Abstract
The relationship between R&D investment and firm/industry productivity has been investigated widely following seminal contributions by Zvi Griliches and others from late 1970s onwards. We aim to provide a systematic synthesis of the evidence, using 1253 estimates from 65 primary studies that adopt the so-called primal approach. In line with prior reviews, we report that the average elasticity and rate-of-return estimates are positive. In contrast to prior reviews, however, we report that: (i) the estimates are smaller and more heterogeneous than what has been reported before; (ii) residual heterogeneity remains high among firm-level estimates even after controlling for moderating factors; (iii) firm-level rates of return and within-industry social returns to R&D are small and do not differ significantly despite theoretical predictions of higher social returns; and (iv) the informational content of both elasticity and rate-of-return estimates needs to be interpreted cautiously. We conclude by highlighting the implications of these findings for future research and evidence-based policy.
Highlights
Productivity effects of research and development (R&D) investment has been a subject of major interest for researchers and policy makers
We suggest that the informational content of the productivity estimates can be enhanced by: (i) availability of firm-level deflators and depreciation rates; (ii) identifying the factors that affect firms differently as they choose their inputs, including R&D capital; and (iii) using richer models preferably with industry-level data to (a) take account of the lag structure in the relationship between R&D capital and productivity, (b) disentangle social returns to R&D from cross-sectional dependence due to unobserved common factors, and (c) incorporate technology class and market power into the technological progress component of the model
Until further progress along these dimensions, all we can infer is that R&D investment is associated with positive private and social returns, but the magnitude of the estimated effects is likely to fall short of reflecting the ‘true’ productivity or rate-of-return concepts implied by the underlying theory
Summary
Productivity effects of research and development (R&D) investment has been a subject of major interest for researchers and policy makers. The fourth finding we report indicates that the gross private rate of return at the firm level (14%) is less than the depreciation rate for R&D capital (15%) usually assumed in the primary studies This anomaly clearly suggests that the existing estimates suffer from a serious downward bias as suspected by Griliches and Mairesse (1991a). The analytical and empirical framework summarized above is fairly tractable and allows for pooling the existing estimates for meta-analysis This quality should not conceal the potential for high levels of heterogeneity due to measurement, identification, sampling and estimation issues discussed widely by leading contributors to and reviewers of the research field (Griliches, 1979; Griliches and Mairesse, 1995; Hall et al, 2010). The expected effect of publication types is informed by meta-analysis studies, which report that selection may be related to publication types (Card and Krueger, 1995; Sterling et al, 1995; Stanley, 2008; Costa-Font et al, 2013)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
