Wednesday, June 29, 2011

Inequality of Mortality

Peter Orszag wrote a column for Bloomberg describing some advances in technology that can help people track their health status. Along the way, Orszag cited a recent study by the National Academy of
Sciences that inequality in mortality was rising in the U.S. He writes: "Among 50-year-old men, for example, those in the highest education group are now projected to live almost six years longer on average than those in the lowest education group -- and this differential has been rising sharply."

I looked up the NAS report, or at least the free pre-publication uncorrected proofs copy available here. The report is generally well-done, as one would expect. But both on issues of mortality inequality within countries and between countries, there seemed to me some narrowness of perspective that left out complementary views.

On mortality inequality within countries

The report explains that inequality or mortality is increasing within countries because mortality rates for those with higher socioeconomic status (proxied by education, income, or a mixture of the two) are experiencing larger gains in mortality than those with lower socioeconomic status. This seems to hold true in recent decades not just in the United States, but also in a number of countries of western Europe. In turn, the report seeks to explain these differences in terms of behavioral patterns (like smoking and obesity) and health issues related to social status.

While this increase over the last few decades in inequality of mortality is certainly worthy of discussion, it is also worth noting an enormous decline in inequality of mortality, in countries all over the world, in the longer term.  In my own Journal of Economic Perspectives, Sam Peltzman took on this topic in Fall 2009 issue in "Mortality Inequality." Peltzman uses a measure of inequality familiar to economists called the Lorenz curve, which is usually applied to measures of income. The top figure is based on data for 1852; the bottom figure on data for 2002. The straight line at a 45-degree angle shows perfect equality of mortality: that is, 20% of the population lives 20% of the total life-years at this time; 40% of the population lives 40% of the life-years for this group, and so on. The curved line is based on actual data. It shows that with high infant mortality, the bottom 30% of the distribution lived close to 0% of the life years. The gap between the perfect-equality line and the data curve shows the degree of inequality. By 2002, life years are MUCH more evenly distributed across the population.

This huge decrease in inequality of mortality outcomes over the last century or two is not just in the United States, but also in a wide array of other high-income and lower-income countries. Peltzman writes: "The substantial increase in longevity over the last century, both in the United States and around the world, is well-known. This essay has documented another aspect of that progress: a considerable contribution to social equality. The dominant fact about this history from a worldwide perspective is how much this aspect of human inequality has diminished. ... Inequality of lifetimes is well along in a historical transformation from a major source of social inequality into a minor one."

The recent trends from the NAS report do not alter Peltzman's basic conclusion.

On differences across countries

The main emphasis of the NAS report is, as the title reveals, "Explaining Divergent Levels of Longevity in High-Income Countries." A particular emphasis is that gains in U.S. life expectancy haven't been keeping up with gains elsewhere. "For US males, life expectancy at birth increased by 5.5 years from 1980 to 2006, the equivalent of 2.04 years per decade. While this is a significant achievement, it is less than the average increase for the other 21 countries examined for this study. Similarly, between 1980 and 2006, life expectancy at birth
for US women increased from 77.5 to 80.7 years, only slightly more than 60 percent of what was achieved, on average, in the same period in the other 21 countries examined."

In turn, it traces these differences back to death rates for lung cancer, heart disease, and stroke. In turn, this is traced back to international differences in smoking behavior in decades past. "Fifty years ago, smoking was much more widespread in the United States than in Europe or Japan: a greater proportion of Americans smoked and smoked more intensively than was the case in other countries. The health consequences of this behavior are still playing out in today’s mortality rates." The report also discusses diet and obesity.

All this is true enough, and intriguing. But there are reasons for differences in mortality across countries that don't trace back to smoking and diet. One interesting comparison from a few years ago by Robert L. Ohsfeldt and John E. Schneider, which appears in their 2006 book The Business of Health, compares actual life expectancy rates across countries to a "standardized rate" that is calculated after taking out fatal injuries due to causes like driving deaths and murder (using OECD data). I was surprised to see that if you leave out fatal injuries, it turns out that that average U.S. life expectancy vaults from near the bottom of the list of high-income countries up to the top. 

To be  fair, the NAS report is more focused on gains to life expectancy for those over the age of 50, and deaths in motor vehicles and from murder typically affect younger people. Still, in all the discussions that do use overall life expectancy numbers (not just life expectancies at age 50), it felt like an oversight to me that these causes of violent death don't come up in the NAS report.