Modeling Myths: On the Need for Dynamic Realism in DICE and other Equilibrium Models of Global Climate Mitigation

Abstract

We analyze and critique how optimizing Integrated Assessment Models, and specifically the widely-used DICE model, represent abatement costs. Many such models assume temporal independence –abatement costs in one period are not affected by prior abatement. We contrast this with three dimensions of dynamic realism in emitting systems: inertia, induced innovation, and path dependence. We extend the DICE model with a stylized representation of such dynamic factors. By adding a transitional cost component, we characterize the resulting system in terms of its capacity to adapt in path-dependent ways, and the transitional costs of accelerating abatement. We formalize a resulting metric of the pliability of the system, and the characteristic timescales of adjustment. With the resulting DICE-PACE model, we show that in a system with high pliability, the optimal strategy involves much higher initial investment in abatement, sustained at roughly constant levels for some decades, which generates an approximately linear abatement path and emissions declining steadily to zero. This contrasts sharply with the traditional formulation. Characteristic transition timescales of 20-40 years result in an optimum path which stabilizes global temperatures around a degree below the traditional DICE behavior; with otherwise modest assumptions, a pliable system can generate optimal scenarios within the goals of the Paris Agreement, with far lower long run combined costs of abatement and climate damages. We conclude that representing dynamic realism in such models is as important as – and far more empirically tractable than – continued debate about the monetization of climate damages and ‘social cost of carbon’.