The Legacy of Policy Inaction in Climate‐Growth Models

ABSTRACT To better understand the structure and core mechanisms of a broad class of climate‐growth models, we study a simplified version of the dynamic integrated model of climate and the economy (DICE) through the lens of growth theory. We analytically show that this model features a continuum of saddle‐point stable steady states. Initial conditions of stock variables, which are notoriously difficult to calibrate, matter for long‐run economic and climate outcomes. However, we also demonstrate that a misspecified initial stock of capital has a significantly smaller impact than a misspecified initial CO 2 stock. These insights have important implications for the consequences of delayed climate policy implementation and the optimal carbon tax. Using a calibrated version of the model, solved numerically for the big transition, we show that postponing optimal climate policy raises both peak temperature and the steady‐state temperature. The findings extend to a large set of analytical and numerical integrated assessment models. The simple DICE, augmented by the finite amplitude impulse‐response (FAIR) carbon cycle model, exhibits a continuum of steady states over time horizons spanning several centuries. Peak and long‐run temperatures depend on initial conditions, a result that is also confirmed for DICE‐2023. We further show that the social cost of carbon to GDP ratio is largely constant despite transitional dynamics; however, its level depends on the timing of optimal policy implementation.