Regional productivity growth in Europe: a Schumpeterian perspective

Using data for the European regions at NUTS-2 level, we test the predictions of a microfounded Schumpeterian growth model with technological interdependence recently developed by Ertur and Koch (2011, EK11). Spatial interdependence is identified by means ofa semiparametric geoadditive spatial autoregressive model which permits us to disentanglethe effect of nonlinearities, spatial heterogeneity and spatial dependence. A control function approach is applied to estimate this particular SAR-type model using the spatial lag of the quality of regional governance and of its components (corruption, rule of law, government effectiveness and accountability) as instrumental variables for the endogenous term Wy. The results corroborate the predictions of EK11’s model: R&D investments and R&D spillovers are important divers of regional growth in Europe. However, spillover effects are much lower after controlling for spatial unobserved heterogeneity. Moreover, important nonlinearities in the effect of physical capital investments emerge, putting into question the strong homogeneity assumption and suggesting a threshold effect in growth behavior.