Description:
We study optimal subsidies for renewable energy (RE) generation to internalize external benefits from intertemporal learning-by-doing spillovers, taking into account increasing marginal costs at the industry level due to limited availability of sites suitable for RE. We find that the optimal RE subsidy is differentiated according to productivity and derive a condition on production and spillovers under which less efficient, i.e. more costly, technologies should receive higher support, as common in actual policy-making. We show that such a support of technological diversification is optimal if (i) the elasticity of learning by doing is large, which means that technologies rapidly mature with little further scope for learning, and if (ii) productive sites are scarce, which limits future utilization of knowledge.