Dynamically optimal R&D subsidization

This paper characterizes the optimal time path of R&D and capital subsidization. Starting from the steady state under current R&D subsidization in the US, the R&D subsidy should significantly jump upwards and then slightly decrease over time. There is a small loss in welfare, however, from immediately setting the R&D subsidy to its optimal long run level, compared to a time-varying R&D subsidy. The results do not depend on the financing scheme, namely lump sum taxation or factor income taxation. The optimal capital subsidy is time-varying under factor income taxation, but time-invariant when subsidies are financed by lump sum taxes.