This paper analyzes the effect of public R&D subsidies on firms' private R&D investment per employee and new product sales in German manufacturing. Parametric and semiparametric two-step selection models are applied to this evaluation problem. The results show that the average treatment effect on the treated firms' R&D intensity is positive. The estimated effects are robust with respect to the different selection models. Further results show that publicly induced R&D spending is as productive as private R&D investment in generating new product sales.