People buy insurance against events they can picture and decline it against events they cannot.
That produces a consistent pattern: against small vivid losses and under-insurance against large abstract ones.
Extended on appliances are the clearest example of the first.
They are priced far above and sell in enormous volume.
Flood cover in an area that has not flooded recently is the clearest example of the second.
Purchase rises sharply after a flood and within about three years to the previous level.
The decay curve is remarkably similar across countries and hazards.
It follows memory rather than risk, which has not changed at all during those three years.
alters the decision more than price does within a substantial range.
An annual expressed as a daily amount sells better, and the same expressed as a percentage of the sum insured sells worse.
Regulators have used this asymmetry in both directions without saying so.
Requiring a percentage disclosure reduces the sale of poor-value products, which is a defensible use of a bias.
cover avoids the whole problem and creates a different one.
A pool that everybody joins is cheaper per person and transfers money from the safe to the exposed.
Where the exposure correlates with wealth, that transfer runs in an uncomfortable direction.
Subsidising flood cover for coastal property is towards the owners of coastal property.
Stating that plainly usually ends the policy, which is why it is stated rarely.
Insurance is the clearest case in economics where a technically correct answer is politically unsayable.