In the United States, the individual states and local authorities are supposed to be in charge of disaster prevention and protection, while the federal government provides insurance against disaster occurrence. And as this insurance is fully with perfect commitment, prevention is sub-optimal.
Timothy Goodspeed and Andrew Haughwout look at the second-best insurance contract where the states provide prevention in a non-cooperative fashion and cannot be monitored. It turns out it is really difficult to coax states to provide appropriate prevention. Fundamentally, this is because of the time consistency problem of the federal government and its soft budget constraint. Once disaster has happened, it is difficult to say no to immediate aid. This is not unlike the problem of humanitarian aid, that keeps going to the same places because they provide no effort for prevention as they know they are insured. Or the banking system.
As long as the insurer wants good things for people, this time-consistency problem will remain. But it can be mellowed by applying a hard budget constraint. This takes strong commitment, which is unlikely in a democracy, as politician cannot afford to refuse emergency aid. Then, why not let the federal government be in charge of prevention investment as well?