Climate scenario analysis has become an important tool to help pension funds understand the financial impact of climate on their investments. Climate regulations often require financial analysis across a range of scenarios from rapid decarbonisation pathways to hot-house worlds of expanding global emissions. But these frameworks are not attuned to the scenarios that are most likely to occur.
What is the value in understanding how a net-zero world impacts investments if that scenario has near-zero chance of occurring? We need to better quantify the likelihood of each climate scenario playing out. Reducing the spread of climate scenarios into a more likely bandwidth would allow pension funds to move from theoretical exercises to quantifying the most likely impact of climate on their investments.
Our team at Emmi have used 25 years of global carbon emissions data from the Global Carbon Budget, led by the University of Exeter, to forecast future carbon emissions over the coming decades. Since climate warming is linearly related to cumulative carbon emissions, knowing where future carbon emissions are heading means you can also understand the most likely warming scenario.
Our results show the world is most likely tracking towards 2.3°C of warming by 2100, with a likely range of 1.7°C to 2.9°C, from a 28% reduction in annual fossil carbon emissions by 2050. Nothing in the method depends on assumptions about future policy, economic growth, population or technology. Historical carbon emissions growth rates are a very good baseline predictor of future emissions.
So what does this approach mean for the likelihood of the existing suite of climate scenarios from the Intergovernmental Panel on Climate Change (IPCC) and the Network for Greening the Financial System (NGFS)? The two traditional scenario anchors of institutional climate analysis (net zero by 2050 and hot-house worlds) have a 0.2% chance of occurring. The most likely scenarios concentrate heavily around the NGFS Fragmented World scenario, which carries 70.6% of the weight on its own, more than the other eighteen scenarios combined. The NGFS Current Policies scenario carries 18.2%, and NGFS Nationally Determined Contributions carries 7.1%. The probability-weighted blend of these scenarios is what produces the 2.3°C most likely pathway.
Others have argued that climate scenarios need to be probability-aware. Carbon data now allows us to quantify those probabilities. This is where the regulatory question becomes hard to avoid. Prudential supervision of pension funds exists to protect member outcomes. Member outcomes are determined by the futures that occur, weighted by their likelihood, not by the boundary cases. That means a prudential climate architecture built for pension member protection is spending a large part of its analytical and organisational effort on climate futures the data suggests have almost no chance of occurring.
A pension fund trustee reading a compliant climate disclosure today learns how the fund fares in a 1.5°C world or hot-house world, both with near 0% chance of occurring, and learns nothing about the most likely scenario. It seems we have defaulted to being more precise about the improbable scenarios, while the risks for the most likely scenarios are hidden.
I am not arguing for removing the scenario tails. Boundary testing is what disclosure was designed for, and a prescribed, common scenario set keeps disclosures comparable across funds. The argument is that the framework is incomplete in a specific, practical way. Alongside the prescribed tails, funds could be asked to focus on their position under an expected central case anchored to observed emissions, refreshed as the data updates, with the likely range of 1.7°C to 2.9°C as the relevant band of futures over which prudence should operate. The tails help quantify the climate risks at the boundary. But the most likely scenarios are where most of the work needs to be focused, allowing trustees to understand whether, and how, member capital is exposed to the world as it is actually unfolding.
Climate disclosure regimes prescribe the tails and leave the baseline to each fund’s discretion. In practice, that often means no baseline at all. Banking settled this question long ago: capital stress testing pairs every tail scenario with a baseline macroeconomic projection, and requires both. Borrowing that convention would not dilute climate stress testing; it would anchor it to the most plausible view of what happens anyway, which is the only way to know what the tail scenarios actually cost.
A 2.3°C warming most likely presents a unique challenge, as it entails both significant transition and physical risks for investors. The climate system adds its own uncertainty in both directions: sensitivity and carbon-cycle feedbacks could carry the same emissions pathway above or below that range. But prudence has always meant calibrating to evidence rather than to aspiration in one direction or alarm in the other. The evidence now identifies a most likely climate future with tighter precision. A prudential framework that ignores it is testing everything except the world its members are most likely to live in.
Ben McNeil is a climate scientist and co-founder of climate risk data provider Emmi. He was previously a researcher at Princeton University and UNSW Sydney, and has served as an expert reviewer for the IPCC.
