Bryan Caplan asks what the value of health care is, and concludes that at the margins, it's not very much. In some ways, he's right: most of the increase in life expectancy over the past thirty years, for instance, has come from improvements in public health rather than advances in medical technology. If we really wanted to boost our life expectancy numbers, the best way to go about it, I think, would be to set up effective public health programs—dieting, exercising, anti-smoking campaigns—and improving income inequality (yes, really). Health care is quite obviously valuable to individuals: that's why we pay so much for it, when we can, and often when we can't. But in the aggregate, it's not everything. (Of course, is there even any use in considering things in the aggregate?)

A related question, meanwhile, is what the value of insurance really is. Economists have found in the past that insured Americans generally spend about 40 percent more on health care than the uninsured. But it's not at all obvious what the effects of this disparity on individual health actually are. These spending numbers could be skewed by the fact that people are buying insurance because they expect to spend a lot on health care in the near future. Or perhaps the uninsured simply need less care, on average, for other reasons.

Earlier this year, Joseph Doyle of MIT did an interesting study on this question by looking at inpatient treatment for victims of severe auto accidents. Since people, usually, don't choose to be in car accidents, this gets rid of any distortion due to self-selection. What Doyle found was that hospitals spent about 22 percent less on uninsured inpatients, who also received 20 percent fewer days of care, on average. The uninsured received fewer spinal fusions, less skeletal traction, fewer organ operations, less plastic surgery. They did get more stitches, however! Now this disparity certainly had an effect: Doyle found that the mortality rate for the uninsured is about 1.5 percentage points higher in a sample with a mean mortality of 3.8 percent. (He also has some very noteworthy data on the differences between different hospitals, along with evidence that Medicaid reimbursement rules encourage more costly treatment.)

Now all that implies a 0.45 percentage point increase in the lifetime risk that an uninsured person will die in a serious auto accident. Solid evidence that there are real consequences to being uninsured. It's not because the uninsured have weird personal characteristics, but because they're treated differently by providers. As you would expect. In an added twist, Doyle argues that the money saved by not buying health insurance more or less compensates for this risk. (He assumes the value of a life is $3 million and the price of a catastrophic health insurance policy with a $1500 deductible is about $300 a year.) So "the benefits and costs of catastrophic insurance are roughly similar."

At any rate, this is all sort of moot, since "the average person" is only so important when it comes to policy. Some people suffer a great deal from not being able to afford insurance, far, far more than that 1.5 extra percentage points of mortality, and those are the individuals who, in particular, have been failed by our health care system.