When last we left insurance, in what seems like a lifetime ago, I had explained the concepts of moral hazard and adverse selection. The Angry Bear provides perhaps a shorter version of these things here. AB also brings up a related concept, "The Market for Lemons," which I probably won't address directly but it does remind me to make sure to include the issue of "market failure"explicitly.
Anyway, to continue...
III. Re-insurance
As I said in the previous post, the role of insurance companies is, for a fee, to remove (or reduce) risk from their customers. In the simple examples I provided, the companies themselves essentially bear no risk themselves. However, in practice for certain types of insurance this isn't actually the case. Obvious examples of when things go wrong for insurance companies are for things like hurricane insurance in Florida or earthquake (or fire) insurance in California. The problem in these instances is that "bad events" are correlated across customers. So, if a hurricane shows up it'll wipe out all of a company's policyholders simultaneously (ditto earthquake, etc...) . Smaller insurance companies likely don't have the geographic scope to diversify their risk portfolios. So, in practice, insurance companies also buy insurance on the reinsurance market. Roughly speaking, this is a way of limiting their own liability by purchasing policies from larger reinsurers. The local hurricane insurance company buys a policy which gives them a rather large deductible, and every claim made on them by their customers over and above that deductible is covered by the reinsurer. A well-known reinsurer is Lloyd's of London. The idea is that these companies have huge portfolios so that if they lose a bundle on a hurricane in Florida, it's no big deal because they have lots of California earthquake premium money coming in. As long as what should be uncorrelated bad events don't happen simultaneously, the reinsurers should be able to cover their liabilities.
This system in general should work, as long as companies have sufficient reinsurance and as long as the reinsurers aren't simultaneously hit with a bunch of hugely unlikely events at once (hurricane in Florida, floods in the Midwest, earthquake and fire in California, a plague of locusts in New England, etc...). However, it's quite possible that such catastrophic events are more likely than we tend to think they are. The problem is due to something called "fat-tailed distributions." The "normal distribution," which is simply one function describing the likelihood of any particular range of outcomes, assigns very tiny probabilities to extreme events happening. However, if those probabilities are in fact "not quite tiny," but somewhat bigger than tiny, then it becomes difficult to set an appropriate price for insurance. It's the difference between "the odds of that happening are so low we don't have to worry about it" and "the odds of having a 500 billion dollar claim are non-negligible, but we're not quite sure what those odds are."
Lloyd's of London itself ran into some problems in recent years due to the high cost of paying out asbestos claims. Lloyd's, along with many other reinsurers, are "unlimited liability" companies. They have shareholders - but unlike the shareholders of most companies, they are responsible for any and all of the company's liabilities. If you own stock in IBM, and it goes bankrupt, you'll lose your investment. If you own a piece of Lloyd's, they can take your bank account and your house too. The scandal with Lloyd's was that many newer owners claimed that the company had conned them into buying into the company, without revealing the extent of their liabilities. The spectacle of a bunch of upper class Brits finding themselves screwed by the Man was interesting.
...this rambled on longer than I intended.... health insurance tomorrow...
...Lloyd's is also known for its famous "inside out" headquarters in London - the plumbing and heating ducts are all on the outside.