Anyone who's visited downtown San Francisco has seen streets like this one:

The US is the richest country on Earth, excluding small states like Singapore with <10 million people. And the Bay Area is one of the US's economic hubs. Why do we see stuff like this? I'll try to break it down quickly. 

First, San Francisco is obviously an expensive place to live. Do housing costs predict high homelessness?

They do! R^2 = 0.69 is really good. (This is a regression on US states, data from here and here. Washington, DC is a high outlier; removing it raises R^2 to 0.73. High outliers include DC, Washington, Nevada, Alaska and New Mexico; low outliers include Maryland, New Jersey, Connecticut, Rhode Island and Virginia, although the first might be an artifact of DC being outside state borders.)

SF's housing costs are 3.3x higher than the US average. Plotting it on this line, we should expect a homelessness rate of 0.7%, 5.8x higher than a city with normal rents. The actual rate is 0.9%, another 1.3x higher than we'd expect compared to rent.

However, there are two other big factors to consider. #1 is that San Francisco has a very strong concentration of homeless people downtown. Here's a breakdown by supervisor district:

Districts are drawn to have equal populations, so District Six, which includes a lot of downtown buildings and office space, has a 5.1x higher rate of homelessness even compared to SF's average. Most cities don't have that. Eg. New York City has roughly the same homelessness rate as SF, but homelessness is more evenly spread through the city; and the worst 10% of neighborhoods are split between the Bronx, Brooklyn, and Queens, instead of being concentrated downtown. 

Factor #2 is homeless people who are sheltered vs. unsheltered. Unsheltered homeless people sleep outside or in vehicles, so they're much more visible. 68% of San Francisco homeless are unsheltered, compared to 25% for the US outside of California. That's another factor of 2.7x.

Summing it up, we have:

Cause of visible homelessnessFactor ofPercent of all causes
SF has higher housing costs5.8x38%
SF has more homeless, adjusting for rents1.3x6%
Homelessness concentrated downtown5.1x35%
SF homeless are more unsheltered2.7x21%

(Note that, since these causes are multiplicative, percentage here is calculated by first taking the log, then dividing.) So, in total, visible homelessness is two orders of magnitude higher in SF than the US average, even though SF as a whole only has 30% more than you'd expect compared to rents.

Many assume that cold weather is a factor in homelessness, but I didn't find any relationship there:

(weather data from here)

although it probably plays a role in whether homeless people are sheltered or not:

so including San Francisco's mild winters as a factor in unsheltered homelessness, we get this table:

Cause of visible homelessnessFactor ofPercent of all causes
SF has higher housing costs5.8x38%
SF has more homeless, adjusting for rents1.3x6%
Homelessness concentrated downtown5.1x35%
SF homeless are more unsheltered (weather)1.6x10%
SF homeless are more unsheltered (other)1.7x11%

I also found that, controlling for rents, the partisanship of a state did not predict homelessness (using the Partisan Voting Index):

Lastly, many people mix together homelessness and crime - assaults, harassment, car break-ins and so on, although many homeless people are not criminals and vice-versa. San Francisco has a dramatically higher rate of property crime than any other county in California, combined with fewer arrests than most places:

The trend with violent crime is a bit less extreme, but still there:

(Data from here; note that this data is from 2014-16 and predates many recent controversies.) High homelessness, high crime, and a low arrest rate probably combine to create an overall, ambient feeling of danger in SF, more than any one factor by itself.


34 comments, sorted by Click to highlight new comments since: Today at 4:07 AM
New Comment

The US is the richest country on Earth, excluding small states like Singapore with <10 million people. And the Bay Area is one of the US's economic hubs.

To highlight out how absurdly rich the Bay Area is, the BEA states that the San Francisco-Oakland-Berkeley Metropolitan Statistical Area had a total GDP of $588,335,543,000 as of 2020, with a population of 4,739,649 as of 2020, according to the Census Bureau. This means that the GDP per capita of the SF metro area was $124,130 in 2020, higher than Singapore, Norway, Ireland, and Switzerland.

But it gets crazier. If you look at just San Francisco county (which basically just includes the city), then its GDP is $201,547,346,000. With a population of 870,014, that means the GDP per capita of San Francisco is a mindboggling $231,660, richer than Luxembourg, Monaco, and Liechtenstein, and every other micronation for that matter.

The idea that solving homelessness in the Bay Area is a matter of just making the city richer, is false. There’s more than enough money.

Many people commute to work in businesses in San Francisco who don't live there. I would expect GDP per capita to be misleading in such cases for some purposes.

Broadening to the San Francisco-San Jose area, there are 9,714,023 people with a GDP of $1,101,153,397,000/year, giving a GDP/capita estimate of $113,357.  I know enough people who commute between Sunnyvale and San Francisco or even further that I'd expect this to be 'more accurate' in some sense, though obviously it's only slightly lower than your first figure and still absurdly high.

But the city of San Francisco likely has a much smaller tax base than its putative GDP/capita would suggest, so provision of city based public services may be more difficult to manage.

Note that
A) zooming in on most city hubs will find you monetary concentrations like this, e.g. Manhattan has a GDP pc of $370k
B) I have never actually heard anyone argue that making the city richer is the path to solving homelessness despite living there for a long time, so suspect this might be an error—are you conflating this with deregulating the housing market? Or do people actually argue somewhere that more money would solve homelessness?

From what I understand, there are two fundamentally different types of homelessness. One is when someone "normal" is just down on their luck and who then tends to live with their friends until they get a new job or something. The other is when someone is severely mentally ill, to the point where it seriously interferes with their functioning in society, which leads to the sorts of homelessness you see on the streets.

When you analyze the correlation between homelessness and housing prices, are most of the cases of homelessness of the first type or of the second type? Is the association equally strong for both types?

Yeah I’m surprised to see this posted without mention of the Michael Shellenberger work on this topic. I actually think the above analysis hints at it: Why is there so much homelessness concentrated downtown here?

The elephant in the room is that many people who have hard drug habits prefer to live in encampments in central cites, and the authorities in the Bay Area (especially SF) are supportive of them doing so. 

There are also a lot of people with housing insecurity due to the challenges of affording rent, but those are not the people living in the visible encampments.

Thanks for writing this!

I've been thinking about homelessness/crime/related social dysfunctions in SF a good bit lately, as I think about moving to Berkeley, and its good to have some hard data put together in one place with relevant context. I've noticed that all I have to go off of on the wider internet, unless of course I go and directly do the research myself, is adversarially selected anecdotes that play into the red and blue political tribe agendas, and I'm epistemically afraid of updating on information selectively filtered like that. This is also probably the single best concise stab at explaining causes of SF's homelessness I've read, as explanations are likewise epistemically distorted by the existence of political tribe pressures -- I've now updated towards SF zoning laws being the chief cause of SF homelessness.

Can you explain how you got from this post to zoning laws? (Not complaining, but I don't see the link and I'm curious)

Almost certainly what he means is: restrictive zoning leads to small amounts of new housing, which leads to high rents, which according to this essay we just read, leads to high homelessness.

Zoning laws restricting housing supply, permanently driving up housing prices in turn, is one of my hypotheses explaining disproportionate SF/large US city homelessness.

Moderate California weather plus some busing of homeless people by other state governments is another.

Based on the above post, I updated in favor of the former hypotheses against the latter.

Some potential notes about "busing of homeless people by other state governments"

San Francisco periodically does a "Point in Time" survey of the homeless population. The last one was in 2017, with an N=1089. 

  • 69% reported they were living in SF when they became homeless 
  • 21% said they were living in California (but not SF) 
  • 10% said they were living out of state

However, San Francisco itself has its own bussing program called "Homeward Bound", which may make the 10% who said they were living out of state hard to interpret. 

There was another Point in Time count in 2019, and more recently a preliminary report for the 2022 count was published: 

SF’s housing costs are 3.3x higher than the US average. Plotting it on this line,

This feels dubious to me. I think there's no a priori reason to assume the correlation is similar for cities as for states.

A full-on Simpson's paradox with reversed correlation seems unlikely, since I guess rich cities are typically in rich states. But I could easily imagine something like: "the housing costs of the state are basically irrelevant, the housing costs of the city matters a lot. When you bin cities into states, you're introducing noise that makes the correlation seem weaker than it is". And I wouldn't be completely shocked by something the other way around, where the housing costs of the state matter a lot more than the housing costs of the city. In which case SF would only get the same multiplier here as anywhere else in California.

Still, if this is the data you had, plotting it on this line seems clearly better than not doing so.

I'm interested to see how Prop C changes things. Proposition C was passed in 2018 and taxes companies that made over 50 million dollars a year so that they contributed money into a fund specifically for ending homelessness. Normally the city budget includes money to create a thousand units of housing for homeless exits, whether the city is buying property to convert into permanent supportive housing or putting money into rent subsidies. A year ago the June budget was able to access the prop c money, and the budget included 4,000 units of housing for homeless exits. About 5,000 people a year become homeless in San Francisco. I did a research project last summer and found that indeed 20% of them are able to be housed through the coordinated entry system. So it'll be interesting to see how that number changes with the Prop C money. Though homeless advocates have to constantly fight with the city to make sure the money is used for homeless exits and not services for homeless people that don't actually help them get housed (like food).

(The research I was doing:

One factor not mentioned yet is that the homeless are allowed to reside on the streets of SF. 

The streets are someone's property (usually - municipal property), and the property owner can implement and enforce rules that ban the homeless from residing there. For example, under a threat of incarceration (including mental hospitals, addiction treatment centers etc).

I don't know if it's a good idea to implement such rules (I haven't thought deeply about it). But it could greatly contribute to the difference in visible homelessness between, say, SF and Beijing. 

That's certainly a good solution to the homelessness problem : force them to leave the streets in the city center. Then they can go on being miserable at some place were we don't see them, which totally solves the problem.

(This comment may contain some form of irony, thread carefully)

I'm not advocating for it, but I can see how relocation could reduce the misery. There is a lot of crime in SF, the access to recreational drugs is too easy, and the rents are too high. 

In particular, for the homeless drug addicts, an Amish-like community in a deeply rural area could be much more healthy. A complete isolation from all recreational drugs alone could be a major factor, if done properly (including medical supervision). If I remember correctly, a large part (if not the majority) of the homeless in SF are drug addicts. 

Drug addicts also commit disproportionately many crimes, so removing them from city centers could noticeably reduce crime (which is causing a lot of misery for the rest of the city population). 

Measuring homelessness by state seems like a deeply flawed approach given that it involves aggregating enormous geographic areas with different conditions. 

I wanted to add a (possible, additional) factor in that I didn’t see included, although I don’t know how you would test it. Guess: because of the size and interconnectedness/maturity of SF’s homeless network, it might be easier to be less isolated/more connected while homeless in SF than in other places. It seems clear that in some parts of SF (I’m thinking of particular streets in SOMA and the Tenderloin, and at Civic Center) the people who live on the streets are connecting with each other and also often enjoying each other’s company. They look like friends. And this may be part of what makes the homeless population more visible in SF than in other places—they’re hanging out together, often talking loudly, and they’re often in the same place all the time, every day. There is a social scene it is possible to be part of, and this might feel better than living alone somewhere much nicer.

(There are still many people who seem both isolated and homeless, too.)

"I also found that, controlling for rents, the partisanship of a state did not predict homelessness""

Did Partisanship correlate with rents if so what direction?


Good post, it makes me think a lot of the homeless crisis is more literally "I can't buy a home". 

I don't buy the housing cost / homelessness causation. There are many poor cities in the US that have both low housing costs and high homelessness. This page mentions Turlock, CA, Stockton, CA, and Springfield, MA as among the top 15 places with the highest homelessness rates; a quick Zillow search indicates they all have a fair bit of cheap housing.

The relationship between homelessness and state-wide housing costs is probably caused by a latent variable: degree of urbanization. Cities are both more expensive and have more homelessness, and states vary widely along the urban/rural dimension.

You also missed a strong countervailing factor which would tend to reduce SF's homelessness: demographics. SF is has fewer blacks than the nation as a whole, and blacks are more likely to be homeless. SF is also disproportionately Asian, and Asians are much less likely to be homeless.

I think SF's homelessness problem is caused by a very simple reason: SF is a relatively pleasant place to be a street person. This is partially because of the weather, as you mentioned, but also because the city is quite tolerant of the homeless population and has a lot of services for them.

On the state level, the correlation between urbanization and homelessness is small (R^2 = 0.13) and disappears to zero when you control for housing costs, while the reverse is not true (R^2 of the residual = 0.56). States like New Jersey, Rhode Island, Maryland, Illinois, Florida, Connecticut, Texas, and Pennsylvania are among the most urbanized but have relatively low homelessness rates, while Alaska, Vermont, and Maine have higher homelessness despite being very rural. There's also, like, an obvious mechanism where expensive housing causes homelessness (poor people can't afford rent).

The correlation between homelessness and black population in a state is actually slightly negative (R^2 = 0.09). Mississippi, Louisiana and Alabama are some of the blackest states and have the lowest homelessness rates in the US; Hawaii, Oregon, Alaska and Vermont have some of the highest despite being <5% black. 

Data from 538's Urbanization Index:

I also don’t buy that there is a causal relationship between high living costs and high homelessness. As a Bay Area resident, it’s pretty clear that people don’t go from “can’t afford a home” to “homeless” – they go from “can’t afford a home“ to ”resident of Boise”. 

Someone else made this point on a post I read recently, but I can’t remember where (Maybe it was Bryan Caplan?) But it extends the observation above as follows: 

The Bay Area (and other high-cost areas) are, all else being equal, desirable places to live. People don’t live here primarily because it is too expensive to have a decent quality of life here. 

A homeless person, however, gets to live here (i.e., a “nice” place to live) without having to face the high housing costs, because they don’t pay those high housing costs.

That seems like a semi-plausible explanation of why high housing costs would be correlated to high homelessness without causing that homelessness. 

It's not true that people move out of the state when they become homeless. Because their friends and family are here, they often stay. 75% of the homeless people in San Francisco are from here. The housing price matters because poor people here are generally only able to stay because of things like rent control, so once they lose their housing (for example through an Ellis eviction or a fire), it's very difficult for them to get housing in the city again. So if they want to live in the city where their support network is, then the only option may be to live unsheltered or in their vehicle.

For those that can make it in another city (usually where a relative is willing to take them in), San Francisco is happy to pay for their transportation through the homeward bound program. Buying somebody a ticket out of San Francisco is the cheapest way for the city to make it look like it's ending homelessness. But the city doesn't actually track whether or not these people are able to make it wherever they end up.

Maybe when you say people go from can't afford a home to moving to Boise, you are talking about people who can't afford to buy a home, not people who can't afford to rent a home

The problem with your "resident of Boise" theory is that it costs an ungodly amount of money to move, all of which must be paid upfront. Moving out of state is even worse, because it often means transferring jobs. This is a huge barrier for a lot of people, and for many its utterly prohibitive.

Then there's the fact that homelessness generally feels like a weird transitional phase, and you bear it with as much grace as possible and hope you have kind friends.

My father is a very soft-hearted person, and ever since I was a child he has let people stay with him who are down on their luck for whatever reason. There have been times where we had like 6 or 7 extra people staying out at my dad's house at once, sharing bedrooms or sleeping on the couches or the living room floor.

I can tell you that most of these people are unable to generate much money at all in spite of their best daily efforts. They simply cannot do it. It's not that they are mentally ill or grossly incompetent. Many of them demonstrate at least a base level of competency in many different directions. The problem seems to be that they quite simply are not needed. There just ends up being no place for them anywhere. They get outcompeted in the workforce as available positions shrink away, rising above their level of proficiency, becoming increasingly niche and specialized or automated, leaving them behind. Many are older adults without living relatives who are established and willing to help them out, and no one else really cares very much what kind of trauma a relatively uneducated 43-year-old man is suffering either, especially if he has a spotty criminal background.

Compounding these troubles is the fact that these people are poor and look poor. They don't use correct grammar (not because they're stupid per se; in fact, this usually comes down to the local culture and the communication styles they're exposed to). They often don't meet the minimum requirements listed on most job applications: smart phone with stable internet connection, professional appearance, reliable transportation, home address, etc.

So you see how people end up getting stuck in a downward spiral. Then they fall out onto the streets and get treated like a nuisance for it - because they more or less are a nuisance. A homeless person is a living, breathing allegory of want.

On the other hand, I have an old friend who got out of prison (drug charges) a few years ago and was placed in a 1-br apartment paid for by the city. He has never struggled with housing since then, since he doesn't have to pay for it. I find that the problem is pretty much invariably financial when it comes right down to it.

On the other hand, I have friends hitting their 30s who very much are in the work force and who are nevertheless struggling to accept the fact that they will probably always be stuck living with no fewer than 9 roommates, come hell or high water.

Tons of people are hanging on by their fingernails or just aren't able to hack it at all and will probably end up sooner or later on the streets of San Francisco or at my dad's house or similar.

It's hard out here.

Thank you! Very clean and solid analysis.

Would it be easy to do the same analysis, but with housing vacancy rate instead of housing costs? Spoilering my prediction to avoid biasing results


 Vacancy and cost will be tightly correlated, but locations with below-the-trendline vacancy will have proportionately more homelessness than locations with above-the-trendline costs

I also found that, controlling for rents, the partisanship of a state did not predict homelessness (using the Partisan Voting Index)


This is not a useful way of looking at this; homelessness would be almost entirely controlled by city, not state, policies. State partisanship in large part measures not how blue or red the states' cities are, but rather how urban or rural the state as a whole is.

Note that, since these causes are multiplicative, percentage here is calculated by first taking the log, then dividing


Could someone explain why you do this? 

I'd love to see a Github version of this post. I'd be interested in re-running the same data to generate results for other cities (e.g. Seattle).

Or maybe some-such site/notebook already exists?  If so, please do tell!

Curated. On one hand, this is a sort of niche, political topic. But I think "how does homelessness work?" is relatively generalizable, and I appreciate Alyssa weighing into the topic with a straightforward, data-driven fact post style approach. 

Does San Francisco look like more of an outlier if you plot unsheltered homeless vs house price?

[+][comment deleted]1mo 1
[+][comment deleted]1mo 1