University as a Knowledge Source of Innovation: A spatial analysis of the impact on local high-tech startup creation

This study contributes to the empirical analysis of specific distances in knowledge spillover effects. We propose a geographical distance-based approach to precisely measure the proximity of knowledge spillover from a university's research activities to high-tech startups in surrounding regions. Most current research measuring knowledge spillover typically use states and cities as the statistical caliber, making it difficult to capture the exact extent of knowledge spillover within cities. In this study, we constructed panel data for Japan for 1998-2018 by dividing the research area into 1*1 km2 meshes and geocoding firms (high-tech startups and firms without patents), university patents, and paper data, and subsequently using each mesh as the basic unit. Additionally, variables containing geographical proximity information were calculated by constructing multiple buffers for each mesh. Our findings show that i) the spillover effects of university research attenuate with distance - rapidly within a 2 km range, and slowly thereafter; and ii) patents are more private and localized than papers. The knowledge spillover effect of university patents attenuates more rapidly with distance.