Peter Orszag walks us through the decision to downgrade the number of American citizens who cannot afford health insurance to "more than 30 million".

Last night, President Obama stated: "We are the only democracy--the only advanced democracy on Earth--the only wealthy nation--that allows such hardship for millions of its people.  There are now more than 30 million American citizens who cannot get coverage."
 
Today, the Bureau of the Census released the most recent data on the number of uninsured Americans. The report, Income, Poverty, and Health Insurance Coverage in the United States: 2008, reveals that 46 million people were uninsured in 2008, the last year for which there are data. These data are based on the Current Population Survey. With two different numbers, there has been some confusion as to which is accurate.  Well, both are -- and the President's version is more focused on the relevant target population for health reform since it excludes unauthorized immigrants.
 
The Census report indicates that of the 46 million uninsured individuals, 34 million were native born and 2.8 million were naturalized citizens.   The report thus shows that there were 36.8 million uninsured U.S. citizens (native born and naturalized) in 2008. An alternative calculation includes legal immigrants, which based on a figure from the Pew Hispanic Center would bring the total to something like 39 million.

Some ambiguity surrounds how to treat individuals who are already eligible for public insurance programs like Medicaid and S-CHIP but do not enroll in those programs, which estimates from the Kaiser Commission on Medicaid and the Uninsured suggest may amount to millions of individuals. These individuals are uninsured but some interpretations would suggest they should not be counted among those who "cannot get" coverage. Subtracting them from the total would produce a number closer to 30 million.
 
To be conservative, the President thus stated that "more than 30 million American citizens" cannot get coverage.

I can't get this math to work.  The actual number of the uninsured, according to the census, is 46.3 million.  Of those, 36.8 million were natives of the US or naturalized citizens.  There is no "alternate figure" for the number of American citizens that includes legal residents.  Legally resident immigrants are fine people.  But they are, definitionally, not American citizens.

Now when I click through to the Kaiser link that Orszag provides, I find that it says that "A quarter of the uninsured (11 million) are eligible for public programs but not enrolled."  36.8mm - 11mm is 25.8 million, not more than 30 million.  But of course, that 11 million figure is from 2007 and will have grown, because that's what population statistics do--there's no evidence that we've gotten better at enrolling people in Medicaid since 2007.  I mean, the Medicare rolls have gone up.  But the number of poor people went up in tandem.   Demographics being what they are, the number of seniors on Medicaid for nursing home care went up as well.

So it's actually something less than 25.8 million. How much less?  Dunno.

A "conservative" estimate would have been "more than 20 million American citizens", not "more than 30 million".  

Of course, if they'd wanted to be really conservative, they would have grappled with the other figures in that study, such as the 15%--6.7 million +/- natural growth--that Kaiser designated as being able to afford health insurance.  To be fair, the study also concluded that some of them might have difficulty obtaining it for various reasons, so it's hard to know exactly how many of them to include.  Still, the number of people who can afford insurance, but choose not to buy it, is not zero, or even close to zero.  Indeed, their existence is implied in the very fact that reform advocates are expecting to use their premiums, through the magic of pooling, to help pay for the uninsurable, and the indigent.  If they don't exist, we're in even deeper financial trouble than we thought. 

But really, in most states, an income of $66,000 for a family of four, or $32,000 for a single, is not too poor to spend several hundred dollars a month on insuring against a catastrophic health event, which is how much insurance costs when you risk rate it, include cost sharing, and don't load it up with mandatory coverage. Obviously, there are exceptions:  people in high cost cities that mandate high cost policies (New York), and people who have a gnarly pre-existing condition that makes them uninsurable.  But you should probably account for the rest, somehow, if you are going to talk about those who are "unable" to obtain insurance.

Someone trying to be truly conservative would also have attempted to control for the fact that government figures tend to show that only around half of the uninsured lack insurance for as much as a year--though of course, that too, is tricky, because it seems that 40% of the very long-term uninsured are Hispanic, and we have to assume that a lot of those are illegal immigrants, who would be the group most likely to lack insurance.  We already threw them out of the sample. Trying to back a reasonable modifier out of that would be hell.

Still, once you adjust for (mostly young people) who are choosing not to buy insurance, and families taking a short term gamble between jobs, neither of whom I think can technically be said to be "unable to afford insurance", you're conservatively going to end up with a number below 20 million.  Generously, a serious "conservative estimate" would be "more than 15 million".  But of course, the administration does not really want to be conservative; they want to have the flashiest numbers possible for their plan.  All administrations do.

Still, even if you say that my modifications are too indeterminate (fair enough), and go back to the figures that Orszag himself laid out, his calculations are off by millions of people.  

Does it matter?  15 or 20 or 25 million people is still a lot of people.  But it matters for the same reason that the difference between 66% and 80% matters.  You can't have a debate where everyone gets to bring their own statistics. 

We want to hear what you think about this article. Submit a letter to the editor or write to letters@theatlantic.com.