A scientific approach to sovereign climate risk assessment
In a webinar hosted by Risk.net, panellists from Scientific Climate Ratings, EDHEC Climate Institute and Amundi discussed the challenges of calculating sovereign climate risk and the steps taken in this direction by Scientific Climate Ratings
The panel
Thibaud Barnaud, Climate risk consulting manager, Scientific Climate Ratings
Nicolas Schneider, Senior research engineer and macroeconomist, EDHEC Climate Institute
Aaron McDougall, Head of climate, Amundi
Moderator: Stella Farrington, Commercial editor, Risk.net
Understanding sovereign climate risk is hugely important but enormously challenging for many organisations. While great strides have been made in recent years when calculating the financial impact of climate risk on corporate and financial firms, far less progress has been made in assessing sovereign climate risk.
EDHEC Climate Institute and Scientific Climate Ratings set out to address this gap, conducting robust scientific research into how physical climate risk will impact the ability of sovereigns to service their debt. The resulting Sovereign Climate Ratings product is designed “to bring early information to investors on the structural vulnerabilities of macroeconomies before these are priced into the spread”, said Nicolas Schneider, senior research engineer and macroeconomist at EDHEC Climate Institute, during the recent Risk.net webinar.
The product, which focuses on chronic physical climate risk, is designed to be integrated into analysts’ broader risk models. “The idea is to provide everyone with a clear, understandable metric that is quantified and that can be ingested into your own risk modelling,” said Thibaud Barnaud, climate risk manager at Scientific Climate Ratings.
While acute physical risks are hard to predict and often make headlines because of the destruction they cause, chronic physical risks are a continuous effect of climate change, compounding cumulatively over time. They tend to yield economic damages on the level of gross regional product (GRP) per capita, which is why the tool takes a granular, bottom-up approach, Schneider explained.
The research covers 191 countries and more than 95% of total world economic production at a granularity of 3,400 regions. It focuses on two time horizons, 2035 and 2050, but also provides annual granularity from 2026 to 2060, said Barnaud. The seven scenarios of the Network for Greening the Financial System (NGFS) are used, complemented by two more extreme scenarios formulated by the EDHEC Climate Institute.
Schneider pointed out that projected temperature increases remain relatively tightly clustered across the nine scenarios until 2035, before diverging much more markedly thereafter. The ratings product translates these temperature increases into economic damages, assigning sovereigns a rating of A to G, with G-rated sovereigns facing structural losses by mid-century.
The methodology underpinning the product is in four parts.
The first stage is historical and entails estimating a climate-econometric model using regional data spanning the last five decades to identify the relationship between climate conditions and growth in GRP per capita.
“That allows us to … capture information over time for years but also, most importantly, over space, meaning over different climate zones, income growth, sectoral specialisation and different idiosyncratic characteristics of these economies,” said Schneider.
This establishes a robust relationship between climate and economy, he stressed.
The second stage moves into projections and uses Nasa’s downscaled climate data across Shared Socioeconomic Pathway (SSP) and Representative Concentration Pathway (RCP) combinations.
“We process an equivalent of 180 billion data points to be able to grasp all the information we then feed into our model to get projected damages across SSP/RCP scenarios,” said Schneider.
However, to produce outputs that could be readily integrated into sovereign-risk analysis, the methodology maps the results onto the NGFS scenarios in stage three.
“We end up with a ‘reprojection’ in stage four, such that you have damages distributed over 3,400 regions,” Schneider said. “And we are able to weight our end-point curves with scenario probabilities, allowing us to compute the ultimate, most fundamental metric of impact, which is the unconditional expectation.”
The US provides an interesting example of the research findings. By 2035 it shows close to a 5% loss, rising to 10.4% by 2050. “We think the US faces a double-digit GDP loss by 2050, and this should be ingested into the sovereign analyst model,” Schneider remarked.
He added that, because of the country’s different climate zones, expected damages from future physical risk are likely to be heterogeneous, with regions closer to the equator likely to face more significant losses than those further north.
“Our belief at the EDHEC Climate Institute and Scientific Climate Ratings is that climate change is a bottom-up realisation, which means we have to start from the region-specific stories to compute a national metric,” said Schneider. “The best way to do that is to start from state-level damages using state-level historical data on GRP and then infer the aggregate effect using density weight, whether population or GDP.”
A focused approach
While there is a huge amount of uncertainty out to 2050, with events such as global population movement and the process of climate adaptation being largely unpredictable, the model does not try to take this uncertainty into account – instead it sticks to what can be known under each scenario, said Schneider. Similarly, the tool is not intended to capture a wide range of risks such as transition risk or political effects, but rather to focus solely on physical climate risk, providing a way to include that risk within a broader model.
“The fundamental goal of this tool is to focus explicitly on physical risk, and particularly on the chronic component of physical risk, which we think is the elephant in the room,” added Schneider.
Aaron McDougall, head of climate at Amundi, Europe’s largest asset manager, welcomed the tool’s clear and transparent approach. He said that climate risk models can become less useful when they incorporate too many variables and assumptions, creating the kind of “black box” outcome that investors want to avoid. In his view, the real value lies in highlighting potential exposure to focused risks, which can then be integrated with broader investment frameworks and capital market assumptions.
McDougall noted that, as a major investor across sovereign bonds and a wide range of financial instruments globally, Amundi requires assessments of physical climate risk that are clear, comparable and reliable. He stressed that these assessments should be considered alongside, not in isolation from, the broader economic context.
He also underlined the relevance of the Sovereign Climate Ratings tool for sovereign engagement, saying it can provide investors with objective, science-based rationale to support dialogue with sovereign issuers and “…to demonstrate why we’re asking about this.”
