Is she "uninsured"? Depends who's asking. Photographer: Joe Raedle/Getty Images
Is she "uninsured"? Depends who's asking. Photographer: Joe Raedle/Getty Images

How many uninsured people have gained insurance as a result of Obamacare? Earlier this month, the Congressional Budget Office projected that the number of uninsured would decrease by 13 million people this year thanks to the Affordable Care Act. Are we on track to make that number?

The answer is "I don't know." And it will remain "I don't know" for a very long time.

Source: Congressional Budget Office
Source: Congressional Budget Office

We heard last week from the administration that five million people had selected a plan. At first glance, that seems to be within spitting distance of the six million projected to buy exchange policies this year. But not all of those people will have paid -- estimates of the rate of attrition that reporters are getting from insurers range between 10 percent and 20 percent. And industry expert Bob Laszewski has suggested that another 2 percent to 5 percent are paying the first month's premium, then failing to pay the second month's.

Even if we had solid figures for the number of people who have actually paid for a policy, we still wouldn't know the net effect on coverage, because many of the people who have bought policies on the exchange already had insurance. How many? We have some suggestive survey numbers, but no solid data yet.

Meanwhile, some previously uninsured people are now insured thanks to Medicaid coverage. How many? Again, we don't know. There's a lot of churn in Medicaid, with millions of new people signing up every year, even in years when the Affordable Care Act isn't launching to much fanfare. It's hard to distinguish new enrollments from normal churn -- and even harder to distinguish people who were previously eligible but finally signed up because of all the publicity surrounding Obamacare, and people who are newly eligible under the law.

Why is it so hard to tell what effect this law is having? Because collecting that sort of data is very hard. Here's an example: New York asks people whether they were previously uninsured when they sign up through the state exchange. Does that mean that we can at least get good data for one big state?

Unfortunately, no, it doesn't. That number probably won't be directly comparable to the data on the uninsured that the CBO is using, because there are lots of different ways to calculate "the uninsured." When I left my last job, I waited a month or so to activate my Cobra coverage. Was I "uninsured" during that time? Different surveys give different answers -- the census defines the uninsured only as people who had no insurance at all during the prior calendar year, while the CBO assigns people whose insurance status changes over the year to a "primary" group, and other surveys pick up anyone who happens to lack insurance at the moment the survey is given. It's a good bet that New York's question is picking up a lot of people who don't have insurance right now (not surprising, or else why are they buying it?) but not people who are "uninsured" as the census or the CBO would define them.

To really know what has happened to insurance coverage, we need to compare responses to the same questionnaire over time. The only such data we have right now comes from the Gallup poll. Eyeballing that, I'd say that the number of uninsured people has fallen by somewhere in the neighborhood of 1 percent, or about 3.1 million people.

But that's just a very rough guess. The data I really want to look at, from the census, won't be out until 2015. Until then, we'll just have to wonder.

To contact the author of this post:
Megan McArdle at mmcardle3@bloomberg.net.

To contact the editor responsible for this post:
Timothy Lavin at tlavin1@bloomberg.net.