A Lifecycle Estimator of Intergenerational Income Mobility

Lacking lifetime income data, most intergenerational mobility estimates are subject to lifecycle bias. Using long income series from Sweden and the US, we illustrate that standard correction methods struggle to account for one important property of income processes: children from affluent families experience faster income growth, even conditional on their own characteristics. We propose a lifecycle estimator that captures this pattern and performs well across different settings. We apply the estimator to study mobility trends, including for recent cohorts that could not be considered in prior work. Despite rising income inequality, intergenerational mobility remained largely stable in both countries.

💡 Research Summary

**
The paper tackles a persistent problem in the measurement of intergenerational income mobility: lifecycle bias that arises when only short snapshots of income are available for the child generation. Using long‑run administrative records from Sweden and the Panel Study of Income Dynamics (PSID) from the United States, the authors first demonstrate that a crucial feature of the income process is systematically ignored by most existing correction methods. Specifically, children from wealthier families tend to have lower initial earnings but experience substantially faster income growth as they age, even after conditioning on education, occupation, and other observable characteristics. For example, among college‑educated Swedish men, those whose fathers are in the top quartile earn less in their mid‑20s but have about 40 % higher earnings by age 40 compared with peers from the bottom‑quartile families. Similar patterns are observed in the US data.

Standard approaches either (i) model the relationship between observed annual income and unobserved lifetime income through an errors‑in‑variables framework, or (ii) estimate partially observed income profiles using age and education controls. Both strategies omit the interaction between parental income and child age, leaving a correlation between the prediction error and parental income that biases the intergenerational elasticity (IGE). Consequently, IGE estimates are highly sensitive to the age at which child income is measured, and recent cohorts (born after the 1980s) are especially vulnerable to over‑ or under‑estimation.

The authors propose a “lifecycle estimator” that explicitly incorporates the parental‑income × child‑age interaction in the first‑stage income‑profile model. The first stage regresses child income on age, education, and a set of controls, adding a term for parental log‑income interacted with child age (and optionally with child’s own income level to capture “fanning out”). This specification captures heterogeneous growth rates across families. In the second stage, the estimated coefficients are used to predict a full lifetime income trajectory for each individual, from which a lifetime average income is constructed. The IGE is then estimated by regressing the predicted lifetime child income on parental lifetime income. By construction, the correlation between the first‑stage prediction error and parental income is eliminated, delivering an unbiased IGE.

The estimator is evaluated in two ways. First, a benchmark sample (cohorts born 1952‑1960) with near‑complete income histories is used to compute a “true” IGE based on actual lifetime earnings. The lifecycle estimator reproduces this benchmark IGE almost exactly, whereas conventional corrections deviate substantially depending on the chosen observation window. Second, robustness checks show that the estimator’s performance is largely invariant to the age range of observed child incomes and to the number of income observations per individual. Even when only early‑career earnings are available (as for recent cohorts), the method yields stable IGE estimates.

Applying the estimator to study mobility trends across birth cohorts (1950‑1989) reveals that earlier claims of declining mobility in Sweden and a sharp rise in mobility in the US for the 1980s cohort are largely artefacts of lifecycle bias. In Sweden, once family‑specific growth effects are accounted for, mobility appears roughly constant from the 1950s through the 1980s, with a modest increase for the 1980s cohort. In the United States, the lifecycle‑adjusted IGE remains around 0.45‑0.50 across all cohorts, contradicting narratives that mobility collapsed for children born in the 1980s despite rising income inequality and larger parental input gaps.

The paper concludes that intergenerational mobility, as measured by the IGE, has been remarkably stable in both Sweden and the United States over the past several decades, even as income inequality has grown. The proposed lifecycle estimator offers a practical tool for researchers working with incomplete income panels, especially for recent cohorts where only early‑life earnings are observed. The authors also note that while their implementation is parametric, the first‑stage prediction problem could be tackled with machine‑learning techniques, potentially improving accuracy further. Future work could extend the framework to incorporate non‑income outcomes (education, health) and to explore heterogeneous mobility across demographic groups.