Has Intergenerational Progress Stalled?

Evidence of regression from IPUMS

I was somewhat surprised to see Eric Levitz arguing that “Generation Z is doing unusually well economically.” An empirical grounding for his argument was provided by a Fed paper from last year. The authors note in the abstract that

At age 36–40, Millennials had a real median household income that was 18 percent higher than that of the previous generation at the same age. This rate of intergenerational progress was slower than that experienced by the Silent Generation (34 percent) and Baby Boomers (27 percent), but similar to that experienced by Generation X (16 percent).

Kevin Corinth and Jeff Larrimore. “Has Intergenerational Progress Stalled? Income Growth Over Five Generations of Americans.” Feb, 2024.

Levitz also cites data from the Survey of Consumer Finances which shows that the median wealth of Americans under 35 in 2022 was the highest on record. He also cites the Economist, which notes that “The typical 25-year-old Gen Z-er has an annual household income of over $40,000, more than 50% above baby-boomers at the same age.”

So, are recent cohorts doing rather well, quite incongruous with the vibes? I decided to investigate the question myself. In what follows we look at an alternate and more comprehensive source of data, IPUMS-USA.

We extract public microdata from the American Community Survey (ACS 5-year, 5% density). We restrict the dataset to non-Hispanic white American citizens to avoid confounding by compositional changes between demographic groups, including through immigration. We have 1,314,715 observations in our dataset, which allows us to obtain high confidence estimates with small standard errors.

We compute population-weighted means of total personal income by 5-year cohort, 5-year age groups, and educational attainment. This means that, unlike the papers cited above, we are able to track progress or regression by class. We present results by 5-year age groups. For each age group, we look at average incomes by birth cohorts and educational attainment categories.

For 20-25 year olds, we find that real incomes have fallen for all educational classes, but they have fallen the most dramatically for those without a college degree. Compared to the 1975-1980 cohort, real incomes have fallen by 18% for those with a college-degree, 48% for those with some college education, 42% for high school graduates, and 54% for those without a high school diploma. The ratio of the incomes of college-educated and high school-educated Americans has increased 41% from 1.34 to 1.89.

It is possible that increasingly delayed entrance into the labor market may exacerbate the decline seen above for those in their early-20s. And indeed, we find less pronounced declines for 25-30 year olds.

Compared to college grads born in the early-1970s, those born in the late-1990s had 13% lower incomes; those with some college saw a decline of 25%; high school grads saw their incomes erode by 29%; and those without a high school diploma had 42% lower incomes. The ratio of the incomes of 25-30 year old college-educated and high school-educated Americans has increased 22% from 1.68 to 2.05.

Moving on to 30-35 year olds, we find declines of 5% (BA+), 13% (some college), 33% (HS), and 32% (<HS). The BA-HS ratio has risen 42% from 2.02 to 2.88. For this crucial prime age group, the diploma divide in real incomes has increased from a factor of roughly two to three.

For 35-40 year olds, we find that real incomes have fallen by 4% for BA holders, 16% for those with some college under their belt, 15% for those with a high school diploma, and 19% for those without. The BA-HS ratio has risen 12% from 2.20 to 2.40.

For 40-45 year olds, we find similar declines: 7% for BA, 14% for some college, 12% for HS, and 11% for <HS. The BA-HS ratio has increased 5% from 2.38 to 2.48.

Millennials will start hitting 45 this year, so higher age brackets are not relevant to the discussion. But you can find all our estimates along with their standard errors in this spreadsheet.

For a rough ballpark, we suggest that the 30-35 year-olds prime age group is a good reference for the discussion, since it is not confounded by delays in labor market participation. Recall that we found declines of 5% for BA+, 13% for some college, and 32-33% for HS or less.

Millennial 30-35 year olds with a high school diploma who were born in the early-1990s were making less than $30,000 per year, while their predecessors who were both in the 1960s and 1970s were making more than $40,000 per year. By any reasonable definition, this is a very large regression. College-educated Millennials are not do so hot either, but their absolute downward mobility pales in comparison.

And it is not just absolute regression either. The working class has also regressed relative to the college-educated class. As documented above, the college premium in real incomes has increased for this reference age group from +100% to +188%.

The incongruence of the above with the results presented in the 2024 Fed paper by Corinth and Larrimore may be due to the fact that they do not condition down on non-Hispanic whites. And, of course, they do not break it down by educational attainment. At the very least, the incongruence suggests that this question deserves further investigation. What we would suggest is that researchers pay attention to the divergence between the classes, especially across the Great Diploma Divide.