Using the Variance Structure of the Conditional Autoregressive Spatial Specification to Model Knowledge Spillovers

Abstract

This study investigates the pattern of knowledge spillovers arising from patent activity between European regions. A Bayesian hierarchical model is developed that specifies region-specific latent factors modeled using a connectivity structure between regions that can reflect geographical as well as technological proximity. This approach exploits the fact that interregional relationships may exhibit industry-specific technological linkages or transportation network linkages, which is in contrast with traditional studies measuring the impact of geographical proximity on regional economic performance that rely exclusively on the geographic or spatial configuration of the regions. A series of model comparisons provides support for the model based on technological proximity over that based on spatial proximity alone. Estimates of the latent factors produced by the model can be used to draw inferences regarding spatial clusters of regions that exhibit high or low levels of knowledge spillovers. The method is illustrated using sample data on patent activity covering 323 regions in nine European countries.

Keywords: Markov chain Monte Carlo, spatial correlation, knowledge proximity, innovation

JEL Classification: C11, O31, R12