Rising inequality is not balanced by intergenerational mobility

The United States currently exhibits more economic inequality than any peer nation, and surveys of US adults support the idea that inequality is acceptable if it is balanced by mobility. Many are untroubled if doctors make 10 or 20 times what janitors make, as long as janitors’ sons have opportunities to become doctors. In an era of rising income and wealth inequality in the United States since the 1970s, that balance of inequality and mobility grows in salience. Enter Song et al.’s paper, “Long-term decline in intergenerational mobility in the United States since the 1850s” (1), which uses linked household and population records on the occupations of generations of US-born white men, along with data from several representative surveys, to describe how social mobility in the United States has changed since before the Civil War and before industrialization transformed economic production. Comparing the occupations of sons to the occupations of their fathers, Song et al. (1) paint a troubling picture of rising intergenerational persistence in occupational status. One’s social class of origin—the class one is born into—has become “stickier” since 1850. That is, sons’ occupations are increasingly predictable from fathers’ occupations. The headline finding is that sons born after 1940—the Baby Boomers, Gen Xers, and Millennials of today—are significantly less likely to surpass their fathers in occupational attainment. Fewer janitors’ sons are becoming doctors today.

Economists, including Raj Chetty, have drawn increased attention to social mobility, a classical sociological concern. Max Weber, a founder of European sociology, conceptualized “life chances” as the set of opportunities faced by people depending upon their current position in society, a position constituted by one’s relationship to the labor market, one’s “status” or social rank, and one’s political power. In the postwar period, sociologists used census data, new nationally representative surveys, and new quantitative methods to distinguish 2 kinds of mobility: exchange mobility and structural mobility. Exchange mobility can be compared to the reshuffling of a deck of cards, where some cards that began at the bottom of the deck get shuffled to the top, and vice versa. Exchange mobility represents a kind of outcome, an absolute upward or downward move. If we allow the deck of cards to grow and change in composition as it is shuffled, we will also encounter structural mobility, which is like adding many hearts to the deck, adding a bunch of jacks to the deck, or both. In the case of structural mobility, movement is relative to opportunity: We will find more jacks at the top of the deck if we add jacks disproportionately as we shuffle. This represents one way of thinking about social mobility during industrialization, a period that saw tremendous upward structural mobility as agriculture shrank, and manufacturing grew, as a percentage of the workforce.

Song et al. (1) unpack this inherent complexity by differentiating between relative mobility and absolute mobility. This is a key contribution of their paper compared to previous work on mobility that cannot distinguish structural from exchange mobility in estimating, for instance, the association between father’s income quintile and son’s income quintile. Song et al. (1) instead measure relative mobility as the association between son’s percentile rank and father’s percentile rank, to capture the fact that the son of a farmer who becomes a salesman is relatively more upwardly mobile in the distribution of occupations if that distribution is not already top heavy with many occupations that outrank salesmen. Conversely, that same salesman son of a farmer is less upwardly mobile, by this measure, if salesmen have been bumped down the status ladder by higher-status occupations. Song et al.’s (1) figure 1 helps to visualize this: Compare the relative positions of salesmen in the 1860 vs. 1890 cohorts.

In contrast, Song et al. (1) measure absolute mobility as the association between son’s occupational status and father’s occupational status. Here, the salesman son of a farmer would be upwardly mobile regardless of how many occupations rank above salesman. The important thing for absolute mobility in this case is that salesmen outrank farmers in the father’s birth cohort as well as the son’s. Both relative and absolute mobility matter because both opportunities and outcomes matter: Sons doing better than their fathers is part of the American Dream, but so is the continuous upgrading of destination occupations (Table 1). If fathers in top-ranking occupations are increasingly likely to monopolize opportunities for their sons, then high-ranking occupations would continue to displace lower ones, such that “moving on up” is not really moving up at all and can even mean moving downward in terms of status-based standards of living or social honor. The sort of downward mobility experienced by clerical workers who are sons of clerical workers (1) would probably feel particularly unjust, because it is “involuntary mobility” that reflects a more closed opportunity structure. This involuntary downward mobility would seem likely to foster a social class variant of the racial resentment reported by Hochschild (2), who uncovers a “cutting in line” narrative that makes sense of working- and middle-class whites’ resentment of nonwhites and the Obama presidency. In this way, Song et al.’s (1) contribution sheds light on current US politics in a way that the well-known trend of declining real wages among all but the highest paid does not. People evaluate their fortunes relative to their hopes and expectations, formed, for many, as a sense of what is normal or abnormal, possible or impossible, in childhood (3).

Table 1.

Relating opportunities and outcomes to mobility concepts and measures

	Opportunity	Outcome
Structure or exchange	Structural mobility	Exchange mobility
Absolute or relative	Relative mobility	Absolute mobility
Measure	Percentile rank	Occupational rank
Metaphor	Involuntary mobility	Shuffling the deck

The use of data from many time points between 1850 and 2015—at least one wave of data per decade—represents another distinctive feature of the Song et al. (1) contribution. They use significantly more data than previous studies, including a cross-validation using no fewer than 9 national surveys, each of which having been used in separate studies of mobility over 50+ y of scholarship on social mobility. The timescale and resolution are particularly important for the question of whether mobility is rising, stable, or declining, because the data are best interpreted in historical context. Only at this scale and resolution can it be shown that short-term changes in mobility are relatively rare across this history, with 1870 to 1940 standing out as a time of remarkable stasis. This makes the reduction in social mobility observed among the post-1940 birth cohorts all of the more troubling. Additionally, because of the scale and resolution of the data, Song et al. (1) are able to show that mobility—in contrast to inequality, as shown by Piketty (4)—was impervious to 2 world wars and the Great Depression. Finally, long-run historical comparison helps to anchor the estimates in a way that addresses Swift’s (5) valuable critique of “perfect mobility” as an inappropriate baseline for comparison in modeling mobility; seeing the historical comparison helps to evaluate the level of mobility in an historically comparative way, rather than in a normative way relative to some artificial referent.

Even more important, perhaps, than the overall trend in social mobility is the decomposition of that trend into the percentage of US-born white males who were upwardly mobile, downwardly mobile, or immobile over this long sweep of history. The rate of immobility fell in a fairly linear fashion from the 1830 birth cohort to the 1980 birth cohort, from about 40% of sons destined for their father’s occupation to about 25%. The rates of upward and downward mobility bounced around more, but there is a troubling trend in the cohorts born in 1940 and afterward. The upward mobility rate fell from around 60% to about 42%, while the downward mobility rate grew from around 17% to about 32%. These men would have also experienced a labor market with growing income inequality, as the 1940 birth cohort reached prime working age around the same time income inequality started ramping up in the 1970s. Not only were these cohorts experiencing more downward mobility; they were also experiencing relative and absolute income declines. There is little evidence of a direct connection between income inequality and income mobility (6), but the lived experience of growing inequality and declining mobility must have been problematic for these men.

At the same time, these men were increasingly likely to live in a household with a woman who was also attached to the paid labor market, which raises a major limitation of this excellent paper (1): It tells us nothing about the daughters of these fathers or the sisters or mothers of these sons. This limitation grows in impact with each successive year, especially between 1960 and 1990, when the percentage of all women active in the paid labor force rose from about 37% to about 57%. The same is true for the racial and ethnic composition of the labor market. White males were 42% of the US population in 1900, and that percentage has steadily declined, to about 25% today. This means that the story told by Song et al. (1) captures less and less of the reality of social mobility in the United States. It also means we miss the crucial role that race and ethnicity play in social mobility in the United States, a role Pfeffer and Killewald (ref. 7, p. 1) vividly visualize in showing how “intergenerational fluctuation coexists with the maintenance of a severely racialized wealth structure.”

This new paper by Song et al. (1) should spark new research that addresses its limitations. An overarching limitation, related to the points about population composition discussed above, is the absence of explanation. While Song et al. (1) are correct in arguing that science should get the empirical facts straight before trying to explain them, it seems there is now more than enough evidence of stable or declining occupational mobility to push the field past the 1975 Featherman–Jones–Hauser hypothesis. The distribution of power in the labor market would seem to be a very promising direction for theory (8). In pursuing explanations of the long-term trend toward less mobility, which implies opportunity hoarding (9) at the top of the social hierarchy in the United States, we must also remember that single-country analyses, even analyses of countries as large as the United States, remain case studies in that they “hold constant” a number of unmeasured variables. Measuring those macroscopic variables can yield explanatory insights that case studies cannot (10). For instance, the explanatory role of institutional arrangements to limit the intergenerational transmission of advantage can be missed if, for instance, the analyst holds constant laws governing land inheritance, a crucial mechanism whereby farmers’ sons become farmers (1). Likewise, farmers’ sons’ opportunities to move “from farms to booming towns and cities to become operative workers, blacksmiths, bricklayers, truck drivers, sales workers…” (ref. 1, p. 3) were surely structured by collective bargaining institutions that shaped those jobs. Thus, it is difficult to conclude that “industrialization is a major—if not the only—determinant of broader trends in social mobility” (ref. 1, p. 5) when only one kind of industrialization is observed.

In the economy, as the United States and peer nations veer toward fixed-term contract labor, deregulated labor markets, more automation, and more offshoring, we would do well to hold the results of this important PNAS paper in mind as we contemplate new rules of the game. In the polity, as voter anger and resentment continue to fuel populist discourse on the right and the left, the revelatory findings of Song et al. (1) offer a vital explanation: People are increasingly falling behind where they probably expected themselves to be.