Self-selection and performance of R&D input of heterogeneous firms: Evidence from China's manufacturing industries

•We estimate self-selection of firm R&D and the net effect of R&D on firm productivity.•We find that R&D input is self-induced by productivity of the manufacturing firms.•The output elasticity of R&D is 3.92% and 5.25% in one year or two after investment.•SOEs are most likely to invest in R&D, but the output elasticity is insignificant.•FDI firms are least likely to invest in R&D, but their R&D input is very productive.

Applying the theories of heterogeneous firms and the propensity score matching difference-in-differences (PSM-DID) method to a rich dataset of Chinese manufacturing firms, this paper examines the self-selection of firm-level R&D input and estimates the net effect of R&D on productivity. The analysis shows that (1) for Chinese manufacturing firms as a whole, R&D input is influenced by firm productivity: more productive firms are more likely to invest in R&D; (2) controlling for the self-selection effect, the net output elasticities of R&D input in one year and two years after R&D input are 3.92% and 5.25%, respectively; (3) although state-owned enterprises (SOEs) are more likely than all other ownership groups to invest in R&D, the R&D input is not productive; (4) although enterprises owned by investors outside of Mainland China are the least likely to invest in R&D, the output elasticity of R&D is more significant and larger in this group than in SOEs and privately owned Chinese firms; and (5) surprisingly, the net effect of R&D is not significant in high-tech industries. Policy implications are derived from the findings.