How architectural principles can help conceptualize and analyze breakups among intergenerational households

The study investigates the stability of children’s intergenerational living arrangements, focusing on the odds that three-generation and skipped-generation households dissolve. Prior work links instability in children’s living arrangements with adverse outcomes, but less is known about how differences in household generational structure and the generation that heads the household (the focal generation) relate to breakups. The authors propose a novel conceptual framework inspired by architectural principles: (1) households are ordered by a generational hierarchy (first, second, third), and (2) a focal generation functions as the central entity maintaining the household. Applying this framework, they define and compare three household types—three-generation headed by grandparents (3G1G), three-generation headed by parents (3G2G), and skipped-generation headed by grandparents (SG1G)—to test whether the focal generation and hierarchy predict dissolution.

Prior studies show that three-generation households are often transitory as young parents move out; only a minority of families remain in such arrangements over multiple years. Evidence indicates teenage childbearing, race, and family structure predict longevity in three-generation households, with Black families tending to sustain coresidence longer than White families. Skipped-generation households, despite greater economic hardship, often persist longer than three-generation households, with caregiving by grandparents commonly extending for multiple years. The literature, however, lacks large-scale comparisons that simultaneously consider both the generational hierarchy and which generation heads the household. The authors hypothesize: (a) three-generation households headed by grandparents (3G1G) will have higher dissolution odds than those headed by parents (3G2G), (b) skipped-generation households (SG1G) will be the most durable, and (c) demographic and economic characteristics of the focal generation (e.g., age, race/ethnicity, marital status, education, disability) and household resources (income, SNAP participation) will further shape dissolution odds.

Data: The study uses the U.S. Census Bureau’s Survey of Income and Program Participation (SIPP) 2004, 2008, and 2014 panels, which provide monthly longitudinal data on household composition, headship, and socioeconomic characteristics. The design enables precise identification of generational hierarchy and focal generation each month by enumerating relationships to the household head. Sample: 5,831 Non-Hispanic White, Non-Hispanic Black, and Hispanic grandchildren who were age 12 or younger at panel start and living with grandparents, generating 116,796 grandchild-months. Distribution at baseline: 61.5% in 3G1G (n=3,587), 19.8% in 3G2G (n=1,152), and 18.7% in SG1G (n=1,092). Outcome: Monthly indicator of dissolution (household transitioning away from the initial intergenerational configuration via exits of parents, grandparents, or grandchildren). Descriptives: Event-history (survival) summaries produce incidence rates and timing to dissolution. Modeling: Because monthly dissolution is a rare event (~1.5% per grandchild-month), the authors estimate rare-events logistic regression with small-sample/rare-event bias correction. Key predictors: indicators for household type (3G1G and SG1G; 3G2G as reference), focal-generation characteristics (age, gender, race/ethnicity, nativity, marital status, education, labor force attachment, work-preventing disability), child characteristics (age group, gender), and household/contextual factors (log monthly household income, SNAP participation, poverty status, metro residence, state of residence, log state unemployment rate, and seasonal variation). Models include month-in-panel and region controls and use robust standard errors. Limitations of data include panel lengths of 48–60 months and lack of pre-panel duration in the initial configuration (left-censoring).

Descriptive differences: SG1G households were the most economically disadvantaged: 45% below poverty vs. 33% (3G1G) and 30% (3G2G); lower household income ($3,246 vs. $5,691 and $5,359 monthly); higher public assistance; and greater SNAP participation (39% vs. 40% in 3G1G and 32% in 3G2G). SG1G heads were older (57.4 years), more likely disabled and unmarried, and more often out of the labor force. Incidence rates and timing (Table 2): Monthly dissolution incidence was highest for 3G1G (0.0137), followed by 3G2G (0.0123), and lowest for SG1G (0.0095). Average months until dissolution: 18.6 (3G1G), 20.4 (3G2G), 20.6 (SG1G). Regression results (Table 3 and text): Relative to 3G2G, 3G1G had higher odds of dissolution, while SG1G had lower odds, confirming the central role of generational hierarchy and focal generation. Focal generation characteristics mattered: household heads who were Non-Hispanic Black or Hispanic had lower odds than Non-Hispanic Whites; older head age modestly reduced odds. Education showed mixed patterns: having only a high school diploma was associated with lower odds compared to less than high school; some college/college degree suggested higher odds but were generally not statistically significant. Unmarried status and work-preventing disability of the head were not significant. Household resources: higher log monthly income significantly reduced dissolution odds; SNAP participation substantially reduced odds, consistent with an alleviation of food insecurity helping households remain intact. Context: Seasonal patterns showed higher odds in summer and lower in winter; higher state unemployment rates were associated with increased odds. Predicted odds by race and type: Averaged monthly predicted odds indicated White 3G1G and 3G2G had the highest odds (~2.3% and ~2.2%), Black 3G2G around 1.83%, White SG1G about 1.3%, and Black SG1G the lowest (~0.8%). Black intergenerational households generally showed lower dissolution odds than White counterparts.

Findings substantiate that both the generational hierarchy and which generation is the focal head are crucial to understanding the durability of intergenerational households. Three-generation households headed by grandparents dissolve faster than those headed by parents, likely reflecting younger parents’ preferences or economic/role dynamics prompting departures from grandparents’ homes. Despite greater economic hardship, skipped-generation households endure longer, consistent with the strengthened caregiving imperative when parents are absent and incentives to avoid alternatives like foster care. Race moderates these patterns: Black intergenerational households, particularly Black SG1G, show the lowest dissolution odds, potentially reflecting kinship and support norms. Economic supports enhance stability: higher income and SNAP participation reduce breakup odds, and seasonal/contextual factors (summer months, higher state unemployment) elevate risks. Overall, while differences are meaningful, dissolution remains a relatively rare monthly event, implying substantial short- to medium-term stability within the observed panel windows.

By importing architectural principles of hierarchy and focal point, the study offers a coherent framework to classify intergenerational households and explain variation in their stability. The approach demonstrates that identifying both the generational structure and the focal generation materially improves prediction of household dissolution: SG1G are most durable; among three-generation households, those headed by parents outlast those headed by grandparents. Racial differences further condition these effects, with Black households exhibiting greater durability. Economic resources and food assistance play stabilizing roles. Policy implications include tailoring supports by household type—e.g., SNAP and other aid can bolster stability or facilitate appropriate transitions depending on the configuration, and legal or service supports (such as custodial rights for grandparents or in-home care access) may be differentially beneficial. Future research should use data enabling richer ecological/contextual measures, address left-censoring by observing households from formation, extend observation windows, and test mechanisms that link focal-generation characteristics, economic adaptations, and seasonal/contextual shocks to household continuity.

Data constraints limit observation to 48–60 months and do not capture pre-panel duration of the initial intergenerational arrangement, introducing left-censoring. The SIPP lacks detailed indicators of broader ecological systems (family history, culture, neighborhood context) and within-household dynamics/conflicts that may precipitate dissolution. Some focal-generation measures (e.g., disability, unmarried status) were not significant, which may reflect measurement, sample composition, or unobserved factors. Results are based on households with grandchildren age ≤12 at baseline, which may not generalize to households with older grandchildren. The rarity of monthly dissolution events, while addressed with bias-corrected logistic models, still constrains precision for some estimates.