Why Do We Need Probabilities to Tackle Climate Scenario Ambiguities?
Climate scenarios have become a cornerstone in helping financial decision-makers navigate the uncertainties of climate risks.
But even with cutting-edge models from Oxford Economics and the NGFS, investors still lack a clear picture of their true exposure.
Here’s the critical gap: these scenarios don’t tell us how likely different outcomes are.
That’s a problem this study aims to solve.
The deliberate avoidance of assigning probabilities makes asset valuation and regulatory decisions less effective. It leads to poor risk management and weaker climate strategies.
Where Current Climate Scenario Frameworks Fall Short: A Probabilistic Lens
The EDHEC Climate Institute (ECI), a research centre well known for its innovative thinking elaborates on this gap in their study, “How to Assign Probabilities to Climate Scenarios”.
The study puts forward two key questions:
- What if we brought probabilities into these scenarios to make them more effective, while still building on the solid foundation they provide?
- What’s the best methodology to complement the existing models by adding probabilities?
The paper underlines that this limitation of current models makes it difficult for policymakers to make assessments based on clear guidance in relation to which risks are urgent or extreme.
Accordingly, investors rely on probability estimates to price assets based on expected future returns. Without those probabilities, the asset pricing efforts would not be effective and would be “stillborn”, in the authors’ words.
The dataset used for probability estimates in this study is initially a massive one, with 5,905 different estimates of Social Cost of Carbon (SCC) derived from 207 studies.
A Two-Pronged Approach to Climate Risk Clarity
The strategy of the study is to build a distribution for the SCC that reflects the information we have on the implemented carbon tax.
Here are the two complementary approaches:
• The informative (Elicitation-based) distribution: It integrates the views of economists and their estimates (not perfect but helpful) on the social cost of carbon.
With this expert-informed route, we use three key information:
- Fiscal, monetary and technological limits on the feasible pace of abatement.
- The distribution of economists’ views on optimal abatement policies.
- The observed gap between expert recommendations on how fast abatement should be and what policymakers have actually implemented.
Figure 1: Distribution of temperature increase in 2100 obtained with the informative approach (1)
• The ‘least-committal’ (maximum entropy) distribution: It makes the fewest arbitrary assumptions and uses only what we know, or the information we were given.
In this route, we do not consider the views of economists on abatement policies useful and simply use the information we have, even when it's limited.
For instance, imagine that the only information we have about a variable is its mean and variance. When you only put in these two pieces of information, then the normal distribution would be the resulting maximum-entropy distribution, in the sense that makes the fewest extra assumptions.
We can think of this route as the “least arbitrary” approach.
Figure 2: Distribution of temperature increase in 2100 obtained with the maximum entropy approach (1)
This structured way to embed probabilities into climate scenarios allows for actionable insights leading to effective risk management and investment.
In both approaches, the study’s probabilistic estimates are grounded in real-world data, such as the cost of traded carbon permits, adjusted for realistic political limits.
The Results: The Probabilistic Model Suggests a High Chance (35–40%) That Temperatures Will Exceed 3°C by 2100
These distributions converge on concerning findings about the severity of climate risks faced.
The study highlights that we are far from reaching some of our key climate pledges globally.
The results—strikingly similar across both approaches—suggest that:
- The chance of remaining under the 1.5°C target is very low:
The likelihood of limiting end-of-century temperature increases to 1.5◦C is very small. Whilst technologically still achievable, reaching this goal would require a dramatic and sudden alignment of abatement policies with expert views.
- The median global temperature rise could be around 2.7°C by 2100:
This average temperature anomaly is well above the end-of-century target of 2.0°C, and achieving this target is highly unlikely.
- There’s a high chance (35–40%) that temperatures will exceed 3°C by 2100:
This is a worrying result, as such high temperatures would push the planet into uncharted territory. This could increase the likelihood of tipping points and could lead to irreversible climate shifts.
The human species has never experienced such temperatures. Adaptation efforts could be severely compromised, potentially leading to significant physical damage.
The study also finds that the physical damage caused by higher temperatures may outweigh the economic costs of transition—a key insight for risk planning.
How to Attach Probabilities to Oxford Economics Scenarios
The sketched approach in this model offers an effective complementary look at the existing climate scenarios and how to make them more useful for financial decision-makers.
One of the most common and widely used climate scenario models is provided by Oxford Economics (OE). Its seven narrative-based scenarios describe pathways for a larger number of macro financial quantities (2023-2050).
But OE scenarios come without probabilities.
To fill that gap, the model presented above aims to assign probabilities to climate scenarios in a way consistent with the assumptions in the climate scenarios developed by OE.
Here are the key steps taken to achieve this goal:
- Calibrate the ECI model so that it behaves as closely as possible to the OE model.
- Generate expected paths for emissions, temperatures and GDP.
- Assign probabilities to OE scenarios that recover these expected paths as closely as possible.
As a result of this method, the study demonstrates a strong “polarisation” of probabilities.
Taken together, the two specific scenarios associated with the slowest pace of emissions account for more than 90% of the total probability. If we add the Baseline scenario, the probabilities add up to more than 95%.
At the opposite end of the spectrum are the very “optimistic” scenarios that have such high-speed abatements that this study shows very low probabilities with them.
Rethinking abatement policies: A damaging “gap” between expert advice and its real-world implementation
The results of a probabilistic exercise on Oxford Economics demonstrate the severity of the climate outcomes facing investors and policymakers.
It also highlights a damaging real-world pattern that has been present for almost 40 years: a persistent gap between climate abatement policy advice offered by economists and the actual implementation of their recommendations.
This policy gap could map directly onto the high-probability, high-temperature outcomes.
These outcomes and their associated risks would be challenging to avoid without urgent policy alignment and a serious shift in course.
The study argues that the recent developments in global politics, however, don’t paint an optimistic picture about whether this policy gap will be meaningfully closed.
The researchers suggest that incorporating climate probabilities can help build more effective adaptation strategies in a world full of uncertainty and political inertia.
The study’s novel approach goes beyond offering theoretical suggestions; it also provides practical tools to assess and manage climate outcomes based on not only possibilities, but also probabilities.
References
