Knowledge, innovation and economic growth : spatial heterogeneity in Europe

In this paper, we aim at re-assessing the undisputed positive relationship between innovation and economic growth by questioning the view that R&D (and formal knowledge in general) equates innovation and innovation equates regional growth. We rather propose that these linkages are strongly mediated by local territorial assets and explore this relationship at the regional level (NUTS2) for 262 regions of the European Union. In doing so, we rely upon an original database encompassing several knowledge and innovation indicators, ranging from R&D expenditures, patent data, to newly released data on different types of innovation: product, process and marketing and/or organizational innovation, derived from the Community Innovation Survey 2002-2004 wave. The data set also includes several variables aimed at capturing different elements characterizing possible different attitudes and patterns of innovation that we control for, such as regional preconditions for knowledge and innovation creation and acquisition (namely, accessibility, trust, structural funds funding, foreign direct investments). The results of the analysis confirm that R&D is an important driver of economic growth. However, this result hides a larger territorial heterogeneity and needs some qualifications. Firstly, only regions strongly endowed with elements supporting knowledge creation processes are likely to benefit from the positive returns to R&D; a critical mass of R&D investments is therefore needed in order to exploit the eventual benefits arising from increasing returns to research expenditures. Secondly, once controlling for innovative behavior, R&D does not show anymore a significant impact on GDP growth. In fact, whereas the growth benefits accruing from R&D look rather selective and concentrated in a relatively small number of regions, the benefits accruing from innovation look not only of greater magnitude but more pervasive and beneficial for a larger number of regions. From these findings, we ultimately draw ad-hoc policy suggestions.