The new business of climate risk assessment: who are the emerging actors and how reliable are their products?

To enhance corporate due diligence and accountability for climate change, in recent years the EU has adopted interconnected legislative instruments such as the Corporate Sustainability Reporting Directive (CSRD) and the EU Taxonomy Regulation. Under this framework, corporations must disclose environmental, social, and governance performance, assessing both their impacts on climate change, water, and pollution and the risks these issues pose to their operations. These reporting needs have created demand for technical skills in climate science and related disciplines. In response, new business activities have emerged: insurance companies have opened climate related branches to support scientific climate risk assessments, and many other companies and startups now provide similar services.related branches to support scientific climate risk assessments, and many other companies and startups now provide similar services. 

The value chain used by these companies typically begins with future climate projections from CMIP6 models and proceeds through the computation of hazard metrics to the final estimation of risk in terms of damages, losses, or productivity impacts. Each step introduces its own layer of uncertainty, which cumulatively affects the final results. Consequently, different methodological choices can lead to widely diverging conclusions, from low to very high risk, for the same type of asset. Moreover, to produce results at scale, many private climate risk providers apply standardized workflows that do not adapt to the distinct physical characteristics of different hazards. This lack of hazard specific treatment can further amplify uncertainty and reduce scientific robustness.specific treatment can further amplify uncertainty and reduce scientific robustness. 

In this work, we present a comparative analysis of several commercial climate risk assessment solutions, benchmarked against our in-house methodology. We quantify how different methodological choices affect the overall uncertainty of the risk estimate for the same asset. Our results show that the selection of the climate model ensemble, in both size and composition, is the first major source of divergence. The ensembles we evaluated range from 5 to 20 models. In some cases, models are selected based on scientifically robust criteria such as models’ interdependency and equilibrium climate sensitivity. In others, the choice is driven only by the availability of specific climate variables. This latter approach can bias the ensemble toward hotter or colder models and may fail to capture a sufficiently broad range of plausible futures. A second major source of uncertainty arises from the use of fixed thresholds for computing hazard metrics. These thresholds can yield more or less conservative results, meaning that two companies may classify the same asset as either high-risk or low-risk for the very same hazard.house methodology. We quantify how different methodological choicesrisk or lowrisk for the very same hazard. 

Overall, our findings highlight the need for a scientifically grounded framework that brings standardization, comparability, and reliability to the entire climate risk assessment chain, ensuring that corporate reporting under EU regulations does not depend excessively on the particular provider selected.