Of Trees and Monkeys. The evolution of technological specialization of European Regions

How do regions develop and evolve along their productive and technological path is a central question in many scientific fields from international economics, to economic geography, from public policy to regional science. Within an evolutionary perspective, we believe that, in general, a given region is most likely to develop new industries or new technologies closer to its pre-existing specialization. Our research builds on an empirical stream of literature, started by Hausmann and Klinger (2007) and Hidalgo et al. (2007), aimed at tracing the world evolution of industrial specialisation, at the country level, following the evolution of export portfolios. We refocus this line of analysis on the regional European technology/knowledge space along the research avenue started by Kogler et al. (2017). We aim at investigating the pattern and the evolution of regional specialisation in the most innovative EU countries in terms of the interaction of three factors: (i) endogenous processes of knowledge recombination and localised technological change, (ii) exogenous technological paradigm shifts and (iii) trans-regional spatial and technological spillovers and networking dynamics. More specifically, our paper maps the technological trajectories of 198 EU regions over the period 1986-2010 by using data on 121 patent sectors at the NUTS2 level for the 11 most innovative European countries, plus Switzerland and Norway. We map the knowledge space following two distinct and complementary approaches: a micro-level one, based on co-classification information contained in patent documents (as in Engelsman and Van Raan, 1992; Kogler et al., 2017), and a macro-level one, based on conditional co-specialisations of regions in the same patent classes (as in Hidalgo et al., 2007). These two representations of the knowledge space serve as reference bases for understanding the evolution of regional technological specialization, being measured in terms of the sector-region relative technological advantage (RTA), and for modelling its dynamics as a function of spatial, technological and socio-cognitive proximity. The results show a significant path dependence in the evolution of the regional technological specialisation, whose changes are significantly shaped mostly by phenomena of localised technological change and recombinant innovation. We also find evidence of a significant role played by spillovers and neighbourhood effects in the form of geographic and technological spillovers.