Predicting things like that is exactly the job of insurers and reinsurers. They're very good at it and it actually becomes easier at scale.
How many houses will be made uninhabitable by subsidence in the US in the next 5 years? I predict more than 100 and fewer than one million. There, that's a prediction. A team of actuaries armed with historic data could massively improve it.
This is much easier than the same kind of prediction for earthquakes or hurricanes, which have tail risk of unlikely but devastating events. If insurers don't want to cover this kind of risk, it's either because it's too small to be worth spending time on, or so predictable that it's not worth spreading the risk (better to just condemn the handful of houses built on top of known mines and move on).
How many houses will be made uninhabitable by subsidence in the US in the next 5 years? I predict more than 100 and fewer than one million. There, that's a prediction. A team of actuaries armed with historic data could massively improve it.
This is much easier than the same kind of prediction for earthquakes or hurricanes, which have tail risk of unlikely but devastating events. If insurers don't want to cover this kind of risk, it's either because it's too small to be worth spending time on, or so predictable that it's not worth spreading the risk (better to just condemn the handful of houses built on top of known mines and move on).