Executive summary
The assessment of climate change and environmental impacts on risk in the financial industry is becoming increasingly dependent on complex models that challenge existing validation functions attuned to traditional credit risk approaches.
Models that assess climate-related risks or which generate other environmental, social and governance (ESG) data are being developed and recast to support strategy, provisioning, stress testing, ESG ratings, and disclosures in an environment where approaches, data sources, scientific research and regulatory oversight are in a state of flux. Regulators in many jurisdictions are requiring banks to assess both, the physical risks of climate change and the transition risks arising from the global shift towards a low-carbon economy.
As these models are becoming ‘mainstreamed’ into banks’ operations, model risk managers are faced with the challenge of validating methodologies and approaches that can be very unlike the credit risk models they are familiar with.
Traditional credit risk modelling methodologies assume a cyclical economy in which the past can be used to predict the future. Hence, validation approaches have relied heavily on back-testing against historical data to give confidence in future predictions. A suite of standard metrics (for e.g., Gini, Kolmogorov–Smirnov, R2) have been developed for quantifying this confidence.
But climate risk does not have suitable historical precedent to evaluate the impacts of global warming on credit risk metrics that can be easily used for model validation. So they do quite the opposite, i.e., extrapolate from present trends, apply assumption-based adjustments to risk metrics, and incorporate expert judgement where data is lacking.
Validation functions must, therefore, place greater emphasis on assessing the conceptual soundness of methodologies and assumptions of these models, apply sensitivity testing and scenario analysis to understand model behaviour, and evaluate whether the models are complete or miss any key drivers of risk.
In this paper, we provide an overview of the types of models encountered in financial climate risk modelling, the key challenges faced by financial institutions in validating them, and how CRISIL can support financial institutions in overcoming these challenges.
