Innovation-Led Transitions in Energy Supply∗

I generalize a benchmark model of directed technical change in order to reconcile it with the historical experience of energy transitions. I show that the economy becomes increasingly locked-in to the dominant sector when machines and energy resources are substitutes, but a transition away from the dominant sector is possible when machines and energy resources are complements. Consistent with history, a transition in research activity leads any transition in resource supply. A calibrated numerical implementation shows that innovation is critical for climate change policy. A policymaker uses a U-shaped emission tax trajectory so as to immediately transition innovation away from the fossil sector, wait for clean technology to improve, and then hasten a transition in resource supply later in the century.