Friday, January 6, 2017

What we stand to lose: Basics of health insurance

On Wednesday I was a guest in the Rural Health class taught by the nursing program at Hartwick College. For many residents of rural areas, social and economic factors are an important aspect of their health situation, along with access to health care. In that context, the professors for the course invited me in to speak about health insurance.

My goal was to make health insurance more comprehensible by setting it in the context of other types of insurance, draw general principles from those, then see what happens when you apply those principles to health coverage.

I expect this to be two posts: one on the insurance background, the other on issues specific to health insurance. We'll see how it goes.

Start with the basic idea of insurance.

You as an individual are risky. Let's say we're talking about whether you'll get hit by a car. And to keep the exercise simple, let's say that getting hit by a car brings with it $20,000 in medical expenses (along with a lot of pain). And let's say there's a 3-in-100,000 chance of you being hit this year.

So there are two possible outcomes. You almost certainly won't be hit, in which case the cost is $0. But there's a 0.003% chance that you will be hit, in which case there's a cost of $20,000.

So, almost certainly $0, but just maybe $20,000. Nothing in between.

Now take a group of 100,000 people, all with that same 0.003% chance of a $20,000 accident. You can't expect there to be exactly 3 accidents. In fact, the chance of that is only 22.4%. But there's a 99.6% chance that the number of accidents will be between 0 and 8. Which means the average cost for the group will almost certainly be between $0 and $1.6.

That's a much easier situation to plan for than your individual situation where the only possible outcomes are $0 and $20,000.

This is the magic of what's known as risk pooling.

When you pay an insurance company, you might think of it as you paying them to bear the financial risk on your behalf: if you get hurt, you won't have to pay - they will. But in some sense, when they put together a pool of people to insure, in some sense they're not bearing risk for you. They're making it go away. That's the magic part.
The insurer charges $1.60 per person, and they will almost certainly have enough money to pay for the accidents that actually happen.

(Actually, they can cut it significantly closer, because an insurance company can itself buy insurance in case it has to pay out more than it expected to.)

There's something else to notice. What happens if you carry this insurance for 20 years and you never have an accident? You will have paid $32 and gotten nothing for it. You never once had the pleasure of having the insurance company pay your medical bills. Unlike your colleague the next cubicle over who totally didn't see that motorcycle coming and spent three weeks in the hospital. Lucky stiff.

Of course that's backwards. You're the lucky one, for not having had to suffer an accident. And yes, your premiums, and the premiums of all the other lucky people, went toward the medical costs of your unlucky colleague who got hit.

That's a basic principle of insurance: the lucky pay for the unlucky.

Next think about buying a house. When you get a mortgage, it turns out you'll have to buy home insurance as well. The house is the bank's collateral, the thing they get to take as compensation if you default on your mortgage.

If the house burns down and it wasn't insured, then the bank has no collateral. And if, after a fire, you decided to stop paying (Why keep paying for a house that doesn't exist?), there'd be nothing for the bank to take in place of your house.

So home insurance is mandatory.

Well, that's not strictly true. If you have enough money to buy a house without taking out a mortgage, then nobody's in a position to force you to insure the house (though it's still a good idea).

And if you just decide not to buy a house at all, then of course nobody will be forcing you to buy insurance.

But if you want to buy a house, and you want or need to borrow to do it, then yes, the insurance is mandatory.

Home insurance also provides an opportunity to talk about another phenomenon.

If your house burns down, there are all sorts of negatives, including physical risk to you and loss of irreplaceable personal items (like family heirlooms), and the inconvenience of needing to find another place to live, at least temporarily.

There's also a financial cost, of needing to pay for that temporary or permanent alternative housing, and the rebuilding of your old house if you're doing that.

Given all those negatives, it makes sense for you to exercise some care to avoid your house catching fire.

If your house is insured, you still have the physical risk, loss of heirlooms, and inconvenience of relocation, but you don't have the financial risk. All else being equal, we expect you to be somewhat less careful, now that you're insured.

The name for this phenomenon is "moral hazard." This crops up any time there's a risky behavior where the benefits of a good outcome accrue to you, while the costs of a bad outcome are borne by others. Your actions are biased in the direction of taking risks you really shouldn't be taking.

It turns out there are some aspects of moral hazard that can be policed. A sprinkler system reduces the chances of your house being destroyed by fire. Considering the decision on your own, you may decide it's not worth it: the sprinkler system costs a fair amount of money, you think a fire is unlikely, and if it does happen - hey, you're insured.

The insurer has a simple way of correcting for this: make it a condition of your insurance that you have a sprinkler system. But you can't always eliminate moral hazard. Maybe the insurer never thought you'd use real candles on your Christmas tree, so they didn't rule it out, and ... FWOOM! There goes your house (with the insurer on the hook for it).

Additional insights come from car insurance. There's a form of this that's voluntary: nobody will force you to buy insurance to pay for damage to your own car. But there's another type that is mandatory: if you want to legally register your car and drive it on the public streets, you have to get insurance for damage you might cause to the property, health, and lives of others.

If you get in an accident and hurt others, you might not be able to pay for that. If you're uninsured, the costs fall on the people you hurt (or their insurance companies). To avoid this outcome, the state simply says, "No insurance? No registration."

As with home insurance, there's a legal way of avoiding the requirement to buy car insurance: don't drive.

In contrast to homeowner's insurance and car insurance, there are types that are entirely voluntary, such as term life insurance. The contract for a product like this might say, "Pay us $100 per year, and if you die before the end of 20 years, we'll pay $100,000 to whomever you've indicated as your beneficiary."

I have a moderate medical condition, while my wife is in good health. 13 years ago, when we bought life insurance for ourselves, she ended up with a premium that was only 1/3 as much as mine - because I'm more likely to be dead by 2025 than she is. I still have excellent odds of lasting that long, but my wife's odds are more excellent than mine. Roughly speaking, the percent chance that I'll be dead by 2025 is about three times as large as the percent chance that my wife will be dead by then.

The principle at work here is known as actuarial fairness. It's three times more likely that I'll die and force the insurer to pay out, than that my wife will die and force the insurer to pay out. So it's fair that I should have to pay a premium three times as large as hers.

The last concept to introduce is adverse selection

Imagine we somehow make insurers blind to who's actually buying their product - say we're selling insurance out of a vending machine, so you can't tell the purchaser's age, sex, or health status. Obviously, you have to charge the same price for everyone. And let's say the chance of any randomly chosen person dying in the next 20 years makes it so that $120 per year is an actuarially fair premium averaged across the entire population.

This product won't be equally attractive to all potential customers. One person is a healthy, 23-year-old female. There's a very low chance she'll die before age 43, so she decides not to buy insurance.

But another person walks by: 55 years old, male, a weak heart. A pretty good chance he won't make it to age 75. $120 a year for his spouse and kids to receive $100,000? That's pretty reasonable.

So the unhealthy, middle-aged man buys life insurance, while the healthy, young woman doesn't. The pool of insured people is no longer representative of the population as a whole - there's a higher-than-average chance that the insurer will need to pay out. In order not to lose money, the insurer has to charge $140 rather than $120. But now the product is even less attractive to healthy young people, so more of them drop out. The pool as a whole is even sicker, so up the price goes again.

This keeps happening until the premium is so high that virtually no young, healthy people want to buy insurance. The pool has fallen apart. The only insurance that's available is very expensive, only marginally attractive to a sick person, and totally uninteresting to someone in good health. Nobody is able to sell insurance that works for young, healthy people, because if you charged a low enough premium to be attractive to them, the older, sicker people would love it and be certain to buy it, so once again you'd end up paying out more than you could afford to based on what you collected in premiums.

This is the problem of adverse selection. It's often described as when there's relevant information about your characteristics that you know and that I (the insurer) don't know. But it works the same if we think of it as relevant information about your characteristics that you know but that I am forbidden from acting upon.

The important thing is simply whether I'm unable to differentiate my price based on who my customer is, whether that inability is based on not knowing, or knowing but being forbidden from using my knowledge. In either case, adverse selection is present, and the insurance pool falls apart.

To wrap up the key ideas of insurance:
  • Risk pooling: the average outcome for a group of people is far more predictable than the specific outcome for any individual. So when we pool a bunch of people, there's a lot less uncertainty about the average cost associated with that group than about the specific cost associated with any individual in the group. Risk isn't merely transferred by pooling; it's actually reduced.
  • The lucky pay for the unlucky: if you never "get" to file a claim on your insurance, it's because nothing bad ever happened to you.
  • Mandatory insurance: under certain circumstances, the government or some private entity might require you to buy insurance as a condition of something else.
  • Moral hazard: when you are spared from the bad possible outcomes of risky behavior, while benefiting from the good outcomes, you'll take more risks than you should.
  • Actuarial fairness: when the premium you pay is proportional to the cost you are likely to represent.
  • Adverse selection: when an insurer is unable to make use of information about characteristics of its customers, that affect how costly they're likely to be. The insurer charges higher premiums to protect itself, the less costly people drop away even more, and the pool falls apart.
In the next post, we'll see what happens when we apply these concepts to insuring people's health, rather than their homes, cars, or lives.

No comments:

Post a Comment