What do we know about R&D spillovers and productivity? Meta-analysis on heterogeneity and statistical power

Endogenous growth theory and the knowledge capital model predict that research and development (R&D) investment is associated with increasing returns and positive externalities. These insights have informed public support for R&D investment directly and indirectly. We aim to establish where the balance of the evidence lies, the extent to which the evidence has adequate statistical power, and which factors may explain the variation in the empirical findings. Drawing on 983 spillovers and 501 own-R&D effect-size estimates from 60 empirical studies, we find that the average productivity effect of spillovers: (i) is smaller than what is reported in most narrative reviews; (ii) is even smaller when only adequately-powered evidence is considered; (iii) differs by spillover types; and (iv) is not larger than that of own-R&D. We also report that the percentage of adequately-powered evidence is low (30%-55%). We highlight the implications of these findings for future research and public policy design.