The risk of global weather connections: Are atmospheric hazards independent?

Major atmospheric-driven catastrophes, such as hurricanes and floods, may appear to be independent events when looked at historically. Yet it is well established in climate science that regional weather and climate conditions in one part of the world can have impacts on other parts. It is important for insurers and reinsurers to know how interconnected climate drivers are as they are required by regulators to hold a level of capital that adequately reflects their exposure to losses from significant weather events.

To establish whether the assumption of independence in insurers’ internal models is appropriate, this research investigates the extent of the links between different global extreme weather perils and the mechanisms for these dependencies. This report answers the question: is it reasonable to assume independence between significant weather risks in certain region-perils around the world? Lloyd’s and the Met Office have disclosed the methodology in full for general review purposes and to encourage debate.

Key findings

• For extreme weather events an assumption of independence is appropriate.
• A number of regions show some correlation between climate drivers but these are not considered to be substantial enough to warrant a change in capital.
• Even when there are high correlation levels between weather events, it does not necessarily follow that there will be large insurance losses.
• Weather events can still occur simultaneously even if there is no link between them.

The results of the modelling presented in this study demonstrate that an assumption of region-peril independence is currently appropriate for use in modelling extreme natural catastrophe risks. This important finding supports the broader argument that the global reinsurance industry’s practice of pooling risks in multiple regions is capital efficient and that modelling appropriate region perils as independent is reasonable.