When does it make sense to use an insurance model, and when is an insurance model less useful? After all, it clearly doesn’t make sense for most of our transactions to be done through an insurance model. We don’t make our food choices using grocery insurance, nor do we buy movie or concert tickets with an entertainment insurance policy.

An initial answer one might give is that we should use insurance to cover highly expensive transactions. But this isn’t quite right. Something being expensive does not, in and of itself, make an insurance arrangement sensible. This is because many things which are expensive are also reasonably predictable. An insurance arrangement makes sense for events which are high cost, low probability, and not predictable at the individual level. To give it a pat description, insurance make sense for events that, individually, are a matter of if, not when, but at the population level, are a matter of when, not if.

Consider homeowners insurance as an example. There are a lot of expenses that come with owning a home, but not all of them are typically covered with insurance. Depending on construction materials, most homes need to have their roof replaced every fifteen to twenty years or so. Replacing a roof is very expensive, but most homeowner insurance policies don’t cover this kind of repair. That’s because needing to replace the roof of your house isn’t a matter of if, it’s just a matter of when. These kinds of easily anticipated expenses are best met through savings, not insurance. This is in contrast to, say, a house burning down. Whether or not your house burns down isn’t a matter of when, it’s a matter of if. Statistically, your house probably won’t burn down. But from the point of view of the insurance company, covering many tens of thousands of houses or more, that somebody’s house will burn down isn’t a matter of if, it’s just a matter of when. This difference in predictability, rather than total expense, is what explains when it makes sense to use insurance or not. And it’s also why a relatively small house fire that causes $5,000 in damages will be covered by insurance, but a roof replacement costing $20,000 typically will not.

Thinking about it in these terms, we can see places where comprehensive homeowners insurance doesn’t make sense – places where a home suffering major damage isn’t a matter of if, it’s a matter of when. There are areas in the United States where particular kinds of natural disasters are all but guaranteed to occur on a frequent basis – hurricanes, or major flooding. If you live on the Atlantic coast in Florida, your home getting hit by a hurricane isn’t really a matter of if, it’s just a matter of when. In a free market, we’d expect to see fairly little insurance for this, and instead people who choose to buy homes in areas with such well known and easily predictable risks would also be responsible for the repairs their homes will regularly and predictably need. Instead, areas with high risk levels see the insurance markets heavily distorted through subsidies that making living in high-risk areas artificially cheap by compelling others to foot a significant portion of the bill for the risks the residents choose to take. This creates a situation where the market is sending a clear signal saying “Hey, maybe don’t build lots of expensive infrastructure here” and legislators respond by saying “Counterpoint – maybe do build lots of expensive infrastructure there, and then when the inevitable and easily predictable results come about, you can just pass off the majority of the costs to your fellow citizens!”

Health insurance, as it currently exists, also makes little sense when viewed through this lens. In a free market, we would expect health insurance to cover treatment for conditions that aren’t a matter of when, but a matter of if. It would be used for cases like car accidents, where you suddenly and unexpectedly have large injuries needing immediate care, or for someone who develops cancer or needs an organ transplant. But it wouldn’t (and doesn’t) make sense for ordinary, predictable health maintenance and routine care to be provided though an insurance model. Some people might think we need health insurance for this because even routine maintenance can be so expensive, but I would argue that causation goes the other way – routine, commonplace health maintenance is so expensive because so much of it is covered through an insurance model.

Imagine if, due to subsidies and legal regulations, even the most routine car maintenance had to be covered by auto insurance. Need an oil change, or some new snow tires for the winter? Imagine having to check around to see which auto shop is in or out of network for your insurance company, trying to determine which sets of tires your policy covers and to what degree, working out what your deductible will be for the tire installation, and never even bothering to ask about the full price of the process, because you don’t ever pay the full price. And suppose you met your out-of-pocket maximum for auto insurance for the year! You’d have every reason to take your car into the shop as many times as you could for the rest of the year. This is one reason why health care reform that’s dedicated to making health insurance even more widespread and even more comprehensive is all but certain to make the problems with our health care system even worse.

Insurance can be a great tool, but it’s just a means to an end, not an end in itself. Making insurance more widespread and more comprehensive is not automatically a good thing – but as a soundbite, it’s very politically popular, and unfortunately that tends to be what carries the day more often than not.