Regional knowledge spillovers: Fact or artifact?

The explanation of different levels of innovation and their spatial distribution represents the central focus of the analysis. The empirical literature documents the incidence of spatial autocorrelation in patenting activities and interprets them as evidence for knowledge spillovers. Alternatively, the authors propose the spatial pattern of input variables in innovation processes as driving forces of patenting autocorrelation. They analyze 51 Nuts 1 regions in Europe and find that the high degree of spatial autocorrelation exhibited by patent applications can be explained comprehensively by the spatial location of the input factors in the knowledge production function. These are traditional indicators on R&D investments and human capital from Eurostat and proxy variables on social capital from the European Values Study. This finding has important implications for the scope of an autonomous regional innovation policy.