Publication bias in the returns to R&D literature

We conduct a meta-analysis of the private returns to R&D literature and find that 26 % of the results in a hypothetical complete literature are missing. The combined estimate for the marginal returns to R&D falls from 18.2 % to 13.8 % when publication bias is accounted for. The latter estimate is very low, as it is gross of depreciation. One interpretation is that publication bias has contributed to masking the need for a revision of the standard R&D capital model.