Estimating the Local Average Treatment Effect of R&D Subsidies in a Virtual Common Pot

We investigate the additionality effects of Eurostars, Europe's largest multilateral subsidy program for R&D-performing small and medium sized enterprises. A specific budget allocation rule serves as an instrument and allows us to identify the local average treatment effect of public R&D grants. This rule, referred to as Virtual Common Pot (VCP), is designed to avoid cross-subsidization between participating countries. We compare the program's effect under a VCP with the counterfactual situation under a Real Common Pot (RCP), where project authorities allocate a single budget according to uniform project evaluation criteria. Our estimates suggest a large positive impact on job creation whereas there is no treatment effect on patenting. In addition, we find a relative inefficiency of 19.4% more jobs which could be created by the program under a RCP.