Beyond DICE: What the New-Generation Integrated Assessment Models Can Teach Us About the Optimal Abatement Policy

printer-friendly version

On september 21, Riccardo Rebonato, Professor of Finance, EDHEC Business School, Programme Director (Implications of Climate Change on Asset Pricing and Investment Management), EDHEC Risk Climate Impact Institute delivered a keynote talk on optimal abatement with IAMs. He discussed "Beyond DICE: What the New-Generation Integrated Assessment Models Can Teach Us About the Optimal Abatement Policy".

Professor Rebonato discussed the follow issues:

Questions

  • What does economic theory tell us about how rapidly we should ramp up our efforts to curb climate change? Shall we wait until we are “richer and smarter” (Nordhaus)? Or should we act immediately and decisively now (Stern)?
  • Why do similar models give such different answers?
  • Is it true that only ‘almost infinite altruism’ can justify aggressive abatement?
  • Is a model-based analysis meaningful, given the deep uncertainty in many model inputs?

Answers

  • If we use utility functions that recover stylized investment facts, even with high discount rates aggressive abatement becomes optimal.
  • Uncertainty about the damage function and future economic growth are the two main drivers of the result – they pull in different directions.
  • In arriving at these results, how much we care about avoiding feast and famine is as important as how ‘altruistic’ we are.
  • Negative emissions play a key role in the optimal policy. We should target net negative, not net zero. Without carbon removal, the social cost imposed by the ‘optimal’ abatement is too high.

Policy relevance

  • The Paris Agreement 1.5-2 C is no longer arbitrary but can be justified by an auditable model – this can re-engage bodies such as the IPCC that have turned their backs on Integrated Assessment models.
  • Subsidies and taxes have been focussed so far only on renewables – this is very inefficient. Substantial subsidies should be considered for NETs as well.