There are two tough questions facing the folks who want to remake Medicare into a premium support (or voucher) program.

The first is that, despite decades of attempts, we don't have a single good example of a competitive health insurance market driving down prices in a big way. It hasn't worked in the Federal Employee Health Benefits Programs or CalPERs. It hasn't worked in Medicare Advantage or the Texas state exchange. Now some people think that perhaps it can work, and these efforts just got it wrong, somehow. I actually count myself in that camp, at least somewhat. But so far, the fact is that it just hasn't worked.  

The second is that we do have evidence that a major way insurers compete is through trying to figure out who's healthy and who's sick and insuring the healthy people and turning away (or pricing out, or underserving) the sick people. And that's a kind of competition we don't much want.

This is of particular concern with proposals to open Medicare to more competition because the results could be so devastating. If private insurers manage to get the healthy seniors, and Medicare gets stuck with the unhealthy seniors, Medicare could effectively be destroyed as it gets caught in a death spiral where it needs to raise prices because it's been stuck with sicker patients, which drives out even more healthy patients, which requires higher prices, and so on. The result will be that premium support both doesn't work, and it wounds the Medicare program.

Which is why we should worry about this graph:

That comes from "Favorable Selection, Risk Adjustment, and the Medicare Advantage Program," a new paper in the journal Health Services Research that health economist Austin Frakt flagged on his blog. It's not a particularly easy to read graph, but what you're seeing is a bias in the Medicare Advantage program -- which is the program where private insurers compete for Medicare beneficiaries -- toward the healthy and away from the sick. 

In particular, that graph shows us that in every single year from 1999 to 2008, the Medicare beneficiaries switching out of the private plans and back into traditional Medicare were sicker and more expensive than the beneficiaries who stayed in Medicare Advantage's private plans. That implies that the private insurers in Medicare Advantage have figured out how to underserve sicker seniors such that they're driven out of the program and back into traditional Medicare. That's good for their bottom line, but bad for the program. If this is how private plans competing with Medicare save money, then they're not saving money. They're shifting costs.

Reihan Salam took a look at this data and concluded that it might "bolster the case for phasing out Medicare FFS, as doing so would mitigate the adverse selection dynamic as MA plans would be competing against each other rather than the traditional program." I'm not really seeing it. If what we're learning is that private insurers can and will try to avoid sicker seniors, it's hard to see how that strengthens the argument for getting rid of the public insurer that doesn't discriminate against sicker seniors and moving entirely to the private insurers that do.

A more natural conclusion is that premium support will live or die based on its ability to penalize plans that select healthy patients and reimburse plans that end up with sicker patients. This is what they do in a variety of European countries, but most health-care experts I've spoken to think we're much worse at risk adjustment than insurers are at risk selection, and we're not likely to embrace the level of regulatory oversight that European countries are comfortable with, and that keeps their insurers (most of which are nonprofit) more or less in line.

As I've always said, I'm open to premium support. But I think it makes a lot more sense to begin by piloting the program so we can see how it works, as the Urban Institute has suggested, rather than by moving Medicare over to it entirely and simply hoping for the best.