This paper estimates regional innovation performance in Italian regions for the 2018-2023 period, using the Summary Innovation Index (SII) as a composite measure. Fixed-effects as well as random-effects panel specifications enable the estimation of various structural factors that favor regional innovation systems. International scientific co-publications (COIN) as well as the share of highly cited publications (CITE) indicate positive and significant relationships with SII, supporting global cooperation in research and scientific excellence. Product-innovating SME prevalence (PRSM) and public-private research co-authorships (PPRV) also emerge as strong indicators in favor of science-industry linkages boosting regional innovation capacities. Confidence intervals for innovation co-collaborations of SMEs (COLS) indicate negative relationships with SII, possibly indicative of regional innovation network ineffectiveness or counter-productive co-operations. Non-technological innovation, in terms of industrial design applications (DSGN) as well as sales of innovative goods (SALE), also contributes positively, in particular in random-effects estimates. Machine learning based on Random Forest regression confirms DSGN, COIN, and PPRV as the most prominent variables for predicting regional innovation outcomes, while CITE, as well as COLS, indicate relatively weak predictive abilities. Cluster analysis identifies six regional innovation profiles. They vary from high-performing, well-integrated regions to weakly structured ones with limited knowledge diffusion and commercialization. The paper's results clearly indicate the multifaceted nature of regional innovation, as well as the need for specific, targeted policy interventions. Specifically, one must foster public-private partnerships, promote openness to the entire world, and favor commercialization channels for balanced and sustainable innovation in the regions.
Innovation Without Borders? International Openness, Design, and the Uneven Geography of Ital-ian Innovation
alberto costantiello
2025-01-01
Abstract
This paper estimates regional innovation performance in Italian regions for the 2018-2023 period, using the Summary Innovation Index (SII) as a composite measure. Fixed-effects as well as random-effects panel specifications enable the estimation of various structural factors that favor regional innovation systems. International scientific co-publications (COIN) as well as the share of highly cited publications (CITE) indicate positive and significant relationships with SII, supporting global cooperation in research and scientific excellence. Product-innovating SME prevalence (PRSM) and public-private research co-authorships (PPRV) also emerge as strong indicators in favor of science-industry linkages boosting regional innovation capacities. Confidence intervals for innovation co-collaborations of SMEs (COLS) indicate negative relationships with SII, possibly indicative of regional innovation network ineffectiveness or counter-productive co-operations. Non-technological innovation, in terms of industrial design applications (DSGN) as well as sales of innovative goods (SALE), also contributes positively, in particular in random-effects estimates. Machine learning based on Random Forest regression confirms DSGN, COIN, and PPRV as the most prominent variables for predicting regional innovation outcomes, while CITE, as well as COLS, indicate relatively weak predictive abilities. Cluster analysis identifies six regional innovation profiles. They vary from high-performing, well-integrated regions to weakly structured ones with limited knowledge diffusion and commercialization. The paper's results clearly indicate the multifaceted nature of regional innovation, as well as the need for specific, targeted policy interventions. Specifically, one must foster public-private partnerships, promote openness to the entire world, and favor commercialization channels for balanced and sustainable innovation in the regions.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
