Widespread and increasing public subsidy for research and development (R&D) has given rise to a large and growing number of evaluation studies. While economic theory identifies market failures that justify public support, theory also suggests reasons why returns might be disappointing. Similarly, the empirical literature investigated – 52 micro-level studies published since 2000 on either input or output R&D – reports a wide range of findings. The lack of conclusiveness both of theory and of the evaluation literature motivate this Meta-Regression Analysis (MRA). This study contributes to policy debate by identifying a representative subsidy effect: after controlling for publication selection bias and for a wide range of sample and study heterogeneities, MRA findings reject crowding out of private investment by public subsidy but reveal no evidence of substantial additionality. In addition, among the research practices explaining the heterogeneous effects reported in this literature, those related to the treatment of unobservable firm heterogeneity are particularly important.