Model Governance in Climate Risk: Meeting Banking Requirements for Capital and Risk

Integrating climate risk into banking operations represents a fundamental challenge for model governance frameworks. As banks face increasing pressure to incorporate climate considerations into their risk assessments and capital allocations, they must ensure that their modelling approaches meet rigorous governance standards.

?Since regulators are expanding existing prudential frameworks to address climate-related risks, model governance is urgently needed when these models inform capital requirements or lending criteria. Banks face a complex web of requirements when implementing climate risk models, particularly for scenarios like assessing flood risk in mortgage portfolios, which can lead to adjusting lending criteria.

These requirements stem from regulatory obligations and the practical need to make sound risk-based decisions. For example, vendor flood scores to pre-screen UK mortgage applications are questionable from model appropriateness, leading to profound societal implications.

?At the heart of climate risk modelling lies the "model stack problem." Modern climate risk assessment typically involves the output of one model becoming the input of another. The complex dependencies will amplify uncertainties and potential errors throughout the modelling chain.

When fundamental physical processes like cloud formation and precipitation patterns directly influencing flooding and other physical risks are simplified, the resulting risk assessments become questionable unless there is an appreciation of what the model cannot do; otherwise, material financial decisions like asset allocation are built on sandy model foundations.

?The recent revision of GDP impact projections by NGFS, driven by the introduction of precipitation modelling, is a stark reminder of how sensitive these model chains can be to underlying assumptions. Given the popularity of NGFS with disclosure teams in financial firms, the propensity to take the results without question will lead to material capital increases without the appropriate scrutiny. Core climate models often simplify the representation of clouds due to computational limitations and incomplete understanding of cloud physics.

This cascading of models creates a critical vulnerability in risk assessment. By rigorously understanding and documenting each layer's assumptions and transformations, banks can build their climate risk decisions on stable foundations. For pragmatic and accurate risk assessment and capital allocation, banks must extend beyond traditional model governance requirements to address the unique challenges of climate risk modelling.

For climate-related models to be acceptable in the current Model Governance framework, they will need to meet the following transparency:-

·?????? Underlying assumptions and construction of climate variable distributions (e.g. tails of the distribution of precipitation at a catchment area)

·?????? Data source lineage (public and vendor), measurement techniques (manual, automatic)

·??????Data curation, quality checks and frequency of updates

·?????? Input data derived by vendor models

·?????? Aggregation process of portfolios of assets by location

·?????? Use of climate scenarios such as RCP 8.5 or migration pathways

·?????? Description of known model weaknesses, uncertainty of output and limitations of appropriate use

?The crucial stage is for practitioners to understand model outputs - a simple risk score, a specific recommendation, or a stress scenario - and their intended use. Banks must document acceptable use cases and how consumers should interpret the outputs to incorporate within their risk-based process.

Most importantly, banks must ensure that end users of model outputs - risk managers, loan officers, or senior decision-makers - fully understand the model's limitations and appropriate use cases. This understanding should extend to both the technical limitations of the models and the practical constraints on their utility.

?The path forward requires immediate, collective action to address these critical model governance challenges. While the task is complex, the financial industry has repeatedly demonstrated its ability to solve everyday challenges through collaboration and standardisation. As climate risk becomes increasingly central to financial decision-making, the industry's model governance teams must catch up and expand their frameworks.

?The good news is that promising early-stage initiatives through industry bodies are laying the groundwork for solutions. Organisations like FINOS, OS-Climate, and ISDA are leading crucial work in developing open-source frameworks and standardising climate risk assessment methodologies. This collaborative approach offers several immediate advantages. Open-source frameworks enhance transparency and enable peer review of methodologies.

?Industry-wide collaboration allows institutions to share the burden of solving these complex challenges while ensuring consistent approaches to risk management. A collaborative approach to model governance will allow finance to:-

·?????? Move to scenario-based analysis that better captures uncertainties

·?????? Develop standardised approaches that acknowledge model limitations and weaknesses

·?????? Create transparent frameworks for documenting and validating model assumptions

·?????? Establish appropriate use of climate risk models in capital and risk decision-making

?By working together through established industry bodies, sharing knowledge, and committing to open standards, we can create governance frameworks that are both practical and robust. The topic of climate risk pushes all participants to move beyond regulatory compliance - it's about building a resilient financial system that can effectively manage climate risks while enabling the transition to a sustainable economy.