Research

This study examines how employed, welfare-connected low-income mothers experience significant earnings losses around childbirth, spending roughly six months out of the labor force on average, though Black mothers tend to return sooner. It finds that social welfare programs, if fully utilized, can compensate for lost earnings after birth but not during pregnancy, with most available benefits being non-fungible nutrition assistance.

This study provides the first comprehensive evaluation of population forecast error for San Diego County and its jurisdictions, analyzing multiple forecast vintages and increment years over more than five decades. It finds that forecast accuracy decreases as the forecast horizon lengthens and is shaped by population growth rate, population size, and jump-off discrepancy, confirming established patterns in the literature while extending them to errors in population change rather than just population level. The analysis also shows that simple naive extrapolation methods often matched or outperformed more complex forecasting approaches, underscoring the value of benchmarking forecast performance.

This paper traces how the U.S. Census Bureau created public use microdata as part of the 1960 Census, highlighting how the resulting public use sample transformed analysis of the U.S. population and spurred a wave of new census-based research. It emphasizes that collaboration between academics and the Census Bureau, particularly through the Population Association of America, was central to developing this first microdata product, and argues that sustained collaboration remains necessary to keep microdata broadly accessible and usable.

This study uses two survey experiments to test whether the order in which fertility preference questions (general ideal, personal ideal, wanted, and expected fertility) are asked influences respondents' answers. Results show that most measures, especially personal ideal and wanted fertility, are relatively stable and indistinguishable regardless of question order, though expected fertility declines when asked later and personal ideal can shift among older respondents when asked after general ideal. The findings suggest general ideal fertility is a distinct, non-interchangeable construct, and the authors offer guidance for survey design regarding comparability of fertility preference measures across studies.

This study examines lifespans of U.S. immigrants born between 1850 and 1890, finding that while these immigrants generally lived shorter lives than U.S.-born Whites, this mortality disadvantage was concentrated among certain origin groups. Notably, migrants tended to outlive both their origin-country populations and their nonmigrant siblings, with migration-mortality patterns varying based on both origin and destination contexts.

This research note examines the reliability of retrospective reports on parity and birth dates using repeated measures from a longitudinal U.S. survey. Findings show that while most respondents report consistent parity and birth date information across waves, about one in ten respondents (and one in five parents) report parity inconsistently, and 11% of those with children report differing first birth dates. Such inconsistencies were more common among males and less advantaged respondents.

Using large U.S. Census and ACS samples from 1940–2013, the authors reanalyze prior findings on assortative mating and the division of labor in marriage. They find robust evidence for changes in the post-marriage division of labor, but more variable evidence on assortative mating, estimating it explains about 29% of earnings homogamy trends, while noting that analytic choices like the definition of newlyweds and handling of missing values significantly affect results.

This study examines how shifting partnership patterns have contributed to recent fertility change across several European Union countries and the United States between 2005 and 2020, using a Kitagawa decomposition to separate composition (partnership exposure) effects from rate (within-union fertility) effects. The analysis finds that declining marriage and cohabitation rates reduced total fertility rates in every country studied, even in nations where overall fertility rose, indicating that fertility trends stem not only from changes in childbearing behavior within unions but also from reduced exposure to union-based childbearing. The findings underscore the need to consider union formation dynamics when explaining contemporary fertility change.

This paper examines the long-term decline in fertility across rich countries, noting U.S. and European birthrates have been below replacement since the mid-1970s, with cohort fertility falling between those born in 1955 and 1975. It argues this decline stems from increased female autonomy combined with a mismatch between men's and women's preferences, since men tend to benefit more from traditional arrangements while women benefit more from departing from them. When men are unlikely to abandon traditions, some career-oriented women forgo or delay having children.

This study is the first to link extreme temperatures with unclaimed deaths in New York City, using daily air and wet-bulb temperature data matched to public burial records. Findings show that a 1°F increase over a 7-day summer period predicts 1.2% more unclaimed deaths, and decedents on extreme heat days were up to three times more likely to go unclaimed, highlighting heightened mortality risk among vulnerable urban populations during extreme heat.

This study uses a novel cohort discontinuity design to estimate the causal impact of the Great Recession on U.S. fertility, examining single-year birth cohorts of women overall and by race (White and Black). The findings show negative fertility effects for women born between 1964 and 1992, with the largest declines observed among younger cohorts and Black women.

This study investigates how chronic conditions contribute to the widening gap in disability and mortality between adults with and without a college degree from 2002 to 2018. It finds that while chronic disease prevalence rose similarly across educational groups, the relationships linking these diseases to disability and mortality diverged by education level, driving the growing disparities. The authors note these diverging patterns are not mainly explained by differences in health behaviors or insurance coverage.

This research note addresses the challenge of accurately projecting race group populations, using the Census Bureau's implausibly low 2050 projection for the American Indian and Alaska Native (AIAN) population as an example. The author argues that standard cohort component models fail because they ignore race response change—shifts in racial identification due to complex identities or administrative processes—and introduces a method to incorporate this factor. Applying this approach to the AIAN population, the study projects growth from 9.7 million in 2020 to 19.8 million by 2050, a much higher estimate than the Census Bureau's figure of 8.7 million.

Using CPS data from 1963–2023, the study compares posttax, posttransfer household income at ages 36–40 across five generations of Americans. It finds that income growth has slowed for younger generations, mainly due to stalled increases in women's work hours, though Millennials still saw 20% higher median income than the prior generation, with lifetime income gains outweighing rising educational costs.

This study uses U.S. Census data linked to administrative death records to examine how well early-adulthood sociodemographic characteristics predict individual lifespan among a cohort of men born in 1910, observed through deaths between 1975 and 2005 (N=121,000). Despite well-documented group-level disparities in life expectancy tied to structural forces like racism and class inequality, the analysis finds that these characteristics explain less than two percent of overall individual-level variation in lifespan. The findings reinforce that between-group variance in life expectancy is small relative to within-group variation, underscoring the nondeterministic nature of how structural inequality shapes individual mortality outcomes.

Using 2006–2023 American Community Survey data, this study finds that the long-standing rise in nonmarital births plateaued during the 2010s, as married women became increasingly more likely than unmarried women to have children, especially among those with a high school education or less. These results challenge deinstitutionalization theories of marriage, suggesting marriage norms remain influential and that declining marriage rates are contributing to the overall decline in U.S. birth rates.

This research note examines 170 years (1850–2021) of U.S. living arrangements by combining life expectancy data with harmonized IPUMS microdata to calculate expected life years spent alone, with primary kin, or in extended households, broken down by sex, age, and cohort. The analysis identifies three historical eras: a "large household" period (1850–1940) marked by stable extended households absorbing life expectancy gains, a "primary kin" era (1940–1980) featuring rapid growth in life years spent with immediate family and declining extended households, and a "diversified" phase (1980–onward) marked by declining two-parent households and increasing diversity in arrangements, including living alone and with extended kin.

This study uses 2010–2019 American Community Survey data to examine marriage patterns among multiracial individuals, focusing on rates of racial endogamy and "partial endogamy" (sharing one racial component with a partner) after adjusting for group size and demographic controls. Findings show that exact and partial racial endogamy are common among multiracial people, with no evidence that they show a general affinity for other multiracial partners without shared ancestry or that race matters less in their partner choices compared to monoracial individuals. The results suggest that racial classification and established racial boundaries remain influential in shaping marriage markets even as racial categories grow more ancestrally complex.

This study projects the demographic impacts of a 2025 executive order ending birthright citizenship for children of certain noncitizen parents. It finds that Latinos would bear the largest absolute impact, comprising nearly 80% of new "unauthorized" births short-term and over 90% by 2050, while Asians would face the largest relative impact due to their higher share of temporary visa holders, with 41 unauthorized births per 1,000 unauthorized Asians compared to 17 per 1,000 among Latinos. The authors call for greater understanding of the societal implications of these disparate effects on millions of children and families.”

This study examines how residential racial segregation and its spatial clustering shape individuals' choices of frequently visited (activity) neighborhoods, comparing White and Black residents in Chicago using discrete choice models. Findings show that all groups tend to gravitate toward White contiguous segregated cluster (CSC) neighborhoods—White residents show geographic isolation by avoiding Black areas, while Black residents from Black CSC areas often choose activity locations in White CSC neighborhoods due to greater institutional, amenity, and safety advantages. The results, robust to alternative model specifications, have implications for understanding routine mobility patterns and designing desegregation policies via behavioral nudges.

This study uses 12 statistical and machine learning models with over 150 predictors from the U.S. Health and Retirement Study to predict individual-level lifespan, finding that despite comparable and relatively high discriminative accuracy across models, most individual-level lifespan heterogeneity remains unexplained. The authors identify consistent inequalities in mortality predictability, with lower accuracy for men, non-Hispanic Blacks, and less-educated individuals, who also show less accurate subjective predictions of their own lifespan, while top predictive features (habits, health history, finances) are similar across groups. The findings highlight the inherent limits of individual mortality prediction even with rich longitudinal data and underscore context-dependent inequalities relevant to future research and public policy.

This study uses a Bayesian multistate life table approach with Health and Retirement Study data (1992–2018) to examine racial, ethnic, and sex differences in marital status and living arrangements after age 50. Findings show White adults typically spend most later years married and living with a spouse, Black adults spend the least time in marriage/spousal coresidence and more time alone or with nonspouse family, and Hispanic adults fall in between with notable time in multigenerational households; minority women experience especially reduced years of marriage and spousal coresidence compared to men. The authors argue these disparities highlight the need for policies that account for diverse family structures supporting aging populations.

This study estimates racial and ethnic welfare disparities among older Americans using longitudinal data and an expected utility framework that accounts for consumption, leisure, health, mortality, and wealth. The findings show that inequality is understated by traditional measures like consumption or life expectancy alone, and that most of the welfare gap stems from conditions already present by age 60 rather than differences in how people age afterward. The authors conclude that reducing late-life health risk factors like hypertension or diabetes has only a marginal effect on closing these gaps, suggesting that policies targeting earlier stages of the life cycle would be more effective.