The research, published in SSM – Population Health by Taegyun Yim, Euna Kim, and JungHo Park, analyzed twelve waves of longitudinal data from the nationally representative Korean Welfare Panel Study, covering the years 2012 through 2023. Rather than imposing a single affordability benchmark, the team constructed a series of binary indicators at one-percentage-point intervals ranging from 1 percent to 30 percent of income, then modeled how the association between housing cost burden and depressive symptoms shifted across each successive cut-off. The result is one of the most granular portraits to date of how housing-related financial strain translates into psychological distress, and it points to a systematic misalignment between the thresholds policymakers use and the lived experience of the people those policies are meant to protect.

The study focused on household heads, who typically bear primary responsibility for managing household finances and making residential decisions. The analytic sample comprised 3,413 households observed consistently across all twelve waves, yielding 40,956 household-year observations. To be included, households had to be continuously observed from wave 7 through wave 18 of the panel, maintain the same household head throughout, and provide complete information on all key variables. Depressive symptoms were measured using the Korean version of the Center for Epidemiologic Studies Depression Scale, an eleven-item instrument in which respondents rate experiences such as poor appetite, sleep disturbance, loneliness, and sadness on a four-point scale. Scores were summed and rescaled to a maximum of 60, with higher values indicating more severe symptoms.

Housing cost burden was defined using the monthly rent-to-income ratio, calculated as monthly housing expenses divided by total monthly household income. Housing expenses included rent, maintenance fees, utility bills such as water, electricity, and gas, and housing-related interest payments. Under the conventional 30 percent benchmark, roughly 12 percent of the sample households were classified as cost-burdened. Descriptive statistics already hinted at a stark divide: household heads in cost-burdened households reported a mean depressive symptom score of 10.8, compared with just 5.9 among those who were not cost-burdened. The cost-burdened group also tended to be older, more likely to be female-headed, less educated, more likely to rent, and substantially poorer on average than their non-burdened counterparts.

To move beyond simple cross-sectional comparison, the researchers employed a panel hybrid model, a statistical approach that separates within-household changes over time from persistent differences between households. This design matters because it allows researchers to distinguish the effect of a household becoming cost-burdened from the possibility that certain households are simply more likely to be burdened and also more prone to depression for other reasons. The hybrid model, implemented through a generalized linear mixed model framework, controls for unobserved household-level heterogeneity while still permitting estimation of effects for time-invariant characteristics such as sex. Standard errors were clustered at the respondent level, and year fixed effects absorbed economy-wide shocks.

The findings from this model were striking on two fronts. Within individuals, when a household became housing cost-burdened, the household head’s depressive symptom score increased by 0.39 points compared with periods when the same household was not burdened. Between individuals, household heads who were cost-burdened on average reported depressive symptom scores 4.33 points higher than those who were not, a substantial gap on a scale where the overall sample mean was 6.7. Employment and higher income emerged as protective in both dimensions, while living in rental housing was associated with elevated symptoms both within and between households.

But the most consequential finding came from the threshold-varying analysis. When the researchers re-estimated their models using cut-offs ranging from 1 percent to 30 percent in one-point increments, the association between housing cost burden and depressive symptoms did not switch on abruptly at 30 percent. Instead, the burden dummy first attained statistical significance in a low tipping zone around 10 to 15 percent of income, and the point estimates then rose almost monotonically as the cut-off increased, reaching their maximum around 27 percent. Below 10 percent, no detectable effect appeared. The pattern is a gradient, not a cliff. In other words, the mental health damage of high housing costs becomes measurable well before the conventional benchmark is reached, and it deepens progressively as burdens climb.

The researchers then asked whether this gradient differed by housing tenure, estimating cut-off-specific interaction models between the burden indicator and rental status. The answer was unambiguous. Among owner-occupied households, the within-person association between housing cost burden and depressive symptoms was small and statistically non-significant across most cut-offs. Among renters, however, the association turned positive and significant starting at the 9 percent cut-off and remained significant all the way through 30 percent, with the renter-owner difference itself statistically significant across many moderate-to-high thresholds, particularly from the mid-teens onward. This heterogeneity suggests that the same rent-to-income ratio carries very different meaning depending on whether a household owns or rents. For owners, moderate housing payments may reflect accumulated equity and economic security; for renters, they more directly cannibalize disposable income and signal residential precarity.

Sensitivity analyses reinforced the robustness of these conclusions. When the team redefined housing burden using the housing expenditure share relative to total household living expenses rather than income, a significant association with depressive symptoms emerged just below 30 percent and grew steadily at higher cut-offs. When depressive symptoms were treated as a binary outcome in logistic hybrid models, odds ratios became significant beginning around the 11 percent cut-off and remained elevated across many subsequent thresholds, closely mirroring the continuous-outcome results.

The theoretical framing of the study draws on the Dahlgren and Whitehead model of health determinants, which conceptualizes health as the product of layered social and economic conditions rather than discrete events. From this perspective, depressive symptoms are understood as outcomes produced through sustained exposure to structural pressures, with housing costs constituting one such core condition. If housing-related stress emerges progressively as costs rise, the authors argue, then reliance on a single administrative threshold may obscure earlier and more moderate forms of strain that nevertheless carry mental health consequences. South Korea makes a particularly instructive case for testing this idea because measured housing cost burdens there tend to be lower than in many Western contexts, yet housing remains central to household economic security through distinctive arrangements such as jeonse, a lump-sum deposit system that conventional monthly rent-to-income ratios may fail to capture fully.

The policy implications are considerable. When eligibility for housing assistance is triggered only after households cross a fixed administrative cut-off, support may arrive too late to prevent early psychological deterioration, particularly among vulnerable populations such as older adults and low-income renters who face disproportionate mental health risks from unaffordable housing. This timing mismatch may also help explain why evaluations of housing affordability interventions have produced mixed evidence on mental health outcomes. The authors suggest that housing policies targeting renters at earlier stages of cost burden, before they approach conventional benchmarks, could mitigate psychological harm before it accumulates. They also point to the protective role of employment and income security, arguing that rental subsidies may be more effective when complemented by labor market supports that strengthen both economic capacity and social integration.

The study acknowledges several limitations. The threshold at which housing costs begin to affect mental health may vary across regions with different housing markets and living costs. The analysis focused on household heads and could not examine whether other household members, such as spouses, adult children, or elderly parents, experience housing strain differently. Constructing a strongly balanced panel required excluding households with gaps in observation or changes in household head, which may introduce selection bias if the most housing-unstable households were also the most likely to drop out, potentially understating the true burden. The data also did not distinguish between mortgaged and unmortgaged owners, nor between jeonse and monthly rent arrangements.

Even with these caveats, the core message stands. The 30 percent rule remains a useful administrative reference point, but it identifies households already experiencing elevated strain rather than marking the onset of mental health risk. Mental health, the authors conclude, functions as a sensitive early indicator of housing-related stress, and policy frameworks that wait for a threshold to be crossed will systematically intervene after the damage has begun. Recognizing housing affordability as a sustained, graded source of psychological strain, rather than a binary condition that materializes at a fixed line, offers a path toward earlier and more targeted interventions aimed at preserving psychological well-being before manageable burdens harden into substantial distress.