Since 1980, Americans born wealthy have had fewer chances of sliding down the social ladder than before, and the probability of escaping the bottom quintile has barely moved. A study by Jonathan Davis and Bhashkar Mazumder for the Chicago Fed identifies factors likely to contribute to the decline in mobility, notably the returns to education and the link between parental income and the probability of marriage. Approximately 40% of the decline in family income mobility is attributed, in this exercise, to the strengthening of the link between parental income and the probability of marriage.

The Essentials

  • Intergenerational mobility has declined in the United States since 1980: the probability of remaining in the top quintile has remained broadly stable, the probability of escaping the bottom quintile has barely moved (Federal Reserve Bank of Chicago, Davis and Mazumder, “The Decline in Intergenerational Mobility After 1980,” WP 2017-05, revised in 2022).
  • The marriage-parental income gradient has been multiplied by 4 since 1980. The strengthening of the link between parental income and marriage is associated with greater intergenerational persistence of family income.
  • Davis and Mazumder estimate that approximately 40% of the decline can be explained by the strengthening of the link between parental income and the probability of marriage, the study finding no significant increase in assortative mating.
  • Education remains a lever, but it is embedded within the link between parental income and marriage, which has strengthened.
  • To reverse the trend, public policies could address the prerequisites for mobility: housing, professional networks, and mixed social spaces.

The Matrimonial Gradient, an Amplifier of Inequality

The coefficient for parental income-probability of marriage rises from approximately 0.05 to 0.22 between two cohorts studied by Davis and Mazumder. This figure, estimated by Davis and Mazumder using the longitudinal surveys NLS66 and NLSY79, indicates a strengthening of the association between parental income and the probability of marriage. Individuals meet where they live, work, and study, but the magnitude of the shift is striking.

Two mechanisms work together. The first is residential: housing segregation has intensified over the past forty years in major American metropolitan areas. Affluent neighborhoods and poor neighborhoods have reinforced their class identity, reducing opportunities for encounters between different social backgrounds. The second is educational: elite universities, despite their financial aid programs, remain spaces massively frequented by children of affluent families, and they now constitute a primary matrimonial market. Harvard and Stanford also select future couples.

The effect on mobility is mechanical. When two high incomes combine in the same household, the household accumulates financial capital, real estate wealth, and social networks at a pace that a median income alone cannot achieve. When two low incomes combine, constraints accumulate: housing instability, limited access to healthcare, chronic stress that affects children’s development. Data from Raj Chetty and his colleagues (NBER, 2020) show that the neighborhood and household environment in the early years of life has lasting effects on children’s income trajectories thirty years later.

Asymmetric Mobility: It Is Easier to Stay at the Top Than to Climb

The Davis and Mazumder study distinguishes two dynamics that aggregate analyses tend to conflate. Since the 1980s, the probability of remaining in the top income quintile has remained approximately stable at around 31%. The probability of escaping the bottom quintile has barely moved. American mobility exhibits an asymmetry: the probability of staying at the top has remained relatively stable while that of escaping the bottom has barely progressed.

This result contradicts the image of an America where elites renew themselves. The American Dream, in its statistical version, presupposed circulation between classes: those at the top descending, those at the bottom ascending. This flux existed, imperfectly but really, until the 1970s. Deindustrialization, financialization, and the rise of highly skilled professions have since widened a gap between those who inherit networks and capital, and those who start with neither.

Homogamy amplifies this asymmetry in a specific way. At the top, two high incomes forming a couple protect their children through accumulation: real estate in secure neighborhoods, private or well-funded public schools, extracurricular activities, access to internships and mentors. At the bottom, two low incomes combined share the same constraints, often in the same under-resourced neighborhoods, without any catch-up dynamic. The effect is asymmetric.

The Limits of Education as a Corrective Lever

The problem also lies in the temporality of effects. Education acts on individuals before the matrimonial filter applies, but it does not modify the spaces where this filter operates. A degree obtained from a non-selective institution opens doors on the job market without necessarily integrating its holder into circles where unions between high earners form. Returns to education measured at the household level include any eventual effect on the partner. The study does not detect a significant increase in assortative mating.

The strengthening of the link between parental income and marriage appears as a probable contributor to intergenerational persistence, although its causal role is not established by these data alone.

The dominant policy response over the past thirty years has been educational: invest in early childhood, broaden access to universities, fund scholarships. These levers have their utility. Chetty’s data on mobility show that certain university institutions, notably public universities and community colleges, produce rates of upward mobility well above their share in higher education. Education can open trajectories.

But education alone encounters the matrimonial filter. Training more graduates from modest backgrounds is not enough if these graduates encounter, upon leaving their studies, segmented job markets and social spaces that remain equally separate. Educational mobility and economic mobility do not automatically overlap. A child from a modest family who obtains a diploma from a second-tier university rarely gains access to the networks of major firms, consulting firms, or investment funds—networks where the highest remuneration is concentrated today and, incidentally, where future life partners are found.

This decoupling is documented. Robotization and wage polarization have compressed median incomes and concentrated gains at the top of the distribution, making crossing the boundary between middle classes and affluent classes both more economically decisive and more socially difficult. Education policy can train more candidates for mobility; it cannot, by itself, reshape the spaces where trajectories are constructed.

The Geography of Mixing: An Underestimated Lever

If homogamy is nourished by spatial separation, it is geography that becomes a terrain for intervention. Research by Chetty and his colleagues on Opportunity Atlas shows notable variations in mobility among American counties: with equal income and education at birth, children who grow up in certain rural Midwestern counties or certain suburbs of Denver have significantly better trajectories than those who grow up in Atlanta or certain areas of the Deep South. These differences are explained in part by the actual social mixing of neighborhoods, family stability, and the quality of local institutions.

Housing policy thus becomes, indirectly, a mobility policy. Allowing construction of housing at varied costs in dynamic employment zones, where high value-added enterprises concentrate, increases the probability that young people from different backgrounds actually cross paths. Exclusive zoning laws, which de facto reserve the best-equipped neighborhoods for affluent families, are an obstacle to matrimonial mixing as much as to school mixing. Studies show that residential voucher programs targeting areas of high mobility can have effects on children’s trajectories.

The question of professional networks is of the same order. Large American companies recruit massively through channels that reproduce social selection: co-optation, unpaid internships accessible only to families who can finance them, networks of alumni from selective universities. Some firms have experimented with blind hiring or paid internship programs targeting less selective universities, with encouraging results on hiring diversity. These initiatives remain marginal relative to the scale of the phenomenon.

Systems That Limit Homogamy: What Other Countries Have Tried

The United States is not alone in observing this phenomenon, but the magnitude of the matrimonial gradient is particularly strong compared to Nordic countries or Canada. This difference stems less from cultural values than from institutions. In countries where access to selective universities is more disconnected from parents’ income, because tuition is low, student housing more mixed, and social benefits more universal, the matrimonial filter is less pronounced.

Intergenerational mobility already documented for the United States shows that the country stands out unfavorably from its developed peers on this ground. This situation results from accumulated institutional choices, some of which are reversible.

The Netherlands and Denmark have maintained homogamy rates lower than those in the United States notably thanks to the existence of more mixed social spaces: daycare centers and public schools attended by all classes, inexpensive universities, less segregated housing markets. These conditions do not guarantee mobility, but they keep open possibilities for encounters between different backgrounds that American segregation is progressively closing.

The Pew Charitable Trusts documented that American adults whose parents belonged to the top quintiles have twice the chance as their Canadian counterparts of remaining in the same quintile in adulthood. The gap between the two countries in terms of homogamy is not the only explanation, but it contributes to it.

Correcting the Mechanism, Not Just Its Symptoms

Educational inequalities and combined increasing returns are associated with an increase of approximately 14% in persistence; changes in the returns to education alone explain only a small part in this model. This result should refocus public policy debates. Education deserves to be supported, but it is not isolated from the effect of the matrimonial filter that recomposes inequalities. Targeting only degrees means training more qualified individuals for a matrimonial and residential market that remains segmented.

The levers that could correct the mechanism itself are known: housing policy favorable to mixing in dynamic zones, reform of public school financing to reduce gaps between rich and poor districts, paid internship programs accessible without family capital, universal early childhood educational institutions with homogeneous quality. These measures have no direct bearing on matrimonial choices—no one regulates marriages—but they act on the conditions in which these choices form.

Merit and individual effort remain constitutive values of the American ideal. If the strengthening of the link between parental income and marriage is a probable contributor to the closure of mobility, and if this link depends on where one lives, where one studies, and the networks one accesses, then meritocracy requires deliberate action on these prerequisites. The cohorts studied show lower intergenerational mobility and a stronger association between parental and children’s incomes, suggesting the importance of the prerequisites for mobility.

Sources

  1. Federal Reserve Bank of Chicago – Intergenerational Economic Mobility (Davis & Mazumder, 2022 NBER)
  2. Chetty, R. Et al. – Opportunity Atlas, NBER 2020 (opportunityinsights.org)
  3. Pew Charitable Trusts – Economic Mobility in America (annual reports)
  4. Columbia Center on Poverty and Social Policy – Residential mobility and intergenerational outcomes, 2026