At the risk of over-hoisting from an important piece of analysis, below are some of the key sections from Anusar Farooqui, who writes as Policy Tensor, on the extent of the “death of despair” catastrophe. His piece, Yes, High-School Graduates Are Dying of Despair, is important because he demonstrates that the rising death rates over time extend even into those with some college education.
Farooqu got in an argument with Matt Yglesias on Twitter over the body of work by Ann Case and Angus Deaton on so-called deaths of despair, in which they found what they called an AIDS-level surge of mortality among less-educated whites in middle age. Not only have Case and Deaton refined their analysis, but other studies have identified an increase in early deaths from suicide and addiction in other groups.
For background, from our first post in 2015 on Case-Deaton findings:
The authors found that from 1999 to 2013, the death rate among non-Hispanic whites aged 45 to 54 with a high school education or less rose, while it fell in other age and ethnic groups. This is an HIV-level silent epidemic: AIDS killed an estimated 650,000 from the mid-1980s to present, while an estimated close to half-million died in half that time period who would have lived had their mortality rates fallen in line with the rest of the population. It is hard to overstate the significance of these findings. From the New York Times:
“It is difficult to find modern settings with survival losses of this magnitude,” wrote two Dartmouth economists, Ellen Meara and Jonathan S. Skinner, in a commentary to the Deaton-Case analysis that was published in Proceedings of the National Academy of Sciences.
This cross-country comparison from the study shows how extreme an outlier these middle aged whites are:
The big culprits are linked to despair, namely “poisoning” which is opioid abuse first and alcoholism second, and suicides. Case and Deaton dug into the underlying statistics, and found distressingly high levels of pain and impaired health in this age group, so pain and physical impairment may well be bigger culprits than economic distress:
And the rise in death rates took place among men and women, in all of the four major regions of the country the authors examined, and obesity rates were not a driving factor.
These pathologies have been showing up in other demographics. For instance, the Wall Street Journal reported in May 2023 that death rates in the 1 to 19 year old age group had risen at an unprecedented rate from 2019 to 2021 due to guns, suicides, car accidents, and drugs.
Now to the immediate discussion. Farooqui took issue with a claim by Matt Yglesias, bolstered by Eric Levitz, that the Case-Deaton data was less significant that it might seem because it lumped together those with no high school degrees with high school diplomas. Yglesias and Levitz argued that the “deaths of despair” were only taking place among high school dropouts and so the white working class was not in as terrible shape as it might seem.
Farooqui pointed out that Case and Deanton had shown that death rates among the middle aged were rising, albeit relatively modestly, even among those with some college education. But then he turns to the crux of the disagreement:
However, they [Case and Deaton] do not distinguish between [less than] HS and HS, and therefore do not address the specific issue highlighted by Levitz: “… it is actually an acute crisis of mortality of the bottom 10%.”
This is an empirical question. We attack this problem using CDC data collated by Wharton. They have age-specific mortality rates by educational attainment. The ordinal categories are
In order to obtain a more representative graph, we use age-specific population weights to combine age-specific hazard ratios. The next graph displays the population-weighted average of hazard ratios for our age-specific cohorts. By construction, this weighted average of hazard ratios is not confounded by any increase in average age within discrete age brackets. And, as explained above, because we’re looking at hazard ratios, it is also not confounded by the common component. This is as kosher as it gets in this business.
We can see that the upward trend in hazard ratios is not confined to high-school dropouts. It is true that the trend is most pronounced for them. But the upward trend is also significant for high-school graduates and those with some college under their belt. The all-cause hazard ratio has increased by 3.28 for high-school dropouts, 2.08 for high school graduates, and 1.27 for those with some college. The upshot is that, on the wrong side of the diploma divide, despair goes very far up indeed.
And then he drives the point home (I’ve omitted the charts in this section, but Farooqui showed he has the goods):
All-cause mortality has a very strong signal. But the evidence for American working class despair is not confined to mortality. Prime-age labor force participation contains the same signal of despair: four out of every nine Americans with only a high-school diploma are not even looking for a job. It’s not like high-school graduates can survive on one pay-check! These are obviously discouraged workers in despair.
Take family reproduction—perhaps the most important factor in human well-being. Following the Sixties revolution in behavioral norms, the rate of family reproduction stabilized for college graduates by the 1990s. But it has continued to fall for high school graduates. See my note from three years ago. And if you’re interested a deeper dive, see Andrew Cherlin’s work.
Or, take divorce rates. For college graduates, 29.7 percent of first marriages end in divorce by age 46. For high-school graduates, 48.2 percent of first marriages end in divorce. (The table’s from here.)
We can keep going. The truth is that American working-class despair is such a robust, large-scale pattern that one can recover the signal from practically anything we care to measure—mortality, morbidity, BMI, labor force participation, marriage rates, child-out-of-wedlock rates, you name it. So, I stand by my challenge to Matthew Yglesias.
Well-off Americans are so removed from these cohorts, usually encountering them in various service, as in servile, positions where they have to put on a good face as a condition of getting paid, that they can pretend that the lesser orders aren’t suffering in financial and emotional terms. One proof is despite Case’s and Deaton’s landmark work, there’s no interest in what to do to alleviate this personal and collective disaster. Our version of “Let them eat cake” has gone from “let them learn to code” to “Deprogram them all”.