The objective of this study is to identify knowledge spillovers that spread across regions in Europe and vary in magnitude for different industries. The study uses a panel of <br/>203 NUTS-2 regions covering the 15 pre-2004 EU-member-states to estimate the impact over the period 1998-2003, and distinguish between five major industries. The study implements a fixed effects panel data regression model with spatial autocorrelation to <br/>estimate effects using patent applications as a measure of R&D output to capture the contribution of R&D (direct and spilled-over) to regional productivity at the industry level. The results suggest that interregional knowledge spillovers and their productivity effects are to a substantial degree geographically localised and this finding is consistent with the localisation hypothesis of knowledge spillovers. There is a substantial amount of <br/>heterogeneity across industries with evidence that two industries (electronics, and chemical <br/>industries) produce interregional knowledge spillovers that have positive and highly significant productivity effects. The study, moreover, confirms the importance of spatial autoregressive disturbance in the fixed effects model for measuring the TFP impact of interregional knowledge spillovers at the industry level.
|Seiten (von - bis)
|1 - 16
|Romanian Journal of Regional Science
|Veröffentlicht - 1 Mai 2007