Despite the growing reliance on detailed climate projections for assessing property-level risks, such as flooding, wind, and wildfires, there is increasing skepticism about the reliability of these models. Many insurers and investors use these forecasts to guide decisions, but recent findings reveal significant inconsistencies between different climate risk models, raising questions about their accuracy and transparency.
Recent analysis by Bloomberg Green highlights that risk models often produce conflicting assessments. For instance, a comparison of flood risk models for properties in Los Angeles County showed only a 20% agreement in vulnerability scores between a private model and an open-source academic alternative.
The climate nonprofit CarbonPlan also reported substantial discrepancies among private risk models. In their review, which involved two models, Jupiter Intelligence and XDI Pty Ltd, CarbonPlan found that the models agreed on only 12% of locations with increased fire risk over the coming decades. For coastal flood data in New York City, the agreement was even lower, with only 21% of locations showing consistent risk assessments between the models.
The differences arise from variations in proprietary methodologies and data inputs used by different firms. Despite requests for data from nine climate-analytics companies, only two provided sufficient information for meaningful comparison. This lack of transparency complicates efforts to evaluate the reliability of these models and limits their usefulness for end-users.
XDI co-founder Karl Mallon and Jupiter co-founder Josh Hacker both acknowledge the importance of model transparency but highlight the challenges of sharing detailed methodologies due to business interests. While both companies strive for accuracy, the broad differences between models suggest that the risk assessments may vary significantly for individual properties.
CarbonPlan’s findings stress the need for greater transparency in climate-risk modeling. The organization argues that without consistent practices for comparing and validating these models, the industry may rely on opaque “black-box” approaches, undermining the effectiveness of risk management strategies.
In summary, as climate adaptation becomes increasingly crucial, the reliability of private risk models remains uncertain. The call for more transparent and comparable analyses reflects a broader need for accuracy in predicting climate-related risks that could affect billions of lives and trillions of dollars.
