Because they flood all of the time. Ergo, private insurers are smart enough to realize that these are sucker's bets. And private insurers don't take sucker's bets. But the government (in the name of compassion) is more than happy to. Well for 40 years the NFIP has been insuring these homes that shouldn't be insured, and rebuilding them, and rebuilding them again, and sometimes rebuilding them some more. I had a friend in Houston whose parents lived in a flood prone plain. Their home was flooded (as in you are sloshing through water on the first floor) at least three times over a 10-15 year period.
For those who may live in the desert, there is a reason why we don't live in the ocean and have more experience exploring the moon than the seas. Water tends to have the nasty habit of destroying things. Even "waterproof" things. Floods have the nasty habit of bringing water (and whatever else is in the vicinity - mud, oil, gasoline, algae, mold spores, etc.) into places we don't want it. All that stuff tends to destroy things faster than water does. Here's an experiment for you. Take a wet washcloth (not dripping wet, just more than damp) and leave it on your 14 step coated (including waterproof polyurethane top coat) dining room table (or chair, or china hutch, etc.) overnight. Be sure to have your Mom or Wife or Significant Other's permission before you do it. Now imagine that the table is immersed in water, for three days, and there happens to be some nice solvents mixed in as well.
My wife and I bought one rental property that also happened to be in the a flood plain. To be more precise, the very back corner of the lot (which dipped down towards the gulley behind it) was in the flood plain. The building itself was at least 50 feet away from the flood plain. The place had never flooded (and didn't have a basement to flood first either) in the thirty years it had been standing. Two houses on one side of it did not have any part of their property in the flood plain. Yet, because of the layout of the building and the street, both of them were actually closer to the boundaries of the flood plain than my structure was. Yet, I had to insure my property with flood insurance.
Flood insurance cost as much (if not a little more) than property insurance. I am sure that the private insurer loved my property since there was little to no chance of it ever flooding. Yet the premium I paid was the same as a building that sat inside the flood plain. Insurance companies make their money based on risk. They try to figure out how much risk there is to action A, and then charge you X amount of dollars to cover that risk (based on the amount of people they insure that are going to engage in action A). They are a great barometer for things like safety since unlike a government dictate that says action B is safe, the insurance company has to put its money where its mouth is.
So, they will give you discounts for wearing a seatbelt (a safe action), but don't give a rat's rear end about how many guns you have in your home (something that is irrelevant to safety). If however, you do want to insure your guns against theft or damage beyond what your home owners insurance will cover (or your jewlery, or rare baseball card collection), then the insurance company may want to know how it is stored (i.e. in the closet = higher premium, case hardened multi bolt safe = lower premium)
For human induced risks (automobile accidents, home owners insurance), they have a lot of data and aren't worried about losing money on the deal. There is not some huge rash of car accidents that is going to bankrupt some company. For environmental related risks (huricanes, flooding, forest fires, etc.) they are more leary because the risk is less predictable. Hence, premiums are higher, policies are more restrictive and in some cases they just won't insure you (hence the NFIP). For crackpot induced risks (war and terrorism), they don't insure at all. Not because the risk is high (it is actually miniscule), however the localized costs are unpredictable and range from 0 to ginormous. Its a lot easier to predict how much damage a Category 5 hurricane will do to a major metropolitan area than what the next terrorist attack will be.
NFIP on the other hand, really doesn't care about the costs. They probably charge something similar to what a private company would charge for a less risky place. If they charged what the market rate was (i.e. what would be needed so that the NFIP was not underwater), then no one would be able to afford to live someplace like that. This is not rocket science, its risk analysis. Every successful business (insurance or otherwise) has figure this out long ago. Why can't our government (frankly, I don't care why not, I don't want the government in that business anyway).
Profit is an excellent incentive to determine the cost of risk. The insurance industry has been doing it successfully for centuries. Government hasn't done it successfully yet.
Yep, and then we won't have to pay to rebuild them over and over and over again. We sold that property after a year of holding it. There was no positive cashflow, the flood insurance basically ate it all up.