Abstract:
The bread-and-butter of integrated assessment modeling has been the study of deep decarbonization needed to meet widely discussed climate policy goals such as stopping the rise in temperatures or cutting net emissions to zero. A troubling element of these models is their lack, to varying degrees, of proper representation of how policies are formulated and the impacts of policies and changes in technology on patterns of investment in clean technologies. In some ways, the models have yielded abundant elegant insights into deep decarbonization strategies that, increasingly, don’t reflect reality. This talk will outline some ways to improve the models and show suggestive results from a decade of collaboration with different modeling teams. The net effect of adding more realism is to increase pessimism about meeting global climate goals in the near term and more optimism about the long term. More realistic model assessments suggest that carbon removal technologies and climate resilience (and possibly geoengineering) need more policy priority. Realism also has implications for the geography of climate policy action and investment—with Europe occupying a more central role and the United States becoming less reliably relevant.
Bio:
David G. Victor is a distinguished professor of innovation and public policy at the School of Global Policy and Strategy at UC San Diego and also a professor in Climate, Atmospheric Science & Physical Oceanography at the Scripps Institution of Oceanography. He directs the UCSD-wide Deep Decarbonization Initiative (D2I) where his research focuses on the engineering, economic and political challenges associated with slashing emissions of warming gases and removing the gases that have already accumulated in the atmosphere and oceans.
Summary:
Focus: how to improve Integrated Assessment Models (IAMs) so that they are more predictive and informative
IAMs are a key tool for assessing climate policy and produce many forecasts that look at climate, energy, agriculture, investment, etc.
E.g. global uniform carbon tax is more tractable but highly unlikely
Modeling Strategy: focus on most important and tractable phenomena
Carbon caps and taxes
Huge variation in Policy challenges across the world and states within US
Question: how inefficient is it for states within a country to set their own climate policies vs a single national one?
IAMs suggest that the inefficiency is not high as long as there is smooth trade of power and po
Decarbonization via electrification / More capital-intensive energy system
small differences in capital costs will inform whether investments are/aren't profitable
Cost of capital varies across the world (higher in more risky countries, lower in more stable ones); makes it harder to upgrade infrastructure in developing world
This raises the carbon price needed to motivate investment
Indeed, we now see that most infrastructure investment in green tech is happening in stable, developed countries
Investors have limited visibility into the future
Modeled by having the modeled agents anticipate the future with varying levels of precision or time horizon (e.g. 0-5y, ~8y, 15+y)
New tech: by assumption and learning-by-doing efficiency improvements
Tech adoption: S-shaped curve:
Slow development and unpopular
Fast adoption and politically popular (lots of investment)
New tech firms have money use it to change competitive landscape
Lobbying for more favorable laws
Maturity, replacement by next tech
Thus, once decarbonization technologies take off, it will get encoded in public policy
Stringent climate goals: carbon removal deployed globally
The longer you wait to stop polluting, the more the carbon stocks rise
As such, if the policy goal is to keep warming below a certain temperature, carbon removal is required
Big Question: how do we get from today’s small carbon removal industry/research to the global multi-GT CO2 deployment we need by 2040-2050
Example: growing more row crops with deeper/more durable roots; better carbon storage (decades to centuries)
How quickly can this be put into practice?
Looking back on the adoption of GMO crops: change happens over 5-10 years (where legally allowed)
Suggests that this transition can remove a few GTons/year and is quickly deployable
Example: increasing ocean alkalinity
How quickly can this be deployed
Analogues?
Crushed stone production (sand, gravel, lime)
Aquaculture, desalinization
Took a few decades to deploy
Example: direct air capture
Analogues: chemical synthesis projects, fertilizer plants
These technologies have taken decades in the past
Recommendation for IAMs: make adoption of carbon removal slower and different adoption curves for each tech
Recommendation for decarbonization: focus on investor, developer incentives for executing deployment and profiting from it
Collaboration in emissions
There are many externalities: cutting emissions has a private cost and public benefit
Best-case scenario: global enforcement of agreements and high joint gains
In a more chaotic world: little global enforcement, while maintaining joint gains
Europe is developing decarbonization clubs
Can bring in China
US is outside of this process because it is currently unreliable
Capital is key: need to pay for the power grid to decarbonize; can only be done in lower risk/cost-of-capital scenarios
Roles in decarbonization for government
Direction/orchestration
Payment
Chiuna: govt does both
Europe: govt directs, private industry pays
US (old): govt pays, private directs
US (today): all private
IAM literature over-focuses on high-confidence results since they’re inter-comparable
E.g. IPCC has many more high-confidence statements than med-low confidence
This short-changes important results that are harder to model
Makes it hard to motivate work on resolving phenomena that have low confidence that are