Robin Hanson’s health care petition

our single clearest data point regarding the marginal value of this spending, the US-funded RAND health insurance experiment,
where from 1974 to 1982, 7700 subjects were randomly assigned to 3 to 5
years of free or not free medicine, found no significant evidence of a
substantial health effect of more medicine, confirming the usual results of continuing aggregate healthmedicine correlation studies,

We the undersigned petition
the US to
publicly conduct a similar experiment again soon, this time with at
least ten thousand subjects treated for at least ten years, which
should be feasible for a half billion dollars, or one part in forty
thousand of annual medical spending.  Whatever other purposes such an
experiment pursues, it should try to make clear the aggregate health
effects of variations in aggregate medical spending, variations induced
by feasible regimes of quality control, including free patient choice
induced by a varying aggregate price.

Here is the link.  I doubt if upping the number of subjects will much change the results.  As long as we are playing mad scientist, I would prefer some disaggregated tests, namely:

1. How much better off are the poor uninsured if they get health insurance?  (Financially much better off but in health terms only slightly better off is my current guess, and yes there is already some evidence here.)

2. How much less healthy would the well-insured be if they had to consume thirty percent less health care?

3. How much healthier would we be if we retargeted expenditures to some commonly recommended areas, such as pre-natal care and prescription drugs?

And my favorite is:

4. How much would health care cost if we simply banned all health insurance and modified forms thereof? 

Except possibly for #1, these are not easy experiments to run, and yes computational modeling would beg the relevant questions.  But I think #3 — or even the thought thereof — poses the biggest problem for Robin’s worldview that medicine doesn’t do us much good.  Robin is a real world innovator, a hands-on, duct tape kind of guy, so he can’t retreat into the claim that we cannot possibly parse current expenditures more effectively.  Lots of health care does lots of good, and from there we can pick up the ball and run with it.

For more on Robin’s revisionist health care views just scroll through the last week’s entries on his blog.


4. How much would health care cost if we simply banned all health insurance and modified forms thereof?

Quite! I'd love to know the answer to that. Maybe it could be modelled then tested on historical data?

The trouble is that historically the majority of the population had virtually no savings or capability for borrowing, and the situation was sometimes complicated (in cities) by powerful guilds.

But I suspect that with no insurance most health care (eg. consultations with doctors, most drugs) would be affordable from income, medium-sized things like hernia operations would be affordable from savings (they would cost about the same as an annual vacation) - which leaves the 'catastrophic' problems and expensive chronic illness to be picked-up by charities.

So far I am persuaded by Arnold Kling's argument that only catastrophic health insurance is really necessary. But I don't understand why that isn't already the system - perhaps it has to do with health insurance being mostly an emploment perk, so that luxurious consumption of routine health care items (which could - in a system with no insurance - easily be paid for from income or savings) has now become so widespread as to generate massive inflation of these prices such that very little health care is currently affordable from income.

I tend to agree, Bruce. There are historical reasons (here in the U.S.) for how health insurance got tied to employment...but basically we created a perverse system where an employee have an incentive to purchase too much health insurance, and an employers enables this behavior. (due to tax subsidies)

I wonder what Arnold Kling and Robin Hanson will think of Victor Fuchs' Health Vouchers?

Steve, we already get substantially less medicine - it is called "Kaiser." :)

The RAND experiment oversampled from the poor in order to see the effect on them separately (there was no effect), and I would expect a successor to do the same.

Regarding item #3, that is they point of my talking about "feasible regimes of quality control." Yes, let's see what can be done there.

Hanson has libelled (sp?) Kaiser. There is no evidence that more efficient medicine is less medicine.

Amen to Pat Lynch. If the concern is that costs are spiraling out of control, by all means let people bear more of them and we'll see a rapid end to that problem.

Most of the heat generated by Hanson's comments come from the "medicine is fraud" bumper sticker. It is just not highly plausible that the benefit observed in pharmaceutical clinical trials are all false positives stemming from bias and the 1-in-20 effect. These are drugs which, to gain licensure, must show effects in two independent well-designed pivotal trials and are invariable supported by a wealth of earlier clinical and pre-clinical data. No one goes into a Phase III study expecting a 19-in-20 risk of failure, and indeed, failure to meet primary endpoint in registrational studies is nowhere near 95%.

Lots of light, by contrast, could come from following up the concern from aggregate studies that much medical spending is wasted. This finding would not surprise many Evidence-Based Medicine advocates. For a shocking amount of medical practice, benefit has not been supported in any fashion.

The fear is that a RAND repeat will simply confirm at great expense what we already know (much medicine is of minimal benefit), but will not identify which elements of health care are, in fact, unnecessary. A better option would be to specifically test practices which account for a relatively high proportion of billing, to determine level of benefit vs. no care or cheaper care.

It's interesting to break out at three different concerns:

a. Your health might be ruined by not being able to afford to go to a doctor.

The RAND study seems to say that, at least over a large range of situations in the US, this didn't apply. Everybody had some kind of maximum per-year expenditure limit, but in general, even having to pay for all their normal medicine out-of-pocket didn't make people much worse off.

b. You might get clobbered by health care costs and lose your house or go bankrupt or something.

Here, the RAND study doesn't tell us much. I don't have hard data, outside of anecdote and WSJ articles.

c. You might get worse care (or no care) because of the lack of insurance. The RAND study doesn't quite tell us about this either, though it seems like you generally can get care if you need it, albeit maybe waiting a little longer or getting a big bill at the end.

Keith, you should read Clive Crook's piece in today's FT on the French health care system. Apparently the reason the French live so long is not because they don't work much, eat small portions of food, or drink red wine. According to Crook their health care system is saving them. I'm old enough to remember when we were "blaming" the wine. Figuring out the relative impacts across countries is really difficult, so you may be right, but again it's easier to jump to the conclusion about health care in this political climate.

And my favorite is:

4. How much would health care cost if we simply banned all health insurance and modified forms thereof?

Suppose it were empirically shown that banning health insurance would reduce health care cost without damaging outcomes. Would you endorse such a ban?

If we're going to do the experiment again, let's limit it to the chronically ill and look at wealth and income effects as well as health spending and health status, so we see what the real impact of substantial cost sharing is on the sick.

I agree with this completely, thanks for the post.

Please come to conan gold, we will give you a great surprise.

Comments for this post are closed