The relative effectiveness of R&D tax credits and R&D subsidies: A comparative meta-regression analysis

• Published in: Technovation 2022-07, Vol.115, p.102450, Article 102450
• Main Authors: Dimos, Christos, Pugh, Geoff, Hisarciklilar, Mehtap, Talam, Ema, Jackson, Ian
• Format: Article
• Language: English
• Subjects: Additionality; Meta-regression analysis; Policy evaluation; Publication bias; R&D subsidies; R&D tax credits
• ISSN: 0166-4972; 1879-2383
• DOI: 10.1016/j.technovation.2021.102450

There are large primary literatures that evaluate the effectiveness of either R&D tax credits or R&D subsidies in promoting private R&D. However, this Meta-Regression Analysis, by investigating these literatures jointly, is the first study that systematically measures and compares the effectiveness of these two policy instruments. After controlling for publication selection and sources of heterogeneity, we find that both tax credits and subsidies induce additional private R&D and that neither instrument systematically outperforms the other. However, whereas subsidy effects are increasing over time tax credit effects are not. Although their respective effects are “small”, they are not negligible: in round terms, an additional $1 of public R&D support of either type induces 7.5 cents of additional private R&D expenditure. Sources of heterogeneity in the reported effects include: tax credits are most effectively delivered as “incremental” schemes, are more effective in economies with a balanced “policy-mix” regime, and are generally less effective for micro firms and SMEs than for large firms; while subsidies are more effective for manufacturing firms, although not for high-tech firms, and are more effective than tax credits in economies predominantly using subsidies. Finally, we argue for the importance of statistical power in the design of evaluation studies. •R&D tax credits and R&D subsidies yield input additionality and are similarly effective.•$1 of either R&D tax credits or R&D subsidies induces 7.5 cents of additional private R&D.•The effectiveness of R&D support is conditional on firm and country heterogeneities.•The effectiveness of R&D subsidies tends to increase over time.•Both literatures suffer from publication bias and lack adequate statistical power.