Subsidized research joint ventures (RJVs) between public research institutions and industry have become increasingly popular in Europe and the US. We study the long-run effects of such a support scheme that has been maintained by the Danish government since 1995. To cope with identification problems we apply nearest neighbor caliper matching and conditional difference-in-difference estimation methods. Our main findings are that (i) program participation effects are instant for annual patent applications and last for three years, (ii) employment effects materialize first after one year and (iii) there are no statistically significant effects on value added or labor productivity. We further show that these overall results are primarily driven by firms that were patent active prior to joining the RJV and that there are no statistically significant effect for large firms. Both types of firms are disproportionally represented in the support program we study.