Modeling myths: OnDICEand dynamic realism in integrated assessment models of climate change mitigation

Abstract

We analyze how stylized Integrated Assessment Models (IAMs), and specifically the widely‐used Dynamic Integrated Climate‐Economy model (DICE), represent the cost of emissions abatement. Many assume temporal independence—that abatement costs in one period are not affected by prior abatement. We contrast this with three dimensions of dynamic realism in emitting systems: (i) inertia, (ii) induced innovation, and (iii) path dependence. We review key evidence from the last quarter century on each of these three components. Studies of stock lifetime, dynamics of diffusion and past transitions suggest typical transition timescales of at least 20–40 years for the bulk emitting systems. The evidence that substantial innovation is induced by both prices and market deployment is unambiguous. Finally, both data and a rapidly growing literature demonstrate substantial path dependence in general, and specifically “carbon lock‐in and lock‐out.” Some stylized models in the past decade have incorporated technology learning, and others have considered inertia, but the combination of these factors is important and not yet evident. More complex hybrid IAMs with technology‐rich energy‐system models incorporate these factors, but their complexity has limited the wider understanding and influence of their underlying insights. Few if any global models fully represent path dependence. We conclude with likely implications drawing upon the empirical and modeling evidence accumulated, including results from extending DICE with a highly stylized representation of such dynamic factors. This suggests that dynamic interdependencies could multiply several‐fold the optimal level of initial abatement expenditure. This is because early abatement then also directly facilitates subsequent emission savings. The diversity of dynamic linkages across sectors and technologies also implies more nuanced policy than a single global carbon price. Thus, the issues explored in this review can radically change the general policy conclusions drawn from models, which, like DICE, neglect dynamic realism. This article is categorized under: Climate Economics > Aggregation Techniques for Impacts and Mitigation Costs

Levelized cost-based learning analysis of utility-scale wind and solar in the United States

“…In contrast, our results add to the body of literature that suggests that endogenous learning is a real phenomenon, and show that learning need not moderate with industry maturity. While the proper treatment of endogenous learning in energy and climate models is a complicated matter, at a minimum, our results suggest further effort in this direction is warranted—adding to other areas of possible model improvement ( Grubb et al., 2021a , 2021b ).…”

“…First, it is important that these communities account for non-constant learning. There is evidence that common energy-sector planning and integrated assessment models have not kept pace with the recent declines in wind and solar costs, and so may be understating the deployment potential for these technologies ( Xiao et al., 2021 ; Grubb et al., 2021b ), though recent research on historical growth rates suggests that other model features may result in overstated potential ( Cherp et al., 2021 ).…”

Impact of COVID-19 type events on the economy and climate under the stochastic DICE model

“…Of course, although the DICE model is a typical reference point for many climate-economy studies, it is important to remember that IAMs and the DICE models in particular have significant limitations (in the model structure and model parameters) and have been criticized and debated in the literature (for example, see the discussions in Ackerman et al (2009); Pindyck (2017); Grubb et al (2021); Weitzman (2011).…”

“…It balances parsimony with realism and is well documented with all published model equations; in addition, its code is publicly available, which is an exception rather than the rule for IAMs. At the same time, it is important to note that IAMs and the DICE model in particular have significant limitations (in the model structure and model parameters), which have been criticized and debated in the literature (see the discussions in Ackerman et al (2009); Pindyck (2017); Grubb et al (2021); Weitzman (2011)). Despite the criticism, the DICE model has become the iconic typical reference point for climate-economy modeling, and is therefore used in our study.…”

Institutional “Paris Agreement Compatible” Mitigation Scenarios Evaluated Against the Paris Agreement 1.5°C Goal

“…Our focus has been on institutional pathways, but similar limitations are valid for the growing literature of bottom-up energy modeling approaches that find the potential for 100% renewable energy based systems by mid-century [42][43][44][45][46][47][48][49][50][51] , which tend to outpace estimates of renewable penetration rates compared to IAMs 27,52,53 . A claim of 100% RE by 2050 may align with power-sector benchmarks for PA-compatibility, but it is not sufficient to guarantee these pathways meet the LTTG.…”