Are health care suppliers to blame for high health care costs in the United States? Uwe Reinhardt and others say yes.

I am not convinced.


The basis of this claim is somewhat weak, however. In health care, it is difficult to measure output. In theory, you want to measure the impact that health care has on people’s lives, but in practice this is difficult to quantify…

The Reinhardt paper tries to use different sorts of approximations to measure health care output. For example, they compare across countries the number of physician visits and the number of hospital admissions per capita, and they find that the United States does not rank especially high on these indicators…

Using inputs as a proxy for outputs is quite problematic, because of productivity differences. If you measured agricultural output by counting the number of farmers, then you would never guess that the United States is such a large food producer relative to other countries.

ore generally, I am concerned that Reinhardt’s approach risks running afoul of the fallacy of indirect inference. Indirect inference means drawing a conclusion about something you have not observed based on what you have observed…

For example, suppose that you find that differences in years of schooling explain only 10 percent of differences in income. You might infer that the all of the other differences — 90 percent — must be due to innate ability. That would be an indirect inference, and it would be wrong. To give just one example of how you could be mistaken, it might be that a better measure of schooling — which takes into account quality, for example — explains 80 percent of the differences in income. (I’m not suggesting that 80 percent is the true figure, only that it is theoretically possible.)

Diagnosing high prices as the cause of high spending on the basis of imperfect measures of the quantity of services is an example of indirect inference. Such an inference is quite tenuous. In this context, it might be considered statistical malpractice.

For Discussion. What direct evidence is there of health care supplier overcharges that have a significant effect on overall health care spending?