Belfer Center, Kennedy School, Harvard Univ. / by Nidhi R. Santen, et al.
We present a new modeling framework for studying optimal generating capacity and public RD&D investments in the electricity sector under decision-dependent RD&D uncertainty and learning. We show that when uncertainty and learning features are omitted, as is typically done in practice, the investment strategy can be considerably different from the optimal solution. We use a bottom-up stochastic power generation capacity expansion model with uncertain endogenous RD&D-based technical change, and focus on solar PV RD&D for its current prominent role in the U.S. national energy and climate policy discussion. Uncertainty in the outcome of RD&D investments is characterized using novel expert elicitation data, allowing for a transparent and consistent integration into the framework. The problem is formulated as a multi-stage decision under uncertainty to represent opportunities for policymakers to learn and adapt to new information between decision stages. Results show that under a carbon constraint, the optimal investment strategy includes lower solar PV RD&D spending upfront but more RD&D spending later—and sometimes higher spending overall—when compared to a strategy under perfect foresight about RD&D outcomes, or based on single-shot decision-making under uncertainty without learning. We also show that when uncertainty is considered without learning, new solar PV capacity investments are depressed. Overall, the results caution that the most common approaches used by the research community today to inform policy on optimal energy technology investments, including scenario analysis and Monte Carlo simulation that assume perfect foresight with no learning, may be systematically over- or under-estimating the optimal investment strategy.