Skip to content

Content Library

Back to Articles

Article May 13, 2025

Banking on a Mirage - Why Climate Risk Models May Be Failing Us

Banking on a Mirage - Why Climate Risk Models May Be Failing Us

Andreas Simou, Managing Director, Center for Financial Professionals

A leading model validation expert calls for a paradigm shift in financial risk modeling, warning that climate risks defy conventional assumptions and demand urgent, data-driven, and forward-looking integration into credit stress tests.
  • Integrating climate risks into credit risk stress testing requires a deep understanding of the assumptions and uncertainties behind climate model inputs to avoid misleading outcomes.
  • The mismatch between short-term financial stress testing horizons and long-term climate risk scenarios presents major challenges, particularly given the limited usefulness of historical data in projecting future climate trends.
  • Static assumptions in modeling, such as fixed carbon prices or unchanging balance sheets, fail to capture the dynamic nature of company behavior and should be replaced by adaptive, bottom-up approaches that account for real-world responses.
  • As regulatory expectations around climate risk evolve, institutions must align internal models with long-term climate objectives, treating ESG and climate considerations not as optional add-ons but as essential components of risk management frameworks.

display-middle-ads-here

At CeFPro’s recent climate stress testing conference in Amsterdam, a leading figure in model validation at a major European bank offered a compelling insight into how transition and physical risks can be integrated into current credit risk stress testing frameworks.

His central argument emphasized the importance of understanding the climate model inputs themselves before integrating them into financial risk models.

Without grasping the underlying assumptions, uncertainties, and dynamics embedded in these climate projections, he argued, any attempt to use them as stress testing inputs risks being superficial or even misleading.

He cited, in particular, the transition from NGFS Phase 3 to Phase 4 as a critical evolution that reflected a growing delay in global mitigation actions that have seen transition risks increase even in scenarios targeting similar long-term temperature outcomes.

One of the most difficult challenges in aligning climate science with credit risk modeling is the temporal mismatch.

Traditional stress testing horizons span three to five years, while climate risk scenarios extend 30 to 50 years into the future.

This drastically increases uncertainty, especially when historical data offers limited guidance for forward-looking climate risk modeling.

Climate variables are largely influenced by external radiative forcings – greenhouse gas emissions – rendering past observations inadequate as indicators of reliable future trends.

The speaker challenged the prevailing econometric mindset – which often assumes the maintenance of the status quo – arguing that such assumptions fall apart entirely in the context of climate modeling.

The presentation delved into the practical complexities of quantifying transition risk using tools like carbon pricing.

The carbon price in Europe, for example, has already seen large fluctuations, moving from €100 to €60 in a short period, and could potentially reach €1,000 according to some scenarios.

Companies will inevitably adapt, so assuming static balance sheets and ignoring second-round effects is a fundamentally flawed approach to meeting critical challenges.

The speaker instead urged those managing climate risk to take a dynamic, bottom-up approach to stress testing that incorporates adaptation behavior within their business models.

While the European Central Bank’s (ECB) top-down approaches remain influential, more granular techniques using sector-level Total Value-at-Risk (TVF) factors, derived from public databases such as EXIOBASE, offer a pathway to link emissions and economic activity.

By calculating the emissions intensity per sector and adjusting for gross value added, it then becomes possible to project stress impacts on credit risk parameters like Probability of Default (PD).

Delegates at the session were told that these projections can then be embedded in models such as the Merton structural framework, allowing for differentiated impacts even within the same sector.

The dispersion among firms in a sector becomes a critical driver of differentiated risk impacts, making ESG differentiation not just desirable but necessary.

Case studies and empirical data back this methodology.

One case study used in the session highlighted a retail portfolio in Southern Europe that was stress-tested using macroeconomic drivers tied to NGFS scenarios.

Events you may be interested in

View all of our upcoming events

Sign Up to Our Newsletter

Sign up

Connect Plus

Unlock more with Connect Plus membership

This content is available to Connect Plus members.