the survey. Dr. Albino Caporale from the Tuscany Regional Administration has supported us with an open dialogue on the challenges of the analysis. Elena Pirani from the University of Firenze provided a valuable contribution to the design of the survey, and Marco Mariani from IRPET gave us some helpful suggestions and comments on earlier versions on the paper. Finally, we thank Helen Lawton-Smith and Klaus Nielsen from Birkbeck, University of London, for helpful comments on a first draft of the paper. All errors and omissions are our own.