The Cost of Letting People Fall

The phone needs charging. The appointment reminder is on the phone. The address requested by the form is an address you no longer have.

From a distance, the task looks simple: attend the appointment. Up close, it has become a chain of smaller tasks, each requiring something that disappeared with the room. Somewhere to wash. Somewhere to keep the document dry. A reliable way to arrive. Enough sleep to understand what is being said.

The loss of a home does not end when the person leaves the building. It can remove the place from which the rest of life was organized.

We should remember that when a person in severe difficulty is described as failing to make use of an opportunity. An opportunity has conditions of entry. If we ignore those conditions, we can build a service that looks generous from the office and remains difficult to reach from the street.

This chapter concerns what happens when ordinary foundations give way, and what we owe one another before the damage becomes a permanent identity. It is also about evidence. There are people who see every hardship as an individual failing, and people who make every hardship the automatic result of one economic system. Neither habit is adequate to the people whose lives are at stake.

A humane account has to be more exact.

The count and the people it cannot contain

HUD estimated that 771,480 people experienced sheltered or unsheltered homelessness on a single night in 2024, an increase of 18.1 percent from the previous year’s estimate. The count includes people in specified shelter settings and places not intended for habitation. It excludes many other forms of unstable housing, including temporarily staying with friends or relatives because there is nowhere else to go.¹

This is a one-night estimate, not the number of distinct people who experienced homelessness during the entire year. Local counting methods and the difficulty of locating people limit its precision. We should not relabel it a verified count for 2026 merely because we are discussing it now.

Even with those limits, the scale demands attention. Each entry concerns a human being who needs somewhere to live. A count is necessary for planning, but a person can vanish inside it. We see the total and forget the task of finding a bathroom before work, keeping medication secure, or getting a child ready for school.

Nor does the count explain itself. Several pathways can end in the same visible condition. Income can be inadequate. A relationship can break down. A person can leave violence, lose employment, develop a serious illness, exit an institution without a secure destination, or face a housing market with too few accessible options. These possibilities can overlap. Establishing their prevalence and their causal weight requires more than looking at somebody on a pavement.

We should stop asking a single photograph to supply a social theory.

A loss that can deepen the loss

It is tempting to assume that poverty causes eviction and then end the explanation there. Research suggests that an eviction can itself make the situation worse.

Robert Collinson and his coauthors studied eviction cases in New York City and Cook County, linking court records with administrative data. They used differences in the tendency of quasi-randomly assigned judges to order eviction as a way to estimate its effects. In the study, an eviction order increased emergency-shelter use in the year after filing by 3.4 percentage points, compared with a 0.9 percent mean for the nonevicted group. The analysis also found harm to earnings and financial outcomes.²

The causal interpretation applies under the study’s assumptions to cases whose outcomes are affected by judge leniency. Shelter records do not capture every form of homelessness. These are results from two settings, not a coefficient we can attach to every eviction in the country.

Still, they help correct a consequential mistake. A person can arrive at court already in distress, and the outcome can add another layer of harm. The sequence is not adequately described by saying the person was poor to begin with.

This should change how we think about prevention. If keeping a viable tenancy stable can avoid further damage, then the relevant comparison includes what happens after the housing loss. That does not imply that every tenancy can or should be preserved in exactly its present form. A dwelling may be unsafe. A household may need to move. A landlord may have legitimate claims. The aim is a secure housing outcome, with fair treatment of the parties, rather than the ritual preservation of any arrangement simply because it exists.

A system capable of helping early would need to discover trouble before the last possible moment. It would need accessible advice, a way to resolve mistakes, and resources for cases in which money is the immediate obstacle. It would also need somewhere for people to go when the existing situation cannot be repaired. A referral to an unavailable home is still an unavailable home.

When a public intervention changes the result

HUD’s Family Options Study randomly assigned intervention offers to 2,282 shelter-recruited families in 12 communities. At roughly 37 months, the composite housing-instability rate was 17.0 percent for families offered priority access to a permanent subsidy, versus 38.1 percent under usual care.³

The composite meant at least one night homeless or doubled up in the previous six months, or emergency-shelter use in the previous twelve months. These findings concern offers to the families studied; they are not estimates for every population experiencing homelessness.

The housing result deserves attention on its own terms. Earlier improvements in alcohol-dependence and combined alcohol/drug-problem measures were not detected at 37 months.⁴ We should not demand miracles from a useful intervention.

I want that precision because it helps us invest intelligently. If an intervention does the job of stabilizing housing, allow it to be good at that job. Do not require it to solve every difficulty before it is worth funding. Equally, do not use its success on one outcome to declare all the other work finished.

A stable room may be the beginning of recovery, study, employment, or a quieter day. Those possibilities are worth creating even when they do not all appear at once in a follow-up table.

The person before the explanation

Substance use deserves honest discussion. It can create serious harm. So can the absence of treatment. It can be involved in the events preceding housing loss, and a person’s circumstances after housing loss can affect use, risk, and access to care. A responsible account allows these pathways to intersect without turning them into a stereotype.

The University of California, San Francisco’s statewide study of adults experiencing homelessness interviewed people across California and asked about their experiences before and during homelessness. Thirteen percent cited their own substance use as a reason for leaving their last housing. Among participants using specified drugs regularly or reporting heavy episodic drinking, one in five wanted treatment but could not obtain it. These are retrospective self-reports from a California study, not national causal estimates.⁵

The research is useful partly because it asks people what happened, rather than treating visible difficulty as a diagnosis. Even then, a report of use, a clinical substance-use disorder, an overdose, and a housing problem are different things. We should not exchange one for another when the sentence becomes more dramatic that way.

A person also has a history that is larger than the problem being studied. They may be a parent, a mechanic, a neighbor, an artist, a person who is funny when they are not exhausted, or someone whose abilities have been hidden by a long emergency. None of those possibilities cancels the need for treatment or accountability. They do remind us that a human life is not an entry under a service category.

I would like public systems to become very practical about this. If someone wants treatment, what prevents access? If a service requires documents, transport, or communication, how will those be supplied? If someone is frightened of a setting, what safe alternatives exist? If a housing placement fails, what is learned before the next attempt?

These are harder questions than whether we feel sympathy. They require institutions that can respond to a real person rather than an ideal applicant.

An Address

The letter has a place to go.
The coat can stay upon its hook.
A cup waits where you left it.
No one has moved the book.
Tomorrow has a starting point,
A key beside the door.
You can put one burden down
And think of something more.

What economic harm can help explain

There is evidence that economic dislocation can contribute to substance-related harm, although the relationship is neither automatic nor complete.

A study by Atheendar Venkataramani and colleagues examined automotive plant closures and opioid-overdose mortality in manufacturing counties. Comparing changes in affected and comparison areas, the researchers estimated a differential increase of 8.6 opioid-overdose deaths per 100,000 working-age adults five years after a closure. The confidence interval was wide, and the design remained observational.⁶

The outcome was mortality, not the number of people developing addiction. Exposure was geographical; the researchers could not identify the result as the fate of individual workers who lost their jobs. Drug supply, treatment access, other local changes, and residual confounding remain relevant. The study supports economic opportunity as one contributor worth taking seriously, not an explanation for every death.

Why does this matter for a book about affordability? Because a society’s economic arrangements can affect more than its shopping basket. The availability of a plausible future, the resources to manage disruption, and the means to obtain help belong in the investigation of wellbeing.

I am careful with the word despair. It can describe a genuine human experience, but it can also become a story imposed on people whose actual experiences were never measured. A death certificate does not tell us the whole inner life. We can acknowledge severe economic distress without pretending to know every person’s motive.

The practical implication is to broaden prevention. Protect access to care. Reduce exposure to dangerous conditions. Make recovery support reachable. At the same time, take seriously the loss of work, stability, and opportunity in a community. These approaches need not compete for a single explanation of the person.

A government should be capable of treating a health emergency and asking why the surrounding life became so hard to sustain.

Alcohol belongs in the picture

Our cultural reactions to drugs are inconsistent. A familiar legal product can seem less alarming than an unfamiliar illegal one even when both are associated with serious harm. Alcohol therefore belongs in this discussion, with its own evidence and without being folded into a vague category of bad choices.

CDC researchers estimated an average of about 178,000 U.S. deaths annually attributable to excessive alcohol use in 2020–2021, up from about 138,000 in 2016–2017. The estimate includes chronic conditions and acute causes; it is not simply a count of alcohol poisonings. It uses methods for estimating the portion of deaths attributable to alcohol.⁷

These are historical-period estimates. They do not establish that economic insecurity caused those deaths. They should not be added indiscriminately to drug-overdose totals, because categories can overlap. Their importance is that substantial harm can exist within familiar routines and legitimate commerce, rather than only in the places that make people uncomfortable to look at.

We also need to recognize improvement where it occurs. In May 2026, CDC released provisional estimates of 69,973 drug-overdose deaths in 2025, compared with 81,313 in 2024, the third consecutive annual decline. The figures can be revised, and the release does not isolate the causes of the improvement.⁸

That is good news. It does not erase the remaining deaths. It does reject a lazy story in which every harm rises without interruption and every intervention must therefore have failed.

The purpose of evidence is to help us discover what makes a life safer. We should welcome a decline and work harder to understand how to sustain it.

The dignity of an effective response

A humane policy must care about the people affected by dangerous conduct as well as the people in difficulty. A neighbor should not have to accept violence, threats, or an unsafe public space as the price of compassion. A person without a home also needs protection from those harms. Treating these concerns as opposing moral camps makes effective action harder.

The response should fit the problem. Immediate danger can require an immediate safety intervention. A medical crisis needs appropriate care. A housing need requires a route to housing. The categories can overlap, but one institution should not be expected to manufacture every missing service by force of personality.

This is where the cost of underinvestment becomes visible as a design problem. If the only reliably available response is an emergency response, situations that might have been addressed earlier reach that system late. To establish the size and causes of such a pattern in a particular city, we would need local evidence. As a principle of design, though, it should concern us whenever the front door to help opens only after a person has fallen far enough.

I want more doors before that point. They should be easy to find, hard to lose through an administrative mistake, and accountable for what happens after referral. The measure of success should include stable housing, access to needed support, safety, and the person’s own account of whether life has become more manageable.

There will be setbacks. Recovery is not a performance staged to reassure the taxpayer. Programs should investigate failure, improve practice, and distinguish a temporary difficulty from a permanent verdict on the person.

We cannot justify every program with the promise that it will save more money than it costs. Some valuable help will require a net public contribution. The moral case should be able to stand in daylight beside the fiscal case. A society can decide that leaving a person without basic protection is unacceptable even when the accountant cannot book a profit from the alternative.

People do not have to become inexpensive before they deserve a place to live.

The next chapter asks what this principle means when public money, technology, and the desire for safety meet. We should want streets on which people can live without fear. We should also want a society in which fewer lives reach the point of crisis. Those ambitions belong in the same budget conversation, with evidence for each and dignity throughout.