The welfare cost of delayed climate policy

Abstract:

Climate policy is often seen as a timing problem: the choice between weaker action now and stronger action later. In the context of our model there is one policy variable, the carbon price. Today that price is close to zero, well below the optimum. We examine the cost to society of further delaying implementation of the optimal price that maximizes social welfare. We do this by studying a constrained optimal solution in a stochastic climate asset-pricing model with Epstein-Zin preferences, where the price is set at or close to zero for N years, before increasing it to the optimal level. Delay is costly because of increased climate damages. In our baseline calibration, with an optimal 2025 carbon price of ~$190/tCO2, a ten-year delay produces a consumption-equivalent deadweight loss (DWL) of ~4% of consumption per year of delay, corresponding to annual compensation of about $3.6 trillion throughout the decade. Once policy is implemented, optimal carbon prices and mitigation increase, but subsequent action cannot undo consumption losses created during the delay. Partial early mitigation has substantial value: allowing \(\sim\)30\% mitigation of baseline emissions, about half of the optimal rate, avoids about two thirds of the no-action welfare loss and implies a 2025 carbon price of ~$63/tCO2. Higher future carbon prices mitigate the delay, but only at a premium.

JEL: D62,G12,Q54
Keywords: Climate policy | Climate risk | Carbon pricing | Welfare economics | Policy delay

Significance statement:

Climate change is costly. Just how costly is the subject of many studies calculating the social cost of carbon and `optimal' carbon prices, but those price paths themselves are poor indicators of the welfare cost of delayed climate action. Higher climate damages increase prices, while innovation in carbon-cutting technologies lowers them. We here zero in on deadweight loss as a proxy for the welfare costs of delay and find enormous costs of even just a moderate delay of five or ten years, in a world that has been pushing off optimal climate policy for decades.

PNAS R&R; please contact for full draft.

Code: github.com/gwagnerGH/delay

Presentation: Dialogues in Complexity Workshop, Princeton, NJ (20-21 November 2025).