A research group at University of Illinois makes population projections for US cities (suburbs? municipalities? this is a little unclear from the article) through 2100, and the top 40 hits are in the metro areas outside Phoenix, Dallas, Houston, Austin, Oklahoma City, Miami, Tampa, Orlando, Riverside (greater greater Los Angeles), Salt Lake City, Las Vegas, Denver, Boise, Fargo. These would seem to be heat, drought, flood, and fire prone areas, so this does not square with the idea that disaster-driven insurance rate increases will force mass population movements out of these areas.

Part of the answer to the insurance paradox is that political pressure causes states to set up “high risk pools” initially intended to assist small numbers of highly vulnerable homeowners, and the scope of these tends to creep up over time. This has particularly happened in Florida, although Florida has taken steps to move people out of their program recently. Another piece of the puzzle is that a big factor in private insurance rates is not disasters but credit scores, and this “mutes the market signals”. I tend to think that insurance companies, evil or at least amoral as they are, know what they are doing in terms of the math, and credit scores must be highly correlated with claims and losses. They also probably have no reason to take a long term view because they can drop policies any time they want as conditions worsen. Mortgage companies might have something to say about this, but remember that they are implicitly government subsidized for the most part.