Fewer Americans are making more than their parents did—especially if they grew up in the middle class

Peer 1 client Robert A. plays hopscotch with his 5-year-old daughter Isabella during Kid's Day at Peer 1 in Denver November 7, 2015. Peer 1 is a drug and alcohol rehabilitation programme for men, many having spent years in and out of prison, in Denver, Colorado. The men have often tried and failed over and over to turn their life around. With histories of abuse as children and living on the streets, they come to Peer 1 hoping to turn away from addiction and crime, to rebuild their lives and learn how to integrate into society. Treatment includes family group therapy, meditation and trust-building exercises.  REUTERS/Rick WilkingPICTURE 25 OF 26 FOR WIDER IMAGE STORY "PEER 1: PRISON, DEATH OR RECOVERY". SEARCH "PEER WILKING" FOR ALL IMAGES - GF20000082782

One of the most striking social science findings of recent years is that only half of today’s 30-year-olds earn more than their parents. Raj Chetty and his coauthors showed that rates of absolute mobility—that is, the share of children with higher inflation-adjusted incomes than their parents—declined from around 90 percent for children born in 1940 to just 50 percent for those born in 1984. Just in case you’ve been snoozing for the last year and half, here’s the chart:

Surprisingly, the bulk of the decline between the 1940 and 1980 cohorts was concentrated toward the top of the income distribution. As Aparna Mathur and Cody Kallen of AEI wrote in “Poor rich kids?”: “[P]erhaps the most puzzling—and least commented upon—finding is the large positive correlation between the parent’s income and the decline in absolute mobility over the years. Put more simply, the richer the parents, the larger has been the decline in mobility for their kids.”

But the 1940 cohort may be unusual for many reasons, most notably the influence of World War II and the postwar economic boom. If we instead begin with the 1950 cohort, then it appears that absolute mobility has fallen most for children with parents in the middle class. Comparisons to the 1960 and 1970 cohorts reveal a similar story of middle-class decline.

Here we dig a little deeper into the rates of absolute mobility for those born on different rungs of the income ladder, and in different birth cohorts. The big drop in absolute mobility for those at the top of distribution occurs between the 1940 and 1950 birth cohorts. Since then, the middle class has suffered greater losses of absolute mobility than those at the top or bottom.

The Chetty-bomb, revisited

The following chart from the original paper shows the share of children at each percentile of the income distribution who out-earned their parents in the five decadal cohorts between 1940 and 1980:

While absolute mobility declined across the income distribution, those born into the very bottom of the distribution are still highly likely to out-earn their parents. This is virtually a mechanical result: we would expect such a trend given the low point from which they started. Children born into the 5th percentile of their parents’ income distribution in 1980 would only have to make it into the 12th percentile of their own cohort’s distribution to earn more than their parents, according to Online Data Table 7. Likewise, low rates of absolute mobility at the very top are not a surprise. As Chetty et al. explain, “Naturally, rates of absolute mobility were lower at the highest parental income levels, as children have less scope to do better than their parents if their parents had very high incomes.”

What is more surprising is that absolute mobility rates appear to have declined more for those near the top (up to around the 96th percentile) than for those in the middle. This conclusion depends on the choice of a comparison cohort. Here we show the percentage-point decline in absolute mobility for the 1980 cohort compared to each of the other four cohorts from the previous chart:

Only the comparison to the 1940 cohort is skewed toward the top. If we compare the 1980 cohort to the 1950 or 1960 cohorts, then the decline in absolute mobility is more evenly distributed across the wide middle of the distribution. This suggests that the unusually large declines at the top of the distribution took place early in the period. Consider the following comparison of each cohort to the one born one decade prior (i.e. the 1950 cohort to 1940, 1960 to 1950, and so on):

As the chart shows, the decline in absolute mobility between the 1940 and 1950 cohorts was greater higher up the income distribution. But this was not true for later cohort comparisons. At the tails, people born in 1960 were only slightly less upwardly mobile than those born in 1950, and those born in 1970 were actually more upwardly mobile than those born in 1960. The tails of the 1980/1970 comparison are less pronounced, but it seems that those in the lower-middle suffered the greatest declines in absolute mobility between these cohorts.

What’s different about the 1940 cohort?

There remains something of a puzzle in the top percentiles, however. Given that top-heavy losses are unique to comparisons that use 1940 as the base cohort, the key question may not be why so many rich children born in the 1950s and beyond failed to out-earn their parents, but rather why so many rich children born in the 1940s succeeded. This answer, at least, seems fairly simple: Most people born in the 1940s earned more than even some of the highest earners in their parents’ generation due to broad-based economic growth in the post-war years (see Figure 2B of the Chetty et al. paper). Under more normal growth conditions, we might expect the distribution of absolute mobility to look more like it did in 1950, with lower mobility levels at higher ranks of parental income.

Who has dropped fastest?

Who, then, has suffered the greatest decline in absolute mobility: the middle class or the “poor rich kids”? The answer depends crucially on the choice of reference group. Empirically, the biggest drop between decadal cohorts is between those born in 1940 and those in born in 1950—and especially for those at the top of the distribution—with a slower decline since, focused more on those in the middle. In the original paper, Chetty et al. estimated that the main explanation for the mobility decline between the 1940 and 1980 cohorts was increased inequality, followed by slower growth rates; it would be interesting to repeat the exercise for the decline between the 1950 and 1980 cohorts, given the very different distributional stories for the later cohorts. But it seems clear that the absolute mobility trends have very little if anything to do with the pulling away of the top 1% in recent decades.

But psychology matters as much as statistics here. In terms of how mobility impacts our sense of our own status, progress, and condition, the question is which benchmark we select to make that assessment. For many people, mobility does consist of doing better than your parents did, in absolute terms. This seems to have become steadily harder to achieve for those born into middle-class families in particular from 1950 onward. The challenge is to learn from these historical trends in order to secure a better future for the middle class.