Why the Homeless Crisis Keeps Getting Worse in the US

On a night in January 2024, a team of outreach workers and volunteers counted over 8,000 homeless people in San Francisco, nearly 600 more than when the city last counted them two years earlier.

It’s a high number to be sure, though I don’t know if anyone was exactly shocked given the city’s reputation.

In that same year, San Francisco adopted a budget of over 800 million dollars for its homelessness programs.

A simple division reveals that the city would have been spending more than 100,000 dollars per homeless person, which is higher than the GDP per capita of the entire United States.

This wasn’t unique to 2024 either. People have long noticed that San Francisco was throwing a hilariously large amount of money at a problem that kept getting worse.

When the city was asked about this back in 2016, they said it was because a large part of their homeless budget did not go to homeless people but to those in permanent housing.

I don’t know if that made you feel any better. In any case, San Francisco was not alone. Homelessness is a problem that permeates the entire United States.

It kept getting worse over the last decade, especially in recent years, to the point that homelessness is now known internationally as an integral part of American culture.

The richest country on Earth apparently can’t figure out how to house its own people, no matter how much taxpayers’ money it throws at the problem.

The apparent discrepancy here confuses me. What’s more confusing are the explanations for it.

An online search reveals a wide range of answers all purport to be the truth.

People say a lot of different things. What they never seem to do is agree.

If only someone could take a sane look at the data and attempt to find the throughlines among the different perspectives.

Well, it’s funny you ask. I figure I’ll take a crack at it today.

I am just a YouTuber, so you probably shouldn’t believe everything I say.

So, over the course of this video, I’ll walk you through the research and data I’m looking at and how I am parsing them. By the time I get to the conclusion, you can keep me honest about whether I’ve been biased or form your own opinions.

Now before we do anything fancy, let’s start with the most basic point.

Homelessness is, by definition, the condition of people who don't have a home, presumably because they can't afford one. It is natural to assume, then, that homelessness is caused by a lack of housing affordability.

But does the data support this rather obvious assumption? It does.

Researchers Colburn and Aldern did the simplest version of this test. They took the median rent of various places in America and plotted it against how much homelessness those places experience.

The result is a clean correlation: the higher the rent in a particular region, the higher its rate of homelessness.

They also tried another measurement called the “first rent quartile,” which roughly measures how expensive the cheapest housing in a region is. If you are poor and at risk of homelessness, those are the housing units you would go for, so it is potentially more relevant.

They found the same correlation: the higher the first rent quartile, the higher the homelessness rate.

This finding has been widely corroborated by other research using different statistical methods and data samples.

In particular, when Pew researchers tracked American cities from 2017 to 2022, the cities where rent rose the fastest were also the cities where homelessness rose the most. The cities where rent barely moved saw homelessness fall.

In Raleigh, North Carolina specifically, over the course of five years, its rent jumped about a third, and homelessness shot up more than 60%.

Then Raleigh built more housing, causing rent to stop climbing. Subsequently, homelessness dropped by a third over the next two years.

In other words, when rent goes up, homelessness goes up. When rent gets under control, homelessness comes down.

So far, the story of why the homeless crisis keeps getting worse seems to just be rent.

In other words, housing has become increasingly less affordable in the US over the years. And that's what seems to be driving homelessness.

Now, a caveat. This is a theory based on the data. The data itself is correlational, not causal. In other words, rent and homelessness are two variables that move together, but this is not definitive proof that rent causes homelessness. To get definitive proof, we would need to run controlled experiments, which would be impractical in this case. We can't arbitrarily raise the rent of an entire city just to see how it changes homelessness.

But this is as clean a dataset as I can get that allows me to theorize about causation.

The fact that rent and homelessness move together across geography and time makes rent the cleanest driver of homelessness that the data reveals.

But unfortunately that's also where the clean data ends. I was surprised to learn this, but rent, or housing affordability more generally, seems to be the only variable that can reliably predict homelessness. It appears that nothing else can do it.

Colburn and Aldern investigated this pretty thoroughly. They tested everything that I can think of that could conceivably be related to homelessness.

They analyzed poverty, unemployment, mental illness, drug use, welfare, migration, race, and even weather.

With the exception of rent, nothing else they looked at produced any meaningful statistical relationship with the homelessness rate.

This was surprising to me because, for example, I would have thought poverty would be relevant, but it isn't according to the data. Cities with high poverty seem to have low homelessness so long as their rents remain cheap.

I also thought drug abuse would correlate, but the data doesn't show that either.

As a sanity check, according to the CDC, the American state with the highest drug-overdose death rate in 2023 was West Virginia.

Yet its homelessness rate was very low because it is cheap to rent there.

By comparison, California's drug-overdose death rate was actually below the national average in 2023. So it was doing well in the drug department relative to the average.

Yet it had one of the worst homelessness rates in the country, because, again, rent is very expensive there.

But at this point, some of you might feel like you are getting gaslit by the data.

When we see homeless people on the street, especially in America, they don’t usually look like your average Joe who just happens to be unable to afford rent. It doesn’t look like what they need is rent money. It looks like they need a different kind of help.

The situation on the street looks much more dysfunctional than a simple rent affordability problem would suggest.

This is why many people believe that drug use and mental illness are causing the homelessness crisis: that explanation maps more accurately onto what people see on the street.

Additionally, a rent-affordability problem sounds fundamentally like a money problem. Logic suggests that a money problem can be solved with money. But the US has been throwing money at the homelessness problem for years, and it hasn’t gotten better, so what gives?

These are good objections. To square these circles, we need to distinguish between two types of homelessness.

The first group consists of people who are pretty much just your average Joes who happened to be unlucky. Maybe they lost their jobs, received large medical bills, had flat tires without enough savings, or experienced a series of these unlucky events. In any case, shit happened: they could no longer pay rent, had no backup, and had to figure out how to sleep in their cars.

This group is often referred to as “transitional homelessness” in the literature.

They make up the majority of the homeless population in America. One study puts their proportion at about 80%. At the same time, they are not very visible, meaning you don't always see them.

People who are experiencing transitional homelessness are probably as uncomfortable about their situation as anyone can be. If seeing a homeless person makes you uncomfortable, trust me, they are much more uncomfortable about it than you are—enough so that they typically make efforts to hide it.

They might stealth-camp in their cars in ways that don't draw attention. They might try a homeless shelter if one is available. If they must sleep in a public space, they might pick a well-hidden spot for their own safety, and sometimes because they are embarrassed.

They also tend to exit homelessness quickly, as the name “transitional” suggests. As you can probably imagine, many of them are quite motivated to get off the street. In fact, that is probably the single most important priority for many of them while they are experiencing homelessness. Most of them do get off the street within weeks to a few months.

The second group is called “chronic homelessness.” This group is much more visible. If you ever encounter a homeless person on the street in America, you are probably encountering someone in this group.

Those shocking images and footage of homelessness in America that go viral on the internet typically come from this group.

This is where things like addiction, drug abuse, untreated mental illness, disabilities, and other risk factors are common.

Chronic homelessness is a much smaller category in America. Estimates put it at about 10% on the low end and about 20% on the high end.

As the name suggests, people in this category tend to stay homeless much longer. In fact, many of them may be on the street for years.

For accuracy, the reality is probably more like a continuum. It is not so much that a homeless person is either transitional or chronic. In reality, there is probably a spectrum, with transitional homelessness on one end, chronic homelessness on the other, and an individual might fall somewhere along that line.

In some literature, there is also a third category called “episodic homelessness” that exists between transitional and chronic homelessness. These are people who may exit homelessness but later fall back into homelessness again, possibly experiencing multiple homeless episodes throughout their lives.

So treat these categories as mental models to understand the phenomenon, but remember that reality is more nuanced.

I won't overcomplicate things for this video. For our time together, all you need to remember is the distinction between transitional and chronic homelessness.

This distinction explains why the data about homelessness doesn't seem to fit our impression of it.

The data tells a story of rent, while our impression seems to suggest something more dysfunctional.

That is because, for those in transitional homelessness, it is a rent problem. The transitionally homeless are the majority of the homeless population. That is the story the data is telling. So the data is not wrong. It is explaining what homelessness primarily is.

But despite being the largest portion of the homeless population, transitionally homeless people are not very visible, as we said. In other words, the story the data suggests is the main story, but it is not the one we see.

What we see on the street is chronic homelessness. So our impression of homelessness comes from that group, even though it is a much smaller part of the homeless population.

This is the Pareto Principle in a way: a minority of the homeless population produces the majority of the public impression.

People think homelessness is a drug and mental-health problem because that is the only slice of homelessness they see.

That impression is not necessarily wrong. It is just not precise.

Homelessness is a drug and mental-health problem, but for the chronic tier, the small minority. Also, for the chronic tier, it is not just drugs and mental health but also disability and other risk factors.

These risk factors don't increase the overall scale of homelessness. That is driven by rent. But risk factors do drive homelessness on an individual level. In other words, they don't predict how many people become homeless, but they do predict who becomes homeless.

So when we see the overall rate of homelessness in a region go up, what is driving that increase is more likely to be expensive rent. But when rent becomes expensive, those with risk factors are more vulnerable to that disruption and more likely to become homeless as a result.

The causality here can go the other way too.

It is true that when someone has risk factors, they are more likely to become homeless and get stuck on the street.

But when someone is stuck on the street for a long time, that can cause them to develop or obtain risk factors that perpetuate their circumstances too.

Being homeless is traumatic: it is sleepless, dangerous, freezing, and humiliating. When someone is stuck in that state for years on end, that is the kind of thing that could create something like an addiction problem or a mental illness as much as it results from one.

This is also why simply throwing money at the homeless problem hasn't fixed what appears to be a money problem.

It is only a money problem for one of the two groups.

For transitional homelessness, it is more or less a money problem. The reason they are homeless in the first place is because they don't have money. The data suggests that if you give them money, they will probably stop being homeless.

There are controlled experiments that test this mechanism.

The best one came from Project New Leaf.

They screened out the transitionally homeless people from the homeless population and basically just handed them cash. The result was that those who received cash got off the street faster than those who did not. They might have gotten off the street so much faster that they ended up saving the government money by not using homeless services.

This experiment was done in Canada. Whether it can be extrapolated to the US depends on how structurally similar the two countries’ homelessness problems are.

I think they are similar, although Canadians would probably get offended if you tell them they are similar to Americans. I don't know how much of a threat you'd consider an angry Canadian to be.

In any case, I would personally want to see this experiment replicated a few more times in America before I would consider it conclusive.

But even if transitional homelessness is just a money problem, we won't be able to solve homelessness by throwing more money at it because of the chronic tier.

Chronic homelessness is not a money problem. If someone is suffering from drug abuse, mental illness, or other forms of debilitating risk factors, simply giving them money is probably not going to help. Chronic homelessness requires a different type of solution, such as dedicated programs offering the specific kind of help they need.

That is what we try to do in practice. In fact, much of the American homelessness budget goes into programs built for the chronic tier, such as supportive housing with full-time staff on site, addiction and mental-health treatment, caseworkers who manage each person's recovery, long-term medical care, and similar services.

The Pareto Principle kind of applies here again. The chronic tier, as a small part of the homeless population, consumes a much larger share of homelessness programs.

There is a bigger conversation here about how to fund, create, and implement effective homelessness programs. For now, I just want to highlight the complexities involved.

Homelessness is not just one problem. Transitional and chronic homelessness, for example, have different needs that require different solutions. There are further complications within each category.

So this is why simply throwing money at homelessness won't solve it.

In closing, I would like to mention a few popular explanations for homelessness that we often hear in media and assess them on their merits in a rapid-fire fashion.

First, you'll hear inequality mentioned as a driver of homelessness a lot.

Inequality is a very big problem in America. It is not a direct driver of homelessness, but it can have an indirect impact that is worth exploring.

Inequality is the gap between how much money poor people make and how much money rich people make.

That is not the gap homelessness cares about.

The gap that is relevant to homelessness is the gap between how much money poor people make and how much rent costs.

Theoretically, the gap between the poor and the rich can be as wide as the distance from the Earth to the Moon, but as long as the poor can afford housing, homelessness would stay low.

But in practice, there can be an indirect relationship here that is connected by rent.

For example, if inequality takes the form of a lot of rich people congregating in a particular region, it can bid up rent in that region. The increased rent would then cause more homelessness. San Francisco is an example of this.

But if the city manages to keep up its housing supply when rich people come and does not let rent get out of control, homelessness would stay low. Austin, Texas, would be an example of that.

Another cause of homelessness you might hear is immigration.

Immigration, specifically asylum seekers, did directly add to the homelessness numbers.

It is not a driver of homelessness in the sense that it is not the reason homelessness keeps getting worse over the years, but it is a reason why homelessness got worse once.

Between 2023 and 2024, the number of homeless people went from 653,104 to 771,480, up about 18%, which was the largest one-year jump on record.

A large part of that increase consisted of people who came to America to seek asylum.

This influx of migrants was the result of a series of unfortunate geopolitical events across a few countries, mostly in Latin America.

The sheer number of asylum seekers overwhelmed the American immigration system, which was not prepared for this scale. Subsequently, many asylum seekers ended up in homeless shelters, especially in these four sanctuary cities.

I am omitting a large amount of detail here. The full immigration story requires a separate video, or frankly a separate documentary.

I won't do that right now. But to parse the part that is relevant to homelessness:

The fact that they were in homeless shelters to begin with was already atypical. They were never supposed to be in homeless shelters in the first place. It was not standard procedure to send asylum seekers into homeless shelters. They were there almost as an ad hoc resort by an overwhelmed immigration system.

In any case, these asylum seekers were counted as homeless people in shelters and ended up in the homelessness statistics. But it was a failure of the immigration programs rather than the homelessness programs.

Since it only happened once, it is not a systemic force that continually drives up homelessness across time. In other words, we don't have large numbers of asylum seekers constantly coming to the US to live in homeless shelters.

It is a one-time thing.

And it is already reversing. I'll end with this since it looks like good news.

In 2025, national homelessness fell for the first time in almost a decade.

A large part of that drop came from asylum seekers leaving the shelter system.

So the good news is that the homelessness number dropped. The bad news is that the homelessness problem is probably still here to stay.

We don't know for sure. This is just one data point, and it is very recent. So these are just predictions.

But from the data I'm seeing, the homelessness problem doesn't look to be getting substantially better.

If we look into the breakdown of the homelessness count, what decreased last year were families, driven by asylum seekers leaving the homeless shelter system. However, chronic homelessness actually hit a record high.

Additionally, I'm not seeing evidence that the structural driver of homelessness is getting better, that being rent.

In fact, it looks like rent has gotten worse. Median rent rose by 4.8% nationally in 2025.

But I do hope I'm wrong. One year is not enough to establish a trend in statistics in any case. So I hope the drop continues, and when I check the stats in a decade, I would love to see that homelessness finally peaked in 2024 and has been on a downward trend from there. We'll see how that turns out.

Alright everyone, that's it. Check out my community for the vibes and the newsletter for the scripts, both linked below. Thank you. See you.