More on Health Care Predictions

Apropos of my earlier post, Ezra Klein writes to complain that I have erroneously grouped him in with Nick Kristof.  I should be clear that Kristof's estimates on their own suggest that the death rate should drop by 20% or so over the next ten years; Ezra's estimates are just gravy.

But this triggered a back and forth about what it means to claim that "hundreds of thousands" of people will be saved by this bill, and similar.  A lot of my correspondents on the left seem to think that I am trying to set up some sort of a gotcha--to make this into a "contest".  Or that making predictions is just "sour grapes".

That seems like a more apt description of people who just induced us to pass a bill that will ultimately cost more than $200 billion a year--the fifth or sixth largest line item in the US budget.  During that debate I heard a lot about the 20-45,000 people who were dying from lack of insurance every year.  I heard about how US mortality indicators lagged behind the rest of the developed world.  I heard about infant mortality.  I heard, over and over again, about medical bankruptcies, and how medical bills were bankrupting America.  I heard about the CBO score that said this bill would be deficit neutral.  Let me know if I've missed anything, but it seems to me that mortality, financial protection, and deficit-improvement were the three major planks upon which this bill was sold.  They are certainly the bulk of the anecdotes that fill heart-rending articles and presidential speeches.

Forgive me, but to my admittedly naive ears, this sounds like what you are saying is that you think that if we cover the uninsured, we will have lower mortality rates, fewer medical bankruptcies, and a lower deficit.

So what I want is not some sort of contest between me and some other blogger, but accountability for the bill.  The bill has passed. The CBO estimates that when it is fully implemented, virtually all native-born Americans will be insured.  That's the best that any health care bill in America was ever going to do, because America is just not going to pay people to come here. (Maybe you think we should cover illegal immigrants, but hey, I think we should privatize Social Security, and I'm sure we'd both like to fly to the land of Oz on a magic unicorn.  All three are irrelevant preferences.)

However "imperfect" this bill is, you got what you wanted: virtually all the uninsured are covered, and those who aren't covered probably aren't particularly unhealthy.  So now you should be willing to state that all the marvelous things you claimed would come to pass, will actually come to pass.  Over a reasonable time frame.  You cannot tell me that we will save hundreds of thousands of lives over a fifty or sixty year time frame.  I mean, you can, but then I don't take you seriously.  That's a few of thousand lives a year, far lower than the number of American lives claimed annually by "non-steroidal anti-inflammatory drugs such as aspirin"--at a cost of $200 billion a year, or $70-100 million per life saved.  I know, every life is priceless, but US policy cannot actually be operated as if this were true.  Moreover, when you stretch out the time frame this way, your theory is non-falsifiable: a few thousand lives a year is too small to be distinguished from statistical noise.

To me, that just won't do.  Americans were not told that American households would be 1% less worried about bankruptcy, or that we'd save a hundred thousand lives over thirty years.  They were regaled with eye-popping statistics on deaths from lack of health insurance--I certainly was, by many of the very same commenters who are now suddenly wary of prediction making.  If you quoted those statistics, you were committing to a pretty strong position on the benefits of this bill.  By my count, since we're now supposed to be covering at least 2/3 of those who are currently uninsured, and the remainder are often immigrants who trend younger than the general population, you believe that we should see a reduction of at least 15,000 deaths a year.  You might argue me down to 12,000, but you couldn't get me as low as ten.  That is what is implied by citing a figure of 20,000 deaths a year.

If you quoted Himmelstein et al's 45,000, obviously you should be expecting deaths to fall by at least 25,000 a year, very conservatively.  If we don't see such improvements, then those studies were wrong. And if you won't commit to saying that you expect such a sizable reduction in our mortality rate, then you were wrong to cite them.

I mean, maybe we say that there are a bunch of combo benefits: we reduce bankruptcies by a third, save five thousand lives a year, get some harder-to-measure morbidity benefits, and so on.  But there have to be some measurable benefits.  If this helps families stave off financial ruin, we should see a meaningful and sustained reduction in the number of bankruptcies.  If it improves health, that should show up in life expectancy.  If it doesn't, then the bill doesn't do what you said you expected it to do.  That's valuable information!  Not so much about you, as about health care bills.

If you don't think that any of the effects of this bill will be large enough to measure and hopefully, large enough to justify the price tag of this bill, then I have to ask two questions:

1)  Why the hell are we spending $200 billion a year, plus the mandated spending by individuals and employers on premiums, plus the new money the states will have to spend on Medicaid?

2)  Why on earth did you bring up all these apparently irrelevant statistics?

I'm all for accountability for beliefs.  That's how you make your beliefs better. That's why I want to see all the people who threw around all sorts of theatrical arguments commit to what they are actually reasonably willing to predict will happen.  Then explain why the outcomes that they are actually confident enough to predict justify spending about $2000 for every household in the country.

I don't think that's unreasonable.  I sure wish people had done it before the bill passed--it would keep us more honest in our debates.  But post-facto accountability is still a lot better than nothing.  Otherwise the bill's supporters, and opponents, will be too tempted to move the goalposts.

Which is why I made my predictions.  I'm sure at least some of them will be wrong, human fallibility being what it is.  But the point is not to prove who the best pundit is.  It's to see how confident people really are in their mental model of the world.  We've got a lot staked on those models, now.  We deserve to know.