Deirdre Bloome (Harvard University) provided an introduction to the topic of social and economic mobility. She began with a definition: “social and economic mobility capture the distance people move between their position in an origin resource distribution and a destination resource distribution, to help us understand society’s openness.” For example, she said, if affluent children become affluent adults, and poor children become poor adults, then income mobility across generations is low. Mobility has
consequences for both individuals and society, said Bloome. An individual’s movement out of poverty can impact a number of outcomes, from how many children they have to who they vote for. On the societal level, she stated, more mobility can: increase economic efficiency, innovation, and growth; shift who has power in society and reduce abuses of power; and increase the likelihood that people have the same prospects regardless of whether they were born rich or poor.
Bloome walked workshop participants through the concepts in her definition of social and economic mobility. A resource distribution refers to how markers of social and economic advantage are distributed among individuals and communities. Markers include income, educational attainment, earnings, occupation, and wealth; Bloome noted that these markers are related to one another but are not interchangeable. For example, workers with more education may be able to obtain a higher wage, which in turn may allow them to accumulate wealth. However, Bloom stated, in some situations, these markers do not rise and fall together. For example, groups with low labor force attachment (such as women in many societies) may have high earnings mobility but low income mobility if the way they replicate their advantages is through marriage more than employment. Since these markers are not interchangeable, she said, it is critical that data infrastructures for mobility studies include multiple resource measures.
Next, Bloome explained the meanings of origin and destination in the definition of social and economic mobility. The meanings of these words, she said, depends on whether they are applied to intragenerational mobility or intergenerational mobility. For intragenerational mobility, which captures change over the course of an individual’s working life, origin refers to early career positions while destination refers to later career positions. For intergenerational mobility, origin refers to the parents’ positions, and destination refers to the offspring’s positions; multigenerational research further considers grandparents’ and great-grandparents’ positions as origins. Intra- and intergenerational mobility are related, said Bloome, but studies of the two are not exchangeable. The data demands are particularly high for studies on intergenerational mobility, which often require capturing long-term social and economic well-being with multiple years of data observed prospectively across decades.
Bloome used an analogy of a ladder to elucidate the difference between mobility, inequality, and poverty. Mobility is measured across time, either an individual’s lifetime or across generations. Inequality and poverty, on the other hand, are measured as snapshots at a single point in time. Mobility is movement between the rungs of a resource ladder. Inequality describes the length of the resource ladder—longer ladders indicate more inequality (greater distance between the ladders’ rungs) while shorter ladders indicate less inequality, with poverty represented by positions toward the bottom
of the ladder. For example, said Bloome, a comparison of poverty between 1990 and 2020 would show whether the share of people living in poverty increased or decreased during this time. However, only a study of mobility would be able to determine whether the people who were poor in 1990 remained poor in 2020.
Bloome described two ways to measure mobility (the distance a person moves on the resource ladder). The first is absolute mobility, in which the destination resource level is compared with the origin resource level. The second is relative mobility, in which the destination resource level is compared with the origin resource rank; these ranks depend on whether and in which direction a person’s peers are moving. For example, Bloome said, a person could experience upward absolute mobility and downward relative mobility if their income rose more slowly than the income of their peers. These two measurements provide “fundamentally different insights into the persistence of advantages and disadvantages,” said Bloome. Relative mobility has traditionally been considered an indicator of equality of opportunity, because of its comparative nature. When relative mobility rises, she explained, people from low and high origin positions become increasingly equal in their chances of obtaining positions toward the top of the destination ladder. Absolute mobility, in contrast, has traditionally been considered an indicator of the extent to which economic growth is widely shared. In theory, everyone can experience upward absolute mobility if strong economic growth is distributed across the population. For relative mobility, in contrast, Bloome said, “for every move up the rankings, someone must move down.” She listed some of the types of statistics that are used to measure absolute and relative mobility. For example, absolute mobility can be measured by the percentage of people who are upwardly mobile, or the typical gain/loss between origin and destination. Bloome explained that relative mobility can be measured by the percentage of people remaining in their origin quintile or occupational class, or through correlation between origin and destination positions.
With these definitions in mind, Bloome turned to consider how these measures help one understand a society’s openness. High mobility in a society is an indirect indicator of high equality of opportunity. She explained that mobility is an indirect indicator of opportunity because, while origin and destination positions can be observed, opportunity is unobserved. Opportunity captures the destination distributions that might be possible for people from a given origin position, but equal opportunities are not expected to generate perfect mobility. Since some association between origin and destination positions is expected, some theorists distinguish between acceptable and unacceptable obstacles to opportunity. While this distinction can be useful, said Bloome, it is fraught with questions that cannot be answered empirically. For example, how should the freedom of parents to
invest in their children’s achievement (which is a freedom that leads to low mobility) be weighted against the freedom of children to not be limited by their parents’ resources?
Bloome said there are three types of questions that mobility studies can answer: causal, predictive, and descriptive questions. Causal analysis can help one understand the processes that generate and undermine mobility, which in turn helps one understand how to disrupt these processes. However, she noted, mobility-generating processes are complex; change in one area may be undone by change in another area, and pathways stretch across multiple life stages and institutional levels. Moreover, interventions designed to alter mobility may not increase relative mobility if the intervention’s impact is equal across the origin distribution.
Mobility can also be used to answer predictive questions, said Bloome, such as what will happen in the future given a certain origin position. Answering these questions requires good models of destination resource distribution at each origin position, she said, and early indicators (e.g., children’s test scores) may be surprisingly uninformative about mobility because of the inherent complexity and multidimensionality of the pathways between origin and destination.
Using mobility studies to answer descriptive questions—such as how mobility in the United States compares to mobility in other countries—can “provide crucial stylized facts that help us understand our world and make decisions about how to act.” Bloome noted that while people may have different opinions about what to do with this information, it is critical that they start from a shared understanding of the “ground truth” that can only be provided by careful and accurate descriptive studies. In order for data to be useful for descriptive studies, said Bloome, there is a need for data at the individual levels and data that includes all members of the population of interest. For example, if a study aims to describe people’s destinations based on their origins, it should include people whose origins are relatively hard to measure, such as immigrants.
Bloome closed by suggesting that empirical work “must not shy away” from relative mobility, where movement up the ladder requires some privileged people to move down. Furthermore, she emphasized, empirical work must focus “with absolute vigilance” on providing accurate descriptive representations of the mobility experiences of all population members, including the most underprivileged members of society, who are often missing or underrepresented in survey and administrative data.
David Grusky (Stanford University) described five key developments in the field of sociology that have led to new findings and contributions in
the study of mobility. First, he said, sociology is no longer laser-focused on class mobility, but instead has undertaken a more expansive examination of the many types of mobility. Grusky gave several examples of sociological research that demonstrates this expanded outlook. Paying more attention to absolute mobility than they did in the past, researchers have found a dramatic decline in absolute socioeconomic mobility, driven by declines in middle-status production occupations and an increase in low-status service occupations.1 Sociological researchers have begun studying educational mobility in its own right, said Grusky, and found a U-turn in educational mobility that mimics the U-turn in income inequality.2 While income and earnings mobility have long been treated as the “province of economists,” sociologists have recently conducted research on topics such as earnings elasticity3 and building a measurement infrastructure for wealth mobility.4
The second development in sociology, said Grusky, is a return to the field’s attention to the demographic foundations of mobility. For example, Hout found a relationship between mobility and family structure,5 and Bloome found that mobility increases as children transition more frequently into new family types (e.g., via divorce and remarriage).6
The third development is a shift away from descriptive analysis and toward causal analysis, said Grusky, with sociologists examining the causal effects of factors such as college and neighborhood. For example, Wodtke and colleagues found a strong effect of exposure to a low-amenity neighborhood during adolescence on high school graduation; this effect was particularly strong for poor families.7 This is an area, said Grusky, where
1 Hout, M. 2018. Americans’ occupational status reflects the status of both of their parents. Proceedings of the National Academy of Sciences, 115(38), 9527-9532. https://doi.org/10.1073/pnas.1802508115
3 Mitnik, P., and Grusky, D. 2020. The intergenerational elasticity of what? The case for redefining the workhorse measure of economic mobility. Sociological Methodology, 50(1), 47-95.
5 Hout, M. 2018. Americans’ occupational status reflects the status of both of their parents. Proceedings of the National Academy of Sciences, 115(38), 9527-9532. https://doi.org/10.1073/pnas.1802508115
6 Bloome, D. 2017. Childhood family structure and intergenerational income mobility in the United States. Demography, 54(2), 541-569.
7 Wodtke, G.T., Harding, D.J., and Elwert, F. 2016. Neighborhood effect heterogeneity by family income and developmental period. American Journal of Sociology, 121(4), 1168-1222. https://doi.org/10.1086/684137
there is a lot of interest in distinguishing confounding selection effects from true causal effects.
With the emergence of linked administrative data, it has become possible to carry out high-quality trend analyses extending back into the 19th century, said Grusky; this new capacity to understand the long arc of history is the fourth key development in sociology. One example of this type of research is a study that found a sharp decline in mobility after the transition into a non-agricultural economy8; this shows the powerful effect of early industrialization and the exit of children from the agricultural sector on mobility, he said.
The fifth key development has been the rise of qualitative work in the area of mobility. Grusky shared a qualitative study on nutritional disparities in early childhood9 that found that “low-income parents resort to unnourishing food because it is the main affordable treat at their disposal; in effect, it is an attractive way to treat their kids within the context of their budget constraints.” He said that this work is a classic example of unpacking the social, psychological, and familial dynamics behind nutritional outcomes, which are critical for mobility. As further examples of the role of qualitative approaches in social mobility research, Grusky pointed to studies showing that that higher-income “helicopter parents” find ways to ensure that their children are evaluated well during primary and secondary school,10 and that low-income students who attended elite secondary schools are substantially more at home at elite colleges than the low-income students who attended public schools11 and “are left to flounder on their own” at elite colleges. These examples led Grusky to suggest that neither primary schools nor colleges should be regarded as “the great equalizer.”
New Directions for Research
With these trends in sociological research in mind, Grusky turned to identifying new directions for research. His first suggestion was to build an integrated mobility model. Currently, there is a specialized research literature for each type of mobility (e.g., occupation, neighborhood), and while there is research that combines some of these types, the lack of an integrated
8 Song, X., Massey, C.G., Rolf, K.A., Ferrie, J.P., Rothbaum, J.L., and Xie, Y. 2020. Long-term decline in intergenerational mobility in the United States since the 1850s. Proceedings of the National Academy of Sciences, 117(1), 251-258. https://doi.org/10.1073/pnas.1905094116
9 Fielding-Singh, P. 2021. How the Other Half Eats: The Untold Story of Food and Inequality in America. Boston, MA: Little, Brown and Company.
10 Calarco, J.M. 2018. Negotiating Opportunities: How the Middle Class Secures Advantages in School. New York: Oxford University Press.
11 Jack, A.A. 2019. The Privileged Poor. Cambridge, MA: Harvard University Press.
model makes it easy to misinterpret the source of trends. For example, a decline in occupational mobility may be accompanied by a similar trend in earnings mobility; without direct measures of occupation, the source of the decline in earnings mobility may be misattributed. Grusky emphasized that the integrated model should do many things at once: model all types of intergenerational reproduction (e.g., education, occupation, class, earnings, income, neighborhood), integrate intergenerational and intragenerational mobility, include multiple generations, measure marital mobility and individual mobility, and represent assortative mating for origin and destination families. Estimating the model empirically with administrative data will soon be possible, he said.
Relatedly, Grusky’s second suggestion was to build a formal theory for the integrated mobility model, with the goal of specifying how different types of mobility are affected by different types of human capital investments, and by predistributional, redistributional, and safety net reforms and changes. Furthermore, he stated, the model could show how changes at one stage in the model propagate to create changes at subsequent stages.
Another area with ample room for contributions, Grusky said, is examining key inequality trends via intergenerational processes. As an example, he pointed to the need for exploring the role of intergenerational processes in trends such as occupational racial and gender segregation, assortative mating, residential segregation, and life expectancy. There is also a need to integrate qualitative and quantitative mobility research. The typical approach, said Grusky, is to carry out two separate studies side-by-side, with the idea that they inform each other. However, he said there is no reason why these approaches cannot be integrated by carrying out qualitative and quantitative analyses with the same respondents, making it possible to directly address possible inconsistencies between these types of research. Grusky concluded that it is imperative that researchers prepare for the big data storm that will occur when linked Census data from 1960 to the present day become available.
In the study of mobility, economists were “quite late to the party,” said Joseph Ferrie (Northwestern University). The earliest work on mobility came from sociology, and while two early economists dealt with mobility, they did so only incidentally, he said. For Karl Marx, mobility was a problem because it precluded the development of a stable class consciousness. In the 1930s, Frank Knight viewed the advantages that were inherited through families as an impediment to market competition. Even further back, said Ferrie, the writings of Alexis de Toqueville convey a clear sense that there was something “fundamentally different about the American experience”
compared with European countries, as seen in the ease with which families could rise to the top of the economic system, and the ease with which families could fall from “prominence to ignominy.”
Economists began to view aspects of mobility as part of the field of economics in the 1960s, as a consequence of urban unrest and conversations about persistent poverty across generations. The initial focus was narrow, looking at how people could move from the very bottom of the ladder to the “next one or two rungs up.” Even here, said Ferrie, the interest in mobility was largely incidental and did not take a holistic view of the entire distribution or consider how mobility could be impeded or promoted at other levels. This research informed the War on Poverty, he said, but did not provide much in the way of an understanding of the processes of mobility from an economic perspective.
The first research that caused economists to “sit up and take notice,” Ferrie said, was a formal theory of mobility by Becker and Tomes in 1979 and 1986.12,13 This two-period model sought to explain the transmission of advantage across generations, looking at both unconscious parental investment (genes) and conscious parental investment (education). The model estimated an intergenerational elasticity of about 0.2, meaning that over the course of two generations, only about four percent of any advantage or disadvantage persists. This low number was “quite surprising” because it did not comport with people’s “casual impression” of how mobility operated in the United States. The major flaws of this theory, said Ferrie, were its narrow focus on genes and education rather than other factors that can be transmitted across generations (e.g., social connections), and the fact that the model did not take into account that the timing of investments can be as important as the amounts invested. It is now estimated that the intergenerational elasticity is about 0.6 in the United States and above 0.2 everywhere in the world.
More recent developments in economics mobility research include reducing noise in the data by using multiple years of parent and child earnings, conducting analyses across multiple countries, and working with larger datasets. The early datasets that were used—such as the National Longitudinal Survey and the Panel Study of Income Dynamics—were often too small to answer some of the most important questions, especially with regard to mechanisms. Another development, said Ferrie, has been the progress on new metrics that address some of the undesirable properties of
12 Becker, G., and Tomes, N. 1979. An equilibrium theory of the distribution of income and intergenerational mobility. Journal of Political Economy, 87(6), 1153-1189.
the intergenerational elasticity, such as mathematical issues that arise when a person has an income of zero.
In closing, Ferrie identified some of the most interesting questions and areas of current and future research:
- Can intergenerational mobility be accounted for solely by the relationship between parents and children, or do other generations matter? In trying to predict children’s outcomes, for example, additional information can be obtained by looking at the grandparents’ outcomes, even after accounting for those of the parents.
- Does inequality matter? What is the nature of the relationship between inequality and mobility? The “Great Gatsby Curve,” for example, indicates that intergenerational mobility tends to be considerably lower in countries that have more inequality.
- For which outcomes (e.g., income, wealth, education) is persistence (“immobility”) observed?
- Why does mobility vary across geographic locations in the United States?
- To what extent has mobility changed or stayed stable over time?
- What are the mechanisms (e.g., genes vs. environment, specific channels) by which mobility can be explained?
Ferrie noted that early work by economists on mechanisms focused on twin studies and comparing outcomes for monozygotic and dizygotic twins, and that some emerging research is being conducted using new tools such as genome-wide association studies, which allow researchers to look for genetic associations across generations in order to explain some of the correlations. This type of research has a number of downsides, said Ferrie: it requires enormous amounts of data, does not answer the question of mechanisms, and “basically has no policy implications.” However, he said, it may help to narrow down the range of areas in which mobility really does change. While it may have promise, said Ferrie, it is too early to discern whether the promise is being realized.
Following the presentations, Courtney Coile (Wellesley College) moderated a general discussion with speakers and workshop participants. She began by asking the speakers to identify the measures of mobility that should receive the most attention. Bloome responded that “we always have to go back to our question.” The measures researchers use depend on what they are trying to learn about, she said. For example, if the issue is how to improve jobs and the labor market, a focus on occupational and earnings
mobility would be appropriate. If the issue is how to improve children’s well-being, broader measures such as income and wealth might be more relevant. Bloome stressed again that measures are not interchangeable; for example, a person might have a high income but low wealth if they haven’t yet received the returns on their investments. Focusing on only one measure “won’t tell us the full story about mobility,” but it can help focus efforts and intervention points.
Need for Integrated Model
Grusky said that there would be value in research that examines multiple types of mobility at once because it could help elucidate the driving force behind trends; for example, trends in income mobility might actually be driven by trends in occupation mobility. By working toward an integrated model that does not “balkanize the field into separate one-off studies,” researchers could further the understanding of how types of mobility do or do not move in synchrony. Ferrie agreed that this is a promising area for future research, noting that it has been rare to have a way of observing a set of different outcomes within the same body of data. With new, more comprehensive data sources available, it is becoming feasible to examine multiple factors at once and to “feel confident that we’re actually looking at different dimensions of the same problem.”
Coile asked Grusky to comment further on his suggestion that researchers need to prepare for the “big data storm” that is coming. Researchers can prepare now, he said, by building methods to analyze various types of mobility all at once, and developing a formal theory for this integrated model. Bloome seconded the need for this work, saying that having a formal model to build on theoretically is important to help make sense of the empirical evidence that is evolving, and to consider how it might apply in different circumstances and at different times. To this point, she said, these theories and models need to incorporate heterogeneity in associations and causal effects across different times, places, and populations. “We know things will be different,” she said, models should explore how and why things are different.
Using Mobility Research to Predict Outcomes
Mobility research is in some sense “inherently backwards-looking,” said Coile; for example, a study can show how policies that were in place 30 years ago have affected long-term outcomes. In order to impact policy, however, researchers need to know what might happen prospectively if certain actions were taken today. Coile asked Bloome to comment on whether there are short-term outcome measures that might give clues about
long-term outcomes, or if there are other ways to obtain forward-looking information from mobility research. Bloome cautioned against conflating ultimate outcomes with proximate outcomes, and advocated for an accounting of the uncertainty in the pathways between proximate and ultimate outcomes. From a predictive modeling perspective, she said, this means that models of both the variation and the mean, along with uncertainty in the prediction, are needed. For example, while on average affluent children become affluent adults, adult outcomes vary widely depending on factors such as job openings and labor regulations. Bloome also pointed out that researchers can leverage insights from machine learning in order to incorporate knowledge about variation and uncertainty into their predictions.
Timing of Measurements
In the general discussion, Michael Hout14 asked speakers to address the issue of when mobility measures are captured, and how this might have an impact on the data. He noted that some types of measures—for example, educational attainment—usually end earlier in life, whereas other measures—for example, consumption—continue indefinitely. Ferrie agreed that the timing at which mobility is measured is an important consideration for studies, and that there are peaks for different types of measures. For example, income peaks in the late 40s and early 50s, whereas wealth peaks at the end of an individual’s work life. This makes integrating different measures of mobility challenging, because it is very rare to have a dataset that includes multiple measures across multiple time frames. These types of data are becoming more available, said Ferrie, and it is important that researchers determine ways to conduct this type of integrated research. Grusky agreed that newer data sources will make real-time measurements across a variety of types of mobility possible, allowing researchers to get a sense of how mobility processes unfold over the lifetime and to identify and differentiate period, cohort, and age effects. Differentiating these effects from one another is important, added Bloome, because of changes occurring across society; for example, income is peaking later in life because people are getting more education and delaying marriage. Fabian Pfeffer (University of Michigan) concurred with the need to consider the temporal dynamics of mobility measures, both across the life course and across time. Having the ability to integrate measures of mobility across time would “fundamentally shift how we study social mobility.”
14 Chair, Division of Behavioral and Social Sciences and Education; National Academies of Sciences, Engineering, and Medicine.
Prioritization of Mobility Measures
A workshop participant said that the Bureau of Labor Statistics is planning a new cohort for the National Longitudinal Surveys of Youth, and asked speakers to comment on what kinds of mobility measures should be prioritized in new research. The participant said that this is a “great opportunity” to ensure that future data are as useful as possible. Ferrie said he would prioritize measures of health and social capital, in addition to the measures already collected. Bloome said it would be helpful to get information about the partners of the respondents, but that the most important thing is to not lose the ability to compare across surveys.
Two workshop participants brought up the issue of family structure and the importance of parenting. One asked for speakers’ perspectives on how “non-cognitive skills” such as education, expectation, and self-control fit into the mobility picture. The other asked about the rise of single-parent families, its impact on mobility, and whether there are policy implications in this area. Grusky said that a full model of mobility would take into account both investments in cognitive capacities but also in non-cognitive areas. Bloome added that test scores are often used as early indicators of other measures, but they are often not the best predictors of outcomes. Other types of measures—such as non-cognitive skills—are more difficult to measure but may be more useful. Ferrie added that “we are doing a disservice” to parents by not making them aware that there are many ways they can help their children be successful. Even if a child’s test scores are not high, “the child is not doomed to a life of failure.” Rather than focusing on improving test scores, he said, parents should be made aware of additional approaches, such as those in the non-cognitive space. Grusky said that it is important to acknowledge that some institutions “illicitly select on noncognitive assets,” and mentioned the college admissions process as an example. Interventions in this area are generally thought of in terms of generating non-cognitive assets on the supply side, he said, but they could also be directed at the demand side.
Regarding the question about single-parent families, Bloome said that rather than attempting to use policy to change family structure itself, policies should focus on supporting families with resources that will help children have the best opportunities. Family structures are changing dramatically and very quickly, she said, and it is unlikely that this trend will reverse; “people are making different choices and we have to respect those choices,” she said.
Inter Vivos Wealth Transfers
A participant asked for comments on how inter vivos (between living persons) wealth transfers impact mobility, and whether researchers need better ways of measuring this phenomenon. Ferrie agreed that the transfer of resources from one generation to another is hugely impactful on mobility. These transfers tend to occur at pivotal times, he said, such as buying a first home, having children, or starting a business. Help from parents or grandparents at these stages “allows people to take a step up the economic ladder.” It is essential, he said, to learn more about how both parental and grandparental resources impact mobility.
Measures of Subjective Status
A workshop participant raised the topic of subjective mobility, asking whether it was important to measure people’s perceptions of their own mobility and social status. Grusky responded that many measures of mobility presuppose that people care about how they compare with others on a variety of dimensions, such as income or wealth. There are also measures that compare specific groups, such as siblings, peers, or neighbors. However, he said, relatively little is known about which type of mobility comparisons are actually relevant for the subjective assessments that people make. This is an important area for future research, he said, because there is good reason to believe that people’s understanding of where they stand relative to others is important for their subsequent behavior. Bloome agreed with the importance of this research, and noted that survey data over time reveals that subjective standard of living has not been declining over time, despite the fact that occupation and income mobility have been. “This is a very important piece of the puzzle to understand,” she said; disconnects between objective and subjective measures of mobility may have implications for people’s well-being. Hout added that it is difficult for people to answer these subjective questions because it can be challenging to process inconsistencies in one’s own status. While there are correlations between measures such as education, income, and occupation, there is also “an awful lot of play” when people are asked to sum up their overall standing.
Impact of Critical External Events on Mobility
The final question from participants focused on the impact of critical events—such as mass disasters or the COVID-19 pandemic—on mobility. The participant asked, “do we have the right kind of data that we need to understand the impact of these kinds of major events and craft the policy
responses that might be needed to address them?” Bloome responded that measuring this impact requires “very fine-grained” data across time and space, which is becoming increasingly available. Critical events impact people differently, depending on their stage of life. For example, there is likely to be a larger impact on the mobility of people who were 20 years old when COVID-19 began compared with people who were much younger or much older. Impact also depends on other characteristics, such as being poor or rich, and it is important for research to collect multiple measures across time and space in order to fully capture the impact of these events.