Setting the Bar for Climate Risk Scenario Analysis

This presentation explains the key flaws in scenarios put forth by regulations for scenario analysis and stress testing, and provides examples that demonstrate the value of the breadth of climate scenarios for analyzing risk.

Scenario analysis of one kind or another forms the backbone of regulatory guidance for climate risk disclosure and management. A complete scenario analysis would address the risk of economic damages from a transition to a green economy (transition risk), and the direct risks from changing environmental factors (physical risk). While flawed modeling has enabled incorrect interpretations of the ties between SSPs and physical outcomes, the same flawed modeling is informing scenario analysis guidelines and regulations. Transition and physical risks interact in ways that are not yet well modeled, and current scenario analysis practices are incomplete in that they do not consider both kinds of risk in a common framework.

An example is that some have argued that SSP5-8.5 should not be used, because SSP5 is not realistic. This conclusion is non-sequitur. It is not rational to reject GCM projections for SSP5-8.5 as a plausible physical outcome, based on SSP5, because the coupling between RCP 8.5 and SSP 5 is based on fundamentally flawed assumptions and modeling. Consistency between socioeconomic pathways and physical climate modeling scenarios cannot be established by the models used to do so. Risk analysis, disclosure, and management are better served by ignoring the supposed connections between SSPs and physical risk. An informative risk analysis would leverage the rich physical risk data informed by CMIP6 to construct scenarios that capture the range of plausible physical outcomes.

Many scenario analysis frameworks follow the recommendations of the Task Force for Climate-Related Disclosure (TCFD). While the TCFD recommends using multiple scenarios, it is non-specific on what scenarios (i.e., combinations of physical risk and transition risk pathways) are most insightful. That leaves scenario definition in the hands of regulatory organizations or non-governmental organizations (NGOs) that add specificity to the analysis, such as the Network for Greening of the Financial System (NGFS).

Scenario analysis remains the most viable path forward because of the deep uncertainty inherent to assessing and managing climate change risks. And it’s becoming more clear that physical risks will cost us more than transition risks (e.g. NGFS Scenarios, 2022). Given that we face real threats and need to have some rational response to assessing climate risk, which requires new scenarios, a practical question is then: do we have the information we need to define and perform useful scenario analysis for climate risk?

The answer to that depends on how we choose the physical and transition risk models and pathways, and whether the data needed to inform those choices exist. Data to inform a range of useful possible and plausible physical climate pathways does exist today. This is possible because of the breadth of climate modeling executed against the IPCC scenarios, irrespective of the socioeconomic pathways that have been coupled with the physical pathways.

By breaking the assumption that we can robustly link transition and physical risk, we are free to design scenarios meant to stress test each risk separately, while continuing to develop storylines or modeling that can build plausible links between the two. Focusing on physical risk, we can think about approaches that can support superior risk analysis with data that we already have. That avoids both the nonviability of combined transition and physical risk, and fundamental challenges associated with accumulating data needed for robust transition risk. And in many cases it enables assessing the risks associated with physical tipping points.