If the petard fits . . .

By Megan McArdle

The Economist's Free Exchange urges caution about the finding that the income-health gradient is even steeper in Canada than it is in America:


If the best evidence is not very good, then we need to be careful about drawing hard conclusions, and more careful still about "screaming them from the rooftop." I thought that the combination of the qualifier "slightly" in the phrase "the health-income gradient is slightly steeper in Canada than it is in the U.S.", along with the questionable nature of the data was sufficient to give us pause before beginning to evangelise. Apparently, that's not the case.

I would suggest that we ought to be very circumspect in analysing the health-income gradient--especially when making cross-country comparisons. For instance, research findings show that the health-income gradient is far flatter among Latinos--in general, Latinos are much healthier than their income levels imply. In America, Latinos make up a far larger share of the population than in Canada, and they are almost certainly overrepresented in the population of the uninsured. Have the NBER paper's authors taken such factors into consideration?

Research also shows that the income-health gradient for the population as a whole depends on the importance of within-group effects of income on health for subpopulations. If within-group effects are strong, then greater economic inequality between population subgroups can actually flatten the income-health gradient for the population as a whole.

It isn't easy to say how these different effects stack up, and that's entirely the point. Screaming from the rooftops inplies certainty of which there is none here. What we do know with certainty is that access to health insurance is more equitable in Canada, and total health outcomes are better in Canada. Without better data on the income-health gradient, I don't believe there is cause to question the arguments of those emphasising the equity benefits of single-payer systems.



Point well taken. Except . . . how come advocates for single-payer are so rarely interested in exploring these holes in the data when they support the argument for equity? Why don't we hear about the problems in cross-country comparisons of commonly used metrics such as infant mortality? Moreover, why doesn't the blogger mention that America also has a much higher percentage of African Americans, who, for reasons that are not clear, have higher mortality and premature birth rates even when things like income and education are controlled for?

You cannot argue, on the one hand, that cross country comparisons make it hard to really know whether Canada's income-health gradient is steeper; and then turn around and say that "for half what we spend, Canada offers insurance to all its citizens and produces better outcomes in practically every category listed: life expectancy at several ages, infant mortality, mortality from chronic conditions, mortality from various other diseases (including respiratory diseases, despite a higher rate of smoking), and so on." I agree that cross-country comparisons are fraught, though also, unfortunately, the best we have. But either you rely on that sort of data, or you don't; you can't use general arguments about the reliability of international data to dismiss specific studies that don't bolster your case.

This article available online at:

http://www.theatlantic.com/business/archive/2007/09/if-the-petard-fits/1983/