The New York Times reports on a recent paper by Harvard’s David Cutler and two co-authors.

The study calculated, however, that Americans of all ages spent an average of $19,900 on medical care for each extra year of life expectancy gained over the last four decades of the 20th century. And that cost is worth it, the study authors say.

I was emailed a copy of the paper, but I do not see it on line anywhere.The methodology is rather crude. They add up the increases in longevity over the past 40 years and they add up the health care spending over the past 40 years, and they compare the two. They do lots of disaggregation by age group and time period–I’m not saying it didn’t take a lot of work. But it’s crude. There are no direct estimates of costs and benefits.

The authors assume that 50 percent of the gains in life expectancy over the past 40 years are due to health care. At my talk the other day, I said that my rule of thumb would be to attribute only 25 percent of the gains to health care. The authors justify the 50 percent number, but if I were right then they are too optimistic about the benefits of health care.

On the other hand, most health care is what is called palliative care, designed to relieve pain rather than extend life. So they are probably, if anything, too pessimistic about the benefits of health care.

I actually believe two seemingly contradictory things. One is that, on average, health care spending is cost effective. The other is that, at the margin, a significant amount of spending is on medical procedures that are not cost-effective. In the health economics literature, my view is known as the “flat of the curve” hypothesis–obviously, I am not alone in believing it.

The flat-of-the-curve hypothesis is that if you were to rank-order treatments by cost-effectiveness, with the highest benefit/cost ratios first, then the first-ranked treatments basically carry the whole health care system. That is, they are so beneficial that in the average of the whole system the good treatments are not canceled out by the bad treatments.

Think of a treatment that costs $25,000 but saves an infant’s life and generates 75 years of life-expectancy. At $100,000 a life-year, that treatment is worth $7.5 million, which means that you can waste over $7 million somewhere else and still say that in the aggregate medical care is cost-effective.

David Cutler’s methodology does exactly that sort of aggregation. If there are policy implications that follow from this type of study, they elude me.