Pick some basket of relevant countries/regions, then gather data on property prices and insurance rates. Have two sets of properties for each region: exposed coastal areas and 'safe' properties inland at higher elevations, perhaps sheltered from storms. The safe properties act as a control for demographics, overall economic activity and local laws/taxes (e.g. in the same US state).
Generate an index as some function over the difference of prices and rates between the two sets. The index would have regional components and some aggregate total.
Data could also be collected on actual damage and claims. These affect later rates, so they have a lagged impact on the index anyway, but it might be useful to isolate multipliers or confounding effects. There could be scaled impact values based on the total exposed property value in the region, but there would be significant non-linearities.
Of course, insurance companies have this data, and surely have internal models that incorporate local and regional indexes for risk. But it would still be useful to have public data, an agreed index methodology, and a wider geographical extent than a single company patch or country market.
Property is notoriously idiosyncratic and illiquid, so it would be hard to trade the index, or use it for hedging risk, but perhaps some prediction market could be built around the index.
toomuchtodo•7mo ago