Evaluating the spatial heterogeneity of innovation drivers: a comparison between GWR and GWPR

In studies focusing on innovation activities, the potential spatial heterogeneity in the relationships between innovation and its triggering factors is an unexplored topic. On this ground, this paper aims to a twofold contribution. First, we verify the existence of spatial variability in the relationships. We evaluate the estimation gains due to local regressions, such as geographically weighted regression (GWR) and geographically weighted panel regression (GWPR), with respect to the classical global methods (e.g., OLS, Fixed Effects panel regression). Second, we compare the GWPR with GWR and global models to evaluate if the joint consideration of time and space dimensions allows for the rise of new insights. We resort to official data on 287 NUTS-2 European regions in 2014–2021. The results confirm that GWPR estimations significantly differ from GWR and global models, potentially producing new patterns and findings.