Actually Justice, This Is How Insurance Works

From the SC hearing today:

JUSTICE ALITO: But isn't that a very small part of what the mandate is doing? You can correct me if these figures are wrong, but it appears to me that the CBO has estimated that the average premium for a single insurance policy in the non-group market would be roughly $5,800 in -- in 2016. Respondents -- the economists have supported -- the Respondents estimate that a young, healthy individual targeted by the mandate on average consumes about $854 in health services each year. So the mandate is forcing these people to provide a huge subsidy to the insurance companies for other purposes that the act wishes to serve, but isn't -- if those figures are right, isn't it the case that what this mandate is really doing is not requiring the people who are subject to it to pay for the services that they are going to consume? It is requiring them to subsidize services that will be received by somebody else.
GENERAL VERRILLI: No, I think that -- I do think that's what the Respondents argue. It's just not right. I think it -- it really gets to a fundamental problem with their argument.
JUSTICE GINSBURG: If you're going to have insurance, that's how insurance works.

That is close enough to being right that it's a forgivable mistake under most circumstances. But in this context that small distance between right and wrong make all the difference.

While you need healthy people to subsidize sick people, you don't need a group of systematically healthy people to do the subsidizing. If an illness which affects 1% of people costs $10,000 to treat, then you need everyone paying $100 for insurance to cover expected medical payments. Yes, at year's end the healthy will be subsidizing the sick but at year's beginning, you have an equal likelihood of being sick as everyone else.

But that's not what Ginsburg was talking about. To see her version, imagine some people have a 1% chance and others have a 10%. You don't need the 1%ers to subsidize the 10%ers. The 10%ers could just pay more (specifically, $1,000). In other words, you don't need an ex ante difference in expected health costs for health insurance to work. You just need to be able to tell people apart (which is why you have to fill out all those forms).

Granted, this seems mean. You can't completely control if you're a 1%er, a 10%er, a 50%er, a 0.1%er, etc. But you also have some control. Your risk is based on three factors:

D, factors that are dependent on your actions such as diet, level of exercise, how sun exposure, etc.
I, factors that are independent of your actions such as genetics, upbringing, etc.
M, factors that are mixed, such as income which comes from both luck and hard work.

p = p(D,I,M), where p is the probability of getting sick.

Forget Obamacare: ideally the government would give people enough money to buy health care, if they choose. The amount of money would be proportional to somewhere between a person's I and their I+M. Born with diabetes? That increases your expected yearly costs by $1,500 so here's $1,500. Ate poorly and became diabetic? You get nothing. Yes, this is too expensive to do perfectly, but a second-best solution would be to use existing medical records and run them through a computer to determine a subsidy. That gets us really close and is quite feasible. Plus, I'd not require people to actually use the money to buy health care. Want to spend it on ice skating clowns? Fine, but make sure they stay off my lawn.