By now, almost everybody in Iowa and quite a few people beyond it have heard Hillary Clinton or John Edwards attack Barack Obama over his health care plan.Edwards and Clinton would both require every American to obtain insurance, a proposal known to policy wonks as an “individual mandate.” Obama would require all children to have insurance, but not all adults. Citing that difference, Edwards and Clinton have said that Obama's plan could leave as many as 15 million people without insurance.

Curious where they that number came from? Well, it seems to come from me. Sort of.

The day Obama unveiled his plan, back in May, I wrote an article about it. I praised the plan as an ambitious, meaningful proposal that would likely improve the lives of millions of Americans. I praised Obama himself for showing a genuine commitment to this issue. Then I criticized him for not including a mandate, which one of his rivals, Edwards, had already done. (Clinton hadn't yet unveiled her plan.) In other words, I said that Obama's plan was good but not great—the same verdict that other close observers of health care policy, like Ezra Klein and Paul Krugman, reached.

In the course of making my case, I thought it important to convey some sense of scale. So, based on my reporting, I suggested somewhere around 15 million people—or roughly a third of the people now without insurance—might still lack coverage, even if Obama's plan were implemented.

That magic 15 million figure has since worked its way into the campaign. Edwards was the first to invoke it, during a televised debate in early June. But it's Clinton who really put the estimate into play in mid-November, when she started using it as part of her attacks on Obama's health care plan—and distributing my article as proof of its validity. Most recently, Clinton cited the figure and my article on Friday, when calling upon Obama to retract an advertisement in which he promises his plan would cover everybody.

Mandates raise lots of interesting and complicated questions, about which I hope to say something more intelligent shortly. In the meantime, though, here's the explanation of how I got that figure—and whether, six months later, it still seems sound to me.

***

It's pretty much conventional wisdom that, without a mandate, a substantial portion of Americans would remain uninsured. But to come up with a figure, I relied heavily on conversations I had with Jonathan Gruber, an economist at the Massachusetts Institute of Technology.

Gruber is a highly regarded economist who specializes in precisely these kinds of issues. Although he served in the Clinton Administration and is generally identified with Democrats, politicians of both parties have sought his advice. Back when Governor Mitt Romney was setting up his universal health care plan for Massachusetts, he brought Gruber into the process. Today, Gruber continues to serve on the board of the Connector, which is overseeing the Massachusetts plan. 

The reason so many people ask Gruber's advice is that he has developed a model, based on past data, for projecting how various policy changes will affect the number of people who obtain health insurance. It is similar to the model used by both the Congressional Budget Office and the Treasury Department. (You can read more about him in this Washington Post story.) Since all three of the leading Democratic contenders, including Obama, were known to be have sought his input this campaign cycle, I figured that made him a particularly reliable source of guidance.

Gruber told me that his projections showed that, without an individual mandate, a program of very generous subsidies and market reforms would bring in close to half the uninsured population. Adding a child mandate, he said, could bump it to two-thirds. Since Census figures showed around 45 million uninsured, I asked if that meant roughly 15 million would still lack insurance. He said that sounded about right. I put that figure in my story (although, in my quick translation of our conversation, I explained the step-by-step math incorrectly—saying that the starting baseline for coverage without a mandate was one-third, not one-half). I didn't attribute this to Gruber directly, though I'd cited his work elsewhere in the article, since that part of our discussion had been on background. He's since made these views public.

*** 

It was a crude, back-of-the-envelope calculation based on a projection. And all projections contain some uncertainty because they make assumptions, some of which might turn out to be wrong. In other words, the figure should not be treated as gospel.

On the other hand, figures like these also represent the best available information we have for evaluating policy proposals.  And a projection like Gruber's is no less reliable, certainly, than a lot of the other numbers campaigns routinely throw around. Take, for example, Obama's promise that his plan will require only $50 to $65 billion a year in new revenue once fully phased in—because he anticipates generating huge savings from better management of disease and use of information technology. It's a number the campaign has provided. It's reasonable to put at least some stock in it because they likely reflect the influence of David Cutler, a Harvard economist who studies health care extensively and is also widely respected for his intellect and honesty. But if you're going to take that figure seriously, then, it seems to me, you certainly have to take Gruber's figures seriously, too.

And, more important, Gruber's views are hardly out of the mainstream. Over the last few days, I've had the chance to do something I couldn't with that first-day story: consult several other leading authorities. One of them is Len Nichols, an economist who worked on the 1993-94 Clinton health care effort and today heads up the health policy project at the New America Foundation. “Every reasonable model out there ... will show you that the kind of subsidies that we could do, 50 percent or so, are going to get you half," he said. "The way you go from half to 15 [million] is the kid mandate.”

These days, Nichols is probably the most visible promoter of individual mandate schemes in Washington. And, for those who want to see hidden agendas at work, Gruber's advocacy of individual mandates is also well-known. Then again, in both cases, they've reached that conclusion based on their respective research histories. If they have a bias, it's in favor of a policy as opposed to a candidate—a policy that they happen to believe is right. 

Still, for the purists out there is always somebody like John Holohan, who directs the Urban Institute's Health Research Center and, as best as I can tell, has no direct connection to the presidential campaign. Holohan commands universal respect, too, having worked on these sorts of problems for two decades. And he's pretty much where Gruber and Nichols are on this question. Without a mandate, he told me, “Obama would still leave about 22 million, 23 million, but he has a mandate for children, about 9 million uninsured kids, so assuming you get most of them, you get pretty close to 15 million.”

Just to be sure this sentiment wasn't purely a project of Washington group-think, I contacted Altarum, a non-profit health care research institute based in Ann Arbor, Michigan. They hadn't modeled a plan like this specifically and warned that, without more details, they couldn't be precise. But with those caveats out of the way, analyst George Miller and economist Charles Roehrig sent me an e-mail explaining that "We've done some very crude hand calculations that suggest that the estimate of 15 million uninsured under an Obama-like plan (no individual mandate, coverage of all children, incentives) is in the right ball park."

By the way, while the Clinton campaign had been circulating my article as evidence to back their claim, an official later told me that they weren't relying on me exclusively. Before using the number publicly, they'd consulted experts independently—presumably, some of the same ones I did.

***

For the record, the Obama campaign continues to dispute the 15 million figure, arguing the estimate is based too heavily on a generic reform package and not a plan with the specific provisions that Obama has. They are particularly enthusiastic about the potential for automatic insurance plan enrollment at the workplace to boost participation. One economist, Sara Collins of the Commonwealth Fund, has gone on the record as saying the 15 million figure sounds too high. (She told that to Factcheck.org, which ran a story on this right after the controversy started).  Holohan, too, told me that auto enrollment might help bump up the participation, though he couldn't say by how much. Jacob Hacker, a Yale University political scientist (and occasional TNR contributor) who has also been working with all three campaigns, concurred.

Gruber, Nichols, and Roehrig, on the other hand, were more dubious that the new enrollment practices would change much. Among other things, Gruber explained, his two-thirds estimate already made optimstic assumptions about Obama enrollment practices. In addition, most of the evidence on automatic enrollment comes from studies of 401K retirement accounts. (Research has shown that when you enroll people automatically but let them opt out, they are much more likely to sign up than if you simply give them the option to enroll. In other words, it's the default choice that matters.) And it's not necessarily true, as Roehrig noted, that people will treat health insurance the same way.

My own opinion—and it is only an opinion, albeit one by somebody who's spent a lot of time trying to figure out health care policy—is that Gruber, Nichols, and Roerhig are probably more right than wrong. Automatic enrollment could be a huge help, but it seems like the crude two-thirds estimate already takes that into account, at least partly. (Also, the paperwork for the Obama plan doesn't actually say anything about automatic enrollment, although Cutler referred to it—a bit obliquely—in this article posted after Obama first unveiled the plan.)

So I guess I'm right back where I started: 15 million is a very, very rough estimate of how many people might still be uninsured if Obama's plan became law. But these are the figures we use in campaigns.  And at least a few well-respected authorities, none of them tied to one candidate, think it makes sense.

***

Of course, that's only half the story—as the Obama campaign will eagerly tell you. For the last few days, they have been concentrating on a different argument altogether: that the estimates of their rivals' plans are too optimistic. Austan Goolsbee, a top Obama advisor who is also an economist at the University of Chicago, has written a memo arguing that Obama's plan may actually cover more people than either of the rival plans would—once you take into account that the Clinton and Edwards plan would leave out millions, too.

Is he right? Is this really basically a wash?  My best guess on that very important question is coming shortly.

Note: The world of health policy is small, so I've gotten to know virtually every expert cited in this article—including Cutler, Gruber, Hacker, and Nichols—well enough to call them friends. I don't think that biases me, particularly since they occupy different sides in this dispute, but I thought I should make that clear.

--Jonathan Cohn