Simulating how rent burden, eviction, and homelessness risk propagate into physical and mental health outcomes across a household population
Housing is increasingly recognized by clinicians and public health researchers as a foundational "social determinant of health" — not a background condition, but an active clinical variable. Sociologist Matthew Desmond's Evicted (2016), based on fieldwork in Milwaukee, documented how eviction is not merely a symptom of poverty but a cause of it, triggering job loss, depression, and worse housing in a self-reinforcing cycle. This simulator starts from a baseline renter population and traces how instability accumulates into measurable health harm — and how targeted intervention reverses it.
The U.S. Department of Housing and Urban Development defines "cost-burdened" as spending more than 30% of gross income on housing, and "severely cost-burdened" as spending more than 50%. These are not arbitrary accounting thresholds — they mark the point at which households begin trading off spending on food, medication, transportation to medical appointments, and preventive care to keep a roof overhead.
Household budget studies consistently show that above the 30% line, discretionary health spending is the first category cut: prescriptions go unfilled, follow-up visits are skipped, and preventive screenings lapse. Below the line, households retain enough slack to absorb an unexpected medical bill or a week of reduced hours without triggering a housing crisis. Above it, the same shock cascades directly into missed rent.
Aggregate eviction and homelessness rates describe an outcome, but obscure the pathway. This simulator represents each household as an individual point moving along a stability spectrum — from stable, to rent-burdened, to eviction notice, to homelessness risk — so the compounding, path-dependent nature of housing instability becomes visible rather than abstract.
The same underlying mechanism recurs across the housing-health literature: instability is rarely a single event. It is a trajectory shaped by income shocks, landlord decisions, local eviction-court procedure, and the availability (or absence) of legal and financial intervention at each decision point along the way.
Princeton's Eviction Lab, founded by Matthew Desmond in 2017, produced the first nationwide database of eviction court records, revealing that roughly 3.6 million eviction cases are filed in the U.S. each year — about one every eight seconds on average pre-pandemic.
The horizontal gauge that anchors every stage of this simulation runs from "Stable" on the left to "Homelessness Risk" on the right. Each small house icon is one simulated household; its position along that axis is its current housing-instability score, and its color follows the same four-band classification used throughout: green for stable, yellow for rent-burdened, orange for eviction risk, red for homelessness risk.
At baseline, the population clusters heavily toward the stable band, with a visible minority already drifting into the rent-burdened zone — the same households the Joint Center for Housing Studies identifies as "marginally housed": not in crisis today, but carrying no financial buffer against the next rent increase or reduction in hours.
The transition from stable housing to crisis rarely begins with a single catastrophic event. It typically begins with an income shock — a cut in hours, a medical bill, a car repair — landing on a household that already has little financial slack because rent consumes a large share of income. The rent-burden threshold slider in this simulator models exactly this sensitivity: the higher the share of income consumed by rent, the smaller the shock required to trigger an eviction notice.
In most U.S. states, a landlord who wants to evict a tenant for nonpayment must first serve a written notice — commonly a "pay or quit" notice — giving the tenant a narrow window, often 3 to 14 days, to pay the arrears in full or vacate. If the household cannot cure the debt in that window, the landlord may file in eviction court.
Critically, most tenants facing eviction have no legal representation, while the overwhelming majority of landlords do. Desmond and colleagues found that unrepresented tenants are far more likely to receive a default judgment simply because they never appear in court — not because they lack a valid defense, but because they lack notice of the process, transportation, or childcare to attend a hearing during work hours.
Raising the rent-burden threshold in this simulator does not just relabel a handful of households — it shifts the entire distribution. Because rent burden and income volatility are correlated (lower-income renters face both higher burden ratios and less predictable income), small increases in the burden threshold rapidly expand the pool of households one shock away from an eviction notice.
This mirrors real housing markets: as median rent rises faster than area median income — as it has in most major U.S. metros since 2019 — a larger share of the renter population crosses the 30% and 50% burden lines simultaneously, even without any change in individual household behavior.
A single eviction filing appears on a tenant's public record indefinitely in many jurisdictions, and can disqualify them from future rental applications even if the case was later dismissed — meaning the mere filing, not just the outcome, can trigger irreversible housing instability.
This simulator's rent-burden threshold slider is a simplified proxy for a well-documented empirical relationship: the share of a renter population living above the 30% burden line rises non-linearly, not proportionally, as local median rent outpaces local median income. A metro where rents have risen 26% while wages rose 14% does not just shift a few marginal households — it can push a large cohort across the burden line within a single lease-renewal cycle.
That non-linearity is why housing counselors and eviction-prevention programs emphasize early intervention: a household one month behind on rent is dramatically cheaper and easier to stabilize with a small emergency rental-assistance grant than the same household is to rehouse after a formal eviction judgment, a lost security deposit, and a damaged rental history.
When an eviction notice is not resolved, households do not simply lose a home — they enter a cascade of forced moves, doubling up with relatives or friends, overcrowded living arrangements, and for a growing share, entry into literal or imminent homelessness. Each transition compounds the last: a forced move mid-lease often means a worse unit, a longer commute, disrupted schooling for children, and weakened social ties that would otherwise buffer the next shock.
Forced moves rarely land households in equivalent or better housing. Displaced tenants typically relocate to lower-quality units in higher-poverty, higher-crime neighborhoods — not because they prefer them, but because an eviction record and compressed timeline eliminate better options. Desmond's Milwaukee fieldwork documented that evicted mothers were significantly more likely to lose their jobs in the following months, not because of unrelated circumstances, but because the eviction process itself — court appearances, moving days, unstable childcare — directly conflicted with work schedules.
Doubling up (moving in with relatives or friends) is the most common response to an eviction and is undercounted in official homelessness statistics, since HUD's formal definition of homelessness generally excludes people staying temporarily with others. Overcrowding that results from doubling up independently elevates infectious disease transmission, household conflict, and child developmental stress.
The homelessness-risk zone in this simulator represents households for whom doubling-up arrangements have also broken down, or who lack any informal safety net to draw on — the population most likely to present at a shelter intake or emergency room. HUD's Annual Homelessness Assessment Report (AHAR) point-in-time count found 653,100 people experiencing homelessness on a single night in January 2023, an increase driven substantially by rising rents outpacing income growth in the years prior.
Entry into this zone is rarely permanent for any single household, but the population flow is continuous: as some households exit through rehousing or reunification, new households enter through unresolved eviction and exhausted informal support — a churn this simulation renders visually as continuous dot movement rather than a static count.
Children who experience a forced move show measurably higher rates of chronic absenteeism and lower standardized test scores in the year following the move — effects that persist even after the family re-stabilizes, according to longitudinal school-district studies cited in housing-mobility research.
A household that avoids formal eviction by moving voluntarily ahead of a court date is often counted, in official statistics, as a routine move rather than a housing-instability event — even though the underlying dynamics are identical. Each additional move within a short window degrades the very resources (steady employment, a fixed school, a known primary care clinic, a stable social network) that would otherwise help a household recover.
This is why the escalation stage of the simulation shows households drifting further right along the gauge even without any single new shock: instability, once triggered, tends to be self-sustaining unless something external — legal aid, emergency assistance, a housing-first placement — interrupts the trajectory.
Housing instability does not merely correlate with worse health outcomes — a growing body of longitudinal and quasi-experimental research shows it causes them. The physiological pathway runs through chronic stress: sustained activation of the hypothalamic-pituitary-adrenal axis under prolonged uncertainty elevates cortisol and inflammatory markers, a state neuroscientist Bruce McEwen termed "allostatic load," which over time degrades cardiovascular, metabolic, and mental health.
Chronic housing instability functions as a persistent, unpredictable stressor — precisely the kind of exposure allostatic load research identifies as most damaging, because the body's stress-response systems never fully down-regulate between episodes. Sustained cortisol elevation contributes to insulin resistance, hypertension, and impaired immune function; the practical clinical consequence is that patients with unstable housing show worse control of diabetes and hypertension even when medication access is equivalent, because stress, disrupted routines, and lost medical records undermine adherence.
Desmond and Kimbro's 2015 study of Milwaukee mothers found that eviction was associated with significantly elevated rates of depression and material hardship up to two years after the event — evidence that the health harm of housing instability is not confined to the acute crisis but extends well into the recovery period, if recovery happens at all.
As housing instability disrupts primary care continuity — patients lose their address of record, miss appointments during moves, or lose insurance eligibility tied to a fixed address — the emergency department becomes the default access point for both acute and, increasingly, manageable chronic conditions. This is visible in the ED-visit-rate metric in this simulator, which rises in step with the instability index even though the underlying disease burden accumulates gradually rather than suddenly.
Pediatric health is especially sensitive: children experiencing housing instability show higher rates of asthma exacerbation (linked to substandard or overcrowded housing conditions), delayed developmental milestones, and behavioral health referrals, compounding the adult health effects within the same household.
A study of Boston Medical Center pediatric patients found that children in housing-insecure families were hospitalized at rates comparable to children with chronic conditions like asthma — despite housing insecurity not being a diagnosis anywhere in the medical chart, prompting the rise of formal social-needs screening in primary care.
A growing number of health systems now administer standardized social-needs screening tools — such as the Accountable Health Communities Health-Related Social Needs screening tool — at primary care and emergency-department intake, asking directly about eviction notices, utility shutoffs, and housing quality alongside traditional vital signs. The rationale mirrors this simulator's design: instability is a leading indicator of health decline, not a lagging one, and it is measurable before the emergency-department metric spikes.
Health systems that screen for housing instability and connect patients to legal aid or rental-assistance resources at the point of care are, in effect, intervening at the second stage of this simulation's pipeline — before rent burden converts into full escalation — which is both clinically and financially far cheaper than treating the downstream chronic-disease and mental-health consequences modeled in this stage.
The Housing First model, pioneered by psychologist Sam Tsemberis with Pathways to Housing in 1990s New York, inverted the traditional sequence of homelessness services: instead of requiring sobriety, treatment compliance, or "housing readiness" before offering permanent housing, Housing First places people directly into permanent housing with no preconditions, then wraps voluntary supportive services around them. Three decades of research — including randomized controlled trials — show this approach out-performs treatment-first models on housing retention, and produces measurable downstream health-cost savings.
The strongest evidence for Housing First comes from Canada's At Home/Chez Soi study (2009–2013), a five-city randomized controlled trial run by the Mental Health Commission of Canada that assigned homeless adults with mental illness to either immediate Housing First placement or "treatment as usual." Housing First participants achieved stable housing far faster and maintained it longer, and the trial documented significant reductions in emergency department visits and inpatient hospital days relative to the control arm.
The logic is counter-intuitive to traditional service models but empirically robust: stable housing is not the reward for recovery — it is very often the precondition for it. Without a stable address, medication adherence, mental health treatment engagement, and chronic disease management are all measurably harder to sustain.
Medical-legal partnerships (MLPs) embed civil legal aid attorneys directly within healthcare settings, allowing clinicians to "prescribe" legal help for patients facing eviction, utility shutoff, or benefits denial — treating the legal problem as a health problem, because it functionally is one. The National Center for Medical-Legal Partnership estimates that the large majority of tenants who receive legal representation in eviction proceedings remain housed, compared to a small minority of unrepresented tenants — a gap driven less by the merits of individual cases than by the basic fact that represented tenants show up, negotiate, and access procedural protections unrepresented tenants often never learn exist.
Cities that have implemented a "right to counsel" in eviction court — New York City being the most studied example — have documented substantial reductions in eviction judgments and displacement rates following implementation.
Beyond the human case for stabilization, the fiscal case is well documented. Culhane and colleagues' landmark study of New York City's supportive housing initiative for homeless adults with serious mental illness found that supportive housing placements were associated with an offsetting reduction in shelter, hospital, and correctional system costs nearly equal to the cost of the housing itself — meaning the intervention was close to cost-neutral even before counting the value of improved health and stability.
The housing-first investment slider in this simulator reflects that logic directly: as investment rises, a larger share of at-risk households are pulled back toward the stable end of the spectrum, the aggregate instability index falls, and the ED-visit-rate metric declines in parallel — the same pattern documented in the At Home/Chez Soi trial and subsequent supportive-housing cost studies.
Housing-first and rapid-rehousing programs are not a guarantee against future instability — relapse into housing crisis remains possible after any single intervention. But the weight of RCT and quasi-experimental evidence is unusual in social policy: stable housing, offered first and without preconditions, is one of the most consistently effective health interventions available, precisely because housing is itself the mechanism, not merely a marker, of the outcomes it predicts.