EXPLAINING THE VARIATION IN THE EMPIRICAL ESTIMATES OF ACADEMIC KNOWLEDGE SPILLOVERS*

ABSTRACT This paper presents a meta‐regression analysis of the empirical studies exploring the relationship between academic knowledge spillovers and regional innovation. The results of the literature are found to depend on country effects, the level of geographical aggregation employed, the method of measuring innovative output, the use of industry‐level data, and on the sectoral composition of the sample. Endogeneity, sample selection, and agglomeration forces are further elements that explain the findings on this issue but only weak evidence, after controlling for publication bias, supports the existence of a genuine spillover effect of university research on regional innovation.