A spatial panel data version of the knowledge capital model

This paper attempts to analyze the impact of knowledge and knowledge spillovers on regional total factor productivity (TFP) in Europe. Regional patent stocks are used as a proxy for knowledge, and TFP is measured in terms of a superlative index. We follow Fischer et. al (2008) by using a spatial-spillover model and a data set covering 203 regions for six time periods. In order to estimate the impact of knowledge stocks we use a spatial autoregressive model with random effects, which allows for three kinds of spatial dependence: Spatial correlation in the innovations, the exogenous and the endogenous variables. The results suggest that there is a significant positive impact of knowledge on regional TFP levels, and that knowledge spills over to neighboring regions. These spillovers decay exponentially with distance at a rate of 8%. Using Monte Carlo simulations we calculate the distribution of direct and indirect effects. The average elasticity of a region's TFP with respect to its own knowledge stock is 0.2 and highly significant. The average effect of all other regions' TFP is about 50% higher, which confirms that the cross-country externalities are important in the measuring of the impact.