In a previous post I wondered whether additional spending on health care in fact brought greater health. Whether we do controlled experiments, or look at cross-national data, a convincing effect is often hard to find.
Brendan Kennelly of Lehigh referred me to this OECD paper, which examines cross-national data to find a positive health effect from health care spending. The result is much stronger for women than for men.
I asked Robin Hanson what he thought of the piece, here is his response:
When a coefficient is truly zero, about 5% of the studies estimating that
coefficient should show a significant difference at the 5% level. And if
people search for specifications that will give this result because they
believe the coefficient is not zero, you might expect 10-20% (or more) of
the reported results will show a significant non-zero coefficient. So
given the current state of econometric practice, the most you could
reasonably hope for when a coefficient is really zero is for about 90% of
the studies to fail to find a significant effect of the desired sign. This
is the current state of the evidence on the aggregate effects of medicine
My take: The whole debate still makes may head spin. I wonder how much of health care expenditure is geared toward alleviating anxiety (“we did all we could…”) rather than improving health per se. We may be mismeasuring the relevant output. If we could measure the total costs of illness, what percentage of the total would come in the form of anxiety? Furthermore, health care is not a single thing. The marginal return from some forms of health care (bypass operations?) is certainly high, in other cases the return much lower or perhaps negative. So perhaps the debate should shift away from aggregates, and toward trying to identify which health care expenditures we can do without, or do not need to subsidize at current margins.