Education Now Ahead of Race in Determining Life Expectancy. NewsWire

NewsWire: 4/5/21

• According to a new study, the life expectancy gap between Americans with and without a college degree more than doubled between 1990 and 2018. Over the same period, racial disparities in life expectancy shrunk by 70%. (Proceedings of the National Academy of Sciences)
• NH: The married-couple team of Angus Deaton and Anne Case (the former an economics Nobel laureate at Princeton) have written often about economic inequality and health. "Deaths of despair," their most famous coinage, is the term they use to describe the growing mortality rate among low-income, midlife, and mostly white Americans over the past couple of decades due to alcohol, drugs, depression, and suicide.
  
• Along the way, the Deaton-Case team has shed new light on the connection between two seemingly unrelated demographic trends.
• The first trend is the decline in the life expectancy at birth of the entire U.S. population since 2014--and especially the dominant role in that decline played by rising mortality in nonelderly adults (age 25-65). The age-adjusted mortality rate of nonelderly Americans has been rising since 2010. (The data are through 2019, but obviously they're only going to get worse in 2020 and 2021.) See "Adults Under Age 65 Driving Decline in U.S. Life Expectancy."

• The second trend is the strong and growing correlation between affluence and life expectancy. According to a recent NAS study, for example, the rich-poor gap in life expectancy between Americans born in 1930 (early-wave Silent, today age 90) and Americans born in 1960 (late-wave Boomers, today age 60) has grown dramatically. In the earlier birth cohort, the most affluent quintile could expect to live 5.1 years longer at age 50 than the least affluent quintile. In the later birth cohort, it could expect to live 12.7 years longer.

• If we put these two trends together, they point in a provocative direction. They suggest that at least part of what's driving the steeper rise in mortality among younger cohorts could be a steeper rise in socioeconomic inequality. Unfortunately, the SAS data on income quartiles and longevity is highly inferential and cannot be generated each year or for separate age groups. From coroners' reports, we know the age of death of pretty much everyone who dies, but we don't have a very accurate measure of their lifetime income or wealth.
• Here's where Deaton and Case come up with a new approach. They point out that, in addition to age, we actually do have a very accurate record of the race/ethnicity and educational attainment of the vast majority of fatalities. So the authors are able to generate an accurate, population-wide estimate of U.S. mortality rates by age that include these variables. To exclude the impact of mortality improvement at age 75 or over, Deaton and Case choose to measure expected years of life at age 25 to age 75. BTW, the life-table code for this is ₅₀e_25, or the expected years of life, up to age 75, after reaching age 25. Thus, the maximum value is 50.
• Going into this research, Deaton and Case hypothesize that whereas thirty years ago race (white versus black) was the most important indicator of socioeconomic difference, today educational attainment (in particular, a four-year college degree) is the most important. They observe that back in 1990, the earnings premium for a B.A. degree over high school could not have been much more than 40%. But by 2019 it had reached 86%.
• Well, the results turn out to be pretty much in line with the authors' expectations. To begin with, let's look at expected years of life (age 25 to 75) from 1990 to 2018 for the entire population. And then let's look at it for the college and the noncollege groups separately.

• As we can see, there's a widening years-of-life divergence between the college and noncollege groups. In 1990, they were 1.9 years apart. In 2018, they were 3.1 years apart. What's more, while the college years-of-life continues to rise in every year, the noncollege years-of-life sinks rapidly after 2010. This noncollege group, which comprised 79% of 25-to-75-year-olds in 1990 and 65% in 2018, pulled the whole population down with them.
• Now let's look at the results by race and ethnicity.

• Here, as expected, we see a declining divergence--or convergence--between whites and blacks. In 1990, they were 3.6 years apart. In 2018, they were 1.6 years apart. We have reported on this convergence before. (See "The Gap in Life Expectancy Between Blacks and Whites Has Narrowed.") The results also highlight the rising Hispanic advantage in life-expectancy over both whites and blacks. This "Hispanic exceptionalism" has attracted much interest from health researchers.
• Now let's put all the pieces together and segment the population by both race and education. These results are striking.

• Note that whites and blacks are rapidly converging with each other, among both men and women, when we look at each group at the same educational level. That is, both whites and blacks with 4-year college degrees are climbing and moving closer together. And both whites and blacks without college are falling and moving closer together. The gap separating races within each educational level (1.0 years) is less than one third of the gap separating educational levels within each race (3.6 years).
• The Hispanic years-of-life are not shown in this chart. But elsewhere the authors show that college Hispanics' life expectancy is only slightly higher than that of college whites. Nearly the entire advantage in overall Hispanic life expectancy is in the much-higher years-of-life of noncollege Hispanics over both noncollege whites and noncollege blacks.
• How do Deaton and Case explain these results? Their discussion points causally to the rising income and wealth gap between college and noncollege Americans. This negatively affects the health of the noncollege group not just through the unaffordability of such things as health insurance and safe neighborhoods, but more importantly (they argue) through loss of professional dignity and social esteem. They quote political philosopher Michael Sandel, to the effect that a college degree is now "a condition of dignified work and social esteem." Noncollege Americans work at less stable jobs, they have fewer opportunities for advancement, they marry less and divorce more, they experience more stress, and they live in less cohesive communities. These are all associated with the higher rates of morbidity and disability and the "deaths of despair" chronicled by the authors.
• That is one possible explanation. But IMO there could be others. For example, it could be that causality works in the opposite direction. Failure to complete college is not a life path that causes poorer health (through lack of income or prestige), but rather it could be the path that is typically chosen by those Americans who have already given up on achievement for unrelated reasons.
• Let's assume, in this case, that changing family and social norms cause a larger share of young whites and blacks (males, especially) to lose the initiative to prepare for college or the perseverance to finish their degree once they have started. That loss would be the driver of later poorer health. And the lack of a college degree would be highly correlated with that loss but only incidental to it. Charles Murray in Coming Apart would seem to support this line of argument.
• This brings us to Case and Deaton's own puzzling counter-example: Hispanics. Despite much higher poverty rates and lower insurance coverage rates, noncollege Hispanics actually have higher expected years-of-life before 75 than all whites (including those with college degrees).
• Why are Hispanics exceptional? We really don't know. Some researchers suspect genetic differences. But most believe the reasons derive from beneficial habits of diet, marriage, family, entrepreneurship, and community--a cultural mix that maintains health despite a lack of educational credentials. The Case-Deaton discussion would benefit by taking their example more seriously. Growing earnings inequality among younger-born cohorts may be a driver. But it's probably not the whole story.
• Educational attainment has long fascinated demographers because it shows up so persistently and prominently in regression analysis. Countless studies have shown that education is highly correlated with life expectancy even after controlling for other variables (like income, health risks, insurance, and so on) that are often regarded as "causal." This is especially true in America. But it is not true for all subcultures within America. Which may tell us that we're still not certain how the causation works.

To view and search all NewsWires, reports, videos, and podcasts, visit Demography World. For help making full use of our archives, see this short tutorial.