Model-based net-zero scenarios, including those of the IPCC, aren’t worth the paper they are written on, say leading economists

16/06/22
Author: 
David Spratt
Teaser photo credit: Plenary session of the COP21 adopting the Paris Agreement in 2015. By UNclimatechange from Bonn, Germany – they did it!, CC BY 2.0, https://commons.wikimedia.org/w/index.php?curid=81571199

Jun 10, 2022

originally published by Climate Code Red

World-leading economists have blown a hole right through the middle of the main tool used to produce the net-zero scenarios embraced by climate policymakers.

In a new paper, Sir Nicholas Stern, Nobel Laureate Joseph E. Stiglitz and Charlotte Taylor conclude that climate-energy-economy Integrated Assessment Models (IAMs), which are the key tool in producing emission-reduction scenarios, “have very limited value in answering the two critical questions” of the speed and nature of emissions reductions and “fail to provide much in the way of useful guidance, either for the intensity of action, or for the policies that deliver the desired outcomes”.  The research paper is The economics of immense risk, urgent action and radical change: towards new approaches to the economics of climate change.

Now this is a big thing, because IAMs are at the centre of the IPCC Working Group III report on mitigation, and “have played a major role in IPCC reports on policy, which, in turn, have played a prominent role in public discussion. They continue to play a very powerful role in the research activities of economists working on climate change.”

And it is not just the IPCC. The net-zero plans which are the focus of international discussion — such as those of the International Energy Agency and the world’s bankers (Network for Greening the Financial System) — are fundamentally dependent on IAMs.

Last year Breakthrough’s report, Degrees of Risk, described how an objectivity is bestowed on IAMs that they do not serve: they are highly subjective tools with unrealistic assumptions that more than anything else reflect modellers’ view of society. Depending on how modellers perceive the roots of the problem to be solved, they will “design the model structure, including possible instruments and relationships within the model accordingly… Hence, the very structure of a model depends on the modeller’s beliefs about the functioning of society”. Consequently, IAM results have the capacity to privilege particular pathways and entice policymakers into thinking that the forecasts the models generate have some kind of scientific legitimacy.

A stunning example is the set of net-zero 2050 (NZ2050) scenarios being promoted by the world’s central bankers. They have up to 50% of primary energy use coming from fossil fuels in 2050, “offset” by use of unreliable carbon accounting and land-based measures — including bio-energy — which impinge upon and may decrease land for cropping even as demand for food increases by half over the next 30 years. This agenda is supported by the fossil fuel industry and by some “climate activist” investor groups.

The economists’ research identifies a number of fundamental problems with IAMs, some of which are intrinsic to their structure and cannot be remedied. Perhaps the biggest is the deep uncertainty that surrounds climate change, let alone the uncertainty in trying to map those physical changes into social and economic impacts. Deep uncertainty exists “where the outcomes cannot be fully described” and hence cannot be quantified. Yet without quantification of future costs and benefits, IAMs simply cannot compute: they break down and can provide no answer.

An allied problem is the failure of IAMs to deal with extreme risk (such as “fat tails”, where there is a higher probability of occurrences at the high end of the range of possibilities). As the authors note, “the world has been much more focused than the IAMs on a different set of issues, the risks of catastrophic consequences”.  They say:

“When we combine the uncertainties—about climate science, about the “right” economic model, about the parameters of the models, about the changes in those parameters over time, about the political processes which affect both the environment and economy— climate policymaking is a quintessential example of decision making under uncertainty, where the decisions themselves affect the magnitude of the uncertainties.

“The standard IAMs pay only limited attention to risk and ignore uncertainty. While in some cases the former can, with some difficulty, be incorporated into the analysis, the latter cannot…

“Deep uncertainty—where individuals do not know all of the possible outcomes and the probabilities of occurrence of different possible events, and know that they don’t know— presents even more fundamental methodological difficulties for the IAM. In particular, it gives rise to incomplete orderings. “This is inconsistent with the underlying (but rarely stated) hypothesis in the IAMs that there is a complete ordering. And, it is not just the incompleteness per se of the ordering that is the problem; it is where the incompleteness arises, which is the inability to describe and evaluate some of the most important, catastrophic outcomes.

“In short, in the presence of deep uncertainty, the framework of maximising expected utility lacks credible foundations, and under plausible assumptions, rational individuals undertake more precautionary behaviour than might be suggested by such a model.”

This is but one of a number of fundamental flaws identified by Stern, Stiglitz and Taylor. Others include:

  •  Systematic underestimation of future damages;
  •  Unjustifiably high discount rates;
  • One-sided assumptions about carbon prices being the primary de-carbonisation driver;
  • Flawed descriptions of how the underlying economy actually works: unworldly assumptions about utility maximisation, and descriptions about structural change, markets and technologies that turn a blind eye to market failures, dislocation and how markets actually function.

As noted in Degrees of Risk, The NGFS scenarios chronically underestimate future damage, using IAMs that basically ignore non-linearities and cascades. The NGFS admits its damage estimate from physical risks “only cover a limited number of risk transmission channels. For example, they do not capture the risks from sea-level rise or severe weather. They also assume socio-economic factors such as population, migration and conflict remain constant even at high levels of warming.” This, in itself, is enough to disqualify these scenarios from being seen as credible stories about alternative futures.

Can more realistic damage functions improve IAMs? Studies with more realistic damages can bring  IAMs into line with Paris goals, but while they might make the IAMs seem more reasonable, these studies simply illustrate, but do not resolve, the extreme sensitivity of results in these models, which make them very weak frameworks for policymaking: “the problems of foundational weaknesses associated with extreme risk, deep uncertainty and the challenges of structural change are still not resolved” say Stern, Stiglitz and Taylor.

And rather than reply on carbon prices, a diversity of policies is required:

“The IAMs have been particularly influential in the calculation of carbon prices for use in appraisal of public programmes. They have been used to suggest relatively low levels for carbon taxes…  Given the severe limitations of the IAMs, and the multitude of key issues that arise in fostering a green transition, it makes sense to adopt a diversity of approaches to analysis and modelling to understand and illuminate these issues and relevant policy responses.”

Stern, Stiglitz and Taylor conclude that even “new and improved” IAMs are likely to be of only limited value in providing policy guidance. Even if the assumptions are made more realistic, these models “have been shown to be very unrobust; they are not helpful for guiding policy decisions, unless we are very confident in all of the key functions and their parameters—which we cannot be”.

And an alternative? Rather than IAMs which are based on cost-benefit analysis in a world where the costs (damages) simply cannot be quantified, it is suggested that a better approach is to set a reasonable, consensus goal, and then find the best way of achieving that goal:

“The difference is analogous to that between cost-benefit and cost- effectiveness analysis. In many arenas of policy where the benefits are hard to evaluate— wars, regulations which affect health, safety and life itself or biodiversity—there is often resort to cost-effectiveness analysis. An agreement is first reached on goals and constraints, and economic analysis centres on the best way to achieve the given goals within the constraints.”

 

Teaser photo credit: Plenary session of the COP21 adopting the Paris Agreement in 2015. By UNclimatechange from Bonn, Germany – they did it!, CC BY 2.0, https://commons.wikimedia.org/w/index.php?curid=81571199