Optimal emissions under exogenous and endogenous learning

Some abatement technologies become cheaper over time due to spillover effects from other sectors. In a climate cost-benefit analysis, this process is “exogenous”, in the sense that it only depends on time. Other abatement technologies become cheaper only if they are deployed on a large scale. This “endogenous” process of learning by doing decreases abatement costs as a function of cumulative abatement. We aim to shed light on the differentiated impact of endogenous and exogenous learning on the optimal mitigation path. This is particularly important in a time when many models and scenarios are ignoring the dynamic characteristic of learning by doing. We develop a cost-benefit integrated assessment model which includes both types of learning dynamics as well as inertia. Theoretically, endogenous learning leads to a supplementary term in the optimality condition: the “learning gains”, whereas exogenous learning only creates an incentive to postpone climate action. We show analytically and numerically that including endogenous and exogenous learning steepens the abatement path. In a cost-benefit analysis, both types of learning leads to lower peak warming. Moreover, endogenous learning leads to negative emissions in the long run. Besides, the common practice
of modelling endogenous learning as an exogenous process underestimates optimal abatement by 9% in 2050.