Fear of eviction makes people ill

Situation

The first phase of the Renters’ Rights Act came into force in England on 1 May 2026. The Act received Royal Assent on 27 October 2025, seven years after Theresa May, then prime minister, promised to end no-fault eviction. It abolishes Section 21, the route by which a landlord could end a tenancy without giving a reason, and replaces fixed-term assured shorthold tenancies with open-ended periodic ones. A landlord can no longer evict in the first twelve months and must give four months’ notice on the grounds of sale, moving in or housing a family member.

The months before the deadline produced a rush. Landlord Action recorded a 60 per cent year-on-year rise in eviction instructions for March 2026, with Section 21 instructions up 43 per cent across the first quarter. In parallel, Savills counted 254,000 buy-to-let homes listed for sale over the past twelve months, 697 a day, 28 per cent above the same period in 2024, with 86 per cent of the sold homes leaving the rental sector. The deadline compressed tens of thousands of forced moves into a few months, ahead of a law meant to make moves like them harder to impose.

The reform is usually argued in financial and legal terms: the effect on landlords, on the size of the rental market, on rents. That argument is real and it runs later in this report. But it skips the prior question. Before the money, a forced move or the standing threat of one does something to the person who has to live through it, and that something is measurable. The official statistics already gesture at it. MHCLG recorded 4,960 households threatened with homelessness as a direct result of a Section 21 notice in the last quarter of 2025, down 15.8 per cent year on year. When Keir Starmer justified the reform he did not talk about money. He said: “For too long, families have lived with the constant fear of eviction.” That is a claim about a state of mind. The honest test of the reform is therefore a health test as much as a market one, and that is the test the debate has mostly left out.

Findings

Where the harm starts

Across this evidence, the damage starts before anyone moves. Katherine Fowler and colleagues counted eviction- and foreclosure-related suicides in sixteen US states over 2005 to 2010 and found 929 of them, with the annual number doubling across the period (Fowler et al., 2015, American Journal of Public Health). 79 per cent of those suicides happened before the actual loss of the home, and more than a third clustered within two weeks of an acute housing event. People were breaking under the approach of homelessness, in the weeks before it arrived.

That is a count of deaths. It is not a controlled experiment, and it lands in the middle of a financial crisis, so the broader economy is tangled into it. The cleaner test came later, by accident of policy. Kathryn Leifheit and colleagues used the patchwork of US eviction moratoria during the pandemic as a natural experiment (Leifheit et al., 2021, JAMA Network Open). States that imposed strong moratoria, ones that blocked the notice and the court filing rather than only the final bailiff step, saw moderate-to-severe psychological distress among low-income renters fall by 12.6 per cent relative to the rest. Weak moratoria, which stopped only the eviction itself, did nothing measurable. The threat was the exposure. Take the threat away early enough and the distress comes down with it.

This is the closest the field gets to causal proof, and it points the same way as Starmer’s sentence. The harm a no-fault regime produces is not confined to the moment of eviction. It runs through every year a household can be given notice without a reason.

What it does to the adults who live with it

Matthew Desmond and Rachel Tolbert Kimbro followed evicted mothers in the Fragile Families study and matched them against similar mothers who were not evicted (Desmond and Kimbro, 2015, Social Forces). The evicted mothers had more material hardship, higher rates of depression and worse health. Two years on, the gap had not closed; the eviction bent the line of the years after it.

Morgan Hoke and Courtney Boen pushed on the causal question with longitudinal data from the Add Health cohort (Hoke and Boen, 2021, Social Science and Medicine). Tracking the same people over time, so that fixed differences between them drop out, they still found more depressive symptoms after an eviction, with about a fifth of the effect operating through heightened social stress. The size of the effect shrank once they adjusted fully, but eviction still preceded more depressive symptoms, not fewer, under the harder test. Hugo Vásquez-Vera and colleagues pulled the literature together in a review of 47 studies and reported that 55 per cent found a mental-health effect of eviction or its threat, most often depression, anxiety and distress (Vásquez-Vera et al., 2017, Social Science and Medicine). Vásquez-Vera and colleagues also note plainly that most of the underlying studies are cross-sectional, so the correlation is firmer than the causation.

What it does to children

The clearest administrative evidence is about children, because schools record what happens to them. Peter Hepburn, Matthew Desmond and colleagues linked eviction court records to school enrolment data across Houston and found that 12.3 per cent of children whose parents received an eviction filing changed school mid-year, almost three times the rate of comparable children (Hepburn et al., 2025, Sociology of Education). Fewer than half of the threatened children stayed at the same school from one year to the next. They lost about two more school days in the year of the filing, picked up extra suspensions afterwards and tended to land in schools with worse results and higher churn. The disruption extended past the move, into the child’s school year.

The strongest single study on childhood moves comes from Denmark, where the national registers let researchers follow everyone. Roger Webb, Carsten Pedersen and Pearl Mok tracked all 1.47 million Danes born between 1971 and 1997 and counted their moves through childhood (Webb, Pedersen and Mok, 2016, American Journal of Preventive Medicine). Frequent moves, and above all several moves in a single year during early adolescence, went with sharply raised later risks of attempted suicide, violent offending, substance misuse, psychiatric illness and premature death. The point that makes this study hard to dismiss is that the raised risk appeared in every income band. That is the best available argument that the churn itself does damage, and not only the poverty that often drives it.

The mirror test

The studies that improved housing on purpose, and measured what happened, run the other direction. Philippa Howden-Chapman ran a cluster-randomised trial in New Zealand that insulated some homes and left others as they were (Howden-Chapman et al., 2007, BMJ). The insulated households roughly halved their odds of reporting fair-or-poor health, took fewer days off school and work and visited the doctor less. It is a trial of housing quality rather than tenure, but it is a real randomised trial, and the effects are large and consistent.

The Moving to Opportunity experiment did the same for neighbourhood. Jens Ludwig and colleagues reported that families randomly given vouchers to move to lower-poverty areas showed higher subjective wellbeing a decade later, by about a tenth of a standard deviation, with a clear dose-response (Ludwig et al., 2012, Science). The effect on clinical mental health was weaker, significant only at the margin. In a separate analysis the same experiment cut extreme obesity and diabetes markers, so the body responded even where the diagnosis of mental illness did not move much (Ludwig et al., 2011, New England Journal of Medicine). Raj Chetty, Nathaniel Hendren and Lawrence Katz then showed that children who made the move young grew up to earn about 31 per cent more as adults, while teenagers who were moved gained nothing and sometimes lost (Chetty, Hendren and Katz, 2016, American Economic Review). Those are economic outcomes, not health ones, but they point to the same conclusion as the health evidence. Timing sets the sign for earnings: Chetty and colleagues found that only children who moved young gained that income boost as adults, while those who moved as teenagers gained nothing and sometimes lost ground. Whether timing matters the same way for wellbeing and physical health is untested.

Position

Put the strands together. The standing threat of a no-fault eviction is a measured source of harm, the threat as much as the event, and the cleanest causal evidence we have says that removing it lowers distress. Read this way, the Renters’ Rights Act is a health intervention, and it should be judged against the wellbeing and distress data.

But a health intervention only helps the people it reaches, and the market decides who that is. The reform abolishes the no-fault route while leaving the sale route open, and a SpareRoom survey attributes 43 per cent of evictions since Royal Assent to a landlord selling. The people pushed out by a sale and the people priced out of the next tenancy inherit exactly the insecurity the reform was built to remove. The 697 sales a day are the scale of that group. The Act redistributes the harm of insecurity rather than removing it: the household that keeps its home gets the health dividend, and the 697 daily sales are the size of the group that does not. The reform has made that second group smaller in some places and, by tightening supply, larger in others.

Research Context

The correlation at the centre of this report is not seriously disputed. Across disciplines, countries and decades, health tracks housing security: it worsens when a home is lost or falls under threat, and it improves when housing improves. Epidemiology, developmental psychology, environmental health and the economics of neighbourhood all point the same way, from independent data.

The causation is the contested part, and the reason is always the same. The people who get evicted are poorer, sicker and more stressed before the eviction, so some of the harm that looks like a consequence of the move is really a property of what forced it. The evidence that survives this objection is the evidence that breaks the link between the two. Leifheit’s moratoria are an external shock. They do not depend on which tenant they reach, and they move distress. Webb’s Danish risk shows up across every income band, so it is not only poverty talking. Howden-Chapman and Moving to Opportunity are randomised, so selection cannot be the whole story. Of the four, only Leifheit tests the eviction threat itself; Howden-Chapman and Moving to Opportunity randomised housing quality and neighbourhood poverty, and Webb remains observational. Together they narrow the confounding objection from different angles. The rest adds correlational support.

The honest counter sits in the same literature. Sarah Burgard and colleagues looked specifically at eviction in a Michigan panel and found that, after adjustment, eviction on its own carried no significant health association, even though homelessness and rent arrears did (Burgard, Seefeldt and Zelner, 2012, Social Science and Medicine). Talmatzky and colleagues, in a 2026 systematic review, graded the overall certainty of the housing-insecurity and mental-health field as low to very low, precisely because so much of it is observational (Talmatzky et al., 2026, PLoS One). And there is a mechanism gap. The pathway from a notice through chronic stress and lost sleep to physical illness is biologically plausible and widely assumed, but it has not been measured directly in evicted people. The pathway is plausible, not proven. Kim McKee and colleagues have at least described the lived version of that pathway, the way insecurity and cost keep low-income renters from putting down roots (McKee et al., 2020, Housing Studies), and the UK Collaborative Centre for Housing Evidence has set out the same case in its work on tenure insecurity.

The market-economics research belongs here too, but as the mechanism that sets the exposed group rather than as health evidence. Rebecca Diamond, Tim McQuade and Franklin Qian showed that rent control in San Francisco cut rental supply by 15 per cent and pushed citywide rents up 5.1 per cent, while raising the chance that sitting tenants stayed (Diamond, McQuade and Qian, 2019, American Economic Review). Scotland ran the nearer experiment. Its 2022 rent cap applied only inside a tenancy, and Zoopla recorded new-let rents rising 12.7 per cent in the nine months that followed, as landlords lifted rents between tenancies instead. What those studies establish is who ends up moving and searching. That is the population on which the health evidence then acts. One caution belongs on the record. Pacitti, in the Resolution Foundation report this analysis draws on for rent trends, concluded that the landlord-exodus narrative was not well supported by the data at the time (Pacitti, 2024), which is a useful brake on reading the Savills sell-off as destiny.

What is missing is the obvious study. No one has yet measured the human outcomes of the Renters’ Rights Act in England, because the Act is weeks old. The data to do it sits with CaCHE, the Joseph Rowntree Foundation and the housing centres, and the cross-map of that data against the next year of evictions, moves and rents is the study still waiting to be done.

Implications

For investors and operators in residential property. The human-outcome evidence changes what tenure stability is worth. A long secure tenancy is the variable that the wellbeing and health studies actually move, beyond yield or void risk or reputation. That gives build-to-rent and longer leases a measurable human payoff the classic short-let buy-to-let does not.

Not every long lease earns the effect this evidence describes. Leifheit traced the effect to moratoria that blocked the notice and the court filing, not only the final bailiff step; weaker protections did nothing measurable (Leifheit et al., 2021, JAMA Network Open). A build-to-rent product that markets security but still allows notice on thin grounds is closer to the weak case than the strong one. The human-outcome payoff attaches to tenancies that remove the standing threat of notice. Term length on its own is not the mechanism.

For policy and for landlords. The reform’s health benefit is real but conditional. It reaches a household only if that household keeps its home. The sale ground and the pricing of new lets are the two leaks, and if they simply move the insecurity from the protected sitting tenant to the renter who has to find somewhere else, the reform will show the same redistribution as above: an improved average masking a sharper harm at the edge. The Scottish sequence is the warning. Scotland’s 2022 cap protected the sitting tenant but left new lets uncapped, and new-let rents rose 12.7 per cent in the following nine months.

The leak has a rough size already. SpareRoom surveyed landlords and found 43 per cent of evictions since Royal Assent attributed to a sale (SpareRoom landlord eviction survey, October 2025 to April 2026), against a background of 697 buy-to-let sales a day (Savills Research, April 2026). Whether that group experiences the same distress the reform was built to prevent is not yet measured for England specifically. CaCHE and the Joseph Rowntree Foundation hold data that could run the cross-map against the next year of evictions, moves and rents. Commissioning that comparison is how a policymaker would actually know whether the reform reduced harm or moved it, more reliably than watching the homelessness-threat count fall on its own.

For anyone weighing whether the Act worked. The market numbers and the health numbers have to be read together. A falling count of Section 21 homelessness threats looks like success on its own. Set against a rising count of sale-driven moves and a tightening supply, it may only mean the same redistribution again, the harm moved to a different household rather than reduced. The next twelve months of English data will show which of the two it is. The measure worth watching is the count of sale-driven moves set beside the health outcomes, with the rent index only one line in it.

The Burgard caution should apply to future readings of this reform as much as to the studies already cited here. Burgard and colleagues found that a raw association between eviction and health washed out once they adjusted for what preceded the eviction (Burgard, Seefeldt and Zelner, 2012, Social Science and Medicine). A first study claiming the Act improved population health should show its work on that same adjustment, prior income, prior health, prior housing stability, before the finding gets credited to the reform itself.