Dynamic Technology Subsidies : Paying People Not to Wait ∗

We analyze how a policymaker should structure subsidies over time in order to induce agents to adopt a new durable good technology. We theoretically demonstrate that, if agents are myopic, a policymaker can price discriminate by using a subsidy that increases over time but can take advantage of technological change by using a subsidy that decreases over time. However, if agents plan their adoption around future subsidies, then the policymaker commits to low future subsidies in order to induce adoption in early periods. In order to measure the relative importance of these different channels, we estimate a dynamic model of household preferences in the California residential market for rooftop solar photovoltaics. We show that households’ expectations of future subsidies and technological progress increase the cost of the efficient policy by 50% and limit the regulator’s ability to reduce total spending by taking advantage of technological progress. Using the efficient subsidy could have allowed the state to spend 31% ($46 million) less to achieve the same level of adoption.