Here’s a very simple model of health insurance: A bunch of people put put some money in a pot on January 1. If any of them gets sick that year, the health care bills incurred that year are paid with the money in the pot. The pot is the “insurance company.” Question: How much money should each of them put in the pot? OK, easier question: What would you need to know to figure out how much money each should put in the pot?
Hopefully you’re thinking that you’d need to know the probability of someone getting sick and the expected cost of health care if they do. Those two pieces of information would be enough to get started on this problem. But there is something else you’d need to pin down first.
Who are these people? That is, over what population would you need to know the probability of someone getting sick? Over what population would you need to know the expected cost of health care conditional on being sick?
Well, it’s the population of the people putting money in the pot. Duh! But, I ask again, who are these people? This is not a simple question. I’m asking now about the determinants of health insurance purchase. People select themselves into purchase. Nobody is forced to buy a policy, and certainly not this policy.
Roughly speaking, people buy insurance if the cost to do so — the amount of money they have to put in the pot — is less than the value they’d get by using the money in any other way. People buy insurance if it is a good enough deal to them.
It’s getting circular. We want to know how much to put in the pot. To figure it out, we have to know who else wants in too (so we can determine the probability of illness and expected cost over the population buying the insurance). But people want in based on how much they would have to pay. But that’s what we want to know!
This illustrates a fundamental property of health insurance. The cost of the product depends on the demand for it. In contrast, the cost of most goods and services are independent of demand. The cost of producing (but not the price to purchase!) another pair of shoes has nothing to do with who will buy them.
So, this is rather complicated. We know (or I’ll tell you) that the cost of insurance is the probability of illness multiplied by the expected cost of health care conditional on illness for the population insured. The price that you pay for insurance is closely related to cost, though the company charges a little more for all the other stuff it does plus profit. Let’s ignore all this other stuff, as I did in the original set up. We’re just putting money in a pot. We don’t know who is “we,” who are the people putting in the money. So, we’re stuck.
Since we can’t answer the questions I originally posed because we don’t know enough about who will buy insurance,* let’s try a simpler question: As the cost of insurance goes up, and hence its price (or the money one must put in the pot), what kind of person is more or less likely to buy? You can answer this.
Relatively more healthy people will not buy insurance as price rises. Since they are less likely to need insurance or the cost of care for them is lower if they do, at some point the price of insurance exceeds the value to them. Conversely, people more apt to get sick or need costly care are more likely to continue to purchase insurance even as the price rises.
Notice what is happening. As price rises, the cost per person likely to buy goes up (because the healthy stay out but less healthy do not). After all, why should someone pay for something they don’t expect to use? This is adverse selection at work. A typical person in the subpopulation of those who actually purchase insurance is less healthy and more costly to insure than the average person drawn from the total population. That’s what adverse selection means.
Don’t take “pay” too literally. Adverse selection occurs for public programs, like Medicaid, for which individuals pay very little or even nothing in premiums. “Pay” is a broader than that. It takes time and effort to enroll, to acquire the necessary information, to fill out the forms, and the like. Any barrier, any work an individual must do to enroll is a form of payment. Adverse selection results.
Fundamentally, where does adverse selection come from? From the fact that demand (who buys insurance) affects the cost of the product, hence its price. This is a feature of insurance that is not common to other goods and services.
* That is, I haven’t told you in this post how to figure out who will buy insurance. In the real world, we know a lot about who will buy insurance, but let’s not go there.
Liran Einav and Amy Finkelstein, “Selection in Insurance Markets: Theory and Empirics in Pictures,” Journal of Economic Perspectives 25(1):115–138, 2011.