Photographer: Joe Raedle/Getty Images
Photographer: Joe Raedle/Getty Images

Supporters of the Affordable Care Act are basically counting on its beneficiaries to make it impossible to repeal. That makes it important to know exactly how many of those beneficiaries there are.

As I’ve written before, it’s hard to know yet -- Medicaid signups will continue for the rest of the year, and while we know how many people selected a policy on the exchanges, we don’t know how many of them have paid. But one number should be pretty easy to figure out: how many young people have gained insurance through their parents’ policies, thanks to the Obamacare provision that allows them to stay on those policies until they are 26.

And yet, that’s not very easy to figure out. You’ve probably heard 3 million. That’s the administration’s preferred figure. But it is almost certainly way too high, as Avik Roy notes:

Sommers justified this estimate using figures from the National Health Insurance Survey from the Centers for Disease Control. The NHIS survey, among other things, breaks down the percentage of U.S. residents with private coverage, public (i.e., government) coverage, and no coverage. Sommers wrote that “from September 2010 to December 2011, the percentage of adults 19 to 25 with insurance coverage increased from 64.4% to 74.8%, which translates into over 3 million additional young adults with coverage.”

There are a number of glaring flaws with Sommers’ calculation. First off, he doesn’t distinguish between growth in private coverage vs. public coverage; the under-26 mandate only applies to private coverage. In addition, not every young adult with private coverage is in a family plan where the mandate applies. That is to say, if you’re between 18 and 26 with private single coverage that you’ve purchased on your own, or obtained from your employer, the under-26 mandate isn’t relevant.

Second, he assumes that the entirety of the growth in coverage is due to Obamacare.
This is highly unlikely.

To state the obvious, 2009-2010 was an unusually bad time for the economy, especially when it came to employment. As the economy improved, the number of uninsured began to fall naturally -- which we can see because it improved in all age groups, not just those ages 18 to 26.

Even if you ignore this problem, however, it’s very hard to figure out how many people gained insurance this way. There are multiple data sets you could use, and multiple meanings of the word “uninsured.” To see what I mean, look at the data from the National Health Interview Survey, which the administration used to generate its figure:

How should we interpret these data? The number of uninsured spikes up and then falls again; compared with 2008, it has improved only slightly, if at all. And if we look at the bottom line, which shows people who were uninsured for a year or more, it’s basically the same as it was before the financial crisis. If you look at the line showing people who were "uninsured at time of interview” -- which includes a lot of people who were only temporarily uninsured -- then it’s improved a lot.

Which group should we look at? I can argue it in multiple ways. Most of us probably would not have endorsed a radical overhaul of the health-care system in order to solve the problem of adult children who are uninsured for a few months between jobs or after graduation. On the other hand, it is statistically likely that a few of those so covered have had accidents or illnesses that otherwise would have been a financial problem. On the third hand, those people are unlikely to form much of a coalition against repeal.

We also have to consider that with employment still depressed, particularly among the young, those numbers would probably look worse if it weren’t for the Affordable Care Act. How much worse? Very hard to say.

But it gets more complicated still, because there are other data sources. There is the Medical Expenditure Panel Survey, which seems to show a much greater gain for the uninsured:

On the other hand, the U.S. Census barely shows any improvement at all:

Both the Census and the MEPS define "uninsured" as people who have been without insurance for the entire year, yet their results differ greatly.

Not complicated enough for you? Then ask yourself why we’re just looking at the percentage of young people who are uninsured. After all, if someone goes on Medicaid, those numbers will improve -- but you can’t attribute it to a policy allowing them to stay on their parents’ insurance until they’re 26. We should really be looking at the percentage of this age group who have private insurance through an employer:

If you graphed youth unemployment over the same period -- which I would love to do for you, except that the Bureau of Labor Statistics' data tool keeps freezing on me -- you’d see that it is basically the reverse of this pattern: It spikes up in 2009 and 2010, then begins to fall. So while it has improved, some of that improvement likely reflects people in this age group getting jobs. But even if you assume that the improvement since 2009 is all Obamacare, the number is nowhere near 3 million.

All right, so it seems likely that there has been some improvement, mostly among people who would have gone without insurance for a couple of months between jobs. How does that translate into numbers?

A couple of caveats about these numbers: They measure different periods. The NHIS only had annual data from 2008 on, but it ran to 2013; the MEPS and the Census had data starting in 2007, which I picked as the last normal year before the financial crisis. But the MEPS only runs to 2011, the Census to 2012.

More difficult is that they measure different age ranges: The MEPS uses 18 to 24, as does the Census. Only the NHIS looks at the whole age group we’re interested in: 19 to 25.

With those caveats, even allowing for that variation, you can see that there is a lot of variation, depending on whether you measure from the start of the data or the worst year, and on how you define “the uninsured.” This is also true of private insurance:

None of these series will get you to 3 million. The MEPS comes closest, but as Aaron Carroll has pointed out, the NHIS and Census provide the most reliable estimates for population subgroups because they have the broadest samples.

If you adjust for the fact that some of these people would have gotten employer-based insurance anyway, as the economy recovered, then reasonable estimates of the number of young people who now have insurance thanks to this particular provision range from “almost none” to “under 1.5 million.” And the majority of people who gained insurance this way are probably people who otherwise would have been uninsured for months, not years.

Is that a good deal or a bad one? That’s a question that will have to be left to the reader. This particular provision probably doesn’t cost much, in aggregate. But given that most of the people covered by it are healthy people who would have gotten insurance pretty quickly anyway, it probably doesn’t provide much aggregate benefit, either.

To contact the writer of this article:
Megan McArdle at mmcardle3@bloomberg.net.

To contact the editor responsible for this article:
Brooke Sample at bsample1@bloomberg.net.