Yesterday I was walking along and it occurred to me that some of the horrific change in death rates among 45-54 year American whites since the late 1990s relative to other groups in America and abroad may be due to changes over time in the average age of 45-54 year olds.
Back in 1998, 45-54 years olds were born in 1944 to 1953, but there weren’t all that many people in that cohort who were born in 1944 or 1945 because the Baby Boom didn’t start until 1946. So the average age in 1998 of the 45-54 year old cohort was lower than you would guess if you naively assumed that 10% of its members were born in each year.
By 2013, the birth years of 45-54 year olds were 1959-1968. The birthrate declined pretty steadily in the 1960s with the usual final year of the Baby Boom said to be 1964, although that’s fairly arbitrary. But in any case, it was lower in 1968 than in 1959, so the average age of 45-54 year olds was a little older in 2013 than in 1998.
So, in 2013, the average 45-54 year old was slightly older than the average 45-54 year old in 1998, so it’s not surprising, all else being equal, that they tended to die more.
And sure enough, today, both Andrew Gelman and Philip Cohen post helpful calculations estimating that half of the absolute increase in middle-aged white death rates might be due to this increase in average age among 45-54 year olds.
But these estimates just account for half of the absolute rise in death rates, leaving not only the other half of the absolute rise but the entire opportunity cost of the fall in death rates seen among other middle-aged groups in America and abroad.
Moreover, while the Baby Boomer effect as purely a statistical artifact has some explanatory power, my suggestion that The Sixties! rather than the Baby Boom played a role in higher death rates due to the three main growing causes: overdoses, suicides, and liver problems. (Actor Philip Seymour Hoffman, who died at 46 in 2014, would be pretty representative — for a celebrity — of whites taking the express checkout lane.) Obviously, the Baby Boom and The Sixties! of sex, drugs and rock and roll are interrelated, but we can use the fact that drugs didn’t become a broad scale mass phenomenon in America as soon as the first Baby Boomer turned 18 in 1964, but instead took 5 or 10 years longer to move from elite to mass consumption.
Let’s take another look at Deaton and Case’s graph of five-year cohorts measuring the spectacular growth in death rates for three causes: overdoses, suicides, and liver troubles. This is for 1999-2013 (while other work they’ve done looks at 1998-2013):
When I eyeball the lines, I come up with the following table of five-year cohorts, sorted from oldest to youngest:
For example, the 60 to 64 year old white cohort (dashed purple line) showed an increase in death rate per year from 1999 to 2013 of 1.7, which is pretty bad, but pretty average for this chart. In 1999, the oldest members of the cohort were born in 1935 and the youngest in 1939, so none of them were Baby Boomers. By 2013, all of the 60 to 64 year olds were Baby Boomers, with the oldest being born in 1949 and the youngest born in 1953. These people in 2013, on rough average, had been 18 in 1969. Drugs were becoming more common but remained more an upscale fashion in 1969, with Woodstock that summer often seen as the last example of the previously upscale drug scene, while Altamont that fall came as a warning that the surly masses were getting out of control.
But the big increases in death rates from 1999 to 2013 were seen in the next two younger cohorts, the Red Dash group of 55-59 year olds and Green line group of 50-54 year olds.
In 1999, the Red Dash 55-59 year olds were born between 1940 and 1944 and were 18 around 1960, while in 2013 they born between 1954-1958 and were 18 around 1974. They had a death rate from the Big 3 new killers that went up 2.9 per year.
Even worse were the Green line 50-54 year olds whose death rate went up 3.1 per year. In 1999, they were mostly but not entirely Baby Boomers born in 1945-1949. On average they were 18 in 1965, when you pretty much had to Brian Wilson to be getting heavy into drugs that early in the Sixties. But in 2013 the cohort was comprised of people born in 1959-1963 who were 18 around 1979.
After that the growth in the death rate falls until the youngest group, the 30-34s (which seems worrisome, but I don’t that much about younger people, so I’ll leave it to somebody else to investigate).
So, some of the growth rate in deaths is purely a statistical artifact, but some of it seems tied to rapidly changing cultural norms about drugs and the like during white Americans’ most famously impressionable years.
It could be that I’m all wrong about my surmise of there being a “generational effect” on the death rate from overdoses etc. in the 2000s. But my theory that white people took a lot of drugs in the late 1960s and 1970s and that’s having some kind of medical or behavioral effect on them in the 21st Century doesn’t seem too far-fetched.
UPDATE: Case & Deaton respond in the NYT to Gelman with a new graph:
We shared the Gelman critique with the study’s authors, and Mr. Deaton sent a reply to our colleague Gina Kolata, who had written a front-page article about the study. The data you see in the chart here is the essence of his reply: Breaking down the 45-to-54 age group into single years of age, which should avoid Mr. Gelman’s concern, still shows the same pattern.
“If we want to be more precise about the age range involved, we could say that for all single years of age from 47 to 52, mortality rates are increasing,” wrote Mr. Deaton, the most recent winner of the Nobel Prize in economics. “So the overall increase in mortality is not due to failure to age adjust.”
He added that some of the causes of death that were rising, such as drug overdose, “are not things that people age into,” unlike cancer or heart disease.
“We stick by our results,” he said.
Update: Mr. Gelman has responded on his blog. He accepts that mortality rose for middle-aged whites between 1999 and 2013, he said, but notes that the rise was substantially less after adjusting for age.
Based on his reading of the more detailed Case-Deaton numbers, Mr. Gelman wrote, “mortality rates among non-Hispanic whites aged 45-54 increased by an average of about 4% after controlling for age.” The increase was 12 percent without the age adjustment, suggesting that age bias accounted for about two-thirds of the increase — but did not entirely explain the increase.
Ultimately, both sides of the exchange agree on a fundamental fact: The recent mortality trends for middle-aged whites look significantly worse than they do for many other groups. “Their key claim,” Mr. Gelman writes, “is that death rates among middle-aged non-Hispanic whites in the U.S. slightly increased, even while corresponding death rates in other countries declined by about 30%.”
Looking at Case & Deaton’s new graph, the age range with declining mortality rates (age 45-46) who were 45 or 46 in 1999 were born in 1953-1954, and turned 18 in 1971-72. People who were 45 or 46 in 2013 were born in 1967-1968 and turned 18 in 1985-1986.
In contrast, look at age 50, the worst increase in death rates: somebody who was 50 in 1999 was 18 in 1967 while somebody who was 50 in 2013 was 18 in 1981.
But what’s the deal about people who were 18 in 1964?
Gelman adds his own new graph with an age adjustment:
With Gelman’s age-adjustment, he finds a sharp increase in death rates for 45-54 year old white men from 1999 to 2005, with stabilization after that. (I’m not sure why he looks at white men rather than whites of both sexes).
In other words, this very bad stuff happened roughly 10 to 16 years ago, but nobody much noticed it until 2015, which says a lot about how much our society is looking out in an organized fashion for the interests of whites qua whites.